According to the spec step2 [1], we should not load resource for the track element which text track is disabled.

[1] https://html.spec.whatwg.org/multipage/media.html#start-the-track-processing-model

According to the spec step2 [1], we should not load resource for the track element which text track is disabled.

[1] https://html.spec.whatwg.org/multipage/media.html#start-the-track-processing-model

According to the spec [1], the user agent must start the track processing model when text track has its text track mode changed.

[1] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:start-the-track-processing-model

Showing track's kind in debug log is helpful.

According to spec [1], it doesn't mention that we have to run track selection in stable state, it just said "the user agent must queue a task to run the following steps".

[1] https://html.spec.whatwg.org/multipage/media.html#honor-user-preferences-for-automatic-text-track-selection

In honor user preferences for automatic text track selection [1], we would set did-perform-automatic-track-selection flag to true [2], and then we won't execute automatic track selection anymore [3].

It means that we would only do automatic track selection one time, and then user has to enable newly added track explicitly by changing its mode.

In this test, we have done the automatic track selection when we added the metadata track to media element's text track list, so we won't run it again, even if the newly added track has default attribute.

Therefore, we have to enable the caption track explicitly. In addition, add the missing event parameter for the function trackAdded().

[1] https://html.spec.whatwg.org/multipage/media.html#honor-user-preferences-for-automatic-text-track-selection
[2] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:did-perform-automatic-track-selection-2
[3] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:did-perform-automatic-track-selection

According to the spec [1], we should empty track's cue list whenever a track element has its src attribute set, changed, or removed.

[1] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:attr-track-src

Failed to load [1] indicates that the text track was enabled, but when the user agent attempted to obtain it, this failed in some way.

According to the spec [2], if fetching fails for any reason (network error, the server returns an error code, CORS fails, etc), or if URL is the empty string, then queue a task to first change the text track readiness state to failed to load and then fire an event named error at the track element.

In addition, when we remove src attribute from a loading or loaded track element, we should also set it to failed to load.

[1] https://html.spec.whatwg.org/multipage/media.html#text-track-readiness-state
[2] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:text-track-failed-to-load

As the src url is the same, we have no need to reload the resource.

There are too many self used in the lambda, we can just capture this to remove redudant self.

To reduce non-neccesary input, we can always capture this and print the address in the log, and use same log level for all of them.

According to the spec [1], the text track list of cues is initially empty. It is dynamically modified when the referenced file is parsed.

Therefore, after reset the track element's url, the number of cues of the text track should be zero, until we start parsing resource and add cue into the list.

[1]
https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:text-track-list-of-cues
https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:text-track-list-of-cues-3

As now we won't start loading for disable tracks, so we have to enable the track by changing its track mode explicitly or adding default attribute before the media element running the automatic track selection.

There are some common changes for these tests.

(1) listen track element's load event
There are some tests listening video's loadedmetadata, but it's wrong. The loading process of media element and track element are completely non-related.
If you would like to check track element's status, you should wait for track element's load event.

(2) enable track explictly
If the text track which has default attribute is added to the media element before the media element starts running automatic track selection [1], then it would be enable by the media element.
Otherwise, you have to enable track explicitly by changing its track mode.

(3) refactor tests
Refactor those tests' structure in order to make them more readable, and add the comment to show what the test purpose is for each test.

[1] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:text-track-7

These elemenets are useless and we can run tests without them.

These tests didn't use region at all, so we have no need to set the pref.

Create test elements in HTML beforehead, which can remove unnecessary JS codes and make test cleaner.

Use named function for callback to reduce the indentation.

This patch do two things in order to trigger loading for track element and wait for correct event to check track's and cues' status after loading finished.

(1) listen track element's load event
There are some tests listening video's loadedmetadata, but it's wrong. The loading process of media element and track element are completely non-related.
If you would like to check track element's status, you should wait for track element's load event.

(2) enable track explictly
If the text track which has default attribute is added to the media element before the media element starts running automatic track selection [1], then it would be enable by the media element.
Otherwise, you have to enable track explicitly by changing its track mode.

[1] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:text-track-7

Refactor those tests' structure in order to make them more readable, and add the comment to show what the test purpose is for each test.

According to spec [1], if the track URL changes so that it is no longer equal to URL while fetching is ongoing, we need to change the text track readiness state tofailed to load and dispatch error.

It didn't mention we have to dispatch error if track element has finished loading.

[1] https://html.spec.whatwg.org/multipage/media.html#sourcing-out-of-band-text-tracks:text-track-failed-to-load-3

This test is used to ensure that we queue 'honor user preferences for automatic text track selection' as a marco task, not a mirco task.

In this test, we would trigger a media event before queuing a text track selection task, and check the text track's mode to know whether the text track selection runs after the task for media event.

The channel might not be created correctly if we pass invaild url (eg. "invalid://url"), we should handle this error.

Sometime wpt runs test even before the document becomes visible, which would delay video.play() and cause play() running in wrong order.

Backed out 22 changesets for browser-chrome failure at browser_cache.js.

Backout link: https://hg.mozilla.org/integration/autoland/rev/100b92fd6219edddc0c788ea7c04e99b85e9c21a

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception&revision=87067f045e1e8c957ce1e3c6b6777a525a1d3fc0&searchStr=%28bc4%29&selectedJob=247906371

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=247906371&repo=autoland&lineNumber=2967

As now we won't automatically load disabled text track, we have to mark track as default in order to trigger loading.

Try result looks good.
https://treeherder.mozilla.org/#/jobs?repo=try&revision=3b4c1323af1641c92fefc9a92f09d19eac2aea1c&selectedJob=248042414

Backed out for causing bug 1548454 to permafail

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=pending%2Crunning%2Ctestfailed%2Cbusted%2Cexception&group_state=expanded&searchStr=windows%2C10%2Cx64%2Casan%2Cmochitests%2Ctest-windows10-64-asan%2Fopt-mochitest-e10s-5%2Cm%285%29&revision=ce4a146da49d57afadd9205edd251c32d5d0bd9a&selectedJob=248075630

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=248075630&repo=autoland&lineNumber=10706

Backout: https://hg.mozilla.org/integration/autoland/rev/6acae89335b04d1d634fe5f6be2637f25e3e81e2

I've already updated the fix in patch17. In order to ensure we won't break any test again, pushing patches to try server before relanding all of them.
https://treeherder.mozilla.org/#/jobs?repo=try&revision=825ad84c20cb0b10b247c6c501e6d6c704420e73

https://hg.mozilla.org/mozilla-central/rev/572e5b0bcab9
https://hg.mozilla.org/mozilla-central/rev/6d16370ca0cf
https://hg.mozilla.org/mozilla-central/rev/a5ed9d3975c4
https://hg.mozilla.org/mozilla-central/rev/c90139d2c836
https://hg.mozilla.org/mozilla-central/rev/7a20deaf11bc
https://hg.mozilla.org/mozilla-central/rev/b19effb108a9
https://hg.mozilla.org/mozilla-central/rev/01efc1080b51
https://hg.mozilla.org/mozilla-central/rev/bb4756110ece
https://hg.mozilla.org/mozilla-central/rev/079555c7d41b
https://hg.mozilla.org/mozilla-central/rev/6c49cf7caac3
https://hg.mozilla.org/mozilla-central/rev/4d15400f8a6d
https://hg.mozilla.org/mozilla-central/rev/f4d1f1b6e2c3
https://hg.mozilla.org/mozilla-central/rev/63dd1f7a3ff4
https://hg.mozilla.org/mozilla-central/rev/bef097125232
https://hg.mozilla.org/mozilla-central/rev/0c1ef149fc1d
https://hg.mozilla.org/mozilla-central/rev/8d7f304e91f7
https://hg.mozilla.org/mozilla-central/rev/9b78a97b3387
https://hg.mozilla.org/mozilla-central/rev/6a593a6deb7e
https://hg.mozilla.org/mozilla-central/rev/590e376de09a
https://hg.mozilla.org/mozilla-central/rev/4159b0ad4c58
https://hg.mozilla.org/mozilla-central/rev/39c66f35a9a7
https://hg.mozilla.org/mozilla-central/rev/5a39102e79ad
https://hg.mozilla.org/mozilla-central/rev/e210709d41e9

According to the spec step2 [1], now we won't start loading for the track element with disabled text track.

[1] https://html.spec.whatwg.org/multipage/media.html#start-the-track-processing-model

Is this a change that actually needs to be documented? Is the web developer or end user really going to notice this, typically? Feels a lot like an implementation detail, so to speak.

It will be good if we can document it. To let users know that if they want to load a track and do any related stuffs on that track, such as waiting load event, that track should be hidden or showing.

Documentation updates:

• Updated TextTrack.mode for bug 1550633; also added example.
• Updated TextTrack for bug 1550633.
• Submitted BCD PR 4579 with the TextTrack updates
• Updated Firefox 69 for developers

There is a little inaccuracy for the TextTrack.mode [1], the mode of the text track, which has default keyword, would only be automatically changed to showing or hidden by executing honor user preferences for automatic text track selection [2]. That selection would only be done ONCE, so if you add default on a diabled track after finished the selection, the track mode would not be changed.

[1] https://wiki.developer.mozilla.org/en-US/docs/Web/API/TextTrack/mode
[2] https://html.spec.whatwg.org/multipage/media.html#honor-user-preferences-for-automatic-text-track-selection