Date: Thu, 21 Mar 1996 17:06:30 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Wolfgang Winkler <winkler AT delorie DOT com>
Cc: djgpp AT delorie DOT com
Subject: Re: floating point error
In-Reply-To: <4iolb6$6ph@brachio.zrz.TU-Berlin.DE>
Message-Id: <Pine.SUN.3.91.960321170238.25106M-100000@is>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On 20 Mar 1996, Wolfgang Winkler wrote:

> Can anybody tell me: Is there any workaround or do I have to use djgpp V1
> instead? Is it possible just to replace the math-library of V2 by that of V1?

Try linking without -lm switch.  That will cause gcc to only scan the
standard libc.a which in v2.0 includes the code to all ANSI C math
functions compiled from the sources very close to what in v1.x was in
libm.a.  (In v2.0, libm.a was compiled from an entirely different
sources.)

If that doesn't help, you can either set the x87 to not generate FP 
exceptions (use the `_control87' library function) or install your own 
signal handler for those exceptions (with `signal' library function).