Re: [rtcweb] Fwd: New Version Notification for draft-uberti-rtcweb-plan-00.txt

[Paul Kyzivat <[e-mail]>]

On 5/8/13 6:01 PM, Ted Hardie wrote:
> On Wed, May 8, 2013 at 12:43 PM, Paul Kyzivat <
>
>     *I* think it is important that clue and rtcweb to be able to coexist
>     in a single endpoint, and for a clue+rtcweb client to interoperate
>     with either another clue+rtcweb client or a pure clue client. Using
>     different indicators in the two environments doesn't help with that.
>
>
> So, can you unpack "coexist"?  A single endpoint could certainly have
> either a clue client or an rtcweb client active at a single moment in
> time--do you want to say that you want them both to be active and share
> the same cameras, microphones, etc?

What I mean is that I want the capability to *implement* a clue client 
in a browser, with the media going directly to the peer clue client (or 
MCU), whether the peer is also a browser or is a native sip clue 
implementation.

Since clue isn't done yet, we can impose some constraints on it to help 
with that. Within reason.

Clue will have its own control protocol, which we intend to define to 
run over DTLS/SRTP. It should be possible for a browser client to use an 
RTCWEB data channel for that.

And of course clue has needs for many media "flows" (captures in 
clue-speak), so those need to map directly onto MediaStreamTracks, or 
something like that in the webrtc world.

Whatever signaling is used to indicate bundling usage in SDP in RTCWEB 
should be consistent with what is used in clue. It would be unpleasant 
if we had to have a translator in the web server that must translate one 
SDP notation to another.

	Thanks,
	Paul