Re: C is 'better' than Ada because...

[clodius@hotspec.lanl.gov (William Clodius)]

In article <01bb7bf9$b89a1740$96ee6fcf@timhome2> "Tim Behrendsen"
<tim@airshields.com> writes: 

   Theodore E. Dennison <dennison@escmail.orl.mmc.com> wrote in article
   <31F613F3.2781E494@escmail.orl.mmc.com>...
   > Tim Behrendsen wrote:
   > > 
<snip>
   > 
   > You have a good point. Unfortunately, your premise is wrong. Ada is in 
   > extensive use outside of the Government. In fact it is the language of 
   > choice wherever a software failure is liable to cause loss of life or a 
   > very large lawsuit. 

<snip>
   Based on your post, Ada is only appropriate for mission critical
                              ^^^^
   applications, and not for other applications? Again, the question,
                     ^^^
   and I ask this respectfully, "If it's so good, why isn't it used more
   universally than it is?"  And, if C is so bad (in many people's opinion,
   not mine), why is it used almost universally, despite its acknowledged,
   real problems?

First, Dennison's post indicates that Ada is believed to be the most
appropriate language for mission critical applications. That by itself
says nothing whatsoever about its applicability, or lack thereof, for
other applications.

Second, several people in this thread have already addressed your
question about why is C used far more often than Ada. The answer has
typically been that this difference in useage is almost entirely due
to social and historical factors, beyond the relative merits of the
languages. You may or may not believe this explanation, but if you
are truly respectful of those that have responded to your question you
would at least recognize that they have given an answer. Ideally you
would also note why you do, or do not, believe that their answer is
valid.

Third, while the intrinsic merits of a language are important, those
merits may differ in importance from problem to problem, and must
include such social and historical factors as: are there lots of
programmers available that are trained in this language; how easy is
it to interface to the operating system in this language; how
expensive are compilers for the language; are there non-compiler tools
available for the language that compensate for the language's
weaknesses; how much control do I want over the details of
implementing a given algorithm (Ada and C are very comparable on this
point); are compilers available for all platforms of interest (the is
portability practical question); etc. Most of the above
(specification of the algorithm independent of system is an
exception), are very much history or culture dependent:

1. It takes time to train students and there is a tendency to train
them in the most popular language.

2.  It takes time for an OS to be written and it is very expensive to
replace it.

3. A large part of the price of a compiler is determined by cumulative
sales volume. An old and popular language has less expensive compilers
than a new language.

4. An older popular language will have more tools available than a
newer language.

5. Many programmers will avoid pure functional languages, because of
their lack of control over the implementation of an algorithm,
although their semantics allows a compiler to generate more optimal
code. 

6. New languages take time to have wide availability.

-- 

William B. Clodius		Phone: (505)-665-9370
Los Alamos National Laboratory	Email: wclodius@lanl.gov
Los Alamos, NM 87545