W3C home > Mailing lists > Public > public-rdf-in-xhtml-tf@w3.org > March 2004

review of "XHTML and RDF"

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Wed, 10 Mar 2004 17:05:22 +0100
To: www-html-editor@w3.org, public-rdf-in-xhtml-tf@w3.org
Message-Id: <200403101705.22526.jjc@hpl.hp.com>


This is a review of XHTML and RDF, valentine's day 2004

http://www.w3.org/MarkUp/2004/02/xhtml-rdf

concentrating on technical issues rather than presentational.

1. I like this design very much, the technical issues should be sent as minor 
nits.

2. The treatment of same doc references is confused.

Background:

A same doc reference is a URIref of the form "" or "#foo".
While browsing this should not cause a further retrieval.

RDF Concepts requires the use of absolute URIrefs (and hence excludes same doc 
refs from the RDF graph. Various implementations, such as cwm and N3 ignore 
this limitation). 

RDF/XML Revised shows how same doc references are turned into absolute URIs 
using the inscope base URL (defaulting to the document retrieval URL)

Comment:

Since the document URL "" or <> is central to this document, at some point it 
needs to be clarified that that URL is either the retrieval URL or that given 
by a base attribute.

3.  the default namespace.

The editorial note at the end of section 3.1 seems incorrect.

A bare property "author" should either be seen as an error (my preference) or 
sitting in the default namespace, which I take to be the XHTML 2.0 namespace

The latter would suggest that "author" would be using an undefined concept, 
however it would be uniform over a site. With such a reading no change would 
be needed to "next" and "stylesheet" (see editorial note in section 3.4).

My preference however would be to require the xhtml2 namespace to be made 
explicit.

4) about used with both qnames and urirefs

e.g. about="p:TonyBlair" and about="#q1"

both in 3.3

This could be permitted by effectively given 'about' a union type of qname 
union uriref, with union understood with an order as in XML Schema datatypes. 
However, when I have proposed that (in other contexts) I have typically 
encountered a vomit factor from others. 

Options:
  choose URIrefs or qnames for about
  have two attributes one with qname, one with about
  if namespace in scope then qname, else uriref

A decision should be made

5) not all RDF graphs can be represented

e.g.

_:a <eg:prop> _:a .

This is not a significant weakness but can be used to help motivate also 
supporting GRDDL.

6) possibility of using XMLLiteral to include markup from document body in the 
RDF graph (expanded in a follow up note - I need to go in the next few 
minutes)

7) It is not clear to me what the intended schema validation technology is.
With DTDs it is not possible (if I understand correctly) to permit an unknown 
set of namespace declarations. Allowing such seems necessary

e.g.

<html xmlns:vocab1="..."
        xmlns:vocab2="..."
       ...
     >
to declare the various semantic vocabs being used.

Personally, I am happy with the answer that XHTML2 depends on relax and relax 
permits this.

8) default namespaces

  It may be a good idea to incorporate dc by default, since this is the most 
widely applicable metadata vocabulary for HTML documents



(More on XMLLiteral later)

Jeremy




 
Received on Wednesday, 10 March 2004 11:06:57 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:14:59 GMT