lclint-interest message 78

From evans@cs.virginia.edu Mon Jun  3 12:30:37 1996
Date: Mon, 3 Jun 96 12:03:44 -0400
From: evans@cs.virginia.edu (David Evans)
To: lclint-interest@larch.lcs.mit.edu
Cc: apang@mindlink.bc.ca
In-Reply-To: Anthon Pang's message of Thu, 30 May 96 00:01:52 -0800 
Subject: Subscription request


Sorry for the slow response...I was out of town last week.

A few things on your wish list are currently done by lclint:

> Let's see...what would be on my wish list for lclint?  How about:
> - being able to parse C++ files

Nope, sorry.  It would be very useful to see a lclint-like tool
developed for C++, but there are no immediate plans to do this here.

> - adding some simple software metrics, eg LOC and cyclomatic complexity

Again, would be worth doing, but not likely anytime soon.  There are
lots of other tools available that do these kind of metrics.  I believe
some perhaps more revelant metrics could be derived that take advantage
of some of the extra information used in lclint checking, but don't have
a clear idea on what kind of metrics would really be useful.

> - integrating Unix portability checks -- autoconf, meta-config

The standard library can be used to check conformance to library calls,
but the problem with porting unix applications is often that the
standard library is not supported (at least, in my experience, this is
usually by things like autoconf and configure are needed).  I'm not sure
how lclint would be useful in checking this, other than providing
smaller subsets of the libraries that more platforms are believed to
support.

> - value tracking (ala PCLint 7)

LCLint does something similar for null pointers (i.e., figures out at a
given point in the code if a pointer may be null), but doesn't do
anything to try to approximate the ranges of numerical values at compile
time --- its not clear to me exactly what PCLint does here, other than
what they show in the examples (which seems to be limited to simple, but
very useful things, like knowing the range of values a loop bound can
hold).

> - support for traditional lint annotations (if not already), eg /*NOTREACHED*/

These are supported already.  See
http://lclint.cs.virginia.edu/guide/appE.html
("Traditional Lint Comments") for details.

--- Dave