W3C home > Mailing lists > Public > www-rdf-interest@w3.org > April 2001

RE: A Simple Analogy

From: Danny Ayers <danny@panlanka.net>
Date: Sat, 14 Apr 2001 23:18:51 +0600
To: "Brian McBride" <bwm@hplb.hpl.hp.com>
Cc: "RDF-Interest" <www-rdf-interest@w3.org>
Message-ID: <EBEPLGMHCDOJJJPCFHEFGEBPDCAA.danny@panlanka.net>
Hi Brian,
Thanks for the reminder - to be honest, I'd completely forgotten about WSDL.
Yes, this is very, very much the kind of thing I had in mind, though I've
been approaching it from a different angle. WSDL is in a very nice
position - it is reasonably inclusive, pretty much any process the semantic
web might be interested in using could be defined as a web service, and it
follows from tried & tested IDL techniques. It also has some pretty good
industry backing (which is considerably more than I can say) though at least
one of the companies behind it has a rather strange approach to
interoperability.

Where I would find fault however, is in the almost-proprietry approach of
trying to build something for the purposes of WSDL as a reworking of IDLs.
Step back, bigger picture, cliches, platitudes. I would see the interface as
being just one particular component of the model of the web
service/agent/process (as it happens, I defined Interface as a subClass of
Component in my RPP proposal [1]). I believe web services (and other
processes) should be viewed as resources, and can and should be modelled
using a framework developed for that purpose,  and expressed in a syntax
used for the description of other resources. In other words, why come up
with yet another ML (WSML) when RDF would not only be suitable, but would
allow the web service metadata to be operated on by exactly the same agents
that can be used for other metadata. There may be big advantages in being
able to plug WSDL directly into e.g. COM, but I reckon the advantages of
being able to plug directly into the semantic web outweigh these.

In the arena of WSDL, I'd better mention UDDI. I don't have a problem with
this - some kind of registry/discovery etc is needed, and SOAP seems a
reasonable protocol for such work. But I'd argue that if you go with RDF
(something like RPP) then for one you don't need any new registries,
existing ones such as search engines and even directory services can be
used. For description you have RDF. The enquiries your agents make are
handled by exactly the same facilities that will handle any other query on
the semantic web.

There's a good chance that WSDL & UDDI will get good adoption - I don't
mind, it's a step in the right direction. I just think that at some point
the WSDL <-> RDF linkage will be needed, whereas if you go the route of pure
(once  it's been ironed) RDF then things like WSDL & UDDI are redundant.

Web services are resources - why not treat them as such?

Thanks again for the reminder Brian - I'm going to have to have another look
at WSDL to see what I can pinch ;-)

[1] http://www.isacat.net/citnames/2001/04/rpp.htm

<- Hi Danny,
<-
<- Have you had a look at WSDL?
<-
<-   http://www.w3.org/Submission/2001/07/
<-   http://www.w3.org/Submission/2001/07/Comment
<-
<- Is that related to the sort of thing you had in mind?
<-
<- Brian
Received on Saturday, 14 April 2001 13:24:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:48 GMT