Re: Minutes for 2001-04-02?

[Paul Grosso <pgrosso@arbortext.com>]

At 11:20 2001 04 16 -0400, John Cowan wrote:
>Lauren Wood scripsit:
>
> > Present: Lauren, David, Norm, Paul, Tony
>
>John's belated views on a few points:
>
> > Attribute question: unprefixed attributes are in the per-element
> > namespace.  An attribute with an explicit namespace prefix is in the
> > global attribute partition. Norm suggests that global attributes in
> > the catalog namespace aren't  recognized and are an error (effects
> > must be ignored, but an error message can  be raised).
>
>I disagree here: prefixed attributes should be ignored and *not* an
>error.  This enables catalog elements to do double duty as XLink or
>RDF.

John, I want to be sure I understand with what you're disagreeing.

Norm's suggestion is not that any prefixed attr be an error, but only
that prefixed attrs whose prefix is associated with our XML Catalog
namespace be an error.  Otherwise, something like:

  <catalog     xmlns="whatever we decide our namespace is"
           xmlns:foo="whatever we decide our namespace is">
    <public publicId="some public id" uri="bar" foo:uri="baz"/>
  </catalog>

would be allowable and (1) we'd have to define just how to
handle this and (2) all catalog processors would have the
added overhead of always having to make two checks for every
attribute.

Instead, since we "own" our own namespace, we felt we should
just say that we define our namespace such that elements in
our namespace do not allow explicitly prefixed attributes 
(which are called "global attributes" by the Namespace spec) 
that are in our namespace.  Unprefixed attributes (which are
therefore in the "per-element-type partitition" and which
we do recognize as having semantic import on elements in our
namespace) and attributes in other namespaces (which we ignore)
are allowed.

Do you object to this?

paul