At 3:35 PM -0700 8/9/01, Brad Hughes wrote:
>Maybe it's just me, then, but I *want* my programs to blow up if there's
>overflow or underflow. It means something's wrong.
That's certainly been the VMS philosophy over the years. You don't
want the milling machine to cut through the shop floor or the
medication dosage to be wrong because of a programming error or
garbage data input. But there are a lot of real-world computing
situations that don't work like that anymore. For example, you don't
want your browser to crash because you moved your mouse too quickly.
I think the ability to know you've hit a boundary will still be
there, but you may have to think more about it rather than rely on
being automatically stopped in your tracks. My biggest concern is
what we will do about VAX.
>Hasn't his always been
>the case, and has only just become an issue due to the bigint test?
I think there have been assumptions there for a while that have not
been tested as thoroughly until now.
>(And yes, I know I'll probably have to convert a couple of thousand data
>files from VAX to IEEE floating point when the Great Port To IPF happens,
>but I'd hoped to wait a few years before doing so.)
I'd guess they'd support G_FLOAT in some fashion. They might
convert->do math->convert back, the same way they implement D_FLOAT
on Alpha.
--
____________________________________________
Craig A. Berry
mailto:[EMAIL PROTECTED]
"Literary critics usually know what they're
talking about. Even if they're wrong."
-- Perl creator Larry Wall
