Re: A proposed solution

> There is nothing in the current
> namespace spec that prevents an application from deducing that two
> things are identical at any level in the processing so I am interested
> in hearing what you build your assumptions on.

Some processor may decide to treat two namespaces in the same way,
but it can't decide two different namespaces are the same namespace.

an XSL engine accepts as XSL instructions elements that are in the XSL
namespace. The example I posted is not in the XSL namespace and so
must not be accepted. An XML parser which presented such a document
to the application as being in the XSL namespace would be unusable.

the namespace spec has essentially nothing to do with processing by
applications, and very little to do with comparison of namespaces. Its
main function is to specify unambiguously what is the namespace name
and local name of an item with qualified name of the form x:abc.

Currently the answer is quite clear and not depending on any
implementation, the local name is "abc" and the namespace name is
the XML attribute value of the closest xmlns:x XML 1.0 attribute.
(ie the literal string as given in the instance, after white space
normalisation that is applied to all attribute values.)

As far as I can tell you propose to replace this definition by an
underspecified situation where the actual namespace name is undefined
as different processors may or may not apply different normalisations.

David

Received on Sunday, 18 June 2000 17:04:59 UTC