Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site calgary.UUCP
Path: utzoo!utcsri!ubc-vision!alberta!calgary!radford
From: radford@calgary.UUCP (Radford Neal)
Newsgroups: net.lang.c,net.arch
Subject: Re: Integer division
Message-ID: <114@calgary.UUCP>
Date: Tue, 4-Feb-86 19:16:02 EST
Article-I.D.: calgary.114
Posted: Tue Feb  4 19:16:02 1986
Date-Received: Wed, 5-Feb-86 09:23:13 EST
References: <332@ism780c.UUCP> <11603@ucbvax.BERKELEY.EDU>
Distribution: net
Organization: University of Calgary, Calgary, Alberta
Lines: 14

First, let me say that I'm firmly in the (-3)%2 == 1, (-3)/2 == (-2) camp. 
This is normally what one wants.

There is one case, however, where the (-3)/2 == 3/(-2) == -(3/2) == -1 
identity is usefull - writing software floating point routines. I offer this
as a suggestion as to why the initial "mistake" was made. 

Now that we "all" have hardware floating point, can we change divide?

Actually, I'd be satisfied if people would at least *document* what their
divide operation does! (E.g. the MC68000 processor manual just says "the
division is done using signed arithmetic"...)

     Radford Neal