Re: Datatype test cases/ datatype entailment/ datatype support.

On Fri, 2002-11-22 at 10:47, Jan Grant wrote:
> 
> The cross-datatype entailment test cases are reasonably clear cut, from
> a mathematical point of view, I think. But Patrick raises a good point
> when he asks (with a weather eye on implementation?) "what does it mean
> to say that datatype X is supported?"
> 
> 
> The intention with an entailment test case that's described as having
> datatype support for DTs X, Y and Z is this: that to be able to run the
> test case, you are able to decide:
> 
> 	for any two datatyped literals, a and b, with the datatypes
> 	of a and b coming from {X, Y, Z},
> 
> 	do a and b represent the same value?
> 
> (You don't need to be able to work out what that value is; just to
> answer that question.)

Yes; that's a very good specification;
please

(1) put it in the manifest, ala

<rdf:Description about="#datatypesEntailmentRules">
  <rdfs:comment>
An entailment test case that's described as having
datatype support for DTs X, Y and Z is this: that to be able to run the
test case, you are able to decide:
 
 	for any two datatyped literals, a and b, with the datatypes
 	of a and b coming from {X, Y, Z},
 
 	do a and b represent the same value?
 
 (You don't need to be able to work out what that value is; just to
 answer that question.)


  </rdfs:comment>
  <rdfs:seeAlso
rdf:resource="http://www.w3.org/TR/2002/WD-rdf-mt-20021112/#dtype_interp"/>
</rdf:Description>

(2) say that in the tests WD somewhere too, and

(3) point to it from the relevant tests; i.e. change

<test:entailmentRules
rdf:resource="http://www.w3.org/2000/10/rdf-tests/rdfcore/datatypes#" />

to

<test:entailmentRules
rdf:resource="http://www.w3.org/2000/10/rdf-tests/rdfcore/datatypes/Manifest#datatypesEntailmentRules" />



> Now, it might be better if the "datatype support" was recorded in the
> test case manifest using a cons list;

I don't think it's essential; i.e. there's no closed-world
assumption that needs to be made explicit. datatype entialment
is monotonic w.r.t. the number of datatypes you support;
i.e. if G1 datatypes(dt1,dt2)-entials G2, then we know
G1 datatypes(dt1,dt2,dt3)-entails G2.


> I'm willing to take advice on
> that, but that is what is meant by the manifest contents at this stage.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 22 November 2002 13:04:20 UTC