Re: [rtcweb] AVPF vs AVP

Hi,

I agree with Hadriel and Harald.

Also, even if there is no SBC, a UAS that doesn't support MULTIPLEX would see an offer of two different streams, and if it accepts both a second o/a would be needed anyway.

And, as Harald said, the backward compability is for entities that do not support multiplexing.

Regards,

Christer

> -----Original Message-----
> From: rtcweb-bounces@ietf.org 
> [mailto:rtcweb-bounces@ietf.org] On Behalf Of Hadriel Kaplan
> Sent: 16. syyskuuta 2011 21:06
> To: Roni Even
> Cc: <rtcweb@ietf.org>
> Subject: Re: [rtcweb] AVPF vs AVP
> 
> 
> On Sep 16, 2011, at 9:03 AM, Roni Even wrote:
> 
> > Magnus,
> > Maybe we can take the approach from
> > draft-holmberg-mmusic-sdp-multiplex-negotiation and offer the same 
> > m-line with the same port number twice, one as AVP and one 
> as AVPF and 
> > let the other side chose, he can answer with a port=0 for 
> the profile 
> > it does not want. If the multiplex proposal has backward 
> > interoperability than this proposal also does.
> 
> You can't do that.  Not only is it not kosher in the sense 
> that you're really offering two media sessions, rather than 
> alternatives for one, but it actually won't work in practice: 
> if the SDP offer crosses an SBC or App-Server B2BUA, some of 
> them will treat them as two media sessions and thus offer 2 
> different port numbers on their UAC side and make it no 
> longer distinguishable from a normal offer of two media sessions.
> (and I've tested this on an SBC)
> 
> -hadriel
> 
> _______________________________________________
> rtcweb mailing list
> rtcweb@ietf.org
> https://www.ietf.org/mailman/listinfo/rtcweb
> 