>>However, User Agents cannot rely on this behavior because
>>proxies that do not support 3326 will likely not include
>>the Reason header in CANCELs they send.
So if any new header wants to be used in a CANCEL, then all proxies
have to be software-updated. New headers for CANCEL won't be
guaranteed to work unless every proxy in the path is updated.
That's an awkward default.
Ok, so I understand the rules, but what is the reason for this?
What are the problems/dangers of copying CANCEL headers across?
Regards,
Attila
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Bob
Penfield
Sent: 09 February 2009 12:39
To: Iñaki Baz Castillo
Cc: [email protected]
Subject: Re: [Sip-implementors] "Reason-Header" inCANCELthroughvariousproxies
RFC 3261 section 9.1 defines how CANCEL requests are constructed. Section 16.10
talks about how a proxy processes a CANCEL request refers to section 9.1.
Section 16.6 (Request Forwarding) does not apply to CANCELs in a stateful
proxy. IMO, a strict interpretation of these sections would prevent a proxy
from blindly copying all headers from the received CANCEL into the outgoing
CANCEL. However, new RFCs, which build on 3261 and define new headers that can
appear in a CANCEL request, can define proxy behavior for processing those
headers. RFC 3326 does in fact does define proxy behavior for the Reason
header. On page 3 it states:
Proxies generating a CANCEL request upon reception of a CANCEL from
the previous hop that contains a Reason header field SHOULD copy it
into the new CANCEL request.
So a proxy which supports 3326 should include the received Reason header in the
CANCELs that it sends.
However, User Agents cannot rely on this behavior because proxies that do not
support 3326 will likely not include the Reason header in CANCELs they send.
Cheers,
(-:bob
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Iñaki Baz
Castillo
Sent: Monday, February 09, 2009 7:14 AM
Cc: [email protected]
Subject: Re: [Sip-implementors] "Reason-Header" inCANCELthroughvariousproxies
2009/2/9 Attila Sipos <[email protected]>:
>>>Yes, in fact I hope you are right :)
>
> I do too!!
>
> I worry that one of the SIP heavyweights is going to correct me ( it's
> happened before!! )
>
> I'm sure they will add something to clarify!
Let's wait for them :)
--
Iñaki Baz Castillo
<[email protected]>
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
