Message-Id: <200005291856.VAA13282@mailgw1.netvision.net.il>
Date: Mon, 29 May 2000 21:55:43 +0200
X-Mailer: Emacs 20.6 (via feedmail 8.1.emacs20_6 I) and Blat ver 1.8.5b
From: "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
To: djgpp AT delorie DOT com
In-reply-to: <cs95jscumbafalh5jj0f8l6dm1u0uiq5o5@4ax.com> (message from Damian
	Yerrick on Mon, 29 May 2000 17:30:06 GMT)
Subject: Re: Random numbers
References: <392CB951 DOT 34B432E6 AT chemistry DOT uq DOT edu DOT au> <Pine DOT SUN DOT 3 DOT 91 DOT 1000525084705 DOT 21360M-100000 AT is> <cs95jscumbafalh5jj0f8l6dm1u0uiq5o5 AT 4ax DOT com>
Reply-To: djgpp AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> From: Damian Yerrick <Bullcr_pd_yerrick AT hotmail DOT comRemoveBullcr_p>
> Newsgroups: comp.os.msdos.djgpp
> Date: Mon, 29 May 2000 17:30:06 GMT
>
> int random_number =
>     low + (double)rand () * (high - low + 1) / RAND_MAX;
> 
> When I do this on my target platform (386 PC), the cast to double
> produces SIGNOFPE.  Including emu387.dxe and emu387.bat (a script
> to insert the required environment variables) fixed it.
>   @ECHO OFF
>   SET 387=N
>   SET EMU387=C:\PINEIGHT\EMU387.DXE
> 
> Does this mean I have to bundle emu387.dxe with all programs that use
> random numbers?

The emulator is not required for the RNG, it is required for FP
operations.  The fragment from the FAQ uses a cast to double, to force
FP computations: that's what triggers SIGNOFP.

> Or can I do this to get rid of the unrandom low-order
> bits?
>   nextPiece = (rand() >> 4) % 5;

You are assuming that bits 4 to 6 are random enough.  I don't think
you can do that without testing.