nextafter and forced rounding

[Richard Henderson]

On Fri, Aug 06, 2004 at 11:31:00AM -0700, Ulrich Drepper wrote:
> I'm seeing this on x86 in test-float:
> 
> testing float (without inline functions)
> Failure: Test: nextafter (0, -1) == -fltmin
> Result:
>  is:         -1.40129846432481707092e-45  -0x1.00000000000000000000p-149
>  should be:  -1.40129846432481707092e-45  -0x1.00000000000000000000p-149
>  difference:  0.00000000000000000000e+00   0x0.00000000000000000000p+0

That's ... odd.  How can you have a miscompare with identical values?


r~