From: <sis...@op...> - 2013-12-15 14:14:45
|
-----Original Message----- From: Peter Rockett > So true! If you do any serious floating-point calculations and end-up > with errors in the 1 ulp range you should be punching the air with > hysterical delight! You'd love perl then - it has 1ulp discrepancies all over the place ;-) As an added bonus, you can get quite a few 2ulp discrepancies - and I think I've seen the odd 3ulp discrepancy there. This was what sent me down the path of seeing what strtod (which perl does *not* use in assigning its floating-point values) was reporting in the first place. Cheers, Rob |