RE: following-sibling and xsl:sort

> > Therefore, any problem, which has solution using the xxx:node-set()
> > extension function should have a solution without using it.
> 
> I tend to disagree with that statement.  

Me too. Turing completeness is not the same as closure over the data model.
To take an obvious example, there is no way of creating a result tree that
contains an unparsed entity, even though the data model allows unparsed
entities to exist.

Closer to the hypothesis in question, I don't believe it is possible in XSLT
1.0 without the xx:node-set() extension to create a result tree containing a
namespace that is declared in neither the source document nor the
stylesheet, if the result tree contains no element or attribute whose name
is in that namespace.

Michael Kay
http://www.saxonica.com/