NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/lang/cplus/18230

Internet Message Format | 1992-12-21 | 3.8 KB

Path: sparky!uunet!dtix!darwin.sura.net!max.fiu.edu!serss0!feathers
From: feathers@serss0 (Michael Feathers)
Newsgroups: comp.lang.c++
Subject: Re:  Give me safe C++
Message-ID: <[e-mail]>
Date: 19 Dec 92 13:11:46 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail] (Usenet Administrator)
Organization: Florida International University, Miami
Lines: 69

In article <[e-mail]> [e-mail] (Richard Bielak) writes:
>In article <[e-mail]> [e-mail] (Richard M. Hartman) writes:
>
>[...]
>
>>Safety (when not being enforced by the language) is to a large
>>extent (IMHO) a matter of good programming practices.  Like a
>>seemingly ever-growing segment of the general populace, the cry
>>comes "oh powers that be, protect us from ourselves", and thus
>>we get laws requiring people to wear seatbelts.  It is no that
>>I am against seatbelts (or safety), but I was perfectly capable
>>of wearing a seatbelt before the law existed.  To my mind, there
>>was no *need* for the law.
>
>This analogy with seat belts bugs me. Here is why. Seatbelts are there
>for the benefit of consumer of cars.
>
>This implies that programmers are *consumers* of programming
>languages. This misses the point, that programmers are the
>*manufacturers* of software.
>
>Don't you agree, that safety regulations forced on the manufacturer
>are a good thing? Rememeber the Ford Pinto?

I don't.  The bad press that Ford got because of the Pinto out stripped any
incentive towards safety that a regulation could generate.  Believe it or not,
safety is a side effect in a free market.

>If you are writing safety critical software - let's say software that
>controls the brakes in *my* car - I'd rather have you code in PASCAL
>and run with all the runtime checks on. I don't want my brakes to stop
>working just because you forgot an ampersand.

I wouldn't rather have the code written in Pascal.  The reason has to do with
one of my experiences as an undergraduate.

Before I started majoring in computer science, I took it upon myself to teach
myself C as a first language.  Everyone I knew at the time said that C was
a hard language to learn and I figured that if I was capable of doing that I
probably wouldn't get half way through my degree and then discover that I was
not cut out for programming.

Well, I spent a good amount of time writing over the edges of arrays and
dereferencing null pointers before I finally figured out that I should really
understand my code before it executes and construct it in such a way that I
know when a run time check is or is not necessary.  I learned this lesson
thoroughly only after getting extremely tired of crashing in C.

In my last year as an undergraduate, I remember looking over my shoulder in the
labs at the terminal of my neighbor.  He had a run-time bounds error message
on his screen.  At that point I had been doing Pascal projects for a few years
at school and I had never seen one of those before.  Checking around, I
discovered that they were not too uncommon.  My conclusion was that run time
checking was being used as a crutch.

If I had it my way, I'd have C taught as a first programming language in CS
curricula.  If nothing else it would teach students that the best safety
measure they can take with their code is to understand it.  The iterative
cycle of getting run time errors, patching them and then getting more run time
errors should never have to occur and should never be relied upon my
beginning students.

I've seen students who learn C first have no problem wih Pascal other than
claustrophobia, and I've seen students who learn Pascal first have an
extremely agonizingly difficult time learning C because it does not have their
familiar crutches.


Michael Feathers     [e-mail]