doats@spam.ua.oz (Donald Oats) (04/25/91)
I have found a problem concerning the float switch in REDUCE.
It seems to cause errors in some arithmetic operations.
Below is the (slightly edited) output from REDUCE 3.3.
Has anyone else had this problem ? If so, is there a version of
REDUCE for which the bug has been fixed?
------------------------------------------------------------------
% REDUCE input file -- 25 April 1991
% REDUCE 3.3, 15-Jan-88
% This program illustrates a problem concerning the use of the
% "on float" switch.
a:=300000000;
A := 300000000
% First, print out a few sample calculations.
1/a;
1
-----------
300000000
10/a;
1
----------
30000000
100/a;
1
---------
3000000
1000/a;
1
--------
300000
% All the above were OK, now repeat, but with float enabled.
on float;
1/a;
3.33333e-09
10/a;
0
100/a;
0
1000/a;
3.33333e-06
% Note that 10/a and 100/a both return 0, yet 1/a and 1000/a are OK.
% What gives ???
shut "bug.dat";
---------------------------------------------------------------------
Don.
--------------------------------------------------------------------------------
"A man may fight for many things: his country, his principles, his friends,
the glistening tear on the cheek of a golden child. But personally, I'd
mudwrestle my own mother for a ton of cash, an amusing clock & a stack of
French porn." --- Lord Edmund BlackAdder
--------------------------------------------------------------------------------