Re: [xml-dev] The purpose of a namespace URI is ...

[Liam R E Quin <liam@w3.org>]

On Sun, 2012-12-02 at 09:33 -0800, Dan Vint wrote:
> This is an ok alternate way of getting the same effect, but it is not 
> supported universally. For instance XML catalogs I don't believe will 
> work with this.

That's an interesting point. It would be an interesting idea to augment
XML Catalogue files with the ability to discriminate based on a version
attribute, although I don't think that would fit well with all existing
APIs for entity resolvers.

The truth is that we had very little time available when we designed
namespaces, and some of the people involved were not up to speed with
Web architecture.

However, there's no intrinsic advantage to a public identifier over a
namespace URI, they're both strings; the Web has been described as an
experiment to find out what happens if you give everything a name, a
URI, and so at W3C there tends to be a lot of pressure to use URIs for
everything.

[...]

> The Public Identifier provided a good mechanism for identifying the 
> vocabulary of a document, why was this not carried forward with Schemas?

The battle you should fight is not public identifiers vs. URIs. That
battle has basically been fought more than a decade ago, and, like most
battles, happened because the people involved failed to reach agreement.
We should not have included public identifiers in XML at all - it would
have been better than including them without defining how they worked,
when SYSTEM urn:foo has the potential to be just as good and SYSTEM
http://... had the potential to be better.

Maybe Oasis (or even a joint Oasis/W3C group?) could enhance XML
catalogs to support simple conditional mappings, but I'm not sure if the
changes would get enough traction today to be worth while.

I agree that when you have, for example, a mixture of XSLT 1 and XSLT 2
and XSLT 3 stylesheets and you want to feed them to different
processors, the namespace identifier is not adequate. Neither, however,
are public identifiers on their own, since by the time an XSLT processor
is resolving a PI, you've already chosen an XSLT processor... so the
logic belongs in whatever you are using to choose the XSLT processor.

Maybe this would be a good requirement to bring to the XML Processing
Model Working Group.

So there are multiple use cases for versioning, and no single answer,
except that more could be done.

Liam

-- 
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
Pictures from old books: http://fromoldbooks.org/
Ankh: irc.sorcery.net irc.gnome.org freenode/#xml