/godot_dev/platform/android/java/src/com/google/android/vending/expansion/downloader/impl/DownloaderService.java:575: Error: The WIFI_SERVICE must be looked up on the Application context or memory will leak on devices < Android N. Try changing to .getApplicationContext() [WifiManagerLeak]
mWifiManager = (WifiManager) getSystemService(Context.WIFI_SERVICE);
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Explanation for issues of type "WifiManagerLeak":
On versions prior to Android N (24), initializing the WifiManager via
Context#getSystemService can cause a memory leak if the context is not the
application context. Change context.getSystemService(...) to
context.getApplicationContext().getSystemService(...).
1 errors, 0 warnings
Now InputDefault is responsible for giving out joypad device IDs to the platform, instead of each platform handling this itself.
This makes it possible for c++ modules to add their own "custom" gamepad devices, without the risk of messing up events in case the user also has regular gamepads attached (using the OS code).
For now, it's implemented for the main desktop platforms.
Possible targets for future work: android, uwp, javascript
All the warnings are factored out of the platform-specific files and moved to
SConstruct. Will have to check that it does not introduce regressions on some
platforms/compilers.
We were reserving 12 bytes from the buffer for ip, port, and length, but since
IPv6 introduction we should be reserving 24 (IPv6 are 16 bytes)
(cherry picked from commit 5dc7c920bf)
We should probably create a specific function for setting the
recv buffer anyway. UDP sockets does not need to bind (listen)
to be able to call recvfrom. This is especially useful for clients
who just call set_send_address and start communicating with a server.
(cherry picked from commit 9336857132)
TCP status polling is always performed as non blocking.
Trying to put a packet on a connecting socket will fail immediately.
(cherry picked from commit fa0cb7da0e)
- TCP:
- `listen` bind to wildcard "*" -> dual stack socket
- `listen` bind to address -> socket from address type
- `connect` -> resolve using best protocol (UNSPEC), socket from address type
- UDP:
- `listen` bind to wildcard "*" -> dual stack socket
- `listen` bind to address -> socket from address type
- `put_packet`/`put_var` -> resolve using TYPE_ANY (UNSPEC), socket from address type
(to change socket type you must first call `close` it)
(cherry picked from commit 88a56ba783)
On KDE (and possibly others) the "default" cursor theme is actually some system default, not the one you've set in the desktop setting.
This was especially annoying when using a white cursor, as Godot would then reset back to a dark one.
In my case it was also keeping the cursor from changing its shape.
(cherry picked from commit fc84ccc468)
The other subfolders of tools/ had already been moved to either
editor/, misc/ or thirdparty/, so the hiding the editor code that
deep was no longer meaningful.
(Manual redo of 49c065d29c)
During runtime godot calls chdir() several times. This doesn't really
matter normally but when using tools such as gprof the location of the
profiling data is kind of hard to intuit.
With this PR we simply store the current working directory at start and
restore it once we're almost done exiting.
This doesn't use the OS abstractions as when we need to get the current
workdir we haven't yet initialized it (by necessity). This would break
if we tried to build X11 for windows, but since the X11 target is
hardcoded to use the UNIX abstractions I don't think it matters.
(cherry picked from commit d0c2015fe1)
Add initialization for OS_OSX.mouse_mode in OS_OSX::OS_OSX(). mouse_mode now defaults to OS::MOUSE_MODE_VISIBLE.
(cherry picked from commit 6921e11805)
debug_release doesn't turn off optimizations for release target now. Ensure that sanitizer options apply to both C and C++ files.
Built-in optimization/debug flags are prepended such that user-specified flags can override them.
Based on and around the discussion in PR #5194.
(cherry picked from commit 7a85d25218)