RE: type test case

> > 2. Will this work with non-XML Schema datatypes?
> 
> I guess so

I'd like to see some proof of that. xsi:type is defined
by XML Schema. If RDF is going to use it for non XML Schema
datatypes, then is it not the case that RDF is extending
the semantics and usage of xsi:type beyond that defined
by XML Schema?!

I've got major alarm bells going off in my head about
this one -- not to mention a very bad taste in my
mouth.

XML Schema is not the only schema formalism for XML,
and even though both XML Schema and RDF are both 
"children" of the W3C, I don't see that they need to
be so tightly integrated in this fashion.

> > 3. Will this work with arbitrarily/user defined primitive types?
> 
> I guess so

Again, does that not mean that RDF is extending
the semantics and usage of xsi:type beyond that
defined by XML Schema?

> > 4. How does this relate to defining datatype ranges of properties?
> 
> well I guess global range constraints
> could be used to detect inconsistencies, no?

So, is it the case that

IF
   :s :p <DDD>"LLL" .
THEN
   <DDD>"LLL" rdf:type <DDD> .

???

That's not adding *anything* new to the functionality 
already provided by the stake-in-the-ground idiom for
local datatyping.

The WG agreed that the stake-in-the-ground would *not*
be changed unless there was strong technical motivation.

Where is it?! How does the datatype property idiom fail?!

Something doesn't smell quite right here...

> (too much guesses I guess)

This is what really bothers me. We have spent many months coming
to consensus on key aspects of datatyping, and IMO there remained
only one single issue left to decide, whether literals had tidy
or untidy syntax/semantics -- and now it seems like we're back
at square one and anything goes.

Was there *any* dissent in the WG about the local idiom and
its semantics? I don't recall any. What's the deal?!

Patrick

Received on Wednesday, 7 August 2002 07:10:52 UTC