Re: Fwd: RE: [ws-caf] Statement for WS-Transactions workshop

["Mark Little" <mark.little@arjuna.com>]

> Eric suggested that the primary authors of the WS-CAF specs would like
> to become authors of WS-C+T in the process of bringing WS-TXM BP into
> the fold. The indication was that WS-TXM Acid and LRA are duplicative of
> AT and BA, and would be dropped. WS-Context was suggested as an aspect
> to be retained, by contrast.

I don't think that's quite right and certainly wasn't the intent (and wasn't
how people from IBM/MSFT took it). Although there is overlap the protocols
aren't exactly the same and there is perceived benefit from perhaps
increasing the size of the WS-T or WS-TXM family of protocols. It was
certainly the intent that merging both sets of specifications somehow rather
then replacing would be the order of the day going into any joint effort. Of
course what the end result would be is another matter.

> In his statement he suggested that pretty much all of the original
> author companies of CAF were supportive of this approach.
>
> We in Choreology made the statement that we considered BTP to be dead,
> i.e. not a contender as leading specification for Web Services
> transactions. (We have held this view since it became clear that IBM and
> Microsoft introduced their own specs in this space, but now seems like a
> sensible time to emphasize the point. We believe BTP is like a
> transition relationship: it's necessary in the absence of more durable
> or authorative solutions; it's good while it lasts, it fills the gap
> while the other solution gets sorted, and it's not going to last. We
> would like to see WS-BA emerge as the vanguard spec, drawing on other
> and prior work, e.g. BTP and CAF.)

At least the commonly held view seems to be that something like WS-BA will
emerge to fill the specific requirement of WS-BPEL. As with WS-CAF,
IBM/MSFT/BEA still agree that there will be different protocols for
different use cases.

I know that there is interest in the LRA and BP models in the "other camp",
and like I said, the differences between Acid and AT may be approached by
having two models in roughly the same space. Anyway, this is all speculation
at this time. Only time will tell.

> We raised a whole series of issues relating to details of WS-AT, and
> more substantive methodological ones relating to WS-BA. There was not
> time to discuss or even raise all of our points and we were invited to
> submit a detailed, prescriptive account in writing for further
> consideration, which we will do. The key items are: need to align
> business promises (e.g. reservations) with protocol promises (PREPARED);
> equality of treatment for the positive and negative final signals
> (CONFIRM/CANCEL in BTP-speak, CLOSE/COMPENSATE in WS-BA speak); need to
> allow selective confirmation of prepared participants (do not introduce
> a rule that ensures a uniform outcome across the whole pool). We also
> believe that an interoperable control protocol (terminator-coordinator)
> protocol is very important (although not strictly critical) for a
> selective confirmation protocol.
>
> We are also interested in the checkpointing and notification aspects in
> WS-TXM BP.

Mark.

----
Mark Little,
Chief Architect, Transactions,
Arjuna Technologies Ltd.


>
> Alastair
>
>
> -----Original Message-----
> From: Newcomer, Eric [mailto:Eric.Newcomer@iona.com]
> Sent: 17 March 2004 13:50
> To: Pete Wenzel
> Cc: ws-caf@lists.oasis-open.org
> Subject: RE: Fwd: RE: [ws-caf] Statement for WS-Transactions workshop
>
> Pete,
>
> My apologies to you and everyone else on the list.  I should have
> thought to send an email right after the meeting.
>
> There was no formal reaction to our statement.  The workshop was set up
> as a feedback workshop, and the WS-CAF statement was taken in that
> context.  We (Mark Little and I) also proposed a "straw horse" proposal
> for merging WS-CAF with WS-AT, WS-BA, and WS-C.  I understand it's being
> evaluated.
>
> Overall the meeting consisted primarily of presentations on the
> BEA/IBM/MSFT specifications with some discussion about them that
> resulted in about a dozen issues being raised.
>
> I think it was a good opportunity to let everyone know what's going on
> in WS-CAF, make the point about common ancestry and technical
> similarities, and highlight where WS-CAF provides extensions to the
> BEA/IBM/MSFT specs, especially around generic context management and the
> business process transaction model.
>
> We will have to wait for a more formal reaction.
>
> Eric
>
> -----Original Message-----
> From: Pete Wenzel [mailto:pete@seebeyond.com]
> Sent: Tuesday, March 16, 2004 2:10 PM
> To: Newcomer, Eric
> Cc: ws-caf@lists.oasis-open.org
> Subject: Re: Fwd: RE: [ws-caf] Statement for WS-Transactions workshop
>
>
> Eric, are you able to report anything about the workshop, or was it
> held under nondisclosure?  Any response to your proposal?  Thanks.
>
> --Pete
> Pete Wenzel <pete@seebeyond.com>
> Senior Architect, SeeBeyond
> Standards & Product Strategy
> +1-626-471-6311 (US-Pacific)
>
> Thus spoke James Bryce Clark (jamie.clark@oasis-open.org) on Mon, Mar
> 08, 2004 at 10:36:30PM -0800:
> > >> --- Below this line is a copy of the message.
> > >>
> > >> Date: Mon, 8 Mar 2004 19:24:08 -0800 (PST)
> > >> From: Eric Newcomer
> > >> Subject: RE: [ws-caf] Statement for WS-Transactions
> > >> workshop
> > >> To: ws-caf@lists.oasis-open.org
> > >>
> > >> Apologies - I typed this in much earlier today, but
> > >> our email system has been out since noon, and is
> > >> apparently still down.  So I'm posting from my
> > >> private account.  Eric
> > >> ------
> > >>
> > >> Hi,
> > >>
> > >> Per today's concall, here is the edited version of the
> > >> statement I plan to give on behalf of the WS-CAF TC at
> > >> the Microsoft/IBM/BEA WS-Transactions feedback
> > >> workshop Wednesday March 10.  Please let me know if
> > >> there are any further comments or suggestions.
> > >>
> > >> The WS-CAF TC would like to recognize the common
> > >> ancestry and technical similarities across the WS-T,
> > >> WS-C, WS-BA and WS-CAF sets of specifications.  During
> > >> our work we've discovered the benefits of separating
> > >> out context management as a generic mechanism, and
> > >> have developed a key additional protocol called the
> > >> Business Process transaction model.  We think the
> > >> WS-T, WS-C, and WS-BA specifications would benefit
> > >> from including these major concepts.
> > >>
> > >> We propose a discussion on finding the best way to
> > >> move forward and bring our work together.
> > >>
> > >> Thanks -
> > >>
> > >> Eric
>