On Sun, Jan 07, 2001 at 03:29:14PM -0800, Neil Schemenauer wrote: > I don't know how test_pow was passing under Linux. Under Linux with the buggy float_pow: >>> pow(10.0, 0, 10) nan >>> pow(10.0, 0, 10) == 1 1 >>> pow(10.0, 0, 10) == 0 1 Under Windows NAN obviously behaves differently. floating-point-is-fun-ly y'rs Neil