Bug 176281 – track.onmute isn't called for a remote MediaStreamTrack when its counter part track is removed from the peer connection

• Component: WebRTC
• Status: RESOLVED FIXED

[Ben]

Created attachment 319733 [details]
Test

There is no way to detect when a remote peer removes a track in a WebRTC call.

I've tried track.onended, track.onmute, stream.onremovetrack:
pc.ontrack = function(e) {
  var track = e.track;
  var stream = e.streams[0];
  // try to detect when a remote track is removed
  track.onended = function() {console.log('onended')};
  track.onmute = function() {console.log('onmute')};
  stream.onremovetrack = function() {console.log('onremovetrack')};
}

In the attached track.html example you should see "video ended" and "audio ended" in Chrome and Firefox but not in Safari 11.1 Release 38.

[Clement Wehrung]

Hitting the same issue. There doesn't seem to be any good workaround. Ideas?

[Radar WebKit Bug Importer]

<rdar://problem/44525674>

[youenn fablet]

Hi Ben, Clement, thanks for the report and test case.
As per spec, the track should be muted so an unmute event should probably happen.

[youenn fablet]

Created attachment 349915 [details]
Patch

[Clement Wehrung]

Happy to test further tomorrow but I did set onmute to the same callback in doubt and it didn't work. Only testing for remote stream. Thanks for looking into it!

[Clement Wehrung]

The proposed workaround from the webrtc-discuss thread works but it's less than ideal: check setRemoteDescription and cache previous media IDs to compare later if one has disappeared

[youenn fablet]

(In reply to Clement Wehrung from comment #5)
> Happy to test further tomorrow but I did set onmute to the same callback in
> doubt and it didn't work. Only testing for remote stream. Thanks for looking
> into it!

Right, WebKit is not currently doing so and the posted patch should fix it.

[Eric Carlson]

Comment on attachment 349915 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=349915&action=review

> Source/WebCore/Modules/mediastream/libwebrtc/LibWebRTCPeerConnectionBackend.cpp:489
> +        auto* backend = static_cast<LibWebRTCRtpSenderBackend*>(transceiver->sender().backend());

Nit: downcast<LibWebRTCRtpSenderBackend> would be better, although it is probably a big enough change that you may want to do it in a separate patch.

[youenn fablet]

(In reply to Eric Carlson from comment #8)
> Comment on attachment 349915 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=349915&action=review
> 
> > Source/WebCore/Modules/mediastream/libwebrtc/LibWebRTCPeerConnectionBackend.cpp:489
> > +        auto* backend = static_cast<LibWebRTCRtpSenderBackend*>(transceiver->sender().backend());
> 
> Nit: downcast<LibWebRTCRtpSenderBackend> would be better, although it is
> probably a big enough change that you may want to do it in a separate patch.

Right, let's do that in a follow-up patch

[WebKit Commit Bot]

Comment on attachment 349915 [details]
Patch

Clearing flags on attachment: 349915

Committed r236090: <https://trac.webkit.org/changeset/236090>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.