Douglas
Crockford
www.crockford.com
And you think you're so clever and classless and free
John Lennon
JavaScript is a class-free, object-oriented language, and as such, it uses prototypal inheritance instead of classical inheritance. This can be puzzling to programmers trained in conventional object-oriented languages like C++ and Java. JavaScript's prototypal inheritance has more expressive power than classical inheritance, as we will see presently.
Java | JavaScript |
---|---|
Strongly-typed | Loosely-typed |
Static | Dynamic |
Classical | Prototypal |
Classes | Functions |
Constructors | Functions |
Methods | Functions |
But first, why do we care about inheritance at all? There are primarily two reasons. The first is type convenience. We want the language system to automatically cast references of similar classes. Little type-safety is obtained from a type system which requires the routine explicit casting of object references. This is of critical importance in strongly-typed languages, but it is irrelevant in loosely-typed languages like JavaScript, where object references never need casting.
The second reason is code reuse. It is very common to have a quantity of objects all implementing exactly the same methods. Classes make it possible to create them all from a single set of definitions. It is also common to have objects that are similar to some other objects, but differing only in the addition or modification of a small number of methods. Classical inheritance is useful for this but prototypal inheritance is even more useful.
To demonstrate this, we will introduce a little sugar which will let us write in a style that resembles a conventional classical language. We will then show useful patterns which are not available in classical languages. Then finally, we will explain the sugar.
First, we will make a Parenizor class that will have set and get methods for its value, and a toString method that will wrap the value in parens.
function Parenizor(value) { this.setValue(value); } Parenizor.method('setValue', function (value) { this.value = value; return this; }); Parenizor.method('getValue', function () { return this.value; }); Parenizor.method('toString', function () { return '(' + this.getValue() + ')'; });
The syntax is a little unusual, but it is easy to recognize the classical pattern in it. The method method takes a method name and a function, adding them to the class as a public method.
So now we can write
myParenizor = new Parenizor(0); myString = myParenizor.toString();
As you would expect, myString is "(0)".
Now we will make another class which will inherit from Parenizor, which is the same except that its toString method will produce "-0-" if the value is zero or empty.
function ZParenizor(value) { this.setValue(value); } ZParenizor.inherits(Parenizor); ZParenizor.method('toString', function () { if (this.getValue()) { return this.uber('toString'); } return "-0-"; });
The inherits method is similar to Java's extends. The uber method is similar to Java's super. It lets a method call a method of the parent class. (The names have been changed to avoid reserved word restrictions.)
So now we can write
myZParenizor = new ZParenizor(0); myString = myZParenizor.toString();
This time, myString is "-0-".
JavaScript does not have classes, but we can program as though it does.
By manipulating a function's prototype object, we can implement multiple inheritance, allowing us to make a class built from the methods of multiple classes. Promiscuous multiple inheritance can be difficult to implement and can potentially suffer from method name collisions. We could implement promiscuous multiple inheritance in JavaScript, but for this example we will use a more disciplined form called Swiss Inheritance.
Suppose there is a NumberValue class that has a setValue method that checks that the value is a number in a certain range, throwing an exception if necessary. We only want its setValue and setRange methods for our ZParenizor. We certainly don't want its toString method. So, we write
ZParenizor.swiss(NumberValue, 'setValue', 'setRange');
This adds only the requested methods to our class.
There is another way to write ZParenizor. Instead of inheriting from Parenizor, we write a constructor that calls the Parenizor constructor, passing off the result as its own. And instead of adding public methods, the constructor adds privileged methods.
function ZParenizor2(value) { var that = new Parenizor(value); that.toString = function () { if (this.getValue()) { return this.uber('toString'); } return "-0-" }; return that; }
Classical inheritance is about the is-a relationship, and parasitic inheritance is about the was-a-but-now's-a relationship. The constructor has a larger role in the construction of the object. Notice that the uber née super method is still available to the privileged methods.
JavaScript's dynamism allows us to add or replace methods of an existing class. We can call the method method at any time, and all present and future instances of the class will have that method. We can literally extend a class at any time. Inheritance works retroactively. We call this Class Augmentation to avoid confusion with Java's extends, which means something else.
In the static object-oriented languages, if you want an object which is slightly different than another object, you need to define a new class. In JavaScript, you can add methods to individual objects without the need for additional classes. This has enormous power because you can write far fewer classes and the classes you do write can be much simpler. Recall that JavaScript objects are like hashtables. You can add new values at any time. If the value is a function, then it becomes a method.
So in the example above, I didn't need a ZParenizor class at all. I could have simply modified my instance.
myParenizor = new Parenizor(0); myParenizor.toString = function () { if (this.getValue()) { return this.uber('toString'); } return "-0-"; }; myString = myParenizor.toString();
We added a toString method to our myParenizor instance without using any form of inheritance. We can evolve individual instances because the language is class-free.
To make the examples above work, I wrote four sugar methods. First, the method method, which adds an instance method to a class.
Function.prototype.method = function (name, func) { this.prototype[name] = func; return this; };
This adds a public method to the Function.prototype, so all functions get it by Class Augmentation. It takes a name and a function, and adds them to a function's prototype object.
It returns this. When I write a method that doesn't need to return a value, I usually have it return this. It allows for a cascade-style of programming.
Next comes the inherits method, which indicates that one class inherits from another. It should be called after both classes are defined, but before the inheriting class's methods are added.
Function.method('inherits', function (parent) { this.prototype = new parent(); var d = {}, p = this.prototype; this.prototype.constructor = parent; this.method('uber', function uber(name) { if (!(name in d)) { d[name] = 0; } var f, r, t = d[name], v = parent.prototype; if (t) { while (t) { v = v.constructor.prototype; t -= 1; } f = v[name]; } else { f = p[name]; if (f == this[name]) { f = v[name]; } } d[name] += 1; r = f.apply(this, Array.prototype.slice.apply(arguments, [1])); d[name] -= 1; return r; }); return this; });
Again, we augment Function. We make an instance of the parent class and use it as the new prototype. We also correct the constructor field, and we add the uber method to the prototype as well.
The uber method looks for the named method in its own prototype. This is the function to invoke in the case of Parasitic Inheritance or Object Augmentation. If we are doing Classical Inheritance, then we need to find the function in the parent's prototype. The return statement uses the function's apply method to invoke the function, explicitly setting this and passing an array of parameters. The parameters (if any) are obtained from the arguments array. Unfortunately, the arguments array is not a true array, so we have to use apply again to invoke the array slice method.
Finally, the swiss method.
Function.method('swiss', function (parent) { for (var i = 1; i < arguments.length; i += 1) { var name = arguments[i]; this.prototype[name] = parent.prototype[name]; } return this; });
The swiss method loops through the arguments. For each name, it copies a member from the parent's prototype to the new class's prototype.
JavaScript can be used like a classical language, but it also has a level of expressiveness which is quite unique. We have looked at Classical Inheritance, Swiss Inheritance, Parasitic Inheritance, Class Augmentation, and Object Augmentation. This large set of code reuse patterns comes from a language which is considered smaller and simpler than Java.
Classical objects are hard. The only way to add a new member to a hard object is to create a new class. In JavaScript, objects are soft. A new member can be added to a soft object by simple assignment.
Because objects in JavaScript are so flexible, you will want to think differently about class hierarchies. Deep hierarchies are inappropriate. Shallow hierarchies are efficient and expressive.
I have been writing JavaScript
for 14 years now, and I have never once found need to use an uber
function. The super idea is fairly important in the classical
pattern, but it appears to be unnecessary in the prototypal and functional
patterns. I now see my early attempts to support the classical model in
JavaScript as a mistake.