- 26 Jul, 2016 1 commit
-
-
- 25 Jul, 2016 4 commits
-
-
-
-
* Guard against negative media index due to manifest vs actual timing differences One of the possible reasons for us getting a negative media index is that we've recorded actual segment timing information and it differs enough from the manifest's timing information that we can't properly place the first segment. In the event that we're at time 0 and the video isn't live, we can safely assume that we should get the first segment. * Change getMediaIndexForTime to return first segment when time and expired are 0
Garrett committed -
-
- 21 Jul, 2016 5 commits
-
-
explicitly ask for versions on each section of additional info remove the actual section as it will always be the same as description
Brandon Casey committed -
-
-
- 20 Jul, 2016 1 commit
-
-
* Removed segment timeout when on the lowest enabled rendition * Move timeout check to only run when needed
Matthew Neil committed
-
- 15 Jul, 2016 2 commits
-
-
- 30 Jun, 2016 1 commit
-
-
- 29 Jun, 2016 1 commit
-
-
* Working implementation of adaptive seeking * Use this.tech_ * Added the ability for the gap skipper to clean up after itself on dispose
Jon-Carlos Rivera committed
-
- 20 Jun, 2016 1 commit
-
-
- 09 Jun, 2016 3 commits
-
-
- 08 Jun, 2016 2 commits
-
-
* Adds a manual rendition selection api to an instance of HlsHandler * Update the doc with information about the representation api
Jon-Carlos Rivera committed -
* updated the README with more up-to-date information fixing code fence formatting remove bandwidth line from constructor as it is done in setOptions_() added tools to the readme * readme: minor spelling/grammar/format fixes
Brandon Casey committed
-
- 06 Jun, 2016 1 commit
- 02 Jun, 2016 4 commits
-
-
-
-
…PlaylistLoader (#724) The PlaylistLoader signals that the active playlist has changed by triggering a mediachange event and the segmentloader needs to know about the new playlist or it will continue to select segments from the old playlist
Jon-Carlos Rivera committed -
-
- 31 May, 2016 3 commits
- 27 May, 2016 2 commits
-
-
* Added bandwidth stats to HLS * moved bandwidth stats to reside on segment loaders * added accessors for stats on master playlist controller * added getters for stats using master playlist controller on hls * updated tests to check for stats
Brandon Casey committed
- 26 May, 2016 5 commits
-
-
-
Segment durations are defined in the spec as being less than or equal to target-duration
Jon-Carlos Rivera committed -
* Fudge segments that are reported as having a zero-second duration * The fetcher logic basically ignores segments with a duration of zero. Give them a tiny duration so that the fetcher will "see" these segments. * Added tests for and fixed getSegmentBufferedPercent_ calculations * Now returns the percent buffered of the entire segment duration instead of the "adjusted" duration * Handles segments reported as having a zero-duration * Reduced the number of segments that we will attempt to "timeCorrect" when the segment chosen by `checkBuffer_` is already more than 90% buffered to 1 * No longer trigger errors from `timeCorrection_` handling, returning to the previous behavior * Use the tech's setCurrentTime function in segment loaders * Moved getSegmentBufferedPercent to `Ranges` module * Moved correction for zero-duration segments from parse-stream to parser proper
Jon-Carlos Rivera committed -
-
* add tests for all options in HLS * add configuration hierarchy testing * Hls.GOAL_BUFFER_LENGTH is settable to a number greater than 0 with a warning * use Hls.GOAL_BUFFER_LENGTH if it exists * added unit tests to verify warning logs * fixed comment about Mbps vs MB/s
Brandon Casey committed
-
- 23 May, 2016 4 commits
-
-
-
This is because media-sources requires video.js but doesn't require browserify-shim itself, so, instead we need to make sure that we browserify-shim all the dependencies.
Gary Katsevman committed -
-
* Fix how and when timeCorrection_ was being applied so that it works for Flash * Always apply timeCorrection_ when we haven't learned anything even if playlists have changed * Apply timeCorrection_ directly to the currentTime when calling getMediaIndexForTime so that we can work around bad segment.end metadata in the playlist * Emit an error if we have been stuck in a "timeCorrection" loop for more than 5 fetch attempts * Make sure to clear the timeout before setting another in monitorBuffer_ * Clearer naming of variables * Use timeCorrection_ when a chosen segment is more than 90% buffered * Break out timeCorrection_ functionality into a reusable piece * Added tests of the new timeCorrection_ functionality * Test that timeCorrection is applied even when the segment.end data is misleading * Verify that an error is emitted when we have failed to make progress after 5 time-corrections * Ensure that we only have 1 timer if monitorBuffer_ is called multiple times * Moved percent-buffered check out of checkBuffer_ to fillBuffer_
Jon-Carlos Rivera committed
-