Window not shown unless get_surface is called.

[backwardspy]

I've put together a minimal example of opening a window with PySDL2. I'm not sure what I'm missing, but my window only shows up if I include a call to window.get_surface() (even if I throw away the result.)

Here's my (working) example:

import sys
import sdl2
import sdl2.ext


def main():
    sdl2.ext.init()
    window = sdl2.ext.Window("PySDL2 Test", size=(320, 200))
    window.show()

    window.get_surface()

    sdl2.ext.TestEventProcessor().run(window)

    sdl2.ext.quit()
    return 0


if __name__ == "__main__":
    sys.exit(main())

If I comment out line 11 (window.get_surface()) then the window is never shown. None of the examples seem to do this, so I feel like I'm missing something.

If it makes any difference, I'm using SwayWM on Wayland.

Thanks!

[a-hurst]

Hey @backwardspy, thanks for the bug report! Oddly enough, your test script creates a window for me regardless of whether window.get_surface() is commented out or not, so it's almost certainly an SDL2 bug and likely a Wayland-specific one. Digging into the docs, the remarks for SDL_GetWindowSurface() (which get_surface() wraps) says:

A new surface will be created with the optimal format for the window, if necessary.

so it might be the case that SDL2 on your system doesn't create a surface by default and get_surface() does the job, resulting in the behaviour you're seeing.

If you modify your local copy of pysdl2 to add a call to SDL_GetWindowSurface() in the show() function in the sdl2.ext.window object like so:

def show(self):
      """Show the window on the display."""
      video.SDL_ShowWindow(self.window)
      video.SDL_GetWindowSurface(self.window)

does that fix the issue on your end? If so, I might officially add it in to the function since it wouldn't really hurt anything and might fix this issue for others.

[backwardspy]

Hey! Unfortunately I am unable to test your solution, as I no longer have access to the box I was using originally and my current machine is a completely different setup (macOS.) I suspect that something between SDL and Wayland was causing it as you say. Perhaps somebody else who is running Wayland can confirm.

[a-hurst]

Totally fine @backwardspy, I guess that's what happens when we don't get around to dealing with bug reports for a while! If I have time over the next few weeks I'll see if I can reproduce the issue with a Linux VM in case I can fix this in time for 0.9.7. Thanks again for the report!

[a-hurst]

Finally tested my proposed fix on a old laptop that I tossed Manjaro Sway on. With the extra line, it seems to work! Fixed in the latest commit.