Re: ISSUE-21: It will be hard to tell where a prefix comes from

On Thu, 2010-04-08 at 14:15 +0000, RDFa Working Group Issue Tracker
wrote:
> ISSUE-21: It will be hard to tell where a prefix comes from
> 
> http://www.w3.org/2010/02/rdfa/track/issues/21

Further, for:

<div profile="http://example.com/p" typeof="foaf:Person">
  <h1 property="foaf:name">Joe Bloggs</h1>
  <p property="http://example.com/foo">Bar</p>
</div>

If the profile defines "foaf" to point to the usual prefix, then it
should result in the following triples:

 _:x a <http://xmlns.com/foaf/0.1/Person> .
 _:x <http://xmlns.com/foaf/0.1/name> "Joe Bloggs" .
 _:x <http://example.com/foo> "Bar" .

However, if the profile document returns a 404, then it results in:

 _:x a <foaf:Person> .
 _:x <foaf:name> "Joe Bloggs" .
 _:x <http://example.com/foo> "Bar" .

i.e. the prefix "foaf:" is treated as a URI scheme, not a CURIE prefix.

Using profiles to define plain terms (not prefixes):

<div profile="http://example.com/p" typeof="Person">
  <h1 property="name">Joe Bloggs</h1>
  <p property="http://example.com/foo">Bar</p>
</div>

If the profile document is unavailable, this generation of weird triples
doesn't happen - the typeof=Person and property=name triples simply
disappear and you end up with a simple subset of the intended graph:

 _:x <http://example.com/foo> "Bar" .

So the wider issue is that if a profile document disappears it's
impossible to even distinguish between a non-safe CURIE and a full URI.
So allowing CURIE prefixes to be defined by profiles threatens our
ability to permit full URIs in the previously CURIE-only attributes.

-- 
Toby A Inkster
<mailto:mail@tobyinkster.co.uk>
<http://tobyinkster.co.uk>

Received on Thursday, 8 April 2010 18:11:09 UTC