xml-dev - RE: Default Namespaces - why don't they apply to attributes?

[Ronald Bourret <rpbourret@hotmail.com>]

Jonathan Borden wrote:

>  I don't understand the distinction you are making: a node having a
>namespace vs. being a member of a namespace. To me a node *has* a name, and
>*belongs* in a namespace, otherwise why use the term *namespace* where
>*name* is perfectly sufficient? I sumbit that having a namespace/prefix and
>being a member of a namespace define identical behavior.

I definitely am making this distinction -- and with good reason, I feel. If 
you read the namespaces spec closely, a "namespace" is really only defined 
as a conceptual entity once -- somewhere up near the top where the term is 
originally defined. The rest of the spec talks about the mapping from 
prefixed names to universal names. It never talks about namespaces as 
entities and, more importantly to this discussion, never defines membership 
in a namespace.

I realize this is to a large extent splitting semantic hairs, but I find it 
unfortunately necessary to split them, especially with respect to the 
question of whether unprefixed attributes are "in" a namespace. My points 
are:

1) The namespaces spec never answers this question and, if you rephrase 
questions like, "Is an unprefixed attribute in a namespace?" as "Can you 
uniquely identify an unprefixed attribute?" you can avoid this problem.

2) Since unique identification is what namespaces were designed to solve, 
you've solved your problem and you don't need to worry about whether 
unprefixed attributes are in a namespace.

3) I have not seen any questions about namespace membership that can't be 
rephrased as questions about identity. (This is what my call for examples 
was about -- to see if I am wrong on this point.)

4) The namespaces spec would have been better off having either (a) been 
written in terms of membership, or (b) been titled "Unique Identification of 
Elements and Attributes through Mapping Prefixed Names to Universal Names" 
or some similar uncatchy title.

5) Since the namespaces spec was not written in terms of membership, it is 
better to think of namespaces in terms of unique identification and mappings 
than in terms of conceptual/physical objects called namespaces.

(Along these lines, a namespace is at least the collection of all elements 
and attributes whose universal name has the same URI. Note that this is a 
logical, but unrequired, step in the mapping from prefixed name to universal 
name.)

>  You have defined an algorithm which has the effect of associating an
>attribute namespace with the enclosing element namespace. This is fine, yet
>what specification is directing me to use this algorithm?

In the case we are discussing, RDF is directing you to do this. The 
namespaces spec is not.

>An application
>which defines behavior(node) based upon namespaceURI(node) would not be
>using your algorithm. Are you implying that this is wrong?

No. But I am saying (and I think we agree here) that the RDF spec is saying 
more than the namespaces spec does on this point.

>  The existence of such an application is the counter example to your
>argument. We never argeed that you could rewrite the application :-)

Touche! I plead guilty to not wording the question clearly enough.

-- Ron
________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com