- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Fri, 20 Apr 2007 21:48:17 +0100
- To: "McDonald, Ira" <imcdonald@sharplabs.com>
- CC: Chris Lilley <chris@w3.org>, John Cowan <cowan@ccil.org>, Sandro Hawke <sandro@w3.org>, semantic-web@w3.org, www-international@w3.org
> This situation is an interoperability mess. I don't think it as bad as it might appear. Things manage to work together for a number of reasons. a) Applications don't need the strings in a URI or IRI slot to conform precisely with any spec whatsoever: what they need are: - ability to resolve relative references - ability to dereference a string sometimes neither of these depends on the bits of the specs that we are discussing. b) some application areas, for example, the semantic web, require the URI or IRI or whatever, to function as a logical identifier - and they tend to use character-by-character comparison These applications do require a higher level of interoperability concerning issues like when to hexify. However, at least for the semantic web world, there has been a strong emphasis on standard test suites, that the application developers do use, and do include, at least a handful of internationalization examples. c) In theory their could be problems when an application focussed on logical identifiers meets an application that carts around URIs simply as addresses to retrieve stuff from. However, such issues very rarely arise - I think because the latter group of applications tend to behave in a lazy fashion - only checking a URI for correctness when they need to, otherwise it's just a string. d) My initial question was precisely one area where the two views of an http URI collide: after a redirect. I haven't had much joy in helpful responses. Jeremy McDonald, Ira wrote: > Hi, > > An onlooker's observation. > > This situation is an interoperability mess. The chance that > a bunch of independent implmentations and validation tools > converged on the same interpretations of allowed/disallowed > is zero - this discussion thread hasn't visibly converged. > > Cheers, > - Ira > > Ira McDonald (Musician / Software Architect) > Chair - Linux Foundation Open Printing WG > Blue Roof Music / High North Inc > PO Box 221 Grand Marais, MI 49839 > phone: +1-906-494-2434 > email: imcdonald@sharplabs.com > > -----Original Message----- > From: www-international-request@w3.org > [mailto:www-international-request@w3.org]On Behalf Of Chris Lilley > Sent: Friday, April 20, 2007 12:30 PM > To: John Cowan > Cc: Jeremy Carroll; Sandro Hawke; semantic-web@w3.org; > www-international@w3.org > Subject: Re: xml:base (was Re: IRI meets RDF meets HTTP redirect) > > > > On Thursday, April 19, 2007, 5:14:49 PM, John wrote: > > > JC> Note: I'm a member of the XML Core WG, which owns the XML Base spec, > JC> and I may speak in accordance with my best recollection of things > JC> discussed there when making statements about intentions. However, > JC> I don't speak for the WG. > > And I am equally drawing on recollections of a working lunch at the Tech Plenary, with Paul Grosso and others, where the notion that the infoset property for the base URI 'represented a URI' rather than 'contained a URI' and thus, was compatible with IRIs, was discussed. > > I agree that the term IRI is a more recent coinage than many of the specifications which allowed a wider range of non-ASCII characters to 'represent URIs' after escaping. > > I disagree with your assertion that simply using XML Base (without dereferencing) forces a hexification. > > -- Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Friday, 20 April 2007 20:50:45 UTC