Usage of binding to constants and binding to types in scopes with Ninject

Question

Which way of creating bindings of single object to interface is preferable, when and why:

Kernel.Bind<IFoo>().ToConstant(new Foo());

or

Kernel.Bind<IFoo>().To(typeof(Foo)).InSingletonScope();

Or, if both ways are incorrect and better to be avoided, what should be used instead?

Answer 1

With both constructions you accomplish the same. However, in the latter approach construction of the single Foo object is deferred until the first Get call. Let me illustrate that with a little example. Consider the following application:

class Program
{
    static void Main(string[] args)
    {
        Console.WriteLine("Starting the app");

        IKernel kernel = new StandardKernel();
        kernel.Bind<IFoo>().ToConstant(new Foo());

        Console.WriteLine("Binding complete");

        kernel.Get<IFoo>();

        Console.WriteLine("Stopping the app");
    }
}

public interface IFoo
{
}

public class Foo : IFoo
{
    public Foo()
    {
        Console.WriteLine("Foo constructor called");
    }
}

This gets you the output:

Starting the app
Foo constructor called
Binding complete
Stopping the app

Now, let's replace the ToConstant call with To(typeof(Foo)).InSingletonScope()

class Program
{
    static void Main(string[] args)
    {
        Console.WriteLine("Starting the app");

        IKernel kernel = new StandardKernel();
        kernel.Bind<IFoo>().To(typeof(Foo)).InSingletonScope();

        Console.WriteLine("Binding complete");

        kernel.Get<IFoo>();

        Console.WriteLine("Stopping the app");
    }
}

public interface IFoo
{
}

public class Foo : IFoo
{
    public Foo()
    {
        Console.WriteLine("Foo constructor called");
    }
}

Now the output is:

Starting the app
Binding complete
Foo constructor called
Stopping the app