W3C home > Mailing lists > Public > www-ws-desc@w3.org > July 2004

RE: Action Item 2004-07-01 Solution to 168/R114

From: Jim Webber <Jim.Webber@newcastle.ac.uk>
Date: Wed, 14 Jul 2004 12:54:05 +0100
Message-ID: <37E80E80B681A24B8F768D607373CA800103DB46@largo.campus.ncl.ac.uk>
To: <www-ws-desc@w3.org>

Paul:

> OK, i'm happy for you and Gudge to publish _your_ WSDLs with 
> a single operation "stuffHappens" or whatever. That's fine - 
> you both have a strong architectural POV, one shared by many others. 

Let me clarify - I understand that WSDL contracts will, in general,
contain many wsdl:operation elements. As Gudge points out, however,
these are *WSDL* level descriptions and do not necessary map onto
physically invokable entities in the service implementation (WSDL isn't
yet another object IDL).

> But there is another, equally valid POV where a WSDL author 
> has gone to the trouble of providing several "buckets", each 
> of which has different semantics: "getTemp", "nukeLocation" 
> etc. Lots of other people want to be able to describe this 
> explicitly, and that's also fine by me.

Indeed I believe that most WSDL contracts will be of this form.

> FWIW i can't really see many services being published in 
> which the 'action undertaken' isn't obvious for a whole host 
> of reasons, not least access control and auditing. After all, 
> if you really don't really know what's going to happen when 
> you send a message, or what you can expect in return, then 
> it's not a very useful service! 

Again I agree that you understand the semantics of a message exchange
otherwise it is pointless. I think the point that I'm trying to make is
that WSDL contracts are a description of messages that a service
understands and the message exchange patterns that it supports. To try
to infer anything more that that, such as where a message is dispatched
to, is both fruitless (you can't because it's hiddent) and dangerous
(because you shouldn't be binding to things that should be hidden).

This issue rather boils down to: does the WG provide a means of
identifying/binding to/influencing a Web Service's dispatch mechanism or
not. My personal preference is that such a mechanism is the business of
the Web Service and should not be advertised through a WSDL contract
(it's not WSDL level as Gudge put it). Others may disagree, in which
case my point of view may seem to be wrong.

Jim
--
http://jim.webber.name  
Received on Wednesday, 14 July 2004 07:54:35 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:32 GMT