RE: 'nother xslt2 engine

> > By default, the Altova XSLT 2.0 Engine strips all whitespace in
> > whitespace-only nodes from the source XML document. Note 
> that the presence
> > and absence of whitespace-only nodes affects the value the 
> position()
> > function returns. 
> 
> They must be joking? Or is this thing different than in XSLT 1.0?

XSLT 2.0 is a little more explicit than XSLT 1.0 in defining exactly what
the rules are.

With both versions, XSLT is defined to take a source tree as input and there
are no constraints on how the source tree is constructed: in effect,
construction of source trees is something that happens outside the XSLT
processor rather than within it. If someone wants to provide a utility that
constructs a tree by deleting every second element, and they think there is
a market for such a utility, they are welcome to do it.

The Data Model specification for 2.0 defines a default way of constructing
the data model from an Infoset or PSVI. This states that whitespace-only
text nodes should be discarded if they appear in "element content" - that
is, if they are children of an element whose DTD or XML Schema content model
does not allow mixed content. However, the conformance rules for XSLT 2.0 do
not require a processor to support this mapping. (Conformance rules,
however, often change in the final stages of completing a spec.)

Michael Kay