[Webkit-unassigned] [Bug 179690] [MSE] Use correct range end checks in sourceBufferPrivateDidReceiveSample()

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=179690

--- Comment #8 from Alicia Boya García <aboya at igalia.com> ---
If we are going to diverge from spec-defined behavior, here is an idea: drop
all three removal steps (1.13, 1.14a and 1.14b) and replace them with this
single one:

Remove every existing frame in the track buffer whose presentation range
intersects with the range [PTS+1µs, PTS+DUR-1µs).

I think that is much simpler to understand and makes for more user-predictable
behavior yet still takes JS float conversion effects into account and
algorithmic complexity need not be affected... the most common case should be
still easy to keep O(1) and there should be no much problem in getting the
O(log(n)) search to work (just look at presentation end for the upper bound,
which should be no problem since the ordering of samples is the same whether we
consider PTS or presentation end because both values are monotonically
increasing...).

This simple rule should handle all the real cases that are handled by the
current approach just fine and fix those that are unspecified by the current
MSE spec.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20171115/ca3ddd25/attachment.html>