On Wed, Feb 3, 2010 at 12:13 PM, Scott Lawrence <[email protected]> wrote:
> On Wed, 2010-02-03 at 11:54 -0500, Dale Worley wrote:
>> On Wed, 2010-02-03 at 10:18 +0000, Sven Evensen wrote:
>>
>> > The scenario is as follows:
>> >
>> > A calls B, B answers
>> >
>> > Our appl. sends a REFER to transfer from A to C
>> >
>> > The REFER has From : A, To : B, Refer-To : C, Referred-By : A
>> >
>> > When using unmanaged GW (Cisco IOS GW), it replies with
>> >
>> > INVITE From : B, To C
>> >
>> > sipXBridge replies with
>> >
>> > INVITE From : A*, To C
>> >
>> >
>> >
>> > A – 8116          (phantom user)
>> >
>> > A* - 01225581810 (DDI of A)
>> >
>> > B – 07795951717   (mobile)
>> >
>> > C – 8216          (phantom user)
>> >
>> >
>> >
>> > The question is: Is Cisco GW doing right or is sipXBridge doing right?
>> > Or maybe both?
>>
>> As a general rule, the initial INVITE from A to B will establish the
>> From/To URIs of the dialog, and when B sends the new INVITE to C, it
>> will use as its From URI a URI which represents B in some way.  Usually
>> phones have an AOR for this purpose, but I suppose a B2BUA could create
>> a URI based on what the "real" UA is on the "other side" of the dialog,
>> or it can use the From/To URI that was used in the original dialog.
>>
>> In this case, sipXbridge seems to be using the From address of the
>> original INVITE that it received, which isn't expected to describe B in
>> any way.  So that's a bug.
>
> I don't think so - at least not in the sense that it's something we need
> to change.
>
> In order for the sipXbridge to construct a To/From field value that
> reflected what was on the 'other' side, it would have to either wait for
> some response from the other side (which would almost certainly mean
> breaking the rule of sending a 100 response within 100ms), or it would
> have to change what it sent later (which would require a dialog-updating
> transaction, and might not be supported well by phones anyway).  Add to
> that the complexity of keeping it correct when calls are transferred and
> the inherent problems with To/From headers essentially being just
> decoration that can't be used for routing anyway (which you pointed
> out), and I'm lead to the conclusion that there's nothing here that
> needs fixing.


Thanks Dale and Scott for the analysis.

Since this is a gray area, I will refrain from making any changes in
the short term.  I did things this way to allow for the right
caller-id to show up (or what I considered to be "right" ).  Perhaps
we need an override ( configuration parameter) for this behavior that
we can consider in 5.0


>
>
>
> _______________________________________________
> sipx-users mailing list [email protected]
> List Archive: http://list.sipfoundry.org/archive/sipx-users
> Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users
> sipXecs IP PBX -- http://www.sipfoundry.org/



-- 
M. Ranganathan
_______________________________________________
sipx-users mailing list [email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users
sipXecs IP PBX -- http://www.sipfoundry.org/

Reply via email to