Re: [RDFa] ISSUE-3 @class and @role for rdf:type

Hi Ivan,

> :-)
>
> There is a difference, though. @about has been introduced for RDFa.
> However, (and I may be wrong with that!) @role was introduced with some
> general, not-necessarily-RDF usage in XHTML2 that RDFa _may_ reuse.
> And, from the RDFa side, there is no way to predict what the evolution
> of @role will be in future versions of XHTML2...

I'm not sure where we're heading with this. :) But perhaps I can just
clarify a few things about @role, which will show why it was on our
radar before (but why it probably no longer needs to be).

First, it's actually a separate module. It was originally distinct
from XHTML 2, then it was included into XHTML 2, and then it was
broken back out again. And as Steven said, it's already implemented in
Firefox. So HTML and XHTML documents will already include @role,
whether we like it or not, which is the first reason it is legitimate
to consider it within RDFa.

Also, @role has a very explicit relationship to RDF...that's the whole
point of it. @role values are URIs, which give you a 'hook' to go and
get more information if you want to. Which means that it's also
legitimate to ask whether the mechanism for establishing that hook is
via RDFa, or by some other means.

Finally, the whole rdf:type debate; in the past those using @role did
define it to be the same as rdf:type, but that was changed. And of
course, in our own discussions there was a period where @role meant
rdf:type, but again, that has been changed.

So...given that no-one is using @role any longer, to represent
rdf:type, I think it would be quite legitimate to defer what RDFa says
about @role until the future. We will need to say something about at
some point though, since it will be increasingly used, but that's no
more than saying we have to say something about <meta> and <link> in
HTML now.

Hopefully you can see that none of the reasons for talking about @role
are to do with some of us trying to 'sneak' XHTML 2 attributes into
RDFa by the back door. :) In fact, speaking for myself, I've been
saying the opposite for a while now--that in the version of RDFa we
are currently finalising, we should restrict ourselves as much as
possible to only using HTML 'as we find it', and then have a
smattering of extra attributes for the more complex constructs.

Regards,

Mark

-- 
  Mark Birbeck, formsPlayer

  mark.birbeck@x-port.net | +44 (0) 20 7689 9232
  http://www.formsPlayer.com | http://internet-apps.blogspot.com

  standards. innovation.

Received on Thursday, 5 July 2007 11:07:10 UTC