Re: Last Call Comment: Support Token-like behavior in @prefix

I see one snag.

At the moment, if there is a @rel="BLABLA" then the algorithm is:

1. first check whether BLABLA is a an NCNAME, then check whether it is a defined term via a @profile. If so, we are done
2. else, check if BLABLA abides to a CURIE syntax with a ':' and then proceed with a CURIE resolution.

What you propose is that #2 is a little bit modified by allowing for an NCNAME to be processed.

This is easy to implement. But the consequence is that if the term 'SOMETHING' is defined _both_ in a @profile _and_ an xmlns:SOMETHING="http://...", then the @profile version will prevail. This may be regarded as fairly unintuitive for HTML users. See, for example, CSS: if I have a CSS class defined in an external CSS file, and then I define the same class in an embedded CSS in the HTML file or as part of a @style attribute, the 'local' definition will prevail.

Ivan

On Jan 5, 2011, at 05:01 , Manu Sporny wrote:

> I just noticed that an issue concerning CURIEs hadn't been raised and I
> think that Mark meant to do so before the LC period ended for RDFa Core.
> We had a brief discussion about why one couldn't do this:
> 
> <div xmlns:Person="http://xmlns.com/foaf/0.1/Person"
>     xmlns:name="http://xmlns.com/foaf/0.1/name"
>     about="#me" typeof="Person" property="name">Frank</div>
> 
> I was convinced that this was not allowed in RDFa 1.0, but
> lo-and-behold, take the text above and plug it into Live Loop:
> 
> http://rdfa.digitalbazaar.com/live-loop/
> 
> and out come these triples:
> 
> <http://example.com/sample.html#me>
>   <http://www.w3.org/1999/02/22-rdf-syntax-ns#type>
>      <http://xmlns.com/foaf/0.1/Person> .
> <http://example.com/sample.html#me>
>   <http://xmlns.com/foaf/0.1/name>
>      "Frank" .
> 
> Now, clearly my parser is wrong because it violates the production rules
> for a CURIE, but we don't have a test in the test suite that checks this
> and it has gone all of this time without being detected.
> 
> This raises the question - if a fully conforming RDFa 1.0 processor can
> pass the test suite /and/ support the mechanism above, why don't we just
> relax the CURIE rules to allow reference-less CURIEs?
> 
> If we do that, why not allow something like this in RDFa 1.1:
> 
> prefix="Person: http://xmlns.com/foaf/0.1/Person
>        name: http://xmlns.com/foaf/0.1/name"
> 
> I don't believe we create any backwards incompatibilities, and this way,
> people that are concerned about not being able to specify RDFa Profile
> Terms without using an external document could use this method instead.
> 
> So, the LC request is to relax the rules for CURIE production to allow
> the mechanism described above.
> 
> -- manu
> 
> -- 
> Manu Sporny (skype: msporny, twitter: manusporny)
> President/CEO - Digital Bazaar, Inc.
> blog: Linked Data in JSON
> http://digitalbazaar.com/2010/10/30/json-ld/
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf