Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utcs!mnetor!seismo!caip!clyde!cbatt!cbosgd!ihnp4!inuxc!pur-ee!uiucdcs!okstate.UUCP!uokvax.UUCP!emjej
From: emjej@uokvax.UUCP.UUCP
Newsgroups: net.lang.c
Subject: Re: PRECEDENT FOR USE OF =
Message-ID: <3000080@uokvax.UUCP>
Date: Thu, 10-Jul-86 10:08:00 EDT
Article-I.D.: uokvax.3000080
Posted: Thu Jul 10 10:08:00 1986
Date-Received: Sat, 12-Jul-86 22:40:15 EDT
References: <1645@brl-smoke.ARPA>
Lines: 26
Nf-ID: #R:brl-smoke.ARPA:1645:uokvax.UUCP:3000080:000:1294
Nf-From: uokvax.UUCP!emjej    Jul 10 09:08:00 1986


/* Written  8:15 pm  Jul  3, 1986 by MEYER@RADC-TOPS20.ARPA in net.lang.c */
     I know this is going to cause a bit more of a stir here, here goes.

Everybody keeps guessing at the reasons for the use of = for asignment
and == for equality tests.  How about a different angle: how many times
in a program do you assign something versus how many times you compare things?
/* End of text from net.lang.c */

If that's a reasonable principle of programming language design, then
let's see if we can't get Huffman-coding for C keywords in the next ANSI
draft standard.

= for assignment is one of many flaws in C syntax.  Combined with the
absence of a Boolean type, it is a considerable source of errors in C
code.  Lint should warn the user of every occurrence of "if (a = b)".

(I predict that at least one C-worshipper will say "*real* C programmers,
as opposed to quiche-eaters, don't make that mistake."  I think they're
wrong, based on the instances of it I've seen.  Second-order comment
will be "you haven't done a study, so you're not worth bothering with":
I admit that I haven't, but (1) I lack the resources, and (2) it's too
bad the designers of programming languages (and Unix utilities) don't
have to do human factors studies before introducing their products.)

					James Jones