Re: What's the best language to learn? [was Re: Should I learn C or Pascal?]

[seebs@solutions.solon.com (Peter Seebach)]

In article <01bb89f1$31be4f60$87ee6fce@timpent.airshields.com>,
Tim Behrendsen <tim@airshields.com> wrote:
>>I am not sure I buy your interpretation.  Thinking like a programmer is
>>largely a matter of learning to find useful abstractions.  How do you write
>>code to test whether a number is even?  Well, you do it by looking at what
>>evenness is, and looking for what characteristics it has that would help you.

>Solving a problem is not an
>"abstraction" question, it's a question of breaking it down
>into fundamental steps of data transformations.

No, that's implementing a solution.  Different stage.  Consider Cantor's
diagonal.  A perfectly good algorithm which you *cannot* implement.

>Testing a
>number as even is transforming a number into a bit based on
>its evenness property.

But how you do that has *NOTHING* to do with the question of whether or not
you have defined a solution to the problem.

Whether or not you can do that will answer the question "is my solution going
to work on existing machines".  But if you can't, you still have a solution,
it's just not one which meets your requirements.

>So what?  You can always learn languages, but you only have
>one shot at teaching someone to think like a programmer.

Nonsense.  People learn new things all the time.

>*Programming is not languages*.  Programming is breaking problems
>down into fundamental steps to achieve a desired result.  The
>way the steps are described is not relevent to the general question
>of a solution.  When we are talking about teaching a student,
>the less extraneous issues such as languages and the more
>experience we can given them of the fundamental procedural
>nature of the computer, the better.

I don't really think we are going to convince each other.  I believe that
the "fundemental procedural nature of the computer" may not be a permanent
thing.  I'd rather teach them to think in the general terms, and see that
procedural nature as just one more current limitation, like computers which
have only 2 MB of RAM.  This will better prepare them for techniques that
depend on *not* thinking of the computer as procedural.  Look at Icon...
Neat language, but if you think in terms of normal procedural code, it will
never work for you.

>> You keep using assembly as an example of what the basics of computing are
>> like.  Please explain how this is a better model *of the problem domain*
>> than lisp or C.  Why should students try to learn machine-level and
>> program-level at the same time?  If you want them to start with
>> architectures, don't start them on algorithms, start them on trivial data
>> manipulation in assembly.

>It's a better model because it's real.  C and all languages are
>artificial constructions intended for the maintainance of large
>projects.

So?  They're just as real.  *computers* are artificial constructions.

>What gets missed in all this is that *computers are simple*!
>All they do in the fundamental sense is take a very simple
>instruction, execute it, and repeat very, very fast.  If this
>is presented the first day, the student can't help but realize
>the simple nature of the computer, and it loses its mystery.

I think you underestimate the idiocy of many college students.  :)

Still, why wouldn't, say, scheme, work for this?  You can make a *VERY*
simple system - much simpler than assembly for any platform - and show
the students how everything is built on it.

If all you want to do is show them that the computer is simple, teach them
in Turing machine assembly.  :)

>Look at ol' Darin; if he didn't have enough brainpower left
>to focus on the algorithm at hand, what was he being confused
>by?  I think it was by all the language and abstraction, which
>is not what programming is.

But it *is* what programming is.  Programming is turning a problem
description into a solution description.  The implementation is merely
coding.  Turning a solution description into a program can be done
by trained monkeys.  (Or, emprically, compilers.)

>If he had been given a solid
>foundation of the simple nature of the computer, I think he would
>always have a "comfort zone" he could return to whenever he
>didn't understand something, because it's *always* implemented
>in the fundamental terms.

But why insist on making the comfort zone so far from the nature of the
task?

This is like teaching people to drive by starting them out with metallurgy.
Sure, it's fundemental to the operation of the car, *but they don't need to
know it*.  It can help, but it's not a good basis for an understanding at
the right level.

-s
-- 
Peter Seebach - seebs@solon.com - Copyright 1996 - http://www.solon.com/~seebs
Unix/C Wizard - send mail for help, or send money for consulting!
The *other* C FAQ, the hacker FAQ, et al.  See web page above.
Unsolicited email (junk mail and ads) is unwelcome, and will be billed for.