Re: Update of RTCRtpSender "doohickey" proposal

On 05/02/2014 08:36 AM, Kiran Kumar Guduru wrote:
> Samsung Enterprise Portal mySingle
>
> Perhaps this will avoid my confustion which I posted some time back [1].
>
> // parameter to the onaddtrack event
>
> interface RemoteTrackEvent : Event {
>
>  readonly attribute RtpReceiver receiver;
>
>  readonly attribute MediaStreamTrack track;
>
>  readonly attribute MediaStream stream;
>
> };
>
>
> Is it required to have both stream and track?
>
> AFAIK stream contains track inside it.
>
> Is there any other requirement to have both track and stream 
> independently?
>

A stream can have multiple tracks. Without track, you don't know which 
track the event is on.
A track can be in multiple streams. Without stream, you don't know which 
stream/track association the event concerns.


> [1] http://lists.w3.org/Archives/Public/public-webrtc/2014Apr/0087.html
>
> ------- *Original Message* -------
>
> *Sender* : Stefan Håkansson LK<stefan.lk.hakansson@ericsson.com>
>
> *Date* : May 02, 2014 15:12 (GMT+09:00)
>
> *Title* : Re: Update of RTCRtpSender "doohickey" proposal
>
> On 29/04/14 18:50, Martin Thomson wrote:
> > On 28 April 2014 23:11, Justin Uberti wrote:
> >> The issue could still exist if you do addTrack(T, S) and then later
> >> S.removeStream(T). We might prefer taking the stream directly for
> >> syntactical reasons, but I don't think it avoids the core issue.
> >
> > That's fine; the stream configuration on the receiving end always lags
> > the configuration at the sending end.  I'm just not convinced that we
> > need a way to specify identifiers in this way, Harald's speculation on
> > rehydration notwithstanding.
>
> In principle we can have the situation that a track added to the PC is
> member of several MediaStreams on the sending side. Should the "stream"
> argument be a list (regardless of it is a list of MS objects or MS
> identifiers)?
>
> >
> >
>
>