Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Ken Garlington]

"Bob" <rfritz@home.com> wrote in message news:3B6A15C3.8B3A9277@home.com...
: Ada is a programming language developed by DOD. The first standard was
published
: in 1983 (after a multi-year selection and refinement process involving
Hundreds of
: people in industry, government, and the military)  and another updated
design was
: published in the mid 90s. It is a strongly typed language like Pascal that
was
: inbtended to support embedded programming. For a while it was mandated by
law for
: programs, but that was relieved some years ago. The 90s version
incorporated
: objects.
:
: The language is fine, but it is expensive because commercial use never
caught on.
: Tools are extremely expensive and available for only a few target
computers.
: Several late-80-s - early 90s aircraft projects were caught in the Ada
mandate
: F-22 among them.

Actually, none of the statements above are quite true:

- Ada is still used today on new commercial projects,
- There are free, open source, and commercial versions of Ada toolsets
available for a variety of platforms, including processors that run
Microsoft Windows, Linux, and a variety of embedded OSs.
- F-22 and other projects used Ada before there was a mandate, and continue
to do after the lifting of the mandate.

: Today most DOD projects use C or C++. C++ is basically Ada with full
polymorphism
: (objects)and C syntax.

This is just plain wrong.

: Actually, most C++ these days is just C compiled with a C++
: compiler. Object oriented systems in my experience do not provide any
advantage in
: real time development. I'm sure  a lot of Ada zealots will flame this
: statement,.but I have worked on C4I, signal processing, and now flight
control for
: UAVs and obect oriented design helps very little, and then in very obscure
parts
: of the system. Simple old techiques of modularity, high cohesion and low
coupling
: are more directly applicable.
:
: IMHO, Ada was killed by its proponents who made it a holy writ rather than
a tool.
: The more it was crammed down peoples throats the more they resisted.
Another
: factor was that the implicit development paradigm shifted from the 80s
model of a
: central computer with a lot of remote terminals (VAX) to the current
: workstation/PC on a net model.
:
: The central computer allowed one compiler to serve many so a
multi-thousand dollar
: price tag was acceptable. But if each programmer had his/her own computer
with
: several times the Vax computing power having a local compiler made sense.
Borland
: and Microsoft provided Pascal or C/C++ for a few hundred dollars for each
PC, and
: SUN/HP/SGI were not very much more per station. In fact the Gnu compiler
for C/C++
: is quite good and is free, and versions are available for both Windows and
Unix.

Not surprisingly, this is also true of the GNU Ada toolset!

: Ada continues as a legacy language, with systems being derived from the
projects
: of the 80s/early 90s, but there are not a lot of new starts. One
interesting note
: is that most of the Boeing digital airliners use Ada, meaning any versions
that
: have fly-by-wire. But my view is that Boeing makes the tool work well for
them
: rather than the tool creating quality systems by virtue of its own
qualities.
:
: I actually like Ada as a tool, but time and economics have passed it by.
:
: It should also be noted that many of the people that gave you Ada went on
to work
: at the DODs Software Engineering Institute (SEI) that gave us the highly
: bureaucratic and expensive 5 level software process ratings.

This is also a highly incorrect characterization of the Capability Maturity
Model (presumably what is meant by "process ratings").

For more information on the Ada language, see comp.lang.ada

: Bob
: UAV Software Lead
:
: Buescher Family wrote:
:
: > What is Ada?  The computer language?  Anyone care to explain?
: >
: > Geoffrey
: >
: > Ken Garlington wrote:
: >
: > >
: > > Well, it's true LM was an early adopter of Ada (anticipating the F-22
EMD
: > > contract requirement, as mandated by the U.S. Congress). The statement
is of
: > > course wrong in every other respect. Considering that the tarverbot
can't
: > > even spell software (literally!), I suppose one true item in a sea of
: > > inaccuracies is the best we can hope for...
:

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Bob Fritz]

If I really cared about Ada I would rebut these remarks essentially calling me a
liar. Suffice it to say that 23 years experience with DOD software including
participation in the Red/Green selection and the early Ada process through the
late 80s result in my statements.


Bob


>

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Tony Gair]

Right break it up boys,
1. Ada is not dead, in fact it has obviously been highly developed with lots
of useful components out there.

2. C++ is, if I get hold of him.

3. You US types make me laugh, laugh, laugh, not necessarily with you
either. Keep up the good work.
( I love your little fat little munchkin faces when you're angry)

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Ted Dennison]

In article <rbkb7.868$Pa.443446345@newssvr16.news.prodigy.com>, Ken Garlington
says...
>
>"Bob" <rfritz@home.com> wrote in message news:3B6A15C3.8B3A9277@home.com...
>: Today most DOD projects use C or C++. C++ is basically Ada with full
>: polymorphism (objects)and C syntax.
>
>This is just plain wrong.

Actually, wrong doesn't even begin to describe it. Ada supports both compile
time and runtime polymorphism (I assume that's what's meant by "full
polymorphism). C++ did indeed borrow some of Ada's ideas (eg: exceptions,
generics, line comments), but by no means all the important ones. For instance
it still does not sport any kind of native concurrency support, which is one of
Ada's big draws. C++ does have some of C's syntax but it has a lot of its own
too. C++ is a definite improvement over C, but its hardly a suitable substitue
for Ada.

I find the emphasis on C++ rather odd anyway. For the job I'm working on, the
argument was that FORTRAN would have been better used. The one before that, the
arguers wanted C. The one before that, they were switching from CMS-2. All of
these were post-mandate Ada jobs. I have yet to see a full-up C++ DoD job
(though I've no doubt some exist). The funny thing is that in each case, none of
the folks who wanted a different language were actually doing the work. People
who actually *use* Ada for a while tend to quickly see its benifits.

---
T.E.D.    homepage   - http://www.telepath.com/dennison/Ted/TED.html
          home email - mailto:dennison@telepath.com

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Ken Garlington]

"Bob Fritz" <rfritz@home.com> wrote in message
news:3B6DE593.D172D839@home.com...
: If I really cared about Ada I would rebut these remarks essentially
calling me a
: liar. Suffice it to say that 23 years experience with DOD software
including
: participation in the Red/Green selection and the early Ada process through
the
: late 80s result in my statements.

And my experience with DoD software, Ada applications, and contributions to
the Ada language standard -- also starting in the early days of Ada, but
then continuing through the 1990s and up to today -- indicates that your Ada
experience (which apparently ended over a decade ago) may be out of date.
Furthermore, there are contemporary references available from the Internet
that support my statements. To choose one of my points at random: If you do
not believe that there is a GNU compiler for Ada that is quite good and is
free, with versions available for both Windows and Unix, you can download a
version and see for yourself. Try the following sites for free downloads of
GNAT version 3.13p:

  wuarchive.wustl.edu/languages/ada/compiler/gnat/distrib/3.13p/

  cs.nyu.edu/pub/gnat/3.13p/

  packages.debian.org/unstable/devel/gnat.html

  linux.davecentral.com/4634_programcomp.html

Additional Ada tools (some released under the GPL) are available at:

  www.ada-france.org/ada-mode/

  www.adapower.net/gtkada/

  www.rrsoftware.com/html/prodinf/claw/clawintro.html

  libre.act-europe.fr/xmlada/

  adabroker.eu.org/

If you like to buy support with your compiler, try www.gnat.com.

(As an aside, I note that I used no personal attacks in my prior post --
certainly the word "liar" was never used. If you generally react this badly
to the possibility that you might be wrong, how did you ever make it through
23 years of software engineering without a nervous breakdown? :)

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Ken Garlington]

"Tony Gair" <tonygair@nospammy.blueyonder.co.uk> wrote in message
news:2Dsb7.72072$_b4.6134056@news1.cableinet.net...

: 3. You US types make me laugh, laugh, laugh, not necessarily with you
: either. Keep up the good work.

: ( I love your little fat little munchkin faces when you're angry)

Another case of USA envy :)

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Matthew V. Jessick]

Ted Dennison wrote:

> Actually, wrong doesn't even begin to describe it. Ada supports both compile
> time and runtime polymorphism (I assume that's what's meant by "full
> polymorphism). C++ did indeed borrow some of Ada's ideas (eg: exceptions,
> generics, line comments), but by no means all the important ones. For instance
> it still does not sport any kind of native concurrency support, which is one of
> Ada's big draws. C++ does have some of C's syntax but it has a lot of its own
> too. C++ is a definite improvement over C, but its hardly a suitable substitue
> for Ada.

I've worked with UAV algorithms and software in Ada and now
work in video games using C++.
C++ to me requires considerably more discipline to program without
silly bugs.

The    if (A=B) {    mistake   where B is copied into A and if non-zero
the if condition is true.


if (A)
    B;
    C;

where C appears to be part of the loop but really isn't
just because the first programmer didn't "waste" time
putting in some "superfluous" curly braces and the maintenance
programmer wasn't sufficiently alert to catch it.

The case problem of having to hand code break statements
to separate case blocks:

switch(a)
  case b:
     sdfdasf;
     adsfdaadsf;
  case c:
    break;

In Ada, every possible value for (a) is required to be handled.
The additional error potential of hand coding break statements
is also removed.

And don't even get me started on {}{}{{{}}{}{}{}{}{{}
;}
or macros or multi-dimensional array syntax
or default argument passing by value or...

It just isn't suited to coding logic statements without
silly mistakes. (There are coding standards you can impose
to reduce these mistakes, but I miss my Ada compiler each time
I make one of them.)


- Matt

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[John Keeney]

Matthew V. Jessick <mjessick@gte.net> wrote in message
news:3B6F73F9.15E98173@gte.net...
> The    if (A=B) {    mistake   where B is copied into A and if non-zero
> the if condition is true.

Actually, I've writen this code more than a few times, and meant it.
Generally more in context of loops though.

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Emmanuel Gustin]

"Matthew V. Jessick" <mjessick@gte.net> wrote in message
news:3B6F73F9.15E98173@gte.net...

> C++ to me requires considerably more discipline to program
> without silly bugs.

For any large project, I suspect that C++/C software is also
more expensive than software written in a modern language.
The compilers may be cheap but that is beside the point;
because of the rather unsafe characteristics of these languages
the debugging stage of development is inevitably longer and
validation more difficult. That quickly becomes much more
expensive than even a US$ 10,000 compiler.

The best excuse for using C/C++ is that, as C is basically
a gold-plated assembler, it is convenient for implementing
low-level interactions with hardware. But for a large and
mission-critical application it too unsafe. You never know
*for sure* that it isn't going to set its pointers to the wrong
address and crash the entire environment. And the aircraft
with it.

--
Emmanuel Gustin <gustin@NoSpam.uia.ac.be>
(Delete NoSpam. from my address. If you can't reach me, your host
 may be on our spam filter list. Check http://www.uia.ac.be/cc/spam.html.)

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Marin David Condic]

C may be a gold-plated portable assembly language, but I don't think its
quite as wonderful as Ada is for low-level interaction with the hardware. To
illustrate - try building a struct in which you can control the
representation as exactly as you can build an Ada record for the same job. C
leaves too many "implementation defined" behaviors and/or simply doesn't
give you control over representation. It ends up a lot easier in Ada. And
for all of C's bit-twiddling capabilities (shifting, and/or ops, etc.) Ada
has exact parallels, so I just don't see it as having an advantage. (And
I've written low-level stuff in both languages, so I don't think there is
any lack of experience with either one coloring my judgement.)

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com
Web:      http://www.mcondic.com/


"Emmanuel Gustin" <Emmanuel.Gustin@skynet.be> wrote in message
news:9krue8$5v03n$1@ID-52877.news.dfncis.de...
>
> The best excuse for using C/C++ is that, as C is basically
> a gold-plated assembler, it is convenient for implementing
> low-level interactions with hardware. But for a large and
> mission-critical application it too unsafe. You never know
> *for sure* that it isn't going to set its pointers to the wrong
> address and crash the entire environment. And the aircraft
> with it.
>

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[Mark]

>
> : ( I love your little fat little munchkin faces when you're angry)
>
> Another case of USA envy :)
>

Envy? Maybe!

However the USA's position is a false one. Why? Because the US is a military
driven economy. Take away the military/space budget and alot of jobs and R&D
would not have happened. The integrated circuit and the internet  are
classic examples.


Why is Mr. President driving on with the SOSW system? Because it will cost
tens of $billions, sustaining the economy. If the USA ever runs out of
potential threats... well that will never happen, because it will always
need to think of some reason to spend the money on.



My Two Pennies Worth. Completely off topic.

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[bendel boy]

"John Keeney" <jdkeeney@iglou.com> wrote in message news:<3b716b22_1@news.iglou.com>...
> Matthew V. Jessick <mjessick@gte.net> wrote in message
> news:3B6F73F9.15E98173@gte.net...
> > The    if (A=B) {    mistake   where B is copied into A and if non-zero
> > the if condition is true.
> 
> Actually, I've writen this code more than a few times, and meant it.
> Generally more in context of loops though.

But why would you want to mean this?

For my money, I would prefer the Algol approach:

if (0 = A := B) then

which would be interpreted as

(i) assign B to A -- assignment clause
(ii) test if A is zero -- logical clause

The C code makes use of a confusing short-cut, allowing confusion
between assignment, numbers, and booleans. Great for the experienced
programmer - fatal for the follow-on maintenance.

Re: Ada (was Rival JSF teams fly final STOVL flights with flair, highlight strengths)

[bendel boy]

"Ken Garlington" <Ken.Garlington@computer.org> wrote in message news:<rbkb7.868$Pa.443446345@newssvr16.news.prodigy.com>...
> "Bob" <rfritz@home.com> wrote in message news:3B6A15C3.8B3A9277@home.com...
> Ada continues as a legacy language, with systems being derived from the
> projects of the 80s/early 90s, but there are not a lot of new starts.

From my experience C/C++ counts as a legacy language. For my work
Fortran 90/95 is the current language.

Legacy language = we got burnt, so no longer use it; OR the college
kids are no longer taught it, so we've switched to the flavour of the
month.

In our case, we got burnt. We took working Fortran, wrote glue C++,
and had a product that failed on three counts: time scale, budget, and
doing the job.

Choose the the language that is best for the job - I wouldn't use
Fortran to write an arcade game, and I wouldn't see C++ as the best
language for solving maths systems. Ada is a lot like Algol-68, and
that was to me the 'crsipest' language I ever used.