- From: Tim Berners-Lee <timbl@w3.org>
- Date: Wed, 7 Jun 2000 22:58:05 -0400
- To: <xml-uri@w3.org>, "Simon St.Laurent" <simonstl@simonstl.com>
-----Original Message----- From: Simon St.Laurent <simonstl@simonstl.com> To: xml-uri@w3.org <xml-uri@w3.org> Date: Monday, June 05, 2000 3:58 PM Subject: Re: Request for status dump and issues check >I don't think anything significant has changed in anyone's position (so far >as I can tell). > On the other hand, I think it's at least clearer that the >underlying assumptions on which people are building their arguments are >wildly (perhaps wickedly) different. Actually, I think there have been shifts. I can say (and have indeed said) that I now understand better now what people were saying before (e.g. when some said they used relative URI-references I assumed they were using them in that sense rather than as just strings, which caused a lot of my initial confusion.) I think others have too. I have also shifted ground in that rather than simply insisting on absolutizing, I can see we may have to compromise with banning relative URIs, because the interpretations of them are simply inconsistent. I see two possible outcomes at the moment which I understand 1) Namespace references are URI-references according to the spec. XML document processing requires a base-URI to be provided, or it may under some circumstances fail, but XML processors should provide a base URI of some sort in all cases (even based on pwd in a unix pipe) The bit in the Namespace Recommendation about character-for-character comparisons was wrong because no one had then spotted that it provided inconsistencies with relative URUI references. It is changed to be compatible to the later Xpath spec. Users note that there are rather specific cases unlikely in real use in which this could conceivably cause well-formedness for documents with multiple namespaces all quoted using relative URI-references to change. This would be my personal preference as I believe getting it right is more important for the long term. 2) Namespace references are URI-references but the use of relative-URIs is not allowed in a conforming document. The results of using one are undefined The namespace specification was found after its release as a Recommendation to be inconsistent in the case of relative URI-references for namespace names. The restriction to disallow relative URI-references was made to allow XML software to check well-formedness of a document without access to its URI. This would be a compromise I could live with. Especially if, as David Durand suggested, it be the short-term course of action and (1) be lined up for XML 2.0 As John Curran just said there have been many wild ideas which have lead to useful discussion but do not remain in the universe of discourse. There is I think one option which I can try to summarize. 3) Namespace names look like (have the syntax of) URI references but are not. They do not share any semantics with the URI references which they happen to match. Within any context in which a URI is expected that URI cannot be treated as a namespace name I could not live with that.
Received on Wednesday, 7 June 2000 22:56:40 UTC