Sylvain Becker

@sam The Android_Window check doesn't seem mandatory on its own. the variable Android_window isn't used. we just check that the window has been created and is not destroyed (nor in...

@AntTheAlchemist copy the text into a file: foo.txt ``` cd SDL patch -p1 < foo.txt ```

Hey @AntTheAlchemist , do you have any news on this ?

@AntTheAlchemist, some anr/crash are weird and maybe only side effect of heap corruption. what you can test I guess: - without the patch. - put a SDL_Delay after the Renderer/Window...

@Green-Sky maybe you can add some debug in SDL code to see what's wrong: show if there are some unknown format we could handle: https://github.com/libsdl-org/SDL/blob/main/src/camera/android/SDL_camera_android.c#L270 https://github.com/libsdl-org/SDL/blob/main/src/camera/android/SDL_camera_android.c#L325 display the number of...

@Green-Sky thanks ! I hope this can help could pixel stride be wrong :/ ? (copy paste of row stride ?)

this old pr shows the code I tested before: (dont look at the PR itself but at the old revision) https://github.com/libsdl-org/SDL/pull/8565/files#diff-9859fc8ca0ebc34d849490517ec49ba8f707b0f28e3b6a4034a5309eb146f38f ``` if (frame->num_planes == 3) { /* plane 2...

to make sure, here https://github.com/libsdl-org/SDL/blob/main/src/camera/android/SDL_camera_android.c#L327 num_planes, should go from 3 to 2

here's that probably incorrect: https://github.com/libsdl-org/SDL/blob/main/src/camera/android/SDL_camera_android.c#L353 we should pack the plane and use the smallest stride edit: no, that's should be ok in fact, SDL should be able to handle width...