An Introduction to Polymorphism in Java

AP Computer Science Curriculum Module: An Introduction to Polymorphism in Java Dan Umbarger AP Computer Science Teacher Dallas, Texas © 2008 The College Board. All rights reserved. College Board, Advanced Placement Program, AP, SAT, and the acorn logo are registered trademarks of the College Board. connect to college success is a trademark owned by the College Board. Visit the College Board on the Web: www.collegeboard.com An Introduction to Polymorphism in Java The term homonym means “a word the same as another in sound and spelling but with different meaning.” The term bear could be a verb (to carry a burden) or it could be a noun (a large, hairy mammal). One can distinguish between the two usages through the use of context clues. In computer science the term polymorphism means “a method the same as another in spelling but with different behavior.” The computer differentiates between (or among) methods depending on either the method signature (after compile) or the object reference (at run time). In the example below polymorphism is demonstrated by the use of multiple add methods. The computer differentiates among them by the method signatures (the list of parameters: their number, their types, and the order of the types.) // A Java program written to demonstrate compile-time // polymorphism using overloaded methods public class OverLoaded { public static void main(String [] args) { DemoClass obj = new DemoClass(); System.out.println(obj.add(2,5)); // int, int System.out.println(obj.add(2, 5, 9)); // int, int, int System.out.println(obj.add(3.14159, 10)); // double, int } // end main }// end OverLoaded public class DemoClass { public int add(int x, int y) { return x + y; }// end add(int, int) public int add(int x, int y, int z) { return x + y + z; }// end add(int, int, int) public int add(double pi, int x) { return (int)pi + x; }// end add(double, int) }// end DemoClass This form of polymorphism is called early-binding (or compile-time) polymorphism because the computer knows after the compile to the byte code which of the add methods it will execute. That is, after the compile process when the code is now in byte-code form, the computer will “know” which of the add methods it will execute. If there are two actual int parameters the computer will know to execute the add method with two formal int parameters, and so on. Methods whose headings differ in the number and type of formal parameters are said to be overloaded methods. The parameter list that differentiates one method from another is said to be the method signature list. There is another form of polymorphism called late-binding (or run-time) polymorphism because the computer does not know at compile time which of the methods are to be executed. It will not know that until “run time.” Runtime polymorphism is achieved through what are called overridden methods (while compile-time polymorphism is achieved with overloaded methods). Run-time polymorphism comes in two different forms: run-time polymorphism with abstract base classes and run-time polymorphism with interfaces. Sometimes run-time polymorphism is referred to as dynamic binding. Types of Run-Time Polymorphism There are five categories or types of run-time polymorphism: 1. 2. 3. 4. 5. Polymorphic assignment statements Polymorphic Parameter Passing Polymorphic return types Polymorphic (Generic) Array Types Polymorphic exception handling (not in AP subset) 2 1. Polymorphic Assignment Statements When learning a new concept, it is often helpful to review other concepts that are similar and to use the earlier, similar skill as a bridge or link to the new one. Look at the following declaration: int x = 5; double y = x; // results in y being assigned 5.0 This is an example of “type broadening.” The int x value of 5, being an int which is a subset of the set of doubles, can be assigned as the value of the double y variable. On the other hand, double x = 3.14; int y = x; results in the compile error message “Possible loss of precision.” The JVM knows that it will have to truncate the decimal part of 3.14 to do the assignment and is fearful to do so, thinking that you have made a mistake. You can assure the JVM that you really do know what you are doing and really do wish to effect that truncation by coding a “type cast.” double x = 3.14; int y = (int) x; At right is some curious code to analyze. The variable value y received from x was originally an int value (5), but we are not allowed to assign that value (5) to the int variable z without a type cast on y. It seems as though the “type broadening ” from 5 to 5.0 has somehow changed the nature of the value. This situation will be helpful to remember in another few pages when we discuss a concept called “downcasting.” public class DownCast { public static void main(String [] args) { int x = 5; double y = x; //int z = y; y = x = 5 right??? int z = (int)y; // now it’s O.K. }// end main }// end class Possible loss of precision (compile error) Consider the following example. In the figures shown here boys and girls enter a gymnasium where they become generic sports fans, but are not allowed to enter gender-specific restrooms without first being converted back (type cast) to their specific gender types. Sports fans boys in a girls gymnasium boys girls boys’ restroom girls’ restroom 3 We now move from discussing primitive variables to object reference variables. The figure at the right pictorially represents an “is-a” relation between two classes. ClassB is an extension of ClassA. ObjY is a type of ClassA, but objX is not a type of ClassB. This relation is not symmetrical. Class A Class B public class PolymorphicAssignment { public static void main(String [] args) { ClassA obj1 = new ClassA(); ClassA obj2 = new ClassA(); ClassB obj3 = new ClassB(); 1) 2) 3) 4) 5) 6) obj1 = obj2; // no problem here...same data types obj1 = obj3; // obj3 is a type of ClassA…ok //obj3 = obj2; // "incompatible types" compile message //obj3 = obj1; // still incompatible as the obj3 value // stored in obj1 (see line 2 above) // has lost its ClassB identity obj3 = (ClassB)obj1; // the ClassB identity of the object // referenced by obj1 has been retrieved! // This is called "downcasting" obj3 = (ClassB)obj2; // This compiles but will not run. // ClassCastException run time error // Unlike obj1 the obj2 object ref. variable // never was a ClassB object to begin with objX objY public class ClassA { }// end ClassA public class ClassB extends ClassA { }// end ClassB } // end main }// end class In the code above, line 1 is a snap. Both object reference variables obj1 and obj2 are of ClassA() type. Life is good. Line 2 works because obj3 is an object reference variable of type ClassB, and ClassB type variables are a type of ClassA. Obj3 is a type of ClassA. Life is still good. Line 3 will not compile, as the code is attempting to assign a ClassA variable value to a variable of ClassB type. That is analogous to trying to assign a double value to an int variable. Line 4 is more complicated. We know from line 2 that obj1 actually does reference a ClassB value. However, that ClassB information is now no longer accessible as it is stored in a ClassA object reference variable. Line 5 restores the ClassB class identity before the assignment to ClassB object reference variable obj3 with a type cast. Life is good again. Line 6 is syntactically equivalent to line 5 and will actually compile because of it, but will result in a “ClassCastException” at run time because obj2 never was ClassB data to begin with. Exercise 1: 1. How is line 2 above conceptually similar to the discussion of “type broadening?” Use the term “is-a” in your response. 2. What is wrong with the code in lines 3 and 4 above? Relate to the discussion on the previous page. 3. Why isn’t line 4 okay? From line 2 aren’t you assigning a ClassB value to a ClassB variable? 4. Lines 5 and 6 are syntactically equivalent. They both compile but line 6 will not execute. Why? Explain the difference between those two lines. 4 Code to Demonstrate Polymorphic Assignment import java.util.Random; public class PolyAssign { public static void main(String [] args) { Shape shp = null; Random r = new Random(); int flip = r.nextInt(2); if (flip == 0) shp = new Triangle(); else shp = new Rectangle(); System.out.println("Area = " + shp.area(5,10)); } // end main }// end class abstract class Shape { public abstract double area(int,int); } // end Shape public class Triangle extends Shape { public double area(int b, int h) { return 0.5 * b * h; } }// end Triangle. public class Rectangle extends Shape { public double area(int b, int h) { return b * h; } }// end Rectangle Output is chosen at random: either Area = 25 (area triangle) or Area = 50 (area rect) Here we see run-time polymorphism at work! The JVM will not know the value of variable “shp” until run time, at which time it selects the area() method associated with the current object assigned to “shp.” The “abstract” specification on class Shape means that that class cannot be instantiated and only exists to indicate common functionality for all extending subclasses. The “abstract” specification on the area method means that all extending subclasses will have to provide implementation code (unless they also are abstract). 5 2. Polymorphic Parameter Passing Early in the Java curriculum we encounter the syntax requirement that actual (sending) and formal (receiving) parameters must match in number, type, and sequence of type. With polymorphism we can write code that appears to (but does not actually) break this rule. As before, we use the idea of “type broadening” of primitives as a bridge to understanding how polymorphism works with parameter passing of objects. If we write the code: int actual = 5; method(actual); and the method public void method(double formal) { implementation code } the int actual parameter is “broadened” to a 5.0 and received by parameter formal. Note that double actual = 5.0; method(actual); and the method public void method(int formal) { implementation code} would result in a “type incompatibility” compile error message unless a type cast were made on the actual (sending) parameter first: method( (int) actual); We now move from the discussion of type broadening of primitives to the idea of polymorphic parameter passing. We create objects for each of the two classes shown at right here and proceed to show their objects being passed as parameters to a method. public class ClassA public class PolymorphicParameterPassing { Class A { public static void main(String [] args) }// end ClassA { ClassA obj1 = new ClassA(); public class ClassB extends ClassA Class B ClassA obj2 = new ClassA(); { ClassB obj3 = new ClassB(); 1) method1(obj1); 2) method1(obj3); 3) //method2(obj1); 4) obj1 = obj3; 5) //method2(obj1); 6) method2((ClassB) obj1); 7) // method2((ClassB) obj2); } // end main public static void method1(ClassA formal) {} public static void method2(ClassB formal) {} }// end class }// end ClassB In line 1, at left, an object reference variable of ClassA type is passed to method1 and received as a ClassA object reference variable. Actual and formal parameter types are the same. Life is good! Line 2 shows a ClassB object reference variable passed to and received as a ClassA type variable. This is okay, as a ClassB type variable “is-a” type of ClassA variable. Line 3 fails, as you are passing a superclass type variable to be received as a subclass type. It seems as though line 5 should work, as obj1 received the value of a ClassB variable, but it doesn’t work unless the ClassB identity is restored through a type cast as shown in line 6. Line 7 will compile, as it is syntactically the same as line 6, but line 7 will result in a “type cast exception” upon program execution. 6 Code to Demonstrate Polymorphic Parameter Passing import java.util.Random; public class PolyParam { public static void main(String [] args) { Shape shp; Shape tri = new Triangle(); Shape rect = new Rectangle(); Random r = new Random(); int flip = r.nextInt(2); if (flip == 0) shp = tri; else shp = rect; printArea(shp); // output will vary } // end main } // public static void printArea(Shape s) { System.out.println("area = " + s.area(5, 10)); }// end printArea() end class abstract class Shape { abstract double area(int a, int b); }// end Shape public class Triangle extends Shape { public double area(int x, int y) { return 0.5 * x * y; } }// end Triangle public class Rectangle extends Shape { public double area(int x, int y) { return x * y; } }// end Rectangle Exercise 2: What is the output of this program? Why? 7 3. Polymorphic Return Types When returning values from return methods we know that there must be a type compatibility between the type of the variable receiving the value and the value being returned. For example: We can code: int x = retMethod(); and public int retMethod(){ return 5;} double x = retMethod(); and public int retMethod(){ return 5;} because the value being returned (5) is a type of double and, after the int 5 is “broadened” to a 5.0, that value can be assigned to the double variable x. The code int x = retMeth2(); with public double retMeth2() { return 5;} results in a “type compatibility” error message (even though 5.0 was originally an integer value) unless we type cast the double return value as: int x = (int) retMeth2(); Again we move from discussion of type broadening of primitives to the idea of polymorphic return types. We create objects for each of the two classes shown at right here and proceed to show their objects being used in return methods. public class PolymorphicReturnTypes { public static void main(String [] args) { ClassA obj1 = new ClassA(); ClassA obj2 = new ClassA(); ClassB obj3 = new ClassB(); 1.) 2.) 3.) 4.) 5.) 6.) obj1 = method1(); obj1 = method2(); //obj3 = method1(); // incompatible types //obj3 = method3(); // incompatible...why? obj3 = (ClassB) method3(); //obj3 = (ClassB) method1(); Class A Class B public class ClassA { }// end ClassA public class ClassB extends ClassA { }// end ClassB } // end main public static ClassA method1() { return new ClassA(); } public static ClassB method2() { return new ClassB(); } public static ClassA method3() { return new ClassB(); } }// end class 8 Exercise 3: 1. 2. 3. 4. 5. 6. Why does line 1 compile and execute? Why does line 2 compile and execute? Why does line 3 fail to compile and execute? Why does line 4 fail to compile and execute? How is line 4 different from line 3? Line 5 is similar to line 4. How does it succeed when line 4 fails to compile? Line 6 is similar to line 5. Line 6 will compile but not execute. Why? 9 Code to Demonstrate Polymorphic Return Types import java.util.Random; public class PolyReturn { public static void main(String [] args) { Shape shp = retMethod(); System.out.println(shp.area(5, 10)); } // end main public static Shape retMethod() { Random r = new Random(); int flip = r.nextInt(2); if (flip == 0) return new Triangle(); else return new Rectangle(); }// end retMethod() }// end class abstract class Shape { public abstract double area(int a, int b); }// end Shape class Triangle extends Shape { public double area(int x, int y) { return 0.5 * x * y; } }// end Triangle class Rectangle extends Shape { public double area(int x, int y) { return x * y; } }// end Rectangle Exercise 4: What is the output of the preceding code? Why? 10 4. Polymorphic (Generic) Array Types Arrays in Java, as with other languages, are homogeneous data types. This means that every array position must have the same type, hence the same length. int arr = new int[1000]; // every array position reserves space for 4 bytes (in Java) The declaration above reserves 4000 (1000 * 4) bytes of contiguous memory—each 4 bytes of memory storing an int value. Combining homogeneous position length together with contiguous memory is what gives arrays their O(1) access. In the figure below, storage or retrieval of data in position x can be determined by the formula “address of arr[x] = base + (x * 4).” Hence if the first item in the array is located in memory position 3b4f then the xth position in the array would be 3b4f + (x * 4). 0 3b4f 1 2 … … … x … 3b4f + (x * 4) 999 With “one-based counting” the same addressing formula would be “address of data in array = base + (x – 1) * 4.” This is why computer scientists like “zero-based counting.” Thus contiguous memory arrays allow us to use a single identifier for many, many different items of data while still having O(1) access. We need only to observe the restriction that every position in the array reserves the same amount of memory. If, instead of working with an array of primitives, we declare an array of objects, we can still achieve O(1) access but also achieve the effect of a “virtual” heterogeneous array. This effect will require some background to understand but is important as it is the “trick” behind polymorphic arrays. 11 The code example below is used to prepare the reader for the idea of polymorphic arrays. public class Poly1 extends java.applet.Applet { public void paint(Graphics g) { Head objRef0 = new Head(); objRef0.draw(g); RightEye objRef1 = new RightEye(); objRef1.draw(g); LeftEye objRef2 = new LeftEye(); objRef2.draw(g); Nose objRef3 = new Nose(); objRef3.draw(g); Mouth objRef4 = new Mouth(); objRef4.draw(g); }// end paint }// end class poly1 public class Head { public void draw(Graphics g) { g.drawOval(75, 75, 400, 400); } }// end class Head public class RightEye { public void draw(Graphics g) { g.drawOval(150, 200, 60, 40); } }// end class RightEye public class LeftEye { public void draw(Graphics g) The code in method paint above is implemented by the JVM by creating five object reference variables, each referencing an object as shown in the figures below. { g.drawOval(300, 200, 60, 40); } }// end class LeftEye public class Nose { public void draw(Graphics g) { g.drawOval(250, 300, 30, 30); } objRef0.draw(g); objRef1.draw(g); objRef2.draw(g); objRef3.draw(g); objRef4.draw(g); }// end class Nose compile time polymorphism public class Mouth { public void draw(Graphics g) Each respective draw() method here has a different implementation which draws a different part of the face. { g.drawArc(175, 300, 200, 100, 180, 180); } }// end class Mouth objRef0 obj0 draw() objRef1 objRef2 obj1 draw() objRef3 obj2 draw() objRef4 obj3 draw() obj4 draw() 12 What we wish to do is to rewrite the code above so that the effect will be as follows: obj3 draw() obj1 draw() 0 1 2 3 obj2 draw() obj0 draw() 4 obj4 draw() If we could place all the object references inside an array, then the code objRef0.draw(g); objRef1.draw(g); objRef2.draw(g); objRef3.draw(g); objRef4.draw(g); for (int k = 0; k < 5; k++) refArray(k].draw(g); could be replaced by Run-Time Polymorphism Compile-Time Polymorphism The benefit of this new code would increase as the number of desired draw methods increased. The problem of achieving such code is the requirement that all contiguous memory arrays be typed—the same type— so that each array position will be the same length. What type could we choose? We might try Object refArray = new Object[5]. But then there would be the problem of “type casting” each object back to its reference variable subclass type. Java 1.5 automatically unboxes for the ArrayList but only if a common type can be specified when the ArrayList is declared: ArrayList refArray = new ArrayList(); What could we use for ? What we need is an abstract “place-keeper” type—sort of like a variable in Algebra I, but representing an unknown type as opposed to an unknown value. abstract class BaseClass draw(); class Head draw(); class RightEye draw(); class LeftEye draw(); class Nose draw(); class Mouth draw(); Classes Head, RightEye, LeftEye, Nose, and Mouth all extend from abstract class BaseClass. Therefore each of those classes “is-a” type of BaseClass. Hence the code we are looking for to instantiate an array of generic type would be: BaseClass [] refArray = new BaseClass[5]; 13 Below is the code to draw the happy-face figure using a “polymorphic array” with an abstract base class type. import java.awt.*; public class PolyArray extends java.applet.Applet { public void paint(Graphics g) { BaseClass [] refArray = new BaseClass[5]; class Head extends BaseClass { public void draw(Graphics g) { g.drawOval(75, 75, 400, 400); } }// end class Head refArray[0] = new Head(); refArray[1] = new LeftEye(); refArray[2] = new RightEye(); refArray[3] = new Nose(); refArray[4] = new Mouth(); class RightEye extends BaseClass { public void draw(Graphics g) for (BaseClass e: refArray) e.draw(g); }// end class RightEye }// end paint }// end class PolyArray abstract class BaseClass { abstract void draw(Graphics g); } class Nose extends BaseClass { public void draw(Graphics g) { g.drawOval(250, 300, 30, 30); } }// end class Nose { g.fillOval(150, 200, 60, 40); } class LeftEye extends BaseClass { public void draw(Graphics g) { g.drawOval(300, 200, 60, 40); } }// end class LeftEye class Mouth extends BaseClass { public void draw(Graphics g) { g.drawArc(175, 300, 200, 100, 180, 180); } } // end class Mouth 14 Exercise 5: Using a polymorphic array and abstract base class, draw the house below. 15 There is another way to effect a polymorphic array, through a technique known as using a generic interface (base class). An interface (base class) is similar to an abstract base class but with differences. The table below compares an abstract base class with an interface base class. They both can serve as “generic” types. Abstract Base Class 1) Must be specified as abstract. Classes are concrete by default. 2) Can have both variables and constants 3) Usually has a mix of concrete and abstract methods. 4) Subclasses extend abstract base classes. 5) Subclasses can only extend 1 abstract base class. 6) Methods can be private, public, or protected. Interface (abstract by default) 1) Abstract may be specified but is default abstract regardless. Does not implement code! 2) Can only have constants, no variables. 3) Can only have abstract methods. Interface methods may be specified as abstract but are default abstract regardless. 4) Subclasses implement an interface (base class). 5) Subclasses can implement more than 1 interface. 6) All methods (actually method headings) are public. Below is the code to draw the happy-face figure using a “polymorphic array” with an abstract interface class type. import java.awt.*; public class Head implements BaseInterface { public void draw(Graphics g) public class PolyArray extends java.applet.Applet { public void paint(Graphics g) { BaseInterface [] refArray = new BaseInterface[5]; { g.drawOval(75, 75, 400, 400); } }// end class DrawHead refArray[0] = new DrawHead(); refArray[1] = new DrawLeftEye(); refArray[2] = new DrawRightEye(); refArray[3] = new DrawNose(); refArray[4] = new DrawMouth(); public class RightEye implements BaseInterface { public void draw(Graphics g) for (BaseInterface e: refArray) e.draw(g); }// end class RightEye { g.fillOval(150, 200, 60, 40); } public class LeftEye implements BaseInterface { public void draw(Graphics g) }// end paint }// end class poly1 { g.drawOval(300, 200, 60, 40); } interface BaseInterface { /* abstract */ void draw(Graphics g); } public class Nose implements BaseInterface { public void draw(Graphics g) { g.drawOval(250, 300, 30, 30); } }// end class Nose }// end class LeftEye public class Mouth implements BaseInterface { public void draw(Graphics g) { g.drawArc(175, 300, 200, 100, 180, 180); } } // end class Mouth 16 5. Polymorphic Exception Handling (Optional, not in AP subset.) To catch only the large fish in the figure below we would use the net with the large mesh, which would allow the two other smaller fish to swim right through the net. Then, to catch the middle-sized fish, we would use the net with middle-sized mesh, allowing the smaller fish to swim through the net. On the other hand, if we used the net with the smallest mesh on our first cast we would catch all three fish at once. Catching exceptions with try-catch blocks is similar to catching fish of different sizes, due to the polymorphic relation of the exception classes. In the API segment below, you can see that the bottom class in the extended inheritance relation indicated is the “ArrayIndexOutOfBoundsException.” Catching an exception with a “ArrayIndexOutOfBoundsException” object is analogous to catching the fish with the large mesh. It can only catch the largest fish. Due to polymorphism, the higher up one goes in the chain of inheritance the more errors can be caught (analogous to using a finer-mesh fishnet). The ability to catch errors selectively allows for different sections of recovery code targeted at specific error situations. 17 import java.util.ArrayList; import static java.lang.System.*; public class PolyException { public static void main(String [] args) { int [] x = new int[10]; String str = “Dallas”; try{ x[-1] = 3; // error #1 int y = str.charAt(-1); // error #2 int z = 5/0; // error #3 } // catches all three errors shown in try block catch(RuntimeException e) { out.println(“RuntimeException”); } // end catch RuntimeException // catches "big fish", then "middle fish", then "little fish" /* catch (ArrayIndexOutOfBoundsException e) { System.out.println("ArrayIndex OutOfBoundsException"); }// end catch ArrayIndexOutOfBoundsException catch(IndexOutOfBoundsException e){ System.out.println("IndexOutOfBoundsException"); } // end IndexOutOfBoundsException catch(RuntimeException e) { System.out.println("RuntimeException"); }// end RuntimeException */ }// end main }// end class Exercise 6: 1. Type in and execute the code above. 2. One at a time, comment out errors 1 and 2, recompile, and rerun. Explain the resulting outputs from the three executions. 3. Reactivate all three errors, comment out the single catch block shown in this code example, then activate the commented three catch blocks shown to the right. Recompile and rerun. Again, one at a time, comment out error 1 and 2 as before, and rerun. Explain the resulting outputs. 18 Answers Exercise 1: 1. Line 2 of PolymorphicAssignment.java assigns a Class B type object reference variable as the value of a Class A object reference variable. This is okay as ClassB extends ClassA and therefore obj3’s type “is a” type of Class A. 2. Line 3 is attempting to assign a superclass type object variable as the value of a subclass object reference variable. This is like trying to assign a double value to an int. This is also true for line 4, but lines 3 and 4 differ from each other as the value being assigned in line 3 (obj2) never was a ClassB object while the value being assigned in line 4 (obj1) was originally a ClassB object prior to the conversion to superclass type (like type broadening from int to double). This distinction between lines 3 and 4 is crucial to understanding the difference between lines 5 and 6. 3. Line 4 is not allowed because even though obj1 was, in line 2, assigned the value of a ClassB variable, obj1’s internal format was somehow changed after line 2 to the superclass type in a manner analogous to the type broadening from int to double. 4. Line 5 works because obj1’s value was originally a ClassB type (see line 2) but line 6 does not work as obj2 never was a ClassB type to begin with. Exercise 2: The output of PolyParam.java is chosen at random to be either “area = 50” or “area = 25.” Assignment of the variable shp is made at random to be either tri, a variable of class type Triangle, or rect, a variable of class type Rectangle. Both those respective classes have overridden the method area in the abstract superclass to reflect the different areas for the triangle and the rectangle. Both the assignment statements shp = tri; and shp = rect; are syntactically possible due to polymorphism—tri and rect are both subclasses (hence types of ) abstract class Shape. Exercise 3: 1. In line 1 method1() returns a ClassA object reference variable and assigns it to obj1, which is also a Class A object reference variable. 2. In line 2 method2() returns a ClassB object reference variable (which is a subclass type of Class A) and assigns it to a ClassA object reference variable. The type assignment is still compatible. 3. In line 3 method1() returns a ClassA object reference and attempts to assign it to a ClassB type variable. This is analogous to assigning a double value to an int variable. 4. In line 4 method3() returns a ClassA object reference value and attempts to assign it to a ClassB object reference variable. Although lines 3 and 4 look very much alike, the “incompatible types” mismatches are actually different because the value returned by method3() was originally a ClassB object type that was “broadened” to ClassA format. This latter point is important to understanding the difference between lines 5 and 6. 5. Line 5 converts the underlying object reference type (a ClassB object value that was broadened to a ClassA value) back to its original type though the use of a type cast. 6. Line 6 cannot type cast the ClassA object reference value returned by method1() as, unlike the object returned in line 5, the object returned in line 6 never was a ClassB object reference to begin with. Exercise 4: The answer to Exercise 4 is the same as the answer for Exercise 2 except that this time the result is obtained by polymorphic return types instead of polymorphic assignment statements. 19 Exercise 5 import java.awt.*; public class HouseBuild extends java.applet.Applet { public void paint(Graphics g) { BuilderClass [] refArray = new BuilderClass[5]; refArray[0] = new Framer(); refArray[1] = new Windows(); refArray[2] = new Door(); refArray[3] = new Roof(); refArray[4] = new Chimney(); for (BuilderClass e: refArray) e.build(g); public class Framer extends BuilderClass { public void build(Graphics g) { g.drawRect(250, 300, 200, 100) }// end build }// end class Framer ; public class Windows extends BuilderClass { public void build(Graphics g) { g.drawRect(280,340, 20, 20); g.drawRect(400,340, 20, 20); }// end build }// end class Windows }// end paint }// end class abstract class BuilderClass { abstract void build(Graphics g); }// end BuilderClass public class Door extends BuilderClass { public void build(Graphics g) { g.drawRect(330, 360, 40, 40) ; }// end build }// end class Door public class Roof extends BuilderClass { public void build(Graphics g) { g.drawLine(250, 300, 350, 200); g.drawLine(350, 200, 450, 300); }// end build }// end class Roof public class Chimney extends BuilderClass { public void build(Graphics g) { g.drawLine(300, 250, 300, 200); g.drawLine(300,200, 330, 200); g.drawLine(330, 200, 330, 218); }// end build }// end class Framer Exercise 6 1. Type in and execute given code. 2. Errors 1 (ArrayIndexOutOfBoundsException), 2 (IndexOutOfBoundsException) and 3 (RuntimeException) will all be caught by the RuntimeException object. This is analogous to catching all three fish at once with the fine-mesh net. This catch is nonselective and does not allow for recovery code specified to correct specific error situation. 3. Here the errors are caught one at a time by the polymorphic exception catches. This allows for recovery code written to correct specific error situations as they occur. 20