lecture 3: more constructors, scope, wrappers, inheritance, and object oriented design. java for the...
TRANSCRIPT
Lecture 3: More Constructors, Scope, Wrappers, Inheritance, andObject Oriented
Design.
Java for the ImpatientJava for the Impatient
• Extending classes & inheriting fields/methods– Using this and super– Constructor chaining– Polymorphism & casting– Single and multiple inheritance– Visibility modifiers
Object Oriented Design
Inheritance
Natural, hierarchical way of organizing things.
Staff Member
Employee Volunteer
Hourly Salaried
Consultant
Think in terms of “is a” relationships: An Employee is a Staff Member, as is a Volunteer. An Hourly worker is a Employee. A Consultant is a(n) Hourly employee.
(subclass of Hourly)
(subclass of Employee)
(subclass of Staff)
(superclass)
class Animal { protected String strName = “”; protected String strNoise = “”; protected int iNumTimesPerformed = 0; // constructors, accessors & modifiers go here public void identifySelf( ) { System.out.println(“My name is “ + strName); } // of identifySelf public void perform ( ) { doYourThing( ); iNumTimesPerformed++; } // of perform public void doYourThing( ) { ; // ‘no-op’ method } // of doYourThing
} // of Animal
Example
Subclasses(Dog extends Animal
i.e. “A dog is an animal” or“All dogs are animals”)
class Dog extends Animal { public Dog() { strNoise = “Woof”; } // of constructor
public void doYourThing ( ) { identifySelf(); System.out.println(“I am a dog”); System.out.println(strNoise); } // of doYourThing
} // of Dog
Recall: The Animal class had a no-op method for doYourThing()
Animal
Dog Cat Human
Subclasses(Cat extends Animal
i.e. “A cat is an animal” or“All cats are animals”)
class Cat extends Animal { public Cat() { strNoise = “Miaow”; } // of constructor
public void doYourThing ( ) { identifySelf(); System.out.println(“I am a cat”); System.out.println(strNoise); } // of doYourThing
} // of Cat
Animal
Dog Cat Human
Example (Cont’d)Dog pickles = new Dog();pickles.setName(“Pickles”);pickles.doYourThing();// output:// “My name is Pickles”// “I am a dog”// “Woof”
Cat abby = new Cat();abby.setName(“Abby”);abby.doYourThing();// output:// “My name is Abby”// “I am a cat”// “Miaow”
Subclasses(Human extends Animal
i.e. “A human is an animal” or“All humans are animals”)
class Human extends Animal { public Human() { strNoise = “I think therefore I am”; } // of constructor
public void doYourThing ( ) { identifySelf(); System.out.println (“I am a sentient being”); System.out.println(strNoise); } // of doYourThing
} // of Human
Animal
Dog Cat Human
Example (Cont’d)
Human descartes = new Human();descartes.setName(“Rene”);descartes.doYourThing();// output:// “My name is Rene”// “I am a sentient being”// “I think therefore I am”
Inheritance and ScopeVariables (e.g. strNoise):• Java first examines current method, looks for local variable or parameter;• Java then examines current class (e.g. Dog);• Java then examines superclass (e.g. Animal); • Java continues up the class hierarchy until no more superclasses to examine.
Methods (e.g. doYourThing() or identifySelf()):• Java first examines current class;• Java then examines superclass;• Java continues up inheritance hierarchy until no more superclasses to examine.
Specifying ScopeSpecifying Scope
Java allows you to override the scope rules by saying which variable/method you’re referring to:
Keyword super:keyword for specifying method or variable from superclass, e.g., super.doYourThing( )
Keyword this:keyword for specifying method or variable in current object, e.g., this.doYourThing( )
class Dog extends Animal { public Dog() { super.strNoise = “Woof”; } // of constructor
public void doYourThing ( ) { super.identifySelf(); System.out.println(“I am a dog”); System.out.println(strNoise); } // of doYourThing
} // of Dog
Same (in this case) asstrNoise = “Woof”;andthis.strNoise = “Woof”;
Same (in this case) asidentifySelf();orthis.identifySelf();
Using super
class Dog extends Animal { // constructor as before
public void doYourThing() {identifySelf();System.out.println(strNoise);
} // of doYourThing
public void identifySelf() { super.identifySelf();
System.out.println(“I am a dog”); } // of identifySelf
} // of Dog
Animal
Dog Cat Human
I.e.this.identifySelf()(newly defined below)
I.e. theidentifySelf()(defined in Animal)
class Shape {
public final double PI = 3.14159; protected String name; public String getName () { return (this.name); } // getName
public int area () { return (0); } // area
} // Shape
A geometry example
Shape
CircleRectangle
class Rectangle extends Shape { private int length, width; Rectangle () { this(0, 0); } // constructor
Rectangle (int l, int w) { this( l, w, “rectangle”); } // constructor
Rectangle (int l, int w, String n) { length = l; width = l; name = n; } // constructor
public int area () { return (length * width); } // area
public String getName () { if (length == width) return ("square"); else return (super.getName()); } // getName
public String toString () { String s; s = new String ("A " +
getName() + " with length " + length + " and width " + width);
return (s); } } // toString
} // Rectangle
Java’s rule:
• If first line of constructor is not an explicit call to a superclass constructor, Java will implicitly put super( ) as the first line, calling the superclass default constructor.
public Dog() { strNoise = “Woof”; } // of constructor
• An exception to this rule: chained constructor call tothis(params)
will defer super( ) call
• To use superclass constructors with params, call them explicitly, e.g., super(strName)
Constructors and Inheritance
implied call to Animal() here
Inheritance and ScopingInheritance and Scoping
Examples:
super(xxx) // calls a superclass constructorsuper.xxx // accesses superclass’ variablesuper.xxx( ) // calls superclass’ method
this(xxx) // calls a current-class constructorthis.xxx // accesses current class’s variablethis.xxx( ) // calls current class’ method
Note: cannot do super.super<something>
(can achieve this effect via casting, but rarely should; details later...)
Inheritance and Scopingclass StaffMember { String strName;
public StaffMember( ) { System.out.println (“in default StaffMem constr; No Name”); setName(“No Name”); } // of constructor
public StaffMember(String strName) { System.out.println (“in 2nd StaffMem constructior; have a Name”); setName(strName); } // of constructor
public void setName(String strName) { this.strName = strName; } // of setName
}// of StaffMember
class Employee1 extends StaffMember { public Employee1(String strName) { setName(strName); } // of constructor } // of Employee1
What happens???
Inheritance and Scoping
Note: Employee has no local setName() method
class Employee2 extends StaffMember { public Employee2(String strName) { setName(strName); } // of constructor
public void setName(String strName) { super.setName(strName); System.out.println (“Name set”); }} // of Employee2
Class Object
• Java provides a base class, Object
• All classes that do not have an extends clause implicitly inherit directly fromclass java.lang.Object
Examples:
public boolean equals (Object o)
public boolean String toString ()
• When you create your own toString( ) method for a class, you are overriding the toString( ) provided by Object.
Object HierarchyObject Hierarchy
Animal
Dog Cat Human
Object
Employee
Salaried Hourly
class Object methods: String toString() boolean equals(Object obj) and a few others...
Animal
Dog Cat Human
Object
Employee
Salaried Hourly
Or how about...
Primitive types (e.g., int) are not classes
But sometimes, we may have need to make use of primitive types in a context that requires that we manipulate objects, not primitivese.g. many collection classes are collections of Objects
Java provides a set of wrapper classes (a.k.a. type wrappers, a.k.a. envelope classes) to support treating primitives as objects.
It does this by providing a specific class that corresponds to each primitive data type
They are in java.lang, so the names are universally available
Wrapper ClassesWrapper Classes
Wrapper ClassesWrapper Classes
Class corresponds to Primitive
Boolean boolean Character char Byte byte Short short Integer int Long long Float float Double double
Each one:• allows us to manipulate primitives as objects• contains useful conversion methods.
E.g. Integer containsstatic Integer valueOf(String s)Integer.valueOf(“27”)is the object corresponding to int 27
• contains useful utility methods (e.g. for hashing)
Using wrappers to bridge between objects and primitives:
// create and initialize an int int i = 7;
// create an Integer object and convert the int to it Integer intObject = new Integer( i );
// retrieve the int by unwrapping it from the object System.out.println( intObject.intValue );
// convert a string into an Integer object String strS = “27”; Integer intObject intObject = new Integer (Integer.valueOf(strS) ); // then to an int i = intObject.intValue;
Wrapper ClassesWrapper Classes
A classmethod
InstanceInstance vs. vs. ClassClass Methods Methods
• Use instance methods whenever each object should have its own behavior. e.g.,
pickles vs. descartes vs. abby doYourThing( ).
• Use a class method whenever the class itself should maintain a single behavior for all instances of the class. e.g.,
converting from one type to another.
• Class methods cannot be used to access instance variables
•So you can sayInteger.valueOf(strValue);but notintObject.valueOf(strValue);
Polymorphism• Several subclasses may have different
methods for accomplishing the same/similar behavior– You don’t care about the details when you call
a method, provided you know that the object can perform a method with that signature
– E.g. you have a simulated ecology
with animals in it (SimZoo) and
want to make the animals move• But animals move in diffrerent ways
PolymorphismPolymorphismclass Animal { public void move ( ) { System.out.println(“I am an animal and am moving.”); } // of move} // of Animal
class Fish extends Animal { public void move( ) { System.out.println(“Glug glug gurgle gurgle”); } // of move} // of Fish
class Bird extends Animal { public void move( ) { System.out.println(“Tweet tweet flap flap”); } // of move} // of Bird
Polymorphism (cont’d)
class Dog extends Animal{ public void move ( ) { System.out.println(“Sniff sniff woof woof”); } // of move
public void bark ( ) { System.out.println(“Arf Arf”); } // of bark
} // of Dog
class Driver { public static void main (String[ ] argv) { Animal[ ] animalArray = new Animal[3]; int iIndex; animalArray[0] = new Bird( ); animalArray[1] = new Dog( ); animalArray[2] = new Fish( ); for (iIndex=0; iIndex < animalArray.length; iIndex++) { animalArray[iIndex].move( ); } // of for
} // of main} // of Driver
Output: Tweet tweet flap flap Sniff sniff woof woof Glug glug gurgle gurgle
Polymorphism
All animals can move, so any member of the array can move
PolymorphismPolymorphism• Polymorphism means “taking many forms” ... an object of a given class can adapt take the formof any of its subclasses.
• Polymorphism means that the correct move( ) method will always be called.
• A subclass can be substituted for its superclass, e.g., a bird for an animal. “A bird is a animal.” Yes.
• The reverse is not true: can’t substitute superclass for a subclass, e.g., CANNOT substitute an animal for a bird. “An animal is a bird?” No.
• A single interface for multiple behaviors: Only one interface for the method call. Multiple behaviors based on the subclass.
class Driver2 { public static void main(String[ ] argv) { Animal[ ] = animalArray[3]; Dog d; int iIndex; animalArray[0] = new Bird( ); animalArray[1] = new Dog( ); animalArray[2] = new Fish( ); for (1=0; 1 < animalArray.length; iIndex++) if (animalArray[iIndex] instanceof Dog) { d = (Dog) animalArray[iIndex]; d.bark( ); } // if } // main} // Driver2
Polymorphism
We cast before calling bark() because only dogs can bark. So some array members cannot execute the method
Casting:• used here to give an object of a superclass the form of the appropriate subclass, e.g.,
if (animalArray[iIndex] instanceof Dog) { animalArray[iIndex].bark(); }
would produce an error because objects of class Animal have no method called bark. So, we first cast what instanceof tells us is a Dog as a Dog.
if (animalArray[iIndex] instanceof Dog) { d = (Dog) animalArray[iIndex] d.bark( ); }
Polymorphism
Keyword instanceof: Used to interogate an object to see if it is an instance of the specified class, e.g.
“Is this particular animal of class Dog?”
Casting … Why?
Question: If Java can determine that a given Animal is or is not a Dog (via instanceof), then:
• Why the need to cast it to a Dog object before Java can recognize that it can bark?
• Why can’t Java do it for us?
Sourcecode
CompileBytecode
JVMInterpreter
Programruns
errors errors
Answer: difference between compile-time and run-time type checking.
Casting… Why?
Casting (con’td)
Compile-time Errors:
• Those that are discernable without the program executing.
• Question of language legality: “Is this a legal statement?” e.g.,
iIndex = chVariable;
Statement is not legal.
Run-time Errors:
• Those that are discernable only when the program is running with actual data values.
• Question of execution legality: “Is it legal for this variable to have the actual value assigned to it?”, e.g.,
animalArray[<badIndex>] = someAnimal
Statement is legal, but particular index value isn’t.
if (animalArray[iIndex] instanceof Dog) { animalArray[iIndex].bark; }
• 1st line is legal.• 2nd line isn’t (unless array has Dog).• We can see that 1st line guarantees 2nd is legal.
• Compiler cannot see inter-statement dependencies… unless compiler runs whole program with all possible data sets!
• Runtime system could tell easily. We want most checking at compile-time for performance and correctness.
Casting… Why?
Casting (Cont’d)if (animalArray[iIndex] instanceof Dog) { d = (Dog) animalArray[iIndex]; d.bark( );
}
• Here, legality of each line of code can be evaluated at compile time.
• Legality of each line discernable without worrying about inter-statement dependencies, i.e., each line can stand by itself.
• Can be sure that code is legal (not sometimes-legal).
A Good Use for Casting:
Resolving polymorphic ambiguities for the compiler.
Multiple Inheritance
Some languages allow multiple inheritance:
Animal Pet (two superclasses)
Dog (a subclass of two)
• Multiple inheritance leads to many potentially confusing naming problems (e.g. Pet.doYourThing() vs. Animal.doYourThing())
•Growing consensus: the benefits of multiple inheritance aren’t worth the problems.
• Java has single inheritance onlyonly.•Java doesn’t allow multiple inheritance•Well, there is a restricted kind that avoids most of the problems. It involves using interfaces, which we’ll cover later)
Visibility and AccessVisibility and Access
Visibility Modifier:
Access by: public protected private
Every class Yes No No
A subclass Yes Yes No
An instance of the class Yes Yes Yes
Always specify a visibility modifier.
Guidelines: Only make public methods that are in the class’s “contract” Make all fields private Make all other “private” methods protected Don’t leave off the modifier unless you know about packages
Can an object use a field or call a method?
Animals (again!)class Animal {
protected String strName = “”; protected String strNoise = “”; // constructors, accessors & modifiers go here
public void identifySelf( ) { System.out.println(“My name is “ + strName); } // of identifySelf
public void doYourThing( ) {; // ‘no-op’ method
} // of doYourThing } // of Animal
So, any object can ask ananimal to identify itself,or do its thing.
Exception Handling:
When Bad ThingsHappen to Good Code
Error HandlingSo far:• have done very little error handling• have assumed that things will work as intended
Rules of Thumb:• programmers must test and debug to find and correct compile-time errors
• compiler doesn’t solve the problem of run-time errors (e.g., incorrect values or state)
• programmer must insure that run-time errors result in “graceful” program behavior
• “graceful” means “the program doesn’t just produce wrong effects or blow up!
Exceptions--Why?Exceptions--Why?
• In traditional procedural programming languages, there were various ways of handling run-time errors.
• One way is to return some kind of value indicating whether the procedure succeeded or not.
For example:
public boolean someMethod( ) { // if this method suceeded, return true // otherwise return false }
Note the blurring of the logical distinction betweenprocedures and functions… a bad engineering habit!
Exceptions--Traditional MethodsExceptions--Traditional Methods
• Traditional means of error handling can quickly become ugly, complex and unmanageable.
For example:
If we intend to do the sequence of calling three procedures followed by one or more instructions, e.g.,
someMethod( );
someOtherMethod( );
someThirdMethod( ); /* do some intended actions*/ we can find ourselves with code that looks like . . .
Exceptions--Traditional MethodsExceptions--Traditional Methods
if (someMethod( ) == true) { if (someOtherMethod( ) == true) {
if (someThirdMethod( ) == true) { // have not encountered errors; do intended actions } else { // handle some error caused by someThirdMethod( ) } } else { // handle some error caused by someOtherMethod( ) } } else { // handle some error caused by someMethod( ) }
Exceptions--Traditional MethodsExceptions--Traditional Methods
int iErrorValue = 0; public void someMethod( ) { // do someMethod’s stuff here // if there is an error, then set iErrorValue = 1 }
public void someOtherMethod( ) { // do someOtherMethod’s stuff here // if there is an error, then set iErrorValue = 2 }
public void someThirdMethod( ) { // do someThirdMethod’s stuff here // if there is an error, then set iErrorValue = 3 }
• Another way error handling is to have the value of a global variable represent the error.
Exceptions--Global VariablesExceptions--Global Variables
public void doIt() { someMethod(); someOtherMethod(); someLastMethod();
if (iErrorValue == 1) ... if (iErrorValue == 2) ... if (iErrorValue == 3) ... }
But: What if the run-time error stopped us from continuing?
For example: What if someMethod( ) failed in such a way that we cannot go on to someOtherMethod( )?
To cope, we find ourselves with code that’s nearly as messy as the earlier example which featured multiple nested-ifs:
Exceptions--Global VariablesExceptions--Global Variables
public void doit( ) { someMethod( ); if (iErrorValue == 1) { ... } // if else { someOtherMethod( ); if (iErrorValue == 2) { ... } // if else { someThirdMethod( ); if (iErrorValue == 3) { … } // if else { do intended actions } // else } // else }// else
Exceptions--Global VariablesExceptions--Global Variables
Note: with this techniquewe potentially must wrapthe ENTIRE programin a series of if/else clauses,duplicating code in places.
(Do we prefer robustnessor clarity/maintainability?)
• As you can see, it gets convoluted very fast even for such a simple sequence of steps.
• Historically, the way programmers got around this was simply to not do any error handling at all!
• Instead, they generally used one of two approaches:
• Upon detecting an error, simply terminate the program, i.e., “recognizethe error but don’t handle it, just quit”, or else …
• Don’t even attempt to detect the error; i.e., “let the program react in anarbitrary and unpredictable manner”(blow up? bad values? wrong behavior?)
• Both of these violate a basic tenet of structured programming:
“Allow only a single point of exit from any procedure or function.”
Exceptions: Traditional ApproachesExceptions: Traditional Approaches
Both of these traditional approaches boil down to a case of the programmer simply ignoring the real problem, which is:
When a run-time error occurs in a method,
• how can we stop the method without allowing it to do any damage?
• how can we take appropriate actions to handle the error without having the program simply blow up or do something else that’s bad?
It is not acceptable for programs to fail or to do “bad behavior”!
• Safety critical programs
• Customer satisfaction
We require some mechanism to recover from unexpected or abnormal run-time situations.
Exceptions: The Real ProblemExceptions: The Real Problem
Exception Handling: the modern programming concept for dealing with run-time errors
Exception: “a run-time event that may cause a method to fail or to execute incorrectly”
Purpose of Exception Handling: “to allow graceful handling of and recovery from run-time errors”
Common examples in Java:
• NullPointerException
• ArithmeticException
• ArrayIndexOutOfBoundsException
• Java API contains over two dozen exceptions provided by Java.
Exceptions and Exception HandlingExceptions and Exception Handling
Exception Terminology:
• When an exceptional condition occurs, an exception is “thrown” (i.e., the exception has been recognized).
• The flow of control is tranferred to the point where the exception is “caught” (I.e., where the exception-handling code responds to it).
In the jargon of some other programming languages, when an exception is recognized, an exception is:
• “raised” (in place of “thrown”), then
• “handled” (in place of “caught”).
Same ideas, different jargon.
Exceptions: TerminologyExceptions: Terminology
One is usually more interested in
the type of the exceptionthan in manipulating it as an object,
so “e” is just an object often thrown away.
The general structure of Java’s exception handling:
try { // here goes the code that attempts to perform the // intended action, but that could throw an exception ... } // try catch (ExceptionType1 e) { // here goes the code to handle exception type 1 ... } // catch Type1 catch (ExceptionType2 e) { // here goes the code to handle exception type 2 ... } // catch Type2
Exceptions: General FormatExceptions: General Format
int iDivisor; int iDividend; float fResult;
try { // get input for divisor and dividend ... fResult = (float) iDividend / iDivisor; System.out.println(fResult); } catch (ArithmeticException e) { System.out.println("The divisor was 0"); ... }
Exceptions: Simple ExampleExceptions: Simple ExampleAn example showing the structure of Java’s exception handling:
See, we don’tCare about the exception,
Just about its type being arithmetic error.
How Java handles Exceptions:
If an exception is thrown in a method, then you can do one of two things in response:
1. Catch the exception right then and there, and handle the exception yourself.
You would do this if you have enough information to know how to handle the error.
2. Declare in the method header that whoever called the method has to handle the error.
You would do this if you don't have enough information to know how to handle the error.
Exceptions: How in JavaExceptions: How in Java
Given a full Queue, where a client tries to enqueue an item . . .
• What should you have the Queue do? • How do you know what it should do? • Should it print out a message? • Should it try to increase the Queue’s size? • Should it not enqueue the item? • Should it dequeue an item to make room for the new item?
What should you do? To put it simply, you don't know.
Solution:
1. Your design/documentation for enqueue should state a precondition:
/** PRE/POST: The queue is not full */
2. The code will let the exception propagate.
Exceptions: ExampleExceptions: Example
When an exception is thrown, it must be caught immediately or declared to be allowed to propagate.
An example of code that will not compile:
class Queue { ... public boolean isEmpty( ) { ... } // isEmpty
public void dequeue(Object o) { if (isEmpty( ) == true) throw new QueueEmptyException( ); ... } // dequeue ...} // class Queue
Exceptions: PropagationExceptions: Propagation
Dequeue is not allowed to do this
• Results in an error saying that dequeue must catch or declare QueueEmptyException.
• To resolve this, modify dequeue so that the exception is declared to be thrown:
public void dequeue(Object o) throws QueueEmptyException { if (isEmpty( ) == true) throw new QueueEmptyException( ); ... }
• The method header above declares that this method can throw a
QueueEmptyException
and that the method calling dequeue( ) must plan on catching the QueueEmptyException.
Exceptions: PropagationExceptions: Propagation
Suppose you want to use this Queue class to simulate a line of Customers, and you do:
class Customer { ...} // Customer
class Cashier { Queue customerQueue = new Queue( ); ...
public void getNextCustomer( ) { Customer cust; ... cust = (Customer) customerQueue.dequeue( ); ... } // getNextCustomer ...} // Cashier
Exceptions: ExampleExceptions: Example
This will result in a compile-time error because method getNextCustomer must:
• catch exception QueueEmptyException, or
• declare that QueueEmptyException propagates upwards
Thus, we can repair getNextCustomer in one of two ways:
• Option 1: have it catch exception QueueEmptyException
• Option 2: have it declare that this method allows QueueEmptyException to propagate
Exceptions: Example (cont’d)Exceptions: Example (cont’d)
public void getNextCustomer( ) { Customer cust;
try { ... cust = (Customer) customerQueue.dequeue( ); ... } // try
catch (QueueEmptyException e) { // handle the QueueEmptyException here ... } // catch
} // getNextCustomer
An Exception: Catching ItAn Exception: Catching ItOption 1
public void getNextCustomer( ) throws QueueEmptyException { Customer cust;
... cust = (Customer) customerQueue.dequeue( ); ... } // getNextCustomer
This option dictates that whoever calls method getNextCustomer( ) has the responsibility of handling the exception.
An Exception:An Exception:
Declaring that it will propagateDeclaring that it will propagate
Option 2
public class Test { public static void A( ) { int array[ ] = new int[5]; try { System.out.println( "In A's try." ); array[ 5 ] = 1; } // try catch( ArrayIndexOutOfBoundsException error ) { System.out.println( "In A's catch." ); } // catch } // A() public static void main( String argv[ ] ) { try { A( ); } // try catch( Exception e ) { System.out.println( "In main's catch." ); } // catch System.out.println( "After try in main." ); } // class Test
Exceptions: Another ExampleExceptions: Another Example
Output:
In A's try. In A's catch.
After try in main.
What happens here?
• To recognize the exceptional state, use standard if-else logic.• To respond to them, you can create your own Exceptions.
Exceptions are objectsSo you really define your own classes of exception
• All Exceptions you create are extensions of java.lang.Exception
For example:
class QueueEmptyException extends Exception { public QueueEmptyException( ) { }
public QueueEmptyException(String strMessage) { super(strMessage); } }
Exceptions: Creating Your OwnExceptions: Creating Your Own
• You can take advantage of inheritance when dealing with exceptions.
• Suppose we had an inheritance hierarchy of exceptions like this:
• You can have more than one catch block for a try block.
• A catch block will only catch that exception or a subclass of that exception.
Exception
IOException QueueFullException
EOFExceptionFileNotFoundException
Inheritance and ExceptionsInheritance and Exceptions
try { ... } // try
catch (QueueFullException e) { ... } // catch QueueFull Exception
catch (FileNotFoundException e) { ... } // catch FileNotFound Exception
catch (IOException e) { ... } // catch IO Exception
Inheritance and ExceptionsInheritance and Exceptions
This sequence of catches works:
Exception
IOException QueueFullException
EOFExceptionFileNotFoundException
try { ... } // try
catch (IOException e) { ... } // catch IOException
catch (FileNotFoundException e) { ... // this code can never be reached because // FileNotFound is subclass of IOException } // catch FileNotFound Exception
This sequence of catches doesn’t work:
Inheritance and ExceptionsInheritance and Exceptions
Exception
IOException QueueFullException
EOFExceptionFileNotFoundException
Something you can do with exceptions: try { ... } // try
catch (QueueFullException e) { ... } // catch QueueFullException
catch (IOException e) { ... } // catch IOException
catch (Exception e) { ... // this will catch any other kind of exception // since all exceptions extend Exception } // catch base Exception
Inheritance and ExceptionsInheritance and Exceptions
“catch-all” handler
• In Java, there are actually two types of Exceptions: • those subclassed from Exception• those subclassed from RuntimeException
A technoid annoyance: Both kinds deal with run-time errors
• Those subclassed from RuntimeException do not have to be
explicitly caught or declared in the method header.
• This is good. It prevents code from being cluttered with exception handling:
// possible ArrayIndexOutOfBounds customerArray[10] = new Customer( );
// possible ArithmeticException x = y / z; • These may still be caught and/or propagated
upwards like normal exceptions.
Exception & RuntimeExceptionException & RuntimeException
• There is a problem with the internal state of the program
• A contract is violated
• A security risk arises (SecurityException)
• There is an error with an object or the data it manipulates. • Coping with bad parameters
• Dealing with Truly Exceptional conditions (memory, stack).
Exceptions: When to UseExceptions: When to Use
When to use exceptions: Internal StateWhen to use exceptions: Internal StateFor example:
public int getAge(int iSocialSecurityNumber) throws RecordKeepingException
{int index = getHashKey(iSocialSecurityNumber);int iAge = myArray[index].getAge(iSocialSecurityNumber);
/* Similar to util.ASSERT( ) statement. */if (iAge <= 0)
throw new RecordKeepingException (“Exception: Age for “ + iSocialSecurityNumber +
“ not in range: “ + iAge);else
return iAge;}
public TreeNode getNodeRecursively (int index, TreeNode currentNode) throws MissingNodeException { if (currentNode == null) { throw new MissingNodeException (“Exception: No node with ” + index + “ found”); } // if else if (currentNode.getNumber() == index) { return currentNode; } // else else if (currentNode.getNumber() > index) { return getNodeRecursively (index, currentNode.getLeftChild()); } // if else { return getNodeRecursively (index, currentNode.getRightChild()); } // else
}// getNodeRecursively
When to Use Exceptions: Contract ViolatedWhen to Use Exceptions: Contract Violated
public void initializeTreeNode(int iNumberNodes) { if (myTree == null) { if (DEBUG)
System.out.println (“Null tree found!”); throw new NullPointerException (“Null tree found”);
/* NOTE: Runtime exception; no need to declare propagation */ } else { for (int i=0; i < iNumberNodes; i++) { { TreeNode newNode = new TreeNode( i ); tree.insertNode(newNode); } }
} // initializeTreeNode
When to Use Exceptions: Error with objectsWhen to Use Exceptions: Error with objects
public Integer convertNumber (String strToConvert) {
for (int i =0; I < strToConvert.length(); i++) {
char chTemp = strToConvert.charAt(i);
if (!Character.isDigit(chTemp)) {
if (DEBUG) System.out.println (“Bad input String: “ + strToConvert);
throw new NumberFormatException(“Exception: “ + strToConvert + “ is not numeric”);
} } } // convertNumber
When to Use Exceptions: Bad ParametersWhen to Use Exceptions: Bad Parameters
public class Chat { public static void doStuff() { Listener.count(”Yes it is "); try { Chit.doStuff(); } //try catch (StackOverflowError e){ System.exit(0); } // catch }} // Chat
When to Use Exceptions:When to Use Exceptions:
Truly Exceptional CircumstancesTruly Exceptional Circumstances
public class Chit { public static void doStuff() { Listener.count(”No it isn’t "); try { Chat.doStuff(); } // try catch (StackOverflowError e) { System.exit(0); } // catch }} // Chit
public class Listener { static int iCount; public static void count(String strSpeaker){ iCount++; System.out.println (strSpeaker + " is number " + iCount); }} // Listener
When to Use Exceptions:When to Use Exceptions:
Truly Exceptional CircumstancesTruly Exceptional Circumstances
public TreeNode getNode(int index){ TreeNode tempNode; try { tempNode =
myTree.getNodeRecursively(new TreeNode(index)); } // try
catch(StackOverflowError e) {System.exit(1);
} // catch
return tempNode;} // getNode Or less obviously
• Print an error message
• Log the exception
• Retry the method(maybe with default parameters)
• Restore the system to some previouslyknown "good" state.
• Set the system to some "safe" state.
•Let exception propagate to whoever called the method in which the exception arose
• Catch it and ignore it
“Catch it and ignore it” is generally bad: If the error was serious enough to throw an exception, it should be dealt
with, not ignored.
When Catching Exceptions you can . . .When Catching Exceptions you can . . .
OOA/OOD/OOP“Who”/what knowsenough to handlethe exception?
local?
high-level?
Exception Defined:
Object that defines an unusual or erroneous situation
What to do?
Not handle the exception (program halts. Ugh!) Handle the exception where it occurs (try...catch...) Handle the exception at another point in the program
(exception propagation)
Throwing an exception:
Part of the contract of a method (throws) Responsibility of the client to handle (try…catch…)
Exceptions: ReviewExceptions: Review
A Few Words on Vectors
VectorsVectors• An array (list) that dynamically resizes itself to whatever size is needed.
• A partial API for Vector:class Vector { public void addElement( Object obj ) public boolean contains(Object elem) public Object elementAt(int index) public Object firstElement() public int indexOf(Object elem) public void insertElementAt(Object obj, int index) public boolean isEmpty() public Object lastElement() public int lastIndexOf(Object elem) public void removeAllElements() public boolean removeElement(Object obj) public void removeElementAt(int index) public void setElementAt(Object obj, int index) public int size()} // of Vector
VectorsVectors• Vector is a class.• Must have an object of type Vector instantiated via new( ) to get an instance of Vector.• All rules of good OO programming apply.• Thus, access by requesting services via methods, not via direct access (such an array).
size( ) returns current number of elements.
elementAt(int index) returns reference to element at specified index.
insertElementAt( Object obj, int index ) insertion into linked list (but slower);
addElement (Object obj) adds to end.
CommonExamples
VectorsVectorsVectors:
• Can be populated only with objects;
• Cannot be populated with primitives;
• Can be populated with objects that contain primitives;
• If you need to populate them with primitives, use type wrapper classes e.g., Integer for int, etc.
• Will allow you to populate them with any type of object . . .
• Thus, good programming requires that the programmer enforce typing within a Vector, because Java doesn’t.
VectorsVectorsVectors and casting:
• Vectors are a subclass of class Object.• Thus, vectors can handle any class of object (i.e., no type checking)
• Thus, must cast any object obtained from a Vector before invoking any methods not defined in class Object.
Vector v = new Vector ( );
Student s = new Student( “Joe” );Student otherStudent;
v.addElement( s );
otherStudent = (Student) v.elementAt(0);
// Assume we have vector v and int k.// Assume int k is an index within the// range [0..(v.size( ) - 1)].
Object o;o = v.elementAt(k); // no cast needed, already Object
if (o instanceof Student){ // do stuff for Student}
if (o instanceof Cafeteria){ // do stuff for Cafeteria}
VectorsVectorsVectors and instanceof:
Design issue:This example is legal Java, but is it good programming?
Vectors and ArraysVectors and Arrays
Arrays: statically sizedVectors: dynamically sized
Arrays: can directly access, e.g., myArray[6]
but shouldn’t (except maybe within the class in which they’re declared IF efficiency concerns; or for testing purposes.)
Vectors: must use methods to access.
Vector services provide a good model forthe Array services you should implement.
Vectors versus Linked ListsVectors versus Linked ListsCan use Vectors to simulate a Linked List:
• Don’t want direct access to data, so . . .
• Provide methods for getPrevious( ), getNext( ), etc. that do the std. Linked List things.
• While the list is implemented as a Vector, the client uses it as if it’s a Linked List.
BUT . . .
There are performance implications (that may or may not matter for a given instance).
What is the cost of:
• insertion?
• deletion?
Vectors versus Linked ListsVectors versus Linked Lists
For ordered Linked Lists:• cost of traversal to locate target: O(N)• cost of insert/delete: O(1)• total cost: O(N)
For ordered Vector:• cost of traversal to locate target: O(N) (if accessible via direct access, then O(1) )• insertion or deletion of element implies (average case), moving O(N) elements• total cost: O(N)
Thus, at first glance, equivalent…
But what does Big Oh hide here?• Linked Lists: search thru N/2, plus insert/delete• Vectors: search thru N/2, plus moving N/2
Thus, Vectors imply twice
the work
Vector CapacityVector Capacity
• Capacity is dynamic
• Capacity can grow and shrink to fit needs
• Capacity grows upon demand
• Capactiy shrinks when you tell it to do so via method trimToSize( )
• Using trimToSize( ) implies performance costs upon subsequent insertion.
• When extra capacity is needed, then it grows by how much?
• Depends on which of three constructors is used . . .
Vector CapacityVector CapacityThree Vector constructors:• public Vector (int initialCapacity, int capacityIncrements);• public Vector (int initialCapacity);• public Vector( );
First constructor (2 parameters):• begins with initialCapacity• if/when it needs to grow, it grows by size capacityIncrements.
Second constructor( 1 parameter):• begins with initialCapacity• if/when needs to grow, it grows by doubling current size.
Third construtor (no parameters):• begins with capacity of 10• if/when needs to grow, it grows by doubling current size.