Sender: nate AT cartsys DOT com
Message-ID: <370A7056.1FE86F18@cartsys.com>
Date: Tue, 06 Apr 1999 13:36:38 -0700
From: Nate Eldredge <nate AT cartsys DOT com>
X-Mailer: Mozilla 4.08 [en] (X11; I; Linux 2.2.5 i586)
MIME-Version: 1.0
To: djgpp AT delorie DOT com
Subject: Re: DJGPP's random functions.
References: <3705018e AT news DOT ismi DOT net> <923349441 DOT 478285 AT eole>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp AT delorie DOT com

Crousto wrote:
> 
> The help file speaks about 2 functions : rand() and random(). I've noticed
> that rand (or random, i don't remember ;) return always the same sequence
> even with a previous srandom(biostime(0,0)), but the other one works fine.

Can you show an example of this?  Whichever one does this, it's a bug.

> And it's a very good random generator : according to Bjarne Stroustrup
> (C++'s creator), low-order bits of a random generator are often suspect, so
> rand()%n is not a good way of generating a random number between 0 and n-1
> (he tips to use "(double(rand())/RAND_MAX)*n" ). BUT i've tested rand()%n,
> and it gives a very good repartition... So, DJGPP is the best C
> implementation!!! ;)

In theory, `random' should be a better RNG, especially if you use some
of its more complex modes (see the source).  `rand' is just a linear
congruential RNG, though a good one.

However, `rand' is ANSI and `random' isn't.  (The distinction between
their algorithms is brought about by some obscure legal issues.)
-- 

Nate Eldredge
nate AT cartsys DOT com