Simplify RealtimeOutgoingAudioSource and RealtimeOutgoingVideooSource observeSource/unobserveSource pattern. Currently, this is handled when adding/removing sinks and also when stopping libwebrtc backends. We hit RealtimeOutgoingVideoSource::~RealtimeOutgoingVideoSource() assertions as shown in tests.
Created attachment 384289 [details] Patch
<rdar://problem/57458345>
Created attachment 384297 [details] Patch
Created attachment 384298 [details] Patch
Ping review
Comment on attachment 384298 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=384298&action=review > Source/WebCore/ChangeLog:9 > + We made LibWebRTCRTCRtpSenderBackend ending the observing in its destructor. Nit: s/ending the observing/stop observing/ > Source/WebCore/platform/mediastream/RealtimeOutgoingAudioSource.cpp:49 > +#ifndef NDEBUG "#if !ASSERT_DISABLED" might be better here. > Source/WebCore/platform/mediastream/RealtimeOutgoingVideoSource.cpp:60 > +#ifndef NDEBUG Ditto.
Created attachment 384622 [details] Patch
Comment on attachment 384622 [details] Patch Clearing flags on attachment: 384622 Committed r252987: <https://trac.webkit.org/changeset/252987>
All reviewed patches have been landed. Closing bug.