Manpreet Singh wrote:
> Paul
>
> Thanks for the response. This is what I had in mind too. Somewhere in
> the host portion or R-URI or the Route header, the UA should be able to
> locate the FQDN resolving to itself ( one either by registration or
> simply provisioning ) or its IP address. Sending a request with a bogus
> host in the host part of the URI should not be right.
>
> The reason I brought this up was because some UAs which are really GWs
> (IP-TDM), don't care for it. All they care for is the user part of the
> URI where they would route on the TDM side and they would simply accept
> anything in the host portion just because they were listening on port
> 5060. I was hoping to find something in the spec which kinda mandates this.
AFAIK there are no *rules* about this.
If a UAS wants to be extra lenient about what it accepts then I think it
is free to do so.
A GW is a problem case, because to be useful people often do fabricate
the R-URI for it rather than use one that was given to them by the gw.
But even then, I think it is reasonable to expect that something
fabricating a URI in this way must follow the rules defined for the
target. If the target calls for a particular value in the domain name,
then they had better do it that way.
Which side of this are you on? If you are the one sending to the GW,
then as long as what you do works, why do you care if other things also
work? If you are developing a GW, I think you can righteously insist
that the URI be formed to your expectations.
Paul
> Manpreet
>
> -----Original Message-----
> From: Paul Kyzivat [mailto:[EMAIL PROTECTED]
> Sent: Friday, November 03, 2006 7:29 PM
> To: Manpreet Singh
> Cc: sip-implementors@cs.columbia.edu
> Subject: Re: [Sip-implementors] Host in the R-URI
>
> In general, an UA should only be receiving requests with an R-URI that
> it published in some way. This might be by registering, or in the
> absence of registration, by some kind of provisioning process.
>
> Callers should not be simply making up URIs that point to you.
>
> It is also true that the R-URI will typically have to resolve to you. An
> exception to that is if there is a Route header, in which case that
> should resolve to you and be something you published. But its not normal
> for there to be a Route header when a request is delivered to a UAS.
>
> (This may change - Jonathan has a draft proposing to use this as a
> feature.)
>
> The story may be different for B2BUAs and SBCs, depending on how they
> get spliced into the call path.
>
> Paul
>
> Manpreet Singh wrote:
> > Hi
> >
> > I know I have had discussions about this before but just to confirm
> > again...when a UA ( UAC, be it an IP Phone or the UAS side of a
> > B2BUA/SBC) recieves an INVITE, doesnt the host portion of the R-URI
> > needs to match the UA's IP or if FQDN resolve to UA's IP? So if I send
> > an INIVTE to an IP phone with something bogus in the host portion,
> > should the IP Phone accept that call based on the spec?
> >
> > For a proxy it would be different as the proxy can simply be an
> > outbound proxy to a client and the host doesnt have to match the
> > proxy's IP or domain and here the proxy would simply route the call
> > based on domain if its not the last hop for that call.
> >
> > M
> > _______________________________________________
> > Sip-implementors mailing list
> > Sip-implementors@cs.columbia.edu
> > https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
> >
>
_______________________________________________
Sip-implementors mailing list
Sip-implementors@cs.columbia.edu
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
