Re: [charmodReview-17] replacing all URIs with IRIs

> [I'm copying, because there is an error
> report for the validator, and some suggestions of how to fix it.]

I (sometimes :( ) maintain my parser (ARP) which underpins the W3C RDF

The comments which related to the parser  seemed like fair-enough
defects against it.

The RDF Core has made a number of clarifications that suggest that ARP
has not correctly implemented RDF.
Most of these are ones that I have voted for, including the one that
effectively amounts to using XML Schema anyURI (aka IRI in this thread)
rather than URIs.

While voting I was making decision that I was prepared to do the
programming work required.

I hope to be able to find the time to address these issues in the next
couple of months.

I note that the level of checking that ARP does, will still require me
to use an IRI %-escaping algorithm, in order to check that the resulting
US-ASCII string does conform with the syntactic restrictions of RFC 2396
(as modified by RFC 2732).

My main architectural level concern is what will happen with XML
Namespaces 1.1.
It would be odd for RDF (and a number of other specs) to be using IRIs
(or something similar but not quite - IRI draft is not cooked) while the
core use of URIs in XML was still US-ASCII only.

I hope that a number of specs that are at or near last call including:
- charmod
- XML 1.1, XML Namespaces 1.1
- RDF *

will all show a consistent approach to URI refs and their
I note that both XML Namespaces and RDF are interested in absolute URI
references, and depend most keenly on the notion of equality between
these rather than identity of the denoted resources.


Received on Monday, 27 May 2002 12:13:42 UTC