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

Re: A little courtesy, please (resource questions)

From: Dan Connolly <connolly@w3.org>
Date: Wed, 24 May 2000 13:09:21 -0500
Message-ID: <392C1AD1.2EAF7096@w3.org>
To: David Carlisle <davidc@nag.co.uk>
CC: xml-uri@w3.org
David Carlisle wrote:
> 
> > Please elaborate in detail about what is underspecified about the
> 
> > Given a base URI,
> That bit. Not all documents have a base URI and conventional use of the
> phrase XML document (and its use in the XML spec for that matter)
> would normally allow notions of copying a document, or expressing it in
> canonical form or lots of other things.

One can consider a base URI to be a property of an XML document,
or as something to be associated with an XML document. I'm not at
all sure that the XML 1.0 spec clearly distinguishes these. The
XML Infoset draft specifies that the [base URI] is a property
of an XML document, and that's the way I'm using the term.

http://www.w3.org/TR/1999/WD-xml-infoset-19991220#infoitem.document

We could draft it the other way, so that the base URI is extrinsic.
That would mean that declarations like this example from
the XML spec

	<!ENTITY hatch-pic
                    SYSTEM "../grafix/OpenHatch.gif"
                    NDATA gif >

wouldn't determine any particular absolute URI as the
[system identifier] property of the hatch-pic entity...
the [system identifier] property would have to be defined
as a URI reference, to be combined with a base URI in order
to determine the actual identifer of the resource to access.


Anyway... as specified in the infoset draft, if you copy
the characters of an XML document from one place to another
in the web, the result is a different XML document.

> None of these makes much sense
> if even something as basic as the names of the elements in the document
> changes if you move the document

That's why the XML infoset is drafted the way it is.

> (or even don't move it at all, but just
> type HTTP instead of http into a browser.)
> 
> > I'm still not sure why you asked those particular questions, but
> 
> Because I wanted the response
> 
> > So don't do that;
> 
> That is, the fact that relative URI references aren't very rdf friendly
> isn't really any reason to change the namespace spec, as there are loads
> of possible URI forms that similarly wouldn't really work with rdf.
> It is up to the rdf spec to say which documents it works with.
> 
> David

My motivation for changing the namespace spec isn't just the way
RDF works... it's the inconsistency between the way relative URI
references are specified to work in the namespace spec (i.e.
that you can meaningfully compare them across documents without
absolutizing them) and the way they're used in every other
spec (HTTP, HTML, XML Linking, XSLT, ...) and every implementation
I know of.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Wednesday, 24 May 2000 14:10:20 UTC

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