Simon St.Laurent wrote:

>  >But in this specific case, I don't really understand how using PEs
>  >allows redefinition of names. Doesn't XML 1.0 preclude that?
> 
>  The trick is to include create a parameter entity for the element or
>  attribute names, assemble the prefix and the name there, and then make the
>  declaration using the assembled PE.
> 
>  This apparently lets you slide through with no space under section 4.4.5
as
>  'reference in literal' rather than 4.4.8.
> 
>  This came up back in March, and this seemed to be the accepted answer. (I
>  started out thinking it was prohibited.) I don't think it lets you go as
>  far as leaving out the colon and having no prefix, though that could
>  probably be accomplished with an additional INCLUDE/IGNORE section for
each
>  declaration...
> 

Fine. In reality, the DTD issue may be more of a DTD processor value issue
rather than an XML spec issue per se. A "namespace conformant" DTD validator
would validate based on expanded names rather than prefixed names. This
could be handled by a change to the "Element Valid" validity constraint (not
really a big deal - IHMO).

"Validity Constraint: Element Valid
An element is valid if there is a declaration matching elementdecl where the
Name [jb insert QName] matches the element type, and one of the following
holds:
..."

This would make such a document "Namespace Valid" rather than "XML 1.0
Valid" -- perhaps this is a new category.

Now that we have a good open source DTD processor...

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org