Path: utzoo!attcan!uunet!cs.utexas.edu!tut.cis.ohio-state.edu!unmvax!ncar!tank!uxc!uxc.cso.uiuc.edu!m.cs.uiuc.edu!p.cs.uiuc.edu!gillies
From: gillies@p.cs.uiuc.edu
Newsgroups: comp.edu
Subject: Re: Texts on fundamentals of programmin
Message-ID: <82400035@p.cs.uiuc.edu>
Date: 4 Apr 89 15:41:00 GMT
References: <354@cbnewsc.ATT.COM>
Lines: 35
Nf-ID: #R:cbnewsc.ATT.COM:354:p.cs.uiuc.edu:82400035:000:1588
Nf-From: p.cs.uiuc.edu!gillies    Apr  4 10:41:00 1989


> The approach to problem sets in the class at MIT was great
> - they would give you lots of code and you would then have to figure
> it out and then modify or extend it. Much more of a real-world
> approach than writing everything from scratch. 
>
>				Peter Desnoyers

I agree that one valid way to teach programming is to have students
modify code.

But I have to disagree furiously with the overall implication.  I
recently reviewed my course notes and problem sets from 6.001, when it
was taught in CLISP ('81).  More than HALF THE PROBLEMS ARE AI TOY
PROBLEMS!!!!  I also took intro to C.S.  for majors at Illinois, and
the problems were much more realistic.

Face it, you're not going to be write lisp interpreters, reprogram
eliza's, simulate infinite lists of primes, or reimplement MacSyma out
in the real world.  A LARGE fraction of our problems in 6.001 could
not easily be translated into other languages.  The course is heavily
dependent on the intricacies of LISP!

The course is very unrealistic.  It is fun, and it teaches recursion
beautifully, it neglects documentation totally, but it is VERY UNREALISTIC....


Don Gillies, Dept. of Computer Science, University of Illinois
1304 W. Springfield, Urbana, Ill 61801      
ARPA: gillies@cs.uiuc.edu   UUCP: {uunet,harvard}!uiucdcs!gillies

P.S.  When I took the course, to get an 'A' you had to understand how
to implement LISP closures on a LISP frame stack.  Yes, I got an 'A',
and this knowledge has proven to be about as useful as being able to
jump through a hoop.  In fact, I can't see the fundamental difference!