xml-dev - Re: Validating parser and namespace declarations

[Ronald Bourret <rpbourret@rpbourret.com>]

Rob Lugt wrote:

> Hello all, I have a namespace question which I hope is not a FAQ:-

It is :)  See section 7 (especialy question 7.6) of the XML Namespaces
FAQ:

   http://www.rpbourret.com/xml/NamespacesFAQ.htm

> Should a namespace-aware validating parser frown on an undeclared default
> namespace declaration?
> 
> <!DOCTYPE test [
> <!ELEMENT test EMPTY>
> ]>
> <test xmlns="http://elcel.com/ns"/>

Yes.

> A strict XML 1.0 validating parser will complain that xmlns is not a
> declared attribute for element <test>, but is this the desired behaviour for
> a namespace-aware validating parser?  I have tried the Xerces-java parser
> which doesn't complain (both with and without namespace support switched on)
> and the Microsoft MSXML Parser complains in the XML 1.0 tradition.

Technically, Xerces' behavior is not correct. This is because validation
is an XML 1.0 concept and namespaces are post-1.0. I will let somebody
else decide if Xerces' behavior is useful or not ;)

(As an interesting aside, an earlier version of MSXML required fixed
xmlns attributes that "declared" namespace prefixes in the DTD. This
forced users to write valid documents (with respect to namespaces) and
allowed MSXML to check for validity in those documents. It also
generated protests about the parser not following the standard, so MS
promised to remove it. (This was last winter, so I assume it is long
gone.) I wonder if anybody will complain about Xerces' behavior, which
is similar...)

-- 
Ronald Bourret
Programming, Writing, and Training
XML, Databases, and Schemas
http://www.rpbourret.com