RE: QName Problem Isn't One

This discussion has become hopelessely confused. When such is the case it
ought be a strong signal that problems exist, and people should be very
hesistant to design systems with so much ambiguity.

I am sympathetic to the goal of using XML Schema datatypes in RDF, indeed
using the simplified/strawman syntax that I and others have proposed (see
http://www.openhealth.org/RDF/rdf_Syntax_and_Names.htm) several of the
'problems' with the DAML proposal to use XML Schema datatypes would vanish.
So my criticism is not so much of the DAML proposal, as pointing out the
need for the 'new RDF' to adopt these proposed changes (specifically insert
a '#' between namespace URI and localname when namespace URI ends in an
alphanumeric character, and assume default parseType="Resource", or allow
this to be specified)

A few points:

1) QNames and URIs are not the same thing. XML Schema defines datatypes by
_QName_ not by URI and although certain builtin datatypes have assigned
URIs:

2) Is is not currently possible to assign a URI to a general XML Schema
datatype ... by this I mean one without a specific id (ids are entirely
optional).

3) The interface between RDF and XML is somewhat broken in that although XML
is billed as a serialization format for an RDF 'model' it is not possible to
assign a QName to an arbitrary URI. (so we see that only in certain cases is
it possible to assign a URI to a QName and similarly a QName to a URI).

4) If you aren't going to use the XML Schema QNames to refer to datatypes,
the so called 'integration' is meaningless. Who really cares if the XML
Schema URIs are the same if the namespace is different. All XML software
works on namespace URIs (e.g. XSLT, SAX etc) and if you are going to change
the namespace URI even by a single character, you mind as well define unique
RDF datatypes. What is the benefit of using XML Schema URIs? Is this benefit
realized when the namespace is different? What software do people expect
will be able to work with this?

5) If you insist on using a new namespace _DON'T_ use the "xsd" prefix, this
is totally confusing to people who assume that "xsd" is bound to
"http://www.w3.org/2001/XMLSchema", if you would simply bind the namespace
"http://www.w3.org/2001/XMLSchema#" to the prefix "fubar" and refer to this
as <fubar:decimal rdf:value="35.2"/> then there would be much less
confusion.

6) I am not sure that the usage of

<Person rdf:ID="Adam">
  <rdfs:label>Adam</rdfs:label>
  <rdfs:comment>Adam is a person.</rdfs:comment>
  <age>
  	<xsd:integer rdf:value="13" />
  </age>
  <shoesize>
  	<xsd:decimal rdf:value="9.5" />
  </shoesize>
</Person>

in the DAML example (http://www.daml.org/2001/03/daml+oil-ex.daml) is valid
RDF ... the "age" and "shoesize" properties should have either an
rdf:resource or literal value yet these are pieces of XML. If the
parseType="Resource" or parseType="Literal" this would be ok, but is that
legal RDF in current form?


Aaron Swartz wrote:

>
>
> Sean B. Palmer <sean@mysterylights.com> wrote:
>
> >> IOW, we're not referring to one of the elements or attributes
> >> "in" the XML Schema namespace -- [...]
> >
> > Huh? We're referring to the names that are in the XML Namespace for
> > W3C XML Schema.
>
> Aha, I see the confusion. We're not. Can I make that more clear?
> These names
> are not in the namespace -- XML Schema never uses them in namespaced XML.
> They do not refer to attributes or elements. They're simply a
> bunch of URIs
> that XML Schema has defined.

again, it is the use of the "xsd" prefix to this namespace URI that adds to
this hopeless confusion.

>
> >> We would not want the namespaces to be equal, since they are not.
> > Well, if you're not using the XSD set of names, then fine... but then
> > you can't refer to them as XSD datatypes - they're a whole different
> > bunch of things.
>
> Why? XML Schema Part 2 clearly defines that they are URIs for XML Schema
> Datatypes.
>
> >> This is a tenuous area, I admit. RDF does not state the meaning
> >> of the namespaces which it uses, but nor does XMLNS.
> > Yep, and this is the crux of the matter. If XMLNS had clearly
> > indicated the semantics behind XML namespaces, then we would not be in
> > this mess.
>
> Rather, I think the opposite. If it had indicated semantics, this
> mess might
> exists. As of now, it does not. Why do you think otherwise?
>

XML Schema clearly defines a datatype in terms of a QName. A "decimal" is
refered to by the QName "xsd:decimal" (xsd bound to
"http://www.w3.org/2001/XMLSchema")

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org

Received on Monday, 7 May 2001 01:09:14 UTC