On 5/8/07, Rainer Joswig <joswig at lisp.de> wrote:
>
> LispWorks 5
> CL-USER 6 > (expt 1.0d+20 2)
> 1.0000000000000016E40
> CL-USER 7 > (* 1.0d+20 1.0d+20)
> 1.0E40
> most Lisp systems I tried are giving the same result for both forms.
>
The Common Lisp spec does not seem to require that (expt <float> <integer>)
be calculated by repeated multiplication, and IEEE float doesn't specify it
at all, so I guess (unfortunately) that this is acceptable.
Simple_equal attempts to perform approximate equality so that this sort of
thing won't cause tests to fail, but it is buggy -- I believe Barton is
working on fixing it.
-s