Re: Ada Pricing & Quality ?? - Vendors

Re: Ada Pricing & Quality ?? - Vendors

[cis.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!howland.]

Since the base note was fully compliant with my spec, I guess I have
to respond, so here goes:

1. Price

Ada prices are declining across the board.  However, you can't compare Unix
workstation prices with PC prices.  PC Ada compilers are priced much less
than your $30k, and (I hear) are quite competitively priced to C compilers.

What do you expect to pay for a non-free C or C++ compiler on a workstation?
I'd really like the input.

Also, cross-compilers generally cost more than native compilers because the 
development cost has to be amortized over a smaller market.

2. Useful libraries

We are planning interfaces to standard support libraries in our next-generation
products.

3. Documentation of limitations; verification of examples

Yes, quality is important, including document quality.  What else can
I say.

4. Are we willing to invest to make Ada succeed?  

YES.  WATCH THIS SPACE.

-- Mike Ryer

Re: Ada Pricing & Quality ?? - Vendors

[Robert Kitzberger]

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

>Less than $900, and I want a variety of CASE capabilities thrown in.  I can
>get this for C++, and I get a CDROM thrown in with lots of goodies.  Further,
>I can save much development time by reusing the many large commercial and
>public domain C/C++ systems available - a factor in choosing a compiler.
>Besides, a Pentium-based system is close enough to a workstation, for PC
>prices to set the standard for workstation prices.

What you describe applies well to the PC development market, but doesn't
scale up the the large systems developed in Ada.

Visual C++ gives you admirable support for single-user development of
Windows based applications.  It provides a great way to speed
development of Windows-based user interfaces.

It does NOT provide features necessary for developing large,
truly mission-critical applications (now there's an overused
buzzterm -- "we support your mission-critical payroll applications!")

This requires a real workstation (e.g. Unix) along with remote file
systems, multiple user configuration management, etc.  PCs with kludgy
networking setups don't cut it, unless you enjoy pain (I know, I know,
NT will solve the world's Unix problems ;-).  It also requires a
development environment engineered from the ground up to _really_
support multiple user development, definition and enforcement of
subsystem interfaces, strong typing (IMHO), inherent portability, etc.

Until you convince the world to start developing 10,000,000-line
applications on PCs under Windows in Visual C++, there will be a
market for development environments that _are_ targetted to that
market.

It's not by accident that Apple develops its software on Unix
workstations rather than PCs/Macs, and it's not by accident that large
systems are developed on Unix systems, VAXes, R1000s, etc. rather
than PCs, often in Ada.

On the other hand, I concede a subset of your point: there don't
appear to be any Ada development systems for PC applications that
provide the capabilities of Visual C++.  But the world does not revolve
solely around Windows and PC applications (I find that to be one of the
biggest mistakes people make when they generalize about development
systems, second to the embedded/host dichotomy).

I also agree that there is enormous opportunity for adding some of the
features that make Visual C++ so appealing to Ada development systems.
There is also enormous opportunity for C and C++ development system
vendors to add features to their environments that support development
of huge projects by hundreds of developers (I've received personal
email on the development of OS/2 that indicate that they really needed
a Rational-like development system, _not_ Visual C++).

Personally, I agree with Robert Dewar and Mike Feldman that the lack
of bindings (especially standardized ones, de-facto or otherwise) to
existing operating systems and user interfaces are the single largest
impediment to more widespread use of Ada, not the lack of OO features,
for example.

>>Also, cross-compilers generally cost more than native compilers because the 
>>development cost has to be amortized over a smaller market.
>
>Given that 85% of the embedded market is done in C/C++, as measured by use
>and sales of products (for example, Intermetrics C compiler), Ada will never
>be competitive in the embedded market as C/C++ is amortized over a much
>larger market.

Greg, most of the embedded systems market is composed of 8-bit
microcontrollers (I'll need to ferret out an issue of Embedded Systems
Programming that contains their survey results).  This is clearly the
realm of C and assembly, not Ada.  If you looked at the market for
large embedded systems, which is a small slice of the market compared
to 8-bit microcontrollers, you'd see mainly Ada and some
C/lint/RCS/cpp/make/xmkmf/imake systems.

>In general, as long as I can get Rational-like capabilities in C++ systems
>at PC prices, Ada will never be competitive.

I submit that you aren't getting Rational-like capabilities in
Visual C++; you're comparing apples and oranges.  

I don't mean to be hostile, and I'm certainly not speaking for Rational.

	.Bob.

--
Bob Kitzberger                          Internet:   rlk@rational.com
Rational, Grass Valley, CA              CompuServe: 70743,1550

Re: Ada Pricing & Quality ?? - Vendors

[Gregory Aharonian]

>1. Price
>Ada prices are declining across the board.  However, you can't compare Unix
>workstation prices with PC prices.  PC Ada compilers are priced much less
>than your $30k, and (I hear) are quite competitively priced to C compilers.

You hear wrong.  Compared to PC compilers such as Microsoft's Visual C++,
PC Ada compilers are uncompetitive.  See my post on Visual C++ killing Ada.
For very reasonable prices, you get many CASE capabilities for C++ systems,
while not so for Ada.

>What do you expect to pay for a non-free C or C++ compiler on a workstation?
>I'd really like the input.

Less than $900, and I want a variety of CASE capabilities thrown in.  I can
get this for C++, and I get a CDROM thrown in with lots of goodies.  Further,
I can save much development time by reusing the many large commercial and
public domain C/C++ systems available - a factor in choosing a compiler.
Besides, a Pentium-based system is close enough to a workstation, for PC
prices to set the standard for workstation prices.

>Also, cross-compilers generally cost more than native compilers because the 
>development cost has to be amortized over a smaller market.

Given that 85% of the embedded market is done in C/C++, as measured by use
and sales of products (for example, Intermetrics C compiler), Ada will never
be competitive in the embedded market as C/C++ is amortized over a much
larger market.

    In general, as long as I can get Rational-like capabilities in C++ systems
at PC prices, Ada will never be competitive.  And you can't just match prices;
you have to offer a significantly lower price, a lower price I don't think any
Ada compiler vendor can offer at this late stage in the language wars.  You
need to sell a price half as much for the same capability in another language
to win market share - something like a Visual Ada at $99.

>2. Useful libraries
>We are planning interfaces to standard support libraries in our nextgeneration
>products.

    Again, too little too late.  First you can't offer all of the library
support that exists in the C world.  Just look at the catalogs from places
like Programmers Shop - hundreds of C/C++ libaries for everything under the
sun.  Interfaces doesn't help, as it puts you behind the market in time, as
you have to wait until the C/C++ is out before you can start the interface.
Besides interfaces to another language always sends out the wrong signal,
like Cubans using the dollar for their economy.

    The Ada industry (if such a thing exists) should have been offering 
libraries starting in the late 1980's.  Ada83 is plenty good to be used for
such libraries.  But what kills the Ada library business (as measured by the
number of companies that went broke trying to do so), and Ada in general,
is that Mandated contractors don't buy libraries - thanks to Defense
procurement regulations and N-I-H syndromes.

>4. Are we willing to invest to make Ada succeed?  
>YES.  WATCH THIS SPACE.

    I have been watching this space, and the general industry, since the
mid-1980's.  Where have you guys been (not just Intermetrics - but all of the
vendors)?  Ada should have been advertised in the general computing media,
pushed at the general computing trade shows, and pitched to the editors many
many years ago.  Now whatever you do will be too little too late.

    When the Ada money was flowing freely in the 1980's, some of it should
have been used to kickstart an non-Mandated Ada industry.  But instead, it
was plowed back into short term sheltered profits of the Mandated world.
Even today, at major software CASE trade shows, there is at best only one
or two Ada booths, while at Tri-Ada there will be forty.  Too many are still
too dependent on the sheltered environment of the Mandate.

    I presented a lot of market data on C/C++'s dominance at WadaS, data
important to crafting a strategy to win market share for Ada outside the
Mandated world.  Since then not one Ada company has shown any interest in
the data.

    At every Tri-Ada since the mid-1980's, there should have been the
equivalent of a war map showing the market shares and trends of programming
languages, and at least one session comparing current Ada systems with
current C/C++ systems, if nothing else, to see how the enemy is doing.

    I will make a bet with you Mike (winner buys dinner at Joyce Chen) -
that relative market shares outside the Mandated world for Ada and C/C++
will not change much over the next three years, based on the collective
marketing strategies of the Ada compiler vendors as they are now formulated.

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

Re: Ada Pricing & Quality ?? - Vendors

[david.c.willett]

>From article <CDEvIJ.4ML@inmet.camb.inmet.com>, by ryer@dsd.camb.inmet.com (Mi
ke Ryer):
> Since the base note was fully compliant with my spec, I guess I have
> to respond, so here goes:
> 
> 1. Price
> 
> Ada prices are declining across the board.  However, you can't compare Unix
> workstation prices with PC prices.  PC Ada compilers are priced much less
> than your $30k, and (I hear) are quite competitively priced to C compilers.
> 
> What do you expect to pay for a non-free C or C++ compiler on a workstation?
> I'd really like the input.
> 
> Also, cross-compilers generally cost more than native compilers because the 
> development cost has to be amortized over a smaller market.
> 
> 2. Useful libraries
> 
> We are planning interfaces to standard support libraries in our next-generati
on
> products.
> 
> 3. Documentation of limitations; verification of examples
> 
> Yes, quality is important, including document quality.  What else can
> I say.
> 
> 4. Are we willing to invest to make Ada succeed?  
> 
> YES.  WATCH THIS SPACE.
> 
> -- Mike Ryer

You were doing just fine until #4, Mike.  I'm a software developer, not an
Ada vendor.  That makes me the Ada vendor's customer.  Why should I be willing 
to invest my Company's money to "make Ada succeed"?  Making Ada succeed is 
the vendor's job.  My job is to deliver quality software products to meet 
the needs of *my* customers.

As long as vendors have the idea that compiler customers should be "willing
to invest" to "make Ada succeed", it won't.  Ada will succeed if and only if
Ada vendors become convinced that they must offer products which are a
clearly superior way for their customers (programmers) to serve their 
customers (users).  Being clearly superior includes being cost effective.


<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

Dave Willett          AT&T Federal Systems Advanced Technologies

Remember Dave, you don't pay an anesthesiologist all that money to put you
to sleep, you pay him to wake you up afterwards!

	-- from a discussion on the value of medical specialists

Re: Ada Pricing & Quality ?? - Vendors

[cis.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!agate!sp]

In article <CDG9Ct.C3C@cbnewsl.cb.att.com> 
  willett@cbnewsl.cb.att.com (david.c.willett) writes:

>From article <CDEvIJ.4ML@inmet.camb.inmet.com>, 
  by ryer@dsd.camb.inmet.com (Mike Ryer):

>> Since the base note was fully compliant with my spec, I guess I have
>> to respond, so here goes:
>> ...
>> 4. Are we willing to invest to make Ada succeed?  
>> 
>> YES.  WATCH THIS SPACE.
>> 
>> -- Mike Ryer
>
>You were doing just fine until #4, Mike.  I'm a software developer, not an
>Ada vendor.  That makes me the Ada vendor's customer.  Why should I be willing
 
>to invest my Company's money to "make Ada succeed"?  Making Ada succeed is 
>the vendor's job.  My job is to deliver quality software products to meet 
>the needs of *my* customers.

I think by "Yes" Mike meant "we" Intermetrics are willing to invest to 
help Ada succeed, and "you" potential customers should keep an eye us 
to see if we fulfill this promise (I'm sure you will ;-).

Please don't flame us if you think we should have already invested
(we already have), or we should have invested more wisely or
visibly.  I was just trying to clarify who Mike meant by "we."

>Dave Willett          AT&T Federal Systems Advanced Technologies

S. Tucker Taft   stt@inmet.com
Intermetrics, Inc.

Re: Ada Pricing & Quality ?? - Vendors

[cis.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!howland.]

[drivel deleted]
>    The Ada industry (if such a thing exists) should have been offering
>libraries starting in the late 1980's.  Ada83 is plenty good to be used for
>such libraries.  But what kills the Ada library business (as measured by the
>number of companies that went broke trying to do so), and Ada in general,
>is that Mandated contractors don't buy libraries - thanks to Defense
                                    ^^^
>procurement regulations and N-I-H syndromes.
 
Mr. Aharonian chooses to ignore the rationale for the DOD's regulations
because it does not square with his business profits.
 
The defense department follows a simple set of rules.
 
1. A contractor may provide a reuse library to the DOD for a fee.
2. That library becomes government property.
3. To cut costs, other contractors should be able to use that software 
   free of charge.
 
The key fact to remember: If a contractor pays for a reuse library
originally developed under DOD sponsorship, they can not charge the 
DOD for its purchase price.  This prevents someone from double-charging 
the government for software.
 
Simple and smart economics. Also why someone like businessman Aharonian
has a chip on his shoulder when it comes to DOD procurement.
Why should a contractor pay a reuse librarian like STO for government 
software when (1) they can get it for free and (2) they won't be able to
charge it against software expenditures anyway?  (Hint: add some value)
 
Lon Smithers
 
"Will C and C++ merge? Or will a new dialect,like C+,emerge from the fusion?"
Grading system courtesy of P.J.Plauger, Dr.Dobb's Journal

Re: Ada Pricing & Quality ?? - Vendors

[haven.umd.edu!cs.umd.edu!afterlife!admii!ovation!hipmac1.pica.army.mil!us]

In article <rlk.748154105@bonnie>, rlk@bonnie.Rational.COM (Robert
Kitzberger) wrote:

[stuff deleted]
> 
> It's not by accident that Apple develops its software on Unix
> workstations rather than PCs/Macs, and it's not by accident that large
> systems are developed on Unix systems, VAXes, R1000s, etc. rather
> than PCs, often in Ada.
> 
Just picking a small nit really, but Apple doesn't develop it's software
on Unix workstations, at least not the majority of it.  They develop their
Mac software on Macs, using their own Mac Programmer's Workshop (MPW) shell
and compilers for that shell.  

Steve Wall

Re: Ada Pricing & Quality ?? - Vendors

[Richard G. Hash]

> >What do you expect to pay for a non-free C or C++ compiler on a workstation?
> >I'd really like the input.
> Less than $900, and I want a variety of CASE capabilities thrown in.  I can
> get this for C++, and I get a CDROM thrown in with lots of goodies.

This ($900) seems pretty low to me. Most folks spend more than that in *time*
on *free* compilers (e.g., gcc)! By the time I add up all the other tools
that we use for C development, I'm looking at a considerably larger number
than $900. On the other other hand, the $35K for XXX-brand Ada (hidden to
protect the atrociously guilty) on an rs6000 was enough to make alot of
folks I know choke to death.

>     In general, as long as I can get Rational-like capabilities in C++ system
s
> at PC prices, Ada will never be competitive.

Where can you get Rational-like capabilities for C++ (serious question, I
really don't know). Since I've used a Rational for the last 6yrs or so I've
gotten pretty used to some of the features. I've seem similar like stuff,
but it was pretty lightweight in comparison, and mostly had a pretty UI.

> >2. Useful libraries
> >We are planning interfaces to standard support libraries in our
> > next generation products.
> 
>     Again, too little too late.

This is oh-so-true. What about this-generation?

--
Richard G. Hash                                      email:  rgh@shell.com
Shell Development Company, Bellaire Research Center  phone: (713) 245-7311
3737 Bellaire Boulevard, Houston TX  77025