Draft 17.

14, 9 December 07 20:19 (Zürich)

1
The industry of pure ideas

1.1 THEIR MACHINES AND OURS

Engineers design and build machines. A car is a machine for traveling; an

electronic circuit is a machine for transforming signals; a bridge is a machine
for crossing a river. Programmers — “software engineers” — design and build
machines too. We call our machines programs or systems.

There’s a difference between our machines and theirs. If you drop one of
their machines, it will hurt your feet. Ours won’t.

Programs are immaterial. This makes them closer, in some respects, to a

mathematician’s theorems or a philosopher’s proposition than to an airplane
or a vacuum cleaner. And yet, unlike theorems and propositions, they are
engineering devices: you can operate a program, like you operate vacuum
cleaners or planes, and get results.

Since one cannot operate a pure idea you will need some tangible,
material support to operate programs or, using the more common terms, to run
or execute them. That support is another machine: a computer. Computers and
related devices are called hardware, indicating that — although they’re
getting ever lighter — computers are the kind of machine that will hurt your
feet. Programs and all that relates to them are by contrast called software, a
word made up in the 1950s when programs emerged as topic of interest.

Here is how things work. You dream up a machine, big or small, and
describe your dream in the form of a program. The program can then be fed
into a computer for execution. The computer by itself is a general-purpose
machine, but when equipped with your program it becomes a specialized
machine, a material realization of the immaterial machine that you defined
through your program.
6 THE INDUSTRY OF PURE IDEAS §1.1

The person who writes the program — “you” in the previous paragraph
— is predictably called a programmer. Others, whom we call users, can then
run your program on your computer, or theirs.

From idea to
A writes a results
Program

which a
programmer
User

runs on a

Computer.

If you have ever used a computer, you’ve run some program, for example to
browse the Web or play a DVD, so you already are a user. This book should
help you make it to the next step: programmer.

Cynics in the software industry pronounce “user” as “loser”. It’s one of the goals
of this book that users of your programs will pronounce themselves winners.

The immaterial nature of the machines we build is part of what makes

programming so fascinating. Given a powerful enough computer you can
define any machine you want, whose operation will require billions upon
billions of individual steps, and the computer will run it for you. You do not
need wood or clay or iron or a hammer or anything that could wear you out
carrying it up the stairs, burn you, or damage your clothes. State what you
want, and you will receive it. The only limit is your imagination.

Well, all right, it is one of two limits; we don’t like to mention the other
in genteel company, but you will likely encounter it before long; it is your own
fallibility. Nothing personal: if you are like the rest of us, you make mistakes.
Lots of mistakes. In ordinary life they are not all harmful, as most human
activities are remarkably error-tolerant. You can press your fork a little too
intensely, drink your water a little too fast, push the accelerator a little too
hard, use the wrong word; this happens all the time and in most cases doesn’t
§1.1 THEIR MACHINES AND OURS 7

prevent you from achieving what you wanted: eat, drink, drive, communicate.
But programming is different! At a dazzling speed — hundreds of millions of
basic operations per second — the computer will run your machine
description, your program, exactly as you prepared it. The computer doesn’t
“understand” your program, it just runs it; the slightest mistake will be
faithfully carried out by the machinery. What you wrote is what you get.

As you learn about programming in the following chapters, this is

perhaps the most important property of computers to keep in mind. You might
still believe otherwise: because computer programs do things that seem so
sophisticated — like finding, in less than a second, your ideal vacation rental
from millions of offers available on the World-Wide Web — you may easily
succumb to the impression that computers are smart. Wrong. Although some
programs may embody considerable human intelligence, the computer that
runs them is like a devoted and unsufferable servant: infinitely faithful, almost
infinitely fast, and definitely stupid. It will carry out your instructions exactly
as you give them, never taking any initiative to correct mistakes, even those
which a human being would find obvious and benign. The challenge for you,
the programmer, is to feed this obedient brute with flawless instructions
representing — in an execution of any significant program — billions of
elementary operations.

If you have used computers you know that they do not always react the
way you like. It doesn’t take very long to experience a “crash”, that state in
which it seems everything goes away and execution stops. But except for the
rare case of a hardware malfunction it’s not the computer that crashed; it’s a
program that did not do the right thing, and behind the program it’s a
programmer who did not foresee all possible execution scenarios.

You cannot learn programming without going through this experience of

programs — yours, or someone else’s — that do not work as they should; and
you cannot become a professional programmer without learning the
techniques that will let you build programs that do work as you want.

The good news is that it is possible to produce such programs, provided

you use the proper tools and maintain discipline, attention to the big picture as
well as the details, and dedication.
8 THE INDUSTRY OF PURE IDEAS §1.2

1.2 THE OVERALL SETUP

In the next chapters we are going to jump right into program development.
Initially we will not need too much detailed knowledge about computers, but
let’s see their fundamental properties, as they set the context for the
construction of software.

The tasks of computers

Computers — “automatic stored-program digital computers” to be precise —

are machines that can store and retrieve information, perform operations on
that information, and exchange information with other devices.

This definition highlights the major capabilities of computers:

What computers do
• Storage and retrieval
• Operations
• Communication

Storage and retrieval capabilities are a prerequisite for everything else:

computers must be able to keep information somewhere before they can apply
operations to it, or communicate it. Such a “somewhere” is called a memory. → A more precise
definition of “memory”
appears below:
Operations include comparisons (“Are these two values the same?”), page 12.
replacement (“Replace this value by that one”), arithmetic (“Find the sum of
these two values”) and others. These operations are primitive; what makes
computers able to perform amazing feats is not the intrinsic power of their
basic mechanisms, but the speed at which they can carry them out and the
ingenuity of the humans — you! — who write programs that will execute
millions of them.

Communication allows us to enter information into computers, and

retrieve information from them (the original information, or information that
has been modified or produced by the computer’s operations). It also enables
computers to communicate with other computers and with devices such as
sensors, telephones, displays and many others.
§1.2 THE OVERALL SETUP 9

General organization
The previous definition yields the basic schematic diagram for computers:

Communication Components of
devices a computer
Rest system
of the Processors
world

Memories

The memories hold the information. We talk of memories in the plural

because most computers have more than one storage device, of more than one
kind, differing by size, speed of access to information and persistence
(whether or not a memory retains information when power is switched off).
The processors perform the operations. Again there usually are several
of them. Occasionally you will still see a processor called a CPU, an acronym
for the older term Central Processing Unit.
The communication devices provide means of interacting with the rest
of the world. The figure shows the communication devices as interfacing with
the processors rather than the memories; indeed, when exchanging
information between a memory and the outside world, you will usually need
to go through some operations of a processor. A communication device
supports input (outside world to computer), output (the other way around), or
sometimes both. Examples include:
• A keyboard, through which a person enters text (input).
• A video display or “terminal” (output).
• A mouse or joystick, enabling you to designate points on the terminal
screen (input).
• A sensor, regularly sending measurements of temperature or humidity to
a computer in a factory (input).
• A network connection to communicate with other computers and devices
(input and output).
The abbreviation I/O covers both input and output. The words “input” and
“output” are also used as verbs, as in “you must input this text”.
10 THE INDUSTRY OF PURE IDEAS §1.2

Information and data

The key word in the above definition of computers is “information”: what you
would like to store into memories and retrieve from them, process through the
processors’ operations, and exchange through the communication devices.
This is the human view. Strictly speaking, computers do not directly
manipulate information, they manipulate data representing that information:

Definitions: Data, information

Collections of symbols held in a computer are called data.
Any interpretation of data for human purposes is called information.

Some people will tell you that “data” should only be used in the plural, because it’s
originally the plural of “datum”. Thank them for the kindness of their advice and
disregard it cheerfully. Unless they intend to continue the conversation in Latin,
their linguistic data is obsolete.
Information is what you want: the day’s headlines, a friend’s picture,
background on someone you’ll be meeting. Data is how it’s encoded for
the computer.
As an example, the MP3 audio format, which you may have used to listen
to music with the help of a computer, is a way to encode enough information
about a piece of music into data that can be stored in a computer, exchanged
across a network, and sent to an audio device so that it will replay the music.
The data will be stored in memory. It is the task of the communication
devices to produce data from information coming from the world out there, store
it in memory, and when the processors transform this data or produce new data,
to send it out to the world so that it will understand it as information. Adapted to
show the functions performed, the original picture looks becomes this:
Information Information Information
and data
processing
Input Output
Data Data

Process

Data

The right-to-left arrow suggests that the process is not just one-way but repetitive,
with information being repeatedly fed back to yield new results.
§1.2 THE OVERALL SETUP 11

Computers everywhere

The familiar picture of a computer is the “desktop” or “laptop” computer,

whose processor and memory components are hosted in a box of a size
somewhere between a textbook like this one and a big dictionary; the terminal
is often the biggest part. All this is at human size. At hand size we find such
devices as mobile phones, which today are really pocket computers with
extended telecommunication capabilities. At the higher end, computers used
for large scientific computations (physics, weather prediction...) can reach
room size. This is of course nothing compared to computers of a generation
ago, which took up building size for much more modest capabilities.

Computers:
desktop (a);
(a) laptop (b); PDA
(c); processor to
be embedded
(c) (d).

(d)

(b)

Reduced to their central processor and memory components, computers can

be much smaller than any of this. Increasingly, “the computer” is a device
included — the technical term is embedded — in products or other devices.
Today’s cars include dozens of small computers, controlling fuel delivery,
braking, even windows. The printer connected to your desktop computer is not
just a printing engine, it is itself a computer, able to produce fonts, smooth
images, restart on the next page after a paper jam. Electric razors include
computers; manual razors might include one some day. (The more expensive
razor blades already contain electronic tracking tags to fight theft.) Washing
machines contain computers, and in the future clothes may have their own
computers, helping to tune the washing process.
12 THE INDUSTRY OF PURE IDEAS §1.2

The computers you will use for exercises of this book are still of the
keyboard-mouse-terminal-box kind, but keep in mind that software techniques
have to cover a broader scope. Software for embedded systems must satisfy
very high quality requirements: malfunctions in (for example) brake-control
software can have terrible consequences, and you cannot fix them — as you
would for a program running on your laptop — by stopping execution,
correcting the error, and starting again.

The stored-program computer

A computer, as noted, is a universal machine: it can execute any program that
you input into it.
For this input process you’ll use communication devices, typically a keyboard and
mouse. Text will appear on your terminal screen as you type it, seemingly as a
direct result, but this is an illusion. The keyboard is an input device, the terminal a
distinct output device; echoing the input text on the screen requires a special
program, a text editor, to obtain this input, process it and display it. Thanks to the
speed of computers, this usually happens fast enough to give the illusion of a direct
keyboard-screen connection; but if the computer responds more slowly, perhaps
because it’s running too many programs at the same time, you may notice a delay
between typing characters and seeing them displayed.
When you input the program, where does it go? Memories are available to host
it. That’s why we talk of stored-program computers: to become a specific
machine ready to carry out the specific tasks that you (as the programmer)
have assigned to it, the computer will read its orders from its own memory.
This property of computers explains why we have not seen a proper
definition of “memory” yet. It could have said that a memory is a device for
storing and retrieving data, where (in accordance with the notion of
stored-program computer) “data” includes programs. But it is clearer to
separate the two notions:

Definition: Memory
A memory is a device for storing and retrieving data and programs.

This ability of computers to treat programs as data — executable data —

explains their remarkable flexibility. At the dawn of the computer age, it led to
visions of self-modifying programs (since a program can modify data, it can
modify programs, including itself) and to some grand philosophizing about
how programs were going, through repeated self-modification, to become ever
more “intelligent” and take over the world. Closer to us, it’s also the reason
why email users are told to be careful about opening email attachments, since
the data they contain could be a maliciously written program, whose execution
will destroy other data.
§1.2 THE OVERALL SETUP 13

For programmers, the stored-program property has a more immediate

consequence: it makes programs amenable, like any other kinds of data, to
various transformations through computer operations. In particular, the
program you write is usually not the program you run. Codes that a processor
can execute are designed for machines, not humans; using them directly to
construct your programs would be tedious and error-prone. Instead you will:
• Write programs in notations designed for human consumption, called
programming languages. This form of a program is called its source text
(or source form, or just source).
• Rely on special programs called compilers to transform such
human-readable program texts into a form (their target form) appropriate → Or “machine code”,
for processor execution. or “object form”, see ...

We’ll often encounter the following terms reflecting this division of tasks:

Definitions: Static, Dynamic

Static properties of a program are properties of its source text, which can
be analyzed by a compiler. Dynamic properties are those characterizing its
individual executions.

The details of all this — processor codes, programming languages, compilers,

examples of static and dynamic properties — appear in later chapters. What
matters for the moment is knowing that the programs you are going to write,
starting with the next chapter, are meant for people as well as for computers.
This human aspect of programming is central to the engineering of
software. When you program you are talking not just to your computer but also
to fellow humans: whoever will be reading the program later, for example to
add new functions or correct a mistake. That’s a good reason to worry about
program readability; and it’s not just a matter of being nice to others, since that
whoever might be you, a few months older, trying to decipher what in the
world you had in mind when writing the original version.
Throughout this book we’ll emphasize, along with practices that make
your programs good for the computer — for example, designing programs so
that they will run fast enough — practices that make them good for human
readers. Program texts should be understandable; programs should be
extendible (easy to change); program elements should be reusable, so that
when later on you are faced with a similar problem you don’t have to reinvent
the solution; programs should be robust, protecting themselves against
unexpected input and abnormal circumstances; most importantly, they should
be correct, producing the expected results.
14 THE INDUSTRY OF PURE IDEAS §1.3

A deck of
punched cards

Touch of history:
It’s all in the holes
Aerospace industry old-timers tell the story of the staff engineer who, in an (Possibly apocryphal,
early rocket project, was in charge of tracking the weight of everything that but a good story still.)
would get on board. He kept pestering the programmers about how much
the control software would weigh. The reply, invariably, was that the
software would weigh nothing at all; but he was not convinced.
One day he came into the head programmers’ office, waving a deck of
punched cards (the input medium of the time, see the picture): “This is the
software”, he said, “Didn’t I tell you it had a weight like everything else!”.
This did not deter the programmer: “See the holes? They are the software.”

1.3 KEY CONCEPTS LEARNED IN THIS CHAPTER

• Computers are general-purpose machines. Providing a computer with a
program turns it into a special-purpose machine.
• Computer programs process, store and communicate data representing
information of interest to people.
• A computer consists of processors, memories and communication devices.
These material devices together make up hardware.
• Programs and associated intellectual value are called software. Software
is an engineering product of a purely intellectual nature.
• Programs must be stored in memory prior to execution. They may have
several forms, some readable and intended for human use, others directly
processable for execution by computers.
§1-E EXERCISES 15

• Computers appear in many different guises; many are embedded in

products and devices.
• Programs must be written to facilitate understanding, extension and
reuse. They must be correct and robust.
New vocabulary
At the end of every chapter you’ll find such a list. Check (this is the first exercise
in the chapter) that you know the meaning of each term listed; if not, find its
definition, as you’ll need all terms in subsequent chapters. To find a definition, look
up the Index, where definition pages appear in bold.

Communication device Compiler Computer

Correct CPU Data
Dynamic Embedded Extendible
Hardware Information Input
Output Memory Persistence
Processor Programmer Programming language
Reusable Robust Software
Source Static Target
Terminal User

1-E EXERCISES
1-E.1 Vocabulary
Give a precise definition of each of the terms in the above vocabulary list.

1-E.2 Data and information

For each of the following statements, say whether it characterizes data,
information or both (explain):
1 • “You can find the flight details on the Web.”
2 • “When typing into that field, use no more than 60 characters per line.”
3 • “Your password must be at least 6 characters long.”
4 • “We have no trace of your payment.”
5 • “You can’t really appreciate her site without the Macromedia Flash
plug-in.”
6 • “It was nice to point me to your Web page, but I can’t read Italian!”
7 • “It was nice to point me to your Web page and I’d like to read the part in
Russian, but my browser displays Cyrillic as garbage.”
16 THE INDUSTRY OF PURE IDEAS §1-E

1-E.3 Defining precisely something that you’ve always known

You know about alphabetical order: the order in which words are listed in a
dictionary or other “alphabetical” list. Alphabetical order specifies, of two
different words, which is “before” the other. For example the word sofa is
before soft, which itself is before software.
The question you are asked in this exercise is simply:

Define under what exact conditions a word is alphabetically “before” another.

That is to say, define alphabetical order. This is a notion that you undoubtedly
know to apply in practice, for example to look up your name in a list of
candidates to an exam; what the exercise requests is a precise definition of this
intuitive knowledge, of the kind you might need for a mathematical notion —
or for a concept to be implemented in a program.
To construct your definition you may assume that:
• A word is a sequence of one or more letters. (It’s also OK to use “zero or
more letters”, that is to say accept the possiblity of empty words, if you
find this more convenient. Say which version you are using.)
• A letter is one among a finite number of possibilities.
• The exact set of letters doesn’t matter but for any two letters it is known
which one is “smaller” than the other. For example, with letters of the
Roman alphabet, a is smaller than b, b is smaller than c and so on.
If you prefer a fully specified set of letters, just take it to include the twenty-six
used in common English words, lower-case only, no accents or other diacritical
marks: a b c d e f g h i j k l m n o p q r s t u v w x y z, each “smaller” than the next.

The problem calls for a definition, not a recipe. For example, an answer of the
form “You first compare the first letters of the two words; if the first word’s
first letter is smaller than the second word’s first letter then the first word is
before the second, otherwise...” etc. is not acceptable since it is the beginning
of a recipe, not a definition. A proper definition might start: “A word w1 is
before a word w2 if and only if any of the following conditions holds: ...”.
Make sure that your definition covers all possible cases, and respects the
intuitive properties of alphabetical ordering; for example it is not possible to
have both w1 before w2 and w2 before w1.
About this exercise: The purpose is to apply the kind of precise, non-operational
reasoning essential in good software construction. The idea is borrowed from a
comment of Edsger Dijkstra, a famous Dutch computer scientist.