I went all the way to the top in my effort to determine whether the
HPDATEDIFF instrinsic really is broken in 6.0.   Jeff Vance recalls a bug
in a lower-level routine that was called by all the date intrinsics, which
would cause the program to abort rather than just returning an error code,
if an invalid date was sent to the intrinsic.  However, he said there was a
patch for that.  He didn't know the MPE/iX release level or patch name off
the top of his head.  Well, what do you expect when you stop someone in
the hall between sessions?  Also, these people in the lab are working on
future releases, and can't possible remember all these details.

I recall that there was a porblem of some kind when the date intrinsics
first came out, but that was fixed in one of the 5.5 powerpatches.  I don't
have access to a 6.0 machine right now, so I can't try out HPDATEDIFF.
One of my clients will be upgrading to 6.0 on their development machine
soon, and then I'll be sure to test this.  In the meantime, if anyone else
cares to check, we'd all appreciate it.

Cecile Chi