As an example, we're going to create a simple class of object for use with my simple video game kernel in a new version I'll be introducing in an article in the near future. The class definition consists mostly of constructor methods, since the class itself is presently not much more than a Rectangle with an added field.
import java.awt.*;
// A Simple class for use in the simple video game examples.
// Mark Graybill, Aug. 2010
public class Brick extends Rectangle{
Color brickColor;
public Brick(int newX, int newY, int newWidth, int newHeight){
super(newX, newY, newWidth, newHeight);
brickColor = new Color(0, 128, 255);
}
public Brick(int newX, int newY){
this(newX, newY, 10, 10);
}
public Brick(){
this(0,0,10,10);
}
public void setColor(Color newColor){ brickColor=newColor; }
public Color getColor(){ return brickColor; }
} // End Brick
In this class, we have three forms of constructor, each with a different set of parameters. The one that takes the most parameters is the "base" version. It starts with a call to super(). This calls the constructor for Brick's superclass, or parent class, Rectangle. A look at the documentation for Rectangle shows that it has a constructor that takes four integer arguments, as we use here in super().
When we use super(), it must be the first thing we do within our constructor method. If we don't use super(), Java will do it automatically as the first thing in a constructor, calling it with no parameters. Since we want to set values for our inherited fields of x, y, width, and height it's better to call super() with those parameters. Otherwise, we could just as well have done something like this:
public Brick(int newX, int newY, int newWidth, int newHeight){
x=newX;
y=newY;
width=newWidth;
height=newHeight;
brickColor = new Color(0, 128, 255);
}
This would have the same effect, and Java would insert an invisible call to super() in front of x=newX;.
The other constructors use the first method. To do this, they use another special method that's like super(). It's called this(), and it calls another constructor for this class. We can't do a call to Brick(), if we try, the compiler will see it as an undefined symbol:
>javac Brick.java
Brick.java:11: cannot find symbol
symbol : method Brick(int,int,int,int)
location: class Brick
Brick(0,0,10,10);
^
1 error
So we use this() instead.
Like super(), the this() method must be the first thing called in the constructor method's body. Since don't call super()--it's in the base constructor--so there's no conflict about which goes first. If you use this(), you don't use super().
By using this() with the other constructor methods, we can keep all our key code code for the constructor in one place. If we had each constructor setting member values and constants without calling the base constructor method, then we'd end up with repeated code--a prime opportunity for bugs to enter our code if we update the code in one place, but not another. We don't want repeated code! Multiple independent constructors are used in the Java Tutorials, but I expect they're used to keep the lesson simple, not because they are a good coding practice!
You can find out more about this() and super() in the Java Language Specification, under Explicit Constructor Invocations.
You can probably see that I should really have my base constructor allow a Color to be passed to it, too. Try adding this yourself as an exercise to try out your understanding of constructor methods, then see what javac thinks of your work.