Last call for 2.0.6 bugs!

[Eric_Wasylishen]

Exiting fullscreen-desktop on macOS currently causes the application to freeze; could https://hg.libsdl.org/SDL/rev/81e6d0f852fc be reverted (this was an attempt to fix https://bugzilla.libsdl.org/show_bug.cgi?id=3697 ) until we come up with a better fix?

[amruthraj]

@icculus, @slouken ,
Can the original issue reported in https://bugzilla.libsdl.org/show_bug.cgi?id=3697 be prioritised for 2.0.6?

It leads to a very bad user experience while a video is played in full screen. Rendering gets stuck, but any audio playing in background continues.

[Eric_Wasylishen]

It should be fully fixed now in this revision: https://hg.libsdl.org/SDL/rev/6709dc9adb16

[amruthraj]

Fantastic Eric.
Thanks for the help.

@icculus @slouken
Can this be picked up for 2.0.6? Or you guys did already

[icculus]

It’s in for 2.0.6.

[Martin_Gerhardy]

Hi.

Current hg doesn’t compile on debian testing

Error is in SDL_sndioaudio.c - INFTIM is not defined on my system. Part of the manpage of poll()

   Some implementations define the nonstandard constant INFTIM with the
   value -1 for use as a timeout for poll().  This constant is not
   provided in glibc.

Regards
Martin

[icculus]

Yeah, I was surprised to find out that sndio is even available on Linux (I thought it was OpenBSD’s sound API!).

There was a bug open for this in Bugzilla, and I just pushed a fix for it:

SDL: changeset 11193:c4ff854c6586

–ryan.

[icculus]

On that note, what’s the status of the shaped window API? Bugs aside, it seems functional, but it’s not documented anywhere, and the API seems a bit iffy (have to include SDL_shape.h separately, camel-cased enums, functions returning error codes without calling SDL_SetError() etc).

It seems like another candidate for a clean-up for SDL 2.1.

Yeah, the shaped window API is bitrotting, partially because it’s not well supported on many OSes even before you get to SDL (looking at you, macOS…!)

We’ll definitely accept patches to improve it, but right now no one is working on it, or really plans to. The 2.1 “cleanup” would probably be removal of the API.

[q3dev]

Hello,

I wanted to ask about the Raspberry Pi keypress leakage issue. I’ve been able to work around the issue by blocking keyboard input to the X11 server with “xinput” while running ioq3. I’m using 2.0.4 compiled with --disable-video-x11. Is this something that might be addressed in 2.0.6?

Thank you!

[maartenengels]

Hello,

Same area I hope with @q3dev. I get the following error from touch screen event.

INFO: The key you just pressed is not recognized by SDL. To help get this fixed, please report this to the SDL forums/mailing list https://discourse.libsdl.org/ EVDEV KeyCode 330

If there is still time for it, that would be amazing!

Thank you.

[Martin_Gerhardy]

There is a build error since a few revisions. See the bugreport here: https://bugzilla.libsdl.org/show_bug.cgi?id=3757

[icculus]

There is a build error since a few revisions. See the bugreport here: https://bugzilla.libsdl.org/show_bug.cgi?id=3757

I’ll fix this in the morning.

[icculus]

I get the following error from touch screen event.

This happens when touching the screen, and not a virtual keyboard on the screen?

[icculus]

I wanted to ask about the Raspberry Pi keypress leakage issue.

I thought this was fixed recently, after 2.0.5…I could be wrong. You should definitely try the latest in revision control (2.0.4 is pretty old at this point).

[icculus]

The 2.1 “cleanup” would probably be removal of the API.

(Actually, there’s some debate about this, so don’t take this as truth at this time.)

[maartenengels]

Hello @icculus,

It happens when you touch the screen.

Greetings.

[q3dev]

Thanks for your reply. I compiled and tested 2.0.5-11303 and 2.0.6-11330 and the keypress leakage issue is still there. I’ve been using 2.0.4 because 2.0.5 won’t allow text to be typed into the ioquake3 player console, but that’s a separate issue. I’m configuring SDL2 with:

./configure --host=arm-raspberry-linux-gnueabihf \
          --enable-alsa \
          --disable-alsa-shared \
          --disable-pulseaudio \
          --disable-esd \
          --disable-video-mir \
          --disable-video-wayland \
          --disable-video-x11 \
          --disable-video-vivante \
          --disable-video-opengl

I can start a new thread if that would be helpful. Disabling keyboard input to the desktop x11 server with “xinput” works around the issue but requires a wrapper script. Thanks for your help.

[David_Ludwig]

One of the fixes applied for SDL 2.0.6, appears to lead to a bug when trying to use SDL_Renderer logical screen size(s), on OS X, in conjunction with SDL_SetWindowFullscreen calls. I’ve written up some details at https://bugzilla.libsdl.org/show_bug.cgi?id=3793 , however, I am unsure of what a good fix would be.

[icculus]

Heads up, we have a 2.0.6 release candidate, so literally last call for bugs now:

Sdl 2.0.6 prerelease

Thanks to all the people who contributed code and feedback, SDL 2.0.6 is now available as a PRERELEASE build! http://www.libsdl.org/tmp/download-2.0.php Please try this with your games and applications and report any issues to bugzilla: https://bugzilla.libsdl.org In addition to lots of bug fixes and build improvements, here are the major changes in this release: General: Added cross-platform Vulkan graphics support in SDL_vulkan.h SDL_Vulkan_LoadLibrary() SDL_Vulkan_GetVkGetInstance…

Please try this with your games and apps asap! We are looking for life-and-death showstoppers at this point.

Thanks!

[Eric_Wasylishen]

@slouken I think there’s a bug in the “Cleans up AdjustWindowEx calls” commit, in WIN_SetWindowFullscreen:

SDL: changeset 11461:a99b2e1e8791

The old code was using window->windowed->x/y/w/h as the input to AdjustWindowRectEx, but after the refactor it’s using window->x/y/w/h. Alt+enter in testgl2 is not restoring the window size when exiting fullscreen as a result.