RE: "canonical" URIs


I'm a little confused here.

The facts, as I understand them:
1. XML Namespaces spec defines equivalence rules for URIs in namespace
names, not any URIs in an XML document.
2. XML Namespaces requires URIs to be absolute.
3. anyURI can be a URI References
4. RFC 2396 section 6 says that URI and URI reference equivalence is scheme
5. RFC 2396 provides absolutization rules for URI references, including base
URI values.
6. RFC 2616 section 3.2.3 says what the rules are for http schemed URIs.

Are you saying that XML Schema is using XML Namespace Name rules rather than
RFC 2616 and 2396 when evaluating URI references?

What I was suggesting is that anyURI comparision should default to the 2616
rules, and I think you are saying that Schema uses Namespace Name rules.
These seem to be in contradiction.

Also, I think that the application view of anyURI equivalence could be
different than the schema view.  An xml app, like XLink/XInclude/XEnc/DSig,
would want to do equivalence for it's own purposes and under it's control.
I think that's where Joseph and Stephen's question is coming from.  I know
it came up in XLink and XInclude.


> -----Original Message-----
> From:
> []On Behalf Of
> Biron,Paul V
> Sent: Tuesday, March 19, 2002 2:15 PM
> To: 'David Orchard';
> Subject: RE: "canonical" URIs
> > -----Original Message-----
> > From:	David Orchard []
> > Sent:	Tuesday, March 19, 2002 10:05 AM
> > To:
> > Subject:	RE: "canonical" URIs
> >
> > TAG members,
> >
> > I don't see URI comparison officially listed as a TAG
> issue.  I'd like
> > Joseph/Stephen's issue added to the TAG issues list.
> >
> > Equivalence rules for URIs are defined by the URI scheme.
> HTTP has a
> > section on URI comparison.
> >
> > However, XML does not have a default comparison function for the XML
> > Schema
> > anyURI data type.  I think a reasonable approach would be
> to say that the
> > default comparision function for anyURI is to use the HTTP
> URI comparison
> > algorithm, but that it is overridable by any scheme.
> >
> Our intention (given enough time I'm sure I could find the minutes
> documenting this) was that equivalence for xs:anyURI was
> exactly as stated
> in the namespaces rec (i.e., "character-by-character", simple
> binary string
> compare) but you are right, that is not spelled out (I
> submitted this issue
> for a potential erratum).  Why?  Because it thought it was to
> be too much of
> a burden on processors to check things on a scheme-specific
> what is a processor to do with a scheme that it has never heard of?
> For what its worth, from a purely schema validation perspective, a
> comparison/equivalence function on anyURI is only necessary
> for testing
> whether a literal is included in a type derived from anyURI
> with one or more
> values given for the enumeration facet.
> pvb

Received on Tuesday, 19 March 2002 22:41:02 UTC