Re: RDFa for Firefox

Karl, Sergio,

I think it's important to distinguish between the specification that
works for XHTML1.1, and the ways in which RDFa can begin to be used in
other version of HTML.

According to the specification, there's only RDFa in XHTML1.1, which
means you should have the right mime type, DTD, and DOCTYPE.

However, because we've defined a DOM-based approach to parsing RDFa,
it's trivial to use it within HTML. Sure, it won't validate, though it
is conformant since it's just extra attributes. In fact, a number of
existing implementations don't check for DTD or XHTML1.1 or mime type,
they just go looking for the RDFa attributes. That's the right approach,
as the latest version of RDFa makes it quasi-impossible for authors to
accidentally declare triples.

Experimenting with RDFa in HTML other than XHTML1.1 is exactly what we
need to begin to understand how this might be integrated into HTML5.

Also, the RDFa profile is *not* required by our specification, though it
can be used if you want to be specific. XHTML2.0 includes RDFa in its
specification, so all XHTML2.0 should be parsed for RDFa.

-Ben


Karl Dubost wrote:
> 
> 
> 
> Shane McCarron (8 oct. 2007 - 21:30) :
>> If it were me, I would trigger on the DOCTYPE - there is only one
>> legal, valid way
> 
> s/legal//. Nothing legal.
> 
>> to deliver RDFa today - XHTML+RDFa, which is a specialization of XHTML
>> 1.1.
> 
> And sent as application/xhtml+xml, which means given [browsers market
> share][1] working in 22% browsers only (September 2007)
> 
>> We do have a profile value, and it might be possible to do something
>> with that as well.
> 
> What is the plan of XHTML 2.0 and profile attributes? It seems to [be
> dropped][2]. It has also been dropped from [HTML 5][3].
> 
> btw, what's happening if an RDFa document (XHTML 1.1) is sent over HTTP
> with a wrong mime-type (aka text/html)?
> 
> 
> 
> [1]: http://marketshare.hitslink.com/report.aspx?qprid=0
> [2]: http://www.w3.org/TR/xhtml2/attributes.html
> [3]: http://www.w3.org/html/wg/html5/
>

Received on Tuesday, 9 October 2007 16:06:57 UTC