Re: [avr-libc-dev] [bug #23677] Request for cbrt() and strdup to be added to libc

[Russell Shaw]

Russell Shaw wrote:
> Andy Hutchinson wrote:
> > URL:
> >   <http://savannah.nongnu.org/bugs/?23677>
> >
> >                  Summary: Request for cbrt() and strdup to be added to 
> > libc
> >                  Project: AVR C Runtime Library
> >             Submitted by: hutchinsonandy
> >             Submitted on: Sunday 06/22/2008 at 21:51
> >                 Category: Feature Request
> >                 Severity: 3 - Normal
> >                 Priority: 5 - Normal
> >               Item Group: libc code
> >                   Status: None
> >         Percent Complete: 0%
> >              Assigned to: None
> >              Open/Closed: Open
> >          Discussion Lock: Any
> >                  Release: Any
> >            Fixed Release: None
> >
> >     _______________________________________________________
> >
> > Details:
> >
> > It seems gcc testsuite assumes presence of widely available strdup() and
> > cbrt() functions.
> >
> > strdup is a derivative of malloc
> > cbrt is cubic root x^(1/3)
> > Here are some generic version that can be used as starter.
> > I'm sure someone can trim them up a bit - or even convert to asm.
> >
> > Best version depends on how we built existing maths  functions.
> >
> > double cbrt(const double x)
> >
> > {
> > return pow(x,1.0/3.0);
> > }
> >
> > OR
> >
> > double cbrt(const double xx)
> > {
> >    double x= xx;
> >    if  (x==0.0) /* log will fail, but answer is easy */
> >       return 0.0;
> >    else if  (x>0.0)
> >        return(exp(log(x)/3.0));
> >    else
> >    x= -x;
> >       return(exp(log(x)/3.0));
> >   }
>
> I *never* use straight fp equality in code (x==0.0 baaad).
>
> http://c-faq.com/fp/fpequal.html
> http://c-faq.com/fp/strangefp.html

I suppose it's ok for this case. It's only checking that
log(x) doesn't go infinite. I'm very slow early in the day.