Closures: Line by Line explanation of "Javascript: Good Parts" example?

Question

I'm reading "Javascript: The Good Parts" and am totally baffled by what's really going on here. A more detailed and/or simplified explanation would be greatly appreciated.

// BAD EXAMPLE

// Make a function that assigns event handler functions to an array  of nodes the wrong way.
// When you click on a node, an alert box is supposed to display the ordinal of the node.
// But it always displays the number of nodes instead.

var add_the_handlers = function (nodes) {
    var i;
    for (i = 0; i < nodes.length; i += 1) {
        nodes[i].onclick = function (e) {
            alert(i);
        }
    }
};

// END BAD EXAMPLE

The add_the_handlers function was intended to give each handler a unique number (i). It fails because the handler functions are bound to the variable i, not the value of the variable i at the time the function was made:

// BETTER EXAMPLE

// Make a function that assigns event handler functions to an array of nodes the right way.
// When you click on a node, an alert box will display the ordinal of the node.

var add_the_handlers = function (nodes) {
    var i;
    for (i = 0; i < nodes.length; i += 1) {
        nodes[i].onclick = function (i) {
            return function (e) {
                alert(i);
            };
        }(i);
    }
};

Now, instead of assigning a function to onclick, we define a function and immediately invoke it, passing in i. That function will return an event handler function that is bound to the value of i that was passed in, not to the i defined in add_the_handlers. That returned function is assigned to onclick.

Answer 1

I think this is a very common source of confusion for newcomers to JavaScript. First I would suggest checking out the following Mozilla Dev article for brief introduction on the topic of closures and lexical scoping:

• Mozilla Dev Center: Working with Closures

Let's start with the bad one:

var add_the_handlers = function (nodes) {
// Variable i is declared in the local scope of the add_the_handlers() 
// function. 
   var i;

// Nothing special here. A normal for loop.
   for (i = 0; i < nodes.length; i += 1) {

// Now we are going to assign an anonymous function to the onclick property.
       nodes[i].onclick = function (e) {

// The problem here is that this anonymous function has become a closure. It 
// will be sharing the same local variable environment as the add_the_handlers()
// function. Therefore when the callback is called, the i variable will contain 
// the last value it had when add_the_handlers() last returned. 
           alert(i);
       }
   }

// The for loop ends, and i === nodes.length. The add_the_handlers() maintains
// the value of i even after it returns. This is why when the callback
// function is invoked, it will always alert the value of nodes.length.
};

We can tackle this problem with more closures, as Crockford suggested in the "good example". A closure is a special kind of object that combines two things: a function, and the environment in which that function was created. In JavaScript, the environment of the closure consists of any local variables that were in-scope at the time that the closure was created:

 // Now we are creating an anonymous closure that creates its own local 
 // environment. I renamed the parameter variable x to make it more clear.
 nodes[i].onclick = function (x) {

     // Variable x will be initialized when this function is called.

     // Return the event callback function.
     return function (e) {
         // We use the local variable from the closure environment, and not the 
         // one held in the scope of the outer function add_the_handlers().
         alert(x);
     };
 }(i); // We invoke the function immediately to initialize its internal 
       // environment that will be captured in the closure, and to receive
       // the callback function which we need to assign to the onclick.

Rather than having the callbacks all sharing a single environment, the closure function creates a new environment for each one. We could also have used a function factory to create a closure, as in the following example:

function makeOnClickCallback (x) {
   return function (e) {
      alert(x);
   };
}

for (i = 0; i < nodes.length; i += 1) {
   nodes[i].onclick = makeOnClickCallback(i);
}

Answer 2

It's all about closures. In the first example, "i" will be equal to "nodes.length" for every click event handler, because it uses "i" from the loop which creates the event handlers. By the time the event handler is called, the loop will have ended, so "i" will be equal to "nodes.length".

In the second example, "i" is a parameter (so a local variable). The event handlers will use the value of the local variable "i" (the parameter).