> Then it is a bug, because it returns false on Saxon 8.8 and 
> 8.8.0.4. See my other mail for a couple of tests I did.

Indeed there does appear to be a Saxon bug here. I tested it on my current
development version and it gave the right answer, but it's wrong in the
issued version.

(The optimization that's dealing with constructs like [position() != (1 to
100)] is incorrect). 

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