Just wondering if anybody has a set of protected arithmatic functions
written for CLISP. I'm using the DOS version, with highest exponent = 708.
I tried a few routines from MCL, but the capacities seem to vary too much.
For example: (+ 2.0 5.5E37) works fine, but (+ 2.0 5.5E38) blows up.
Also, something very interesting: the MOST-POSITIVE_DOUBLE-FLOAT has a log
of 709.xxxxx, however if you try (exp 708) it blows up. So the following piece
of code does not work, when under MCL it does:
(exp (floor (log MOST-POSITIVE-DOUBLE-FLOAT)))
Upon further examination, I found that exp has a unpper boundary of 88, and
a lower boundary of -88.
So rather than go testing all of the boundaries, I thought that I'd ask
to see if anybody else (probably working on GP) has done it yet.
Thanks.
--
::: Rob Cattral :::
::: :::
::: ·····@Freenet.Carleton.Ca :::