Alex Balashov <abalas...@evaristesys.com> writes:
> Due to the way the RTP relay works on the server side, this results in
> two different SDP offers from the two respective outgoing branches being
> sent back to the caller:
>
> 1. 183+SDP on branch #1.
>
> 2. 183+SDP' on branch #2.
> 200 OK+SDP' on branch #2.
>
> I am not sure offhand whether this is a protocol semantics violation. On
> the one hand, RFC 3261 section 13.2.1 ("Creating the Initial INVITE") says:
As Paul says, the RFCs aren't so very clear about it, but each branch
has a separate to-tag and so is a different dialog, and each dialog has
a seperate state machine. So the section 13.2.1 you quote applies to
each branch independently.
> Well, the first branch is disposed of with a 5xx reply. But the UAC
> cancels nothing, in spite of getting two different early responses
> from two different dialogs.
You should have mentioned the 5xx reply in your original message, as
that's how the first dialog ends.
> Oh, yes -- they're different dialogs, for sure. I just wasn't sure if
> that would nevertheless pose a problem for some low-budget UAs.
A lot of low-budget UAs handle many situations incorrectly. Ultimately,
the answer is to avoid using them, because otherwise you're spending all
your time adjusting your software to try to figure out what the UA is
and avoiding its flaws.
Dale
_______________________________________________
Sip-implementors mailing list
Sip-implementors@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
