Re: [rtcweb] A problem with both A and B

[Paul Kyzivat <[e-mail]>]

On 5/20/13 4:53 AM, Kevin Dempsey wrote:
> I would say that at present only the sending application knows what a
> stream is for. The receiving side needs to determine the purpose by
> matching the RTP packets to the signalling the applications have done.

ISTM that the legacy cases have depended upon doing this in a simplistic 
way. E.g., the app only supports one audio media stream to be played out 
over one audio device. So it refuses anything other than a single audio 
m-line and then maps all the audio packets it receives, regardless of 
SSRC, to that one device. So it needs minimal signaling.

This has gradually evolved to slightly more complex cases, such as 
audio+video using similar logic. The a=content attribute was added to 
cover simple cases of multiple media of the same type, but still 
assuming a separate m-line for each device.

All of this assumes that there is a very small range of fundamentally 
different applications, so signaling is not used much to distinguish them.

But clearly this is beginning to break down. ISTM the current plan is to 
require that the two ends of a session be running the same logical 
application, and that they then define how to sort out the use of 
multiple media in an application-specific way. This is what both CLUE 
and RTCWEB seem to be doing. In the CLUE case the assumption is that 
initial SIP session setup verifies if the assumption is true, and then 
defines different actions if it is or isn't. In the case of RTCWEB, I 
guess the assumption is that it is probably a single web server 
application controlling both ends. But in cases where that is not true - 
gatewaying to the sip world - I guess this is left for the web server to 
figure out.

I don't know how this is going to work when a Google+ with no Facebook 
account wants to interact with a Facebook user that has no Google+ 
account. ISTM in that case you need separate but equivalent web 
applications on both providers, and somehow they still need to have a 
consistent signaling mechanism between them.

	Thanks,
	Paul