*BSD News Article 11203


Return to BSD News archive

Received: by minnie.vk1xwt.ampr.org with NNTP
	id AA1385 ; Tue, 23 Feb 93 14:38:47 EST
Newsgroups: comp.unix.bsd
Path: sserve!manuel.anu.edu.au!munnari.oz.au!mel.dit.csiro.au!its.csiro.au!dmssyd.syd.dms.CSIRO.AU!metro!ipso!runxtsa!bde
From: bde@runx.oz.au (Bruce Evans)
Subject: Re: 386BSD: cc1 got fatal signal 6
Message-ID: <1993Feb16.215658.29848@runx.oz.au>
Organization: RUNX Un*x Timeshare.  Sydney, Australia.
References: <DERAADT.93Feb11012907@newt.newt.cuc.ab.ca> <C2EwAp.88K@sugar.neosoft.com>
Date: Tue, 16 Feb 93 21:56:58 GMT
Lines: 23

In article <C2EwAp.88K@sugar.neosoft.com> peter@NeoSoft.com (Peter da Silva) writes:
>In article <DERAADT.93Feb11012907@newt.newt.cuc.ab.ca> deraadt@newt.cuc.ab.ca (Theo de Raadt) writes:
>> I've heard a lot of floating point issues discussed on here, but I have
>> never seen a definitive one that declared exactly why this happens.
>> I cannot even compile up gcc2.3.2 with this problem.
>
>> What is the DEFINITIVE answer?
>
>The definitive answer is bad constants in math.h and float.h. The patch

Nope.  The definitive answer is that printf and atof/scanf are inaccurate
and turn good constants into bad ones, and gcc trusts them too much.

>kit we tried to distribute from NeoSoft should fix it. I only hope Nate
>gets our changes integrated or gives us back control so you all can get
>them.

Suitably fudge constants have been posted several times, including once
by Karl.

If you install gcc-2.3.3, don't forget to change DBL_MAX in gcc's float.h.
-- 
Bruce Evans  (bde@runx.oz.au)