RE: Required cardinality checks

> This makes sense, but there are some less obvious cases where 
> it is still not clear how the above would apply. 

Indeed. As I say, there has been a great deal of debate about exactly what
is and is not allowed under these rules. The specification for error
handling in the face of optimization is informal and leaves a lot to the
good sense of implementors. Don't expect all implementations to behave the
same way.

Regards,

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