- From: Paul W. Abrahams <abrahams@valinet.com>
- Date: Thu, 25 May 2000 12:01:02 -0400
- To: michaelm@netsol.com
- CC: abrahams@acm.org, John Cowan <cowan@locke.ccil.org>, timbl@w3.org, jcowan@reutershealth.com, xml-uri@w3.org
Michael Mealling wrote: > On Thu, May 25, 2000 at 10:39:24AM -0400, Paul W. Abrahams wrote: > > John Cowan wrote: > > > Paul W. Abrahams scripsit: > > > > > > > Is there an explicit definition of ``the > > > > strict data type `URI' '' in 2396 that I've missed? > > > > > > Seemingly not. However, 2396 supplies a generic syntax for URIs > > > and URI references. > > > > For URI references, yes. For URIs, no. There are no syntax rules with URI > > on the left side. > > Based on the concensus at the time the strict data type 'URI' was > denoted in the ABNF as the term 'absoluteURI'. The fact that the > ABNF term isn't called 'URI' was for clarity since that > simple acronym was used to many times within the ABNF. That leaves `URI' without a syntactic definition. The denotation in the first sentence should have been recorded. > I.e. the error may be on your part in insisting that since the ABNF > does not contain the term URI then it doesn't apply here. Indeed it > does and the question before everyone is, since the XML Namespace > document attempts to redefine the ABNF term 'URI-Reference' to > something that specifically violates the ABNF and the semantics the > document specifies for those terms, what part of the ABNF will > the next version (if there is a versioning) of the document adhere > to and not attempt to redefine. > > As far as I can tell the ABNF gives you really two useful choices: > 'URI-reference' or 'absoluteURI'. Both of those are unambiguous, I agree. The namespace spec currently uses URI reference; the alternative of forbidding relative URIs would change it to `absoluteURI'. > This leaves you with three choices for moving forward to fix the problem: > 1) pick one of the two ABNF terms > 2) say your not using URIs at all > 3) rewrite 2396..... Pragmatically, I agree that (1) solves the immediate problem. But I also believe that (3) should be done sooner or later, since there are problems with its implicit assumptions. As I noted in another message, RFC 1630 introduced the term `complete URI', and that's probably the right term for something that might be either an absolute URI or a resolved relative URI. Paul Abrahams
Received on Thursday, 25 May 2000 12:01:12 UTC