adminandClaude Opus 5.5
beb95c17a7
A run that waits for its driven videos does not tick its world, so the clock of every free video sprite stands. Its player played on anyway. The picture ran ahead of the clock by the length of the wait, so On video finished came late and a Pause video sent the picture back. - VideoLibrary.setHeld holds the free players next to the toolbar pause. PlaySession and Player switch it with the wait at the start, after a Load scene and while a recording sets up its encoders. - The hold starts in waitForVideos, not on the next tick. Otherwise a free video started at Play would be a frame ahead. - A free player without its first picture plays on until it has one. Held at once, it would never get one. - startRun and endRun clear the hold, so a run that ends while it waits leaves no held preview behind. Found on the way: - Resuming called play() on a clip without Loop that already stood at its end, and play() starts such a video over. This happened after the toolbar pause already and would have happened after every waiting Load scene. Such a clip now stays at its end. Tests: - Unit video-library 14/14, 2 of them new. 7 deliberate breaks, all red. - e2e 66 (atx): ALL PASSED, 40 checks. New 66t: the driven video came 1.5 s late, the wait took 1686 ms. The free video stayed red all along, blue came at 0.983 s, white at 1.483 s and On video finished at 2.017 s. Without the hold the free video was white in the first frame and only 66t was red. - Gate (atx): CHECK EXIT 0, 339 files, 5253 tests. The guard for the clip end came in while the gate ran. It is checked with tsc, lint and its test file with 3 deliberate breaks. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>