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

Hi Ivan,

I also prefer a new attribute.

The only thing I feel really strongly about is that we don't use @role
for rdf:type, since I think that will come back to bite us in the
future. So for me, that leaves two choices, use @class or use a new

I can live with using @class, but I do agree that it comes with some
baggage. I don't mean that from the point of view of some kind of
'backlash', since I think people are using @class semantically already
(even without using microformats). What I mean is that I can easily
imagine people forgetting to put foaf:Person (for example) in the
class attribute on the containing element, since novice authors would
probably see it as 'I must set the CSS class to foaf:Person for this
to work'.

I also believe that rdf:type is so important that it should be part of
the core RDFa attributes, that are independent of any host language. A
host language may have an additional way of doing this, and we might
even decide in the future to use @class in HTML after all. But by
having our own attribute, it means that there is a core way of marking
up rdf:type that is always there, no matter what language is the host.

So to summarise; my preferred approach would be to leave @class
undefined for now--we can always come back to this in a future
version--and use @isA or something like that, to indicate the rdf:type
of something.



On 28/06/07, Ivan Herman <> wrote:
> I am a bit uneasy with the usage of @class. _I know_ that the formal
> semantics of @class allows this type of usage, and I also know that the
> microformat community uses that trick, I still feel that usage @class is
> putting a semantics into the attribute that a lambda user would not
> expect. (And yes, I am also uneasy with the way the microformats reuse
> attributes like title, class, or abbr...).
> I would prefer to use a dedicated attribute if we need it (or simply
> stick to the rel="rdf:type", which is at disposal anyway).
> I must admit I am not fully familiar with the discussion behind @role to
> decide whether @role should be introduced in RDFa for XTHML1, too, to
> cover this usage, or whether a different @type or similar should be
> introduced. I guess this discussion should start _if_ ISSUE-3 is not
> resolved for @class
> Ivan
> P.S. Having said all that: I do not consider this issue as life
> threatening:-) Ie, resolving it quickly is probably more important than
> spending lots of time finding the best solution.
> Ben Adida wrote:
> >
> > Another issue up for discussion.
> >
> > ISSUE-3
> >
> >
> > The question is what @class and @role should yield in XHTML1.1+RDFa. We
> > don't have complete consensus on this (we specifically note Steven
> > Pemberton's worries about the reuse of the @class attribute), but the
> > current solution, as accepted in the Primer and in many use cases, is as
> > follows:
> >
> > @class yields rdf:type only if the value is namespace-qualified. @class
> > contains a space-separated list of values. Only those values which are
> > namespace-qualified yield rdf:type triples.
> >
> > @role does not exist in XHTML1.1, so it is not used here. In XHTML2, it
> > is expected to yield a triple with predicate xhtml2:role.
> >
> > +1 if you agree, otherwise email your disagreements and explanation.
> >
> > -Ben
> >
> --
> Ivan Herman, W3C Semantic Web Activity Lead
> URL:
> PGP Key:

  Mark Birbeck, formsPlayer | +44 (0) 20 7689 9232 |

  standards. innovation.

Received on Thursday, 28 June 2007 10:02:22 UTC