Re: Ada 83 - Sometimes still chosen

[Richard D Riehle <laoXhai@ix.netcom.com>]

In article <7dfqr1$g5j$1@nnrp1.dejanews.com>,
	robert_dewar@my-dejanews.com wrote:

>I disagree, I would say that sometimes the use of obsolete
>equipment is the right solution. I just bought a 1987 Ford
>F250 truck, definitely an obsolete piece of equipment, but
>just fine for my purposes.

 I suddenly have this image of Robert in a tattered baseball
 cap and faded bluejeans, a little spot of grease on the tip of his
 nose from tinkering under the hood of his pickup truck.  Those
 who know Robert might also be amused by this image.  Next, I
 expect him to reveal that he has abandoned the pipe organ for
 an avocation as a country and western singer with a steel string
 guitar.  :-)

 Seriously, though, we simply have different ideas of the meaning
 of obsolete.  I am typing this on one of my office computers, a
 Pentium 233.  It is of course, obsolete by one meaning.  It is not
 obsolete for this purpose.  The 1987 Ford is not obsolete if it
 does the job intended.

>Actually we see very few projects selecting the 1750A for
>new projects. There are other alternatives these days, such
>as the ERC.

 I just recently had yet another inquiry from a [prospective]
 client who has selected the 1750A for a space application. It
 turns out that the alternatives to 1750A are not universally
 trusted by developers of communications satellites.  BTW, is
 there an Ada 95 for ERC?
 
>These days, it is quite inexpensive to generate a new
>Ada 95 compiler, and if there are systems for which Ada 95
>is not used, 

 Some developers of space applications are conservative.
 If a technology is proven over time, has been successful
 on other projects, and personnel are experienced with
 it, the risk of staying with that technology might be deemed 
 less severe.  This is the well-known "If it aint broke don't
 fix it" viewpoint.  A surprising number of program managers
 adopt this peculiar stategy when designing for safety-critical
 software.  

>it is an indicator that very few new projects
>are choosing the hardware in question (yes, I realize that
>there is a chicken and egg problem, but it is minimal given
>the low cost of porting an Ada 95 compiler). If no Ada 95
>compiler is available for a given system, it almost
>certainly means that the customer demand for such a
>compiler is minimal or non-existant.

 I interpret this to mean that very few projects are inquiring
 of ACT about the availability of a GNAT solution.  I wonder if
 a compiler publisher such as DDCI is having the same experience?
 The last I heard, they were still selling a lot of Ada 83 compilers
 to their international customer base.  

>My fundamental point here is that the general impression
>of the community is that Ada 83 was a failure. Yes, that
>is hyperobole of course, but on the other hand, at this
>stage, Ada 83 is pretty creaky.

 Creaky?  I hear the passenger's door of Robert's Ford truck opening
 on a winter morning.  I hear the engine cranking its first rotation
 of the morning.  

 The success of Ada 83 on real projects would not lead me to use
 the word "creaky" to describe it.  I am, as are you, an advocate
 of transitioning to Ada 95 whenever feasible.  I am delighted
 with the improvements in the standard and agree that Ada is the
 current Ada 95.  On the other hand, many of those improvements are
 superfluous in the design of working software.  For the space
 applications mentioned earlier, one of the few really useful 
 improvements is unsigned numerics.  Many communication satellite
 designs avoid tasking, have no use for polymorphism, and are generally 
 conservative in the use of many other Ada features.    

>
>Just yesterday, I talked to someone doing research into
>object oriented component design. He was trying to fit
>into C++, and having various troubles. He made a list of
>deficiencies of C++, all fixed in Ada 95. He then was
>playing with a modified Java to accomodate his ideas, all
>of which could have been used directly in Ada 95.

 The list of deficiencies must have been long and mournful.
 I have decided that, since there is not perfect programming
 language, such comparisons are useful only for reinforcing
 a previously held viewpoint. 

>When I asked him about Ada, he replied "Oh, does Ada have
>object oriented features, I didn't know that".

 We need to do more to get the word out on the new Ada standard.
 Notice that few computer conferences have any participants from
 the Ada community.  An exception to this is the annual TOOLS USA
 conference which has been accepting Ada tutorials and speakers for
 several years now - along with other languages.  

>So of the people asking questions about Ada (without
>specifying which version), the overwhelming majority are
>talking about Ada 95, and it is a definite disservice to
>tell people that Ada 83 is still in wide use and that they
>must worry about the Ada 83 answer as well as the Ada 95
>answer. 

 Ada 83 is still in wide use.  It may be in wider use than Ada 95.
 Those who are using it also have questions.  Most of those who use
 Ada 83 do not frequent this list.  They are in the trenches grinding
 out code.  Many are new to Ada.  When they do find this forum and 
 need to get the answer to a question, they could be unaware of the
 existence of a new standard (as was you C++ colleague).  All they
 want is to get a question answered. If they fail to specify the 
 version of Ada, it is because they are still tilling the fields 
 originally sown with MIL-STD 1815, the only Ada they know.

 Sometimes, because so many of us are active in the current version
 of the language, it is hard to remember that so many are still 
 struggling with the issues of the previous standard.  

 It will not cause any permanent psychological damage for us to be
 charitable toward those gentle jungle folk seeking help from
 an unrehabilitated crocodile.

This would be like asking a question on the C++
>group, and having responders worry that you might really be
>asking about C!

 HmmMMMMmmmmm. Does C++ have templates? Depends of what version
 of C++ you are using.  Does C++ have something like an Ada 
 package?  It depends on which version of C++ you are using and
 how you understand "namespace."   

 I agree with most of your observations, Robert.  Ada 95 is Ada.
 Not everyone programming in Ada 83 knows this.

 Richard Riehle
 richard@adaworks.com
 http://www.adaworks.com