RE: new resource model

 

> -----Original Message-----
> From: Mark Baker [mailto:distobj@acm.org] 
> Sent: Thursday, December 18, 2003 9:14 PM
> To: Francis McCabe
> Cc: www-ws-arch@w3.org
> Subject: Re: new resource model
> 
> 
 
> > I have removed the confusing service model aspect of discovery, and 
> > added in agents using discovery services to discover resources.
> 
> I wouldn't think that a resource model needs a separate 
> discovery service.  Representations facilitate discovery.

I think we in the WG generally see Google, etc. as well as UDDI registries
(and ebXML registries?), not to mention the Semantic Web, as  discovery
services.  It's hard for me to see how we could ignore all that and just say
that somehow "representations facilitate discovery".  Sure, discovery
happens by all sorts of means, manual and automatic, semantically aware and
not, but it *does* happen, and agents that facilitate it are "discovery
services."
 
> 
> In general, I'd prefer the removal of service, discovery 
> service, and resource description for similar reasons.  If 
> the group wants to describe how resources relate to services, 
> you'd need to introduce a "Web server" into the diagram I 
> think. 

A Web server is a particular kind of service that provides representation
transfer services, IMHO.     I see the Web as an instance of a service
oriented architecture.  I'm not sure we have the relationship between SOA,
the abstract resource model, and the concrete Web right in the current
diagrams, but I don't think your suggestions take us in the right direction.
It seems to go without saying that a model that purports to talk about
"services" in the abstract and "Web services" specifically needs to have the
concept of "service" in a very central place!  Sure, all services are
"resources," but so are all sorts of things that WSA has no interest in. 

> 

> 
> As above, I don't see the need to distinguish between 
> resource description and representation.

We've toyed with trying to fit the idea of the representation of a service
resource into the WSA model, without much success IMHO.  (You and Dave
Orchard, and perhaps other TAG members, were in on the thread IIRC).  The
WSDL description of the service protocol, an OWL description the service
semantics, and the SOAP message that invokes a service could all be thought
of "representations of the service resource."  This question seems like
something for the TAG to think about if they ever get around to worrying
about Web services.  It's definitely too abstract for my tastes, and I think
for the [current] intended audience of the WSA document.  Recall that we've
abandoned all hope of writing a W3C Recommendation and fitting this
rigorously in with the Webarch, and are now trying to write down what *we*
have learned from the exercise so that future travellers into this tar pit
know where the really sticky places are.  

> BTW, I'd also suggest that the agent discovers the URI rather 
> than the resource.

Has the TAG said anything relevant to this point?  If not, and unless this
is well established in linguistic philosophy or whatever,  I would prefer
not to get into such a fine distinction.  In common language, one discovers
the thing itself, not the identity of the thing.  In Web services, one
probably wants to discover all the information about a service to see if it
really does what one needs to do, not simply the URI of a service that is
purported to do what one thinks one wants to do.  Just posing the issue
makes my head hurt!

Received on Friday, 19 December 2003 01:06:19 UTC