Re: Computational scientists ignoring and ignored by Ada

Re: Computational scientists ignoring and ignored by Ada

[Robert Dewar]

"It will shame vendors into action '''

What a wierd academic view of the world. Mike, companies do things because
they make profits doing things. If you want a vendor to do X, show that
vendor that doing X will earn at least enough in revenue to cover the
cost of doing X, and hopefully more. The idea of "shaming" a company to
take a decision not in its financial interest is silly!

Re: Computational scientists ignoring and ignored by Ada

[Michael Feldman]

In article <264rob$u3t@info2.rus.uni-stuttgart.de> paus@ifr.luftfahrt.uni-stutt
gart.de writes:
>In article <1993Sep1.142413.6105@sei.cmu.edu> wellerd@ajpo.sei.cmu.edu  
>(David Weller) writes:
>> I think it's a shame that Verdix wasn't
>> the "first kid on the block" to provide us with an early version
>> of protected records (which is, to a limited extent, what the
>> special features of MP-Ada do).
>
>We are using the Verdix compiler on a SiliconGraphics for about a
>year now. From what I have experienced I would not even expect Verdix
>to be the second or third or ... kid. They are neither able to provide
>a standard ISO math library nor are they able to provide a POSIX
>interface to the operating system. As long as they do not come up
>with these simple things I won't dare to dream of any 9X features.
>
Sigh...when I first starting dealing with Verdix in 1986 I thought they
had a real edge. In the last few years I've had a feeling that they were
losing their edge. This confirms it. It also tends to confirm my uneasy
feeling that they really don't care.

Mike Feldman

Re: Computational scientists ignoring and ignored by Ada

[Gregory Aharonian]

>Some of the people who collect a copy of Lowtran do it only to compare
>Lowtran to their favorite model, so they can blast Lowtran in their
>publications.  
>
>You sound completely naive about how models really get used.

   I base my heavily used claim for Lowtran on a few things.

   1) The large number of articles I have read in Applied Optics, SPIE
      conference proceeedings, and other magazines and journals in
      which Lowtran results are displayed.  I add to this the many
      references to Lowtran in DoD technical reports available through
      DTIC.

   2) The fair amount of computer programs in my massive database of
      information on all of the government's (especially DoD) software
      that use Lowtran as an adjunct program.

   3) The fact that at least two companies have commercial versions of
      Lowtran implying there is enough demand to market and profit, to
      such an extent that the Air Force applied for and received a
      patent for the changes in the source code from Lowtran 6 to 
      Lowtran 7, in order to be able to license access to the program.

    Trust me, having examined more DoD models than anyone else on the
planet, and being able to partially measure their reuse rates, I say that
Lowtran is heavily used and would serve as a model code to be translated
to Ada to help get Ada into DoD scientific computing.  Save your naiveness
comments for the rest of the DoD reuse statements.

    Another way would be to translate BLAST into Ada, assuming you know
what it does, who has, and why it is used so much.

-- 
**************************************************************************
 Greg Aharonian                                      srctran@world.std.com
 Source Translation & Optimization                            617-489-3727
 P.O. Box 404, Belmont, MA 02178

Re: Computational scientists ignoring and ignored by Ada

[Gary J. Cernosek]

In article <CCnAw5.rt@cerc.wvu.edu> vanscoy@cs.wvu.wvnet.edu (Frances L VanScoy
) writes:
>We who have an appreciation for Ada and software engineering and
>experience in scientific computing need to continue to encourage
>parallel computer vendors and Ada compiler vendors to provide Ada
>compilers for parallel computers.

In an effort to build on the points being made in this thread, 
I wanted to make reference to the latest Communications of the 
ACM (Sept. '93).  The issue's theme is "Concurrent 
Object-Oriented Programming."  After a quick glance through the 
eight articles presented on the subject, I saw little mention 
of Ada.  Now obviously, since Ada is not quite yet in the 
category of an "object-oriented programming language," it might 
make sense to exclude it in this issue.  But with the new OO 
features of Ada 9X on the way, Ada could (and should) become a 
real player in this barely touched market of integrating object 
technology and concurrent programming.

We all know that in spite of its shortcomings in the tasking 
model, Ada does represent 10+ years of experience with 
concurrent programming within a STANDARD programming language.  
I emphasize "standard" here because I see that several of the 
CACM articles are based on unique, non-standard, or otherwise 
disjoint extensions to existing languages.  This is indeed one 
area in which Ada has a discriminator -- that is, in 
integrating concurrent programming constructs directly into 
the standard source code language.

The reason why I wanted to make these points is that I think 
the Ada community should capitalize on both the technical and 
the marketing opportunities associated with this subject.  
Assuming one believes that the future of computing rests in 
parallel processing technologies, and that the object paradigm 
is here to stay (at least for a while), the convergence of the 
two presents a real growth potential for languages such as 
Ada.  I think that with Ada 9X enabling Ada to escape the "...but 
it's not really object-oriented" syndrome, the next technical 
(and marketing) challenges lie in exploring exactly how to 
integrate Ada's new OO features with its old (and new) concurrent
programming mechanisms.

Now, I know that some work has already been done by Colin 
Atkinson and David Weller in this very subject, and will in 
fact appear in this year's Tri-Ada proceedings.  And there does 
indeed appear to be some significant technical issues facing 
the integration effort.  However, with all of the (mostly 
negative) rhetoric that I see in this newsgroup regarding Ada's 
marketability, I thought I'd try to generate some discussion on 
what to do about the future of Ada that doesn't pertain to the 
mandate issue :-).




-- 
Gary J. Cernosek
Fastrak Training Inc.
Houston Office: (713) 280-4768
E-mail: cernosek@source.asset.com

Re: Computational scientists ignoring and ignored by Ada

[Esther Lumsdon]

srctran@world.std.com (Gregory Aharonian) writes:

>    For example, one of the most heavily reused pieces of DoD R&D software
>is a program called Lowtran.  Yet not once in ten years of Ada spending has
>anyone funded its translation into Ada, which at 20,000 lines of code, 10,000
>of which are tables of propagation coefficients, would not be very expensive
>to translate into Ada.  Doing so would help promote DoD R&D software into
>Ada - if anyone really cared.

Lowtran is an interesting model.  However, you ignore the political
considerations of who in DoD-related research uses what model.  It's
impossible to make any sort of statement about 'heavy reuse' and possible
benefits of translating a model to Ada without looking at the political
considerations.  Some of the people who collect a copy of Lowtran do it
only to compare Lowtran to their favorite model, so they can blast Lowtran
in their publications.  

You sound completely naive about how models really get used.
-- 
-- Esther Lumsdon, not speaking for Verdix.   esther@verdix.com
"It's time to cut bait and talk turkey.  It takes 2 snakes to cross a
puddle. You have to bale hay while the tractor is warm."
  ---- either H. Ross Perot or Dave Barry

Re: Computational scientists ignoring and ignored by Ada

[agate!howland.reston.ans.net!xlink.net!rz.uni-karlsruhe.de!news.uni-stutt]

In article <1993Sep1.142413.6105@sei.cmu.edu> wellerd@ajpo.sei.cmu.edu  
(David Weller) writes:
> I think it's a shame that Verdix wasn't
> the "first kid on the block" to provide us with an early version
> of protected records (which is, to a limited extent, what the
> special features of MP-Ada do).

We are using the Verdix compiler on a SiliconGraphics for about a
year now. From what I have experienced I would not even expect Verdix
to be the second or third or ... kid. They are neither able to provide
a standard ISO math library nor are they able to provide a POSIX
interface to the operating system. As long as they do not come up
with these simple things I won't dare to dream of any 9X features.

Michael

************************************************************************
* Dipl.-Ing. Michael Paus | Phone: (+49) 711-121-1434                  *
* University of Stuttgart | FAX:   (+49) 711-634856                    *
* Forststrasse 86         | Email: paus@ifr.luftfahrt.uni-stuttgart.de *
* 70176 Stuttgart, FRG    |        (NeXT-Mail welcome)                 *
************************************************************************

Re: Computational scientists ignoring and ignored by Ada

[Michael Feldman]

In article <CCnAw5.rt@cerc.wvu.edu> vanscoy@cs.wvu.wvnet.edu (Frances L VanScoy
) writes:
>
>We who have an appreciation for Ada and software engineering and
>experience in scientific computing need to continue to encourage
>parallel computer vendors and Ada compiler vendors to provide Ada
>compilers for parallel computers.
>
Is this a job for a GNAT port? Maybe when some fuzzy-headed academics
do a good Ada for machines like this, it'll shame the vendors into action.

Mike Feldman

Re: Computational scientists ignoring and ignored by Ada

[Michael Feldman]

In article <CCnAw5.rt@cerc.wvu.edu> vanscoy@cs.wvu.wvnet.edu (Frances L VanScoy
) writes:
>
[deletia]
>
>We who have an appreciation for Ada and software engineering and
>experience in scientific computing need to continue to encourage
>parallel computer vendors and Ada compiler vendors to provide Ada
>compilers for parallel computers.
>
Yes!

Yes!

Yes!

Vendors, anyone listening out there? If they can do Fortran and C, you
can port Ada. You've already got the front ends...

Mike Feldman
------------------------------------------------------------------------
Michael B. Feldman -  co-chair, SIGAda Education Committee
Professor, Dept. of Electrical Engineering and Computer Science
The George Washington University -  Washington, DC 20052 USA
202-994-5253 (voice) - 202-994-0227 (fax) - mfeldman@seas.gwu.edu (Internet)
"We just changed our CONFIG.SYS, then pressed CTRL-ALT-DEL. It was easy."
-- Alexandre Giglavyi, director Lyceum of Information Technologies, Moscow.
------------------------------------------------------------------------

Re: Computational scientists ignoring and ignored by Ada

[dog.ee.lbl.gov!agate!howland.reston.ans.net!math.ohio-state.edu!cis.ohio-]

In article <1993Sep1.142933.15335@relay.nswc.navy.mil> bwallet@apssgi.nswc.navy
.mil (Brad Wallet) writes:
>In article <CCnAw5.rt@cerc.wvu.edu>, vanscoy@cs.wvu.wvnet.edu (Frances L VanSc
oy) writes:
>|> [deletia]
>|> We who have an appreciation for Ada and software engineering and
>|> experience in scientific computing need to continue to encourage
>|> parallel computer vendors and Ada compiler vendors to provide Ada
>|> compilers for parallel computers.
>|> 
>
>As a computational scientist, I agree completely.  But, there is hope.
>The new Ada compiler for SGI machines (Verdix, I believe) offers true
>support for multiprocessor machines.  Ada is so well suited for it.
>Different tasks using different processors.  Oh, the structure is all
>built in.
>
>Brad

Quite right, Brad.  The MP-Ada product is a significant advance
for Ada compilers.  However, I would like to point out that the
structure isn't _quite_ "all built in".   Like any product that
has "special" support, there is an (obvious) element of non-portability
in the code you write.  I think it's a shame that Verdix wasn't
the "first kid on the block" to provide us with an early version
of protected records (which is, to a limited extent, what the
special features of MP-Ada do).  I hope that folks that are at
Tri-Ada and OOPSLA put Verdix on the spot for this: There are several
constructs within MP that are supported with PR syntax.  This
will probably lead to inconsistencies in the future, once
9X compilers are released.

Don't get me wrong, I think MP-Ada is a wonderful product.  I just
also happen to think that this is a case of "Excellent tactics,
Horrible strategy".  Certainly, if I were working at Verdix or
SGI, I might claim that PRs aren't "stable" enough to attempt
implementation, but I don't :-)

Perhaps Dave McAllister (boy, I hope I spelled that right) can
add some commentary from the SGI side?

-- 
type My_Disclaimer is new Standard.Disclaimer with record
	AJPO, SEI : Cognizance := Disavow_All_Knowledge;
end record;--)

Re: Computational scientists ignoring and ignored by Ada

[dog.ee.lbl.gov!agate!howland.reston.ans.net!math.ohio-state.edu!darwin.su]

In article <CCnAw5.rt@cerc.wvu.edu>, vanscoy@cs.wvu.wvnet.edu (Frances L VanSco
y) writes:
|> 
|> 
|> I am a faculty member in computer science at West Virginia and a big
|> fan of Ada.  (I first taught a unit on Ada in a course in 1980,
|> taught Ada as a first programming language to freshmen in 1987, and
|> was the major advocate for my department's move to teach Ada to all
|> freshmen cs majors starting in fall 1989.)
|> 
|> Currently I am working with physics and engineering researchers
|> at WVU on a computational materials science project.  The university
|> has its own new (less than one year old) parallel computer.
|> I would *love* to develop new software for this project in Ada.
|> However we only have compilers for dialects of Fortran and C.
|> In discussions with many vendors of parallel computers before we
|> purchased ours I was disappointed to learn that Ada compilers for 
|> parallel computers are essentially nonexistant.  In three+ years on
|> this project I have often missed the protection of strong typing and
|> the support for abstraction provided by Ada.
|> 
|> We who have an appreciation for Ada and software engineering and
|> experience in scientific computing need to continue to encourage
|> parallel computer vendors and Ada compiler vendors to provide Ada
|> compilers for parallel computers.
|> 
|> Frances Van Scoy, Associate Professor of Computer Science
|> Dept. of Statistics and Computer Science, West Virginia University
|> Morgantown, WV  26506-6330      phone 304-293-3607   fax 304-293-2272

As a computational scientist, I agree completely.  But, there is hope.
The new Ada compiler for SGI machines (Verdix, I believe) offers true
support for multiprocessor machines.  Ada is so well suited for it.
Different tasks using different processors.  Oh, the structure is all
built in.

Brad

Re: Computational scientists ignoring and ignored by Ada

[agate!howland.reston.ans.net!europa.eng.gtefsd.com!darwin.sura.net!RBSE.M]

I am a faculty member in computer science at West Virginia and a big
fan of Ada.  (I first taught a unit on Ada in a course in 1980,
taught Ada as a first programming language to freshmen in 1987, and
was the major advocate for my department's move to teach Ada to all
freshmen cs majors starting in fall 1989.)

Currently I am working with physics and engineering researchers
at WVU on a computational materials science project.  The university
has its own new (less than one year old) parallel computer.
I would *love* to develop new software for this project in Ada.
However we only have compilers for dialects of Fortran and C.
In discussions with many vendors of parallel computers before we
purchased ours I was disappointed to learn that Ada compilers for 
parallel computers are essentially nonexistant.  In three+ years on
this project I have often missed the protection of strong typing and
the support for abstraction provided by Ada.

We who have an appreciation for Ada and software engineering and
experience in scientific computing need to continue to encourage
parallel computer vendors and Ada compiler vendors to provide Ada
compilers for parallel computers.

Frances Van Scoy, Associate Professor of Computer Science
Dept. of Statistics and Computer Science, West Virginia University
Morgantown, WV  26506-6330      phone 304-293-3607   fax 304-293-2272

Re: Computational scientists ignoring and ignored by Ada

[agate!usenet.ins.cwru.edu!magnus.acs.ohio-state.edu!math.ohio-state.edu!s]

In <CCKyu8.KtD@world.std.com> srctran@world.std.com writes:
> 
>     One area that Ada has had no luck in penetrating (or more particularly,
> one area where the Ada community has had no interest in penetrating), is
> that of the Computational Science community, which until quite recently was

Well we are attempting to use Ada for just this. So Ada is not being totally
ignored. I think the biggest problem is that Computational Computing is 
often done by Scientist who are not Programmers and self taught themselves
FORTRAN back when there were card punches. Unfortunately Ada is not easy
for them to learn, and they almost have no desire to learn whats not directly
needed for them to do their job, Science not Programming. If I could only
get them to write good FORTRAN code I'd be happy.

The other problem is that SOME Ada compilers do not generate Numerical
code as good as FORTRAN. For example from one Ada compiler the code
takes 1.5 times longer than FORTRAN, and with another compule 3 TIMES longer.
This gives Ada a bad name. I get around this by coding the main structure 
of the program in Ada and the Number Crunching Loops in FORTRAN. Its not 
what I'd like to do but it gives me the performance I need.


Don Berryman
Defence Research Establishment Pacific
Canadian Forces Base Esquimalt
Victoria, BC, CANADA, V0S-1B0
604-363-2731    604-363-2856fax
berryman@orca.drep.dnd.ca