[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: Re: URIs, concrete (was Re: Un-ask thequestion)

• To: <xml-dev@l...>
• Subject: Re: Re: URIs, concrete (was Re: Un-ask thequestion)
• From: Norman Gray <norman@a...>
• Date: Thu, 1 Aug 2002 18:13:01 +0100 (BST)
• In-reply-to: <200208011641.RAA08260@n...>

Greetings,

On Thu, 1 Aug 2002, David Carlisle wrote:

> I don't _want_ to think that way, but it seems your proposed
> chaneg would force me to think of them that way.
> If I have two elements <x:foo a=".."/> and <x:bar a=".."/>
> then currently I can think of the a attributes being unnamespaced
> and so not having any global definition, and so in particular having
> definition derived from their elements.

Yes, which I think is another way of saying that the meaning of unprefixed
attributes is effectively implementation defined, or at least that that
is the practical effect (or maybe even the secret intention!) of the
current text of the spec.

> If on the other hand things are changed so that each of these is
> considered to be the a attribute in the namespace  bound to x:
> then don't doesn't that lead to the conclusion that these attributes
> having the same globally unique namespaced name ought to be the same
> attribute and taht furthermore, being a an attribute with a gloably
> unique name, it ought to be a global attribute that can be used
> anywhere?

I don't think that either of these conclusions follows.

Namespaces (as I understand them) are a purely syntactical feature.
If some of the elements and attributes in an instance are in a certain
namespace, then I can look at that instance with namespace-shaped
spectacles and `see' only the things in that namespace.  That is, I
turn <x:p x:a=""/><foo bar=""/><x:q x:a=""/> into <p a=""/><q a=""/>.
What I do with that result is entirely up to me.  Presumably I have
some sort of schema (formal or informal), and some set of documents
giving the semantics of that namespace-extracted document, which tell
me whether p's a attribute and q's a attribute are the same thing, or
indeed whether q is allowed an a attribute at all.  I can't draw any
conclusions about that simply because they were once in a namespace,
though I take you to be suggesting that I can, or ought to, or might
be expected to.

[...which is to say, I have always understood Namespaces to be AF-lite,
with less rococo syntax]

All the best,

Norman


-- 
---------------------------------------------------------------------------
Norman Gray                        http://www.astro.gla.ac.uk/users/norman/
Physics and Astronomy, University of Glasgow, UK     norman@a...

• Follow-Ups:
  • Re: Re: URIs, concrete (was Re: Un-ask the
    • From: John Cowan <jcowan@r...>

• References:
  • Re: Re: URIs, concrete (was Re: Un-ask the question)
    • From: David Carlisle <davidc@n...>

• Prev by Date: Re: Re: URIs, concrete (was Re: Un-ask the question)
• Next by Date: Re: Re: URIs, concrete (was Re: Un-ask the question)
• Previous by thread: Re: Re: URIs, concrete (was Re: Un-ask thequestion)
• Next by thread: Re: Re: URIs, concrete (was Re: Un-ask the
• Index(es):
  • Date
  • Thread