Christer,
I hope that is not what the authors are proposing, because UPDATE
is certainly useful outside the context of INVITE/re-INVITE transactions, e.g.,
for session timer refreshes.
John
From: sipping-bounces@xxxxxxxx [mailto:sipping-bounces@xxxxxxxx] On Behalf Of Christer Holmberg
Sent: 17 February 2009 14:00
To: Elwell, John; wang.libo@xxxxxxxxxx
Cc: sipping@xxxxxxxx
Subject: Re: About "rollback of re-Invite"Hi John,My understanding of the proposal is that you would only send UPDATEs if they ARE nested to the re-INVITE transaction. Otherwise you would send a new re-INVITE. That way you would know that an UPDATE is nested even if it arrives after the ACK.That of course means that the ongoing re-INVITE transactions needs to finish first (either by itself, or by being cancelled).But, I may have missunderstood, so it's better if the authors speak for themselves :)Regards,Christer
From: Elwell, John [mailto:john.elwell@xxxxxxxxxxx]
Sent: 17. helmikuuta 2009 15:51
To: Christer Holmberg; wang.libo@xxxxxxxxxx
Cc: sipping@xxxxxxxx
Subject: RE: About "rollback of re-Invite"I haven't been following all of this discussion, but there is nothing explicitly to bind an UPDATE transaction to a parent INVITE or re-INVITE transaction. It is implicit in the vast majority of cases, but timing issues break this, particularly where unreliable transport is concerned. For example, if UPDATE arrives while awaiting ACK, is it because UPDATE was sent before the ACK, or because the ACK was lost? If UPDATE arrives before sending a final response to re-INVITE, is it because UPDATE is nested within the re-INVITE transaction, or has there been a CANCEL request that has been lost and will be repeated later? Without explicit binding, the only consistent thing to do is to treat each transaction separately, and not roll back. If this means pre-conditions are broken, the focus should be on fixing pre-conditions somehow.John
From: sipping-bounces@xxxxxxxx [mailto:sipping-bounces@xxxxxxxx] On Behalf Of Christer Holmberg
Sent: 16 February 2009 18:00
To: wang.libo@xxxxxxxxxx
Cc: sipping@xxxxxxxx
Subject: Re: About "rollback of re-Invite"Hi,
>The key point of this discussion is how to find out the proper state of the rollback.
>I think ,Some rules of using re-invite may be helpful.Such as, we restrict only
>one offer/answer pair in a re-invite,I don't think we can do that, because at least some pre-condition scenarios require more offer/answer pairs.And, even if you would only allow e.g. a single UPDATE within the re-INVITE, you would still have the same question: if the UPDATE suceeds, but the re-INVITE then fails, what is the state?Regards,Christer
"Christer Holmberg" <christer.holmberg@xxxxxxxxxxxx> 写于 2009-02-16 21:41:59:
>
> Hi,
>
>
> >>I also pay high attention to the topic of "rollback of re-Invite".
> >>As lots of service is implemented on the network using SIP, and
> >>different ways may be carried out by different corporations, it's
> necessary to define
> >>a complete and clear method to solve the failure when the unsuccessful
> re-Invite
> >>happens.
> >
> >I agree, and we thought that having a generic rule, not related to
> >specific use-cases, would be the most clear, clean and complete
> >solution.
> >
> >[gaoyang] It is simple and clear. But it disregard nature property of
> nested transaction. I concluded incorrectness and drawback of such way.
> More details in
> http://www.ietf.org/internet-drafts/draft-gaoyang-sipping-session-state-
> analysis-01.txt.
>
> I agree it's not the most "beautiful" solution - that's why we spent a
> long time discussing it.
>
> >>And the problems "Commit any session parameters that have been
> sucessfully changed"
> >>metioned in "draft-gaoyang-sipping-session-state-analysis-01.txt" need
> further discuss,
> >>I think.
> >
> >The draft does not define a generic rule. It more or less says that
> >every time there is a new use-case, the offer/answer rollback aspects
> >needs to be documented for that use-case. I think that is something we
> >wanted to avoid.
> >
> >[gaoyang] It is not use-case oriented, but rules/definition oriented.
> >And the definition of how o/a pairs form nested transaction should be
> open for future.
> >Such as "a=chain" and so on extension can form more o/a pairs as nested
> transaction. It is not use-case.
> >We can have a further talk.
> >
> >If I understand chapter 4.3.3 correct, it says that if you have a
> >pending re-INVITE transaction, a NEW modification must be sent in a
> >re-INVITE request. That way we would get rid of the race condition
> >problem. That is an interesting point. I guess the question is whether
> >it's too lake to make such a rule.
> >
> >[gaoyang] Is the "lake" late? I think it can be BCP, not rules :).
>
> Perhaps, yes.
>
> Regards,
>
> Christer
>
>-------------------------------------------------------- ZTE Information Security Notice: The information contained in this mail is solely property of the sender's organization. This mail communication is confidential. Recipients named above are obligated to maintain secrecy and are not permitted to disclose the contents of this communication to others. This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the originator of the message. Any views expressed in this message are those of the individual sender. This message has been scanned for viruses and Spam by ZTE Anti-Spam system.
_______________________________________________ Sipping mailing list https://www.ietf.org/mailman/listinfo/sipping This list is for NEW development of the application of SIP Use sip-implementors@xxxxxxxxxxxxxxx for questions on current sip Use sip@xxxxxxxx for new developments of core SIP
