Re: In Defense of the Mandate

Re: In Defense of the Mandate

[Gregory Aharonian]

>	I claim that an arbitrary embedded system (J) written in Ada is
>(on average) easier for an independent programming team to understand and
>subsequently enhance than one written in any previous language.  Thus,
>it is easier for independent programming teams to create variants of that 
>system (J', J'', J'''.....) than it would be if J were written in some older
>language.  That capability is needed if we have to fight protracted conflicts 
>like World War II, Korea, or VietNam.

Unfortunately, a growing number of real soldiers disagree with you, and
are using C/C++.  For example, there was that Army group involved with
JINTACCS that developed a fielded communications system in C++ using Motif.
Or people trhoughout the Air Force using C++ for database development.
And if you saw Ralph Craft's comment in Government Computer News, you
will understand that this use of C/C++ is common by people in the Armed
Services.

What is hurt Ada is that the DoD has refused to fund a truly honest
assessment of the microeconomics of defense software development vis-a-vis
Ada and C/C++.  The old Mosemann critiques were a joke, the new ones
probably not that much better, and people will continue to make decisions
based on meaningless data.

Also, given Bosnia, the protracted conflicts arguments is kind of moot.

Greg Aharonian

-- 
**************************************************************************
 Greg Aharonian
 Source Translation & Optimization
 P.O. Box 404, Belmont, MA 02178

In Defense of the Mandate

[david.c.willett]

D.C. Willett said:

DCW:	I claim that an arbitrary embedded system (J) written in Ada is
DCW:(on average) easier for an independent programming team to understand and
DCW:subsequently enhance than one written in any previous language.  Thus,
DCW:it is easier for independent programming teams to create variants of that 
DCW:system (J', J'', J'''.....) than it would be if J were written in some olde
r
DCW:language.  That capability is needed if we have to fight protracted 
DCW:conflicts like World War II, Korea, or VietNam.

To which Greg Aharonian responded:

GA>Unfortunately, a growing number of real soldiers disagree with you, and
GA>are using C/C++.  For example, there was that Army group involved with
GA>JINTACCS that developed a fielded communications system in C++ using Motif.
GA>Or people trhoughout the Air Force using C++ for database development.
GA>And if you saw Ralph Craft's comment in Government Computer News, you
GA>will understand that this use of C/C++ is common by people in the Armed
GA>Services.
	
	Apples vs. oranges, Greg.  The database example is a support 
system, not an embedded one.  I don't know about the communications system,
but it would not surprise me if that was a one-of-a-kind prototype.  It 
makes sense for the military to use COTS tools/product where such use doesn't
compromise combat readiness or reproduceability.  The Mandate makes sense 
where use of COTS would.

GA>
GA>What is hurt Ada is that the DoD has refused to fund a truly honest
GA>assessment of the microeconomics of defense software development vis-a-vis
GA>Ada and C/C++.  The old Mosemann critiques were a joke, the new ones
GA>probably not that much better, and people will continue to make decisions
GA>based on meaningless data.
	
	I believe that economics is irrelevant.  We are talking about the 
mission of DoD, which is to kill the bad guys.

GA>
GA>Also, given Bosnia, the protracted conflicts arguments is kind of moot.
	
	History's jury is still out on this one.  The notion that the lack 
of a competing superpower renders the protracted conflict passe remains 
unproven.  Precedents otherwise include the French in VietNam, the Israeli
experience, and North/South Korea.  I'll conceed that with the U.S. lead 
in technology, a protracted conflict is not likely in the next decade,
but that doesn't mean it's safe to dismantle the infrastructure required to
support one.


-- 
Dave Willett          AT&T Federal Systems Advanced Technologies
If God had wanted us to go around without any clothes, we all would have
been born naked!....er...ahh... Let me rephrase that.

Re: In Defense of the Mandate

[Dave Griffith]

In article <C82G4M.DE0@cbnewsl.cb.att.com> willett@cbnewsl.cb.att.com  
(david.c.willett) writes:
> 
> 	I've been sitting on this issue of whether the Ada Mandate
> is a GOOD THING or not, and I've decided to put forward a defense of the 
> Mandate.  The defense rests on some historical factors concerning 
> weapons procurement and production.  Given the victory in the Cold War,
> it can be argued that these presumptions no longer apply, but nevertheless
> they applied for the past several years and may still today.
> 
> First, the groundwork....
> 
> 	Consider World War II.  That war was important because it marked 
> the emergence of the United States as one of the two superpowers.  It also
> marked the beginning of the "Military-Industrial Complex" as we know it.
> Much fundamental U.S. Defense procurement policy can be traced to the 
> procurement practices adopted to win World War II.
> 
> 	World War II taught the U.S. what it takes to win a protracted 
> conflict.  Specifically, that one must be able to produce large quantities
> of standard weapons to overwhelm the enemy at the front.  Producing these
> quantities means that *ANY PRODUCER* must be able to manufacture "from
> scratch" any weapon currently in the U.S. arsenal.  During World War II,
> appliance manufacturers made rifles and tractor factories produced airplanes.
> Today the MIL-STDs are written so that this sort of cross production can
> be implemented again.  It's one reason DOD stuff costs so much more than its
> civilian counterparts.
> 

If this were what the MIL-STDs are trying to do, they've done so in a very odd 
 
way.  The classic counter to this argument is the M-16.  Using advanced  
plastics and alloys, it is only produced at a handful of factorie (2?).  If  
production had to be ramped up 100-fold in a matter of months, it simply  
couldn't be done.  OTOH, the M-16's main competitor, the AK-47/74, is produced 
 
out of bar stock and wood.  Numerous suppliers are available.  In a pinch,  
replacement parts for AK's have been produced by Afghani village blacksmiths.
The AK is acknowleged to be more reliable, if heavier, than the M-16.  Now  
_that's_ weapons procurement.   I see no evidence that current military  
procurement follows such "protracted conflict" scenarios.  Chances of anyone  
but McD managing to put out an F-16 are pretty small.

> 
> 	Consider that embedded software, the kind Ada was developed to 
> produce, is a critical part of modern weapons.  An F-16 simply isn't an 
> F-16 without the fly-by-wire software.  Fire control radars and ECM systems
> depend more on software than on hardware to accomplish their mission.
> 
> 	I claim that an arbitrary embedded system (J) written in Ada is
> (on average) easier for an independent programming team to understand and
> subsequently enhance than one written in any previous language.  Thus,
> it is easier for independent programming teams to create variants of that 
> system (J', J'', J'''.....) than it would be if J were written in some older
> language.  That capability is needed if we have to fight protracted conflicts
 
> like World War II, Korea, or VietNam.

Even accepting your claim on the merits of Ada for this requirement, the chance
  
of a modern war lasting long enough that major program revs would have to be  
shopped out to independent contractors is pretty small.  This would only really
  
be likely to occur in a "total war" on the scale of WWII.   

Your WWII analogy can be used to _oppose_ the mandate.  If WWII style ramp up  
was necessary, the current supply of Ada programmers would be grossly  
inadequate.  There would be a fairly sizeable time lag as programmers were  
retrained to Ada, and a (temporary) quality lag as they brought their skills up
  
from usable to professional.  With this in mind, the WWII analogy suggests that
  
defense procurement use the language for which an adequate supply of  
programmers can be quickly supplied in case of emergency.   C or Cobol,  
probably.

None of this is to put down current military procurement or the Mandate.   
Merely to show the the "protracted conflict" scenario given is incapable of  
explaining either.  On the actual merits of the Mandate, I return you to your  
regularly scheduled flame war.

--
Dave Griffith, Information Resources, University of Chicago,
Biological Sciences Division               dave@delphi.bsd.uchicago.edu
If it's too commercial, you're too old.

Re: In Defense of the Mandate

[agate!howland.reston.ans.net!usc!sol.ctr.columbia.edu!news.kei.com!ub!csn]

In article <1993Jun4.142757.28992@midway.uchicago.edu>, dave@blackjoke.bsd.uchi
cago.edu (Dave Griffith) writes:
|> In article <C82G4M.DE0@cbnewsl.cb.att.com> willett@cbnewsl.cb.att.com  
|> (david.c.willett) writes:

|> > (on average) easier for an independent programming team to understand and
|> > subsequently enhance than one written in any previous language.  Thus,
|> > it is easier for independent programming teams to create variants of that 
|> > system (J', J'', J'''.....) than it would be if J were written in some old
er
|> > language.  That capability is needed if we have to fight protracted confli
cts 
|> > like World War II, Korea, or VietNam.
|> 
|> Even accepting your claim on the merits of Ada for this requirement, the cha
nce  
|> of a modern war lasting long enough that major program revs would have to be
  
|> shopped out to independent contractors is pretty small.  This would only rea
lly  
|> be likely to occur in a "total war" on the scale of WWII.   

Actually, this happens all the time -- war or no war.  Some software systems I'
ve
worked with are 10 to 20 years old and have changed contractor hands 2, 3, 4 
times.  Pieces are spun off, subcontracted, maintained operationally by militar
y
programmers...  This is the whole basis of the readability vs. writability
tradeoff.  

|> Your WWII analogy can be used to _oppose_ the mandate.  If WWII style ramp u
p  
|> was necessary, the current supply of Ada programmers would be grossly  
|> inadequate.  There would be a fairly sizeable time lag as programmers were  
|> retrained to Ada, and a (temporary) quality lag as they brought their skills
 up  
|> from usable to professional.  With this in mind, the WWII analogy suggests t
hat  
|> defense procurement use the language for which an adequate supply of  
|> programmers can be quickly supplied in case of emergency.   C or Cobol,  
|> probably.

This is a scary thought...  It brings up that old question of what is a
programmer?  Is it someone who translates an explicit design, or someone who
makes the design as well.  In the first case, you're talking mostly syntax --
not such a big hurdle to overcome.  In the second case, I hope they're good
enough that the language doesn't matter as much as good design techniques
matter.  (I'm afraid, however, that that's not true -- the "adequate supply
of programmers" out there are not good large scale designers).

This ties into a favorite saying of mine...  "Using Ada makes a bad design 
obvious, using C doesn't"

-- 
Christopher A. Warack, Capt, USAF
Computer Science Department, US Air Force Academy

cwarack@kirk.usafa.af.mil                (719) 472-2401

Re: In Defense of the Mandate

[enterpoop.mit.edu!news.kei.com!sol.ctr.columbia.edu!howland.reston.ans.ne]

>>> On Fri, 4 Jun 1993 04:42:37 GMT, srctran@world.std.com (Gregory
>>> Aharonian) said:


(someone)> 	I claim that an arbitrary embedded system (J) written in
(someone)> Ada is (on average) easier for an independent programming
(someone)> team to understand and subsequently enhance than one written
(someone)> in any previous language.  Thus, it is easier for independent
(someone)> programming teams to create variants of that system (J', J'',
(someone)> J'''.....) than it would be if J were written in some older
(someone)> language.  That capability is needed if we have to fight
(someone)> protracted conflicts like World War II, Korea, or VietNam.

And it is terribly important. But I would like to refer to a Feynman
argument that is terribly important, especially in war: quick is
beautiful. One could argue that Ada programs can be developed more
quickly as the language is "better", but unfortunately most of an
application is not bespoke, and Ada has a distinct disadvantage when it
comes to environments/components:

Aharonian> Unfortunately, a growing number of real soldiers disagree
Aharonian> with you, and are using C/C++.  For example, there was that
Aharonian> Army group involved with JINTACCS that developed a fielded
Aharonian> communications system in C++ using Motif.  Or people
Aharonian> trhoughout the Air Force using C++ for database development.

Which is a catstrophe, as C/C++ are far more insecure, problematic,
difficult to teach languages than Ada *for applications*.

Unfortunately Ada has two huge problems, and the second is a consequence
of the first:

* cheap compilers are not available, if at all.

* a distinct lack of environmental, ready mode software.

A lot of C's success is because of the abundant supply of essentially
free compilers, being bundled with most Unix variants, and most Unix
variants were in any case free to Universities. Every time Unix has been
ported to a new architecture, a new compiler has become available as a
side effect, and usually for free with Unix.

It was also possible to license the C compiler source from AT&T for
something like $4,000, and a similar price used to apply to the C++
frontend. Every little sw house has been able to offer their own C++
compiler product as a consequence.

I tend to agree with Aaronian's comment that it is the mandate, and the
price floor it guarantees to Ada compiler vendors, the main cause for
Ada's stilted diffusion.

Also, no major application/system sw is written in Ada and requires an
Ada compiler. If X windows had been coded in Ada, say, now every X port
would have as a side effect an Ada port, and all people using X would
have a storng incentive to learn Ada.

But there is no major application/system sw in Ada simply because to
become 'major' in the wide world (outside the mandated area) a product
has to be essentially free, and require essentially free development
tools, otherwise research centers and Unviersities, that have th
institutional incentive to try new things, will not adopt it.

If, for example, 386BSD/Linux were written/rewritten in Ada, or if the
conditions for them to be written in Ada existed, Ada would become more
popular than C++, probably.