NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part1/4311

Text File | 1996-08-06 | 5.3 KB | 100 lines

Newsgroups: comp.lang.c++
Path: tank.news.pipex.net!pipex!swrinde!gatech!psinntp!psinntp!psinntp!psinntp!bbnews1!trsvr!news
From: [e-mail] (Benjamin Romer)
Subject: Re: C, C++, or both..??
Sender: [e-mail] (cnews news id.)
Message-ID: <[e-mail]>
Date: Mon, 29 Jan 1996 16:34:12 GMT
X-Nntp-Posting-Host: bmr1.tr.unisys.com
References: <4ehif2$[e-mail]>
Organization: Unisys Corporation
X-Newsreader: WinVN NT 0.92.6

In article <4ehif2$[e-mail]>, [e-mail] (The Crow) says:
>
>I am 17 years old, I have a rudimentary understanding of both C and C++ at this
>point.(I know Pascal well) I want to really delve into something at this point.
>My eventual goal is to either start my own software company, or work for one as
>a programmer. Should I concentrate only on C++ for now, only C for now, or
>both? Or does it not really matter? I have read up on C++ and I really don't
>see much benefit from it. People have kept applications in order (mostly) for
>years without 'object oriented' programming...and C seems to be faster and much
>less of a pain.

C++ has *a lot* on "good old C", one aspect being that it is Good Old C, but
with a whole bunch of useful extensions. Although from a rudimentary standpoint,
it may seem like there's little use for OOP, once you begin to realize some of the
advantages that it gives, you'll wonder how you got along without them. For instance,
writing an application for Windows in straight C requires you to do a lot of low and
mid-level programming, everything from interpreting the message to resize a window
to sending a keyboard keypress to the right place. In a higher-level language like
C++, you can write code to do these things _once_, then reuse the code for
multiple projects. This allows the programmer to concentrate on _solving_the_problem_
rather than dealing with system architecture.  Several premade libraries of code exist
for just this purpose; I can think of two examples, Borland's OWL system, and the
Microsoft MFC system, that you could look into for a better idea of what I mean.

Another benefit can be seen when you begin to work in large groups on a project.
In C, when two people would have to work together, they'd have to agree on a specific
way to name every function and structure to prevent them from redefining another's
code. With C++, The programmers can each be handed a design specification by
the project leader, and they can use each other's objects without having to know
specifics. For instance, suppose a colleague and I are writing a program, and we
need to implement both a tree and a linked list. in straight C, we'd have to define
TreeGetParent() and ListGetParent(), while in C++, we can both use GetParent() and
let the compiler worry about which one to use. Although this example is trivial,
I can easily imagine cases in which the same _semantic_ action must be applied
to many different data structures, making calling the correct function much harder.

This leads me to a third point. There are a few things that you can do with C++ that
you just cannot match with straight C. Using C++ inheritance, we can define a set
of things, say, cars, which may all work in different ways, but are all basically the
same. I can make a Car class in C++, then derive DieselCar and ElectricCar from
that class. Then, suppose I have a Race() function [for a game, perhaps!] to which
I get a pointer to a Car:

void Race(Car *move)
{
    move->Accelerate();
}

Not only do I not need to know which kind of Car it is, electric or diesel, but I also
don't need to wonder what is the proper function to call! In straight C, you'd need
the type of car as well as a switch statement, which could get *UGLY* if there's
thirty or fourty types of cars.

>I have a few programs in mind were object orientation would be a more natural
>way of going about the program (evolution similuations) but a lot of programs
>don't really benefit from it at all.  The biggest improvements in C++ that I
>can see at this point are COUT/CIN, the comments, and NEW and DELETE...the rest
>just seems to make me type and think more.  Is it because I am just not used to
>the idea yet or is it really more complicated? The thing I hate most is the way
>member functions have to be outside of their class, totally counter intuitive
>and makes things very hard to follow.

Actually, you can define the functions inline. Usually it is not done so that the
interface to the code can be examined without having to read the implementation.
This is another basis of OOP: the programmer, given an interface to the object,
shouldn't have to know how it works "on the inside". Suppose you've got a digital
watch; there are buttons for displaying the date, setting the alarm, etc. Do you
need to know how the internal circuitry works to press the buttons?

I'd suggest you get a good book on OOP. Try "Developing C++ Software" by
Russel Winder, published by Wiley Professional Computing. It should fill out
some of the holes in your experiences with C++, and answer some of the
questions you have about using C++ effectively. If you'd like to contact me,
send mail to [e-mail], or preferrably [e-mail].

Ben Romer
Software Engineer
Unisys Corporation

#include <stddisclaim.txt>



>The Crow - [e-mail]
>"It can't rain all the time"
>-Kryptology
>