Hi Gonzalo,
I also agree to Mike's suggestion. Can't a recommendation be made on
the INGRESS GW that the CSeq for an overlapped dial message should
monotonically increase, though each of the INVITEs constitute a different
dialog. This can help the EGRESS GW to drop out-of-order INVITEs.
Regards,
Prasanna
Huawei Technologies.
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
[EMAIL PROTECTED]
Sent: Friday, November 29, 2002 11:38 PM
To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]; [EMAIL PROTECTED]
Subject: Re: [Sip] different messages with same call-id before 100 is sent
out
In a message dated 11/28/2002 5:08:11 AM Eastern Standard Time,
[EMAIL PROTECTED] writes:
in the overlap case we do not care if the nth INVITE reaches the other
end before the (n-1)th INVITe, so you do not have to wait for anything
before sending the new INVITE.
[MAP] This may be true as long as the "other end" knows to throw away the
old INVITE (with fewer digits). However, I believe that the very fact that
INVITE's can arrive out of order leads to significant difficulty, since
their correct interpretation depends on the order.
Going back to some earlier comments, I believe there must be a requirement
that, for the PSTN-IP-PSTN case (which I think is the focus in this whole
discussion), that a single IAM received on the ingress side (of IP) must
result in exactly one IAM being sent on the egress side back to the PSTN.
There may be fewer (or no) SAMs sent than received, for example, if the IP
portion, at either the ingress or egress GW, does overlap to en-bloc
operation, or simply has the smarts to put multiple digits (from SAMs)
together.
I believe that a solution to meeting this requirement (single IAM) has a
significant impact on all the other discussions about out-of-order INVITES,
same or different Call-IDs, etc. Therefore, unless it can be agreed that
this is a requirement, all the other discussion is meaningless.
I raise this point especially with relation to my proposals in IEPREP for a
priority-based service for a military application which includes preemption
in the circuit switched world using SS#7 (MLPP). Multiple IAMs (for high
priority calls) would have the unacceptable result of preempting calls and
then releasing the IAM when another arrived (or took another path).
Is it possible to agree that this single IAM rule should be followed?
Mike Pierce
Artel
_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
