Re: Unknown prefix xml?

Martin Gudgin wrote:
> 
> Hi Eric,
> 
> Thanks very-much for you input, comments inline.

You're welcome, this is an interesting point (IMHO).
> 
> Martin Gudgin
> DevelopMentor
> 
> ----- Original Message -----
> From: "Eric van der Vlist" <vdv@dyomedea.com>
> To: "Martin Gudgin" <marting@develop.com>
> Cc: "XML Schema Dev" <xmlschema-dev@w3.org>
> Sent: Monday, January 29, 2001 11:06 AM
> Subject: Re: Unknown prefix xml?
> 
> > Martin Gudgin wrote:
> > >
> > > Just to be sure... You are saying I have to put
> > >
> > >     xmlns:xml='http://www.w3.org/XML/1998/namespace'
> > >
> > > in my schema document?
> >
> > Yes.
> >
> > > That seems very weird. The xml namespace prefix is always in scope, why
> > > should I have to declare it.
> >
> > My take on this is that W3C XML Schema is trying to kill 2 birds with a
> > single declaration.
> >
> > When you write xmlns:foo="http://bar" in a schema, you do 2 different
> > things:
> >
> > 1) you declare a namespace prefix per the namespaces in XML 1.0 rec
> > that, most of the time, you will not use as a namespace prefix.
> 
> [MJG]
> Well, I will use it as a namespace prefix. Maybe not in the name of an element or attribute but
> elsewhere where QNames are needed ( element and attribute decls for example ).

What I meant is that you are declaring xmlns:foo="http://bar" without
any intention to write somewhere in the document <foo:element> or
<element foo:attribute="...">. 

> >
> > 2) you declare to the schema processor that you will be using the prefix
> > 'foo' inside W3C XML Schema attributes to identify the namespace
> > "http://bar".
> 
> [MJG]
> I don't really see the distinction. Element names and attributes names are QNames. Certain
> attributes in the schema language are QNames.

Sorry for the rant. IMHO, this usage is creating unnecessary adherence
between namespace prefixes and schema information.
> >
> > The xml namespace prefix is always in your scope per the namespace rec
> > to perform 1), but not ins your scope to perform 2).
> 
> [MJG]
> Hmmm. This seems counterintuitive to me. I know Henry was keen that he be allowed to map
> http://www.w3.org/1998/XML/namespace to an arbitrary prefix but I didn't think it was because the
> xml prefix wasn't in-scope for attributes of type QName. But I could be misremembering...

Again, it's my personal take on the subject.
> >
> > > MSXML sees this as an error, complaing that the prefix xml is invalid.
> > > Xerces and Oracle accept it. Either way it doesn't solve my problem. A modified schema with the
> > > above declaration still gives the same error[3]
> >
> > I am confused with your schema. It has no global element definition...
> > Is it intended to be included in another one ? Otherwise I don't see
> > which instance document can be validated ? Could you provide one ?
> 
> [MJG]
> The schema is the instance. I'm trying to validate it against the schema-for-schemas. This is a
> minimum repro test case from a much larger schema I'm trying to validate. I can provide a schema
> with an element decl if you like but it won't make any difference to the problem at hand :-(

No, I had just missed the point.

I think your problem is that there is no namespace definition for the
schema target namespace.

When I add one, it validates using either XML Spy 3.5 or XSV (I haven't
the most recent version installed, though).

Eric
-- 
------------------------------------------------------------------------
Eric van der Vlist       Dyomedea                    http://dyomedea.com
http://xmlfr.org         http://4xt.org              http://ducotede.com
------------------------------------------------------------------------

Received on Monday, 29 January 2001 06:39:37 UTC