Re: [sipcore] Question on draft-camarillo-sipcore-reinvite-01.txt

[Gonzalo Camarillo <[e-mail]>]

Hi,

I have put together a new version of the draft:

http://users.piuha.net/gonzalo/temp/draft-ietf-sipcore-reinvite-pre01a.txt

Let me know if you have some comments before I submit it.

Regarding the decision about when a given change in the session state 
has been executed when preconditions are used, I have added a paragraph 
that summarizes our discussions and the following draft:

http://www.watersprings.org/pub/id/draft-camarillo-sipping-precons-00.txt

Please, read Section 5.

Sections 7 and 15 contain new rules to avoid glare situations and race 
conditions. These rules avoid nasty call flows that were able to get the 
UAs out of synch.

Some time ago, we discussed whether or not unreliable provisional 
responses could refresh the remote target. There are several problems 
with unreliable responses updating remote targets. First, we would need 
to define even more rules to avoid glare situations and race conditions. 
Second, SIP does not provide any ordering for unreliable provisional 
responses. Responses arriving out of order at the UAC could easily get 
both UAs out of synch. Therefore, only reliable responses can refresh 
remote targets, as was specified in RFC 3261 anyway.

We also discussed the applicability of the target refresh rules to 
initial INVITEs. The rules in the draft cover them as well so I do not 
think we need to add anything else to that end.

Thanks,

Gonzalo