From: Sengan DOT Short AT durham DOT ac DOT uk
Message-Id: <25900.9605051801@bylands.dur.ac.uk>
Subject: Re: libm.a/linker *BUG* with test program
To: ao950 AT FreeNet DOT Carleton DOT CA
Date: Sun, 5 May 1996 19:01:36 +0100 (BST)
Cc: djgpp AT delorie DOT com
In-Reply-To: <4me6om$5md@freenet-news.carleton.ca> from "Paul Derbyshire" at May 4, 96 00:01:26 am
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

> This may be a stupid question but: why is it that when the math library is
> not on the command line, it not only links and resolves the calls to pow
> and ldexp, but it *works* too???
> And, of course, if this program works without the library and crashes with
> it, maybe you might try just not using the library for this situation...
> 
> --
>     .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
>  -()  <  circles, and bark is not smooth, nor does lightning travel in a
>     `*'  straight line."    ,------------------------------------------------
>          -- B. Mandelbrot  |  Paul Derbyshire (PGD) ao950 AT freenet DOT carleton DOT ca
> 


There appears to be a bug in the cygnus math library as far as pow and ldexp
go. The solution is to use the functions in the standard libaries instead. They
appear to work. (ie pow and ldexp are in two different libraries, and you can
link in one or the other)

Sengan