Thanks for the response.

> I once proposed using a "19x" provisional response to 
> indicate that a early dialog has been terminated, but I 
> havent done anything onnit since then.

Figure 10 of the following draft presents using a 155 with Reason header
indicating 487 (or similar cause).

http://www.jdrosen.net/papers/draft-rosenberg-sip-unify-00.txt


Draft-jbemmel-herfp-solution-00 and draft-mahy-sipping-herfp-fix-00 mention
the use of 130.  Within these drafts, 130 currently appears to only be used
for "repairable" responses.  If this remains, another response would be
needed for non reparable releases of individual early dialogs.


> I guess the only current way is to send a BYE in the backward 
> direction, but I have a feeling many clients would either 
> reject it or release the whole session setup.

That was my assumption also.  Additionally it would require a proxy to morph
into a B2BUA for cseq handling and late answers.

 
> The problem, no matter which of the above mechanisms is used, 
> is: what does a UAC do if the only so far established early 
> dialog is terminated? There could me responses coming in 
> order to create other early dialogs...

I agree that race conditions, dropped packets, and delays before next 18x
could cause problems.  Hopefully these type of issues will be addressed
during the herfp effort.

Until then, is there a recommended solution?  I assume some vendors treat
the latest 18x as the most likely still active.  And I assume some vendors
get creative with SDP usage, UPDATE, NOTIFY (dialog package), etcetera.  Is
anyone aware of a draft or RFC presenting the best common practice until a
1xx solution becomes adopted?



_______________________________________________
Sip-implementors mailing list
[email protected]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to