Iñaki Baz Castillo wrote:
> El Thursday 08 May 2008 16:53:32 Paul Kyzivat escribió:
>
>> The offeranswer draft explains. THere can be only one answer per offer
>> per dialog. Sometimes the same answer is send in multiple messages, but
>> technically only one of them is the actual answer.
>
> But note that UAC just receives one final response (the 200 OK in case phoneB
> answers).
> In the diagram
I can't understand the diagram. The alignment is all messed up for me.
But I have seen similar pictures many times, so I think I know the gist
of what is happening.
> we can see that AS transforms the first 200 OK from
> Announcementserver into a 183 befire sending it to UAC.
> So in conclusion:
>
> - UAC sends INVITE to AS (offer-1)
> - AS forwards it to AnnouncementServer.
> - AnnouncementServer replies with 200 (answer-1, To_tag-1)
> - AS converts it into 183 (maybe acting as a B2BUA more than a proxy) and
> sends it to UAC.
So far all is fine.
> - UAC receives 183 (To_tag-1), there is no final answer yet.
Note that 3264 has no notion of an "early answer" and a "final answer" -
only an answer. 3261 has a notion of "early dialog" but that is not the
same thing.
Assuming no reliable provisionals being used...
The *answer* will only be in a reliable response, which will be the 200
to User(A). However User(A) may treat SDP received in a provisional
response *as if* it were the offer. However the SDP in the answer is
required to be the same as this. In the offeranswer draft we started
calling this a "preview" of the offer.
> - During early-dialog UAC does somehting (DTMF and so) and AS calls to
> phone-B
> who rings.
> - AS receives the 180 of phone-B (To_tag-2) and forwards to UAC.
> - UAC receives 180 (To_tag-2).
As long as this is a different to-tag, then it is a different dialog,
and so it is the first answer in this dialog, and so is ok.
> - Phone-B answers.
> - UAC receives 200 OK (To_tag-2).
As long as the 200 contains the same sdp as was in the prior 180 this is ok.
> Why is it not valid?
It is valid as long as different to-tags are used.
But often people try to do this exact same scenario, but send the
responses, with changed SDP, using the *same* to-tag. That is wrong.
This is all explained in the o/a draft. It exists because questions like
this come up so often.
Thanks,
Paul
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
