Sometimes, we write the following code while working with EventHandler and that may create some problems. I will discuss this later on. Let us see the code:

private void RaiseOperationComplete(EventArgs<string> e)
{
    if(OperationComplete != null)
    {
        OperationComplete(this, e);
    }
}

In the above case, suppose two threads subscribed to the same OperationComplete handler and one unsubscribed the event. In such a scenario, the OperationComplete event will become null and for the next thread since the event is null, it will not execute. This happens mainly in multithreaded applications, but you can't guarantee for single threaded applications too.

So what do we do? How can we make sure that our application will not come into such a situation and work properly?

Yup, let us modify the code a bit. First, we will store the event handler to a local variable and then we will raise the event from the locally cached copy. Have a look into the following code:

private void RaiseOperationComplete(EventArgs<string> e)
{
    EventHandler<string<eventargs><string>> handler = OperationComplete;
    if(handler != null)
    {
        handler(this, e);
    }
}

Here, when we execute the OperationComplete event handler, we will store the current value in a local variable named "handler". Hence, both "handler" and "OperationComplete" will point to the same object. Thus, we will create a cached copy of the same value for the handler and then raise the event.

The reason you do this is because the assignment is a copy of the delegate instance at that point, and not a by-value copy. Doing this, the reference will be cached just in case another thread, or re-entrant code, nulls out the reference before you call it. The original value will not change though it will set the event to null.

So, remember to always follow the second approach and be in the safer place of coding.
Thanks to Pete O'Hanlon for pointing this to me in this post.

CodeProject