Hi, I guess we should also look at the text in 3261 regarding this as part of the fix-the-bugs-in-3261 work that was initiated in Prague.
Regards, Christer > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf > Of Christer Holmberg (JO/LMF) > Sent: 12. huhtikuuta 2007 9:58 > To: Paul Kyzivat > Cc: [EMAIL PROTECTED] > Subject: Re: [Sip-implementors] VS: SDP offer/answer for delay media > > > Hi Paul, > > I am definately not saying that people should put their SDPs > wherever they want. > > I am saying that people should look at this issue from a o/a > state machine perspective, and not only focus on whether > there is an SDP body in a message or not. > > For example, you say that if the offer was sent in a reliable > 18x, you find no rationale to include a copy of it again in > 200. I am not sure I think it's that clear, but that we could > argue about forever. The important thing is that IF a SDP has > already been sent reliable in 18x, it should not matter > what/if you include in 200. > > Regards, > > Christer > > > > -----Original Message----- > > From: Paul Kyzivat [mailto:[EMAIL PROTECTED] > > Sent: 11. huhtikuuta 2007 22:57 > > To: Christer Holmberg (JO/LMF) > > Cc: Bu, Wenfei (Leo); [EMAIL PROTECTED] > > Subject: Re: VS: [Sip-implementors] SDP offer/answer for delay media > > > > Christer, > > > > Your comments indicate how including SDP in extra places would not > > constitute offers or answers and so how it ought not cause problems. > > > > But regardless of that, SDP in those places does not seem to be > > allowed by existing standards. There is already enough problem with > > redundant SDP that was grandfathered in by the wording of > the specs, > > and it has proven to cause trouble. So I would not want to > explicitly > > allow this in other places. > > > > If you want to put SDP in these places, in a non-conformant > way, with > > the expectation that others will not be bothered by it, > they you are > > free to try that. But don't complain if somebody handles it > in a way > > different from what you expect. > > > > Paul > > > > Christer Holmberg (JO/LMF) wrote: > > > Hi, > > > > > >>> All, > > >>> > > >>> This was talked before but it still looks not clear: > > >>> > > >>> UAC UAS > > >>> > > >>> ----------->F1 INVITE w/o SDP > > >>> > > >>> <-----------F2 18X w/ SDP1 > > >>> > > >>> ----------->F3 PRACK w/ SDP2 > > >>> > > >>> <-----------F4 200 for PRACK > > >>> > > >>> <-----------F5 200 for INVITE w/ SDP1 > > >>> > > >>> ----------->F6 ACK (w/ SDP2 ???) > > >>> > > >>> > > >>> > > >>> My questions: > > >>> > > >>> 1. Could 200 for INVITE in F5 carry an SDP? > > > > > >> No. There has already been one o/a in the transaction > initiated F1. > > >> There cannot be another offer in the same transaction. > > > As far as I understand, the 200 for INVITE does not carry a > > new offer. It carries the same answer that was carried in 18x (F2). > > > > > > > > >> In the case where the invite carries an offer sdp, and > > there is SDP > > >> in a provisional response that same sdp can be in the 200. That > > >> wording in > > >> 3261 that specifies this is intended for unreliable > > provisionals, but > > >> has been interpreted to also be possible when there was > a reliable > > >> response with an answer. But I can find no rationale for > > allowing an > > >> offer that was sent in a reliable provisional to be sent > > again in the > > >> 200 response to the invite. > > > > > > Since SIP only allows at most one o/a exchange per SIP > > transaction, I think the o/a state machine should know that > the SDP in > > 200 is of no meaning - no matter if it's a copy of a > previously sent > > (in a reliable 18x) offer or anwer. > > > > > > It's the same thing if you send an offer, or answer, in a > > reliable 18x, and then copy it in subsequent reliable 18x responses > > for the same transaction - from an o/a perspective they have no > > meaning. > > > > > > > > >> Also I can find no rationale for allowing a *new* offer to be > > >> included in the 200 response to an invite if the initial > > o/a has been completed. > > > > > > Correct. Again, at most one o/a exchange per SIP > > transaction. Keeping > > > that in mind will take you pretty far, I think :) > > > > > > > > >>> 2. If 200 in F5 carries the same SDP with that in 18X, > should UAC > > >>> send an ACK with SDP? > > >>> > > >> Definitely not. > > > > > > Agree. > > > > > > Regards, > > > > > > Christer > > > > > > _______________________________________________ > > > Sip-implementors mailing list > > > [EMAIL PROTECTED] > > > https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors > > > > > > > _______________________________________________ > Sip-implementors mailing list > [EMAIL PROTECTED] > https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors > _______________________________________________ Sip-implementors mailing list [EMAIL PROTECTED] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
