Data Abstraction and Object Oriented Programming CS 351

Data Abstraction and Object. Oriented Programming CS 351 – Programming Paradigms

Object Oriented Programming The development of complicated computer applications has seen the need for data abstraction with the software engineering field. The main advantages of data abstraction are: l l 1. 2. 3. l It reduces the conceptual load for the programmer. It provides a method of fault containment by preventing the programmer from using code in an inappropriate way. It increases the independence among program components. Programmers can work on individual parts of the code without affecting the interface between components. The final point is difficult to achieve in the ``wild’’

Defining a class – C++ Style class list_node { private: list_node* prev; list_node* next; list_node* head; public: int val; list_node(); ~list_node(); list_node* get_prev(); list_node* get_next(); void insert(); void remove(); }; Access Specifiers Destructor ; after the }

Defining a class – C++ Style class list { public: int empty(); list_node* head(); void append(list_node* ln); ~list(); }; Booleans are ints in C++! In C++, the boolean type can be represented by either the bool type which works identically to the Java version or through the use of an integer value ( carried over from C). l Any non-zero integer value evaluates to true in C++. l

Defining a class - cont This class definition is usually contained within a ``header’’ file, typically labelled file. h To create instances (objects) of the list_node class: l l 1. 2. l list_node element; list_node * element = new list_node(); Both are equally valid ways of creating the object in C++, what is the difference between the two? The invocation of the special method to create an object is called the constructor. In programming languages that do not feature automatic memory management, the objects must be deleted manually via a special method known as the destructor.

Data Access l l l OOP allows data to be hidden and protected in a way that is different to imperative programming. We can declare data that belongs to a class as either public or private. This allows us to control the visibility of the data when classes interact. In C++, every member is private by default, but the use of the public: syntax allows any fields that follow to be declared as public. This is in contrast to the simple nature of the struct in C. All of the data members in structs are public in C but private in C++!!

Subroutines These are referred to as methods in OOP. l As data members can be declared to be private, it is convention to supply get and set methods for each private member. l Our code for list_node should now show the following changes: l private: … int val; public: void set_val ( int v ); int get_val (); …

Derived Classes l l In OOP we can choose to have an abstract idea coded in a base class. We can then derive a concrete class from the base class. This derivation allows the derived class (child class, sub class) to inherit all of the features of the parent class. By deriving new classes from old ones, the programmer can create arbitrarily deep class hierarchies with added functionality at every level of the hierarchy. Our code from earlier now looks like this: l class queue : public list { public: void enqueue (list_node * n); list_node * dequeue(); }; Inheritence in C++

Accessing Derived Methods l In the base class, the method definiton for remove could look like this: void list: : remove() { head = head->next; } In the derived class, queue, how do we call methods from the parent class? l In Java we can use the super keyword to call methods and fields from the parent class. l To call methods belonging to a class in C++, we use the scope resolution operator : : l void queue: : remove() { list: : remove(); }

Encapsulation & Inheritance Encapsulation is a technique that allows the programmer to group data and the subroutines that operate on that data and place them together in a single place. With the ability to inherit fields and methods from base classes, some rules for data hiding and data management must be defined. Some questions to be answered include: l l l 1. 2. 3. Should private members of a base class be visible to methods of a derived class? Should public members of a base class always be public members of a derived class? How much control should a base class have over the visibility of its members?

Example class list { public: int empty(); list_node* head(); void append(list_node* ln); ~list(); }; class queue : private list { public: using list: : empty; using list: : head; void enqueue (list_node * n); list_node * dequeue(); }; We are extending the class list privately. What difference does this make? We must explicitly make public methods that are privately derived accessible. We use the using keyword for this purpose.

C++ Visibility Rules Any class can limit the visibility of its members. 1. i. iii. iv. A derived class can restrict the visibility of members of a base class but can never increase it. 2. i. iii. 3. Public members are visible anywhere the class declaration is in scope. Private members are visible only inside the classes methods. Protected members are visible only inside the methods of the class or any of its descendants. An exception, the friend keyword, what does that do? Private members of a base class are never visible in a child class. Public and protected members of a base class are public or protected respectively in a derived class. Public and protected members of a privately derived base class are private members of the derived class. A protected or privately derived base class can have its members visibility restored through an explicit using declaration.

Comparison versus Java l l 1. 2. Java also features the keywords public, private and protected. There are slight semantic differences In Java though: Classes cannot be derived as protected or private, hence a derived class cannot increase or restrict the visibility of members of the base class. The protected keyword has a different meaning In Java also. A protected member is visible not only in the base class and derived classes but also throughout the whole package it is declared in.

Constructors and Destructors l l l l Most object-oriented languages provide some features for initialising an object at the start of its lifetime. This special method is known as a constructor. This constructor initializes the space in memory that has been allocated. How does the compiler know to allocate the space? Some other languages provide methods, known as a destructor, to provide some housekeeping after the object has reached the end of its lifetime. We have seen that the tilda, ~, is used in C++ to indicate that the special method is a destructor. How do you create a destructor in Java?

Execution Order for Constructors When an object of a derived class is created in C++, the compiler guarantees that the constructor for any base classes will be executed, outermost first, before the constructor of the derived class. l How do we manage this in C++, given the absence of a super keyword? l This is the same as in Java: class queue : public list queue ( args ) { { super ( list args ); … } public : queue: : queue( args passed to queue ) : list ( args for list ) {…} … }; l Similar syntax can also be used to pass default values to a constructor. list_node: : list_node( ) : val ( 0 ) { }

Dynamic Method Binding One of the main advantages of inheritance is that some derived class D has all of the members of its base class B. l Once D is not hiding any of the public members of B, then it is possible for an object of D to represent B in any context where a B could be used. l This feature is known as subtype polymorphism. l

An Example of Dynamic Binding class person { … void person : : print() { // prints details for the person } }; class student : public person { … void student : : print () { //adds more specific information } }; class lecturer : public person { … void lecturer : : print () { //adds more details } }; … student *s = new student (); lecturer *t = new lecturer (); s -> print person *x(); = s; This is a polymorphic tperson -> print*y(); = t; x -> print () ; y -> print () ; call

Example cont… person *x = s; person *y = t; x -> print () ; y -> print () ; l Does the choice of the method to be called depend on the types of x and y? l l Does the choice of the method to be called depend on the classes of the objects s and t to which those variables refer ? l l If so then this is known as static binding. If so then is known as dynamic binding. Dynamic binding is central to OOP and ensures that even if we are using the base class to refer to a child class, the correct methods will always be called.