999e154b2a
This optimization always reports buffers as 'skipped' (i.e. deliberately not shown), which makes sense for the target case of high FPS content on a lower refresh rate screen, when lots of the frames will **never** be shown. However the optimization also results in reporting buffers as 'skipped' when decoding is a bit slow, resulting in a frame being released one vsync late, which then means we have two frames to release in the same vsync (when the previous vsync was empty). In this case, it would be more correct to report this as a 'dropped' frame (since it was due to slow decoding). Until we can change the logic to distinguish these cases and report them separately, this CL disables the optimization completely in GTS tests. This is needed because we often assert there were zero skipped frames, so slight decoding slowness can cause spurious/flaky test failures (our threshold for dropped frames is non-zero). #minor-release PiperOrigin-RevId: 600406443
ExoPlayer playback test module
Tests playback using ExoPlayer.