Path: utzoo!attcan!uunet!lll-winken!ncis.llnl.gov!helios.ee.lbl.gov!pasteur!ucbvax!ucsd!rutgers!cmcl2!adm!smoke!gwyn
From: gwyn@smoke.BRL.MIL (Doug Gwyn )
Newsgroups: comp.unix.questions
Subject: Re: C and lint: default declarations
Message-ID: <9472@smoke.BRL.MIL>
Date: 21 Jan 89 15:20:17 GMT
References: <501@larry.UUCP>
Reply-To: gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>)
Followup-To: comp.lang.c
Organization: Ballistic Research Lab (BRL), APG, MD.
Lines: 23

In article <501@larry.UUCP> jwp@larry.UUCP (Jeffrey W Percival) writes:
>The question is, do "good" programmers typically take advantage of these
>concessions, or do they assiduously declare their integer functions and
>keep their "scopes" as collapsed as possible?

Both approaches are considered proper, since both remain within the
official definition of the language.

Personally, I declare all externs (not done by headers) just after
the inclusion of headers near the start of the source file.
Occasionally this catches a bug that would otherwise be hard to find.
More importantly, it documents the interdependency between modules.

Usually, when I need some variables with extremely localized scope
and a function call doesn't seem appropriate, I'll declare the
variables in a local block.  Besides protecting against bugs, this
also helps some compilers do a better job of register allocation.

>Does anyone have a lint-like tool that does not pander to these
>"forgiving" scope rules and default declarations?

There are many variants of "lint", but I don't recall hearing of
one that would squawk about legal use of implied function declaration.