Godot supports many different compilers and for production releases we
have to support 3 currently: GCC8, Clang6, and MSVC2017. These compilers
all do slightly different things with -ffast-math and it is causing
issues now. See #24841, #24540, #10758, #10070. And probably other
complaints about physics differences between release and release_debug
builds.
I've done some performance comparisons on Linux x86_64. All tests are
ran 20 times.
Bunnymark: (higher is better)
(bunnies) min max stdev average
fast-math 7332 7597 71 7432
this pr 7379 7779 108 7621 (102%)
FPBench (gdscript port http://fpbench.org/) (lower is better)
(ms)
fast-math 15441 16127 192 15764
this pr 15671 16855 326 16001 (99%)
Float_add (adding floats in a tight loop) (lower is better)
(sec)
fast-math 5.49 5.78 0.07 5.65
this pr 5.65 5.90 0.06 5.76 (98%)
Float_div (dividing floats in a tight loop) (lower is better)
(sec)
fast-math 11.70 12.36 0.18 11.99
this pr 11.92 12.32 0.12 12.12 (99%)
Float_mul (multiplying floats in a tight loop) (lower is better)
(sec)
fast-math 11.72 12.17 0.12 11.93
this pr 12.01 12.62 0.17 12.26 (97%)
I have also looked at FPS numbers for tps-demo, 3d platformer, 2d
platformer, and sponza and could not find any measurable difference.
I believe that given the issues and oft-reported (physics) glitches on
release builds I believe that the couple of percent of tight-loop
floating point performance regression is well worth it.
This fixes#24540 and fixes#24841
The current system for capturing the mouse and generating motion events on X11
has issues with inaccurate and lopsided input. This is because both
XQueryPointer and XWarpPointer work in terms of integer coordinates when the
underlying X11 input driver may be tracking the mouse using subpixel
coordinates. When warping the pointer, the fractional part of the pointer
position is discarded.
To work around this issue, the fix uses raw motion events from XInput 2. These
events report relative motion and are not affected by pointer warping.
Additionally, this means Godot is able to detect motion at a higher resolution
under X11. Because this is raw mouse input, it is not affected by the user's
pointer speed and acceleration settings. This is the same system as SDL2 uses
for its relative motion.
Multitouch input on X requires XInput 2.2. Raw motion events require
XInput 2.0. Since 2.0 is old enough, this is now the minimum requirement to
use Godot on X.
This mirrors behavior on Windows and MacOSX where Godot tries to default
to a dGPU if it exists. This doesn't work for Nvidia optimus yet but
this can maybe be added later.
Contrarily to what #23434 assumed, this is not a memory leak,
the VisualServerRaster instance is passed as a parameter to
VisualServerWrapMT's constructor.
Fixes#23437.
Fixes the following Clang 7 warnings:
```
editor/editor_help.h:123:7: warning: 'EditorHelpIndex::popup' hides overloaded virtual function [-Woverloaded-virtual]
editor/editor_help.h:95:7: warning: 'EditorHelpSearch::popup' hides overloaded virtual function [-Woverloaded-virtual]
editor/editor_help.h:96:7: warning: 'EditorHelpSearch::popup' hides overloaded virtual function [-Woverloaded-virtual]
editor/plugins/curve_editor_plugin.h:141:15: warning: 'CurvePreviewGenerator::generate' hides overloaded virtual function [-Woverloaded-virtual]
editor/plugins/script_editor_plugin.h:70:7: warning: 'ScriptEditorQuickOpen::popup' hides overloaded virtual function [-Woverloaded-virtual]
editor/quick_open.h:69:7: warning: 'EditorQuickOpen::popup' hides overloaded virtual function [-Woverloaded-virtual]
main/tests/test_io.cpp:53:15: warning: 'TestIO::TestMainLoop::input_event' hides overloaded virtual function [-Woverloaded-virtual]
servers/audio/effects/audio_effect_record.h:69:15: warning: 'AudioEffectRecordInstance::process_silence' hides overloaded virtual function [-Woverloaded-virtual]
core/os/memory.h:119:2: warning: destructor called on non-final 'ContextGL_X11' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
core/os/memory.h:119:2: warning: destructor called on non-final 'EditorScriptCodeCompletionCache' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
core/os/memory.h:119:2: warning: destructor called on non-final 'Engine' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
core/os/memory.h:119:2: warning: destructor called on non-final 'PhysicalBone::JointData' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
core/os/memory.h:119:2: warning: destructor called on non-final 'VisualServerScene' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
core/os/memory.h:119:2: warning: destructor called on non-final 'VisualServerViewport' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
```
Namely:
[-Wunneeded-internal-declaration]
[-Wunused-comparison]
[-Wunused-const-variable]
[-Wunused-function]
[-Wunused-private-fields]
Fixes the following Clang 7 warnings:
```
editor/plugins/script_editor_plugin.cpp:1417:20: warning: function '_find_node_with_script' is not needed and will not be emitted [-Wunneeded-internal-declaration]
editor/scene_tree_dock.cpp:1859:14: warning: function '_find_last_visible' is not needed and will not be emitted [-Wunneeded-internal-declaration]
modules/gdscript/gdscript_parser.cpp:7838:19: warning: equality comparison result unused [-Wunused-comparison]
scene/resources/mesh.cpp:549:35: warning: unused variable '_array_types' [-Wunused-const-variable]
scene/resources/mesh.cpp:563:18: warning: unused variable '_format_translate' [-Wunused-const-variable]
drivers/gles3/rasterizer_scene_gles3.cpp:54:28: warning: unused function 'store_transform2d' [-Wunused-function]
core/io/file_access_network.h:50:6: warning: private field 'ml' is not used [-Wunused-private-field]
core/io/file_access_zip.h:93:14: warning: private field 'archive' is not used [-Wunused-private-field]
core/io/resource_format_binary.h:122:6: warning: private field 'bin_meta_idx' is not used [-Wunused-private-field]
core/message_queue.h:47:9: warning: private field 'mutex' is not used [-Wunused-private-field]
main/tests/test_gui.cpp:63:11: warning: private field 'control' is not used [-Wunused-private-field]
modules/gdscript/gdscript_parser.h:558:7: warning: private field 'completion_static' is not used [-Wunused-private-field]
platform/x11/os_x11.h:148:11: warning: private field 'ip_unix' is not used [-Wunused-private-field]
platform/x11/os_x11.h:180:7: warning: private field 'net_wm_icon' is not used [-Wunused-private-field]
platform/x11/os_x11.h:189:6: warning: private field 'audio_driver_index' is not used [-Wunused-private-field]
platform/x11/os_x11.h:190:15: warning: private field 'capture_idle' is not used [-Wunused-private-field]
servers/physics/body_pair_sw.h:79:6: warning: private field 'cc' is not used [-Wunused-private-field]
servers/visual/visual_server_raster.h:62:7: warning: private field 'draw_extra_frame' is not used [-Wunused-private-field]
```
Also finally move freetype to its own env and disable warnings for it.
Still needs some work to fix the awkward situation of the freetype and
svg modules used in scene/ and editor/ respectively.
The two POSIX style crash handlers (OSX and X11) now remove their signal
handlers when they are destroyed.
Additonally if they are called while no OS singleton is set, they will
simply abort(). This should not happen now that they remove themselves,
but if a future change seperates OS object and crash handler lifetimes,
this may be easier to report/debug than hanging on SIGSEGV.
This allows more consistency in the manner we include core headers,
where previously there would be a mix of absolute, relative and
include path-dependent includes.
On X11 when we send an XResizeWindow request to the X server it is happy
to say it is done when the request has been handed over to the window
manager. The window manager itself may however take some time to
actually do the resize. Godot expects that a resize request is
immediate. To work around this issue we could implement the whole
_NET_WM_SYNC_REQUEST protocol. However this protocol does not fit very
well with the way we currently process X events and would when
implemented in the current framework still cause a 1 frame delay between
a resize request and the actual resize happening.
This fixes#21720
When we start the engine we haven't yet gotten any X11 motion events so
we don't yet know where our mouse cursor is located. Instead we now
query the X server for this information when we start and update the
appropriate values.
In addition when we move the window we also update the mouse position
based off of X server knowledge as we will also not have received any
mouse motion events.
this fixes#8145 (for X11 only)
When setting an icon that is too large previously Godot would die with a
X Error of failed request: BadLength error. To avoid this we install an
error handler right before we set an icon. If the error handler triggers
we halve the icon size until it works or until we've reached a 0 size on
either width or height.
We print a warning when this happens to alert developers.
This fixes#19716
This adds a static is_viable() method to all rasterizers which has to be
called before initializing the rasterizer. This allows us to check what
rasterizer to use in OS::initialize together with the GL context
initialization.
This commit also adds a new project setting
"rendering/quality/driver/driver_fallback" which allows the creator of a
project to specify whether or not fallback to GLES2 is allowed. This
setting is ignored for the editor so the editor will always open even if
the project itself cannot run. This will hopefully reduce confusion for
users downloading projects from the internet.
We also no longer crash when GLES3 is not functioning on a platform.
This fixes#15324
Modern distributions such as Fedora do not ship 'xdialog' with their
default deployment. This commit adds support for Gnome's Zenity as well
as KDE's kdialog.
- Refactored all builder (make_*) functions into separate Python modules along to the build tree
- Introduced utility function to wrap all invocations on Windows, but does not change it elsewhere
- Introduced stub to use the builders module as a stand alone script and invoke a selected function
There is a problem with file handles related to writing generated content (*.gen.h and *.gen.cpp)
on Windows, which randomly causes a SHARING VIOLATION error to the compiler resulting in flaky
builds. Running all such content generators in a new subprocess instead of directly inside the
build script works around the issue.
Yes, I tried the multiprocessing module. It did not work due to conflict with SCons on cPickle.
Suggested workaround did not fully work either.
Using the run_in_subprocess wrapper on osx and x11 platforms as well for consistency. In case of
running a cross-compilation on Windows they would still be used, but likely it will not happen
in practice. What counts is that the build itself is running on which platform, not the target
platform.
Some generated files are written directly in an SConstruct or SCsub file, before the parallel build starts. They don't need to be written in a subprocess, apparently, so I left them as is.
This commit makes operator[] on Vector const and adds a write proxy to it. From
now on writes to Vectors need to happen through the .write proxy. So for
instance:
Vector<int> vec;
vec.push_back(10);
std::cout << vec[0] << std::endl;
vec.write[0] = 20;
Failing to use the .write proxy will cause a compilation error.
In addition COWable datatypes can now embed a CowData pointer to their data.
This means that String, CharString, and VMap no longer use or derive from
Vector.
_ALWAYS_INLINE_ and _FORCE_INLINE_ are now equivalent for debug and non-debug
builds. This is a lot faster for Vector in the editor and while running tests.
The reason why this difference used to exist is because force-inlined methods
used to give a bad debugging experience. After extensive testing with modern
compilers this is no longer the case.
-Project/Editor settings now show tooltips properly
-Settings thar require restart now will show a restart warning
-Video driver is now visible all the time, can be changed easily
-Added function to request current video driver
[x11] Preserve window size when calling this method.
[osx] Make sure it don't make the window resizable if it's not needed.
[windows] clean up the code.
- Fix a bug when mouse is confined don't update the cursor shape.
- Don't let the mouse leave the window when resizing to a smaller
resolution when MOUSE_MODE_CONFINED.
- Fix set_borderless_window to preserve the actual video_mode.widht/height.
Now generating mouse events from touch is optional (on by default) and it's performed by `InputDefault` instead of having each OS abstraction doing it. (*)
The translation algorithm waits for a touch index to be pressed and tracks it translating its events to mouse events until it is raised, while ignoring other pointers.
Furthermore, to avoid an stuck "touch mouse", since not all platforms may report touches raised when the window is unfocused, it checks if touches are still down by the time it's focused again and if so it resets the state of the emulated mouse.
*: In the case of Windows, since it already provides touch-to-mouse translation by itself, "echo" mouse events are filtered out to have it working like the rest.
On X11 a little hack has been needed to avoid a case of a spurious mouse motion event that is generated during touch interaction.
Plus: Improve/fix tracking of current mouse position.
** Summary of changes to settings: **
- `display/window/handheld/emulate_touchscreen` becomes `input/pointing_devices/emulate_touch_from_mouse`
- New setting: `input/pointing_devices/emulate_mouse_from_touch`
[Linux] Ensures that the custom cursor will be used when changing to
MOUSE_MODE_VISIBLE. Fix#3086
[Windows] Fix cursor flickering when MOUSE_MODE_HIDDEN.
[Mac] Fix possible cursor flicker when MOUSE_MODE_HIDDEN.
This commit adds a new rendering backend, GLES2, and adds a
project setting to enable it.
Currently this backend can only be used on the X11 platform,
but integrating into other platforms is planned.
Found via `codespell -q 3 --skip="./thirdparty,./editor/translations" -I ../godot-word-whitelist.txt`
Whitelist consists of:
```
ang
doubleclick
lod
nd
que
te
unselect
```
We were already linking libstdc++ statically for official binaries,
protecting us against most portability issues. But apparently since
we started using GCC 7 for official builds, we also need to link
libgcc statically for at least 32-bit builds to be portable.
Fixes#16409.
The target of the TARGETS type should be XA_ATOM and not XA_TARGETS when
requested. Since we are sending a number of ATOMS the size should be set
to the integer size and not the char size.
The size field of the atoms is also the number of atoms and not the size
of the array. This caused some clients to wrongly interpret the data and
read garbage in the X11 packet.
I also add the more modern representation for UTF-8 and clarify the
error message if a client attempts to request a type we don't know
about.
This fixes#10431
This adds a separate_debug_symbols option to the x11, windows, and osx
targets. This will default to adding normal debugging symbols to the
artifacts and only splits them when separate_debug_symbols=yes on the
Scons command line.
Also made LINK and CXXFLAGS configurable as command line options.
Note that LINK currently expects the *compiler* that will be used
for linking and will call its configured linker behind the scenes
(so g++, clang++, etc., not ld.gold). See #15364 for details.
Using `misc/scripts/fix_headers.py` on all Godot files.
Some missing header guards were added, and the header inclusion order
was fixed in the Bullet module.
-Fixes to unwrapper (remove degenerates), makes Thekla not crash
-Added optional cancel button in EditorProgress
-Added function to force processing of events (needed for cancel button)
Setting the vsync in the main thread, after the rendering thread starts
and takes the OpenGL context fails, so we need to do that before.
Also, for some reason, the main thread cannot make current the context
anymore.
Fixes#13447
Spec version 0.7 from https://standards.freedesktop.org/basedir-spec/basedir-spec-0.7.html
(latest as of this commit).
Three virtual methods are added to OS for the various XDG paths we will use:
- OS::get_data_path gives XDG_DATA_HOME, or if missing:
~/.local/share on X11, ~/Library/Application Support/ on macOS and %APPDATA% on Windows
- OS::get_config_path gives XDG_CONFIG_HOME, or if missing:
~/.config on X11, ~/Library/Application Support/ on macOS and %APPDATA% on Windows
- OS::get_cache_path gives XDG_CACHE_HOME, or if missing:
~/.cache on X11, ~/Library/Caches on macOS and %APPDATA% on Windows
So for Windows there are no changes, for Linux we follow the full split spec
and for macOS stuff will move from ~/.godot to ~/Library/Application Support/Godot.
Support for system-wide installation of templates on Unix was removed for now,
as it's a bit hackish and I don't think anyone uses it.
user:// will still be OS::get_data_path() + "/godot/app_userdata/$name" by
default, but when using the application/config/use_shared_user_dir option
it will now use XDG_DATA_HOME/$name, e.g. ~/.local/share/MyGame.
For now everything still goes in EditorSettings::get_settings_dir(), but
this will be changed in a later commit to make use of the new splitting
where relevant.
Part of #3513.
There are still some left in the Android Java code, even stuff to swap between
GLES1 and GLES2 support from early Godot days... would be good to see some cleanup
there too one day.
The "graphics/api" option for Android exports is removed, as only GLES 3.0 is supported.
It can be readded when GLES 2.0 support comes back. Fixes#13004.
Travis always has massive backlog of macOS builds, so we can't rely on them
too much.
The iphone build was mostly useful to spot tools=no or target=release_debug
issues, so replacing it by an appropriate X11 build.
Now that #12009 is merged, we should let the system find the binary on
the users $PATH and don't assume we know where to look for them in
different distributions.
This release hides many struct members which provides easier forward
compatibility but is a break from previous releases. A few small macros
provide compatibility between both 1.1.0 and 1.0.x.
Fixes#8624.
Rename user facing methods and variables as well as the corresponding
C++ methods according to the folloming changes:
* pos -> position
* rot -> rotation
* loc -> location
C++ variables are left as is.
Now that we have a built-in stacktrace on a segfault it would be useful
to have debug information on debug_release builds so that bugreports can
include this information. Without this debug info we will still get
function names in the backtrace but not file location.
This commit will by default build all targets with minimal debug info
and then strip the information into separate files. On MacOS this is a
.dSYM file, on Linux/MingW this is a .debug file. MacOSX will
automatically load a dSYM file if it exists in its debugger. On
Linux/MingW we create a 'gnu debuglink' meaning that gdb and friends
will automatically find the debug symbols if they exist.
Existing workflow for developers does not change at all, except that we
now create two instead of one build artifact by default.
This commit also adds a 'debug_symbols' option to X11, MacOS, and MingW
targets. The default is 'yes' which corresponds to -g1. The alternatives
are 'no' (don't generate debug infos at all) or 'full' which runs with
-g2. A target=debug build will now build with -g3.
Apparently -ffast-math generates incorrect code with recent versions of
GCC and Clang. The manual page for GCC warns about this possibility.
In my tests it doesn't actually appear to be measurably slower in this
case, and this is used in a batch process so it seems safe to disable
this.
This fixes#10758 and fixes#10070
The pattern and replacement matching behaviour has been changed purely
due to the nature of switching to a standards-compliant library.
One mistake in the previous behaviour was that named groups didn't have
a number. This has been corrected.
As names are actually just an alias of numbered groups,
RegExMatch::get_name_dict() is now get_names() and is a dict
referring to the group number it represents.
Duplicate names are enabled and the with the first matching instance
used.
Due the lack of a suitable equivalent in PCRE2, RegExMatch::expand() was
removed.
Setting the class hint before mapping the window will allow some
window managers to determine if a window should be treated specially.
This is also in accordance with the ICCCM spec which says that
WM_CLASS should only be changed when a window is in a
withdrawn (unmapped) state.
Fixes#10429
Hein-Pieter van Braam
Rémi Verschelde
Lee Pugh
lupoDharkael
Mikko Mustonen
Cosmic Chip Socket
Marcelo Fernandez
Ibrahn Sahir
George Marques
M. Huri
Fabio Alessandrelli
Aaron Franke
Guilherme Felipe
LikeLakers2
Martin Capitanio
Juan Linietsky
elasota
Viktor Ferenczi
mm
Hugo Locurcio
Duy-Nguyen Ta
Saracen
Max Hilbrunner
Pedro J. Estébanez
bruvzg
karroffel
luz.paz
Iskustvo
BTaskaya
Paolo Perkovic
geequlim
Andreas Haas
Emmanuel Leblond
Juan Linietsky
Enzo Nocera
Colin Kinloch
Nathan Warden
Stefano Bonicatti
Rhody Lugo
AndreaCatania
N0hbdy
santouits
J08nY
Poommetee Ketson
Felix M. Cobos
Elliott Sales de Andrade
letheed
Thomas Herzog
Maxim Sheronov
Wilson E. Alvarez
Zher Huei Lee
Richard Adenling
TwistedTwigleg