- 10 Nov, 2015 1 commit
-
-
Fixes #430. If drainBuffer is called after updating is reset on the sourceBuffer but before updateend fires, a segment could be appended twice. Check to make sure a segment hasn't already been appended before proceeding with draining the buffer.
David LaPalomento committed
-
- 06 Nov, 2015 3 commits
- 05 Nov, 2015 1 commit
-
-
- 02 Nov, 2015 2 commits
-
-
- 22 Oct, 2015 1 commit
-
-
- 21 Oct, 2015 3 commits
-
-
Buffered will not be present in segment info when a segment processing is aborted due to seeking. Make sure bufferedAdditions_ does not error or add incorrect timeline information to the segment in that case.
David LaPalomento committed -
Media sources implicitly track expired time by retaining the mapping between presentation timestamp values and the media timeline in the buffer. That allows us to simplify a good deal of code by not tracking it ourselves. Finish updating tests to work against the new timeline start and end annotations on segments instead of the old PTS values. Remove metadata cue translation because that is now handled by contrib-media-sources. Update key fetching in HLSe to occur concurrently with the segment download. All tests are now passing.
David LaPalomento committed
- 15 Oct, 2015 1 commit
-
-
If the old and new buffered ranges have a shared start or end point, that edge should not be interpreted as a new buffered boundary. Fix up a number of the tests. Some tests are still failing.
David LaPalomento committed
-
- 08 Oct, 2015 1 commit
-
-
When playlists are not segment-aligned or began at different times, we could make bad decisions about which segment to load by just incrementing the media index. Instead, annotate segments in the playlist with timeline information as they are downloaded. When a decision about what segment to fetch is required, simply try to fetch the segment that lines up with the latest edge of the buffered time range that contains the current time. Add a utility to stringify TextRanges for debugging. This is a checkpoint commit; 35 tests are currently failing in Chrome.
David LaPalomento committed
-
- 30 Sep, 2015 1 commit
-
-
There are cases when a mediasource is re-opened and we need to account for that
Jon-Carlos Rivera committed
-
- 29 Sep, 2015 8 commits
-
-
-
-
For #400. Flash doesn't currently fire "loadstart" in data generation mode. Triggering that event should probably happen in video.js itself but fire it once the source handler is selected to make things behave somewhat sanely for now.
David LaPalomento committed -
All the transmuxing code has moved into mux.js so we can get rid of the remnants in this project.
David LaPalomento committed
-
- 28 Sep, 2015 2 commits
-
-
Fix up the tests and remove debug logging. Put in one example case for playlist re-synchronization.
David LaPalomento committed -
When switching renditions or dealing with a live stream with unaligned variant playlists, we may discover that the segment we buffered isn't associated with the time range we expected it to be. In that case, adjust our information about timeline positioning and try buffering again.
David LaPalomento committed
-
- 26 Sep, 2015 2 commits
-
-
Live streams in particular may have variant playlists with content windows that are out of sync. Keep playback rolling in the simplest way possible by seeking into the new buffered region whenever this occurs. This is not an ideal user-experience.
David LaPalomento committed -
Fix the stats page to work with video.js 5. Get the measured bitrate graph updating in real time.
David LaPalomento committed
-
- 24 Sep, 2015 7 commits
- 23 Sep, 2015 1 commit
-
-
- 22 Sep, 2015 4 commits
-
-
-
"extends" is disallowed in IE8 so videojs renamed the inheritance helper. Update our usage to match.
David LaPalomento committed -
When a browser could support HTML or Flash based playback modes, allow video.js's techOrder to have precedence. It's very handy to be able to override the mode selection for debugging purposes, for instance.
David LaPalomento committed
- 21 Sep, 2015 2 commits