Re: scoping of namespace prefix

Right.  There would be all sorts of strange problems that could arise if 
types/element/attribute declarations could change in an uncontrolled 
manner during a validation.  It would at best add complexity.  Locally 
scoped elements provide a limited capability in that direction, but 
otherwise schema is based on an organizing principle that the definitions 
and declarations used in a validation are conceptually invariant (you can 
pick them up as you go, but there must be no visible side effects of such 
incremental processing.)  Among other things, this rule ensures that 
systems like compilers will yield results consistent with, e.g. streaming 
implementations.

------------------------------------------------------------------------
Noah Mendelsohn                                    Voice: 1-617-693-4036
Lotus Development Corp.                            Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------------







Jeni Tennison <jeni@jenitennison.com>
Sent by: xmlschema-dev-request@w3.org
12/03/01 05:44 AM
Please respond to Jeni Tennison

 
        To:     "Naomi Dushay" <Naomi@cs.cornell.edu>
        cc:     xmlschema-dev@w3.org, (bcc: Noah Mendelsohn/CAM/Lotus)
        Subject:        Re: scoping of namespace prefix

Hi Naomi,

> I'm running into a problem with prefix scoping.
>
> My instance document (simplified):
>
> <bar:A xmlns:bar="ignored" xmlns:foo="http://foo.namespace"
> xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
>  xsi:schemaLocation="ignored bar.xsd">
>         <foo:B xsi:schemaLocation="http://foo.namespace fooB.xsd">
>                 <foo:Q />
>                 <foo:R />
>         </foo:B>
>         <foo:C xsi:schemaLocation="http://foo.namespace fooC.xsd">
>                 <foo:Y />
>                 <foo:Z />
>         </foo:C>
> </bar:A>
>
> I have XML schemas bar.xsd, fooB.xsd and fooC.xsd, all of which are
> valid. I can't find anything in the XML Schema spec that implies the
> above instance doc shouldn't work. But there seems to be a namespace
> prefix collision for prefix "foo." This is true for XSV and for the
> topologi Schematron Validator.

The problem is that you can't change the association between a
namespace and a schema half way through a document. On any single
assessment, a schema validator will only consider one schema for a
particular namespace.

The xsi:schemaLocation (or xsi:noNamespaceSchemaLocation) attribute
that defines the association between a namespace and a schema has to
be on or before the first element in the document with that namespace,
but then has a scope beyond (after) that element as well. This is
described in http://www.w3.org/TR/xmlschema-1/#schema-loc.

Effectively this means that you may as well define all your
associations between namespaces and schemas on the document element.
If you have different schemas for different sections of the document,
I'd recommend that you use different namespaces for those different
sections.

Cheers,

Jeni

---
Jeni Tennison
http://www.jenitennison.com/

Received on Tuesday, 4 December 2001 11:50:29 UTC