Upcasting is allowed in Java, however downcasting gives a compile error.
The compile error can be removed by adding a cast but would anyway break at the runtime.
In this case why Java allows downcasting if it cannot be executed at the runtime?
Is there any practical use for this concept?
public class demo {
public static void main(String a[]) {
B b = (B) new A(); // compiles with the cast,
// but runtime exception - java.lang.ClassCastException
}
}
class A {
public void draw() {
System.out.println("1");
}
public void draw1() {
System.out.println("2");
}
}
class B extends A {
public void draw() {
System.out.println("3");
}
public void draw2() {
System.out.println("4");
}
}
I believe this applies to all statically typed languages:
String s = "some string";
Object o = s; // ok
String x = o; // gives compile-time error, o is not neccessarily a string
String x = (String)o; // ok compile-time, but might give a runtime exception if o is not infact a String
The typecast effectively says: assume this is a reference to the cast class and use it as such. Now, lets say o is really an Integer, assuming this is a String makes no sense and will give unexpected results, thus there needs to be a runtime check and an exception to notify the runtime environment that something is wrong.
In practical use, you can write code working on a more general class, but cast it to a subclass if you know what subclass it is and need to treat it as such. A typical example is overriding Object.equals(). Assume we have a class for Car:
@Override
boolean equals(Object o) {
if(!(o instanceof Car)) return false;
Car other = (Car)o;
// compare this to other and return
}
In this case why Java allows downcasting if it cannot be executed at the runtime?
I believe this is because there is no way for the compiler to know at compile-time if the cast will succeed or not. For your example, it's simple to see that the cast will fail, but there are other times where it is not so clear.
For instance, imagine that types B, C, and D all extend type A, and then a method public A getSomeA() returns an instance of either B, C or D depending on a randomly generated number. The compiler cannot know which exact run-time type will be returned by this method, so if you later cast the results to B, there is no way to know if the cast will succeed (or fail). Therefore the compiler has to assume casts will succeed.
@ Original Poster - see inline comments.
public class demo
{
public static void main(String a[])
{
B b = (B) new A(); // compiles with the cast, but runtime exception - java.lang.ClassCastException
//- A subclass variable cannot hold a reference to a superclass variable. so, the above statement will not work.
//For downcast, what you need is a superclass ref containing a subclass object.
A superClassRef = new B();//just for the sake of illustration
B subClassRef = (B)superClassRef; // Valid downcast.
}
}
class A
{
public void draw()
{
System.out.println("1");
}
public void draw1()
{
System.out.println("2");
}
}
class B extends A
{
public void draw()
{
System.out.println("3");
}
public void draw2()
{
System.out.println("4");
}
}
Using your example, you could do:
public void doit(A a) {
if(a instanceof B) {
// needs to cast to B to access draw2 which isn't present in A
// note that this is probably not a good OO-design, but that would
// be out-of-scope for this discussion :)
((B)a).draw2();
}
a.draw();
}
I believe this applies to all statically typed languages:
String s = "some string";
Object o = s; // ok
String x = o; // gives compile-time error, o is not neccessarily a string
String x = (String)o; // ok compile-time, but might give a runtime exception if o is not infact a String
The typecast effectively says: assume this is a reference to the cast class and use it as such. Now, lets say o is really an Integer, assuming this is a String makes no sense and will give unexpected results, thus there needs to be a runtime check and an exception to notify the runtime environment that something is wrong.
In practical use, you can write code working on a more general class, but cast it to a subclass if you know what subclass it is and need to treat it as such. A typical example is overriding Object.equals(). Assume we have a class for Car:
@Override
boolean equals(Object o) {
if(!(o instanceof Car)) return false;
Car other = (Car)o;
// compare this to other and return
}
Using your example, you could do:
public void doit(A a) {
if(a instanceof B) {
// needs to cast to B to access draw2 which isn't present in A
// note that this is probably not a good OO-design, but that would
// be out-of-scope for this discussion :)
((B)a).draw2();
}
a.draw();
}
I believe this applies to all statically typed languages:
String s = "some string";
Object o = s; // ok
String x = o; // gives compile-time error, o is not neccessarily a string
String x = (String)o; // ok compile-time, but might give a runtime exception if o is not infact a String
The typecast effectively says: assume this is a reference to the cast class and use it as such. Now, lets say o is really an Integer, assuming this is a String makes no sense and will give unexpected results, thus there needs to be a runtime check and an exception to notify the runtime environment that something is wrong.
In practical use, you can write code working on a more general class, but cast it to a subclass if you know what subclass it is and need to treat it as such. A typical example is overriding Object.equals(). Assume we have a class for Car:
@Override
boolean equals(Object o) {
if(!(o instanceof Car)) return false;
Car other = (Car)o;
// compare this to other and return
}
We can all see that the code you provided won't work at run time. That's because we know that the expression new A() can never be an object of type B.
But that's not how the compiler sees it. By the time the compiler is checking whether the cast is allowed, it just sees this:
variable_of_type_B = (B)expression_of_type_A;
And as others have demonstrated, that sort of cast is perfectly legal. The expression on the right could very well evaluate to an object of type B. The compiler sees that A and B have a subtype relation, so with the "expression" view of the code, the cast might work.
The compiler does not consider the special case when it knows exactly what object type expression_of_type_A will really have. It just sees the static type as A and considers the dynamic type could be A or any descendant of A, including B.
Downcast works in the case when we are dealing with an upcasted object. Upcasting:
int intValue = 10;
Object objValue = (Object) intvalue;
So now this objValue variable can always be downcasted to int because the object which was cast is an Integer,
int oldIntValue = (Integer) objValue;
// can be done
but because objValue is an Object it cannot be cast to String because int cannot be cast to String.
Consider the below example
public class ClastingDemo {
/**
* @param args
*/
public static void main(String[] args) {
AOne obj = new Bone();
((Bone) obj).method2();
}
}
class AOne {
public void method1() {
System.out.println("this is superclass");
}
}
class Bone extends AOne {
public void method2() {
System.out.println("this is subclass");
}
}
here we create the object of subclass Bone and assigned it to super class AOne reference and now superclass reference does not know about the method method2 in the subclass i.e Bone during compile time.therefore we need to downcast this reference of superclass to subclass reference so as the resultant reference can know about the presence of methods in the subclass i.e Bone
Using your example, you could do:
public void doit(A a) {
if(a instanceof B) {
// needs to cast to B to access draw2 which isn't present in A
// note that this is probably not a good OO-design, but that would
// be out-of-scope for this discussion :)
((B)a).draw2();
}
a.draw();
}
Consider the below example
public class ClastingDemo {
/**
* @param args
*/
public static void main(String[] args) {
AOne obj = new Bone();
((Bone) obj).method2();
}
}
class AOne {
public void method1() {
System.out.println("this is superclass");
}
}
class Bone extends AOne {
public void method2() {
System.out.println("this is subclass");
}
}
here we create the object of subclass Bone and assigned it to super class AOne reference and now superclass reference does not know about the method method2 in the subclass i.e Bone during compile time.therefore we need to downcast this reference of superclass to subclass reference so as the resultant reference can know about the presence of methods in the subclass i.e Bone
Downcasting transformation of objects is not possible. Only
DownCasting1 _downCasting1 = (DownCasting1)((DownCasting2)downCasting1);
is posible
class DownCasting0 {
public int qwe() {
System.out.println("DownCasting0");
return -0;
}
}
class DownCasting1 extends DownCasting0 {
public int qwe1() {
System.out.println("DownCasting1");
return -1;
}
}
class DownCasting2 extends DownCasting1 {
public int qwe2() {
System.out.println("DownCasting2");
return -2;
}
}
public class DownCasting {
public static void main(String[] args) {
try {
DownCasting0 downCasting0 = new DownCasting0();
DownCasting1 downCasting1 = new DownCasting1();
DownCasting2 downCasting2 = new DownCasting2();
DownCasting0 a1 = (DownCasting0) downCasting2;
a1.qwe(); //good
System.out.println(downCasting0 instanceof DownCasting2); //false
System.out.println(downCasting1 instanceof DownCasting2); //false
System.out.println(downCasting0 instanceof DownCasting1); //false
DownCasting2 _downCasting1= (DownCasting2)downCasting1; //good
DownCasting1 __downCasting1 = (DownCasting1)_downCasting1; //good
DownCasting2 a3 = (DownCasting2) downCasting0; // java.lang.ClassCastException
if(downCasting0 instanceof DownCasting2){ //false
DownCasting2 a2 = (DownCasting2) downCasting0;
a2.qwe(); //error
}
byte b1 = 127;
short b2 =32_767;
int b3 = 2_147_483_647;
// long _b4 = 9_223_372_036_854_775_807; //int large number max 2_147_483_647
long b4 = 9_223_372_036_854_775_807L;
// float _b5 = 3.4e+038; //double default
float b5 = 3.4e+038F; //Sufficient for storing 6 to 7 decimal digits
double b6 = 1.7e+038;
double b7 = 1.7e+038D; //Sufficient for storing 15 decimal digits
long c1 = b3;
int c2 = (int)b4;
//int 4 bytes Stores whole numbers from -2_147_483_648 to 2_147_483_647
//float 4 bytes Stores fractional numbers from 3.4e-038 to 3.4e+038. Sufficient for storing 6 to 7 decimal digits
float c3 = b3; //logic error
double c4 = b4; //logic error
} catch (Throwable e) {
e.printStackTrace();
}
}
}
In this case why Java allows downcasting if it cannot be executed at the runtime?
I believe this is because there is no way for the compiler to know at compile-time if the cast will succeed or not. For your example, it's simple to see that the cast will fail, but there are other times where it is not so clear.
For instance, imagine that types B, C, and D all extend type A, and then a method public A getSomeA() returns an instance of either B, C or D depending on a randomly generated number. The compiler cannot know which exact run-time type will be returned by this method, so if you later cast the results to B, there is no way to know if the cast will succeed (or fail). Therefore the compiler has to assume casts will succeed.
I believe this applies to all statically typed languages:
String s = "some string";
Object o = s; // ok
String x = o; // gives compile-time error, o is not neccessarily a string
String x = (String)o; // ok compile-time, but might give a runtime exception if o is not infact a String
The typecast effectively says: assume this is a reference to the cast class and use it as such. Now, lets say o is really an Integer, assuming this is a String makes no sense and will give unexpected results, thus there needs to be a runtime check and an exception to notify the runtime environment that something is wrong.
In practical use, you can write code working on a more general class, but cast it to a subclass if you know what subclass it is and need to treat it as such. A typical example is overriding Object.equals(). Assume we have a class for Car:
@Override
boolean equals(Object o) {
if(!(o instanceof Car)) return false;
Car other = (Car)o;
// compare this to other and return
}
In this case why Java allows downcasting if it cannot be executed at the runtime?
I believe this is because there is no way for the compiler to know at compile-time if the cast will succeed or not. For your example, it's simple to see that the cast will fail, but there are other times where it is not so clear.
For instance, imagine that types B, C, and D all extend type A, and then a method public A getSomeA() returns an instance of either B, C or D depending on a randomly generated number. The compiler cannot know which exact run-time type will be returned by this method, so if you later cast the results to B, there is no way to know if the cast will succeed (or fail). Therefore the compiler has to assume casts will succeed.
@ Original Poster - see inline comments.
public class demo
{
public static void main(String a[])
{
B b = (B) new A(); // compiles with the cast, but runtime exception - java.lang.ClassCastException
//- A subclass variable cannot hold a reference to a superclass variable. so, the above statement will not work.
//For downcast, what you need is a superclass ref containing a subclass object.
A superClassRef = new B();//just for the sake of illustration
B subClassRef = (B)superClassRef; // Valid downcast.
}
}
class A
{
public void draw()
{
System.out.println("1");
}
public void draw1()
{
System.out.println("2");
}
}
class B extends A
{
public void draw()
{
System.out.println("3");
}
public void draw2()
{
System.out.println("4");
}
}
To do downcasting in Java, and avoid run-time exceptions, take a reference of the following code:
if (animal instanceof Dog) {
Dog dogObject = (Dog) animal;
}
Here, Animal is the parent class and Dog is the child class.
instanceof is a keyword that is used for checking if a reference variable is containing a given type of object reference or not.
We can all see that the code you provided won't work at run time. That's because we know that the expression new A() can never be an object of type B.
But that's not how the compiler sees it. By the time the compiler is checking whether the cast is allowed, it just sees this:
variable_of_type_B = (B)expression_of_type_A;
And as others have demonstrated, that sort of cast is perfectly legal. The expression on the right could very well evaluate to an object of type B. The compiler sees that A and B have a subtype relation, so with the "expression" view of the code, the cast might work.
The compiler does not consider the special case when it knows exactly what object type expression_of_type_A will really have. It just sees the static type as A and considers the dynamic type could be A or any descendant of A, including B.
Downcasting is very useful in the following code snippet I use this all the time. Thus proving that downcasting is useful.
private static String printAll(LinkedList c)
{
Object arr[]=c.toArray();
String list_string="";
for(int i=0;i<c.size();i++)
{
String mn=(String)arr[i];
list_string+=(mn);
}
return list_string;
}
I store String in the Linked List. When I retrieve the elements of Linked List, Objects are returned. To access the elements as Strings(or any other Class Objects), downcasting helps me.
Java allows us to compile downcast code trusting us that we are doing the wrong thing. Still if humans make a mistake, it is caught at runtime.
Using your example, you could do:
public void doit(A a) {
if(a instanceof B) {
// needs to cast to B to access draw2 which isn't present in A
// note that this is probably not a good OO-design, but that would
// be out-of-scope for this discussion :)
((B)a).draw2();
}
a.draw();
}
Downcast works in the case when we are dealing with an upcasted object. Upcasting:
int intValue = 10;
Object objValue = (Object) intvalue;
So now this objValue variable can always be downcasted to int because the object which was cast is an Integer,
int oldIntValue = (Integer) objValue;
// can be done
but because objValue is an Object it cannot be cast to String because int cannot be cast to String.
Downcasting is very useful in the following code snippet I use this all the time. Thus proving that downcasting is useful.
private static String printAll(LinkedList c)
{
Object arr[]=c.toArray();
String list_string="";
for(int i=0;i<c.size();i++)
{
String mn=(String)arr[i];
list_string+=(mn);
}
return list_string;
}
I store String in the Linked List. When I retrieve the elements of Linked List, Objects are returned. To access the elements as Strings(or any other Class Objects), downcasting helps me.
Java allows us to compile downcast code trusting us that we are doing the wrong thing. Still if humans make a mistake, it is caught at runtime.
We can all see that the code you provided won't work at run time. That's because we know that the expression new A() can never be an object of type B.
But that's not how the compiler sees it. By the time the compiler is checking whether the cast is allowed, it just sees this:
variable_of_type_B = (B)expression_of_type_A;
And as others have demonstrated, that sort of cast is perfectly legal. The expression on the right could very well evaluate to an object of type B. The compiler sees that A and B have a subtype relation, so with the "expression" view of the code, the cast might work.
The compiler does not consider the special case when it knows exactly what object type expression_of_type_A will really have. It just sees the static type as A and considers the dynamic type could be A or any descendant of A, including B.
In this case why Java allows downcasting if it cannot be executed at the runtime?
I believe this is because there is no way for the compiler to know at compile-time if the cast will succeed or not. For your example, it's simple to see that the cast will fail, but there are other times where it is not so clear.
For instance, imagine that types B, C, and D all extend type A, and then a method public A getSomeA() returns an instance of either B, C or D depending on a randomly generated number. The compiler cannot know which exact run-time type will be returned by this method, so if you later cast the results to B, there is no way to know if the cast will succeed (or fail). Therefore the compiler has to assume casts will succeed.
To do downcasting in Java, and avoid run-time exceptions, take a reference of the following code:
if (animal instanceof Dog) {
Dog dogObject = (Dog) animal;
}
Here, Animal is the parent class and Dog is the child class.
instanceof is a keyword that is used for checking if a reference variable is containing a given type of object reference or not.
Downcasting transformation of objects is not possible. Only
DownCasting1 _downCasting1 = (DownCasting1)((DownCasting2)downCasting1);
is posible
class DownCasting0 {
public int qwe() {
System.out.println("DownCasting0");
return -0;
}
}
class DownCasting1 extends DownCasting0 {
public int qwe1() {
System.out.println("DownCasting1");
return -1;
}
}
class DownCasting2 extends DownCasting1 {
public int qwe2() {
System.out.println("DownCasting2");
return -2;
}
}
public class DownCasting {
public static void main(String[] args) {
try {
DownCasting0 downCasting0 = new DownCasting0();
DownCasting1 downCasting1 = new DownCasting1();
DownCasting2 downCasting2 = new DownCasting2();
DownCasting0 a1 = (DownCasting0) downCasting2;
a1.qwe(); //good
System.out.println(downCasting0 instanceof DownCasting2); //false
System.out.println(downCasting1 instanceof DownCasting2); //false
System.out.println(downCasting0 instanceof DownCasting1); //false
DownCasting2 _downCasting1= (DownCasting2)downCasting1; //good
DownCasting1 __downCasting1 = (DownCasting1)_downCasting1; //good
DownCasting2 a3 = (DownCasting2) downCasting0; // java.lang.ClassCastException
if(downCasting0 instanceof DownCasting2){ //false
DownCasting2 a2 = (DownCasting2) downCasting0;
a2.qwe(); //error
}
byte b1 = 127;
short b2 =32_767;
int b3 = 2_147_483_647;
// long _b4 = 9_223_372_036_854_775_807; //int large number max 2_147_483_647
long b4 = 9_223_372_036_854_775_807L;
// float _b5 = 3.4e+038; //double default
float b5 = 3.4e+038F; //Sufficient for storing 6 to 7 decimal digits
double b6 = 1.7e+038;
double b7 = 1.7e+038D; //Sufficient for storing 15 decimal digits
long c1 = b3;
int c2 = (int)b4;
//int 4 bytes Stores whole numbers from -2_147_483_648 to 2_147_483_647
//float 4 bytes Stores fractional numbers from 3.4e-038 to 3.4e+038. Sufficient for storing 6 to 7 decimal digits
float c3 = b3; //logic error
double c4 = b4; //logic error
} catch (Throwable e) {
e.printStackTrace();
}
}
}
Source: Stackoverflow.com