Re: Which first-course languages? (was: What schools use Eiffel (was: No top schools use Ada)) ?

[scottw@bmtech.demon.co.uk (Scott Wheeler)]

In Article <SPENCER.95Apr20102028@calvin.ERA.COM> Spencer Allain 
writes:

>> I'm hoping that the only reason for the statement

>> >>   but we wouldn't even have considered employing him if he hadn't 
        learnt C under his own steam.

>> was because the job involved real-time systems where every 
millisecond counts or the company didn't want to invest a couple of 
weeks teaching a new person the syntax of a particular language.

No. As a matter of fact, we don't want him to program in C - at present 
our main language is C++, but I don't have a religious attachment to 
that (as a matter of fact, I used to use Modula as my main language 
before moving to C). Also in a computer science course I would prefer 
that people start with a "purer" language: my preference at the moment 
would be for Eiffel. My point (obscurely made) was that a CS course 
should not *only* deal with academic languages, particularly ones as 
far removed from everyday life as Miranda. If we take Eiffel as an 
example, it has certain specific lessons to teach regarding 
encapsulation, data hiding, re-use by templates. The main lesson, of 
course, is the very practical idea of programming by contract, in order 
to build reliable software. Fine. 

Now suppose someone takes a course which emphasises Eiffel heavily. 
Almost certainly the student gets a job involving programming, and 
almost certainly it won't be in Eiffel. To profit from the lessons 
learned, the ex-student has to apply the same lessons to the 
environment being used - C++ is quite likely these days, and it is 
sufficiently flexible to support most of the same ideas. In practice, 
what we get is likely to be very poor, because the course didn't cover 
adapting the techniques to the environments which will probably be 
encounted, and most students won't think this through in depth 
themselves. 

Now why don't we use Modula, or Oberon, or Eiffel, or whatever? It's 
not because C++ is the flavour of the month, or not directly. It's 
because the factories we work with want to use PCs, preferably running 
Windows (don't tell me you don't like it - *we* don't like programming 
it!). While all of those languages are supported on PCs, the compilers 
and environments (debuggers, GUI class libraries, browsers, integrated 
editors) are very weak compared to C++, the proprietary Borland 
implementation of Pascal, and "Visual BASIC". As I mentioned, I prefer 
Eiffel myself, but the compiler I have is simply not a viable 
development environment, and NICE hasn't standardised the vital class 
libraries yet so we would be locked in to one vendor. I would prefer it 
if the "teaching" languages could also be "working" languages, but as 
yet only Pascal has made that transition, albeit heavily extended.

An final point is that it is certainly possible to learn the syntax of 
most languages in a couple of weeks, as you mention. The trouble is 
that this is like teaching someone to drive by showing them the 
controls of a car: it's only a beginning, it doesn't show them how to 
drive safely or comfortably.

>> Personally, I've found that individuals who have a background in 
many languages (ie. different types of thinking) are generally "better"
programmers.  They may not be able to optimize a loop as well as a
hacker who really knows one language, but they often find a way to
remove the loop completely by using a different concept.

I absolutely agree. I think the only way we can argue over this is 
whether modern programmers should learn some assembler (which side do 
you want to argue?)

Scott (pass me that COBOL manual) Wheeler