Re: Gauntlet gathering rust and dust

[cis.ohio-state.edu!magnus.acs.ohio-state.edu!math.ohio-state.edu!darwin.s ura.net!source.asset.com!vand@ucbvax.Berkeley.EDU (Laurence VanDolsen)]

In article <1993Jul27.184346.24663@iplmail.orl.mmc.com> jcrigler@theopolis.orl.
mmc.com (Jim Crigler) writes:
> 
>Agreed that published stats seem cockeyed (listening, Greg?  Keep
>listening...), but what % is classified?  (That statistic is probably
>classified, too :-) But I have a feeling that it's probably < 25%.
>Anybody got real numbers?

Although classified projects probably do contribute something, the mix
of languages used in such projects is probably about the same as that
used in non-classified DoD work.  It amy even be argued that it is
easier, and therefore more likely, to violate the mandate in a
classified environment.

What REALLY skews the numbers is that nobody is counting.  The vast
majority of the 'statistics' are just estimates.  

"What is your source for that revelation?", he asked.

In over thirty years of software and system engineering for the DoD,
involving work on projects totalling several millions of lines of code,
I have never, repeat NEVER, been associated with a contract which
required that we report Lines of Code.  We have been asked, not
required, to provide estimates, wild-assed-guesses, source-listing-
page-counts with an estimate of LOC/page, person-months with an estimate
of LOC/person-month, even a count of the line-feeds in the source files,
but NEVER have we been asked to provide a certifiable count of LOC
against an official definition of a LOC.

I am sure that there are many contracts which DO require such reporting,
but my experience indicates that there are many which do not.  Given the
absence of data, how can one draw ANY conclusions?

Of course, finding out anything useful about the size of commercial
software must be just as difficult. :-)

--
Laurence L. Van Dolsen - Der fliegender Hollander
My opinions are my own, but you are welcome to them.
Paramax - (805) 987-9302 - vandolsen@cam.paramax.com