Article View: gnu.gcc.help
Article #51699Re: Floating point differences
From: Richard Stallman
Date: Wed, 14 Apr 1993 19:18
Date: Wed, 14 Apr 1993 19:18
17 lines
1607 bytes
1607 bytes
This is disturbing news, but I think there's no way to find out why it happens except by figuring out where in the program the execution diverges. It's possible you'll find a GCC bug. It's also possible that you'll find an undefined C construct which does different things in different compilers. For example, you might find an expression which increments a variable and uses that variable another time, such as `foo (*p, *p++)'. There is no way a priori to guess the explanation. It would be pure luck for test cases to help find this; you will probably get reports of various "failures" in various compilers, and then there will be no way to draw any conclusion from them. The easiest way to track down where in the source code the difference occurs is by running two executables that give different results, both under debuggers, then doing a binary search with breakpoints to find the place where the difference happens.
Technical Details
Message-ID:
<9304141725.AA02954@mole.gnu.ai.mit.edu>Path:
arc1-ca.pugleaf.net!gmd.de!Germany.EU.net!mcsun!uknet!doc.ic.ac.uk!agate!spool.mu.edu!uwm.edu!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!gnu.ai.mit.edu!rmsReferences:
<1qhbioINN55a@fmsrl7.srl.ford.com>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 11318c,e3259ca760e1e814 X-Google-Attributes: gid11318c,public X-Google-ArrivalTime: 1993-04-14 17:10:49 PST Path: arc1-ca.pugleaf.net!gmd.de!Germany.EU.net!mcsun!uknet!doc.ic.ac.uk!agate!spool.mu.edu!uwm.edu!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!gnu.ai.mit.edu!rms From: rms@gnu.ai.mit.edu (Richard Stallman) Newsgroups: gnu.gcc.help Subject: Re: Floating point differences Date: 14 Apr 1993 15:18:16 -0400 Organization: Gnus Not Usenet Lines: 18 Sender: daemon@cis.ohio-state.edu Distribution: gnu Message-ID: <9304141725.AA02954@mole.gnu.ai.mit.edu> References: <1qhbioINN55a@fmsrl7.srl.ford.com>