Lab / / observation

The screenshot tool reports success while the panel stays blank

`SDL_WINDOW_FULLSCREEN_DESKTOP` makes every `eglSwapBuffers` fail on Pi vc4. GLES still renders, so `--screenshot` produces a real image of something the panel never receives.

Rendering, Measurement methodologyraspboard
Machine
UNKNOWN — Pi vc4 is named as the failing target
Commit
UNKNOWN
Number of frames involved
none — this is a mechanism, not a measurement

The project's boot caveats contain this, verbatim:

`SDL_WINDOW_FULLSCREEN_DESKTOP` makes every `eglSwapBuffers` fail on Pi vc4 (GLES still renders, so `--screenshot` looks fine while the panel stays blank/console).

raspboard boot caveats

I am recording it because of the shape of the failure. A full-screen-desktop window mode breaks the swap, and the swap is the thing that puts a picture on the panel. GLES keeps rendering regardless, so the screenshot path still captures a real render of real content.

Why this is worse than a crash

A crash tells you where the problem is. This tells you the output is fine. Anyone debugging a blank display with `--screenshot` gets a good-looking PNG back and concludes the renderer is healthy, which is true, and that the panel is being fed, which is false.

The debugging instrument is rendering the same frame the display path fails to present, so it cannot see the failure it is being used to diagnose. There is no number attached to this entry and there should not be one.