Re: Study shows Ada tools ten times more costly

Re: Study shows Ada tools ten times more costly

[Robert Dewar]

-- Sort of irrelevent.  We are talking cross-development tools.  Also
-- the Ada/Ed translator is currently available from the FSF, and the
-- GNAT Ada 9X front end will shortly be bundled with the above.

This information is incorrect. The Ada/Ed translator is available by
anonymous FTP from NYU, but there is no connection with the FSF. It
is true that we hope to bundge GNAT with the standard GCC distribution

Re: Study shows Ada tools ten times more costly

[Robert I. Eachus]

In article <1993May26.233758.26489@power.amasd.anatcp.rockwell.com> milo@power.
amasd.anatcp.rockwell.com (John DiCamillo) writes:

 >  Hey, put your money where you mouth is:

 >	   Tool			Platform		Price
 >	   ---------------	----------------	--------------
 >	   GNU C/C++ Compiler	Sun/PC/etc.		FREE
 >	   GNU SOURCE		Sun/PC/etc.		FREE

-- Sort of irrelevent.  We are talking cross-development tools.  Also
-- the Ada/Ed translator is currently available from the FSF, and the
-- GNAT Ada 9X front end will shortly be bundled with the above.

  >	  TI C30 C Compiler	Sun-> 'C30 DSP		~ $ 5,000.00
  >	  TI C30 Library Source Sun-> 'C30 DSP		~ $ 5,000.00
  >	  Tartan Ada Compiler	Sun-> 'C30 DSP		~ $36,000.00
  >	  Tartan Ada RTS Source	Sun-> 'C30 DSP		~ $50,000.00

-- I think that these are old prices, but look similar to what we
-- paid.  But, we have several Ada compilers running on the same Suns,
-- some of which were less expensive than the C30 C compiler.  The
-- cross-compilers are the most expensive. Native host the least.

  > Now, I realize that not all Ada Compilers are prohibitively expensive.
  > The point is that many are.  Find me a platform where the Ada compiler
  > is comparable in price to the C compiler!

-- Have you looked at Sun's prices for SunAda and for their SPARC C
-- compiler recently?  (Of course, I recommend you use SunAda but not
-- Sun C.)

  > And if you think that DTP systems are the real source of DoD tool
  > expenditures, think again:

  >	   FrameMaker for Windows		~   $750.00
  >	   Interleaf for Sun 4			~ $1,500.00

   > More expensive than WinWord, yes. Ten (or a hundred) *times* more, no.

   How about Statemate, Cadre Teamwork, Software through Pictures, SES
Workbench, DCE?  I'd have to go pull out old invoices to put prices
together, but between the tools mentioned I think we have spent more
than ten times what we have spent on all compilers.  (Yes we have the
Tartan TMS320C30 compiler, but we only need one copy of that...)

   And I haven't even mentioned the cost of ICEs, etc.  Simulation and
software design tools, most of them not Ada specific, make up a much
larger share of our tool expenditures than all compilers for all
languages put together.  I'm sure the situation at government
contractors is similar.

--

					Robert I. Eachus

with Standard_Disclaimer;
use  Standard_Disclaimer;
function Message (Text: in Clever_Ideas) return Better_Ideas is...

Re: Study shows Ada tools ten times more costly

[dog.ee.lbl.gov!overload.lbl.gov!agate!howland.reston.ans.net!usc!news.ser]

shimeall@cs.nps.navy.mil (timothy shimeall) writes:

>In article <C784HG.5E0@world.std.com> srctran@world.std.com (Gregory Aharonian
) writes:
>>    This comment to some extent supports the contention that the Ada
>>Mandate, by sheltering the Ada tools companies from competition, is distortin
g
>>the marketplace and making use of Ada more expensive than other languages.
>>(I assume in his study that for the most part, military software developers
>>only are buying Ada tools).

>Sorry, the data quoted does NOT support the conclusion Greg makes.
>Sorry, the fact that Military software developers pay higher prices
>for more thorough, capable and higher quality tools [perhaps targeted
>at Ada] does NOT mean that all Ada tools are automatically more
>expensive.  Greg should be more careful of the conclusions he draws.

Hey, put your money where you mouth is:

	Tool					Platform		
		Price
	---------------			----------------		-------
-------
	GNU C/C++ Compiler		Sun/PC/etc.			
	FREE
	GNU SOURCE				Sun/PC/etc.		
		FREE

	TI C30 C Compiler		Sun-> 'C30 DSP			~ $5,00
0.00
	TI C30 Library Source	Sun-> 'C30 DSP			~ $5,000.00

	Tartan Ada Compiler		Sun-> 'C30 DSP			~ $36,0
00.00
	Tartan AdA RTS Sourc	Sun-> 'C30 DSP			~ $50,000.00


Now, I realize that not all Ada Compilers are prohibitively expensive.
The point is that many are.  Find me a platform where the Ada compiler
is comparable in price to the C compiler!

Vendors of non-mandated language (C/C++, Pascal, FORTRAN) compilers 
don't have the luxury of charging these prices because THEIR COMPETITION
WILL EAT THEM ALIVE if they do!

And if you think that DTP systems are the real source of DoD tool expenditures,
think again:

	FrameMaker for Windows		~   $750.00
	Interleaf for Sun 4			~ $1,500.00

More expensive than WinWord, yes. Ten (or a hundred) *times* more, no.
-- 

------------------------------------------------------------------------------
  John DiCamillo    |   milo@power.amasd.anatcp.rockwell.com
  (MILO)            |

Re: Study shows Ada tools ten times more costly

[agate!howland.reston.ans.net!zaphod.mps.ohio-state.edu!math.ohio-state.ed]

In article <C784HG.5E0@world.std.com> srctran@world.std.com (Gregory Aharonian)
 writes:
...stuff deleted

>    One comment has some relevance to the issue of the Ada Mandate.  To
>quote (page 99):
>
>	"The cost of software engineering tools ranges from less"
>	than $1 per function point, to more than $20 per function
>	point.  Developers using PC, Unix and Apple Macintosh
>	platforms have the largest selection of tools priced at
>	less than $1 per function point.  Manufacturers and
>	military software developers often use tools priced at
>	more than $10 per function point.
>
>
>    This comment to some extent supports the contention that the Ada
>Mandate, by sheltering the Ada tools companies from competition, is distorting
>the marketplace and making use of Ada more expensive than other languages.
>(I assume in his study that for the most part, military software developers
>only are buying Ada tools).
>
>Greg Aharonian
>

  Unless there is more to that article this paragraph does not seem
to imply that it is only Ada tools that are costing more.  The DoD developers
use lot of tools that are expensive that many people working on their
own (or even in the commercial world) do not always use. These include
word processors with tons of "features" that cost not 10 times but often
100 times more than a PC editor (ie Sun based interleaf editor vs PC MS
word). Configuration managment tools (other than SCCS, RCS types) not to
mention the fact that DoD developers usually end up having to  use tools
that are not as mainstream as 80x86 680X0 compilers (even when they are
C compilers).

  I won't argue that Ada compilers are not too expensive but there are
plenty of other overpriced tools that drive up that ratio.

Jeff Creem
            
The opions expressed are mine and do not reflect the opinion, blah
blah..

Re: Study shows Ada tools ten times more costly

[Robert I. Eachus]

In article <C784HG.5E0@world.std.com> srctran@world.std.com (Gregory
Aharonian) does it again:

   First he quotes T. Capers Jones on the cost of tools:

	   "The cost of software engineering tools ranges from less"
	   than $1 per function point, to more than $20 per function
	   point.  Developers using PC, Unix and Apple Macintosh
	   platforms have the largest selection of tools priced at
	   less than $1 per function point.  Manufacturers and
	   military software developers often use tools priced at
	   more than $10 per function point.

   then Greg assumes a lot:

       This comment to some extent supports the contention that the
   Ada Mandate, by sheltering the Ada tools companies from
   competition, is distorting the marketplace and making use of Ada
   more expensive than other languages.  (I assume in his study that
   for the most part, military software developers only are buying Ada
   tools).

   Having shopped for, and helped to set prices for, development
tools, the most important factor in price is the size of the potential
market.  (Second most important is the number of other vendors trying
to get a slice of that market.)  Now the first two markets mentioned
by Jones are by far the largest development tool markets around.  In
terms of development tools the military market is one of the smallest.
I wonder if there is any connection between that and cost per function
point...


--

					Robert I. Eachus

with Standard_Disclaimer;
use  Standard_Disclaimer;
function Message (Text: in Clever_Ideas) return Better_Ideas is...

Re: Study shows Ada tools ten times more costly

[dog.ee.lbl.gov!overload.lbl.gov!agate!howland.reston.ans.net!usc!cs.utexa]

In <C78BBy.CM2@taurus.cs.nps.navy.mil> shimeall@cs.nps.navy.mil (timothy shimea
ll) writes:

>Frankly, tools to do for C (to take an example) the types of analysis 
>cannonically done for Ada are going to be MUCH more expensive, due
>to the ambiguous syntax of C.  Consider my favorite C ambiguity:

It's not ambiguous at all.  It means exactly one thing.  It means that
p is a pointer to a pointer to a pointer to an int.  

>void proc(p)
>int ***p;
>{
>...
>}
>The first * in the declaration of p indicates that it is an in-out 
>parameter.  The second indicates that the in-out parameter is a
>single-dimenstioned array.  The third indicates that each array
>element accesses an integer.
>or
>p is an in parameter for a three-dimensional array of integer
>or
>p is an in-out parameter for a two-dimensional array of integer
>or
>... (any other interpretations?)

>Now consider the design of a tool that will perform bounds-checking
>on p, prior to run-time.

You can't ever do this in C, particularly with this type of
declaration.  It's not as if there is some law against giving the
array declaration form if that is what is intended.  Once again, it is
a case of the language not forcing people to do things a certain way
being seen as 'bad'.

This is simply a matter of doing the appropriate interface design,
which you would have to do in any language.  There *are* places where
C syntax is ambiguous, but this is hardly one of them.

-- 
"Insisting on perfect safety is for people who don't have the balls to live
 in the real world."   -- Mary Shafer, NASA Ames Dryden
------------------------------------------------------------------------------
Fred.McCall@dseg.ti.com - I don't speak for others and they don't speak for me.

Re: Study shows Ada tools ten times more costly

[David Emery]

Does Capers Jones define what he means by "function point"?  

Another relevant point is the effect of 'group tools'.  It's very
easy to equip a PC for individual work.  It's a lot harder (but less
hard than it used to be) to equip a collection of PC's for a group
project.  

A third point that I'd like to make here on the specific topic of DoD
tools is the significant cost of embedded system tools, such as
cross-compilers, In-Circuit Emulators, etc.  These tools are very
expensive even for commercial applications, and the additional
requirements for DoD software doesn't help the price any.  

It would be very interesting to see a collection of cost curves
showing 'capitalization' costs for PC-based, workstation-based,
mainframe-based, embedded targets, for single-programmer to large
(>100 programmer) projects, and from low-reliability (games?) to very
high reliability (e.g. avionics) applications.

					dave

Re: Study shows Ada tools ten times more costly

[dog.ee.lbl.gov!overload.lbl.gov!agate!howland.reston.ans.net!zaphod.mps.o]

In article <1993May18.101854.1528@sei.cmu.edu> wellerd@ajpo.sei.cmu.edu (David 
Weller) writes:

>Dadgummit, Greg!  This is NOT related to Ada!  Tool cost is almost
>diretcly related to the PLATFORM, not the language.  I spent 10
>years in the military, five in electronic warfare and five in software
>development.  The costs I found were related to using tools on specialized
>hardware platforms: Honeywell, Tandem, HP Special Test Harnesses,
>and aging IBM boxes.  All these products are either low-volume commercial
>products or special-purpose military equipment.

That's right.  The DoD still thinks it's living in the 1940s, when it
was the dominant developer and buyer of high-technology equipment.  Those
funky special-purpose development systems are outrageously expensive
mainly because the development costs can be amortised over only a handful
of systems, rather than over millions.  Add, of course, the fact that
the DoD procurement system typically trebles the cost of everything.

And it has nothing to do with language. The outrageous cost of non
functioning special purpose Ada development systems is simply a replay
of the similarly outrageous costs of special purpose Jovial, CMS/2,
&c systems, ad nauseam.   But I agree the mandate did nothing to help:
instead of every service having a different programming language, every
service just "had to" procure a different implementation of that language,
with minimal commonality and reuse.  That's why most NATO countries long
ago implemented a single, defence-wide procurement system, both saving
money and gaining military efficiency.

>Now, I'm not excusing these costs either.  Our military culture is more
>paralyzed than the culture of large corporations -- changing from mainframes
>to low-cost PCs is prohibitively expensive in the short-run, which is
>how the military gets its budget.  Worse yet, even IF they change to
>PCs/Macs, security requirements drive up the hardware costs outrageously.
>(New oxymoron: TEMPEST-approved Laptop :-)  

No, it's outrageously cheap.  Typically, you can replace that mainframe with
a set of PCs of equivalent power for less than the *annual maintenance cost*
of the mainframe.  I've seen it done, and you can't get shorter run than that.
The problem is the bureaucracy, along with the fact that the people spending
that maintenance budget will fight tooth and nail to keep it.

And what security, kemosabe?  The electromagnetic emissions of a big iron
mainframe are an order of magnitude stronger than those of a set of PCs with
similar computing power.  I well remeber sitting in a van outside the
perimeter fence and watching the big discs thrash, over half a mile away!
The only thing you really need to secure is the communications lines, and
the simplest, low-cost solution is to use fibre optic cables.  Again, I
saw that done in a building in London, and the main cost was making sure
that every metre of cable was visible, which involved cutting a few holes
in the thick Victorian walls and floors.  It was expensive then - nearly
fifteen years ago - today, it's cheaper than copper wire for the same
bandwidth.  As for Tempest-approved laptops, the BATES program had approved
portable battlefield fire-control terminals, again nearly fifteen years ago.

Re: Study shows Ada tools ten times more costly

[dog.ee.lbl.gov!overload.lbl.gov!lll-winken.llnl.gov!taurus.cs.nps.navy.mi]

In article <C784HG.5E0@world.std.com> srctran@world.std.com (Gregory Aharonian)
 writes:
>   In the January 1993 issue of Software Magazine, T. Capers Jones has a
>nice article titled "Equipping the software engineer"
...
>    One comment has some relevance to the issue of the Ada Mandate.  To
>quote (page 99):
...
>       Manufacturers and
>	military software developers often use tools priced at
>	more than $10 per function point.
>
>    This comment to some extent supports the contention that the Ada
>Mandate, by sheltering the Ada tools companies from competition, is distorting
>the marketplace and making use of Ada more expensive than other languages.
>(I assume in his study that for the most part, military software developers
>only are buying Ada tools).

Sorry, the data quoted does NOT support the conclusion Greg makes.
Consider that military software developers (compared to developers of
civilian software):
  + Are required to more thoroughly document their efforts
    (and tools need to support that documentation)
  + Are required to adhere to specific standards (and tools to 
    validate those standards are often specially-built to do so)
  + Are building software with longer lifespan (the A-6 flight
    software was originally written in the Vietnam War era ..
    and is still flying [albeit modified] today), so tools need
    to be held to higher quality standards
  + Are building larger software, so tools need to be built with
    fewer size constraints

So, disregarding these major tool cost drivers, Greg attibutes the higher
cost to the choice of language.

Sorry, the fact that Military software developers pay higher prices
for more thorough, capable and higher quality tools [perhaps targeted
at Ada] does NOT mean that all Ada tools are automatically more
expensive.  Greg should be more careful of the conclusions he draws.

Frankly, tools to do for C (to take an example) the types of analysis 
cannonically done for Ada are going to be MUCH more expensive, due
to the ambiguous syntax of C.  Consider my favorite C ambiguity:

void proc(p)
int ***p;
{
...
}
The first * in the declaration of p indicates that it is an in-out 
parameter.  The second indicates that the in-out parameter is a
single-dimenstioned array.  The third indicates that each array
element accesses an integer.
or
p is an in parameter for a three-dimensional array of integer
or
p is an in-out parameter for a two-dimensional array of integer
or
... (any other interpretations?)

Now consider the design of a tool that will perform bounds-checking
on p, prior to run-time.
				Tim
Disclaimer: I speak for NO organization, and especially am not a
spokesperson for the Naval Postgraduate School, the Navy, the
Department of Defense or the US Government. 

Re: Study shows Ada tools ten times more costly

[dog.ee.lbl.gov!overload.lbl.gov!agate!howland.reston.ans.net!zaphod.mps.o]

In article <C784HG.5E0@world.std.com> srctran@world.std.com (Gregory Aharonian)
 writes:
>    One comment has some relevance to the issue of the Ada Mandate.  To
>quote (page 99):
>
>	"The cost of software engineering tools ranges from less"
>	than $1 per function point, to more than $20 per function
>	point.  Developers using PC, Unix and Apple Macintosh
>	platforms have the largest selection of tools priced at
>	less than $1 per function point.  Manufacturers and
>	military software developers often use tools priced at
>	more than $10 per function point.
>
>
>    This comment to some extent supports the contention that the Ada
>Mandate, by sheltering the Ada tools companies from competition, is distorting
>the marketplace and making use of Ada more expensive than other languages.
>(I assume in his study that for the most part, military software developers
>only are buying Ada tools).
>

Dadgummit, Greg!  This is NOT related to Ada!  Tool cost is almost
diretcly related to the PLATFORM, not the language.  I spent 10
years in the military, five in electronic warfare and five in software
development.  The costs I found were related to using tools on specialized
hardware platforms: Honeywell, Tandem, HP Special Test Harnesses,
and aging IBM boxes.  All these products are either low-volume commercial
products or special-purpose military equipment.

Now, I'm not excusing these costs either.  Our military culture is more
paralyzed than the culture of large corporations -- changing from mainframes
to low-cost PCs is prohibitively expensive in the short-run, which is
how the military gets its budget.  Worse yet, even IF they change to
PCs/Macs, security requirements drive up the hardware costs outrageously.
(New oxymoron: TEMPEST-approved Laptop :-)  


-- 
-Comments above aren't neceessarily the opinion of the SEI, AJPO, or CAE-Link-
David Weller  |  Have you hugged your DRAGOON lately?
----I'm the Ultimate International Masochist: I speak Ada AND Esperanto!-----

Study shows Ada tools ten times more costly

[Gregory Aharonian]

   In the January 1993 issue of Software Magazine, T. Capers Jones has a
nice article titled "Equipping the software engineer", in which he talks
about trends in software development tools as measured by their function
point capability.  He talks about investment in tools for software
engineers, as compared to other industries.

    One comment has some relevance to the issue of the Ada Mandate.  To
quote (page 99):

	"The cost of software engineering tools ranges from less"
	than $1 per function point, to more than $20 per function
	point.  Developers using PC, Unix and Apple Macintosh
	platforms have the largest selection of tools priced at
	less than $1 per function point.  Manufacturers and
	military software developers often use tools priced at
	more than $10 per function point.


    This comment to some extent supports the contention that the Ada
Mandate, by sheltering the Ada tools companies from competition, is distorting
the marketplace and making use of Ada more expensive than other languages.
(I assume in his study that for the most part, military software developers
only are buying Ada tools).

Greg Aharonian

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