Re: [ws-caf] Future of WS-Context

[Mark Little <mark.little@jboss.com>]

James, what you're not seeing, and perhaps is not recorded sufficiently 
in the minutes of the meetings over the past 9 months or so, is the fact 
that we have discussed this subject on a number of occasions. My 
original email was deliberately not detailed because I think the people 
left on this group from companies such as Oracle, IONA and JBoss know 
the issue sufficiently well.

James Bryce Clark wrote:
>     I am interested in Mark's questions, too.  [1]
>
>     The simplest one to answer is whether WS-CF can go out for public 
> review.   It already has.  [2]
>
>     A less simple one is whether it is worth doing work, if one of 
> three large US software companies don't front it.  (Or, overlooking 
> Mark's choices, any similar list.)   That's for this committee to 
> decide.  It's a fine thing for TCs to look up occasionally, and ask, 
> "is this project still needed"?   By all means, if a project is a 
> waste of time, it ought to stand down.

I don't think this is necessarily a waste of time and in an ideal world 
all of our companies would have sufficient people and time to put on 
every specification that's out there. Unfortunately we don't live in an 
ideal world and the time has come for us all to either commit to doing 
more work in WS-CAF, or to agree that it has done its job once 
WS-Context is adopted. I don't want WS-CAF to suffer the kind of long, 
slow death that some other OASIS TCs have (names removed to protect the 
innocent!)

>     But is it?  With apologies to Mark, his criteria may be a bit 
> myopic.  It's fairly clear from experience that success and failure in 
> XML and SOA are not driven exclusively by large US software companies.

In this space, I think they are.

> Sometimes success absolutely happens because of them.  Other times it 
> happens without them.

In general I agree, but specifically I can't: you only have to look at 
the history of transaction processing systems to see that.

>     The better predictor of success for OASIS output, across our 
> history and that of similar orgs, seems to be early adopter pickup by 
> buy-side users in genuine production. (As opposed to vendor 
> announcements of inclusion in tools -- also a factor, but less of one, 
> I think, for reasons we could discuss if you like.)
>     In that light -- whether actual users express a demand for WS-CAF 
> or its unique functions -- how does WS-CAF look?

In the area of (Web Services) transactions, without the backing of 
*several* major players such as IBM, Oracle, BEA, MSFT and JBoss, a 
specification is highly unlikely to be adopted by end users. Take a look 
at BTP for example: irrespective of what you may think of it 
technically, as soon as BEA, HP and Oracle dropped out, it was doomed. 
To that end, we as a TC need to ask ourselves the following questions. 
Is there any point in continuing if the main thrust of this work is 
being done elsewhere (WS-TX)? I think we're all agreed that there are 
deficiencies in WS-TX that don't appear in WS-CAF and we could provide 
this added-value work on top of WS-TX, but that would require a 
re-charter of WS-CAF. Do we re-charter, or start up another TC? Without 
the backing of other members from the current WS-TX TC, would either 
approach be worth while?
>
>     A while ago, I had the impression that WS-CAF was working from a 
> base theory close to the old MSFT/IBM proprietary WS-C and WS-T.  So 
> is OASIS WS-TX, I gather.  OK, if so, that raises a simple *technical* 
> question.  Will WS-TX do everything WS-CAF can?

Ignoring WS-Context (which I want us to adopted in WS-CAF), then WS-TX 
will do 80%-90% of what WS-CF/WS-TXM does.

>   Or everything that will matter to customers?   Is this just a 
> VHS-Betamax thing, with a delta uninteresting to users, or is there 
> significant functional distinction?

WS-Context is a significant difference. There's also an additional 
transaction model in WS-TXM that is no provided for in WS-TX. If we had 
commitment from all TC members to continue in WS-CAF, then, as I said in 
my first email, I'd be happy to continue. But in this space it really 
does require a critical mass.

>    Can anyone point us to recent info, or reasonably articulate public 
> conjectures, about this?

Maybe the minutes of some teleconferences, but I can't be sure.
>   Apologies for asking -- I recall some earlier chat, such as [3], but 
> not so much recently.   I am sure there is some discussion about this, 
> well known to those of you who concentrate on the area, that I just 
> haven't readily located.

Mark.

>
>     Regards  Jamie
>
> ~   James Bryce Clark
> ~   Director, Standards Development, OASIS
> ~   jamie.clark@oasis-open.org
>
> [1]
> At 03:27 AM 4/3/2006, Mark Little wrote:
>> I'd like to continue the discussion around the future of WS-CAF and 
>> WS-Context. If memory serves, the last time we discussed WS-Context, 
>> Martin argued that rather than move to adopt as an OASIS Standard, we 
>> should first take the currently agreed specification and do some 
>> interoperability testing against it. This is because the last time we 
>> did interoperability (Q2 2005?) it was against a much changed version 
>> of the specification. I believe we all agreed that this was a good 
>> idea but we didn't get commitments for this effort from everyone, so 
>> things stalled.
>> What I'd like to suggest is that we either agree to do some kind of 
>> interoperability workshop (could be purely remote) within the next 
>> two months (say a deadline of end of May 2006), or that we simply 
>> adopt the current specification as an OASIS standard.
>> This then brings me on to the subject of WS-CAF. Firstly, whatever 
>> happened to WS-CF? I believe we fulfilled all of the OASIS rules for 
>> getting it to a public review. Was it put up for public review? If 
>> so, when does that period end?
>> Secondly, I'd propose that as soon as WS-Context is adopted as an 
>> OASIS standard, we close the TC. Although there is some interest in 
>> doing "added value for WS-TX" in the scope of WS-CAF, I don't think 
>> there's enough of a critical mass to make it work. I'd personally 
>> love to do the work, but without companies such as Oracle, IBM and/or 
>> MSFT, I don't think it is worth it.
>> Mark.
>
> [2] http://lists.oasis-open.org/archives/tc-announce/200602/msg00015.html
>
> [3] http://lists.oasis-open.org/archives/ws-caf/200403/msg00030.html