- 12 Mar, 2015 1 commit
-
-
Moving live start time calculation earlier in the code flow to eliminate a race condition, repeating it on first play to create a smooth experience, and making sure currentTime is updated to match these changes.
Brandon Bay committed
-
- 11 Mar, 2015 6 commits
-
-
David LaPalomento committed
-
Update a test case to verify that calls to fillBuffer() and drainBuffer() are not scheduled after the player has been disposed.
David LaPalomento committed -
Add a test case to show that buffer checks begin immediately on setting a new source.
David LaPalomento committed -
Make sure the playlist loader is immediately disposed when the source is changed instead of waiting for sourceopen. Previously, it was possible that the source would be changed and fillBuffer() would run before the new set of playlists had been loaded. In that case, HLS would download a segment from the old playlist and block drainBuffer() from making progress because playlist.segments in the segmentBuffer element would be undefined. Switch from using timeupdate to trigger buffering to using straight window.timeout which fixes #160 and fixes #133.
David LaPalomento committed
-
- 02 Mar, 2015 7 commits
-
-
David LaPalomento committed
-
The standard specifies uppercase ASCII hex digits should be used for MPEG2 text track dispatch types.
David LaPalomento committed -
Feature detect support for addTextTrack so HLS remains functional with older video.js versions.
David LaPalomento committed -
Parse the elementary stream descriptor out of the PMT declaring the metadata stream, if present. Use those bytes to build the inBandMetadataDispatchType field on the generated text track.
David LaPalomento committed -
Describe ID3 metadata parsing and translation to metadata text tracks. Link to helpful background docs.
David LaPalomento committed
-
- 28 Feb, 2015 1 commit
-
-
Listen for metadata events during transmuxing and expose each ID3 frame as a cue at the appropriate time. Assume TXXX and WXXX frames will use UTF-8 encoded text. Start work on exposing the dispatch type for the generated text track.
David LaPalomento committed
-
- 27 Feb, 2015 1 commit
-
-
Decode the fields of simple ID3 tags. Adjust the starting PTS value for the metadata stream so that it aligns with video and audio streams.
David LaPalomento committed
-
- 25 Feb, 2015 3 commits
-
-
Fix up karma configuration so tests pass from the command line. Attach the bytes of the ID3 frame body in metadata stream events.
David LaPalomento committed -
Create a test file for metadata streams and add a basic test for parsing an ID3 tag. The parser is not implemented yet so this test fails.
David LaPalomento committed
-
- 09 Feb, 2015 1 commit
-
-
David LaPalomento committed
-
- 08 Feb, 2015 6 commits
-
-
David LaPalomento committed
-
Key URLs should be fetched relative to the master playlist, if one exists, or their containing playlist otherwise.
David LaPalomento committed -
Hopefully all agree is a better name choice. Also, maybe this will trigger Travis CI to a better build.
Steve Mayhew committed -
The EXT-X-KEY URI attribute is not processed to spec. It should be interpreted as absolute or relative to the playlist containing the EXT-X-KEY tag. This is in the IETF spec (https://tools.ietf.org/html/draft-pantos-http-live-streaming-13#section -3.1). The fix is simple, use the same URL resolution code the TS segments are using. Here is a stream that only works correctly with this fix: http://flashls-test.s3.amazonaws.com/7_copy.m3u8
Steve Mayhew committed
-
- 05 Feb, 2015 4 commits
-
-
As requested by @BCdmlap, changing the discontinuity seek to be a new test instead of a modification to an existing test.
bc-bbay committed -
On seeking to a discontinuity point, the buffer would not refill and the player would stop working. This commit fixes that issue, and adjusts the test to catch the previous failure.
Brandon Bay committed
- 27 Jan, 2015 1 commit
-
- 22 Jan, 2015 1 commit
-
-
Matthew McClure committed
-
- 21 Jan, 2015 1 commit
-
-
On seeking to a discontinuity point, the buffer would not refill and the player would stop working. This commit fixes that issue, and adjusts the test to catch the previous failure.
Brandon Bay committed
-
- 14 Jan, 2015 1 commit
-
-
The old headers were claiming full copyright and no license. The videojs-hls.js header was modified to include a link to the main license from the repo.
Gary Katsevman committed
-
- 07 Jan, 2015 3 commits
-
-
David LaPalomento committed
-
If the downstream developer has a better way of estimating bandwidth for the initial playlist, use that value.
David LaPalomento committed
-
- 06 Jan, 2015 3 commits
-
-
David LaPalomento committed
-
Multiply the bandwidth measurement from the master playlist by five to account for the higher ratio of latency to throughput for this request relative to segments. Using the bandwidth number directly was almost always resulting in very low initial bandwidth estimates and poorer quality startup than necessary. The scaling factor was obtained by testing a number of videos from a high throughput/low latency connection as an upper bound and the same connection throttled to "DSL" levels with Network Link Conditioner for the lower bound. Update the playlist switching simulator to apply initial bandwidth estimates and reduce the simulation duration a bit so that the early behavior is more visible.
David LaPalomento committed
-
