On Thu, 2005-10-27 at 08:52 -0500, Dennis Dupont wrote:
> Thanks for the RFC sections.  As many times as I have read the spec, I
> have until now missed the reference to hairpinning.

That surprises me, since the text I quoted is in the sections of 3261
specifically devoted to the From and Call-ID headers when generating
requests.

> But your extra
> comments have me confused.  Can you elaborate on the idea that "there
> can be multiple early dialogs generated by one request"?

Perhaps reading section 13.1 would make it clearer.

> From section
> 8.1.1.4, the Call-ID must be the same for all messages in a dialog, but
> your comment implies that there can be multiple dialogs for the same
> Call-ID.  This is where I have trouble:
> - What is a dialog and why would there be more than one in a call?

As it notes in section 13.1, if the INVITE is forked (forwarded to more
than one destination), each destination will respond separately,
establishing multiple dialogs from the one INVITE.  Of necessity, each
dialog will have the same Call-ID and from-tag, but each will have a
different to-tag.  (That's why there are to-tags.)

> - What is the difference between an "early dialog" and any other?  

An "early" dialog is one for which a (non-100) provisional response to
the dialog-initiating request has been send (thus establishing a to-
tag), but a final response has not.  A "committed" dialog is one for
which a final response has been sent.  (This terminology was not
established until after 3261 was written, so 3261 is not very clear on
the matter, though this framework was intended by 3261.)

> - And the big kicker: if "Every dialog-starting request must have a 
> unique Call-Id", how can there be multiple dialogs started from one 
> request without multiple From tags?

See above.

Dale


_______________________________________________
Sip-implementors mailing list
[email protected]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to