>   Yes I know.
I know you know:-)


>   I need compiler and runtime help to detect my errors as
> soon as possible.

I really wish I could agree with this whole heartedly. For quite a long
time my main prgramming language was standard ML, which has one of the
strictest static typing systems, it was great, if you got the thing to
compile, it was probably going to evaluate to the right result:-)

But unfortunately I'd say that most of the (offical) static typing of
xpath/xquery (the support of which is really why there's two sets of
comparison operators) is entirely bogus. As can be seen by recent
attempts to use the XQuery test suite with systems implementing the
static typing, almost all the extra errors that it flags are bogus and
the program would have run correctly, and does run correctly on systems
that don't implement the static typing. Principally this is because
(unlike Standard ML, where you almost never need to explictly declare
types of expresions as the type inference system is automatic) the type
inference system of xpath is closely bound up with cardinality checking
that can never be automatic.  Thus basically I can't help but see "eq"
and friends as an ugly wart on xpath2, that fortunately doesn't really
intrude too much, you can just not use it and use the = operator instead.

David