Re: [Fwd: Re: [dquan@mit.edu: jena evaluation]]

Rereading this having forwarded it - I'm not sure my comments on the predicate
qname/uri question were very clear or accurate ...

What I was trying to say was that predicates in RDF have to be expressible as
qname pairs and that jena gives you the option of just giving a single URI
string and relying on an internal split algorithm to guess the
namespace/localname division. This split algorithm does have some known defects.
However, in any case this is just an option - the predicate constructors also
allow you to specify an explict namespace/localname pair and in fact it is only
this version that is exposed through the normal API viz:
   model.createProperty(namespace, localname)

I believe any legal RDF predicate name can be expressed through this interface
you should only get error messages if the namespace is not a legal URI or the
localname component is empty.

Dave

Received on Friday, 19 July 2002 05:47:46 UTC