Lists Home |
Date Index |
By definition, there are no namespaces in content, unless that content
is of type xs:QName and validated by a W3C XML Schema processor.
In untyped documents, content is just a sequence of characters, as far
as the XML parser is concerned. Any namespace interpretation given to
that content is application semantics, not XML semantics.
We've dramatically increased the complexity of the XPath/XSLT/XQuery
data model because we want to (1) be able to preserve namespace bindings
in case they have application semantics that we are not aware of, even
if the namespace bindings have no effect on the interpretation of the
XML as a parser sees it, and (2) allow queries and stylesheets to do so
without any advance knowledge of the namespace bindings that will need
to be preserved.
These requirements do exist if you want to be able to query schemas and
stylesheets without knowing what is in them. It would be very nice to
wish them away...