correct round of reals?

[Fredrik Lundh]

Tim Peters <tim_one at email.msn.com> wrote:
> That's pretty!  I like it.

thanks!

> > (this assumes IEEE doubles)
> 
> And inherits the rounding mode in effect.

that's on purpose, of course.  quoting from SUSV2:

    NAME
        rint - round-to-nearest integral value 

    SYNOPSIS
        #include <math.h>

        double rint(double x);

    DESCRIPTION
        The rint() function returns the integral value
        (represented as a double) nearest x in the direction
        of the current rounding mode. The current rounding
        mode is implementation-dependent.

> Unclear whether Peter wants that, or wants to force
> nearest/even.

well, he wants to emulate "rint" for platforms that don't
support it natively, so I think his only remaining problem
is to port our python code to C...  (see his other post).

</F>