Re: encoding of URI references in RDF/XML

From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
Subject: RE: encoding of URI references in RDF/XML
Date: Fri, 7 Nov 2003 13:50:03 +0100

> 
> Hi Peter,
> 
> we are treating this comment as a comment on RDF Concepts, particularly:
> 
> section 6.4 RDF URI References
> http://www.w3.org/TR/2003/WD-rdf-concepts-20031010/#section-Graph-URIref
> 
> > By the way, my understanding of the RDF specifications are that both
> > documents are indeed legal RDF/XML documents, but that they do not entail
> > each other because according to Section 6.4 of RDF Concepts
> > http://www.w3.org/foo{bar} and http://www.w3.org/foo%7Bbar%7D are
> > different RDF URI references.
> 
> Your understanding was the intent of the documents. Hence there does not
> seem to be an issue here.
> 
> There is of course the related tag issue
> 
> http://www.w3.org/2001/tag/issues.html#URIEquivalence-15
> 
> which was resolved with the following draft
> 
> http://www.textuality.com/tag/uri-comp-4
> 
> the key text being in
> http://www.textuality.com/tag/uri-comp-4#ladder
> 
> RDF follows XML Namespaces in using "Simple String Comparison" whereas the
> alternative in which the two are treated the same is covered in the TAG
> finding under "%-Escaping Issues", where I draw your attention to the
> observation that due to the potentially unknown character encoding of the
> %-escape there is a "(slim) chance of a false-positive in finding these
> equivalent"
> 
> Thus, the text seems to have been sufficiently clear for you to have
> understood the intent, and seems to be in accord with the TAG finding (at
> least not at odds with it).
> 
> I note that a significant part of the TAG finding is the last section "Good
> Practice When Generating URIs".
> [[
> Those who generate URIs and transmit them, or include them in resource
> representations, can make a major contribution to this common good by
> understanding the rules in RFC2396 and generating URIs in an at-least-partly
> canonicalized form.
> ]]
> 
> 
> Does this adequately address this comment? Or do we need to go further?
> 
> 
> thanks
> 
> Jeremy

You have convinced me that the situation was already so broken that there
is really no way to apply intuition to it, and that therefore the extra
brokenness that RDF brings to the mess is not worthwhile arguing about.

I do suggest, however, that there be some indication in RDF Concepts that
applications that wish to generate RDF URI references that contain non
US-ASCII characters not escape these characters.  See Section 6 of
http://www.textuality.com/tag/uri-comp-4 for a similar indication.  (This
text may have been incorporated into RFC2396bis.)

Peter F. Patel-Schneider

Received on Friday, 7 November 2003 09:27:03 UTC