Quotes from G. Pascal Zachary

Highly creative people don't necessarily excel in raw brainpower. They are misfits on some level. They tend to question accepted views and to consider contradictory ones.

No matter how good a jockey, he can't turn a plow horse into a thoroughbred. It was the same with chips and software. Indeed, an operating system depended on a reliable chip.

Code writers and engineers often maintain the fiction that their own psychology has little bearing on their work. Reason rules.

There are invariably many ways to achieve roughly the same technical ends. Technical choices are often highly personal. While shaped by commercial considerations, technical decisions also reflect human values and psychology. Cutler

If you don't measure the performance, you're just guessing, and if you're guessing, you're not very likely to write top-notch code.

If one character, one pause, of the incantation is not strictly in proper form, the magic doesn't work. Human beings are not accustomed to being perfect, and few areas of human activity demand it. Adjustment to the requirement for perfection is, I think, the most difficult part of learning to program." The

It was now the age of visualization, when abstract concepts as well as basic needs and wants were increasingly expressed in visual terms. From its origins as a number cruncher, the computer had gone Hollywood; it was now an image maker of vast power. Thus, graphics in many ways defined the look and feel of computing. Cutler

The great advantage of digital media is that it can be stored, retrieved and massaged by a computer—at lightning speed and with unerring accuracy.

Code writers, like engineers generally, tend to get sidetracked by interesting but irrelevant conundrums.

The harbinger of a revolution, the Altair was the first mass-marketed personal computer. For the first time a computer was dedicated not just to a single task but to one person. The old guard of computing entirely missed the significance of this.

But Unix had a serious shortcoming: No common version existed. Over the years different versions of Unix had proliferated like weeds, so that an application written for one would not run unmodified on another. While DOS presented a single target to consumers, applications writers and computer makers, Unix did not.

NT is alarmingly complex. Consisting of six million lines of code, the program is among humanity's most intricate handiworks. "No one mind can comprehend it all," Cutler says. A

Compact code was inherently good since it consumed less internal memory. Consisting of slivers of silicon chips called DRAM ("dynamic-random-access-memories"), internal memory was like a gas tank. The larger the tank, the farther the car would go. The smaller the operating program, the more gas was left for all other programs. Because

It wasn't until five years after the first 360 hardware was introduced in 1964 that all of its software ran well. By then, IBM had spent nearly as much writing the software as designing the hardware. This astonished the company's managers and vividly highlighted "the greatest impediment to advances in computer technology," the problem of managing large software projects. At

The need for reliability greatly influenced the design of the operating system. What was the best way to isolate applications, so that their failure would not bring down NT too? Cutler's answer was to split the operating system into two major pieces. One piece was the "kernel," which never interacted directly with applications and thus couldn't be contaminated by them. The other piece was the graphical, visible portion of the operating system. To

After a time, Fogelin shifted into writing and editing manuals. Often lampooned, these texts, if studied with a fanaticism ordinarily reserved for the Bible, revealed a multitude of secrets.

In programming terms the piece of the operating system that sustained activity when all else failed was the kernel. It protected itself by imposing certain restrictions on applications, the most important being that only it, and never the applications, directed the hardware. The

Muglia insisted that a computer program, while certainly inspired and created by code writers, must reflect the currents of the market and the desires of customers. No great program was created by slavishly following the market or crudely regurgitating the requests of shoppers. But creators lived in a cocoon. The very demands of their craft made it hard to step outside the bounds of their imaginations. Muglia

Formed a decade earlier by a Whirlwind engineer, Digital was a rising star among minicomputer suppliers, who broke sharply with tradition. In the past, computer designers had promoted large mainframes that shared their power between many jobs. Minicomputers, often priced well below a hundred thousand dollars, made it practical for the first time to dedicate a computer to a single job, such as keeping track of parts, the data for an experiment or the operation of a machine tool. Though

Often pieces of the kernel called a bedrock layer of code, the "microkernel," which was the ultimate chief of staff within NT.

A kernel design had two main benefits. First, the kernel ensured reliability by allowing a user to keep other applications active or launch new ones even if one program accidentally halted. Second, the kernel made it possible for an operating system to display multiple personalities. Each operating-system personality called the kernel in order to issue commands to the computer.

Most code writers were like gifted athletes; they learned by doing and could not explain their actions. They just did it. This method, while fine for getting started, often hampered efforts at making code faster, which required unblinking self-analysis. "The secret to optimizing speed" he said, "is to ask yourself, 'What does this code actually need to do? What's the least work I can do to solve this problem?' " All

Bugs were the backdrop of the code writer's life. A comma in the wrong place. An "if" where there should be a "then." An erroneous call from one piece of the program to another. Each could instantly cause a seizure, the collapse of a finely wrought abstraction into a puddle of ones and zeroes. Only human, software was born to fail. If not catastrophically, then aesthetically. Every software captain knew this.

Coming from the field of aviation, where mundane errors could cost lives, he was exact. For all these reasons, Dunie was drawn to testing, the grubby and anonymous side of programming.