RE: datatypes status

> ... then even
> though it feels a bit icky, I could be persuaded to go with
> xsi:type, ...

I realized I should be a bit clearer about what I mean
by "icky" here:

1. xsi:type introduces qnames into RDF/XML
2. xsi:type has rdf:type semantics but isn't rdf:type
   for reasons having nothing to do with RDF
3. xsi:type implies an XML Schema datatype, even
   if any arbitrary type might be specified

As for the last point, I'm presuming that 

(a) it is in fact OK to specify any datatype whatsoever
    as the value of xsi:type, even one that violates
    constraints defined by the XML Schema specification

(b) the XML Schema folks are OK with us borrowing their
    term as an official term in the RDF vocabulary

If either of the above turn out to not be true, then that
precludes our use of xsi:type.

Patrick