singleton pattern in C# programming language

 singleton patten

A class follows the singleton pattern if it can have one instance at most. Design patterns are very important in object-oriented design and development, so many interviewers like questions related to patterns. Singleton is the only pattern that can be implemented in dozens of lines of code, so it is quite suitable for interviews.
 Single-Threading Applications 

When the constructor is defined as a private method, none of the code outside the class can create its instances. A static method inside the class is defined to create an instance on demand. The class Singleton1 is implemented based on the solution in Listing 2-13.
Listing 2-13. C# Code for Singleton (Version 1)
public class Singleton1 {
    private Singleton1() {
    }
 private static Singleton1 instance = null; public static Singleton1 Instance {
        get {
            if (instance == null)
                instance = new Singleton1();
            return instance;
} }
}
In this class, an instance is created only when static field Singleton1.instance is null, so it does not
have the opportunity to get multiple instances.
Works with Multiple Threads but Is Inefficient
Singleton1 works when there is only one thread, but it has problems when there are multiple threads in an application. Supposing that there are two threads concurrently reaching the if statement to check whether instance is null. If instance is not created yet, each thread will create one separately. It violates the definition of the singleton pattern when two instances are created. In order to make it work in multithreading applications, a lock is introduced as shown in the Singleton2 in Listing 2-14.
Listing 2-14. C# Code for Singleton (Version 2)
public class Singleton2 {
    private Singleton2() {
    }
    private static readonly object syncObj = new object();
    private static Singleton2 instance = null;
    public static Singleton2 Instance {
        get {
            lock (syncObj) {
                if (instance == null)
                    instance = new Singleton2();
}
            return instance;
        }
} }
Suppose there are two threads that are both going to create their own instances. As we know, only one thread can get the lock at a time. When one thread gets it, the other one has to wait. The first thread that gets the lock finds that instance is null, so it creates an instance. After the first thread releases the lock, the second thread gets it. Since the instance was already created by the first thread, the if statement is false

An instance will not be recreated again. Therefore, it guarantees that there is one instance at most when there are multiple threads executing concurrently.
The class Singleton2 is far from perfect. Every time Singleton2.Instance.get executes, it has to get and release a lock. Operations to get and release a lock are time-consuming, so they should be avoided as much as possible.
Double-Check around Locking
Actually a lock is needed only before the only instance is created in order to make sure that only one thread get the chance to create an instance. After the instance is created, no lock is necessary. We can improve performance with an additional if check before the lock as shown in Listing 2-15.
Listing 2-15. C# Code for Singleton (Version 3)
public class Singleton3 {
    private Singleton3() {
    }
    private static object syncObj = new object();
    private static Singleton3 instance = null;
    public static Singleton3 Instance {
        get {
            if (instance == null) {
                lock (syncObj) {
                    if (instance == null)
                        instance = new Singleton3();
} }
            return instance;
        }
} }
In the class Singleton3, it locks only when instance is null. When the instance has been created, it is returned directly without any locking operations. Therefore, the time efficiency of Singleton3 is better than Singleton2.
Singleton3 employs two if statements to improve time efficiency. It is a workable solution, but its logic looks a bit complex, and it is error-prone for many candidates during interviews.
Utilization of Static Constructors
It is guaranteed that a static constructor in a C# class is called only once at most. If it only creates an instance in the static constructor, there is one instance at most. A concise solution for the singleton pattern with a static constructor is shown in Listing 2-16.
Listing 2-16. C# Code for Singleton (Version 4)
public class Singleton4 {
    private Singleton4() {
    }
    private static Singleton4 instance = new Singleton4();
    public static Singleton4 Instance {
        get {
            return instance;
} }
}
In the class Singleton4 above, an instance is created when the static field instance gets initialized. Static fields in C# are initialized when the static constructor is called. Since the static constructor is called only once by the .NET runtime, it is guaranteed that only one instance is created even in a multithreading application.
The time to execute the static constructor is out of the programmers' control. When the .NET runtime reaches any code of a class the first time, it invokes the static constructor automatically. Therefore, the time to initialize instance is not the first time to invoke Singleton4.Instance. If a static method is added into Singleton4, it is not necessary to create an instance to invoke such a static method. However, the .NET runtime invokes the static constructors automatically to create an instance when it reaches any code for Singleton4. Therefore, it is possible to create an instance too early, and it impairs the space efficiency.
Creating an Instance When Necessary
In the last implementation of the singleton pattern in Listing 2-17, Singleton5, it creates the only instance on demand.
Listing 2-17. C# Code for Singleton (Version 5)
public class Singleton5 {
    Singleton5() {
}
    public static Singleton5 Instance {
        get {
            return Nested.instance;
}
class Nested {
        static Nested() {
        }
        internal static readonly Singleton5 instance = new Singleton5();
    }
}
There is a nested private class Nested in the code for Singleton5. When the .NET runtime reaches the code of the class Nested, its static constructor is invoked automatically, which creates an instance of type Singleton5. The class Nested is used only in the property Singleton5.Instance. Since the nested class is defined as private, it is inaccessible outside of the class Singleton5.
When the get method of Singleton5.Instance is invoked the first time, it triggers execution of the static constructor of the class Nested to create an instance of Singleton5. The instance is created only when it is necessary, so it avoids the waste associated with creating the instance too early.
Solution Comparison
Five solutions are introduced in this section. The first solution is workable only in a single-threading application. The second one works in a multiple-threading application, but it is inefficient because of unnecessary locking operations. The first two solutions are not acceptable from an interviewer's perspective. In the third solution, it employs two if statements and one lock to make sure it works in a multiple-threading application efficiently. The fourth one utilizes the static constructor to guarantee only an instance is created. The last solution improves space efficiency with a nested class to create the instance only when it is necessary. The last two solutions are recommended for interviews.
Source Code:
    003_Singleton.cs
 www.cinterviews.com appreciates your contribution please mail us the questions you have to cinterviews.blogspot.com@gmail.com so that it will be useful to our job search community

No comments: