Re: [xml-dev] XPath 2.0 Best Practice Issue: Graceful Degradation

[Philippe Poulard <philippe.poulard@sophia.inria.fr>]

noah_mendelsohn@us.ibm.com a écrit :
> Philippe Poulard writes:
> 
>> Schema technologies were primarily designed to express constraints 
>> on XML document classes.
> 
> To add a bit more detail, the following is the text of the section of the 
> Schema Recommendation that describes the purpose of a schema [1]
> 
> 
>> Unfortunately, schema technologies are still trailing behind. (One 
>> of) the missing feature(s) is the support of semantics data types.
> 
> See those last two sentences in the "purpose" section.   It's not at all 
> clear to me that Schema structures is the right place to go after these 
> semantic types,

Well, so far you added more details -interesting details- but, as you 
said, you are talking about W3C XML Schema ; my thoughts weren't 
restricted to a specific schema technology ; DTD, Relax NG and 
schematron are also concerned.

  except insofar as the existing ability to define named
> types, and inheritance hierarchies in which both the intensional 
> refinement of a type (I.e. it's base type chain) as well as its 
> extensional refinement (a restriction allows no more than its base) 
> matter.  To make that last bit clearer, what I'm trying to say is, you can 
> have two vacuous restrictions of integer, one named "employeeAge" and one 
> named "partNumber".  They both accept the same numbers, but you can't put 
> xsi:type="partNumber" on an element that's expecting an "employeeAge". 
> That's what I meant by intensional as well as extensional, and it takes 
> you just a bit of the way toward semantic typing.

True. Just a bit.
I'm upset that there is no more : in many languages you can deal with 
semantic data types, I complain that we have not the same support on XML 
technologies. The examples Roger and I were talking about in the thread 
are not so complex ; very simple solutions can be considered to enhance 
the type system in XML ; this is what I have experimented in the 
implementation of RefleX.

-- 
Cordialement,

               ///
              (. .)
  --------ooO--(_)--Ooo--------
|      Philippe Poulard       |
  -----------------------------
  http://reflex.gforge.inria.fr/
        Have the RefleX !