Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/17/84; site abic.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!bellcore!decvax!cwruecmp!abic!jst
From: jst@abic.UUCP (Shack Toms)
Newsgroups: net.lang
Subject: Re: Integer division
Message-ID: <752@abic.UUCP>
Date: Fri, 7-Mar-86 18:07:40 EST
Article-I.D.: abic.752
Posted: Fri Mar  7 18:07:40 1986
Date-Received: Sat, 8-Mar-86 23:49:23 EST
References: <11603@ucbvax.BERKELEY.EDU> <4917@alice.UUCP>
Organization: Allen-Bradley Co., Highland Heights, OH 44143
Lines: 40

> > In article <736@abic.UUCP> jst@abic.UUCP (Shack Toms) writes:
> > >> In article <731@abic.UUCP> jst@abic.UUCP (Shack Toms) writes:
> > >> >However:  One might use a%b<0 iff a<0 in an algorithm which printed
> > >> >the value of an integer in a given radix.  The least significant
> > >> >digit of a in radix b would then be |a%b|.  :-)
> > >> 
> > >> So would |a|%b, and it works under either convention.  :-)
> > >
> > >Except that |a| is not available for the full range of a.  In
> > >particular, on a 16 bit computer |-32768| is not expressible.
> > 
> > I don't know about you, but I'm too paranoid about the least negative
> > number to begin with.  In fact, I try not to get close, if possible.
> > 
> > Frankly, if it's a question of a language getting integer division done
> > correctly on -32767 to 32767 or getting it done incorrectly on -32768
> > to 32767, I think there is no debate about which is preferable.  Or,
> > to express my point in less prejudicial terms: in arguing A vs B in a
> > language, the question of what happens with the least negative number 
> > is almost always irrelevant.
> > 
> > ucbvax!brahms!weemba	Matthew P Wiener/UCB Math Dept/Berkeley CA 94720
> 
> In my experience, people start caring about least negative numbers when
> they want to use these weird values in the form:
> 
> #define		CantHappenFlag -32768
> 
> so that they can test for absurd values.  Instead, these people should be
> using enumerated types.  (I will admit to doing things like this before
> some gave me a C book with enumerated types described in it.)

In my experience, people care about least negative numbers any time they
define a function which requires an integer argument which can meaningfully
be from the full range.  This could be anything from a random number generator
(which requires a seed) to an integer modulus function :-).  I don't
believe I have ever used them as a CantHappenFlag,  it is generally
too easy to come across them as the result of ordinary computation.

Shack Toms