Ernst,

Sorry for the delayed reply. I've been busy the latter part of the week and just now catching up on some standards e-mail.

Here's the problem: a.. Suppose a T.38v2 devices calls a T.38v0 device, proposing T.38v2 in Fast Connect. b.. The called T.38v0 device will accept the T.38v2 proposal... there is no text to tell it not to. c.. The called T.38v0 device returns a fastStart element containing v2! d.. The calling device then sends IFP packets encoded according to the 2002 syntax, which will fail to decode. If we had text in H.323 Annex D or elsewhere that said that an endpoint shall not accept a proposal for a version higher than it understands, that would be OK. However, we don't have such rules. If we forced the endpoints to at least pass the capability set messages in the Setup and response to Fast Connect, that would also address the issue... but we don't have that requirement either.

So, we need two things: 1.. We need to address this versioning issue quite explicitly 2.. We need some "kludge" to allow v2 devices to work properly with v0 devices that accept a v2 fastStart proposal

Paul

----- Original Message ----- From: Horvath Ernst To: 'paulej@PACKETIZER.COM' Cc: itu-sg16@external.cisco.com Sent: Wednesday, September 17, 2003 4:55 AM Subject: AW: H.323 Fast Connect and Versioning

Paul,

I still don't see why the existing version field in the t38FaxProfile isn't sufficient and how an additional syntax2002 field would help much.

If an old implementation returns an unrecognised version field, why should it drop an unrecognised new field? So those implementations still won't behave as expected.

The only case where syntax2002 would improve things seems to be with version 0/1 implementations that understand the version field and answer incorrectly by returning a "version = 2" field unchanged. Am I right?

Ernst -----Ursprüngliche Nachricht----- Von: paulej@PACKETIZER.COM [mailto:paulej@PACKETIZER.COM] Gesendet am: Mittwoch, 17. September 2003 06:26 An: itu-sg16@external.cisco.com Betreff: Re: H.323 Fast Connect and Versioning

Anatoli,

So, I think the first step for T.38 is to add the syntax2002 field. Following that, we need to go back and more clearly define versioning issues for all sub-protocols, including T.38, V.150.1, V.ToIP, V.VBD, etc. All of those will present similar issues.

Paul