Re: RDFa test suite addition

Toby,

I get what you are saying here... but you scared me with something.  
What is Cognition learning about a "namespace" in the context of RDFa?  
It should not be learning anything?  There are no namespaces in RDFa.  
There are prefix values and suffixes.  And the concatenation of these 
creates URIs (IRIs) that, when dereferenced, could mean something.  In 
my opinion Cognition could be learning about those URIs... but the 
prefix value is meaningless.  Hell, the prefix value could be "http".  
It could be "h".  It would be completely legitimate.

What am I missing here?

Toby A Inkster wrote:
>
> The current behaviour of Cognition is that it will happily deal with 
> relative URIs - including normalisation of ".." components - for the 
> subject and object of a triple.
>
> But for the predicate, no normalisation currently takes place. In 
> fact, I think that using something like 
> xmlns:ex="http://example.org/ns/" rel="ex:../foo" is likely to cause 
> problems all over the place - e.g. if outputting RDF/XML, Cognition 
> would get confused, because the predicate http://example.org/foo would 
> be outside any defined namespaces.
>

-- 
Shane P. McCarron                          Phone: +1 763 786-8160 x120
Managing Director                            Fax: +1 763 786-8180
ApTest Minnesota                            Inet: shane@aptest.com

Received on Tuesday, 29 July 2008 17:33:40 UTC