Re: [rtcweb] No Plan

[Stefan Håkansson LK <[e-mail]>]

On 2013-05-30 11:19, Bernard Aboba wrote:
> Comments below.
>
> On May 30, 2013, at 1:24, "Stefan Håkansson LK"
> <[e-mail]> wrote:
>
>> Hi Emil,
>>
>> a couple of comments:
>>
>> "1.  Expose the SSRCs of all local MediaStreamTrack-s that the
>> application may want to attach to a PeerConnection."
>>
>> I don't think that would be possible, or desirable. SSRC _only_
>> come into play when a MediaStreamTrack is to be transported over
>> the network - it is useless in local cases. And, the same local
>> MediaStreamTrack could be sent to several peers (using different
>> PeerConnections) and could have different SSRC's (and even
>> different number of SSRC's) for the different peers.
>>
>> I think SSRCs would only be available for MediaStreamTracks that
>> are attached to a PeerConnection.
>
> [BA] If ICE fails, it may be necessary to use alternative transport,
> outside of PeerConnection. So if SSRCs are to be used in signaling
> for that case, the API needs to make them available.

Definitely.

But my point was that unless there is an intention to transport, SSRCs 
are useless (and they only have a meaning if there is RTP involved, 
right?). There are a lot of local use cases for MediaStream's 
(containing MediaStreamTrack's), so it does not make sense to add SSRC 
APIs to those objects; it belongs to PeerConnection IMO.

But when (if) another transport object in JS space than PeerConnection 
is defined, it would also need an API where SSRCs could be accessed.

Stefan

>
>>