Anyone have any first-hand experience with Gimpel's PC-Lint (or any
alternatives?)  I've been thinking about getting it, but would like to hear
any feedback before I plunk down $240.00 I can ill-afford to spend.  I'm
thinking it will be useful for tightening up my code.

    Thanks!
    C.
(to reply via email, remove the 'NS' in the address)

It's worth taking a look at www.splint.org.

Andy

I used many years ago with 16-bit development.  It had good info.

It's excellent.  It's saved my {*filter*}many many times.  The great thing about
it is that you can set it to be as picky or forgiving as you like so you can
apply it to your existing code base without a huge barrage of complaints.
You can set it to be picky for new projects to keep them nice and clean. It
also comes with config scripts for a lot of popular coding standards so
teams can standardise on what they will or won't accept in the code with
ease.  It also integrates well with most compilers, MS included (don't know
if there's a .NET integration available yet).  The support guys are pretty
responsive as well.  The only caveat is that you should 'Be afraid, be very
afraid'.  You'll almost certainly be terrified by the amount of stuff it
turns up in existing code. ;o)

Mike H.

Tue, 3 Sep 2002 16:01:07 in microsoft.public.vc.language, "David Crow

I have been using the product for more than 10 years.
I use it for C; I have little experience of using it with C++.
I am a Beta tester and no longer pay money for the product.
I find it ESSENTIAL. It takes time to get the best from it. The
demonstration below shows a class of error I used to have.

c:\WFB> cl -nologo -W4 badscan.c
badscan.c

c:\WFB> type badscan.c
/*
 * badscan.c - scanf argument inconsistency demonstration
 */

#include <stdio.h>

int main(void) {
  int i = 0;

  scanf("%i", i);
  printf("%d");
  return 0;

c:\WFB> cl -nologo -W4 badscan.c
badscan.c

c:\WFB> lin badscan

c:\WFB> c:\lint\lint-nt +v -ic:\lint\ std.lnt c:\wfb\bin\c0-delta.lnt
+os(_lint.tmp)  badscan
PC-lint for C/C++ (NT) Ver. 8.00g, Copyright Gimpel Software 1985-2002
--- Module:   badscan.C
c:\lint\lint-nt +v -ic:\lint\ std.lnt c:\wfb\bin\c0-delta.lnt
+os(_lint.tmp)  badscan

--- Module:   badscan.C
  scanf("%i", i);
                ^
badscan.C(10) : Warning 560: argument no. 2 should be a pointer
  printf("%d");
              ^
badscan.C(11) : Warning 558: Too few arguments for format (1 missing)

^
badscan.C(13) : Note 953: Variable 'i' (line 8) could be declared as
const
---
PC-lint for C/C++ output placed in _LINT.TMP

c:\WFB>

There are free alternatives; splint from http://www.splint.org is one.
Earlier this year, I put a lot of work into using it as I was hoping
that its function semantic descriptions would give me benefits
additional to those from PC-Lint. I found it unreasonably difficult to
silence the spurious diagnostics. There is a code corresponding to each
diagnostic which suppresses that diagnostic. PC-Lint uses error numbers
and supports wild cards for diagnostic suppression. For example, I could
(perversely) suppress the warnings above with lin -e5* badscan
YMMV. Look further before spending your 240USD at http://www.gimpel.com
--
Walter Briscoe