What is the point of invokeinterface? - jvm

I'm reading this article about how JVM invokes methods, and I think I got most of it. However, I'm still having trouble understanding the need for invokeinterface.
The way I understand it, a class basically has a virtual table of methods and when calling a method with either invokevirtual or invokeinterface this virtual table is consulted.
What is the difference, then, between a method that's defined on an interface and a method defined on a base class? Why the different bytecodes?
The description of the instructions also looks very similar.
The article seems to claim that the method table of an interface can have "different offsets" every time a method is called. What I don't understand is why an interface would have a method table at all, since no object can have the interface as its actual type.
What am I missing?

Each Java class is associated with a virtual method table that contains "links" to the bytecode of each method of a class. That table is inherited from the superclass of a particular class and extended with regard to the new methods of a subclass. E.g.,
class BaseClass {
public void method1() { }
public void method2() { }
public void method3() { }
}
class NextClass extends BaseClass {
public void method2() { } // overridden from BaseClass
public void method4() { }
}
results in the tables
BaseClass
1. BaseClass/method1()
2. BaseClass/method2()
3. BaseClass/method3()
NextClass
1. BaseClass/method1()
2. NextClass/method2()
3. BaseClass/method3()
4. NextClass/method4()
Note, how the virtual method table of NextClass retains the order of entries of the table of BaseClass and just overwrites the "link" of method2() which it overrides.
An implementation of the JVM can thus optimize a call to invokevirtual by remembering that BaseClass/method3() will always be the third entry in the virtual method table of any object this method will ever be invoked on.
With invokeinterface this optimization is not possible. E.g.,
interface MyInterface {
void ifaceMethod();
}
class AnotherClass extends NextClass implements MyInterface {
public void method4() { } // overridden from NextClass
public void ifaceMethod() { }
}
class MyClass implements MyInterface {
public void method5() { }
public void ifaceMethod() { }
}
This class hierarchy results in the virtual method tables
AnotherClass
1. BaseClass/method1()
2. NextClass/method2()
3. BaseClass/method3()
4. AnotherClass/method4()
5. MyInterface/ifaceMethod()
MyClass
1. MyClass/method5()
2. MyInterface/ifaceMethod()
As you can see, AnotherClass contains the interface's method in its fifth entry and MyClass contains it in its second entry. To actually find the correct entry in the virtual method table, a call to a method with invokeinterface will always have to search the complete table without a chance for the style of optimization that invokevirtual does.
There are additional differences like the fact, that invokeinterface can be used together with object references that do not actually implement the interface. Therefore, invokeinterface will have to check at runtime whether a method exists in the table and potentially throw an exception.

Comparing both instructions in the JVM Spec, the very first difference is that invokevirtual checks the accessibility of the method during the lookup, while invokeinterface doesn't.

Related

Benefit of non-empty method of non-abstract class and abstract method of abstract method?

I don't understand why we use abstract method (abstract class) while we can use empty method of non-abstract class and then we override it. Does it sound fine? I am seeking to clarify this issue.
I give 2 examples
public abstract class MyClass {public abstract void foo();}
public MyChildClass extends MyClass {public void foo() {//..TODO}}
public class MyClass {public void foo(){//empty}}
public class MyChildClass extends MyClass {public void foo() {//..TODO}}
Which one is better?
I'll start by saying that you should try to use interfaces instead of abstract classes. Abstract classes couple the subclass to the implementation of the superclass. In a language like Java, the subclass can override any method even if the superclass did not intend to do so, and most people don't qualify their methods with "do not override" all the time.
At the lowest level, abstract methods give you two concrete protections at compile time:
They force you to override the method in a subclass
They disallow the creation of the abstract class
Before listing the use cases for abstract methods, I'll just say that "common functionality" is NOT a good reason for an abstract base class. If you need common functionality, just create a class that has the common methods, and let the various classes call these functions as they see fit.
So when would you use an abstract class? Here are some examples:
Template Method
In the template method pattern, you have all of your functionality, but there's just one internal aspect that's polymorphic, so you have subclasses that override that particular aspect.
For example, if you're implementing a cache, but the cache invalidation policy is polymorphic, you may have an abstract invalidate() method that is called internally by other methods, but it's up to subclasses to implement invalidate().
If there is a preferred default cache invalidation policy, then invalidate() could implement that default. But if that default is downright destructive in some cases, then it shouldn't be a default - it should be abstract, and the code that creates the cache should be forced to explicitly choose the invalidation policy.
This can also be achieved by passing an Invalidator class to the constructor (Strategy pattern), but if the invalidation logic needs to call methods of the cache, it's better to make those method protected and call them from a subclass (i.e. Template Method pattern).
Default implementation of other methods
In languages where interfaces cannot have default methods (e.g. Java 7), you can emulate it using abstract classes. All the interface methods will be abstract, but the default methods would be regular public methods.
Common Interface and Functionality
This is just a more generic version of the template method pattern. The difference is that the polymorphic methods are part of the API.
If your common functionality has a lot of overlap with the functionality you want to expose, and you don't want mountains of boilerplate code, you use an abstract class. For example:
interface File {
abstract Buffer read(int size);
abstract void write(Buffer buf);
abstract long getSize();
abstract void setSize();
// ... get/set creation time, get/set modification time, get
// file type etc.
abstract long getOwner();
abstract void setOwner(long owner);
}
abstract class AbstractFile extends File {
DataMap dataMap;
MetadataMap metaMap;
protected getDiskMap() { return dataMap; }
protected getMetaMap() { return metaMap; }
public Buffer read(int size) { /* loop here */ }
public void write(Buffer buf) { /* loop here */ }
public long getSize() { /* logic */ }
public void setSize() { /* logic */ }
// ... implementation of get/set creation time, get/set modification
// time, get file type etc.
}
abstract class HardDriveFile extends AbstractFile {
OwnershipMap ownerMap;
abstract long getOwner() { /* logic */ }
abstract void setOwner(long owner) { /* logic */ }
}
abstract class ThumbDriveFile extends AbstractFile {
// thumb drives have no ownership
abstract long getOwner() { return 0; }
abstract void setOwner(long owner) { /* no-op */ }
}
abstract class SomeOtherfile extends AbstractFile {
...
}
If we cut the middleman and have HardDriveFile and ThumbDriveFile (and possibly other types of files) implement File and spell out all the common methods, each calling a method of some common class, we would get mountains and mountains of boilerplate. So we inherit from an abstract base class, that has the abstract methods we want to specialize (e.g. based on the existence of an ownership map).
The naive thing to do would be to combine File and AbstractFile into a single class, which is where you'd get the abstract methods getOwner() and setOwner(), but it's better to hide abstract classes behind actual interfaces, to prevent the coupling between consumers of an API and the abstract class.

Avoid adding/extending methods to interface

I have a scenario , where my current interface looks like
public interface IMathematicalOperation
{
void AddInt();
}
After an year i expect the interface to be extended with AddFloat method and also expect 100 users already consuming this interface. When i extend the interface with a new method after an year i don't want these 100 classes to get changed.
So how can i tackle this situation ? Is there any design pattern available already to take care of this situation ?
Note: i understand that i can have a abstract class which implement this interface and make all the methods virtual , so that clients can inherit from this class rather than the interface and override the methods . When i add a new method only the abstract class will be changed and the clients who are interested in the method will override the behavior (minimize the change) .
Is there any other way of achieving the same result (like having a method named Add and based on certain condition it will do Float addition or Integer addition) ?
Edit 1:
The new method gets added to the interface also needs to be called automatically along with the existing methods(like chain of responsibility pattern).
There are at least two possible solution I can think of:
Derive your new interface from your old interface
public interface IMathematicalOperation
{
void AddInt();
}
public interface IFloatingPointMathematicalOperation : IMathematicalOperation
{
void AddFloat();
}
Have simply a parallel interface which contains the new method and have all classes which need the new interface derive from it
I'd suggest the second solution, since I don't understand why you would want an established interface to change.
I encountered a similar issue some time ago and found the best way was not to try and extend an existing interface, but to provide different versions of the interface with each new interface providing extra functionality. Over time I found that was not adding functionality on a regular basis, may once a year, so adding extra interfaces was never really an issue.
So, for example this is your first version of the interface:
public interface IMathematicalOperation
{
void AddInt();
}
This interface would then be implemented on a class like this:
public class MathematicalOperationImpl : IMathematicalOperation
{
public void AddInt()
{
}
}
Then when you need to add new functionality, i.e. create a version 2, you would create another interface with the same name, but with a "2" on the end:
public interface IMathematicalOperation2 : IMathematicalOperation
{
void AddFloat();
}
And the MathematicalOperationImpl would be extended to implement this new interface:
public class MathematicalOperationImpl : IMathematicalOperation, IMathematicalOperation2
{
public void AddInt()
{
}
public void AddFloat()
{
}
}
All of your new/future clients could start using the version 2 interface, but your existing clients would continue to work because they will only know about the first version of the interface.
The options provided are syntactically viable but then, as is obvious, they won't apply to any previous users.
A better option would be to use the Visitor pattern
The pattern is best understood when you think about the details of OO code
this.foo(); // is identical to
foo(this);
Remember that there is always a hidden 'this' parameter passed with every instance call.
What the visitor pattern attempts to do is generalize this behavior using Double dispatch
Let's take this a hair further
public interface MathematicalOperation
{
void addInt();
void accept(MathVisitor v);
}
public interface MathVisitor {
void visit(MathematicalOperation operation);
}
public class SquareVistor implements MathVisitor {
void visit(MathematicalOperation operation) {
operation.setValue(operation.getValue() * 2);
}
}
public abstract class AbstractMathematicalOperation implements MathematicalOperation {
public void accept(MathVisitor f) {
f.visit(this); // we are going to do 'f' on 'this'. Or think this.f();
}
}
public class MyMathOperation extends AbstractMathematicalOperation {
}
someMathOperation.visit(new SquareVisitor()); // is now functionally equivalent to
someMathOperation.square();
The best bet would be for you to roll-out your initial interface with a visitor requirements, then immediately roll-out an abstract subclass that gives this default implementation so it's cooked right in (As the above class is). Then everyone can just extend it. I think you will find this gives you the flexibility you need and leaves you will the ability to work with legacy classes.

Composition over inheritance 2

Please have a look at the following question: Favor composition over inheritance
The accepted answerer says: " it extends Hashtable, in order to reuse its methods and to avoid reimplementing some of them using delegation". I am not sure what the answerer means by: reimplementing some of them using delegation. What does the answerer mean?
I am familiar with Delegates and the Observer design pattern.
When using composition, if you want to support a method that the underlying class has, you must define your own implementation that simply delegates (or uses) the same method on the underlying class. It can be tempting to use inheritance in this case to avoid writing that simple (delegated) method, but inheritance really should be used only when an IS-A relationship exists.
For example,
public class Foo
{
public virtual void Bar()
{
// do something
}
}
public class InheritedFromFoo : Foo
{
// we get Bar() for free!!!
}
public class ComposedWithFoo
{
private Foo _foo;
public void Bar()
{
_foo.Bar(); // delegated to the Foo instance
}
}

Singleton subclass

I have an abstract base class and an implementation class like:
public abstract class Base
{
public Base getInstance( Class<? extends Base> clazz )
{
//expected to return a singleton instance of clazz's class
}
public abstract absMeth();
}
public A extends Base
{
//expected to be a singleton
}
In this example I can make A to be a singleton and even write getInstance in Base to return a singleton object of A for every call, doing this way:
public abstract class Base
{
public Base getInstance( Class<? extends Base> clazz )
{
try
{
return clazz.getDeclaredMethod("getInstance").invoke(null,null);
}
}
public abstract void absMeth();
}
public A extends Base
{
private static A inst;
private A(){}
public static A getInstance( )
{
if( inst!= null)
inst = new A();
return inst;
}
public void absMeth(){
//...
}
}
But my concern is how do I ensure that if somebody writes another class class B extends Base it should also be a singleton and it necessarily implements a static method called getInstance?
In other words I need to enforce this as a specification for all classes extending with the Base class.
You cannot trust classes that extend you to create a single instance of themselves1: even if you could somehow ensure that they all implement getInstance, there is no way to tell that inside that method they check inst before constructing a new instance of themselves.
Stay in control of the process: create a Map<Class,Base>, and instantiate the class passed in through reflection2. Now your code can decide whether to create an instance or not, without relying on the getInstance of a subclass.
1 A popular saying goes, "If you want a job done right, do it yourself."
2 Here is a link describing a solution based on setAccessible(true)
Singleton is a design pattern, not a language feature. It is pretty much impossible to somehow enforce it on the inheritance tree through syntax.
It certainly is possible to require all subclasses to implement a method by declaring it abstract but there is no way to control implementation details. Singleton is all about implementation details.
But why is this a concern at all? Do not make your app dependant on internal details of someone else's code. It is Bad Design™ and having this issue is a sure sign of it. Code against a well-defined interface and avoid relying on internal details.

What is the use of making constructor private in a class?

Why should we make the constructor private in class? As we always need the constructor to be public.
Some reasons where you may need private constructor:
The constructor can only be accessed from static factory method inside the class itself. Singleton can also belong to this category.
A utility class, that only contains static methods.
By providing a private constructor you prevent class instances from being created in any place other than this very class. There are several use cases for providing such constructor.
A. Your class instances are created in a static method. The static method is then declared as public.
class MyClass()
{
private:
MyClass() { }
public:
static MyClass * CreateInstance() { return new MyClass(); }
};
B. Your class is a singleton. This means, not more than one instance of your class exists in the program.
class MyClass()
{
private:
MyClass() { }
public:
MyClass & Instance()
{
static MyClass * aGlobalInst = new MyClass();
return *aGlobalInst;
}
};
C. (Only applies to the upcoming C++0x standard) You have several constructors. Some of them are declared public, others private. For reducing code size, public constructors 'call' private constructors which in turn do all the work. Your public constructors are thus called delegating constructors:
class MyClass
{
public:
MyClass() : MyClass(2010, 1, 1) { }
private:
MyClass(int theYear, int theMonth, int theDay) { /* do real work */ }
};
D. You want to limit object copying (for example, because of using a shared resource):
class MyClass
{
SharedResource * myResource;
private:
MyClass(const MyClass & theOriginal) { }
};
E. Your class is a utility class. That means, it only contains static members. In this case, no object instance must ever be created in the program.
To leave a "back door" that allows another friend class/function to construct an object in a way forbidden to the user. An example that comes to mind would be a container constructing an iterator (C++):
Iterator Container::begin() { return Iterator(this->beginPtr_); }
// Iterator(pointer_type p) constructor is private,
// and Container is a friend of Iterator.
Everyone is stuck on the Singleton thing, wow.
Other things:
Stop people from creating your class on the stack; make private constructors and only hand back pointers via a factory method.
Preventing creating copys of the class (private copy constructor)
This can be very useful for a constructor that contains common code; private constructors can be called by other constructors, using the 'this(...);' notation. By making the common initialization code in a private (or protected) constructor, you are also making explicitly clear that it is called only during construction, which is not so if it were simply a method:
public class Point {
public Point() {
this(0,0); // call common constructor
}
private Point(int x,int y) {
m_x = x; m_y = y;
}
};
There are some instances where you might not want to use a public constructor; for example if you want a singleton class.
If you are writing an assembly used by 3rd parties there could be a number of internal classes that you only want created by your assembly and not to be instantiated by users of your assembly.
This ensures that you (the class with private constructor) control how the contructor is called.
An example : A static factory method on the class could return objects as the factory method choses to allocate them (like a singleton factory for example).
We can also have private constructor,
to enfore the object's creation by a specific class
only(For security reasons).
One way to do it is through having a friend class.
C++ example:
class ClientClass;
class SecureClass
{
private:
SecureClass(); // Constructor is private.
friend class ClientClass; // All methods in
//ClientClass have access to private
// & protected methods of SecureClass.
};
class ClientClass
{
public:
ClientClass();
SecureClass* CreateSecureClass()
{
return (new SecureClass()); // we can access
// constructor of
// SecureClass as
// ClientClass is friend
// of SecureClass.
}
};
Note: Note: Only ClientClass (since it is friend of SecureClass)
can call SecureClass's Constructor.
You shouldn't make the constructor private. Period. Make it protected, so you can extend the class if you need to.
Edit: I'm standing by that, no matter how many downvotes you throw at this.
You're cutting off the potential for future development on the code. If other users or programmers are really determined to extend the class, then they'll just change the constructor to protected in source or bytecode. You will have accomplished nothing besides to make their life a little harder. Include a warning in your constructor's comments, and leave it at that.
If it's a utility class, the simpler, more correct, and more elegant solution is to mark the whole class "static final" to prevent extension. It doesn't do any good to just mark the constructor private; a really determined user may always use reflection to obtain the constructor.
Valid uses:
One good use of a protected
constructor is to force use of static
factory methods, which allow you to
limit instantiation or pool & reuse
expensive resources (DB connections,
native resources).
Singletons (usually not good practice, but sometimes necessary)
when you do not want users to create instances of this class or create class that inherits this class, like the java.lang.math, all the function in this package is static, all the functions can be called without creating an instance of math, so the constructor is announce as static.
If it's private, then you can't call it ==> you can't instantiate the class. Useful in some cases, like a singleton.
There's a discussion and some more examples here.
I saw a question from you addressing the same issue.
Simply if you don't want to allow the others to create instances, then keep the constuctor within a limited scope. The practical application (An example) is the singleton pattern.
Constructor is private for some purpose like when you need to implement singleton or limit the number of object of a class.
For instance in singleton implementation we have to make the constructor private
#include<iostream>
using namespace std;
class singletonClass
{
static int i;
static singletonClass* instance;
public:
static singletonClass* createInstance()
{
if(i==0)
{
instance =new singletonClass;
i=1;
}
return instance;
}
void test()
{
cout<<"successfully created instance";
}
};
int singletonClass::i=0;
singletonClass* singletonClass::instance=NULL;
int main()
{
singletonClass *temp=singletonClass::createInstance();//////return instance!!!
temp->test();
}
Again if you want to limit the object creation upto 10 then use the following
#include<iostream>
using namespace std;
class singletonClass
{
static int i;
static singletonClass* instance;
public:
static singletonClass* createInstance()
{
if(i<10)
{
instance =new singletonClass;
i++;
cout<<"created";
}
return instance;
}
};
int singletonClass::i=0;
singletonClass* singletonClass::instance=NULL;
int main()
{
singletonClass *temp=singletonClass::createInstance();//return an instance
singletonClass *temp1=singletonClass::createInstance();///return another instance
}
Thanks
You can have more than one constructor. C++ provides a default constructor and a default copy constructor if you don't provide one explicitly. Suppose you have a class that can only be constructed using some parameterized constructor. Maybe it initialized variables. If a user then uses this class without that constructor, they can cause no end of problems. A good general rule: If the default implementation is not valid, make both the default and copy constructor private and don't provide an implementation:
class C
{
public:
C(int x);
private:
C();
C(const C &);
};
Use the compiler to prevent users from using the object with the default constructors that are not valid.
Quoting from Effective Java, you can have a class with private constructor to have a utility class that defines constants (as static final fields).
(EDIT: As per the comment this is something which might be applicable only with Java, I'm unaware if this construct is applicable/needed in other OO languages (say C++))
An example as below:
public class Constants {
private Contants():
public static final int ADDRESS_UNIT = 32;
...
}
EDIT_1:
Again, below explanation is applicable in Java : (and referring from the book, Effective Java)
An instantiation of utility class like the one below ,though not harmful, doesn't serve
any purpose since they are not designed to be instantiated.
For example, say there is no private Constructor for class Constants.
A code chunk like below is valid but doesn't better convey intention of
the user of Constants class
unit = (this.length)/new Constants().ADDRESS_UNIT;
in contrast with code like
unit = (this.length)/Constants.ADDRESS_UNIT;
Also I think a private constructor conveys the intention of the designer of the Constants
(say) class better.
Java provides a default parameterless public constructor if no constructor
is provided, and if your intention is to prevent instantiation then a private constructor is
needed.
One cannot mark a top level class static and even a final class can be instantiated.
Utility classes could have private constructors. Users of the classes should not be able to instantiate these classes:
public final class UtilityClass {
private UtilityClass() {}
public static utilityMethod1() {
...
}
}
You may want to prevent a class to be instantiated freely. See the singleton design pattern as an example. In order to guarantee the uniqueness, you can't let anyone create an instance of it :-)
One of the important use is in SingleTon class
class Person
{
private Person()
{
//Its private, Hense cannot be Instantiated
}
public static Person GetInstance()
{
//return new instance of Person
// In here I will be able to access private constructor
}
};
Its also suitable, If your class has only static methods. i.e nobody needs to instantiate your class
It's really one obvious reason: you want to build an object, but it's not practical to do it (in term of interface) within the constructor.
The Factory example is quite obvious, let me demonstrate the Named Constructor idiom.
Say I have a class Complex which can represent a complex number.
class Complex { public: Complex(double,double); .... };
The question is: does the constructor expects the real and imaginary parts, or does it expects the norm and angle (polar coordinates) ?
I can change the interface to make it easier:
class Complex
{
public:
static Complex Regular(double, double = 0.0f);
static Complex Polar(double, double = 0.0f);
private:
Complex(double, double);
}; // class Complex
This is called the Named Constructor idiom: the class can only be built from scratch by explicitly stating which constructor we wish to use.
It's a special case of many construction methods. The Design Patterns provide a good number of ways to build object: Builder, Factory, Abstract Factory, ... and a private constructor will ensure that the user is properly constrained.
In addition to the better-known uses…
To implement the Method Object pattern, which I’d summarize as:
“Private constructor, public static method”
“Object for implementation, function for interface”
If you want to implement a function using an object, and the object is not useful outside of doing a one-off computation (by a method call), then you have a Throwaway Object. You can encapsulate the object creation and method call in a static method, preventing this common anti-pattern:
z = new A(x,y).call();
…replacing it with a (namespaced) function call:
z = A.f(x,y);
The caller never needs to know or care that you’re using an object internally, yielding a cleaner interface, and preventing garbage from the object hanging around or incorrect use of the object.
For example, if you want to break up a computation across methods foo, bar, and zork, for example to share state without having to pass many values in and out of functions, you could implement it as follows:
class A {
public static Z f(x, y) {
A a = new A(x, y);
a.foo();
a.bar();
return a.zork();
}
private A(X x, Y y) { /* ... */ };
}
This Method Object pattern is given in Smalltalk Best Practice Patterns, Kent Beck, pages 34–37, where it is the last step of a refactoring pattern, ending:
Replace the original method with one that creates an instance of the new class, constructed with the parameters and receiver of the original method, and invokes “compute”.
This differs significantly from the other examples here: the class is instantiable (unlike a utility class), but the instances are private (unlike factory methods, including singletons etc.), and can live on the stack, since they never escape.
This pattern is very useful in bottoms-up OOP, where objects are used to simplify low-level implementation, but are not necessarily exposed externally, and contrasts with the top-down OOP that is often presented and begins with high-level interfaces.
Sometimes is useful if you want to control how and when (and how many) instances of an object are created.
Among others, used in patterns:
Singleton pattern
Builder pattern
On use of private constructors could also be to increase readability/maintainability in the face of domain-driven design.
From "Microsoft .NET - Architecing Applications for the Enterprise, 2nd Edition":
var request = new OrderRequest(1234);
Quote, "There are two problems here. First, when looking at the code, one can hardly guess what’s going
on. An instance of OrderRequest is being created, but why and using which data? What’s 1234? This
leads to the second problem: you are violating the ubiquitous language of the bounded context. The
language probably says something like this: a customer can issue an order request and is allowed to
specify a purchase ID. If that’s the case, here’s a better way to get a new OrderRequest instance:"
var request = OrderRequest.CreateForCustomer(1234);
where
private OrderRequest() { ... }
public OrderRequest CreateForCustomer (int customerId)
{
var request = new OrderRequest();
...
return request;
}
I'm not advocating this for every single class, but for the above DDD scenario I think it makes perfect sense to prevent a direct creation of a new object.
If you create a private constructor you need to create the object inside the class
enter code here#include<iostream>
//factory method
using namespace std;
class Test
{
private:
Test(){
cout<<"Object created"<<endl;
}
public:
static Test* m1(){
Test *t = new Test();
return t;
}
void m2(){
cout<<"m2-Test"<<endl;
}
};
int main(){
Test *t = Test::m1();
t->m2();
return 0;
}