The clone()
method is used to create and return a copy of an object. This method arguable should be avoided as it is problematic and a copy constructor or some other approach for copying should be used in favour of clone()
.
For the method to be used all classes calling the method must implement the Cloneable
interface.
The Cloneable
interface itself is just a tag interface used to change the behaviour of the native
clone()
method which checks if the calling objects class implements Cloneable
. If the caller does not implement this interface a CloneNotSupportedException
will be thrown.
The Object
class itself does not implement this interface so a CloneNotSupportedException
will be thrown if the calling object is of class Object
.
For a clone to be correct it should be independent of the object it is being cloned from, therefore it may be necessary to modify the object before it gets returned. This means to essentially create a "deep copy" by also copying any of the mutable objects that make up the internal structure of the object being cloned. If this is not implemented correctly the cloned object will not be independent and have the same references to the mutable objects as the object that it was cloned from. This would result in inconsistent behaviour as any changes to those in one would affect the other.
class Foo implements Cloneable {
int w;
String x;
float[] y;
Date z;
public Foo clone() {
try {
Foo result = new Foo();
// copy primitives by value
result.w = this.w;
// immutable objects like String can be copied by reference
result.x = this.x;
// The fields y and z refer to a mutable objects; clone them recursively.
if (this.y != null) {
result.y = this.y.clone();
}
if (this.z != null) {
result.z = this.z.clone();
}
// Done, return the new object
return result;
} catch (CloneNotSupportedException e) {
// in case any of the cloned mutable fields do not implement Cloneable
throw new AssertionError(e);
}
}
}