Re: URIs, names and well known RDDL names, was: Re: Quick edit

[ht@cogsci.ed.ac.uk (Henry S. Thompson)]

Jason Diamond <jason@injektilo.org> writes:

> Hi.
> 
> > 	Right, emphasis on "whenever possible" and realize that
> > http://www.w3.org/2000/10/XMLSchema may work for XSD under most conditions
> > but we need to retain the flexibility to support, for example, multiple
> > schemas on the same namespace URI. For example, suppose we have one schema
> > for  editing and another for runtime validation. Each may need a different
> > arcrole URI (or given a constant URI some other mechanism to distinguish
> > between related resources).
> 
> Perhaps a more likely scenario would be multiple CSS/XSL stylesheets. In my
> trial implementation of RDDL, I thought that a method returning the resource
> with a specified arcrole would be useful. Since there's no restriction on
> how many resources in a given RDDL directory can use the same arcrole,
> however, I realized that that method would be returning some arbitrary
> resource (in my case, the first one it found with a matching arcrole). How
> would we be able to programmatically differentiate between the two XSD
> resources in your example?
> 
> I agree with Henry that we should not be defining new names for types that
> already have them. Right now xlink:role is required to be
> http://www.rddl.org/#resource. It's obviously a resource--that's what RDDL
> is designed to describe. I propose that we use xlink:role to describe the
> type of resource being referenced. For a RDDL document containing multiple
> schemas, the xlink:role could be http://www.w3.org/2000/10/XMLSchema. The
> xlink:arcrole attributes should then describe the context in which those
> resources (XSDs) should be used. They could be
> http://www.rddl.org/arcrole.htm#edit or
> http://www.rddl.org/arcrole.htm#validate or whatever URI the directory
> maintainer felt was appropriate.
> 
> XLink says this about arcrole: 'For example, a resource might generically
> represent a "person," but in the context of a particular arc it might have
> the role of "mother" and in the context of a different arc it might have the
> role of "daughter."'
> 
> (I'm actually unclear on how xlink:role is supposed to be used. XLink says:
> 'the role attribute indicates a property that the resource has'. Can we use
> it to indicate the type a resource is? I don't see anything preventing
> this.)
> 
> In your example, both resources are XML Schemas. One of those links is used
> in the context of editing an instance document. The other is used for
> runtime validation. They're type didn't change--just how they're used. Right
> now, the resources in arcrole.htm define types. I think it should be an
> error for multiple resources in the same directory to have the same
> arcroles. But, it should obviously be permissable for multiple resources to
> have the same types.
> 
> I'm not entirely certain that this is what RDDL is supposed to do, though.
> The beauty of it was that it was simple as well as useful. Would anybody
> care to draft a list of requirements that RDDL should adheed to for a first
> release? To be honest, the utility it's supposed to provide keeps drifting
> in and out of focus for me. If we had a succint set of goals, we might be
> able to avoid ambiguities like this.

I strongly agree with Jason's analysis here.  I certainly want _some_
attribute of rddl:resource to _always_ be
http://www.w3.org/2000/10/XMLSchema if what's pointed to is in fact an
XML Schema!

ht
-- 
  Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh
          W3C Fellow 1999--2001, part-time member of W3C Team
     2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
	    Fax: (44) 131 650-4587, e-mail: ht@cogsci.ed.ac.uk
		     URL: http://www.ltg.ed.ac.uk/~ht/