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 01:11:36 +0100
Message-ID: <37E80E80B681A24B8F768D607373CA800103DA89@largo.campus.ncl.ac.uk>
To: <www-ws-desc@w3.org>

Hi Paul: 

> i'm OK with not preventing WSDL from using 'voodoo 
> dispatching' since i don't want to restrict or prevent other 
> unforeseen dispatching mechanisms being employed in the future.

Surely all dispatching is "voodoo-esque" since you have no idea what the thing the WSDL contract describes actually is. Tying WSDL-level operation names to, say, the method names on a particular class is perhaps one convention that might not be considered as "voodoo" but it sucks :-)
 
> i'm also concerned that the sender may be expecting '32C' in 
> response but instead receives 'DEFCON 1'.
> 
> can you reassure me this isn't going to happen in practice?

No I'm afraid I can't - the WSDL contract specifies the message structure and message exchanges. However I think that while in practice most WSDL contracts will seem normative (and indeed will behave normatively w.r.t. their associated service) there is guarantee that the WSDL you get is a true reflection of the implementation of a service.

Now if you make the dispatch public you might think you can get round that problem, but I contend that how I dispatch my messages is of no importance to my consumers. If I tell you how I'm going to dispatch you will use that in your code (you know you will because it will make your life easier) but then when I go and change something I break you.

Jim
--
http://jim.webber.name 
Received on Tuesday, 13 July 2004 20:10:04 GMT

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