Re: [rtcweb] Convey "label"

[Stefan Håkansson LK <[e-mail]>]

On 2011-09-09 18:14, Justin Uberti wrote:
>
>
> On Fri, Sep 9, 2011 at 9:13 AM, Stefan Håkansson LK
> <[e-mail]
> <mailto:[e-mail]>> wrote:
>
>     On 2011-09-09 14:58, Justin Uberti wrote:
>
>
>
>         On Fri, Sep 9, 2011 at 3:55 AM, Olle E. Johansson
>         <[e-mail] <mailto:[e-mail]>
>         <mailto:[e-mail] <mailto:[e-mail]>>> wrote:
>
>
>             9 sep 2011 kl. 09:07 skrev Colin Perkins:
>
>          > On 9 Sep 2011, at 07:53, Olle E. Johansson wrote:
>          >> 9 sep 2011 kl. 08:51 skrev Stefan Håkansson LK:
>          >>> there was a discussion yesterday of how to convey the Id (aka
>         "label") of a MediaStream object when it is sent over a
>             PeerConnection. Presumable this info would be in the SDP
>         offer (or
>             whatever we end up using). There was a discussion on if
>         CNAME could
>             be used or if we need something new.
>          >>>
>          >>> You mentioned that it would be quite easy to add a new element
>             or attribute if we would need that. Could you explain a bit
>         more?
>          >>
>          >> Just a kind reminder that there are application specific fields
>             that we can use in RTCP too...
>          >
>          >
>          > I was actually thinking of a new RTCP SDES item type to convey
>             the label.
>          >
>             Side note: I know we're not talking SIP, but it actually touches
>             Hadriel's Session ID proposal too...
>
>
>         It's not clear to me that we need a MediaStream to have a label,
>         or any
>         identifier other than CNAME, unless it's meant to be some sort of
>         "display name" for that particular stream. If there are other
>         expected
>         usages of the label, I'd be interested to know what they are.
>
>     The main purpose of an identifier of MediaStream's I see is to be
>     able to reference them after transport over a PeerConnection. If
>     several MediaStreams are sent over a PC, you need to be able to
>     reference them - all the receiving app gets is "addstream" events.
>
>     Of course you could add them one by one in a determined order, but
>     that would mean that set up takes longer than required.
>
>     I'm not sure CNAME can fulfill this purpose in all cases or not.
>
>
> Right, I agree the app needs to be able to say "this video feed is for
> user A" and "this video feed is for user B", if for no other reason than
> to be able to display a user identifier underneath the appropriate video
> feed. However as mentioned previously I think that this implies that
> each feed (i.e. track) requires an identifier, not necessarily the
> stream. So I propose we add a unique label attribute for tracks, which
> is essentially a handle; this handle allows clear identification of a
> track, and can be passed around in application messaging if needed. We
> may also want to allow a non-unique display name to be provided as well.

This is an interesting idea. However, in the html5 spec media elements 
deal with 'media resources' which can have multiple tracks of audio and 
video. To be able to reuse what is already in place (e.g. audio and 
video elements) I think we should conform to this.

Besides, if you can identify the MediaStream, it is quite 
straightforward to reference a specific track in it as it is the same 
tracks on both sides of the PeerConnection.

>
> I see no harm in having MediaStream's label be the CNAME, as a portable
> handle for MediaStreams, but I haven't yet heard a use case where one
> would use this handle instead of the track handle.
>
>
>
>     _________________________________________________
>     rtcweb mailing list
>     [e-mail] <mailto:[e-mail]>
>     https://www.ietf.org/mailman/__listinfo/rtcweb
>     <https://www.ietf.org/mailman/listinfo/rtcweb>
>
>