Re: [SIP] Deleting Media Streams section in rfc2543bis

[Henning Schulzrinne <[e-mail]>]

Igor Slepchin wrote:
> 
> > Brett Tate wrote:
> > It
> > would be worth mentioning in this section that a media
> > stream can also be deleted by the absence of a "m=..."
> > line which was in the previous SDP for the call leg.
> 
> This is wrong: it cannot. The m= lines are never removed, otherwise you
> wouldn't be able to match m lines in response and request. Consider a
> situation where the request contains
> 
> c=IN IP4 1.2.3.4
> m=audio 54678 RTP/AVP 0 1 3
> m=video 7346 RTP/AVP 10 13 (face)
> m=video 7880 RTP/AVP 10 11 (presentation)
> 
> If the response contained something like
> m=audio 6540 RTP/AVP 0 1
> m=video 6578 RTP/AVP 10
> 
> there would be no way to tell which of the two offered media streams was
> accepted and which one rejected.

While I don't disagree, practically speaking this
multiple-instance-of-one-type situation is fairly uncommon. We probably
still need to define what happens if you get a new SDP (it replaces the
old one). Realistically, we can't expect that

m=audio 4321 ...
m=video 5678 ...

will be answered by

m=audio 8888 ...
m=video 0

I suspect most implementations simply reply 

m=audio 8888

if they can't do video.


I think there was some discussion at some point of labeling streams in
SDP, which would simplify this.
(http://www.ietf.org/internet-drafts/draft-camarillo-sip-sdp-00.txt
seems to solve the opposite problem of "splitting" a single logical
stream across ports, but maybe the two concepts can be married.)


> 
> > It might also be worth mentioning somewhere in
> > Appendix B that receiving a completely new SDP within
> > a call-leg should render the previous SDP obsolete.
> 
> See above. There is currently no way to completely replace SDP, you can
> only update it either by rejecting some media sessions (by setting their
> port # to 0) or by adding the new ones at the end of the list.
> 
> Thank you,
> Igor Slepchin
> 
> _______________________________________________
> SIP mailing list
> [e-mail]
> http://lists.bell-labs.com/mailman/listinfo/sip

-- 
Henning Schulzrinne   http://www.cs.columbia.edu/~hgs


_______________________________________________
SIP mailing list
[e-mail]
http://lists.bell-labs.com/mailman/listinfo/sip