A Quick Look at Object-Oriented Programming

Object-Oriented Programming is a modern approach to the old ideas of code modularity and reusabilty. Many programming languages provide support for OOP features, including C++, Smalltalk, LISP and Pascal. C++, due to the popularity of C, is the most widely used OOP environment so we will focus our short tour on it. COOOLhas been developed using the Free Software Foundation's C/C++ compiler gcc, which is freely available via ftp from prep.ai.mit.edu.

Anything other than a superficial account of OOP is beyond the scope of this paper. For more details see the book by Stanley Lippman cited at the end. Here we content ourselves to mention the basic ideas without really explaining them.

The objects in OOP, called classes in C++, can be abstractions of either the primary problem or its supporting implementation. The principal tasks are to define the classes and their inter-relationship among one another. These relationships can be of two types. One class can be an example of another-a house is a kind of building-or one class can be a member of another-a house has windows. We also must define the set of operations provided by each class; this is the class hierarchy.

All OO languages provide some support for encapsulation, sharing, and inheritance of code. We briefly explain these feature in terms of C++.

• Encapsulation is a mechanism for achieving code isolation: separating the external aspects of an object, which are accessible to other objects and users, from its internal implementation details. This isolation is highly desirable for large software projects since you don't want fiddling with the internal algorithm of some object to cause other code using this object to break.

• Sharing is accomplished in C++ via
  1. Overloading:
     We could, for example, define a new object called an array, and enrich the meaning of the binary operator + so that adding two arrays element by element is as simple as specifying .

  2. Templates:
     Templates allow one to define a family of similar objects that share most data attributes as well as methods. For example, we could define a sparse-matrix template which could represent either double, float, or integer sparse matrices.

• Inheritance is the way we share similarities among different classes while preserving their differences. In C++, a class can be derived from other classes which have similar, but more general attributes and methods. We could define a class representing, say, our abstract notion of a building. Then, later on, if we needed the more specific concept of a house, it could automatically inherit its building-ness from the base class building; adding on only what was spcific to the house type of building. The ability to derive one class from another enables one to define abstract classes which contain the common features of all their descendents.