appsinkNewSample() contained code to skip samples whose presentation time starts after media duration. This is paradoxical, because later sourceBufferPrivateDidReceiveSample() extends media duration whenever the presentation end time of the sample is past the original one. It does not sound reasonable that samples extending the duration are okay on one case but outright rejected in the other. Also, if it was about skipping samples, sourceBufferPrivateDidReceiveSample() could do it itself, and already does in other cases. For all these reasons I was very doubtful of the need for this if() and indeed removing this condition didn't cause any new test failure.
Created attachment 403372 [details] Patch
After discussing this removal with Alicia, checking where the original code came from before being upstreamed[1] (I can't recall any rationale for discarding samples beyond the duration), and Alicia checking that it doesn't break any YouTube TV test[2] nor LayoutTest, I agree with removing the code block. [1] https://github.com/eocanha/webkit/commit/86764b20df56#diff-1a0340714c802e905fa36d804906e36eR3270 [2] https://ytlr-cert.appspot.com/2019/main.html
Committed r263962: <https://trac.webkit.org/changeset/263962> All reviewed patches have been landed. Closing bug and clearing flags on attachment 403372 [details].