W3C home > Mailing lists > Public > xml-uri@w3.org > May 2000

Re: Toward the self-describing web [was: Irony heaped on irony]

From: David Carlisle <davidc@nag.co.uk>
Date: Thu, 18 May 2000 17:58:43 +0100 (BST)
Message-Id: <200005181658.RAA18576@nag.co.uk>
To: xml-dev@xml.org, xml-uri@w3.org


> In a syntactic, but not semantic sense.  This line of argument presumes no
> judicatory relevance in dissenting opinions by Supreme Court Justices.

I don't see this as a good analogy. The point at issue is not "what is a
namespace" in the abstract, or in the law of the United States, but 
"how are namespaces defined in the W3C namespace rec"

> A fundamental construct of meaning is the evolved history

The outline of namespaces in the note is not unreasonable and
it may have represented a dissenting opinion in the deliberations
that produced the final rec, but for people who have in the last year
implemented products and written documents  taking the
w3c namespace rec on good faith, it is rather uncomfortable to see
senior figures in w3c suggesting undermining the basic principle
on which xml namespaces work which is that it does _not_ imply any
lookup of the namespace name. (But does not of course forbid that
something may in fact be there.)

This implication that two namespace names which have URI pointing to
identical copies of a schema document are in any way the same
namespace is deeply worrying. (That is not implied by the relative
uri proposals, but appears to be the motivating reason for using
schema documents at the namesapce uri).

The relative namespace issue is really not so important, despite the
heat it generates. Most documents won't use relative namespace uri, and
most of those that do will work the same way whether a literal string or
absolute URI approach is taken (because, for example in xslt
stylesheets, all references are normally at the same base URI, so if the
absolute interpretation is taken, everything will be changed together
and still work). 

In my estimation the few documents that work differently given the two
interpretations would almost all have an undesired behaviour with the
"absolute" approach, but clearly others have different viewpoints and
if it ends the discussion and allows things to move on I'd live with the
absolute approach.


But this

> I think that sentence gets exploited to suggest that it's OK
> to use http://example.org/foo as a namespace name and then
> allow 404s for requests to that address, and so we should
> take it out if/when we next revise the Namespace spec.

suggestion that there _must_ be a resource, at the namesapce uri
would just be a complete change in the way namespaces work.

If you want a reliable URI to retrieve a schema why not use
the schemaLocation attribute that is designed for that purpose?

XHTML 1.0 has three DTD, XHTML 1.1 another and XHTML basic another,
that's 5 DTD and presumably 5 schema (with more to come as XHTML
modularisation kicks off, but they are all XHTML and all in the
XHTML namespace, just how is this "good practice" of putting a schema
document at the namespace URI supposed to work?

David
Received on Thursday, 18 May 2000 12:59:20 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:32:42 UTC