lclint-interest message 161
From email@example.com Tue Sep 23 11:12:49 1997
References: <9709221936.AA04593@larch.lcs.mit.edu>; from David Evans on
Mon, Sep 22, 1997 at 03:36:00PM -0400
Content-Type: text/plain; charset="us-ascii"
Date: Tue, 23 Sep 1997 11:08:57 +0100
To: Harald Kirsch
From: Richard Brooksby
Subject: Re: How to annotate
Harald Kirsch wrote:
> On Mon, Sep 22, 1997 at 03:36:00PM -0400, David Evans wrote:
> > Richard Brooksby's point that the code should be rewritten not annotated
> > is a good one. ...
> I must admit that I was first a bit offended by Richard Brooksby's
> remark, ...
I apologise. My mail was too short. I should have explained myself. I
have just returned from holiday and have a very large pile of mail to get
through. It tends to make me terse.
> ... I then reconsidered the code and found that rewriting made it
> in fact less complex and, as you say, easier to understand and maintain.
This illustrates the important difference between defect _detection_ and
To merely detect defects you can use LCLint to make you double-check your
code, make you think harder about tricky constructs. You then supress the
warnings when you are satisfied that your code is correct. This will help
you find defects.
However, to realize longer term benefit from LCLint you should modify your
conventions so that such warnings do not occur. This will reduce the
number of defects that you introduce into your code in the first place.
It is _much_ cheaper to avoid introducing defects than to remove them
later. The later your remove them, the more they cost. The greatest cost
is when they have reached the end-user.
Now, in this case there was no defect, but it cost you quite a bit of
effort to make sure. It would also cost everyone else who has to maintain
the code a similar amount of effort. LCLint is trying to tell you that.
You will know when LCLint has been most effective in your development
process when it no longer produces output.
Leader, Manager, and Developer / Memory Management / Harlequin
Tel. Int. +44 1954 785451 (voice) +44 1954 785444 (fax)
University of Virginia, Computer Science