Re: [rtcweb] Another consideration about signaling

[Randell Jesup <[e-mail]>]

On 10/10/2011 5:45 PM, Ted Hardie wrote:
> On Mon, Oct 10, 2011 at 1:34 PM, Iñaki Baz Castillo <[e-mail]
> <mailto:[e-mail]>> wrote:
>
>      > If we agreed on a semantic approach (e.g. solicit/propose or
>     offer/answer),
>      > then the syntactic difference would be, I admit, largely a matter
>     of taste.
>
>
>     Hi Ted. We should come to the reality:
>
>     This is about establishing media sessions, and this is based on SDP,
>     and SDP works as follows:
>
>     - Alice sends, via some signaling protocol, a SDP offer to Bob.
>     - Bob prompts the human user and replies a SDP anwser (if it accepts
>     the call).
>     - After that media session(s) starts.
>     - At some point, Alice or Both could send a new SDP offer to modify
>     the session(s) (for example, for adding video, putting on hold or
>     whatever).
>
>     And that's all. SIP and XMPP/Jingle do that at the end. Both have
>     different semantics but similar target (exchange SDP information).
>
>
> If you believe that each RTCWeb Javascript/server pair needs to
> implement an SDP-based offer/answer protocol, but that it may choose
> among different syntaxes for carrying the SDP, then I think you're a lot
> closer to what I would call "having standard signaling" than not.

Yes - this was not what I had assumed Iñaki was arguing.

I made a similar proposal some time ago (and so did others - Harald, 
etc) for standardizing O-A in some manner, and I suggested providing an 
*optional* simple signalling package that developers could use to 
quickly get up and running.  It would not be a replacement for a 
fully-featured stack, but it might be a small subset of an existing full 
stack.  I also think I proposed a way to handle the O-A portion 
(modification of Harald's semantics) that would help resolve simple 
forking.  I'll dig the reference out, but it also helped solve the 
start-of-call-clipping problem.


-- 
Randell Jesup
[e-mail]