David Ronis wrote:
> Here it is:
>
> (%i1) double_factorial(10.0);
> (%o1) 3839.999999999999
That's what I get. Something is wrong when the testsuite is run. If
you use the old testsuite/threshold, does the test still fail?
>
>
> BTW you mentioned you were on Solaris. If memory serves me correctly,
> doesn't a sun have 1 more digit in the mantissa as opposed to the
> exponent? That might explain the difference.
Nope. Sun has straight IEEE754 double-float numbers, unlike x86 with
it's 80-bit long floats and its kind-of 64-bit double-float numbers.
Ray