*BSD News Article 14640


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!think.com!enterpoop.mit.edu!ai-lab!hal.gnu.ai.mit.edu!not-for-mail
From: mycroft@hal.gnu.ai.mit.edu (Charles Hannum)
Newsgroups: comp.os.386bsd.questions
Subject: Re: Problems with g++2.3.3 and leda
Date: 18 Apr 1993 04:02:38 -0400
Organization: dis
Lines: 17
Distribution: world
Message-ID: <1qr1uu$1hck@hal.gnu.ai.mit.edu>
References: <1qjd46$vc4@obelix.uni-muenster.de>
NNTP-Posting-Host: hal.ai.mit.edu
Keywords: g++2.3.3, leda


In article <1qjd46$vc4@obelix.uni-muenster.de>
frido@math.uni-muenster.de writes:
>
> After debugging the programm with gdb I found that the line where the
> exception happens seems quite harmless. It is a comparison of a
> double variable and zero (double l; if (l==0) ...).

Unfortunately, floating point exceptions are not precise; i.e. the
processor is not left in the state it was at the time the instruction
was executed.  The bug occured in the *previous* floating point
operation, which may be an entirely different piece of code.

-- 
 \  /   Charles Hannum, mycroft@ai.mit.edu
 /\ \   PGP public key available on request.  MIME, AMS, NextMail accepted.
Scheme  White heterosexual atheist male (WHAM) pride!