On Mon, 5 May 2003 22:19:05 +0100
Andy Powell <[log in to unmask]> wrote:

> I've finally got round to looking back thru the last set of comments on
> this working draft and have put together a new version.
>
>   Expressing Qualified Dublin Core in HTML/XHTML meta elements
>   http://www.ukoln.ac.uk/metadata/dcmi/dcq-html/
>
> I hope I've taken all the relevant comments on board - plus some extra
> ones! ;-)
>
> The major changes are
>
> - use of 'dc:' and 'dcterms:' rather then 'DC.' and 'DCTERMS.' as prefixes
>   for DCMI property names
> - addition of 'dcterms:' prefix to encoding scheme names
> - use of <link> tag for properties with a value that is a URL for another
>   resource
>
> To summarise, this means that
>
> <meta name="dcterms:modified"
>       scheme="dcterms:W3CDTF"
>       content="2003-05-05" />
> <link rel="dc:relation"
>       href="http://www.ukoln.ac.uk" />
>
> is the recommended form, rather than
>
> <meta name="DC.date.modified"
>       scheme="W3CDTF"
>       content="2003-05-05" />
> <meta name="DC.relation"
>       content="http://www.ukoln.ac.uk" />
>
> though both these forms (and other variations of case) are explicitly
> stated as continuing to be acceptable (but not recommended).  Therefore,
> there is no problem with backwards compatability with this draft -
> documents with embedded metadata that conform to previous
> recommendations will continue to be conformant.
>
> I guess these are pretty major changes, and that not everyone will be
> happy with them!?  However, I wanted to see what the document looked like
> using the colon-separated form thoughout. My gut feeling is that it looks
> much better and is much more intuitive for those people who are working
> across the RDF/XML, XML and XHTML encoding syntaxes.
>
> What do other people think?

Changing from DC. to DC: is a bad idea.

The main reason is that it makes people think that the
'dcterms:modified' things inside HTML are XML Qualified Names (Qnames).

This will be
  - confusing since no part of HTML or XHTML uses QNames yet

  - suggest that these QNames work like XSLT, XML Schema, etc. - they
    do not and are just strings here.

  - QNames in attribute values are tricky from web architecture point
    of view and introducing them without careful thought is hard
      Using Qualified Names (QNames) as Identifiers in Content
      http://www.w3.org/2001/tag/doc/qnameids.html

  - not necessary, dc. works just fine and is deployed

In particular this will likely be conflicting with some emerging work
to see how to better embed metadata inside HTML & XHTML that has been
going on at the W3C between the W3C and RDF groups.  This is looking
at how to ship things encoded in the RDF family (RDF, OWL, CC/PP, DC,
FOAF, CC, ...) inside HTML.  Dublin Core is a very important
vocabulary for RDF and I want to make sure it doesn't take a
conflicting path.

This work has been emerging quietly in the last few months but hasn't
yet got anything public to show.  I'll give some pointers to earlier
discussion[*]

  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Feb/0103.html

which shows a POTENTIAL approach rewriting the RDF example I gave in
the RDF/XML syntax specification:

<rdf:Description rdf:about="http://www.w3.org/TR/rdf-syntax-grammar">
  <ex:editor>
    <rdf:Description>
      <ex:homePage>
        <rdf:Description rdf:about="http://purl.org/net/dajobe/">
        </rdf:Description>
      </ex:homePage>
      <ex:fullName>Dave Beckett</ex:fullName>
    </rdf:Description>
  </ex:editor>
  <dc:title>RDF/XML Syntax Specification (Revised)</dc:title>
</rdf:Description>

into a 'meta' form, using qnames.

<meta rdf:about="http://www.w3.org/TR/rdf-syntax-grammar">
    <meta name="dc:title">RDF/XML Syntax Specification (Revised)</meta>
    <meta name="ex:editor">
        <meta name="ex:fullName">Dave Beckett</meta>
        <meta name="ex:homePage">http://purl.org/net/dajobe/>
    </meta>
</meta>


This immediately conflicts with the DCQ-HTML approach since you can't
tell which one you are dealing with and will stop these approaches
(strawman approach) being interoperable.

I'd prefer you stick with 'DC.' (or 'dc.' if you must change)

Cheers :)

Dave

[*] Pointers to other and ongoing RDF and HTML work is in the ESW wiki:
  http://esw.w3.org/topic/EmbeddingRDFinHTML