Suppose you have an object ‘a’ of class ‘A’. Sometimes you may need another new object ‘b’. It also belongs to class ‘A’ and has the same data with object ‘a’. But if you do some modification on b, it has no effect to the value of ‘a’. We call this process which produced new object ‘b’ as clone object ‘a’. The commonest time that you need to clone an object is when it is a parameter or return value of one of your public methods. If it is a parameter that you save somewhere, then you don't want the caller to be able to modify it later. So you save a copy of the object. Likewise, if you are returning an object that is part of your class's internal state, you need to return a copy instead so that callers can't accidentally or deliberately change that internal state.
Conventions of clone
1. x.clone() != x // x.clone() will return a new object
2. x.clone().equals(x) // this is the meaning of ‘copy’
3. x.clone().getClass() == x.getClass()
The object returned by clone method should be independent of the object (which is being cloned).
These are not absolute requirements but are general intends of clone method which is also recommended in Java Documents.
How to write clone method
By convention, the approach of writing clone method is:
1. Implements Cloneable interface
This approach ensures your clone method can directly or indirectly call Object.clone(). Otherwise, calling Object.clone() will throws CloneNotSupportedException. Why we need to call Object.clone() in our clone method? Please see approach 2.2.
2. Override the clone method
2.1 Make the clone method to public method
Please be noted that the clone method type of Object class is:
protected Object clone()
throws CloneNotSupportedException
In order to support other class can use our clone method, we should define it as public method.
2.2 Call super.clone() to produce the new object
By convention, the object returned by clone method should be obtained by calling super.clone (this means it’s better to produce the new object by super.clone() than directly use “new” operator). If a class and all of its superclasses (except Object) obey this convention, it will be the case that x.clone().getClass() == x.getClass().
Key point: why we should use super.clone() to produce the new object instead of directly use “new” operator?
v First of all, if all classes obey this convention, our clone method will directly or indirectly call Object.clone method. This method is a native method, it will be more efficient than directly “new” an object.
v Secondly, Object.clone method can recognize the class type which called the clone method using RTTI mechanism. And it will return the new object which has the correct class type. For example:
class A implements Cloneable
class B extends A implements Cloneable {
public Object clone() throws CloneNotSupportedException{
B b = null;
b = (B) super.clone(); // It seems that super.clone() is
//A.clone(), so it will return an
//object of Class A. This is incorrect.
//If the clone method of class A calls
//super.clone method too, it will
//return a new object belongs to
//class B. Thus, we can cast it to
//class B. This is the benefit of
//Object.clone().
return b;
}
}
Now, let’s consider another case, if we write clone method of class A like this:
class A {
public Object clone() {
A a = null;
a = new A();
// Then do some copy data operation.
return a;
}
}
When B.clone() calls super.clone(),unfortunately we can only get the object whose class is A. And we can’t cast the new object to class B since B is a subclass of A.
That’s why it’s strongly recommended that clone method of all classes obey the convention that obtained the new object by calling super.clone().
2.3 Clone members
There are two cases: If the member supports clone, it’s better to call the clone method of the member to return a copy object of this member. If the member doesn’t support clone, you should create a new object which is the copy of the member. After this approach, it will be ensured that x.clone.equals(x) and x.clone() is independent with x.
Examples
/**
* class B support clone
* @author xzhu2
*
*/
class B implements Cloneable {
private int intMember;
public B(int i) {
intMember = i;
}
public void setIntMember(int i) {
intMember = i;
}
public Object clone()
throws CloneNotSupportedException {
B clonedObject = null;
// Firstly, call super.clone to return new object
clonedObject = (B)super.clone();
// Secondly, clone member here
clonedObject.setIntMember(intMember);
// The end, return new object
return clonedObject;
}
}
/**
* class C doesn't support clone
* @author xzhu2
*
*/
class C {
private int intMember;
public C(int i) {
intMember = i;
}
public void setIntMember(int i) {
intMember = i;
}
public int getIntMember() {
return intMember;
}
}
class A implements Cloneable {
private int intMember = 0;
private String stringMember = "";
private B supportCloneMember = null;
private C notSupportCloneMember = null;
public void setIntMember(int i) {
intMember = i;
}
public void setStringMember(String s) {
stringMember = s;
}
public void setB(B b) {
supportCloneMember = b;
}
public void setC(C c) {
notSupportCloneMember = c;
}
public Object clone()
throws CloneNotSupportedException {
A clonedObject = null;
// Firstly, call super.clone to return new object
clonedObject = (A)super.clone();
// Secondly, clone members here
// For basic type member, directly set it to clonedObject
// Because basic type parameter passes value. Modify
// clonedObject.intMember can not effect the intMember
// of itself.
clonedObject.setIntMember(intMember);
// For immutable member, directly set it to clonedObject.
// Becasue we can not change the value of immutable
// variable once it was setted.
clonedObject.setStringMember(stringMember);
// For member which support clone, we just clone it and
// set the return object to the member of new object.
B clonedB = (B)supportCloneMember.clone();
clonedObject.setB(clonedB);
// For member which do not support clone, we need to create
// new object.
C clonedC = new C(notSupportCloneMember.getIntMember());
clonedObject.setC(clonedC);
// The end, return new object
return clonedObject;
}
}