Random problem (that is, problem with random())

Ralf Corsepius ralf.corsepius at rtems.org
Tue Mar 29 02:22:40 UTC 2005

On Mon, 2005-03-28 at 15:12 -0500, Peter Dufault wrote:
> On Mar 28, 2005, at 9:00 AM, Sergei Organov wrote:
> >
> > Thus, random() is BSD beast, not ISO or POSIX.
> >
> > ISO C equivalent is called rand().
> >
> Good point.  As is well known, "classic" rand() has its own problems, 
> plus I still don't understand why the definition of random() isn't 
> being published since it is in the header.

It's "#ifndef'ed __STRICT_ANSI__"
inside of the header but isn't implemented into libc.a.

IMO, this qualifies as a bug or at least as a weak coding in newlib.

>   It must need a 
> feature-enabling define before it shows up in the name space.
Yes - "Not using -ansi" (i.e. if you'd be using "gcc -ansi" "random()"
won't be available.)


More information about the users mailing list