Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!rochester!ritcv!ccivax!rb
From: rb@ccivax.UUCP (rex ballard)
Newsgroups: net.lang.c,net.arch
Subject: Integer division
Message-ID: <387@ccivax.UUCP>
Date: Mon, 10-Feb-86 22:37:37 EST
Article-I.D.: ccivax.387
Posted: Mon Feb 10 22:37:37 1986
Date-Received: Fri, 14-Feb-86 05:53:46 EST
References: <731@abic.UUCP> <4947@alice.uUCp>
Reply-To: rb@ccivax.UUCP (What's in a name ?)
Organization: CCI Telephony Systems Group,  Rochester NY
Lines: 27
Xref: linus net.lang.c:7306 net.arch:2317
Summary: FORTH-83 better?

As I recall, FORTH-83 has adopted the 'theorists' approach.
so that
A B SWAP OVER /MOD ROT * +
gives the original value of A
translated that gives
A B SWAP	( Stack = B A )
OVER		( Stack = B A B )
/MOD		( Stack = A%B A/B B ) (A/B is floored)
ROT		( Stack = A/B B A%B )
*		( Stack = (A/B)*B A%B )
+		( Stack = (A/B)*B+A%B ) (=A)

This was a deliberate break from FORTH-79 which would have required
the 'twiddle test'.  Appearently, most applications either were
using the unsigned versions, the positive case, or had rewritten the
primitive.

the example (taken from the draft standard of May 83)

	dividend	divisor		remainder	quotient
	10		 7		3		1
	-10		 7		4		-2
	10		 -7		-4		-2
	-10		 -7		-3		1

Is this any better?
I guess religion (because thats the way we always do it) isn't everything.