Functions in JavaScript are objects. Like any other object, function is just a collection of name and value pair. The only difference between a function object and a regular object is that function can be invoked and additionally associated with two hidden properties: function context and function code. This might come as a surprise and unnatural, especially if you are coming from a more procedural programming background. Nevertheless this is the critical insight most of us need, to make sense of some of the strange ways that D3 uses function.

Now with this insight in mind, let's take a look at the code snippet again:

  var instance = {}; // <-- A

  var headline, description; // <-- B

  instance.render = function () {
    var div = d3.select('body').append("div");

    div.append("h3").text(headline); // <-- C

    div.attr("class", "box")
      .attr("style", "color:" + spec.color) // <-- D
      .append("p")
      .text(description); // <-- E

    return instance; // <-- F
  };

At line marked as A, B, and C we can clearly see that instance, headline, and description are all internal private variables belonging to the SimpleWidget function object. While the render function is a function associated with the instance object which itself is defined as an object literal. Since functions are just an object it can also be stored in an object/function, other variables, arrays, and being passed as function arguments. The result of the execution of function SimpleWidget is the returning of object instance at line I.

function SimpleWidget(spec) {
...
  return instance; // <-- I
}

Curious readers are probably asking by now how the variable scoping is resolved in this example since the render function has seemingly strange access to not only the instance, headline, and description variables but also the spec variable that is passed into the base SimpleWidget function. This seemingly strange variable scoping is actually determined by a simple static scoping rule. This rule can be thought as the following: whenever searching for a variable reference, variable search will be first performed locally. When variable declaration is not found (as in the case of headline on line C) then the search continues to the parent object (in this case SimpleWidget function is its static parent and headline variable declaration is found at line B). If still not found, then this process will continue recursively to the next static parent so on and so forth till it reaches global variable definition, if still not found then a reference error will be generated for this variable. This scoping behavior is very different from variable resolution rules in some of the most popular languages such as Java and C#; it might take some time to get used to, however don't worry too much about it if you still find it confusing. With more practice and keeping static scoping rule in mind you will be comfortable with this kind of scoping in no time.

for(var i = 0; i < 10; i++){
  for(var i = 0; i < 2; i++){
    console.log(i);
  }
}

This pattern is usually referred as functional when compared with the more popular prototype-based Pseudo-classical pattern. The advantage of the functional pattern is that it provides a much better mechanism for information hiding and encapsulation since the private variables—in our case the headline and description variables—are only accessible by nested functions via the static scoping rule therefore the object returned by the SimpleWidget function is flexible yet more tamper-proof and durable.

Something strange happens on line G:

instance.headline = function (h) {
  if (!arguments.length) h; // <-- G
  headline = h;
  return instance; // <-- H
};

You might be asking where this arguments variable on line G came from. It was never defined anywhere in this example. The arguments variable is a built-in hidden parameter that is available to functions when they are invoked. The arguments variable contains all arguments for a function invocation in an array.

var newArgs = Array.prototype.slice.apply(arguments);

This hidden parameter when combined with the ability to omit function argument in JavaScript allows you to write a function like instance.headline with unspecified number of parameters. In this case, we can either have one argument h or none. Because arguments.length returns 0 when no parameter is passed; therefore the headline function returns h if no parameter is passed, otherwise it turns into a setter if parameter h is provided. To clarify this explanation let's take a look at the following code snippet:

var widget = SimpleWidget({color: "#6495ed"})
    .headline("Simple Widget"); // set headline
console.log(widget.headline()); // prints "Simple Widget"

Here you can see how headline function can be used as both setter and getter with different parameters.

The next interesting aspect of this particular example is the capability of chaining functions to each other. This is also the predominant function invocation pattern that the D3 library deploys since most of the D3 functions are designed to be chainable to provide a more concise and contextual programming interface. This is actually quite simple once you understand the variable-parameter function concept. Since a variable-parameter function—such as the headline function—can serve as setter and getter at the same time, then returning the instance object when it is acting as a setter allows you to immediately invoke another function on the invocation result; hence the chaining.

Let's take a look at the following code:

var widget = SimpleWidget({color: "#6495ed"})
  .headline("Simple Widget")
  .description("This is ...")
  .render();

In this example, the SimpleWidget function returns the instance object (as on line I). Then, the headline function is invoked as a setter, which also returns the instance object (as on line H). The description function can then be invoked directly on return which again returns the instance object. Then finally the render function can be called.

Now with the knowledge of functional JavaScript and a working ready-to-go D3 data visualization development environment, we are ready to dive into the rich concepts and techniques that D3 has to offer. However before we take off, I would like to cover a few more important areas—how to find and share code and how to get help when you are stuck.

One of the great things about D3 when compared with other visualization options is that it offers a wealth of examples and tutorials that you can draw your inspiration from. During the course of creating my own open source visualization charting library and the creation of this book, I had drawn heavily on these resources. I am going to list some of the most popular options available in this aspect. This list is by no means a comprehensive directory but rather a starting place for you to explore:

Even with all the examples, tutorial, and cookbook like this, you might still run into challenges when creating your visualization. Good news here is that D3 has a broad and active support community. Simply "googling" your question can most often yield a satisfying answer. Even if it does not, don't worry; D3 has a robust community-based support: