- From: Gregg Kellogg <gregg@greggkellogg.net>
- Date: Mon, 14 May 2012 15:00:01 -0400
- To: Toby Inkster <tai@g5n.co.uk>
- CC: W3C RDFWA WG <public-rdfa-wg@w3.org>
On May 14, 2012, at 11:54 AM, Gregg Kellogg wrote:
> I'll add a spec ref, but @property is a TERMorCURIEorAbsIRI, which requires that it resolve to an Absolute IRI. For _:invalid to be an absolute IRI, "_" would need to be a valid IRI scheme, which it's not. A scheme is an NCName.
Actually, it's probably not clear enough that TERMorCURIEorAbsIRI should resolve to an absolute IRI. My implementation checks to be sure that everything does, because the content model includes AbsIRI. I believe this is the intent, but perhaps this should be clarified someplace.
As you note, a BNode is a CURIE, and could be considered to be valid. Of course, the choice of output serializations will probably limit you here, as I'm not aware of any in which a BNode can be used in the predicate position, even N-Triples.
Gregg
> This should probably reference 7.4.4.
>
> Gregg
>
> On May 14, 2012, at 3:36 AM, Toby Inkster wrote:
>
>> Is there a spec reference for this?
>>
>> Seems to me that:
>>
>> property="_:invalid"
>>
>> conforms to CURIE syntax. Obviously blank nodes are not allowed as
>> predicates in any current version of RDF. However it is a syntactically
>> valid CURIE, so arguably should be treated the same as any other CURIE -
>> there will be no expansion for the "_" prefix, so it's treated as an
>> absolute URI, equivalent to the following token in Turtle:
>>
>> <_:invalid>
>>
>> Unless there's a spec reference I'd suggest either dropping this test
>> case, or changing the SPARQL to:
>>
>> ASK WHERE {
>> <$TCPATH/0140.xhtml> ?p "Test".
>> FILTER { isBlank(?p) }
>> }
>>
>> Until I see a spec ref or the test is dropped/altered, it's going on
>> my "wilful violations" list.
>>
>> --
>> Toby A Inkster
>> <mailto:mail@tobyinkster.co.uk>
>> <http://tobyinkster.co.uk>
>>
>>
>
Received on Monday, 14 May 2012 19:01:10 UTC