Re: RDFa Profiles, terms, and predicates (oh my!)

Hi Shane,

two things:

---------

The current text says:

[[[
Attempt to retrieve the content of the URI. If the retrieval fails, continue with the next URI in the value.
]]] (Section 9)

but also

[[[
If any referenced RDFa Profile is not available, then the current element and its children must not place any triples in the default graph.
]]] (Section 7.5)

I think the text in section 7.5 is what we accepted, but the text in section 9 seems to contradict that. In my view if, in section 9, attempt to retrieve the content of the URI fails, then an error condition occurs and the evaluation of the profile also stops.

-----

The issue you refer to in section 9 is moot now. In 4.2 we already say that host languages may define a default RDFa profile.

-----


Thanks!

Ivan

On Jul 23, 2010, at 19:45 , Shane McCarron wrote:

> I have tightened the conformance language in the current source version of the spec.  You can see it at http://www.w3.org/2010/02/rdfa/sources/rdfa-core/#s_profiles
> 
> 
> 
> On 7/23/2010 12:05 PM, Shane McCarron wrote:
>> Among your excellent points, you say:
>> 
>> On 7/23/2010 8:59 AM, Ivan Herman wrote:
>>> This model is nice because the target URI can refer to RDFa or to any other RDF formats, ie, instead of a recursive call it can also go and invoke any RDF parser that the underlying library provides.
>>> 
>> 
>> I think this is the key reason that relative URIs won't work.  While we REQUIRE that an RDFa Profile be defined in RDFa, that profile might have other formats and might return a JSON representation, for example, if the requestor indicated in the HTTP Accept header that format was okay.  In that situation, clearly we can't assume any special processing of the returned data.
>> 
>> On the other hand... since the server controls the content of that profile, obviously the server (or profile author) could ensure that the alternate representations were in their appropriate forms; this would include ensuring that a URI mapping was absolute.  That requirement is in my mind independent of whether we require that an RDFa Processor handle relative URIs in object literals associated with 'rdfa:uri'.  We control the horizontal and the vertical here. If we want to mandate special processing for a specific predicate, we can obviously do that.  We already do it for the RDF:XMLLiteral datatype for example.
>> 
>> Anyway... just because we can do a thing doesn't mean we should do a thing.  Requiring that the object literal associated with 'rdfa:uri' be an absolute URI seems the cleanest solution.  I again bow to your wisdom.  Thanks for clearing this up.  I will add text to Section 9 so the next idiot doesn't ask this again!
>> 
> 
> -- 
> Shane P. McCarron                          Phone: +1 763 786-8160 x120
> Managing Director                            Fax: +1 763 786-8180
> ApTest Minnesota                            Inet: shane@aptest.com
> 
> 


----
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