AJPO is still clueless about the reality of DoD Ada rejection

AJPO is still clueless about the reality of DoD Ada rejection

[Gregory Aharonian]

    My recent posting stirred some phone calls.  What a joke.   A programmer
from a DoD contractor tired of the pathetic rationalizations coming out of
AJPO (like that COTS is compatible with the Ada Mandate - what a lie), sent
me the following observation of how programming really gets done inside the
DoD.  The fundamental Ada management policy is still dishonesty - dishonest
analysis of Ada vendors and contractors, dishonest measurement of DoD
language use, dishonest justifications for porky STARS, etc. etc.

                              ====================

Greg, I laugh when I see:

> Ada is still required any time new software is written.

I work on a large military project.   We develop 80% of our new code in
C, 10% in C++, and 10% in Ada (my best estimate).  We have to work lots 
of COTS software into our system and we integrate on Unix platforms.   
That is why we write most of our code in C.   It's a whole lot easier
than retrofitting Ada to a myriad of C interfaces.   Writing Ada bindings
for the C universe and COTS software takes a lot of time and is often an
error prone process, as it is with many foreign language interface endeavors.

The project I work on has 80 software engineers.   Has 2 million lines
of delivered code and is a mission critical DoD system.

It is a perfect candidate for Ada in the DoD mindset.   It is a large
software engineering project.   Ada is avoided with every caveat possible.

We are working in a rapid-prototyping environment where integrating 
3rd Party COTS software quickly is paramount.   Our schedules are very
aggressive.  Using C instead of Ada helps us meet our schedules quickly.
Our C SDEs are far more advanced and robust than our Ada SDEs.  Our Ada
SDE symbolic debugger won't even attach to a UNIX process and the compiler
vendor is a major player in the Ada market.   This is laughable in 1996.

The language lawyers can argue this anyway that they want, but the facts
are the facts.  Software engineering economics prevail.  C is a lot
cheaper, quicker, and more reliable than Ada for this project.

                              ====================

Ada avoided as much as possible.  Ada tools still trailing C tools in 1996.
HEAR THAT AJPO????????????????????????

                              ====================

   Here is what originally provoked the programmer:

Ken Garlington wrote:
> 
> Gregory Aharonian wrote:
> >
> > Reflecting this, in a recently Air Force Scientific Advisory Board report
> > titled "New World Vistas", the panel, along with pushing COTS greatly, also
> > suggested that the Air Force should cease developing software tools and
> > compilers, abandon the DoD's Ada computer language mandate; and depend on
> > aircraft manufacturers to design aircraft cockpits.
> 
> I called Joan McGarity at AJPO (703-681-2463) to find out more
> about the SAB report. Here's what she told me:
> 
> She confirmed that Mr. Paige's office is investigating the
> paragraph in the recent Scientific Advisory Board report
> that was critical of Ada. Both John Goodenough and Larry
> Druffel have indicated that Ada was never discussed by the
> SAB, and that the words in the report were not coordinated
> with the panel prior to publication.
> 
> As a result, I doubt that the SAB report will carry much weight. It
> certainly was not reflected in the new DoD acquision reform direction:
> 
> The new DoDD 5001. and DoDI 5000.2 does not change DoD
> Ada policy. In particular, Ada is still required any time
> new software is written, and DoDD 3405.1 still provides
> the details regarding waivers, etc. There will be a new
> DoDD 3405.1 released very soon (it was held up to allow
> DoDI 5000.2 to be released first), but it will not
> significantly change.

Look, COTS and the Ada Mandate are incompatible, and where they aren't,
Ada is still rejected anyways by DoD programmers.  The Ada Mandate is the
55-mph-speed-limit regulation for DoD programmers.  Sounds nice in theory,
ignored by everyone in practice, because the waiver policy is still a joke.
How can Paige's office expect anyone to take them seriously about enforcing
the Mandate when they can't even track what's written in DoD software reports?

And I don't know what high schools they go trawling through looking for
people to serve on Defense Science Boards, but the only thing worse than
having the SAB panel discuss ditching Ada is that in its discussions of
Air Force software activities (according to Goodenough and Druffel) is that
Ada wasn't discussed at all.


Greg Aharonian

Re: AJPO is still clueless about the reality of DoD Ada rejection

[Ken Garlington]

Gregory Aharonian wrote:

> Look, COTS and the Ada Mandate are incompatible, and where they aren't,
> Ada is still rejected anyways by DoD programmers.

I can't speak for the rest of the world, but I know of a few million lines of
F-22 Ada code that says otherwise.

Re: AJPO is still clueless about the reality of DoD Ada rejection

[Byron B. Kauffman]

Ken, you know we're both on the same side, BUT you and I both know of
cases where more money was spent (in manhours) justifying a Ada waiver
than would have been used if they had just done the Ada like good little
boys and girls. Case in point, a few years ago General Dynamics laid off
a few hundred Ada folks (thank you, Dick Cheney), while the VERY NEXT
WEEK other avionics-related departments were trying to justify an Ada
waiver because they didn't have anyone with Ada experience!!!
Personally, that news didn't sit very well with me while I was standing
in the unemployment line with some very good engineers, and may have
permanently effected my opinion of anti-Ada (alias gutless weenies)
management types...

Please forgive me, but it's a rainy Monday morning and my alma mater got
their butts kicked in the NCAAs.

Is COTS the answer?

[Ted Dennison]

Robert Munck wrote:
> 
> On Fri, 15 Mar 1996 21:06:34 GMT, srctran@world.std.com (Gregory
> Aharonian) wrote:
> 
> > ... (like that COTS is compatible with the Ada Mandate - what a lie)
> 
> I'm beginning to see indications that the DoD fad for COTS software
> is fading, and I'll go out on a limb to predict that there will be an
> anti-COTS backlash by DoD within five years. Reason: COTS software
> may well require more maintenance than home-grown.  See, those
...
> So your nice DoD application, with two or three COTS packages
> and running on MS Windows 3.0 on a 286, needs a HUGE and
> CONTINUOUS upgrading and re-distribution activity.  If it were
...
> So has anyone else seen this happening?

I haven't seen this happen with the DoD yet, but I have to agree with
you that it will. For instance, suppose someone is replacing an AN/UYK
console on an U.S. Navy vessel with a mostly COTS system consisting of
a WindowsNT PC with at COTS GUI driver and some Ada code. The AN/UYK
console it is replacing has been on these ships for over a decade. The
Navy likes to keep their equipment around for 20 years or more. 

Now you tell me, what are the odds that in 15 years when this thing 
breaks the Navy will be able to go out and buy a new shock-resistant 
486DX PC and get a new copy of Windows NT 3.51 from Microsoft, etc? 

-- 
T.E.D.          
                |  Work - mailto:dennison@escmail.orl.mmc.com  |
                |  Home - mailto:dennison@iag.net              |
                |  URL  - http://www.iag.net/~dennison         |

Re: AJPO is still clueless about the reality of DoD Ada rejection

[Robert Munck]

On Fri, 15 Mar 1996 21:06:34 GMT, srctran@world.std.com (Gregory
Aharonian) wrote:

> ... (like that COTS is compatible with the Ada Mandate - what a lie)

I'm beginning to see indications that the DoD fad for COTS software
is fading, and I'll go out on a limb to predict that there will be an
anti-COTS backlash by DoD within five years. Reason: COTS software
may well require more maintenance than home-grown.  See, those
greedy commercial folk find themselves competing with each 
other in saturated markets with diminishing original sales of
their packages. The solution is to churn out new releases as
often as possible with plenty of spiffy new features that
unfortunately tend to change the behavior of the old ones.
Likewise the platform suppliers are upgrading their OSs and
windowing systems, and even the hardware is changing enough
to make two-generation-old machines worthless.

So your nice DoD application, with two or three COTS packages
and running on MS Windows 3.0 on a 286, needs a HUGE and
CONTINUOUS upgrading and re-distribution activity.  If it were
all in good old stable Ada (ignoring the once-a-decade revisions),
the compiler vendor would handle it all by upgrading the run-time
and code generator.

So has anyone else seen this happening?

Bob Munck@acm.org

Re: AJPO is still clueless about the reality of DoD Ada rejection

[David Emery]

In article <314e3103.370551677@news.interramp.com>, munck@acm.org wrote:

> On Fri, 15 Mar 1996 21:06:34 GMT, srctran@world.std.com (Gregory
> Aharonian) wrote:
> 
> > ... (like that COTS is compatible with the Ada Mandate - what a lie)
> 
> I'm beginning to see indications that the DoD fad for COTS software
> is fading, and I'll go out on a limb to predict that there will be an
> anti-COTS backlash by DoD within five years. Reason: COTS software
> may well require more maintenance than home-grown.  See, those
> greedy commercial folk find themselves competing with each 
> other in saturated markets with diminishing original sales of
> their packages. The solution is to churn out new releases as
> often as possible with plenty of spiffy new features that
> unfortunately tend to change the behavior of the old ones.
> Likewise the platform suppliers are upgrading their OSs and
> windowing systems, and even the hardware is changing enough
> to make two-generation-old machines worthless.
> 
> So your nice DoD application, with two or three COTS packages
> and running on MS Windows 3.0 on a 286, needs a HUGE and
> CONTINUOUS upgrading and re-distribution activity.  If it were
> all in good old stable Ada (ignoring the once-a-decade revisions),
> the compiler vendor would handle it all by upgrading the run-time
> and code generator.
> 
> So has anyone else seen this happening?
> 
> Bob Munck@acm.org

What I've seen happen even more often is when a system is constructed
by cobbling together a variety of COTS packages (including O.S., DBMS,
GUI builders, etc.)  Then you get bit by the "upgrades cycles", when
Version 1.2 of Product X is not compatable with Version 3.7 of Product Y.
Version 3.8 of Product Y is compatable with Version 1.2 of Product X,
but requires Version 5.6beta2 of Product Z, which requires Version 1.3c
of Version X to run, etc.

Basically, COTS integration turns out to come too close to a N! integration
problem over the life-cycle.

Maybe this is why "COTS" sounds like a German slang term for vomit...

               dave

p.s.  The "Open Systems" movement was one way to beat this problem.  
Unfortunately, it turns out to be harder to do good Open Systems development
than anyone thought.  The problem is teaching the programmers to "code
within the lines" defined by the standards.

Re: AJPO is still clueless about the reality of DoD Ada rejection

[David Emery]

After I re-read Bob Munck's article, I saw that we're agreeing...

Unfortunately, too many managers learn their technology from the 'free
trade journals', whose articles are written by people pushing their
specific technologies/upgrades/etc.  I think it will take some major
public disasters before the managers learn their lessons.  (I know of
several non-public disasters, but none that I can talk about publically.
Fortunately, none of these systems are either safety-critical or 
operationally significant defense applications...)

            dave

Re: AJPO is still clueless about the reality of DoD Ada rejection

[Ken Garlington]

> Ken, you know we're both on the same side, BUT you and I both know of
> cases where more money was spent (in manhours) justifying a Ada waiver
> than would have been used if they had just done the Ada like good little
> boys and girls. [yet another anecdone published].

I absolutely agree that there are cases where Ada policy is abused.
I also absolutely agree that there are cases where traffic laws are abused.
I'm not quite ready to say that either is useless, however.

Re: Is COTS the answer?

[AdaWorks]

Ted Dennison (dennison@escmail.orl.mmc.com) wrote:

: Now you tell me, what are the odds that in 15 years when this thing 
: breaks the Navy will be able to go out and buy a new shock-resistant 
: 486DX PC and get a new copy of Windows NT 3.51 from Microsoft, etc? 

  Of equal importance.

  Commercial software is one of the few products that includes
  a fine-print disclaimer on every copy.  It is somewhat like buying
  a used-car, "As-Is."

  The disclaimer usually says something to the effect,

      "This software might do what you expect it to do but
       if it doesn't we, the software publisher, cannot be
       held liable.  Also, we abandon all repsonsbility for this
       product it does any damage to you, your business, your 
       clients, or anything or anyone else.."

  On the other hand, the software publisher does give you a full
  warranty on the quality of the media on which they distribute 
  their product. 

  When you buy a COTS product, you do not buy the product. Rather,
  you purchase a license to use the product.  

  When you license a COTS software product, it is restricted to the
  platform indicated on the license.  

  I know of no COTS software that is suitable for "safety-sensitive"
  applications.  This includes a lot of operating systems.

  Richard Riehle
-- 

richard@adaworks.com
AdaWorks Software Engineering
Suite 27
2555 Park Boulevard
Palo Alto, CA 94306
(415) 328-1815
FAX  328-1112

Re: AJPO is still clueless about the reality of DoD Ada rejection

[Peter Finney]

Ken Garlington <garlingtonke@lfwc.lockheed.com> wrote:

>> Ken, you know we're both on the same side, BUT you and I both know of
>> cases where more money was spent (in manhours) justifying a Ada waiver
>> than would have been used if they had just done the Ada like good little
>> boys and girls. [yet another anecdone published].

>I absolutely agree that there are cases where Ada policy is abused.
>I also absolutely agree that there are cases where traffic laws are abused.
>I'm not quite ready to say that either is useless, however.

This reminds me of my years working for NATO in Naples (Italy :-)).

I asked one of my (neapolitan) staff why the traffic signals were
largely ignored in Naples. He replied " humans know better than
robots".

I then pointed out that there were one or two intersections where the
lights were obeyed - reply " those are the dangerous crosssroads".

When I pointed out that it seemed a bit pointless having the signals
at all he said ( I paraphrase) "they are very useful in deciding who
pays when there is an accident".
Peter Finney
Principal Consultant
BAeSEMA Ltd
Portchester UK