The changes in bug 1324883 regressed YouTube, so back them out.
The changes in bug 1324883 were trying to cause the media cache to be cleared
on tab close and on CTRL+F5 reload (i.e. a bypass cache-reload) but they are
causing problems on YouTube, which doesn't use the media cache and instead
uses MSE.
MozReview-Commit-ID: Hx2cehZ2wm1
Mostly-mechanical additions:
- Log constructions&destructions, usually by just inheriting from
DecoderDoctorLifeLogger, otherwise with explicit log commands (for internal
classes for which DecoderDoctorTraits can't be specialized),
- Log links between most objects, e.g.: Media element -> decoder -> state
machine -> reader -> demuxer -> resource, etc.
And logging some important properties and events (JS events, duration change,
frames being decoded, etc.)
More will be added later on, from just converting MOZ_LOGs, and as needed.
MozReview-Commit-ID: KgNhHSz35t0
The process of |TryRemoveMediaKeysAssociation()| is a 2-step async procedue in mainthread.
mMediaKeys might be set to null inside |NotifyOwnerDocumentActivityChanged()| in between
|TryRemoveMediaKeysAssociation| and |RemoveMediaKeys|.
MozReview-Commit-ID: HtiADt3UTvp
This patch is mainly reverting the changing of bug1382574 part3, but not all the same.
Since youtube would call load() when user clicks to play, and then call play()
later. For the old pref (checking user-input-play), we should still allow the
following play() even it's not triggered via user input. It's also same for
seeking, Youtube would call play() after seeking completed.
In this patch, we would allow the script-calling once play() if user has called load()
or seek() before that.
MozReview-Commit-ID: 1UcxRCVfhnR
This is necessary in order to parse style attributes using the subject
principal of the caller, rather than defaulting to the page principal.
MozReview-Commit-ID: GIshajQ28la
AutoplayPolicy is used to manage autoplay logic for all kinds of media,
including MediaElement, Web Audio and Web Speech.
MozReview-Commit-ID: R1TxMkarIw
There are currently two types of sinks for a MediaStreamTrackSource.
Actual MediaStreamTracks and HTMLMediaElement::StreamCaptureTrackSource.
A source needs actual tracks as sinks to not stop() the underlying source.
A StreamCaptureTrackSource, however, should not count toward keeping a source
alive (otherwise HTMLMediaElement.mozCaptureStream() would prevent track.stop()
from working on the track feeding the media element).
MozReview-Commit-ID: 9MBAyZFZUIQ
This is a drive-by fix in that it is not directly related to what the bug is
solving. However, making HTMLMediaElement register as a sink is important,
and pairing it with the WeakPtr<Sink> patch reduces risk greatly.
MozReview-Commit-ID: 7pMDw3MG0ZB
This patch implements HTMLMediaElement::GetEventTargetParent and set
aVisitor.mCanHandle to false to mouse/touch/pointer events, when
the media control is present. This tells the event dispatcher that
these events are supposed to be handled exclusively by the
videocontrol binding within the media element, and should not
dispatch nor consumed by the content.
MozReview-Commit-ID: BXWZX9SYsuC
This patch implements HTMLMediaElement::GetEventTargetParent and set
aVisitor.mCanHandle to false to mouse/touch/pointer events, when
the media control is present. This tells the event dispatcher that
these events are supposed to be handled exclusively by the
videocontrol binding within the media element, and should not
dispatch nor consumed by the content.
MozReview-Commit-ID: BXWZX9SYsuC
The MediaKeys status inside a HTMLME cannot be reflected correctly if the mSetCDMRequest is disconnected in HTMLME::ShutdownDecoder.
This may happen when a page calls load() or sets new src right after setting MediaKeys to null.
MozReview-Commit-ID: 3BZOmw7BNFO
1. move all checks to CanActivateAutoplay()
2. don't cache the pref's value in advance, it might cause wrong result
if user changes pref after media was binded to tree.
MozReview-Commit-ID: 3BeOeaq9wGa
