Apply the supplied xdg-toplevel bounds to resizable windows during initial mapping. Libdecor functionality will have to be added separately, as the functionality needs to be added to the library first.
In reality, this condition will never occur, since the index is checked before calling the display retrieval function, but aggressive LTO with jump threading can generate a warning if this isn't explicitly checked.
Fixes:
src/video/directfb/SDL_DirectFB_render.c: In function ‘DirectFB_CreateRenderer’:
src/video/directfb/SDL_DirectFB_render.c:1153:35: error: assignment to ‘void (*)(SDL_Renderer *, SDL_Texture *, SDL_ScaleMode)’ from incompatible pointer type ‘void (*)(void)’ [-Wincompatible-pointer-types]
1153 | renderer->SetTextureScaleMode = DirectFB_SetTextureScaleMode;
| ^
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
---
Notes:
- declaration was just recently changed by commit 'Fix warning for Android
NDK compiler: "function declaration without a prototype is deprecated in
all versions of C [-Wstrict-prototypes]"', see
ccade50587
WIN_AddDisplay asserted that the returned index is the expected next
one in the list. This can be an unexpected number when displays are
add and removed in a batch (eg, Remote Desktop disconnect and reconnect).
Instead always just make use of the next index, even if it is not the
expected next index value. This fixes issue #9105
(cherry picked from commit bee8874629)
This reverts commit 120b8d4189.
The issue this was patching over in Vanilla-Conquer was the lack of an exposure event when showing a window, which has since been remedied.
Attaching EGL window objects can also cause protocol violations now that the explicit sync protocol is in use, if SDL creates one and then the client tries to attach one itself, so they really shouldn't be created unless the client specifically requested it.
Suppose host has some three-finger gesture. Then we get the following sequence
of events:
DOWN-DOWN-DOWN-MOTION-CANCEL
Note that there is no UP in this sequence. So if we don't handle CANCEL then
we end up thinking that fingers are still touching the screen. Ideally we
should inform the application that cancel has happened as not to trigger
spurious taps but still this is way better than being stuck with phantom
finger touch.
Modern kernels (v5.8+) allow non-root usage of drmDropMaster(), so
we can hold on to our fd after dropping master on it. This fixes
populating drm_fd in the KMSDRM SysWMinfo when using Vulkan.
Also add a missing error check for open() while we're here.
On most Unix platforms supported by SDL, the canonical name used to load
a library at runtime includes its ABI major version, and the name
without a version is not guaranteed to exist on non-developer systems.
libX11-xcb.so.1 is correct on Linux, and probably on other Unix
platforms like FreeBSD.
A notable exception is OpenBSD, which apparently does not use
ABI-suffixed names, so continue to use libX11-xcb.so there.
Signed-off-by: Simon McVittie <smcv@collabora.com>
(cherry picked from commit 7713a7eec7)
Send the fake warp coordinates when emulating warps with relative mode to ensure that it appears that the cursor is always reset to where the application expects it to be.
Fixes cases where games can continue moving the camera if the mouse is lifted during motion.
Cursor visibility in the SDL input layer only reflects whether ShowCursor/HideCursor was called. In the case of relative mode, the cursor can be hidden, but the SDL_Mouse visibility flag will be true.
Track cursor visibility separately in the X11 driver. Fixes the cursor becoming visible when using the warping relative mode with XWayland.
(cherry picked from commit b0713a7d30)
