Hi,

>Christer
>You are right, of course. Often one realizes certain subtleties only during
>detailed implementation and does not see them rightaway on reading the spec.

Been there - done that :)

>I was simply stating that it is not explicitly stated that the REQ-URI of
>the CONTACT in a reliable 1xx be used to send the PRACK. One has to deduce
>this from the statement that a CONTACT is mandatory on a reliable 1xx.
>Of course, if we do this then the PRACK will bypass the proxy, which is
>probably for good (but unstated) reasons that we don't want a stateless
>proxy to fork the PRACK, we want the PRACK to be delivered speedily etc.

Yes, but if the proxy record route, it will still receive the PRACK.
 
>Further, if we CANCEL the original request, we must not send it to the
>Contact, but to the REQ-URI of the original INVITE,  right?

That is true. There has, however, been a discussion about this on the
SIP mailing list. The issue has been if it should be possible to CANCEL
specific call-legs.
 
>Also, if the call is setup and a RE-INVITE is issued directly to the
>CONTACT, the UAS must not send a different CONTACT in the 1xx.

In which 1xx? The 1xx to the RE-INVITE?

>If it did, what should the UAC do? Ignore it and continue using the original
>CONTACT/terminate the call/ update the CONTACT and direct the PRACK/CANCEL
>to the new CONTACT?

Strange things may always take place, in very often it's an
implementation issue how/if they are handled by the nodes, but the basic
rule is to always use the latest CONTACT. After all, it's there for a
reason :)

Regards,

Christer Holmberg
Ericsson Finland



> 
> ----- Original Message -----
> From: "Christer Holmberg" <[EMAIL PROTECTED]>
> To: "A Venkatraman" <[EMAIL PROTECTED]>
> Cc: <[EMAIL PROTECTED]>
> Sent: Friday, July 20, 2001 4:47 PM
> Subject: Re: [Sip-implementors] Req-URI of PRACK
> 
> >
> > Hi,
> >
> > Isn't what you say true for ANY request after the INVITE? Ie, if the
> > proxy doesn't record route, it will not receive the next request, no
> > matter if it is a PRACK, INFO, BYE.... or whatever.
> >
> > Or, did I missunderstand your question?
> >
> > Regards,
> >
> > Christer Holmberg
> > Ericsson Finland
> >
> >
> >
> >
> >
> > A Venkatraman wrote:
> > >
> > > I would like affirmation (or negation+correction) of my understanding of
> how
> > > the Req-URI is to be built for a PRACK when the proxy does not add a
> Record
> > > Route.
> > > Since the CONTACT is mandatory on a reliable provisional response, the
> PRACK
> > > must use the CONTACT to build the REQ-URI.
> > > This means that the PRACK will bypass the proxies, since none of them
> have
> > > record routed.
> > > A PROXY THAT DOES NOT RECORD ROUTE DOES NOT EVER SEE A PRACK (unless it
> is
> > > the default outbound local proxy). Correct?
> > >
> > > The 100rel draft does not specifically mention this, probably because
> this
> > > is the only obvious conclusion from the statement in the UAS section
> about
> > > CONTACT being mandatory on the reliable prov response.
> > >
> > > _______________________________________________
> > > Sipping mailing list  http://www.ietf.org/mailman/listinfo/sipping
> > > This list is for NEW development of the application of SIP
> > > Use [EMAIL PROTECTED] for questions on current sip
> > > Use [EMAIL PROTECTED] for new developments of core SIP
> > >
> > > _______________________________________________
> > > Sip-implementors mailing list
> > > [EMAIL PROTECTED]
> > > http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
begin:vcard 
n:Holmberg;Christer
tel;cell:+358-40-5604412
tel;work:+358-9-2992943
x-mozilla-html:FALSE
org:Ericsson;IP Multimedia / Advanced Signalling Research Laboratory
adr:;;;;;;
version:2.1
email;internet:[EMAIL PROTECTED]
title:System Designer
fn:Christer Holmberg
end:vcard

Reply via email to