lclint-interest message 40

From evs Mon Mar  4 18:20:33 1996
Date: Mon, 4 Mar 96 18:16:28 -0500
From: evs (David Evans)
To: lclint-interest@larch.lcs.mit.edu
Cc: mathieu@hptnos04.grenoble.hp.com
In-Reply-To: Mathieu Antoine's message of Mon, 04 Mar 96 17:11:08 +0100 <199603041611.AA137305874@hptnos04.grenoble.hp.com>
Subject: Looking for LCLint success stories


Hello Mathieu,

I don't think any quantitative studies exist but there are some
anecdotal reports --- I hope readers of this list may be able to provide
you with more recent accounts, too.

Currently, I'm the only one who has really used LCLint 2.0 (which was
publically released last Friday.)  You can find an account of my
(perhaps somewhat biased, of course) experiences using LCLint on its own
source code in Section 7 of

  David Evans.  Static Detection of Dynamic Memory Errors.  To appear in
  SIGPLAN Conference on Programming Language Design and Implementation
  (PLDI '96), Philadelphia, PA, May 1996.
  
  This is availble under Publications on the LCLint home page, or at
  http://www.sds.lcs.mit.edu/~evs/pldi96-abstract.html.

An account or an industrial use can be found in
  http://lclint.cs.virginia.edu/lclint-interest/7.html

>
> - estimate the amount of effort required to introduce LCLint into a software
>   development process for a CMM Level 1 organization:
> 
> - estimate the return on investment 

LCLint is designed so that its use can be introduced gradually.
Programmers should be able to start using it in -weak mode on existing
code with very little effort.  As it is used more, different flag
settings can be used to select stricter checking and enforce stricter
coding standards.  If more effort is spent adding source-code
annotations or writing specifications, significantly better checking can
be done.

Its too soon to have enough experience with LCLint 2.0, but I hope other
readers of this mailing list will be able to provide some more concrete
reports on the costs and benefits of using LCLint in an industrial
setting.

--- Dave