RE: [xml-dev] Re: determining ID-ness in XML

[Don Park <donpark@docuverse.com>]

> I like James' proposal.
>
> However I'd suggest that the 80/20 split lies at not allowing
> more than one ID attribute per element. (So I'm disagreeing
> with DavidB and DonP, at my peril, I'm sure! :)

Yeah, I don't like having to parse attribute value so I understand.  Also, I
have been thinking about the namespace issue complicating the mechanism.
While xml:idatt (XML ID Attribute Declaration?) is most likely placed within
a container element belonging to the same namespace as the attribute of type
ID like this:

<foo xmlns:fb="http://foobar.com/ns/1.0" xml:idatt="id">
  <bar id="b1"/>
  <bar id="b2"/>
  <barfight href="#b1"/>
</foo>

the mechanism gets tedious quickly when there isn't a convenient container
to declare ID type attribute for contained elements.  I guess the question
is which side of 80/20 does this case fall under.

For sake of discussion, here is an example of XML Namespace Declaration-like
approach to XML ID Attribute Declaration.

<foo
    xmlns="http://foo.com/ns"
    xmlns:ds="http://www.w3.org/2000/09/xmldsig#"
    xmlid:id="http://foo.com/ns"
    xmlid:Id="http://www.w3.org/2000/09/xmldsig#">
  <bar id="abc"/>
  <ds:Signature Id="def">...</ds:Signature>
</foo>

Hmm.  It looks rather ugly.  We could opt to ignore the whole namespace
issue and use the same mechanism to specify target ID-space although I am
not sure if there is a huge demand for multiple ID-space per document.

Don Park