2008/9/15, krishna kalluri <[EMAIL PROTECTED]>: > > I think that RFC 3665 shows a correct sequential SIP flow. Anyway it > > could use different Call-ID/CSeq (but the flow is valid). > > [Krishna] If the RFC assumes correct sequence flow then Section 2.4 should > have CSeq = 2.
I see no season. Note that this is *not* a dialog (REGISTER doesn't create a dialog), Call-ID and CSeq math in regitstration process for other purposes. I think is 100% correct section 2.4 to have CSeq=1. > > > 2) 2.4 Cancellation of Registration (RFC 3665): In this case CSeq number > > > MUST be greater than the one used in previous messages. This is from my > > > interpretation from Step 6 of Section 10.3 of RFC 3261. > > > > But that is needed just if "Contact" matches a binding, but note that > > the "Contact" in 2.4 is: > > Contact: * > > > > [Krishna] Step 6 of Section 10.3 of RFC 3261 is describing about Contact = > * scenario. You are right. Since Call-ID matches, CSeq must be greater in this cancellation REGISTER with "Contact: *". So I think that step in 2.4 should be discarded by the server because CSeq is not greater than the CSeq foe each stored binding with same Call-ID. Anyway note that maybe RFC 3665 doesn't paint a real flow. Regards. -- Iñaki Baz Castillo <[EMAIL PROTECTED]> _______________________________________________ Sip-implementors mailing list [email protected] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
