[Webkit-unassigned] [Bug 139813] [MSE] Implement per TrackBuffer buffered.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=139813

--- Comment #4 from Bartlomiej Gajda <b.gajda at samsung.com> ---
> PlatformTimeRanges has a maximumBufferedTime() which would work well here.
> Again, maximumBufferedTime() could be used here.

will fix next week.

> recalculateBuffered()?

Good name, but ultimately I think I will drop caching.

I thought it's safe to cache intersection, and at the moment of calling buffered() just : if (ended) cachedBuffered.add(cachedBuffered.end, highestEndTime), but this does not work for:

1st trackBuffer buffered: (0,2) (7,10)
2nd trackBuffer buffered: (4-6)

"Correct" answer from spec would be (7-10) as both trackBuffers would be extended to 10th second by point 4.2 of buffered.

My approach would have cached buffered intersection = (0,0), so it would add (0-10).

So the alternative would be to remember "did MediaSource ended state changed", and then do recalculate buffered but that's duplicating code and I don't see any improvment by that appraoch..

I think I will just call this algorithm directly each time, but that will be done next week, along with more layoutTest for 'ended' scenarios.

-- 
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/20150109/f0bf7e4e/attachment-0002.html>