[
Lists Home |
Date Index |
Thread Index
]
- From: Jonathan Borden <jborden@mediaone.net>
- To: "Simon St.Laurent" <simonstl@simonstl.com>,XML-Dev Mailing list <xml-dev@xml.org>
- Date: Fri, 27 Oct 2000 18:00:05 -0400
Simon St.Laurent wrote:
> At 02:07 PM 10/27/00 -0400, Jonathan Borden wrote:
> >We agree. ... (btw: by 'global attribute' do you mean a non namespace
> >prefixed attribute?)
>
> No, they're quite definitely prefixed.
>
> I'm referring to the discussion in Namespaces in XML, Section A.2:
> http://www.w3.org/TR/REC-xml-names/#ns-breakdown
I understand. This term was confusing to me because of the overloaded use of
the term 'global' with respect to namespaces ... for example a member of the
global namespace (in other contexts) is often unprefixed.
It seems to me that this sort of language arises because of certain notions
concerning the relationship of an attribute to its defining element. There
exists a strong desire to make more of the relationship between the
attribute and its containing element than actually exists, for example that
an attribute is a child of the parent element, or that an attribute inherits
the namespace of the element, or that somehow unprefixed attributes have
some 'special' relationship with the element. What? (if we go back to the
idea of a QName as being a way to prevent name clashes, there's scant
justification for this).
One of the reasons the term "grove" is used to represent an abstract
document, rather than simply a "tree" is that things like the attribute set
are attached to but distinct from the parent-child heirarchy.
Both the SAX and DOM APIs make no distinction between namespace prefixed and
non-prefixed attributes in that they are always in the same set i.e. there
is no method el.getGlobalAttributeList()
Perhaps I am missing something?
Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org
|
