From:	Wojciech Galazka <wgalazka AT chem DOT uw DOT edu DOT pl>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: Floating Point Exception
Date:	Wed, 20 Aug 1997 13:40:56 +0200
Organization:	University of Warsaw
Lines:	33
Message-ID:	<33FAD7C8.72E7@chem.uw.edu.pl>
References:	<01bcabc3$a01da100$5af382c1 AT damien> <33F8549A DOT 4CB5 AT cornell DOT edu> <01bcac2d$edf69fc0$LocalHost AT damien> <33FACBA4 DOT 1DFF7DBB AT Mathematik DOT TU-Chemnitz DOT DE>
NNTP-Posting-Host:	animal.chem.uw.edu.pl
Mime-Version:	1.0
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Robert Hoehne wrote:
> 
> Rich Birch wrote:
> >
> > Here's the code that I am sure is producing the crash (remember it doesn't
> > crash in Rhide) :-
> >
> > float Function::Result(float x) {
> >        float result = 0;
> >        for (int count = 0; count < num_results; count++) {
> >            if (form.Denominator[count])  //so not to divide by zero

From my good old FORTRAN experience it semms that 
basicly it's a very bad idea to write if(!foo)
when foo is a floating point number. You'd be
better off writing 
if (abs(foo) > SOME_SMALL_VALUE)
assuming abs is a function returnig the absolute value of a variable
and SOME_SMALL_VALUE is something really small, say 10^(-10)

Other than this the code is perfectly legal

> >               result += (form.Numerator[count]
> >                             / form.Denominator[count])
> >                             * int_power (x, form.Power[count]);
> >        }
> >        return(result);
> > }
> 

-- 
Wojciech Galazka
Computer Security Advisor

- Raw text -