Attributes with Intent (and namespaces)

[james anderson]

please bear with me; maybe we're getting to something here...

MURATA Makoto wrote:
> I strongly disagree.  The whole point of the namespace extension is to
> allow applications to tell which namespace a name belongs to.  A qualified
> name is not an opaque object.  Application programs should and will care
> the attached URI.

why?

so far, as i've understood namespaces (wrt xml):
1. the NSDef will bind an URI literal. beyond that the value is of unspecified
form and content.
2. the prefix has a bearing on encoding and decoding only.
3. the universal name arises through the uniqueness of values in the relation
(NSDef.SystemLiteral X QName.LocalPart)

i presume everyone agrees up to here. ...

now, to the reasons the <EM>value</EM> of the uri does not matter to an application:
4. the purpose of the mechanism "articulating" the namespace is to control the
scope of
   declarations made about a name. this is, among other things, the
schema-reuse goal in
   the wd.
5. the application's concern is that, given a universal name, it can determine the
    declarations or effect the declared behaviour.
6. to accomplish this, the application will depend on functions of the form
   (UniversalName) -> (Declaration + Behaviour)
7.  there are no occasions when the application will need a function of the form
   (URI) -> (Declaration + Behaviour)
8. functions of the forms
   (URI X LocalPart) -> UniversalName or UniversalName -> (URI X LocalPart) or even
   (URI X UniversalName ) -> UniversalName
   are necessary for dom construction operations only.
   (yes, this is not a w^3 conform dom, and yes i advocate that the
application use a dom)
9. functions of a form such as
   (URI X LocalPart) -> (Declaration + Behaviour)
   are unnecessary, since, at the application level, <EM>given a dom</EM>
operation of the
   form
   (URI X LocalPart) -> UniversalName
   simply makes no sense. they whould be well constructing names which have no declarations.
   in order words, "the uri does not matter to the application".
10. where the intent is to establish a relation between two universal names,
    (see the forgoing discussion in this thread)
    there may be a need for functions in domains such as
   (URI X URI) -> ? or (UniversalName X UniversalName) -> ?
   but the effects of these (and here i again cross the line to advocacy)
should be declared
   at the decoding level, just as with operations using functions
   (Prefix X URI) -> NSRegion
   which declare namespace regions, .

given the exposition above, where do we diverge?

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)