🚀 go-pugleaf

In article <1989Sep25.184425.20936@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes:
>Decimal *floating* point?  For money?  Are you sure you don't mean
>decimal *fixed* point?  (That is, integers plus a bit of compile-time
>bookkeeping for decimal-point positions.)  If you're thinking of COBOL
>arithmetic, that's definitely fixed-point.

Forgive me for not being more descriptive.  I live with this stuff every
day and forget that what I am doing is *not* very typical.  What I am
talking about is a decimal floatingpoint representation of numbers
that are normally used for business calculations which includes an
arbitrary amount of precision.  The user may specify a desired precision
at any time for any calculation.  Fixed-point is not as practical
under these circumstances.  Data items do not carry a precision attribute
with them.  Any data item must be prepared to take on any precision
(up to 14 digits) at any time.

I'm not trying to convert the whole world over to this manner of doing
math.  I merely suggested that a little help in the hardware (such as
access to the half-carry) would be of use to us and perhaps many other
developers who are in a similar situation.

-- 
Scott Amspoker
Basis International, Albuquerque, NM
(505) 345-5232