Object Construction

---

You have seen how to write simple constructors that define the initial state of your objects. However, because object construction is so important, Java offers quite a variety of mechanisms for writing constructors. We go over these mechanisms in the sections that follow.

Overloading

Recall that the GregorianCalendar class had more than one constructor. We could use:

 GregorianCalendar today = new GregorianCalendar(); 

or:

 GregorianCalendar deadline = new GregorianCalendar(2099, Calendar.DECEMBER, 31); 

This capability is called overloading. Overloading occurs if several methods have the same name (in this case, the GregorianCalendar constructor method) but different parameters. The compiler must sort out which method to call. It picks the correct method by matching the parameter types in the headers of the various methods with the types of the values used in the specific method call. A compile-time error occurs if the compiler cannot match the parameters or if more than one match is possible. (This process is called overloading resolution.)

NOTE

Java allows you to overload any method not just constructor methods. Thus, to completely describe a method, you need to specify the name of the method together with its parameter types. This is called the signature of the method. For example, the String class has four public methods called indexOf. They have signatures

 indexOf(int) indexOf(int, int) indexOf(String) indexOf(String, int) 

The return type is not part of the method signature. That is, you cannot have two methods with the same names and parameter types but different return types.

Default Field Initialization

If you don't set a field explicitly in a constructor, it is automatically set to a default value: numbers to 0, Booleans to false, and object references to null. But it is considered poor programming practice to rely on this. Certainly, it makes it harder for someone to understand your code if fields are being initialized invisibly.

NOTE

This is an important difference between fields and local variables. You must always explicitly initialize local variables in a method. But if you don't initialize a field in a class, it is automatically initialized to a default (0, false, or null).

For example, consider the Employee class. Suppose you don't specify how to initialize some of the fields in a constructor. By default, the salary field would be initialized with 0 and the name and hireDay fields would be initialized with null.

However, that would not be a good idea. If anyone called the getName or getHireDay method, then they would get a null reference that they probably don't expect:

 Date h = harry.getHireDay(); calendar.setTime(h); // throws exception if h is null 

Default Constructors

A default constructor is a constructor with no parameters. For example, here is a default constructor for the Employee class:

 public Employee() {    name = "";    salary = 0;    hireDay = new Date(); } 

If you write a class with no constructors whatsoever, then a default constructor is provided for you. This default constructor sets all the instance fields to their default values. So, all numeric data contained in the instance fields would be 0, all Booleans would be false, and all object variables would be set to null.

If a class supplies at least one constructor but does not supply a default constructor, it is illegal to construct objects without construction parameters. For example, our original Employee class in Example 4-2 provided a single constructor:

 Employee(String name, double salary, int y, int m, int d) 

With that class, it was not legal to construct default employees. That is, the call

 e = new Employee(); 

would have been an error.

CAUTION

Please keep in mind that you get a free default constructor only when your class has no other constructors. If you write your class with even a single constructor of your own and you want the users of your class to have the ability to create an instance by a call to

new ClassName()

then you must provide a default constructor (with no parameters). Of course, if you are happy with the default values for all fields, you can simply supply

public ClassName()
{
}

Explicit Field Initialization

Because you can overload the constructor methods in a class, you can obviously build in many ways to set the initial state of the instance fields of your classes. It is always a good idea to make sure that, regardless of the constructor call, every instance field is set to something meaningful.

You can simply assign a value to any field in the class definition. For example,

 class Employee {    . . .    private String name = ""; } 

This assignment is carried out before the constructor executes. This syntax is particularly useful if all constructors of a class need to set a particular instance field to the same value.

The initialization value doesn't have to be a constant value. Here is an example in which a field is initialized with a method call. Consider an Employee class where each employee has an id field. You can initialize it as follows:

 class Employee {    . . .    static int assignId()    {       int r = nextId;       nextId++;       return r;    }    . . .    private int id = assignId(); } 

C++ NOTE

In C++, you cannot directly initialize instance fields of a class. All fields must be set in a constructor. However, C++ has a special initializer list syntax, such as:

 Employee::Employee(String n, double s, int y, int m, int d) // C++ :  name(n),    salary(s),    hireDay(y, m, d) { } 

C++ uses this special syntax to call field constructors. In Java, there is no need for it because objects have no subobjects, only pointers to other objects.

Parameter Names

When you write very trivial constructors (and you'll write a lot of them), then it can be somewhat frustrating to come up with parameter names.

We have generally opted for single-letter parameter names:

 public Employee(String n, double s) {    name = n;    salary = s; } 

However, the drawback is that you need to read the code to tell what the n and s parameters mean.

Some programmers prefix each parameter with an "a":

 public Employee(String aName, double aSalary) {    name = aName;    salary = aSalary; } 

That is quite neat. Any reader can immediately figure out the meaning of the parameters.

Another commonly used trick relies on the fact that parameter variables shadow instance fields with the same name. For example, if you call a parameter salary, then salary refers to the parameter, not the instance field. But you can still access the instance field as this.salary. Recall that this denotes the implicit parameter, that is, the object that is being constructed. Here is an example:

 public Employee(String name, double salary) {    this.name = name;    this.salary = salary; } 

C++ NOTE

In C++, it is common to prefix instance fields with an underscore or a fixed letter. (The letters m and x are common choices.) For example, the salary field might be called _salary, mSalary, or xSalary. Java programmers don't usually do that.

Calling Another Constructor

The keyword this refers to the implicit parameter of a method. However, the keyword has a second meaning.

If the first statement of a constructor has the form this(. . .), then the constructor calls another constructor of the same class. Here is a typical example:

 public Employee(double s) {    // calls Employee(String, double)    this("Employee #" + nextId, s);    nextId++; } 

When you call new Employee(60000), then the Employee(double) constructor calls the Employee(String, double) constructor.

Using the this keyword in this manner is useful you only need to write common construction code once.

C++ NOTE

The this reference in Java is identical to the this pointer in C++. However, in C++ it is not possible for one constructor to call another. If you want to factor out common initialization code in C++, you must write a separate method.

Initialization Blocks

You have already seen two ways to initialize a data field:

• By setting a value in a constructor

• By assigning a value in the declaration

There is actually a third mechanism in Java; it's called an initialization block. Class declarations can contain arbitrary blocks of code. These blocks are executed whenever an object of that class is constructed. For example,

 class Employee {    public Employee(String n, double s)    {       name = n;       salary = s;    }    public Employee()    {       name = "";       salary = 0;    }    . . .    private static int nextId;    private int id;    private String name;    private double salary;    . . .    // object initialization block    {       id = nextId;       nextId++;    } } 

In this example, the id field is initialized in the object initialization block, no matter which constructor is used to construct an object. The initialization block runs first, and then the body of the constructor is executed.

This mechanism is never necessary and is not common. It usually is more straightforward to place the initialization code inside a constructor.

NOTE

It is legal to set fields in initialization blocks even though they are only defined later in the class. Some versions of Sun's Java compiler handled this case incorrectly (bug # 4459133). This bug has been fixed in JDK 1.4.1. However, to avoid circular definitions, it is not legal to read from fields that are only initialized later. The exact rules are spelled out in section 8.3.2.3 of the Java Language Specification (http://java.sun.com/docs/books/jls). Because the rules are complex enough to baffle the compiler implementors, we suggest that you place initialization blocks after the field definitions.

With so many ways of initializing data fields, it can be quite confusing to give all possible pathways for the construction process. Here is what happens in detail when a constructor is called.

1. All data fields are initialized to their default value (0, false, or null).

2. All field initializers and initialization blocks are executed, in the order in which they occur in the class declaration.

3. If the first line of the constructor calls a second constructor, then the body of the second constructor is executed.

4. The body of the constructor is executed.

Naturally, it is always a good idea to organize your initialization code so that another programmer could easily understand it without having to be a language lawyer. For example, it would be quite strange and somewhat error prone to have a class whose constructors depend on the order in which the data fields are declared.

You initialize a static field either by supplying an initial value or by using a static initialization block. You have already seen the first mechanism:

 static int nextId = 1; 

If the static fields of your class require complex initialization code, use a static initialization block.

Place the code inside a block and tag it with the keyword static. Here is an example. We want the employee ID numbers to start at a random integer less than 10,000.

 // static initialization block static {    Random generator = new Random();    nextId = generator.nextInt(10000); } 

Static initialization occurs when the class is first loaded. Like instance fields, static fields are 0, false, or null unless you explicitly set them to another value. All static field initializers and static initialization blocks are executed in the order in which they occur in the class declaration.

NOTE

Here is a Java trivia fact to amaze your fellow Java coders: You can write a "Hello, World" program in Java without ever writing a main method.

 public class Hello {    static    {       System.out.println("Hello, World");    } } 

When you invoke the class with java Hello, the class is loaded, the static initialization block prints "Hello, World," and only then do you get an ugly error message that main is not defined. You can avoid that blemish by calling System.exit(0) at the end of the static initialization block.

The program in Example 4-5 shows many of the features that we discussed in this section:

• Overloaded constructors

• A call to another constructor with this(...)

• A default constructor

• An object initialization block

• A static initialization block

• An instance field initialization

Example 4-5. ConstructorTest.java

  1. import java.util.*;  2.  3. public class ConstructorTest  4. {  5.    public static void main(String[] args)  6.    {  7.       // fill the staff array with three Employee objects  8.       Employee[] staff = new Employee[3];  9. 10.       staff[0] = new Employee("Harry", 40000); 11.       staff[1] = new Employee(60000); 12.       staff[2] = new Employee(); 13. 14.       // print out information about all Employee objects 15.       for (Employee e : staff) 16.          System.out.println("name=" + e.getName() 17.             + ",,salary=" + e.getSalary()); 19.    } 20. } 21. 22. class Employee 23. { 24.    // three overloaded constructors 25.    public Employee(String n, double s) 26.    { 27.       name = n; 28.       salary = s; 29.    } 30. 31.    public Employee(double s) 32.    { 33.       // calls the Employee(String, double) constructor 34.       this("Employee #" + nextId, s); 35.    } 36. 37.    // the default constructor 38.    public Employee() 39.    { 40.       // name initialized to ""--see below 41.       // salary not explicitly set--initialized to 0 42.       // id initialized in initialization block 43.    } 44. 45.    public String getName() 46.    { 47.       return name; 48.    } 49. 50.    public double getSalary() 51.    { 52.       return salary; 53.    } 54. 55.    public int getId() 56.    { 57.       return id; 58.    } 59. 60.    private static int nextId; 61. 62.    private int id; 63.    private String name = ""; // instance field initialization 64.    private double salary; 65. 66.    // static initialization block 67.    static 68.    { 69.       Random generator = new Random(); 70.       // set nextId to a random number between 0 and 9999 71.       nextId = generator.nextInt(10000); 72.    } 73. 74.    // object initialization block 75.    { 76.       id = nextId; 77.       nextId++; 78.    } 79. } 

 java.util.Random 1.0 

• Random()

  constructs a new random number generator.

• int nextInt(int n) 1.2

  returns a random number between 0 and n - 1.

Object Destruction and the finalize Method

Some object-oriented programming languages, notably C++, have explicit destructor methods for any cleanup code that may be needed when an object is no longer used. The most common activity in a destructor is reclaiming the memory set aside for objects. Because Java does automatic garbage collection, manual memory reclamation is not needed and so Java does not support destructors.

Of course, some objects utilize a resource other than memory, such as a file or a handle to another object that uses system resources. In this case, it is important that the resource be reclaimed and recycled when it is no longer needed.

You can add a finalize method to any class. The finalize method will be called before the garbage collector sweeps away the object. In practice, do not rely on the finalize method for recycling any resources that are in short supply you simply cannot know when this method will be called.

NOTE

The method call System.runFinalizersOnExit(true) guarantees that finalizer methods are called before Java shuts down. However, this method is inherently unsafe and has been deprecated. An alternative is to add "shutdown hooks" with the method Runtime.addShutdownHook see the API documentation for details.

If a resource needs to be closed as soon as you have finished using it, you need to manage it manually. Supply a method such as dispose or close that you call to clean up what needs cleaning. Just as importantly, if a class you use has such a method, you need to call it when you are done with the object.

---