comments inline...

On Mon, Apr 2, 2012 at 10:28 PM, Paul Kyzivat <[email protected]> wrote:

> On 4/2/12 11:55 AM, Nataraju A.B wrote:
> > Ravi,
> >
> > I partially agree with your comments. If you consider the following
> > scenario, it is better to select the local sequence based on the cseq of
> > the received ***response*** than from one of the dialogs (d1 / d2).
>
> Not sure I understand, and if I do I disagree.


[ABN] The received **responses** for the INVITE will always carry the same
cseq as that of the original INVITE. Hence there is no issue, if we use
that cseq for the cloned dialog.

>


> The cseq values sent in opposite directions on a dialog are
> *independent* of one another. The starting value in each direction is
> supposed to be a random number in range [0,2^31-1].
>
> Of necessity these need to be managed independently for each dialog,
> though the the initial value from the UAC will be the same for all
> dialogs resulting from the forking of an initial dialog establishing
> request.
>
[ABN] I second this statement.

>
>        Thanks,
>         Paul
>
> > Also there is no requirement (from 3261) to clone/copy the local sequence
> > number from the existing dialog set.
> >
> > If we go according to your theory, then we have search through all the
> > existing dialoigs and select the one with highest cseq's (may not be
> worth
> > doing). Instead if we select the local seq based on the cseq of the
> > received response (OR one with matching half dialog), then it would be
> > simple and all of the entities (proxies / UA's) would work without any
> > issue with cseq value
> >
> > INVITE (cseq:1)---->
> >
> > <--- 180 (100rel), cseq:1 Tag-1
> > PRACK (cseq:2) --->  Tag-1
> > <--- 200 -PRACK, cseq:2, Tag-1
> >
> > <--- 183 (100rel), cseq:1 Tag-1
> > PRACK (cseq:3) --->  Tag-1
> > <--- 200 -PRACK, cseq:3, Tag-1
> >
> > <--- 180 (100rel), cseq:1 Tag-2
> > PRACK (cseq:2) --->  Tag-2
> > <--- 200 -PRACK, Tag-2
> >
> > <--- 180 (100rel), cseq:1 Tag-3
> > PRACK (cseq:2) --->  Tag-3           *** here cseq should be 3 or 4 ????
> ****
> > <--- 200 -PRACK, (cseq:2), Tag-3
> >
> > Cseq:4 would not create problem for any entity, but is it worth searching
> > across all dialogs to find the highest cseq number for an outgoing
> request.
> >
> > Even this may not be a problem with CSF (Call Stateful proxies) since
> each
> > request is handled within the purview of a dialog, NOT session.
> >
> > Typical contents of a dialog include
> >
> > dialog creating method (INV / UPD / SUB)  - SAME for all dialogs, hence
> can
> > be copied from other dialogs
> > Call-ID          - SAME for all dialogs, hence can be copied from other
> > dialogs
> > Local tag       - SAME for all dialogs, hence can be copied from other
> > dialogs
> > Local seq num    - differ from dialog to dialog depends on the state and
> > time  of the dialog,  hence **should not** clone from other dialogs
> >
> > remote tag           - Always differ from dialog to dialog, hence should
> > not clone from other dialogs
> > remote seq          - Always differ from dialog to dialog, hence should
> not
> > clone from other dialogs
> >
> > My 2 cents....
> >
> > Thanks,
> > Nataraju A B
> >
> > On Tue, Mar 27, 2012 at 11:10 AM, Ravi Kumar<[email protected]>
>  wrote:
> >
> >> Hi,
> >> Thanks for reply.
> >>
> >> So for better interoperability  (with proxies for which RFC is not clear
> >> about CSeq for forked leg) if UAC send request with higher CSeq (in
> below
> >> mention scenario CSeq 3) then call be always successful.
> >>
> >> Please share your opinion.
> >>
> >> Regards,
> >> Ravi Kumar
> >>
> >>
> >>
> ----------------------------------------------------------------------------
> >> ---------------------------------------------------------
> >> This e-mail and its attachments contain confidential information from
> >> HUAWEI, which is intended only for the person or entity whose address is
> >> listed above. Any use of the information contained herein in any way
> >> (including, but not limited to, total or partial disclosure,
> reproduction,
> >> or dissemination) by persons other than the intended
> >> recipient(s) is prohibited. If you receive this e-mail in error, please
> >> notify the sender by phone or email immediately and delete it!
> >>
> >> -----Original Message-----
> >> From: Worley, Dale R (Dale) [mailto:[email protected]]
> >> Sent: Monday, March 26, 2012 9:56 PM
> >> To: Ravi Kumar; 'harbhanu sahai'; 'Vinay V';
> >> [email protected]
> >> Subject: RE: [Sip-implementors] Clarification regarding the Cseq to be
> used
> >> in mid dialog req on cloned leg.
> >>
> >>> From: Ravi Kumar [[email protected]]
> >>>
> >>> But in my opinion it should send request with CSeq 3. Because if some
> >> node
> >>> in network use CSeq of dialog on which clone response is received then
> >> above
> >>> implementation will be more flexible.
> >>> UAS will not have any impact because as per RFC 3261 this is not error
> >>> condition.
> >>
> >> True, a UAC (request sender) can choose to use a higher CSeq.  But a
> >> UAS (request receiver) must not *require* that.
> >>
> >> However, I believe that your complaint is based on a conceptual error:
> >> Forking is an *inherent* part of SIP, and all devices (especially
> >> proxies and other "middle boxes") must be prepared to see multiple
> >> forks of a single request and must handle them correctly.
> >>
> >> Once you assume that all SIP elements will handle forked requests
> >> properly, then there is no need to worry about behavior that is "more
> >> flexible".
> >>
> >> Dale
> >>
> >> _______________________________________________
> >> 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
>



-- 
Thanks,
Nataraju A.B.
_______________________________________________
Sip-implementors mailing list
[email protected]
https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors

Reply via email to