RE: Mapping Specs to the Architecture

So I mentioned one "proactive" area in the f2f.  It was arrived at by the
systematic approach of comparing REST vs SOA architectural styles, and
seeing which properties had been potentially lowered using SOA.  I then
proposed this area, and I got some positive feedback and frankly some
negative feedback.  In some circles, it was classified as a "trout pond",
and I was actually surprised it didn't spawn (so to trout-speak) more
discussion.  BTW, the area was how to do caching of SOA responses for higher
network perf.  I don't really know how to proceed with this.
 
I think a very relevent question to ask wrt mapping specs to architectureis:
are we documenting functional areas, specifications that fit these areas, or
both?  For example: imagine we want to talk about transaction context
propagation.  Shall we talk only about ws-transaction and ws-coordination,
or shall we also talk about btp?  For choreography, shall we talk about
bpel4ws, wsci, wscl, bpss, bpml, bpmi, various research projects?
 
If standards process to date can be my guide, we will end having an
inclusive, rather than exclusive list.  The "stds" body won't want to make
the hard choice of picking winning specifications.  Which means, to be
blunt, my company isn't too terribly interested.  We have made our bets on
which things will "win".  Why would we want to be part of an effort that
legitimizes things that we don't believe in?  Of course, other companies
have made different bets and have the same concerns that we have wrt their
choices.  Or will we try to slice and dice, and offer each "constellation"
of specs?  Companies A, B, C believe in XYZ specs.  Companies D,E,F believe
in 123 specs.  hmmmmm, the mind boggles as to how or why this would be
useful to external companies or developers.
 
Cheers,
Dave

-----Original Message-----
From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org]On
Behalf Of Ugo Corda
Sent: Monday, March 10, 2003 7:24 PM
To: Cutler, Roger (RogerCutler); www-ws-arch@w3.org
Subject: RE: Mapping Specs to the Architecture


I think we did some type of informal work in this area when thinking about
which new W3C WGs we could spawn as a result of our investigations. That way
we first started talking about security, then orchestration, and finally
reliability. Our process has not been very systematic so far, but more based
on some of the most urgent and obvious current needs.
 
I agree that a more systematic process could pull up areas that the industry
has not being chatting about yet. The risk, of course, is going too much
ahead of the industry's currently expressed needs and spending efforts on
areas that could be found to be too specialized and niche-related, of low
current relevance, etc. 
 
I have the impression that many people in the industry are starting to get
the idea that Web services consist of a never ending sequence of new
identified technology areas and associated standards (not yet available) and
wonder whether Web services will ever be in their grasp. I find myself
wondering many times how far standards organizations are from having WS
covered. In order to get an at least approximate answer to that question we
might need a more systematic exploratory work.
 
Ugo

-----Original Message-----
From: Cutler, Roger (RogerCutler) [mailto:RogerCutler@chevrontexaco.com]
Sent: Monday, March 10, 2003 11:13 AM
To: www-ws-arch@w3.org
Subject: Mapping Specs to the Architecture



I had a chat with TimBL about the WS Arch work in which he asked a very
interesting question.  He wanted to know whether we were producing a diagram
that would make clear what parts of the architecture currently have specs in
place, what parts have specs in progress and what parts need specs but there
is nothing in sight.  I don't think that this kind of thing is explicitly a
part of our requirements, and in fact I think that we have not thought too
much about this particular audience -- the people that launch or coordinate
the spec efforts as opposed to the spec writers themselves -- but I think
that the expectation is a pretty reasonable one.

I made some really dumb response like, "Gee, got me.  I suspect that it's a
little early to see results like that but we may be doing things that could
go in that direction".  But now I wonder -- could specs be mapped onto the
spaghetti diagram that we have been working on?  If not, does that indicate
that some other view is needed?  It seems to me that some specs go onto that
diagram pretty naturally, but that some involved with the "ilities" might
not.  Or perhaps that is just because the diagram is incomplete??