JavaScript Inheritance and

Object Programming

Preview Edition
Preview Edition
For the full book:
Amazon ( http://www.amazon.com/JavaScript-
Inheritance-Object-Programming-
Rinehart/dp/1490463046 )
The CreateSpace eStore (
http://www.createspace.com/4327336, preview
edition discount: MAMMBM3G )
The extended eBook from Software
Developer's Journal (
http://sdjournal.org/javascript-inheritance-and-
object-programming-ebook/ )
JavaScript Inheritance and
Object Programming

Martin Rinehart
Copyright 2013, Martin Rinehart
Ebook published by Software Developer's Journal.
Print book printed on demand (POD) by
CreateSpace, a subsidiary of Amazon. Designed
and written by the author in the beautiful Hudson
Valley, New York, USA.
Versions 2e and 2p completed 9 July, 2013.

All text, code and illustrations by Martin Rinehart,

now a full-time frontend engineer. He can be
contacted through http://www.martinrinehart.com/.

Many thanks to Technical Editors Robert Franzese,

founder of Zen40.com, Eric Schmidt of Microsoft
Corporation and Assaf Shemesh, an expert in
software development and management. Without
their efforts this book would have many more errors.

Dedicated to Brendan Eich, who wrote JavaScript

and gave us object programming.
 JavaScript Inheritance and Object
Programming
Introduction 1
1 Objects 5
Reasons for Objects 6
Class-Based v. Prototypal 7
Objects Up Close 8
Introducing JSWindows 10
2 JavaScript Objects 21
Built-Ins 21
Ex Nihilo Objects 22
Custom Objects 25
Object Prototypes 27
The Prototype Chain 29
Coding a Class 31
3 Object Programming 37
OP Defined 38
Programming with Properties 38
Object Programming Examples 40
OP in the JSWindows Library 42
Coding Object Programs 48
4 Inheritance 53
Classes 54
Class-Based Inheritance 57
Prototypal Inheritance 61
Inheritance v. Composition 62
Coding Composition 66
5 JavaScript Inheritance 69
Cascading init() Methods for Data 70

i
 JS Inheritance and OP

Prototypes for Methods 79

JSWindows Inheritance 85
Coding OOP Inheritance 86
6 Inheritance Alternatives 91
Multiple Inheritance 92
Interfaces 93
Capabilities 94
The Window[_M[_BS]] Problem 96
Mixins 98
Calling Capability Methods 99
Examples 101
Capabilities Summary 113
Coding a Capability 114
7 Designing for JavaScript 115
Use ex nihilo Constantly 116
Use Composition Liberally 117
Use Capabilities Liberally 120
Use Inheritance Conservatively 121
Summary 123
Notes 125

ii
 A Note for the Implementers
JavaScript, or more exactly, the subset of JavaScript Crockford
identifies as The Good Parts, is a beautiful language. It is
small, yet expressive. It's combination of C with functional
programming and object programming gives it extraordinary
depth. In nearly a half century of programming I have used
dozens of languages. Only two of them, JavaScript is one, have
been languages I've loved.
Today there are people working to free JavaScript from the
browser, to further empower JavaScript (FileWriter and WebGL,
to mention personal favorites) and to bring it up to professional
speed. My apologies to the latter group. In many ways, object
programming is the enemy of compiled speed.
So a word of encouragement and advice to our courageous
implementers. First, making JavaScript run at some reasonable
fraction of C's speed is a magnificent goal. More power to you!
(And yes, that's self-serving. You are giving more power to all of
us who write JavaScript. We love it and we thank you for it.)
Second, removing object programming to gain speed cuts out the
heart to save the patient. Object programming is not your enemy,
it is the essence of the language. Look on it as a challenge, as the
Everest of your profession. The view from the top will be
spectacular. Object programming at half the speed of C will be
breathtaking.
Martin Rinehart, 8 June 2013
Hudson Valley, New York
 Website
This book is the sixth knowit at the author's website,
MartinRinehart.com. From the home page, click Knowits and
then JIOP.
The gateway page holds the notes (both footnotes and
bibliography). Hyperlinks are inline in the eBook, online in the
printed book (so the printed text of this book is uninterrupted,
and so you can click on the references).
From the gateway page, sub menus link you to the User Guide
to the JSWindows system, the engineer's reference material and
the source code.
 Introduction
Within the JavaScript community there is a great deal of
misinformation regarding inheritance. Some comes from object-
oriented programmers (C++ and its offspring like Java and C#)
looking for, and not finding, their familiar classes. Some comes from
JavaScript programmers looking for Self-style prototypal
inheritance, which JavaScript's prototypes are not very good at.
In this small volume we begin at the beginning, discussing objects,
their place in software and the benefits we expect from them. We
continue to first define terms like class (three meanings) and
inheritance. With a common terminology we then look at how
class-based and prototypal inheritance can be implemented in
JavaScript. To leap ahead, we will propose ways in which JavaScript
is best served by minimizing the long inheritance chains associated
with object-oriented programming languages.

1
2 Introduction

OOP inheritance is very different from prototypal inheritance. They

are widely misunderstood as two ways of achieving the same result.
Their similarities are more superficial than profound. We have to
understand both to understand JavaScript's hybrid class/prototypal
object model.
Throughout the book we look at our JSWindows sample system.
These are windows, JavaScript style:

These windows can be dragged to new positions, resized, minimized,

maximized and so on. They are programmed in JavaScript but with
OOP inheritance.
JSWindows does not use prototypal inheritance. It's goal, in addition
to being a working window system, was to demonstrate class-based
inheritance in JavaScript without additional library code. It was
intended to demonstrate the use of long inheritance chains without
the use of JavaScript's prototype chain.
Along the way, we also examine JavaScript's object programming
(OP) abilities, abilities that are the true distinction between class-
based OOP and JavaScript's hybrid object model. JSWindows library
code (not inheritance-related) makes extensive use of object
JavaScript Inheritance and Object Programming 3

programming. JavaScript programmers almost all use object

programming, though many are not aware of its profound
importance.
Last, we show how JSWindows uses capabilities, a hybrid of Java's
interfaces and JavaScript's mixins. We show that capabilities give us
the benefits of inheritance without the problems associated with
mapping complex real world relationships to OOP-limited models.
(Java replaced multiple inheritance with interfaces, and we were big
fans.) Object programming enables our capabilities.
Finally, we note that inheritance, specifically OOP inheritance, is a
valuable tool for the advanced JavaScript programmer. (Prototypal
inheritance may also be valuable, for those so inclined. If you
understand that these are two different models, you can use both
together.) Once you understand the importance of object
programming, however, inheritance assumes a far less prominent
part in your systems' architectures.
We make assertions about JavaScript's abilities along the way. The
eBook version has two chapters that may interest those who accept
nothing without proof. The first discusses JavaScript's relationship to
traditional OOP concepts, such as encapsulation and polymorphism.
The second elaborates on the details of the JavaScript constructor
mechanism.
 1 Objects
Objects were a software experiment that worked. They began in
research labs in the late 1970s. They became a new, mainstream
programming paradigm in the 1980s. By the '90s, languages that
were being created (Java, JavaScript, Python, Ruby) were all object-
based. Languages that predated objects were being retrofitted. Today
even 50-year old languages (Basic, Cobol, Fortran) have adopted
objects.
In this book we will be examining two forms of inheritance and
alternatives to inheritance. We need to understand the benefits of
objects to see how these alternatives provide (or not) the benefits of
programming with objects.
We will also be drawing examples from our JSWindows
demonstration system. It is introduced at the end of this chapter.

5
6 1 Objects

Reasons for Objects

There are a lot of reasons to prefer programming with objects. We'll
discuss three here.

Objects Do Methods
First, the syntax has things doing things just as happens outside of
the world of software. The things objects do are small software
programs, functions that the object can perform, commonly called
methods. If you program a dog (object) to speak (method) you
express it directly:
dog.speak(); // says Woof, woof

Many objects can implement the same action (method) with

variations appropriate for each, as Listing 1-1 shows.
Listing 1-1
parrot.speak(); // Polly want a cracker.
kitten.speak(); // mew, mew

If your parrots and kittens use different methods for speaking, object
systems will choose the appropriate method for you. For those who
are academically inclined, the eBook version's Chapter 8 explains
subtype polymorphism, the principle underlying the selection of
object-appropriate methods.

Event-Driven Programming
Second, few programs now do anything except by user direction.
Most programs paint an interface (menus, icons, buttons) and wait
for a user command. This is called event-driven programming.
When the user clicks the save icon, the user's document is written
to disk. Internally, the user action could trigger a very small program:
user_data.save();

Again, it uses the noun.verb() syntax. This brings us to the main

reason for the success of objects.
JavaScript Inheritance and Object Programming 7

Taming Exponential Complexity

As programs grow in size, complexity increases exponentially. We
have all used systems that never seemed to be robust. Fixing this
broke that. If the complexity is ever successfully tamed, it is after
immense expense. Objects help shrink the size and limit the
complexity. That reduces the effort (expense).
Let's think about a simple example, a small non-object program
might require a thousand lines of code. The same job using objects
might require ten methods, each only a hundred lines of code. The
ten small methods will be far cheaper to write and vastly cheaper to
debug.
As we look at JSWindows' code, you will see many features that take
surprisingly little code and live comfortably outside of other parts of
the system.

Class-Based v. Prototypal
Objects come in many guises. We will be using class-based object-
oriented programming (OOP, based on classes) and JavaScript's
hybrid class/prototypal object model.
Object systems go back to Simula and SmallTalk, in the late '70s.
These found the mainstream when Bjarne Stroustrup, at Bell Labs,
wrote a compiler that output programs in plain C from an object-
enhanced form he called C++. (During the '80s, C was the dominant
language for professional programming on non-IBM mainframes and
almost all minicomputers and microcomputers.) C++ was first
available in 1983. By the end of the decade it had become
ubiquitous.
To add objects to C, Stroustrup made some sacrifices. His enhanced
C used class modules (more on these shortly) that were not objects,
but that defined objects. Despite protests from purists, Stroustrup's
limited features were well chosen and programmers were delighted
with its improvements over original C. Other languages, such as Java
(1995) adopted the C++ object model. Programming in C++ and its
progeny is called object-oriented programming or class-based
(aka classical) OOP.
8 1 Objects

Class-based OOP relies on class software that defines the objects

created. The programmer decides what features each group of objects
will need and programs an appropriate class software module that
will create objects with those features.
By contrast, the prototypal object paradigm does not use class
modules. The programmer creates a prototype object and other
objects are then cloned from the prototype. This was an experimental
object model from Self, a language that evolved from SmallTalk.
JavaScript's original author, Brendan Eich, adopted the prototypal
model, partially, in a scripting language for Netscape's then market-
leading web browser, Netscape Navigator. JavaScript, born in 1995,
combined class-based and prototypal forms. As JavaScript is, to date,
the only language available for writing programs that run in all
browsers, it has exploded in popularity as the web has exploded.
Class-based OOP and JavaScript perform similar functions and
provide similar benefits.
Programmers have begun to discover that JavaScript supports almost
complete object programming (defined in Chapter 3) and this is a
major advance over the limitations of the class-based model, quite
separate from and far more important than JavaScript's prototypal
features.
So what are objects?

Objects Up Close
An object is a collection of properties (often a set, but set has a
mathematical meaning we do not want here). Properties are named
values. Values may be, depending on the language, simple values
(boolean, integer, character), composite values built from other
values (arrays, objects) or, in languages capable of functional
programming, functions or other blocks of code. (JavaScript also
implements functional programming borrowed from the Scheme
language, a dialect of Lisp.)
Some authors use the word property to specifically mean what we
call data propertiesproperties that are not methods. We find this
misleading in a language where functions are first-class objects.
JavaScript Inheritance and Object Programming 9

In JavaScript, property names must be strings. In most class-based

OOP languages they must be strings that are limited by the
restrictions imposed on variable names.
Objects also are permitted direct access to a collection of functions
(commonly called instance methods) that are part of the class
software (in class-based languages) or the prototype (in JavaScript).
These functions are separate from, but available to, the objects.
All dogs, in the example above, could access the speak() method:
dog.speak(); // says Woof, woof

Here, speak() is a property of dog objects. In the text, trailing

parentheses indicate that the property is executable code, called a
method. This is one of the two main categories of object properties.

Data Properties
Objects may have data properties. These are often said to describe an
object's state. A dog might have properties such as name, breed
and date-of-birth. Each dog (called an object instance or, in
class-based OOP, an instance of the dog class) will have space
allocated to store each of these properties' values. The key point is
that each instance has its own set of data properties.

Methods (Code Properties)

Unlike data, the methods are stored in the class software, in class-
based OOP, or in the prototype, in JavaScript. (As methods require
storage space, it would be extremely inefficient to store a separate
copy of each method with each instance of the class.)
Methods operate on the data properties of each instance. If each dog
had a message property, a small breed could say, Yap, yap while
a large breed said Woof, woof. If an application's dogs could speak
a bit of English, the programmer might combine fixed values with
the dog's name property to achieve a result like the one shown in
Listing 1-2.
10 1 Objects

Listing 1-2
collie.speak(); // My name is Lassie.

beagle.speak(); // My name is Snoopy.

Next we put objects to use. Our JSWindows system is one example.

Introducing JSWindows
The result you see in Figure 1-1 takes exactly four lines of JavaScript
code. One creates the page title and three create the three windows.
(Is there a law that says windows must be rectangles?)
The gateway to the online portion of this book is
http://www.martinrinehart.com/frontend-
engineering/knowits/op/knowits-op.html. You can use the system
your self (and enjoy the full color) by clicking Using JSWindows
from the gateway page. Let's take a look.
JavaScript Inheritance and Object Programming 11

Windows with Border Radii

Figure 1-1
The shield in the center of Figure 1-1 is, as you can see from the
motto, the frontend engineers' family crest. As you can see from the
buttons on its upper right, it is also a window. You can close,
maximize, restore and minimize with these buttons (right to left).
The title is also a drag handle.
JSWindows is written in JavaScript for applications on all devices
that run HTML4 or 5 and CSS2 or 3. The shapes here are done with
CSS3 border radii.
12 1 Objects

JSWindows Contain JSWindows

Figure 1-2
In Figure 1-2 we've changed the size of the shield, eliminated the
border radii specifications from each window and changed the two
inner windows from Window_M (window, movable) to
Window_M_BS (window, movable with buttons for sizing).

Note that there are, by default, three intermediate sizes (between min
and max), not just the single restored size of some other systems.
We are also using the CSS3 RGBA (transparency) capability for the
#2 window.
JavaScript Inheritance and Object Programming 13

Click Small, Click Large

Figure 1-3
Figure 1-3 is exactly two button clicks away from Figure 1-2. Look
at the sizing buttons. We clicked the large button for one small
window, the small button for the other.
Next we go on to minimizing. Note that we are minimizing windows
within their containers. The shield (the one with the motto) is
contained within SCREEN. The two smaller windows are contained
within the shield.
14 1 Objects

Minimized Windows in a Window

Figure 1-4
Notice that the minbox (Figure 1-4, holding the two minimized
windows) also has a restore (now disabled) and a minimize sizing
button. If you click its minimize button, the minbox is reduced to a
one-pixel-tall box, topped by its own sizing buttons, taking almost no
screen real estate.
But let's return to the windows with the CSS3 border radii. They're
more fun.
JavaScript Inheritance and Object Programming 15

The Original Windows, Almost

Figure 1-5
In Figure 1-5 the original windows are shown, except that the two
smaller ones are still sizable. Here their large buttons have been
clicked.
16 1 Objects

Maximizing in the Shield

Figure 1-6
In Figure 1-6, we clicked the maximize button of the translucent
window. (If you compare with Figure 1-5, you can see that the
Window_M Rules! window is masked.)
Isn't this the way you would have coded maximize? (It's easier
than you might guess. It works inside circles and ovals, as well as
shields, too.)
JSWindows is more fun, and in full color, when you try it online
http://www.martinrinehart.com/frontend-
engineering/knowits/op/knowits-op.html.
JavaScript Inheritance and Object Programming 17

The JSWindows system is programmed with class-based inheritance,

using long inheritance chains typical of class-based languages. (It
does not use OOP-emulating library code, nor the JavaScript
prototype chains typical of library OOP implementations in
JavaScript.) We return to this system frequently as we discuss the
implementation of inheritance in JavaScript.
In Chapter 2 we back up, beginning at the beginning.
 2 JavaScript Objects
We begin, in this chapter, by looking at JavaScript objects, how they
are created and how their prototypes work.

Built-Ins
Programmers new to JavaScript quickly learn that there are common
needs that are met by built-in JavaScript objects. For example,
common mathematical operations are performed by the Math
object's methods.

Math
You want to round a number to the nearest integer. Listing 2-1 shows
an example.

19
20 2 JavaScript Objects

Listing 2-1
var width =
content_width + padding + border_width;

div.style.width = Math.round(width) + 'px';

The Math object is really just a collection of library functions, all

related to mathematical operations. You should not manipulate it as
an object.

Date
The Date constructor returns real objects, time stamps that include
both date and time to the nearest millisecond. The Date also
provides the methods you want to manipulate dates (to set a time for
an appointment next week, for example). The bit of code in Listing
2-2 might be included when you were comparing the running times
of alternative functions.
Listing 2-2
var start = new Date();
var result = sample_long_function();
elapsed = start new Date();

alert('Time was: ' + elapsed);

The programmer is teased by these built-in objects. Hopefully, you

decide that you want to go further and create your own objects.

Ex Nihilo Objects
In JavaScript, you can create objects from a constructor (from the
equivalent of class software as in C++) or from a prototype (as in
Self). JavaScript combines both. It also lets you create an object ex
nihilo (from nothing).
There are two direct ways to create a JavaScript object from nothing.
You can use the Object constructor or an object literal.
JavaScript Inheritance and Object Programming 21

The Object Constructor

In JavaScript there is an object constructor, Object.
my_object = new Object();

As the name suggests, this creates an object. The initial capital letter
tells you that the Object() method was intended for use after the
new operator as a constructor. When you create an object this way
you can assign properties, as Listing 2-3 shows:
Listing 2-3
my_object.size = 'large';
my_object.color = 'blue';

Veterans of static, class-based languages (such as C++, Java and

many others) find this capability novel. They are accustomed to
names and types of object properties being fixed at compile time.
Dynamic properties (as in Python) remove this restriction. (Static
properties give optimal performance. Just-in-time compilationa
technique pioneered in Selfminimizes any speed difference,
however.)
The prototype (discussed below) of objects created from nothing is
Object.prototype.

Object Literals
Listing 2-4 shows another way the above ex nihilo object could be
created.
Listing 2-4
my_object = {
size: 'large',
color: 'blue'
};

As in C, use of whitespace is for readability. The following line does

the same job.
my_object={size:'large',color:'blue'};
22 2 JavaScript Objects

JavaScript object literals also create instances of the Object family

(as if you used the Object constructor) and their prototype is, again,
Object.prototype.

The object literal notation (documented by Crockford as JSON,

JavaScript Object Notation) is still the preferred method for creating
objects from nothing. The Object constructor has been assigned
powerful new capabilities by the most recent versions of the
JavaScript standard (ECMAScript). You will want to consider these
new capabilities when they become universally available.
Objects created from the Object constructor and objects created
from object literals all share Object.prototype as their prototype.
This gives them all, for example, a toString() method that reports
[object Object], telling you that the type is object, its constructor
is Object. (This is almost never helpful. Creating custom objects
lets you supply useful toString() methods. If you alert() or
console.log() an object you will be looking at its toString()
version. You usually know that it is an object.)

More ex nihilo Objects

Functions and arrays are also objects in JavaScript. Whenever you
create a function (named or anonymous, with the function
keyword or the Function constructor), you create an object. When
you create an array (via the Array constructor or the array literal
notation) you are also creating objects.
This is another way of creating an ex nihilo object:
arr = ['dog', 'cat', 'mouse'];

From now on, however, we will reserve the adjective ex nihilo for
objects that are neither functions nor arrays.

Custom Objects
To create your own objects, ones that will have their own prototypes
(the next section), you write your own constructor functions, just as
you do in class-based languages.
JavaScript Inheritance and Object Programming 23

In JavaScript you can begin in much the same way that OOP
programmers begin, by writing a constructor function.

Constructors
By convention, the name of a constructor begins with a capital letter.
The constructor function assigns properties, commonly data
properties, and adds initial values. Listing 2-5 shows a simple
example.
Listing 2-5
function Dog(name, breed) {
var new_dog = this; // the 'new' object

new_dog.name = name;
new_dog.breed = breed;
}

If you chose the parameter names carefully, chances are those same
names are also good names for the object's properties. (Study the
example to convince yourself that this does not lead to ambiguities.)
The pattern in Listing 2-5, where the value of the parameter name
becomes the value of the name property, is very common.

Assigning Initial Property Values

Using your Dog constructor, your Dog objects can start their lives
with two properties, name and breed, and each Dog will have values
for those properties. Listing 2-6 shows this constructor being used to
create Dog objects.
Listing 2-6
var snoopy = new Dog('Snoopy', 'Beagle');
var lassie = new Dog('Lassie', 'Collie');

Familiar object.property notation can be used to access these

properties' values.
alert(snoopy.breed); // 'Beagle'
24 2 JavaScript Objects

Non-Trivial Property Values

In addition to assigning arguments passed into the constructor, the
constructor can create any properties that you wish. This could
include arbitrarily complex computations. In class-based OOP
programming, the object's property names and types are specified by
the class software at compile time. In JavaScript, the constructor
function serves this requirement. (In Self, the properties, called
slots, could be added whenever the programmer chose. As in Self,
in JavaScript, and other languages such as Python, the properties are
dynamic and may be changed, added or deleted, during execution. In
traditional class-based OOP languages, only the property values may
be changed.)
In JavaScript, the values you assign to properties can include any
JavaScript values, including other objects. You could assign
functions as property values, but this would seldom be useful. If your
constructor assigns functions, it will be assigning the same code to
every object it creates. You normally want the function code to be
part of the object's prototype (as it is part of the class software, in
class-based OOP languages).

Methods as Prototype Values

Methods (functions) can be assigned as object properties, but that
would be wasteful. The method code can be assigned to the
prototype where it can be accessed by all objects that are members of
the family (generated from the same constructor).
A data property could also be assigned to the prototype, but that, too,
would be wasteful. Assume you have several types of animals and
they have diets such as carnivore and herbivore. Putting this data
value into the prototype would make it available to all animals of
each type, as Listing 2-7 shows.
Listing 2-7
Dog.prototype.diet = 'carnivore';
alert(lassie.diet); // She's a 'carnivore'.

The lookup, however, is not needed. In lassie.diet, JavaScript

looks to the lassie object for a property named diet. It does not
JavaScript Inheritance and Object Programming 25

find it. Then it looks to the prototype, where it finds the value it
seeks. If the property value is the same for all family members, the
property can be made a property of the constructor (shown in Listing
2-8), saving the wasted lookup.
Listing 2-8
Dog.diet = 'carnivore'; // not Dog.prototype
alert(Dog.diet); // same for Lassie, Snoopy...

The JavaScript usage of the constructor is almost identical to the

OOP programmer's use of the class software. Instance data is
assigned to each object. Methods, and data that applies to the whole
family, are part of the class software. (Dog.diet is 'carnivore';
lassie.diet is undefined.) Instance methods are part of the
prototype for the convenience of using object.method() calls, for
example, lassie.speak().
Now we take a longer look at object prototypes, risking repetition in
favor of being absolutely clear.

Object Prototypes
When JavaScript sees an object property name, such as color in
object.color it looks to the object for a property of that name. If
the name does not exist it looks to the object's prototype for a
property of that name. (The prototype is another object.)
Assume that you created singers, as Listing 2-9 shows.
Listing 2-9
function Singer() {}; // no properties
var patty = new Singer(),
maxene = new Singer(),
laverne = new Singer();

patty.sing(); // error, undefined method

We have no sing() method, so these girls don't know how to sing.

In Listing 2-10 we teach Patty, individually.
26 2 JavaScript Objects

Listing 2-10
patty.sing = function () {
alert('...boogie woogie bugle boy...');
}
patty.sing(); //...boogie woogie bugle boy...
Patty can sing, but her sisters cannot. We should have put this
method in the prototype, as in Listing 2-11.
Listing 2-11
Singer.prototype.sing = function () {
alert('...blows eight-to-the-bar...');
};
maxene.sing(); // she sings!
laverne.sing(); // she sings, too!

How did we know that Singer.prototype was the prototype for

our singing sisters? The prototype for any object you create from a
constructor is the property named prototype of the constructor
function from which the object was instantiated. If Singer is the
constructor, then Singer.prototype is the prototype for Singer
family objects. (The eBook's Chapter 9 fills in all the details of the
underlying mechanism by which Singer.prototype becomes the
prototype for all Singer family instances.)
In the class-based model, instance methods are written in the class
software. In Java you would have created a sing() method in the
Singer.java module. In JavaScript you assign the sing() method
to the Singer function's prototype property. The result is the
same. The objects created from the constructor can sing().
(The Andrews SistersPatty, Maxene and LaVernehad their biggest hit, of many,
with Boogie Woogie Bugle Boy in 1941. http://www.dump.com/andrewssisters/,
2:21)

The Prototype Chain

Now we repeat the paragraph that started the previous section:
JavaScript Inheritance and Object Programming 27

When JavaScript sees an object property name, such as color in

object.color it looks to the object for a property of that name.
If the name does not exist it looks to the object's prototype for a
property of that name. (The prototype is another object.)

What happens if JavaScript does not find the name in the prototype?
The prototype is another object. Simply read the paragraph again.
JavaScript will look in the prototype's prototype. In JavaScript, all
objects you create are either created from the Object constructor,
and therefore Object.prototype is their prototype, or their
constructor's prototype is an object created from Object, or their
constructor's constructor is an object created from Object, and so
on.
Searching prototypes stops at Object.prototype. This is called
the prototype chain. Ultimately, Object.prototype is the
prototype of every object you create, directly or indirectly.
Bear in mind the following two important facts as you consider the
prototype chain. First, there are large families of objects (not ones
that you create) that may not have prototypes. JavaScript does not
know how to do input or output. It depends on a host environment
for all I/O. Most commonly, JavaScript runs in a browser that
provides the host environment through objects. These are called
host objects and unless the browser's authors were meticulous
(most weren't) the host objects do not have prototypes. Host objects
seldom have prototype chains.
Second, you can see how the prototype chain could be used to
implement inheritance. If you could connect your Extend family
objects so that properties not found in Extend.prototype would
be sought in Base.prototype, you would have inheritance. What
does not follow is that this would be a good way to achieve
inheritance. There are other ways that are preferable.
Many authors suggest you use the prototype chain to achieve
prototypal inheritance. In a language such as Self, creating one
object from a prototype object is referred to as inheritance. In
JavaScript, creating objects ex nihilo involves no inheritance.
Creating objects via a call to a constructor is analogous to creating
objects in class-based OOP. (JavaScript's ex nihilo objects inherit
from Object.prototype, so they aren't really ex nihilo.)
28 2 JavaScript Objects

Is the term inheritance appropriate for creating objects from a

constructor? In OOP, creating base objects from the Base
constructor is not called inheritance. That term refers specifically to
the creation of objects from a family such as Extend that inherits
from a family such as Base. In fact, you can code JavaScript as if the
prototype were part of an OOP family. JSWindows is coded this way.
Object methods can be found in their prototypes but properties not in
the prototype simply do not exist. (Assume that there is no prototype
chain. A property not found in the prototype does not exist.)
The only extensive use of the prototype chain is made by
programmers who wish to provide fundamental additions to a
JavaScript built-in class. For example, you might wish to add a
capability to all Array objects. You could add additional method(s)
to Array.prototype.
Most experienced JavaScript programmers are opposed to this idea.
What happens if one library adds extensions to Array.prototype
and another also adds extensions to Array.prototype? We have
lots of new capabilities and we're all happy until one library's author
inadvertently (and inevitably) picks the same name chosen by
another. Then we are in trouble. Whichever code is loaded last gets
ownership of that duplicated name. The library that depended on the
first-loaded (and therefore overwritten) version of that name
becomes buggy. The history of JavaScript is replete with once
reliable libraries that suddenly became buggy because of these
naming conflicts.
It follows that if the prototype chain is not used for modifications to
JavaScript basics and it is not used for inheritance, the prototype
chain could be removed from JavaScript. It could and JSWindows,
for one, would not miss it. (JSWindows, as we will see, uses
extensive inheritance, none of it using the prototype chain.)
In you want to experiment with Self-like object prototypes, in
JavaScript: The Good Parts Crockford advocates them and provides
a function to create true prototypal inheritance. (It only takes a half
dozen lines of JavaScript.)
Now that we have introduced JavaScript objects, we can look at
JavaScript's object programming (OP) capability which we will use
constantly. Before we go on, let's put this much to use.
JavaScript Inheritance and Object Programming 29

Coding a Class
Ready to start writing code? In this chapter and the next four we
have guided programming tutorials. The road map is this:
Create a useful Box class (here in chapter 2).
Add a styles configuration object using object
programming (3).
Create a Borders class and combine it with Box objects
using composition (4).
Create a Button class that inherits from Box (5).
Add a Maskable capability and an implements() method
that adds this capability to a Box (6).

A) Create Your Template

If you already have a template, or your editor/IDE can create one,
test it against these requirements. Upgrade as needed. If need be,
create a new template.
A minimal HTML template will include at least:
A doctype
File path and name comments
<html>, <head> and <body> tags
<title> in the head
<script> at the end of the body
use strict; in the script

Compare your template to the one at

http://www.martinrinehart.com/frontend-engineering/knowits/op/op-
tutorial/op-tut-2/op-tut-2a.html. Make sure that yours is better than
ours. (But don't overload it with bits you might not need.)

B) Add a Class Skeleton

Turn your template into an HTML file. We'll be working
continuously on this one file for all the tutorials in this book. Place it
in some convenient folder.
30 2 JavaScript Objects

Now we start to write JavaScript.

Every class will need a constructor, an init() method (property of
the constructor, not in the prototype) and a toString() method (in
the prototype). Add these and compare against our sample at
http://www.martinrinehart.com/frontend-engineering/knowits/op/op-
tutorial/op-tut-2/op-tut-2b.html.

C) Add Parameters
Now we need to think about the actual work for our class. That
means deciding on the parameters for our constructor. Think
carefully here, but don't get stuck. (Think like a backpacker. When
in doubt, leave it out.)
Our sample includes a
parent (document.body or another Box)
id (unique, for the DOM element and for our own use)
pos_size (left,top, width,height array)
color (background)

The init() method should copy these parameters into properties of

the new_box. Writing these with the toString() method will let
you test your work this far. (A simple test is in our sample.)
Our sample is at http://www.martinrinehart.com/frontend-
engineering/knowits/op/op-tutorial/op-tut-2/op-tut-2c.html. You'll be
able to find our samples (Next on the right, bottom or the letter on
the menu) without any more links, we hope.

D) Finish Box.init(), Add a Mainline

To finish init(), document.createElement() a div as a
Box.delem (DOM element) property. Give it an ID (with your id
property). Add to the delem's style property: a background color
and styles for your position and size values. Don't forget to add a
position style (absolute or relative). And don't forget to
appendChild() it to the parent element.
JavaScript Inheritance and Object Programming 31

Now convert that alert() message to an actual mainline, with

sensible values in the pos_size property. Run. Fix typos. Repeat
until you have a Box on the screen.
It's good to see your work on the screen. This is always the point
where we start to smile. Test out your pos_size array with different
positions and sizes. Convince yourself that your color works.
(Create more than one Box as you test.)

E) Add Styles
Next chapter we'll do a little object programming, using a styles
configuration object. For now, we just add styles to the mainline. We
use textAlign, fontSize and padding. Go ahead and add any
that you like. (We'll do a Borders class in Chapter 4, so you might
skip border styles for now.)

F) Finish the Mainline

And finally, finish your work with some content in the Box. Extra
credit if you don't slavishly follow our example.
You've now coded a constructor and an init(), so you can inherit
from it easily. And you've added an instance method ( toString())
to the prototype. Not quite ready to graduate from the School of
JavaScript OOP, but well on your way.
Now, on to object programming. It's easier to do than to say.
 3 Object Programming
We are now ready to consider object programming (OP), the
fundamental difference between JavaScript and OOP languages.
JSWindows makes continuous use of object programming. (An early
attempt to write JSWindows without OP was abandoned. It may have
been possible, but the extra effort was not in our time budget.)
In class-based object-oriented programming the objects are defined
before they are created. Once an object exists, you can execute its
methods or assign values to its data properties. But you cannot add or
delete properties or change methods, for examples. Object
programming gives you the ability to do these, and more.

OP Defined
The OOP model combines object instances, sets of name/value pairs,
with class software. The class software provides a routine to create

33
34 3 Object Programming

instances, to store instance methods and it provides other services.

This is a subset of object programming, which includes these
abilities.
An object programming (OP) system allows the creation,
modification and disposal of objects during program
execution.
Objects are collections of properties. Properties are
name/value pairs.
"Modification" of objects means the ability to create, modify
and delete properties. Modifying properties includes
modifying names and/or values.
An OP system also provides for services provided by OOP
classes, such as storing data and code at the class software
level (to avoid duplicating either in each instance object).
Note that by this definition, the JavaScript OP model is incomplete
as it does not allow direct modification of property names. (It takes a
three-line utility function to modify a property name. A direct
approach to modifying the name would be preferable, but you can
live without it.)

Programming with Properties

JavaScript leads programmers into object programming, almost
without their being aware of it. It's two ways of identifying object
properties play a major role.

Dot Notation
In the familiar object.property notation, a period separates the
object reference from the property name. The latter is a constant in
source code.
var x = object.prop_name;
JavaScript Inheritance and Object Programming 35

In the above, object is the name of an object reference.

prop_name is the name of a property of that object (directly, or via
its prototype or prototype chain).
This is the common OOP notation, as well. Few OOP languages have
the equivalent of JavaScript's subscript notation.

Subscript Notation
Subscript notation allows the use of variable property names. First,
with a constant, the above example is repeated here.
var x = object['prop_name'];

Listing 3-1 shows an example of an expression used to select a

property.
Listing 3-1
var e = func_returning_string(args);
var x = object[e];

Listing 3-2 shows both notations being used to create a new property
within an object. Constant and variable property names are shown.
Listing 3-2
Object.new_prop = value;
Object['new_prop'] = value;

var e = func_returning_string(args);
object[e] = value;

The pair of features above provides surprising power and grace when
programming.

Object Programming Examples

Object Sum
JSWindows uses styles objects to hold lists of CSS styles. These
are the JavaScript versions of CSS declaration lists. Each name/value
36 3 Object Programming

pair in the object corresponds to a CSS property name and value.

This is a styles object:
{borderWidth: '8px', borderColor: '#a0a0ff'}

As styles become known, they are added to an object's styles

object. We often want to add the properties of a new object to an
existing object in the process. Listing 3-3 shows this situation in
pseudocode.
Listing 3-3
var bstyles = borders.get_styles();
all_styles += bstyles; // pseudocode

The += operator does not apply to objects, but we can write a sum
function to do the same job. We want the styles in the incoming
bstyles object to add new properties to the all_styles object. If
properties of the same name already exist in all_styles, we want
to replace their values. As Listing 3-4 shows, this is more trouble to
explain than to code.
Listing 3-4
sum = function (old_object, new_object) {
var ret = {};

for (var prop in old_object) {

ret[prop] = old_object[prop]; }
for (prop in new_object) {
ret[prop] = new_object[prop]; }

return ret;
} // end: sum()

This function creates a new, empty object. Then the first for/in loop
copies the old object's property names and values into the new
object. The second for/in loop copies the new property names and
values into the new object. In the process it will create new
properties, if required, or override existing properties where there is
a name conflict.
If your project wants a sum() function where the values in the old
object are preserved (not overridden by values in the new object) it is
JavaScript Inheritance and Object Programming 37

simple to modify the above. As you loop through the properties of

new_object, just check:

if (old_object[prop] === undefined) ...

Do not overwrite the existing property if it is already defined.

OP for Inheriting Prototypes

We also use the sum() function to copy prototypes, underneath the
extends() function. The latter provides a meaningful name and
saves typing, two reasons worthy of our support. Listing 3-5 shows
extends().
Listing 3-5
extends = function (extend, base) {
extend.prototype = sum(
extend.prototype,
base.prototype);
}

OP in the JSWindows Library

For those who have not totally understood the idea (and for those
who are asking, Is it really that simple?) we provide additional
examples of object programming from the JSWindows library.

DOM related
The library functions are divided into a DOM related group, for
dealing with the browser's host environment, and a Utility group,
for everything else. Almost 80% of both make use of some form of
object programming. Listing 3-6 shows the function that deletes a
single DOM element. This would be simpler if one browser's bugs
did not make it necessary to assign null to the deleted reference.
38 3 Object Programming

Listing 3-6
delete_delem = function (delem) {

while (delem.firstChild) {
delete_delem(delem.firstChild);
}
delem.parentNode.removeChild(delem);
delem = null; // Some MSIE needs this.

} // end: delete_delem()

The delete_delem() function calls itself recursively to remove

children from the delem (DOM element). When the children are
gone, it removes the selected element from its parent. As JavaScript's
main use is for DOM manipulation, it shows how OP is an integral
part of its job.
For a second example, Listing 3-7 shows the library function that
attaches an event listener to a DOM element. As with so much
DOM-related work, one of its jobs is to smooth over differences
between browsers. As with so many of these differences, Internet
Explorer, especially older versions, is the problem.
JavaScript Inheritance and Object Programming 39

Listing 3-7
/** Add an event listener. */
listen_for = function (
wobj, event_name, func) {
var delem = wobj.delem;

if (delem.addEventListener !==undefined) {
delem.addEventListener(
event_name, func, false); }
else if (delem.attachEvent !==undefined) {
delem.attachEvent(
'on' + event_name, func); }
// IE before 9
else { delem['on' + event_name] = func; }
// old school!

} // end: listen_for()

Again we are working with object properties, in this case, event

listening methods. Here we use both dot notation and subscript
notation as we fall back to progressively older ways of adding the
event listener.
An interesting feature of this library function is that the Wobj
reference (Window object, the root of our family hierarchy) is passed
as an explicit parameter, Fortran style. We prefer to call these
functions in the object style:
wobj.listen_for(event_name, func);

Listing 3-8 shows the method in Wobj.prototype that lets us use our
preferred style.
Listing 3-8
Wobj.prototype.listen_for = function (
event_name, listen_func) {

listen_for(this, event_name,
listen_func);
}

When you call object.method(), the object reference on the left

is converted to the this parameter within the method. We simply
40 3 Object Programming

put this back into the Fortran-style parameter list, explicitly. That
gives us object-style method calling for our Fortran-style library
functions.
Our third, and final, example from the DOM-related library functions
removes an event listener that had been added by the old-fashioned
method:
element.onclick = click_func;

Listing 3-9 shows the listener removing function.

Listing 3-9
stop_listening_on = function (wobj, type) {
var delem = wobj.delem;

delem['on' + type] = undefined;

}

JavaScript's subscript notation makes this job simple.

Utility
We have been big fans of object programming for a long time so it
came as no surprise that most of our DOM-related utilities
manipulated object's properties. After all, the DOM is an object tree.
What we were surprised to find was how many of our utility (non-
DOM) functions also used OP. We'll start with a simple example.
Whenever we write a constructor, we also write a toString()
method. It seems you always want to have a readable version of an
object as you are developing. But what about your ex nihilo objects?
The default toString() (from Object.prototype) reports that
you have [object Object] (an object created by the Object
constructor). This is almost never helpful. Listing 3-10 shows a
simple utility that creates a readable version of an object that is
frequently helpful.
JavaScript Inheritance and Object Programming 41

Listing 3-10
o2s = function (obj) {
var ret = [];
for (var pname in obj) {
var prop = obj[pname];
if (typeof prop !== 'function') {
ret.push(pname + ': ' + prop);
}
}
return 'object{' + ret.join(',') + '}';
}

This loops through the properties by their names (in pname) and, if
they are not functions, pushes them onto an array of property
name/value pairs, as strings. That array is used as the center of the
returned string.
Listing 3-11 shows an example of object programming applied to
arrays. (Arrays are objects, in JavaScript.) It removes an element
from an array, shortening the array by one (ensuring that there are no
undefined elements created).
Listing 3-11
remove_element = function(arr, element) {

var index = find_index(arr, element);

if (index === -1) { return; }

ret = [];
for (var i in arr) {
if (i !== index) { ret.push( arr[i]; )
return ret;

} // end: remove_element()

One of the most common mistakes in any object programming is to

assign a second reference to an object when a second object is
needed.
var not_really_second = first;
42 3 Object Programming

When not_really_second is changed, the same change appears

in first as they are both references to a single object. Making a
shallow copy of first is normally correct.
var really_second = shallow_copy(first);

Listing 3-12 shows our shallow_copy() function.

Listing 3-12
shallow_copy = function (obj) {
var ret;

if (obj instanceof Array) { ret = []; }

else if (obj instanceof Function) {
ret = obj; }
else { ret = {}; }

for (var name in obj) {

ret[name] = obj[name]; }

return ret;
} // end: shallow_copy()

The code that copies arrays and non-array objects is identical except
that the return value is initialized differently. The calling code does
not care. Note that using a for/in loop to copy the array correctly
handles sparse arrays and arrays that have been modified via
splice() calls.

For those new to shallow and deep copying, a shallow copy means
that values that are references to other objects are copied. A deep
copy would duplicate the objects and arrays referenced. During a
shallow copy, statements such as the following are executed:
copy.prop_name = original.prop_name;

This creates a second property with the same name and value as the
first, but it is totally separate. After the assignment, there is no
connection between the two. It's as if we had lassie.breed =
'Collie' and snoopy.breed = 'Beagle'.

If the code subsequently assigns to the copy, it does not impact the
original.
JavaScript Inheritance and Object Programming 43

copy.prop_name = some_other_object;

The above statement replaces the value of copy.prop_name with

another object reference. It has no effect on the value of
original.prop_name.

We are now equipped to consider inheritance in JavaScript, both

class-based and prototypal. Before we start on code, we will take a
close look at what is really meant by inheritance when
programming with objects. But first, let's do a little object
programming. You will be an ace in a few minutes.

Coding Object Programs

We'll be working with objects. To test our work, we'll want to take a
look at our objects. If we use an object where it will be coerced to a
string (such as in an alert() or console.log()) JavaScript may
tell us that it's an [object Object]. (An object made from the
Object costructor, as every ex nihilo object is.) This is seldom
helpful. We want to look at the properties of the object. So we begin
with a simple, hard-working utility to tell us more.

A) Object to String
We'll use an o2s() function (object to string) constantly as we
develop. It's job is to take an object, any object, and list its
properties. We want the name and value of each property. (You'll
want a smarter o2s() if you need to show function objects, as their
valuesthe function source codetend to be very long.)
The for (name in object) loop iterates through any object,
returning each property's name. Try using it to write an o2s()
function. Test it with object literals:
alert( o2s({breed:'Beagle', name:'Snoopy'}) );

Don't fuss overly much with the beauty of the output. It's only for use
during development. And it's only for use with objects that don't have
a nice toString() method, which can do a much better job.
44 3 Object Programming

Our version is shown at http://www.martinrinehart.com/frontend-

engineering/knowits/op/op-tutorial/op-tut-3/op-tut-3a.html.

B) A get_ps_styles() Function
Next, we'll want to put together our Box's styles in the init()
method, adding styles objects as we go along. Right now,
however, the body of Box.init() is rather crowded with the dull
work of turning a pos_size array into CSS style specifications.
We've moved this into an inner function whose job is to take the
pos_size array and return the CSS we need. (A value of 100 for
pos_size[0]-leftbecomes left: '100px'.)

Take a peek at ours, if you like, and then write your own. Test it with
an alert() or console.log() inside the Box.init() method.

C) A Library Function to sum() Obects

Our new Box.init() will do its work by the sum() library
function. You want to write and test your sum() carefully, as this one
will definitely be at the heart of your production code. Don't forget
any of these:
sum( {}, {foo:'bar'} )
sum( {foo:'bar'}, {} )
sum( {foo:'bar'}, {foo:23} )

You can look in the code (3, C) if you haven't a clue where to start.
(And you can look back a few pages in this chapter, for a second
opinion.) You'll find that sum() is hard-working, but that doesn't
mean it's hard to code.
Note that the sum operation is not commutative. sum(a, b) is not
equal to the sum(b, a) if a and b have like-named properties. In
the third example, above, the value of foo will be 23 after the sum,
as properties of the second object override like-named properties of
the first object. This becomes important in styles objects as we
want styles we explicitly define to override default styles.
JavaScript Inheritance and Object Programming 45

D) Putting sum() to Use

Begin by replacing the color parameter with a styles parameter.
(Three places: in the parameter list in the constructor, in the
Box.init() parameters and in the argument list in the constructor's
call to Box.init().) Styles will be a configuration object you can
use to specify any CSS styles you like. In your mainline, replace the
color specification with an object, like {backgroundColor:
'#c0e0ff'}. Don't forget to replace the line new_box.color =
with new_box.styles = .
Replace the part of Box.init() that assigns values to your delem's
styles object. This takes more thinking than code. Ensure that the
specific styles from the styles configuration object override any
other styles. (Do you want to look at our version first? That's fine.
But write your own! This is a very fine example of a little bit of OP
making short work of a hard task.)
Our new core of the Box.init() method is woefully lacking in
comments. That's your job. Feel free to copy our code but add
enough comments so that a maintenance programmer will be able to
make sense of it.
In the end, test by adding to the styles configuration object. Any style
explicitly specified in styles should override anything else. For
example, you should be able to right align text with appropriate
styles.

E) Teach toString() About Styles

We've laid the groundwork and made this one easy. The
toString() method should now be able to show the styles you
include in your styles object. You've already got the o2s()
method that will make this easy.

F) A Mainline with Styles

As a next-to-last step, replace those explicit assignments from the
work you did in Chapter 2 with assignments to a styles
configuration object. (This has very little real value with just one
46 3 Object Programming

box. Picture dozens of boxes, each using one of a handful of different

styles objects. That will give this approach lots of value.)
At this point you should have a system that relies on styles
configuration objects and you can happily forget the fact that
someplace, deep inside, these are assigned to a delem.style
object.

G) An alert() to Test
Finally, if you alert() or console.log() your boxes, you'll see
toString()s showing readable styles objects.

Now we can move on to inheritance.

End,
Preview Edition

For the full book:

Amazon ( http://www.amazon.com/JavaScript-
Inheritance-Object-Programming-
Rinehart/dp/1490463046 )
The CreateSpace eStore (
http://www.createspace.com/4327336, preview
edition discount: MAMMBM3G )
The extended eBook from Software
Developer's Journal (
http://sdjournal.org/javascript-inheritance-and-
object-programming-ebook/ )
 Notes
The eBook's Chapters 8 and 9 disect OOP object principles
(polymorphism, encapsulation and so on) and the more obscure
details of the new operator/JavaScript constructor pair,
respectively. The eBook is available from
http://sdjournal.org/javascript-inheritance-and-object-
programming-ebook/.
References, as in a bibliography, should be online, not in print.
(At least until you can place your mouse on a printed page and
click, online will be preferred.)
The gateway to the online portion of this book is
http://www.martinrinehart.com/frontend-
engineering/knowits/op/knowits-op.html.
The gateway page provides links to references.

Defined Terms
Often the most controversial portion of an analysis is the
definitions. We have attempted to use terms that have (more or
less) common definitions and avoid terms that are ambiguous
(e.g., polymorphism). Wikipedia is particularly valuable in
having a community editing process that often achieves
consensus on definitions. We also check other sources, however.
Our defined terms and their references include:
Object-oriented programming Oop-W, Oop-1-5
Prototype-based programming Prtp-W, Prtp-1-5
Classes Clss-W, Clss-1-6
Instances (objects) Inst-W, Inst-1-4
Methods Mthd-W, Mthd-1-4
Inheritance Inhr-W, Inhr-1-6
Composition Cmp-W, Cmp-1-4
(A second dash, as in Oop-1-5, denotes the full range: Oop-1,
Oop-2, , Oop-5.)

Support for Selected Statements

Today even 50-year old languages (Basic, Cobol, Fortran) have
adopted objects. Oop-3, Inhr-4, Clss-6, Mthd-4
Other languages, such as Java (1995) adopted the C++ object
model. Clss-1-6, particularly Clss-2
By contrast, the prototypal object paradigm does not use
classes. The programmer creates a prototype object and other
objects are then copied from the prototype. Prtp1-4
An object is a collection of properties (often a set, but set has
a mathematical meaning we do not want here). Properties are
named values. Inst-1-4
Objects also are permitted direct access to a collection of
functions (commonly called instance methods) that are part of
the class (in class-based languages) or the prototype (in
JavaScript). Inst-3, Mthd-1-4, Prtp-1, Prtp-2, Prtp-4
In OOP, a class is the software that creates and supports a set
of objects, including the constructor and methods that
instances of the class can perform. A class may also have
methods and data of its own (class statics, in Java). Clss-
1-6

'In class-based OOP, when Bo is empty we say that E extends

B, or E inherits from B.' Inhr-2-6.

[In prototypal inheritance] Objects inherit directly from each

other. The base object is called the prototype of the inheriting
object. Prtp-1-4
If an object includes another type of object as a property, it is
using composition. Inhr-1-2, but see Inhr-3 and Inhr-4 for
qualifiers.
[JavaScript] also lets you create an object ex nihilo (from
nothing). Prtp-2
We are not endorsing inheritance-based architecture; Inhr-1
'We extend interfaces to capabilities which borrow from and
extend both Java's interfaces and JavaScript's mixins.' Trts-W,