Code guy

Charles Petzold's new book -- this one for nonprogrammers -- is a remarkable accomplishment.

A decade ago, back when MS-DOS's rule of the desktop was being challenged by OS/2, Windows 3.0, GEM, and a handful of lesser operating systems, I bet on Windows and started a little magazine called Windows Tech Journal. WinTech was the first developers' magazine devoted to the Windows platform. My friends in the publishing industry were skeptical -- Windows was too small a niche market, they said, to support a whole magazine. My bet paid off -- I won't trouble you with the details -- and I spent a decade living, sleeping, eating, and dreaming Windows programming.

Real programmers used C in those days. Their bible -- and mine -- was Charles Petzold's remarkable Programming Windows. This essential book did more than Microsoft's documentation, megabytes of sample code, and the rest of the publishing industry's books and magazine articles combined to explain how to write code for Windows. Petzold simultaneously managed three difficult tasks: He explained how Windows worked, how it was structured, and how application programs should interact with it. He demonstrated through a series of crystal-clear examples how to write Windows programs. And he somehow made the text lively and interesting even though the book is one of the densest, most information-packed documents I've ever read.

I met Petzold long ago when we were recruited to participate in a programming-conference panel discussion together. I've gotten to know him pretty well over the years. In addition to being a meticulous programmer and a gifted technical writer, Charles is one of our industry's nicest and most thoroughly decent fellows. He has a shy grin, a quick wit, and an infectious laugh. He always finds time to share a few minutes of friendly gossip when our paths cross at programming conferences.

CHANCE ENCOUNTER

I don't really keep up with Charles's work anymore. I'm struggling to learn Linux, not Windows. I rarely read PC Magazine, where his articles most often appear. And Charles's devotion to ANSI C, which makes his programming examples so clear, also makes them seem less relevant in these days of newfangled high-level object-oriented component-based data-managing Internet-enabled RAD environments.

Nonetheless, I keep my eye open for Petzold's byline. I was delighted to find it in an airport bookstore a few weeks ago on the cover of Code: The Hidden Language of Computer Hardware and Software. I flipped the book open to the back flap and there was Charles's smiling face. I had to have it. I bought the book and began reading as I waited to board my flight.

With Code, Charles Petzold and Microsoft Press are attempting to cross over from Petzold's proven audience -- professional programmers -- to the broad consumer market of nonprogrammers who are curious about how computers work. Code is a very technical book about computers and software intended for a nontechnical audience.

Charles Petzold is surely among the world's half-dozen most qualified authors to write such a book. His understanding of computers and software is comprehensive. His writing style, though true to the technical-writing tradition, is clear and approachable. And Charles has an undisputed gift for understanding and explaining complex topics. If anyone could make computers and programming comprehensible to nonprogrammers, surely it is the author of Programming Windows.

FROM HUMBLE BEGINNINGS

Code contains all the information you need to build a computer of your own. Petzold begins with a battery, a loop of wire, and a bulb from a flashlight. Then he adds a switch to demonstrate on and off. The addition of a second switch leads to binary numbers: He shows how the circuit's four states can be expressed in two binary digits: 00, 01, 10, 11.

Presented in parallel with the construction projects are explanations of various codes. Petzold shows how Braille, Morse code, and ASCII work to encode text, and explains how these codes might be related to the binary numbers computers use.

The hardware projects get successively more hairy. By Chapter 12 we're building an adding machine out of wires, switches, relays, and light bulbs. Step by step, Petzold introduces more complex circuits and more advanced topics. Gates and oscillators. Edge-triggered latches. Memory circuits. It's like an EE degree in 400 pages.

Eventually, Petzold demonstrates how the machines he's diagrammed can be programmed, and he explains how assembly language works. By this time, the patient reader has all the low-level details of an eight-bit, 64K computer in his head. So Petzold explains what operating systems do and sketches out how the reader might write one.

It is a remarkable accomplishment. Petzold explains computers the way Aristotle would have: methodically, exhaustively, along a meticulously planned course from simple ideas to complex concepts. I didn't use wire and light bulbs to build the hardware projects along the way as I read the book, but I'm sure I could have. All the details were there.

WHAT'S NOT TO LIKE?

Let's start with an unassailable truth: This book is a remarkable achievement. Petzold's trademark clarity is a match for the material. Squeezing this much material into 400 pages is a miracle of logical thinking. Organizing material this complex into a volume that proceeds naturally, with each chapter building upon material presented previously, requires meticulous step-by-step planning. I don't know who else could have written this book.

I am not the nontechnical reader Petzold and Microsoft Press envisioned for this book. I came to the party with rusty but thorough knowledge of truth tables, ASCII and other data-encoding schemes, assembly language, and the role of the operating system. My background in software prepared me to appreciate Petzold's thoroughness and clarity in presentation.

I do not have an EE degree, however, and I have always relied upon computer hardware with only the fuzziest understanding of how the hardware worked. I understand the programmatic effects of assembly-language opcodes, but I lack the background to envision what physical changes result in the computer's circuitry. So Petzold's lucid explanations of computer hardware were new and interesting to me.

It is at this point that I come to my first reluctant criticism of this book. Even though I have a technical background, and even though I was motivated to learn the material, I found portions of Code mind-numbingly dense. There must be a way to allow laymen to understand how computers work without presenting circuit diagrams and exhaustive descriptions of level-triggered D-type flip-flops, frequency dividers, and comparators. The schematics in the first few chapters were clear; they added to the text. But by the time I reached the middle of the book I found my eyes glazing over every time I encountered another diagram. I was so grateful to find software-oriented chapters sprinkled among the schematics!

If my first criticism regards Petzold's judgment in presenting too much information on some topics, my second regards the places he provided too little. Code runs out of steam near the end of the book, just about where Petzold begins his discussion of operating systems. Petzold sketches operating systems, peripherals, high-level languages, and graphic-oriented operating environments with the lightest of touches. After devoting hundreds of pages to switches and relays, Petzold devotes just 19 to graphical operating environments like Windows and the Macintosh. Just as the story gets interesting -- as the book begins to cover the kinds of computers we have around us today -- the in-depth coverage ends.

Charles is a friend of mine, and it pains me to bring up even the slightest criticism. But I think this latter flaw is significant. It would have been better, I think, to end the book with an eight-bit computer and roll up all the industry's achievements since CP/M into a brief concluding chapter. Alternately, Petzold could have written another three or four hundred pages and extended his in-depth discussion into modern computing environments. Given his rare gift for explaining technology, Petzold is just the man to make modern computer architectures, distributed computing, GUIs, objects, and the Internet make sense.

It is not too much to hope, then, that Charles Petzold might one day write a companion volume that picks up where Code leaves off. That would be a book well worth reading!