Date: Tue, 18 Feb 2003 16:04:57 -0800 (PST) From: Paul Herman <pherman@frenchfries.net> To: FreeBSD Hackers <hackers@freebsd.org> Subject: arc4random() range Message-ID: <20030218153651.A240-100000@mammoth.eat.frenchfries.net>
next in thread | raw e-mail | index | archive | help
Hi,
...a potential quick commit for someone. :-)
What's the concesus that arc4random() should be a drop-in
replacement for rand()/random()? Consider the following that
caclulates the average of a bunch of random numbers on [0.0, 1.0]:
bash$ cat rand.c
#include <stdlib.h>
int i;
double avg;
#define ITERATIONS 1000000
#define FIND_AVG(func) do { \
for (i=0, avg = 0.0; i<ITERATIONS; i++) \
avg += (double)func()/(double)RAND_MAX; \
printf("avg:\t%f\n", avg/ITERATIONS); \
} while (0)
int main() {
FIND_AVG(rand);
FIND_AVG(random);
FIND_AVG(arc4random);
return 0;
}
bash$ cc rand.c && ./a.out
avg: 0.500404
avg: 0.500007
avg: 1.001370 <--- !!!!!!
bash$ uname -r
4.7-STABLE
bash$
Is is just understood that arc4random() doesn't obey RAND_MAX, or
should it? Whatever the answer is, perhaps the manpage should
state what the range is like the manpages for rand(3) and random(3)
do.
-Paul.
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030218153651.A240-100000>