TEAM-ADA Archives

Team Ada: Ada Programming Language Advocacy


Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Wesley Groleau <[log in to unmask]>
Reply To:
Wesley Groleau <[log in to unmask]>
Thu, 14 Jan 1999 12:21:39 -0500
text/plain (102 lines)
I realize this was not solicited from me but I'd like to provide you with
my professional opinion regarding the selection of the programming language
for [project].  I don't want to bore you with a lot of technical pros and cons
regarding this issue, but I strongly believe that you can do the job
cheaper and quicker if you use c/c++.  This choice makes a lot of sense for
several important reasons.

   "I don't want ... "  Translation: I don't have any empirical data, so I'll
   give you my non-expert preference.

First, if the OS is Solaris or HP-UX, you will get for free a very
extensive set of c libraries for free.  These will include everything from

   You already answered this quite well.

math libraries to interprocess communications facilities to memory
management functions.  My experience with Ada is that the compilers,
support libraries, and tools will always lag behind what is available for

   This is a genuine difference, but the difference is small and getting
   smaller.  It is more than drowned out by the fact that C in general is a

c, especially in the Unix environment.  You can expect that things which
could be implemented in c very easily will be very difficult if not
impossible with Ada.  Workarounds will be required.  And, it will be

   I hear variations of this all the time.  The followup question is "What do
   you mean by 'implemented'?"  The truth is, in a race to get something to
   compile, first place is the expert C programmer, close for second & third are
   the expert Ada programmer and the average C programmer, last is the average
   Ada programmer.  In a race to get something to run, it's about the same.  In
   a race to get something to run and not crash, Ada pulls ahead.  In a race to
   get it to run correctly, the C programmers are several laps behind.  Have you
   read John McCormick's description of the experience in switching from C to
   Ada in the classroom?  Basically, even though over the years, much of the
   low-level code for a project in C had been developed and GIVEN TO the
   students, a large majority never completed the project.  When they switched
   to Ada,  there were no libraries to give them, yet over half completed.
   After a few basic packages had been developed, the percentage was 75%.  If
   you can't find this report, and you want it, let me know and I'll search.
   See also the Ziegler report at

necessary to use assembly language with Ada in time-critical areas.

   Sometimes true, but in general, see

Performance requirements will more easily be met using c since c compilers
generate more compact code, and the programmer has more control over the
resultant machine code that is generated.

   This is blatant bull.  Even though C compilers have a twenty-year head-start
   on Ada, Ada compilers (on average)  produces more efficient code than C
   compilers (on average), IF the programmers doesn't do a lot of clever tricks
   to make their code unmaintainable (and unreliable).  When these kind of
   clever tricks are present, often as not the code is LESS efficient.  In a
   book older than C, Kernighan and Plauger wrote:  "Let the compiler do the
   simple optimizations."  The reason Ada compilers generate better code is that
   the language tells the compiler more about the constraints, AND the
   programmer has less freedom to interfere with the optimizer doing its job.

Second, the evolution of c to c++ has eleviated many of the government's
concerns which originally drove the development of Ada as a separate

   This guy obviously has ZERO knowledge of the history of Ada, and not enough
   about the history of C++.  Before the name Ada was coined, Bell Labs
   unequivocally admitted that C was so far below the reliability and
   readability requirements that it should not even be evaluated as a possible
   starting point.  The evolution of C++ has, as far as I can tell, alleviated
   NONE of the points where C fails to meet Steelman requirements.

language.  In particular, c++ is now a fully object-oriented language.  As
a consequence, the government is no longer requiring Ada on many (most?)

   OO is entirely unrelated to the relaxing of the mandate.  And Ada is now a
   fully object-oriented language.  The difference is that C++ evolved (in a
   rather disorganized manner) while Ada's OO features were carefully

programs.  And, if we use Rational Rose, using its code generation
capabilities it may be possible to auto-generate a significant number of
coding modules in c with Rose.  I don't believe Rose can generate Ada code.

   You already answered this quite well.

Third, although in this business we have a some people who have familiarity
with Ada, the rest of the programming world is using c.  It will be easier
to find people with the requisite programming skills if we use c, and to
interest them in working on the project.  Programmers will always prefer c
to Ada.

   I am a programmer who has programmed in C.  I still do occasionally.  As a
   result, I will never accept a job offer from a company that is so
   short-sighted as to prefer C over Ada across the board.  As for preference,
   find a hundred good Java programmers, and teach half of them C or C++ and
   half Ada.  More than a few will prefer Ada over Java.  NONE of them (or
   almost) will prefer C or C++ to Java.