Re: RDFa in HTML 4

Toby A Inkster wrote:
> 
> My suggestion, and bear in mind that I haven't implemented this yet, is 
> that we allow something like:
> 
>     <link about="#node-wttj" rel="schema.dct"
>         href="http://purl.org/dc/terms/">
>     <link about="#node-wttj" rel="schema.audio"
>         href="http://purl.org/media/audio#">
>     [...]
>     <div id="node-wttj"
>          about="#welcome-to-the-jungle" typeof="audio:Recording">
>        <span property="dcterms:title">Welcome to the Jungle</span>
>     </div>

Hi Toby,

Thanks for chiming in on this. I do want to point out that, as Mark 
mentioned, we looked at this quite thoroughly.

The above solution doesn't work for, e.g., Creative Commons: we have to 
hand out *one* chunk of HTML that is inserted somewhere in the body of 
the page and that is self-contained. This self-containment requirement 
is more and more important in a widgetized web world, where a web 
producer often doesn't control the whole page anymore.

In the attempt to fix this, you point out:

 > The limitation here to fit (largely) into the existing RDFa parsing
 > pattern is that the element which defines a CURIE prefix must be
 > encountered prior to the element which uses the CURIE prefix.

The resulting solution wouldn't work in a straight DOM implementation 
anymore, as you'd have to inform one recursion branch with another... 
which I think indicates an incorrect design. I think it would be easier 
to retrofit xmlns into HTML4 (and I'm not saying that would be easy.)

Note that we have to be conscious of where we are: RDFa for XHTML 1.1 is 
in CR, and the comments we received in Last Call have been quite 
positive, with adoption from Yahoo, Digg, and Creative Commons. I don't 
see how a radical change in the syntax for HTML4/5 could be in the cards 
at this point, given that many authors will want to produce chunks of 
HTML that work in both.

-Ben

Received on Thursday, 17 July 2008 22:57:21 UTC