W3C home > Mailing lists > Public > www-ws-desc@w3.org > September 2003

RE: On WSDL "operation"

From: Jacek Kopecky <jacek.kopecky@systinet.com>
Date: Wed, 17 Sep 2003 14:52:13 +0200
To: Jim Webber <jim.webber@arjuna.com>
Cc: 'WS Description List' <www-ws-desc@w3.org>
Message-Id: <1063803133.3354.36.camel@localhost>

Jim, please see below.

On Tue, 2003-09-16 at 17:48, Jim Webber wrote:
> No. Services are entites which exchange messages, period. Web Services are
> characterised by the exchange of structured documents described with XML
> Schema (normally). WSDL is a means of describing those documents and how
> they will be exchanged between a single client/requester/consumer (or insert
> today's WSA term here) and a service (similarly, insert today's WSA term
> here).

In your model, do you need message exchange patterns? I think the point
of message exchange patterns is that they are greater than just the sum
of the messages comprising them. What is the difference in value of an
MEP and the sum of the messages? The difference is in some additional
semantics. For example in an input-output MEP, the additional semantics
is that "something happens as a result of receiving the input, which in
turns results in producing the output". I don't want to lose this
"something that happens". That exactly is an operation.

If we only have messageExchanges with no semantics to them apart from
the message exchange, we don't need MEPs and can only have two kinds of
messageExchanges - one way there and one way back. Even if we don't go
there but still only have messageExchanges with no additional semantics,
it is conceivable that two today's operations could collapse into one
messageExchange because they have the same inputs and the same outputs
and only differ in semantics. Say  int sum(int[]);  int product(int[]).

I'm for keeping operations. I think services aren't there to exchange
messages, they are there to *serve*, and if message exchange is
necessary for that (as it usually is), it ought to be specified in the
container that identifies (not defines, though) the particular serving
operation that the service is capable of.

Best regards,

                   Jacek Kopecky

                   Senior Architect
                   Systinet Corporation
                   http://www.systinet.com/
Received on Wednesday, 17 September 2003 08:52:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:26 GMT