September 08, 2016

The grandfathers of DOS

Courtesy PC Magazine, June/July 1982.
Ken Olsen and Digital Equipment Corporation disrupted the IBM mainframe cartel with the minicomputer, that became the micro, that became the personal computer. Except DEC failed to make the critical transition itself and was acquired by a PC maker that also challenged IBM at the height of its powers: Compaq.

One of the tech pioneers who navigated the rocky transitional period was Gary Kildall (1942–1994). Kildall's CP/M operating system played a key role in shifting the software paradigm from the mainframe and minicomputer to the personal computer.

Kildall came a generation after Ken Olsen, half a generation before Gates, Wozniak, and Jobs. Olsen served in WWII. Kildall was a graduate student at the University of Washington when he was drafted into the Vietnam War. He would spend his enlistment teaching computer science at the Naval Postgraduate School.

He later became a tenured professor at NPS while consulting in Silicon Valley. In the early 1970s, he started work on CP/M, an 8-bit operating system designed to power the new microcomputers that ordinary people could afford.

Like the Altair, the 8080-based PC kit that Ed Roberts was building in Albuquerque, world-of-mouth ignited a tidal wave of interest and curiosity in the burgeoning "home brew" computer community. Kildall retired from teaching and together with his wife started Digital Research to develop and market CP/M.

By 1978, the company (headquartered in their house in Pacific Grove, California) had achieved sales of $100,000 a month.

Along with CP/M, two more Kildall innovations made the PC possible. On the technical side, the BIOS chip created a hardware "abstraction" layer that allowed an operating system to work "out of the box" with various hardware configurations without being hand-tuned for the particularities of each one.

On the business side, with the BIOS chip in hand, Digital Research could divorce the OS from dependency on a single hardware platform or manufacturer and sell CP/M to all comers, a marketing model that Microsoft would follow with great success.

The Apple I had debuted in 1976, built on "that horrible MOS Technology 6502 processor," as Kildall described it. But CP/M remained the dominate general-purpose microcomputer OS, running on the 8-bit Intel 8080 and Zilog Z-80. For a time, Microsoft sold the Z-80 SoftCard, enabling CP/M to run on an Apple II.

The SoftCard was Microsoft's number one revenue source in 1980, making Microsoft a major CP/M vendor. And was probably the reason IBM thought Microsoft was also an OS developer.

During the late 1970s, Kildall got distracted customizing the PL/I compiler for Intel CPUs. Development of CP/M languished for almost two years.

Apple released the Apple III in 1980. It was plagued by reliability problems, a lack of software, and like the later Lisa, carried a "sky-high" price. On sabbatical at the time, Steve Wozniak returned to Apple in order to supervise production of the highly successful Apple IIe. Apple regained its footing in 1983.

But in 1981, the microcomputer industry was without a technological leader. In August of that year, IBM changed everything with its 16-bit Intel 8088-based PC.

In Triumph of the Nerds,  Jack Sams recounts how his IBM team, in desperate need of an operating system for the IBM PC, approached Digital Research (on the recommendation of Bill Gates) but couldn't get anybody to sign the strict non-disclosure agreement or agree to their tight production schedule (accounts differ).

The second time IBM raised the issue with Microsoft, Bill Gates signed in a heartbeat. Gates didn't care about the licensing terms as long as it was non-exclusive and Microsoft could sell MS-DOS to other hardware manufacturers. IBM agreed and Microsoft changed the world.

Except Microsoft didn't have an OS in development. So it licensed 86-DOS from Seattle Computer Products for a song and hired the guy who designed it, Tim Paterson.

Kildall later protested that Tim Paterson hadn't reversed-engineered CP/M but had copied his source code. He never pursued this claim. (When Compaq later reversed-engineered the IBM BIOS, it documented every step with legal precision and was never sued by the litigious IBM.)

Paterson, employee number 80 at Microsoft, remembers his historic role with something of a philosophical shrug.

It's been pooh-poohed as Seattle Computer being suckers or something for taking the deal because it made Microsoft so much money. I don't know how many people would have said the guy who provides the operating system to IBM is going to make it rich. I have the impression Bill Gates and Paul Allen felt it was a gamble, not that they were sitting on a gold mine and knew it.

Tim Paterson.
Paterson's Microsoft stock options did allow him to retire at age 42. And when Seattle Computer Products found out who Microsoft's client really was and went to court about it, that "song" became a million dollars.

In any case, Kildall's 16-bit version of CP/M for the PC didn't come out until April 1982, and then initially at six times the price of MS-DOS. Alas, it wouldn't be competitive at any price. The computing world finally had a software and a hardware standard and was sticking with it.

These latter details don't make it into Kildall's memoir, which concludes at the end of the 1970s. Or at least the version we have. Kildall never published the manuscript. The first half was recently made available by his estate as a free PDF download.

Titled Computer Connections: People, Places, and Events in the Evolution of the Personal Computer Industry, Kildall writes with a readable style, not overburdened with technical jargon (although there is plenty of that). It's a compelling personal account about the roots of the PC operating system.

The theme of his recollections might be: "You kids don't know how tough we old-timers had it!" He says this with a wink and a nod, but he's absolutely right. Compared to the hoops programmers once had to jump through, the floppy disk drive and the command line were absolutely amazing steps forward in usefulness.

Kildall's account ends at the end of the 1970s, before the stormy advent of the IBM PC (read the rest of the story here). But he does mention two other times he and Bill Gates crossed paths. The first sounds like a script straight out of Hollywood.

When Kildall was at the University of Washington, two high school students hacked into C-Cubed, a time-sharing computer company run by the director of the UW Computer Center. The kids were none other than Bill Gates and Paul Allen, the future founders of Microsoft. And what happened to them? C-Cubed hired them.

A decade later, a fledgling Microsoft was based in Albuquerque, creating software tools for the Altair. In 1977, Gates came to Pacific Grove to discuss the future of the business with Kildall, who was then running the world's most successful microcomputer software company.

Kildall remembers them getting along like oil and water, "his manner too abrasive and deterministic, although he mostly carried a smile through a discussion of any sort." Kildall had no desire to "compete with his customers," and turn Digital Research into a one-stop that sold both tools and applications.

Exactly what Gates was planning to do. Recalled Kildall,

We talked of merging our companies in the Pacific Grove area. Our conversations were friendly, but for some reason, I have always felt uneasy around Bill. I always kept one hand on my wallet, and the other on my program listings. It was no different that day.

So Microsoft ended up back in Seattle, where Gates and Allen grew up.

Kildall didn't think highly of Gates as a computer scientist. But in all fairness, I'll point to this landmark interview by Dave Bunnell in the debut issue of PC Magazine. As early as 1982, a young Bill Gates demonstrated a remarkably insightful grasp of where the personal computer industry was headed.

Gary Kildall may not have liked the man he ended up passing the baton to, but there's no denying that Bill Gates grabbed it and ran like a bat out of hell.

Related posts

Computer Connections
The accidental standard
MS-DOS at 30
Digital Man/Digital World
The cover

Labels: , , , ,


# posted by Blogger Joe
The accusation that Paterson copied CP/M has been proven false as far as such things can be proven. Given how primitive CP/M and 86-DOS were at the time, this is extremely likely. The irony is that Digital Research later did copy MS-DOS to create DR-DOS. What makes the latter peculiar is that it failed spectacularly to follow the MS-DOS API strictly, even though the Int 21 interface was extremely well documented by that point. This made it horrendous to work with, especially on low-level DOS code (it often changed registers on return in ways that MS-DOS didn't. Then there were several internal structures, which DR-DOS had changed.)

(As a side note; in the early 90s, conspiracy theories of Microsoft deliberately making Windows incompatible with DR-DOS were running rampant. However, at Artisoft we never got LANtastic to work with DR-DOS. Later Novell also had extreme difficulty getting their stuff to run on DR-DOS, AND THEY OWNED IT by that point. To my knowledge, Novell's software never did fully run on DR-DOS and since we couldn't find anyone actually using DR-DOS, it quickly became a pointless task.

In a further irony, in 2005 it was discovered that DR-DOS had taken code from FreeDOS.)

9/09/2016 11:12 AM