Hi Pete,
> I'm interested in a SOAP-based approach, but I think I agree with Tod
> that having a simple HTTP GET/POST interface which returns an RDF/XML
> document would be nice.
I guess I don't see the advantage of this. Why would we want to create our
own interface, query syntax, etc. when everyone else seems to be
implementing Web services?
> Is it not putting a (slightly) heavier burden on user applications to
> insist that they use SOAP? Or is the difference really
> minimal? If it is
> minimal, then that seems OK.
No. Not if we implement the Web services in a protocol-neutral manner.
Publishing the WSDL abstracts the SOAP implementation (the WSIL document
would include a pointer to the WSDL). The clients need not be aware the
service is implemented using SOAP.
> On the issue of transmitting RDF/XML in SOAP messages... I
> must admit it
> never occurred to me there was a problem, but I looked briefly at
>
> http://www-106.ibm.com/developerworks/webservices/library/ws-soaprdf/
>
> I think (but I'm not sure?) that all the resources described
> in the DCMI
> schemas (or at least the resources for which we'd expose descriptions
> through this interface) have URIs assigned to them and so don't use
> rdf:ID, so wouldn't that fit into the simple case illustrated
> by listing
> 6?
The provenance information for the translation relies heavily on rdf:ID. If
we do not offer the provenance info as a service then we avoid this problem,
but are still faced with the encoding problem (see below).
> In, the SOAP primer at
>
> http://www.w3.org/TR/2002/CR-soap12-part0-20021219/
>
> section 5.2 shows RDF/XML encoded in SOAP using an RDF-specific
> env:encodingStyle. Will something like that work for the DC
> descriptions? Or are there subtleties I'm not seeing? Quite probably!
I need to play around with this a bit, but I believe this will look, to the
client, like one big structure. I'm not sure that is particularly useful.
I think it would be more useful to deliver an XML document that conforms to
a published schema. Anyone else have any thought on this?
Regards,
Harry
|