Re: Ada as a first language

["Brian" <phaedrusalt@hotmail.com>]

"John McCormick" <mccormick@cs.uni.edu> wrote

>Perhaps if you read the portion of my articles that explains WHY Ada
>succeeds where C fails you could expand the domain far beyond my
>trains.  My conclusions on the problems with scalars has been
>confirmed in numerous domains.

>I agree that Ada is not suitable for many domains, but embedded
>systems is one domain for which it is generally extremly well
>suited.

John,

Ada works great, better than other languages, for certain situations.  Not 
all situations.  So, let's list a few "rules of thumb" about programming:
1. Short programs tend to be better (More successful) when they're written 
in the language that the implementer knows best.  (Barring prior knowledge, 
then these short programs tend to be more successful in languages that are 
easier to learn.)
2. Some languages make handling some domains easier than others.
3. For programs under 100 SLOC, it might be a wash as to which language was 
better suited.  (As long as the language supports the domain.)

Now, what should we expect from a person with no programming experience, 
experimenting with embedded robotics?  Probably small programs, for a good 
while.  Probably most interested in seeing action from the end effectors 
than anything else.  Now, what environment(s) would be most likely to give 
this person the greatest success with the least pain?  Not Ada.

The conclusions reached in your article are correct, up to a point. 
However, to make your point you abstracted two variables, the user's prior 
experience and the complexity of the problem space.  (I don't blame you, the 
addition of those two variables would have turned your paper into a 
ponderous tome!)  But, in this case these two variables are very important, 
in fact they're the only information given.  Roughly applying my rules of 
thumb (Paragraph 1) to my assumptions (Paragraph 2), we get:

A. The user will probably want to write short programs because he has no 
prior experience.  He might tend to be more successful in a language that is 
easy to learn.  And for programs under 100 SLOC, it might be a wash as to 
which language was better suited.
B. The user will probably most want to see action from the end effector(s), 
and some languages might make handling the robotic domain easier.

So, language probably doesn't matter and user success is of paramount 
importance. In this case, don't you think that an easier environment might 
be a better choice?  If NXT-G (The Lego Mindstorms graphical language) is 
too simple, then there are many other choices available before I'd recommend 
Ada to a beginner.  I realize that we all love using Ada, but shouldn't we 
consider the person's needs first?

Brian Salter