Re: why? US Army Ada software converted to C++?

[David Emery <demery@cox.rr.com>]

In article <9pnumq02o3u@drn.newsguy.com>, <rob@NOTHERE> wrote:

> Air defense moving to C++ from Ada? are they crazy? with
> all the evidence that Ada is safer and more reliable than any C variation
> based language, who would make such a decision? Our tax payers money
> is being wasted here. What a waste. Is this supposed to make our defense
> software better?  

One of the most common reasons I hear for contractors moving from Ada is
that they can't find Ada programmers.  My own hypothesis is that they're
not willing to pay enough.  Anecdotally, I know a lot of Ada people that
make -much more- money doing C (because they're generally substantially
better than their peers who have only C experience.)  Of course, though,
if they do not advertise for Ada programmers, they are unlikely to get
any :-(

But in general language/sw technology decisions are not based on any
particular engineering analysis, but rather business case and
perceptions of what the customer wants to hear and what the managers
believe will be most profitable/least cost.  At the risk of raising
old ghosts, DoD is guilty, in my view, for not performing the full
life-cycle analysis that was possible about 7 or 8 years ago, when Mr.
Emmett Paige held his "Ada Dual-Use Symposia".  I said then, and I still
believe, that the DoD has sufficient data to determine if Ada does 
provide life cycle (or even purely development) cost savings.  The
only study I saw on Ada maintenance (done in 1990 with preliminary
data from a few programs) said that Ada costs were _LINEAR_ on SLOC
counts.  COCOMO shows the cost to be _EXPONENTIAL_ for other languages.

               dave