Re: Resolution for rdfms-fragments

Aaron Swartz wrote:
[...]
> > http://www.w3.org/2000/03/rdf-tracking/#rdfms-fragments

> > Can you point out which part of whatever spec bugs you?
> 
> [[[
> If a fragment identifier is included in the URI-reference then the resource
> identifier refers only to a subcomponent of the containing resource; this
> subcomponent is identifed by the corresponding anchor id internal to that
> containing resource and the extent of the subcomponent is defined by the
> fragment identifier in conjunction with the content type of the containing
> resource, otherwise the resource identifier refersto the entire item
> specified by the URI.
> ]]]
>  - http://www.w3.org/TR/1999/REC-rdf-syntax-19990222
>    Resource Description Framework (RDF) Model and Syntax Specification
> 
> By this definition, <http://www.w3.org/1999/02/22-rdf-syntax-ns#value> would
> either refer to an XPath node set (via the XPointer spec), or just
> <http://www.w3.org/1999/02/22-rdf-syntax-ns>.

OK, I can see why that text bugs you. I'm happy throwing
out all that text, just going forward with the model theory,
syntax, test cases, and primer docs. (with schema somewhere nearby).

> > Or give a use case that you think is insufficiently
> > specified? (i.e. not a foo/bar/baz example)
> 
> I don't know what http://www.w3.org/1999/02/22-rdf-syntax-ns#value means.

How does this relate to fragments?
Do you know what http://dm93.org/2002/frobnitz means?

Do you expec the RDF spec to tell you what it means?

I expect the RDF spec to tell me what sort of inferences
I can make when I use such a symbol in RDF documents;
but that's all the meaning I expect to get from the RDF
spec.


> > Or some piece of code that's acting funny,
> > or difficult to write or interoperate with?
> 
> Well, the entire installed base of HTTP servers would be a good place to
> start, seeing as they don't support URIs with fragments.

Sure they do, vacuuously: the fragment is stripped off before the
request,
so the server never sees it. There's no problem interoperating
between HTTP servers and RDF software whatsoever.

If there is, please explain.

> Similarly, systems
> like WebDAV or access control built on top of that don't support them.

Again, yes, they do, vacuously. (i.e. by not doing anything).
Fragment support is all in the clients.

> Add
> on to that everyone with a tool that conforms to the URI RFC.

Where is the interoperability difficulty? All the RDF
software I know of gets along just fine with all
sorts of URI-RFC-happy software.

>  Finally,
> their mapping to Resources isn't well defined, or defined at all depending
> on your point of view.

Their mapping to resources is just as defined as any other
sort of URI. Please re-read the model theory draft.

> Basically, RDF isn't compatible with the rest of the (non-W3C) Web.

I see repeated claims of that; I see no justification, however.
This is called "argument by assertion." Two can play
at that game: yes, RDF is compatible with the
rest of the Web, W3C or otherwise.

> I guess another solution would be to just rename RDF to be Random
> Description Framework or something, and not claim that the things it
> described were Resources.

I still don't see the problem for which this "solution" is needed.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 6 December 2001 22:20:18 UTC