- From: Dan Brickley <danbri@w3.org>
- Date: Thu, 12 Jul 2001 07:29:33 -0400 (EDT)
- To: Jan Grant <Jan.Grant@bristol.ac.uk>
- cc: Sergey Melnik <melnik@db.stanford.edu>, Aaron Swartz <me@aaronsw.com>, Graham Klyne <Graham.Klyne@Baltimore.com>, RDF core WG <w3c-rdfcore-wg@w3.org>
On Thu, 12 Jul 2001, Jan Grant wrote: > On Wed, 11 Jul 2001, Sergey Melnik wrote: > > > Aaron Swartz wrote: > > > > > > On Wednesday, July 11, 2001, at 12:27 PM, Graham Klyne wrote: > > > [...] > > > > > > > No specific mechanism for generating such URIs is mandated, but > > > > the following options might be considered: > > > > > > This is the problem I have. I think all parsers should spit out > > > equivalent genids for the same document -- the spec should > > > mandate the genid to use. > > > > I agree with Aaron. Otherwise, testing parsers for compliance is tricky. A quick note to bookmark a point: emitting the 'same' invented-URIs for un-named resource mentions is risky: if we make the generated ID purely a function of the bag-of-bytes XML document we could get situations where relative URIs are used to mention _different_ resources, but result in a parser emitting the same generated ID. I don't have time to cook up an example right here, but for eg. <foo rdf:resource="../index.html"/> in different contexts means something different, so an rdf/xml doc including that might be mentioning a different pair of resources. Dan
Received on Thursday, 12 July 2001 07:29:58 UTC