> This is not true as written.
>
> XPath 2.0 is schema aware for anonymous types too, and
> leverages such types appropriately for operators, including
> comparisons. What you can not do is refer to such a type by
> name.
Roger's primary aim was to take advantage of type information to get good
compile-time error checking, in particular to detect "impossible" path
expressions.
In both XQuery and XSLT this depends on declaring the types of parameters to
functions (and in XSLT, the type of the match pattern for a template rule).
The processor won't detect statically that $x/a/b/c is "impossible" unless
it knows that $x is an element whose complex type is a statically known type
T. And if $x is a parameter to a function, you can only achieve that if
either the element declaration or the type (or both) are global.
Michael Kay
http://www.saxonica.com/
