(fwd) article on open source

[rlk at alum.mit.edu (Robert L Krawitz)]

   From: "Jerry Feldman" <Gerry.Feldman at compaq.com>
   Date: Mon, 27 Mar 2000 09:57:01 -0500

   While certainly not open source, when at Raytheon this past year, I had 
   to estimate the number of lines in the HP-UX driver I was writing because 
   all the code metrics on the project were based on number of lines of 
   code (this was mandated not by Raytheon, but by the prime contractor, 
   and probably by the Army in this case).

I once had a manager who used lines of code as a marketing tool,
believe it or not.  He asked me to count the lines of code in
different versions of our product (a math library), and then he put
those numbers on a slide to show the sales force how much
functionality the product had gained!  I don't know if customers ever
saw this.  The other engineer (as opposed to a bunch of scientists
with no software engineering background) and I tried to talk him out
of it, but to no avail.  Evidently it was more compelling to do this
than to plot, say, the number of user entrypoints (even that would
have made more sense).  I guess that by showing our library had 500K
lines of code he could demonstrate how big and important his project
was.

Lines of code are a useful zeroth-order approach to estimating
complexity and how much effort will be required to support something.
Even accounting for differences in style, a 50KLOC program almost
always will be faster to develop and easier to maintain than a 500KLOC
program in the same language and in the same field of endeavor.  It's
completely worthless for anything else.  Using it as any part of a
compensation package severely distorts the incentives away from
producing any kind of quality product.
-
Subcription/unsubscription/info requests: send e-mail with
"subscribe", "unsubscribe", or "info" on the first line of the
message body to discuss-request at blu.org (Subject line is ignored).