Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!caip!topaz!husc6!seismo!umcp-cs!smd
From: smd@umcp-cs.UUCP (Stanley Dunn)
Newsgroups: net.lang
Subject: Re: Proofs &c (was Re: Trichotomous String Comparison Example)
Message-ID: <2313@umcp-cs.UUCP>
Date: Wed, 9-Jul-86 07:51:08 EDT
Article-I.D.: umcp-cs.2313
Posted: Wed Jul  9 07:51:08 1986
Date-Received: Fri, 11-Jul-86 04:49:20 EDT
References: <12221138454.23.PATTIS@WASHINGTON.ARPA> <2308@umcp-cs.UUCP>
Reply-To: smd@maryland.UUCP (Stanley Dunn)
Organization: University of Maryland, Dept. of Computer Sci.
Lines: 66

In article <2308@umcp-cs.UUCP> chris@maryland.UUCP (Chris Torek) writes:
>
>[Incidentally, the introductory programming course at the University
>of Maryland actually *does* avoid integers.  For their first year,
>students use something called `CF Pascal', in which the only legal
>data types are characters and files.  

For the first semester course (CMSC 112) the students use CF Pascal.
In the second semester course (CMSC 113) the students are exposed
tothe full language in a disciplined fashion.

>                                       The students are supposed to
>be able to prove their programs correct, and having few data types
>(one) and `Turing-tape-like' stores (files) helps keep the proof
>complexity down.

They are able to prove their programs correct.
But more importantly, they have learned how to solve 
problems rigorously.


>  I do not believe that these restrictions are
>sensible beyond the stage of learning to write and prove algorithms,
>and I am not even certain they are sensible there.  It does provoke
>novel solutions to old problems, however.]
>

I am sure that many others agree, however this is exactly
the point of view that people have to change.

Many students feel that CF Pascal is a "crutch" and that they should
be allowed to use the full language.  At least the students have an
excuse in that they are young and inexperienced.  The fact
of the matter is that the opposite is true - everything beyond
CF Pascal are merely shorthand notations for cumbersome operations
in CF Pascal, and that if the programmer has mastered the 
concept of Procedural Abstraction there should be no conceptual
difference.

Learning a science is not learning all the details about
the tools you are working with - that sort of training
makes you a technician.  To be a scientist or an engineer
requires that you have a deeper level of understanding 
of the properties of the tools you are using to solve problems.
Computer Science should be no different, and CF Pascal
helps development the thought processes of the 
students (with appropriate exercises, of course).
The projects you use have to be geared for CF Pascal without 
doing any arithmetic or random access manipulation.

Two examples of projects I gave my students to write
were an interpreter and a program for symbolic evaluaton of boolean
expressions.  There are many problems that are hard to solve 
because the concepts are difficult, and not because of CF Pascal.
In fact, many of the projects I used were just character and 
string processing.

This approach to teaching should not be condemmed without
a thorough understanding of the material, after which you 
can give a mature, educated evaluation.  Next week a short
course is being offered here (Dept. of Computer Science, UMCP)
for those interested in teaching Computer Science with 
the Mills' text and this structured approach (CF / D / O Pascal).
Harlan Mills, Vic Basili, John Gannon, and myself will
be going over the text, and relating personal experience
from teaching the two courses CMSC 112  and CMSC 113 here.