David Weller wrote:
>
> > The study is extremely accurate, and although a bit old, I heard that Reifer
> > later made new measures that simply confirmed his first findings.
>
> Let me make a comment to the contrary:  Reifer's latest study
> concluded that C++ may give almost as much (if not more) productivity
> as Ada 83.  I believe he deferred judgement on Ada95, claiming lack of
> sufficient empirical evidence.  I lack a sufficient reference, but it
> should be found in the c.l.a archives (through DejaNews) or the Team Ada
> archives.  If I recall, his study basically concluded that, provided
> you selected a "modern" programming language that supported things
> like modularity, OO features, and system interfacing, your language
> choice didn't yield sufficient variances.

Let me add my own comments, which probably back up what David is saying:

Last August, I emailed Reifer the following:

Subject:
            "Quantifying the Debate: Ada vs. C++" - July 1996 Crosstalk
       Date:
            Wed, 28 Aug 1996 10:30:35 -0500
       From:
            James Squire <[log in to unmask]>
Organization:
            McDonnell Douglas Aerospace
         To:
            [log in to unmask]


Donald,

I had a couple of questions for you in response to your article:

1.  "Table 2.  Dollar Cost per Delivered Source Line of Code (1995)" -
does this cost include the costs up to delivery or does it also include
the cost of maintaining it post-delivery?

2.  Why didn't you choose to compare Ada 83 vs. C and Ada 95 vs. C++
instead of Ada against both C and C++, which effectively becomes Ada 83
against both since there is not enough data yet on Ada 95?  You compare
C++ favorably to Ada 83 when the former is an OOP and the latter is not.
 Isn't the jury really out on C++ too, until there is enough Ada 95 data
to compare it with?

Thank you.

His response to 1) was:  "These are end-to-end development costs.  They
do not include maintenance."

His response to 2) was:  "That is how the data fell out.  No ulterior
motive."

My reaction to his response to 1) was:  "Then I have an observation:  In
Tables 3 and 4 (which both pertain to post-delivery maintenance), Ada 83
fares better than C (and even C++ where there is data; the one data
point
available on Ada 95 fares worse than C++).

"Would not this mitigate (and potentially overtake) your conclusion that
"C and C++ have become more cost-effective"?  At least for C, with the
jury still out on C++?  Maintenance costs are just as important as
development costs, aren't they?  Perhaps you did not have a way to
calculate this, but I'm surprised you didn't mention this in your
article and hold back on the above conclusion."

My reaction to his response to 2) was:  "I didn't mean to suggest any
ulterior motive, but can you say more about "that's how the data
fell out"?  I'm not sure I meant the data anyway, I think I was
referring to the title of the article."

Unfortunately I got no response to this followup.

I was quite unsatisfied with his response, and for the reasons stated
here (which were not answered adequately), I consider his study to be
severely flawed.
--
James Squire           mailto:m193884 no junk mail
[log in to unmask]
MDA Avionics Tools & Processes
McDonnell Douglas Aerospace              http://www.mdc.com/
Opinions expressed here are my own and NOT my company's
"Would you prefer to be conscious or unconscious during the mating? I
would
 prefer conscious, but I don't know what your...pleasure threshold is."
        -- G'Kar (to Lyta Alexander), "The Gathering"