Bug 225511 – [GStreamer][MediaStream] Emit black frames for disabled video tracks

• Status: RESOLVED
• Resolution: FIXED
• Component: Platform

[Philippe Normand]

imported/w3c/web-platform-tests/mediacapture-streams/MediaStreamTrack-MediaElement-disabled-video-is-black.https.html fails.

[Philippe Normand]

Created attachment 427995 [details]
Patch

[Xabier Rodríguez Calvar]

Comment on attachment 427995 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=427995&action=review

> Source/WebCore/platform/mediastream/gstreamer/GStreamerMediaStreamSource.cpp:276
> +        pushSample(blackSample.get());

Nit (maybe for other time): wouldn't it be nice pushSample was void pushSample(GRefPtr<GstSample>&&) ?

[Philippe Normand]

Comment on attachment 427995 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=427995&action=review

>> Source/WebCore/platform/mediastream/gstreamer/GStreamerMediaStreamSource.cpp:276
>> +        pushSample(blackSample.get());
> 
> Nit (maybe for other time): wouldn't it be nice pushSample was void pushSample(GRefPtr<GstSample>&&) ?

I don't think that's possible currently because when the track is enabled the sample being pushed is not owned by the observer, but by the track that generates it. And a track can have more than one observer.

[Philippe Normand]

Committed r277175 (237461@main): <https://commits.webkit.org/237461@main>

[Radar WebKit Bug Importer]

<rdar://problem/77658069>

[Xabier Rodríguez Calvar]

(In reply to Philippe Normand from comment #3)
> Comment on attachment 427995 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=427995&action=review
> 
> >> Source/WebCore/platform/mediastream/gstreamer/GStreamerMediaStreamSource.cpp:276
> >> +        pushSample(blackSample.get());
> > 
> > Nit (maybe for other time): wouldn't it be nice pushSample was void pushSample(GRefPtr<GstSample>&&) ?
> 
> I don't think that's possible currently because when the track is enabled
> the sample being pushed is not owned by the observer, but by the track that
> generates it. And a track can have more than one observer.

It doesn't mean you can't create another GRefPtr<GstSample> by reffing the sampled owned by the track, right?