From owner-freebsd-questions  Mon Aug 26 01:17:18 1996
Return-Path: owner-questions
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id BAA26661
          for questions-outgoing; Mon, 26 Aug 1996 01:17:18 -0700 (PDT)
Received: from eac.iafrica.com (196-7-192-206.iafrica.com [196.7.192.206])
          by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id BAA26656
          for <questions@FreeBSD.org>; Mon, 26 Aug 1996 01:17:05 -0700 (PDT)
Received: (from rnordier@localhost) by eac.iafrica.com (8.6.12/8.6.12) id KAA02909; Mon, 26 Aug 1996 10:12:52 +0200
From: Robert Nordier <rnordier@iafrica.com>
Message-Id: <199608260812.KAA02909@eac.iafrica.com>
Subject: Re: Programming Question
To: hoek@freenet.hamilton.on.ca
Date: Mon, 26 Aug 1996 10:12:51 +0200 (SAT)
Cc: questions@FreeBSD.org
In-Reply-To: <199608251921.PAA01735@james.freenet.hamilton.on.ca> from "hoek@freenet.hamilton.on.ca" at Aug 25, 96 03:21:24 pm
X-Mailer: ELM [version 2.4 PL24 ME8a]
Content-Type: text
Sender: owner-questions@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

hoek@freenet.hamilton.on.ca wrote:
  
> It's been suggested before that this area is as good as any for asking 
> questions WRT to a certain aspect of one of the supplied compilers.  
> While I'm not certain I agree, here goes...
> 
> Is there any way to make the C compiler use much stronger typing.  For 
> example, I would really like it to warn about assigning signed variables 
> to unsigned ones.  (I say this after spending way too much time hunting 
> down an unnecessary bug caused by the lack of such a warning).
> 
> Heck, optimally, I sure wouldn't mind if it were to give me a warning when
> assigning a variable of type TYPDEF to another identical variable declared
> without the TYPDEF'd definition.
> 
> For example,
> 
> --
> signed int a;
> unsigned int b;
> 
> typedef int AT;
> 
> AT aa;
> int bb;
> 
> void
> main () {
> 	a = b;
> 	aa = bb;
> }
> --
> 
> would yield two warnings.
> 
> I've tried just about every -W option listed, but none of them seem to 
> even come close.  Is there anyway to massage the compiler into 
> complaining about these things?

Note that 'typedef' doesn't create types, just synonyms for existing
types.  So the assignment <AT> = <int> doesn't involve a type
conflict in any sense.

For automated nitpicking, I like 'lclint' (from the 'lcc' compiler
people):

| LCLint 2.0 --- 21 Mar 96
|
| x.c:10,1: Function main declared to return void, should return int
|   The function main does not match the expected type.  Use -maintype to
|   suppress message.
| x.c: (in function main)
| x.c:11,2: Assignment of unsigned int to int: a = b
|   Types are incompatible.  Use -type to suppress message.
|
| Finished LCLint checking --- 2 code errors found

--
Robert Nordier