Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!apple!rutgers!att!cbnewsl!cbnewsk!pegasus!hansen
From: hansen@pegasus.att.com (Tony L. Hansen)
Newsgroups: comp.lang.c++
Subject: Re: Lint for C++
Summary: yes, there are some things a lint++ can do for us
Keywords: C++, lint
Message-ID: <1990Dec20.042841.16639@cbnewsk.att.com>
Date: 20 Dec 90 04:28:41 GMT
References: <93Rcu1w163w@turbo.atl.ga.us> <CIMSHOP!DAVIDM.90Dec19125537@uunet.UU.NET>
Sender: hansen@cbnewsk.att.com (tony.l.hansen)
Distribution: comp
Organization: AT&T Bell Laboratories
Lines: 25

< Isn't the C++, itself, enough of a LINT program??  LINT is usually used
< to detect potential problems in the source code that the compiler would
< not normally check for.  C++, though, seems to check for everything
< that a LINT program would, so what's the need?

Off the top of my head, here are some of the things that a lint++ could do
that C++ is incapable of doing:

    o	check the types of the arguments to printf- and scanf-like functions
    o	know about functions which never return
    o	check for falling through switch cases
    o	look for external symbols that could be declared static
    o	find unused external variables

These are all capabilities of the version of lint in System V release 4.
Most capabilities cannot be folded into the C++ compiler.  It might be
possible to introduce some of them through language modifications.  Others
might possibly be introduced through #pragma's. But it's also unclear whether
any of these SHOULD be caught by the compiler.

So, yes, there is a need for a lint++.

					Tony Hansen
				att!pegasus!hansen, attmail!tony
				    hansen@pegasus.att.com