Re: RDFa Lite and non-RDFa @rel values

Hi Stéphane,

First of all, just to make an admin issue clear: any modification here affect the HTML5+RDFa document, and _not_ the RDFa Lite document. Ie, what we discuss here does not jeopardize the PR transition of RDFa 1.1.


On Apr 24, 2012, at 04:49 , Stéphane Corlosquet wrote:

> (by RDFa Lite, I really mean the new behavior of the property attribute)
> 
> It is common for web developers to add a rel attribute in 'a' elements, e.g. rel="nofollow" or rel="author". According to the current processing steps of RDFa 1.1, if this kind of @rel is used in an element containing a property attribute (such as an RDFa Lite snippet), it can impact the processing and change the triples which are generated. I'd like to explore ways to avoid this in HTML+RDFa.
> 
> First of all let's start with an example:
> <p vocab="http://schema.org/" typeof="Person">
>    My name is <a href="http://example.com/" property="homepage">Stéphane Corlosquet</a>.
> </p>
> 
> which generates:
> <> a schema:Person;
>    schema:homepage <http://example.com/>.
> 
> Add a @rel to the mix:
> <p vocab="http://schema.org/" typeof="Person">
>    My name is <a href="http://example.com/" property="homepage" rel="nofollow">Stéphane Corlosquet</a>.
> </p>
> 
> and @property no longer targets the @href but the innerText instead:
> <> a schema:Person;
>    schema:homepage "Stephane Corlosquet";
>    schema:nofollow <http://example.com/>.
> 
> 
> Option #1: ignore @rel if it only includes HTML Link types as defined at http://www.w3.org/TR/html5-author/links.html#linkTypes
> 
> Option #2: ignore @rel if it does not include any explicit CURIE. I think it's safe to assume that the intentional use of RDFa terms in @rel will be limited in comparison to the use of RDFa Lite and the new @property behavior. Therefore requiring CURIEs to be used in @rel would allow us to detect whether @rel is used in an RDFa context or not, and from there decide whether it should be ignored or not.
> 

My vote goes for option #1. 

- HTML5+RDFa has a very clear attitude towards the 'predefined' @rel/@rev values: apart from some specific values, that are in the initial context, the predefined values are ignored (as opposed to XHMLT1). I think option #1 is perfectly consistent with this view and, as the example above shows, is also necessary to avoid problems.

- I think option #2 would be pretty much in a strong contradiction with RDFa 1.1 Core, and would impose a change in the processing steps that are way beyond what I would believe is acceptable. Indeed, what this means is that @vocab would have no effect on @rel/@rev. That is way too strong for me.

> In both options, it is important to emphasize the "ignore", as an empty @rel would still have the same effect of switching the behavior of @property to the old way.

Indeed.

Ivan

> 
> thoughts?
> 
> Steph.


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