RE: Describing the "nature" of a resource

Hello Norm,

I think the assertions that tried out below are assertions about nature
of some representations, rather than assertions about resources (except
indirectly in cases where the resource has a single representation that
is either invariant over time or the assertions are maintained in sync
with changes).

Stuart
--
Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks
RG12 1HN
Registered No: 690597 England

> -----Original Message-----
> From: www-tag-request@w3.org [mailto:www-tag-request@w3.org] 
> On Behalf Of Norman Walsh
> Sent: 16 May 2007 21:38
> To: www-tag@w3.org
> Subject: Describing the "nature" of a resource
> 
> At the last TAG f2f, we discussed namespaceDocument-8. We're 
> going to do it again at our next f2f and I'd really like to 
> make progress.
> 
> As I understood discussion[1] at our last f2f, one of the 
> most significant issues with our current draft[2] was raised 
> by Dan. Dan expressed skepticism over the way we'd proposed 
> to model natures.
> 
> (Note that we've introduced a conceptual model that doesn't 
> have to bear a 1:1 correspondence with any particular RDDL 
> syntax so this discussion is, at least for the moment, about 
> the model and not the particular URIs used by RDDL.)
> 
> The current draft says that natures have URIs and we identify 
> the nature of a resource with those URIs.
> 
> For example:
> 
> <http://docbook.org/xml/5.0b1/rng/docbook.rng>
>    assoc:nature <http://relaxng.org/ns/structure/1.0> .
> 
> asserts that the nature of the docbook.rng file is a RELAX NG grammar.
> 
> The concern raised was that such a statement is a statement 
> of opinion and not of fact. This is even easier to see in 
> cases like HTML 4 where we assert the nature of a resource by 
> pointing to its normative specification. It's not hard to 
> imagine the existence of formats for which the normative 
> specification is *clearly* a matter of opinion.
> 
> Dan suggested instead that we should ground natures in fact. 
> For example,
> 
> <http://docbook.org/xml/5.0b1/rng/docbook.rng>
>    xxx:docRootEltName ("http://relaxng.org/ns/structure/1.0" 
> "grammar") .
> 
> This "docRootEltName" property points to a list which 
> identifies the namespace name and local name of the root 
> element. This clearly is a matter of fact, not opinion.
> 
> Personally, I'm just as happy with assertions in this case, 
> but I did agree to attempt to find facts we could use. I've 
> taken the current list of natures and considered how we might 
> ground them in fact:
> 
> CSS                     content-type    text/css
> DTD                     content-type    application/xml-dtd
> Mailbox                 ???
> Generic HTML            content-type    text/html
> HTML 4                  ???
> HTML 4 Strict           ???
> HTML 4 Transitional     ???
> HTML 4 Frameset         ???
> XHTML                   docRootEltName  xhtml:html
> XHTML 1.0 Strict        ???
> XHTML 1.0 Transitional  ???
> RDF Schema              ???
> RELAX NG Schema (RNC)   content-type    
> application/relax-ng-compact-syntax
> RELAX NG Schema         docRootEltName  rng:grammar
> Schematron Schema       ???
> OASIS Open Catalog      ???
> XML Catalog             docRootEltName  cat:catalog
> XML Scheam              docRootEltName  xs:schema
> XML Character Data      ???
> XML Escaped             ???
> XML Unparsed Entity     ???
> IETF RFC                ???
> ISO Standard            ???
> 
> For some, we can rely on the content-type, I think. And for 
> others, the document root element name. But for many, I don't 
> see an obvious answer and so I'm not sure how to proceed.
> 
> Suggestions most welcome.
> 
>                                         Be seeing you,
>                                           norm
> 
> [1] http://www.w3.org/2001/tag/2007/03/07-morning-minutes
> [2] http://www.w3.org/2001/tag/doc/nsDocuments/
> --
> Norman Walsh <ndw@nwalsh.com> | The perfect man has no method; or
> http://nwalsh.com/            | rather the best of methods, which is
>                               | the method of no-method.-- Shih-T'ao
>

Received on Thursday, 17 May 2007 09:53:02 UTC