I don't know exactly when the concept of the Atari Computer was developed within the corporate mind of Atari, Inc., nor do I know all of the people responsible for nursing that concept into reality. The following history covers the relationship with Atari, Inc., during the evolution of the system software.
     Sometime in early 1978, when the Atari 800 and 400 were still called “Coleen” and “Candy” and were still in the breadboard stages, Atari bought a copy of the source for Microsoft 8K BASIC. This version of BASIC was fundamentally the same product that was implemented by Commodore in the early PETs, was used by OSI, and was a close ancestor of Applesoft. Six months and many, many Atari man-hours later, that 8K BASIC was almost functioning properly on the Atari prototypes. But buying source for a program buys you just that: source. Generally, you also receive little documentation, sometimes obscure code, no guide to modification, and no real support. What to do? The products were due to be shown in early January, 1979, at the Consumer Electronics Show (CES) in Las Vegas, Nevada.
     Enter Shepardson Microsystems, Inc. (SMI), my employer at that time. Though little known by the microcomputer public, SMI had already produced some very successful, private labeled microcomputer software. Among our better-known efforts were the original Apple DOS, Cromemco 16K Extended BASIC, and Cromemco 32K Structured BASIC (just being completed at that time). Also, we had done some work for Atari on a custom game processor. (Which used a 12-bit ROM and 5-bit RAM configuration and was well received at Atari, hut never produced.)
     Coincidentally, about that same time SMI had also purchased source for Microsoft 6502 BASIC. After producing Apple’s DOS, we had the bright idea of mating the Apple II peripheral bus with the KIM/SYM/AIM system bus (and it still seems like a good idea to us, but ...). The idea was to provide a disk system (Apple’s) to the Single Board Computer market. Needing a BASIC to sell with the system, we plunked down a few grand and purchased Microsoft’s. Though it looked to us like it would be difficult to modify, we were intending to resell it with a minimum of changes, so it seemed appropriate.

A New BASIC?
Re-enter Atari, some time in the late summer of 1978, asking if SMI could help them. With Microsoft BASIC? Well ... we really didn’t want to, but ... Could we propose a new BASIC? We talked. And had meetings, and a study contract, and more meetings, and finally we wrote a specification for a 10K, ROM-based BASIC. (I still have a copy of that spec, and it’s amazing how little the final version deviated from that original.)
     Of course, in the middle of all these discussions, Atari naturally divulged how their (truly superb) ROM-based Operating System would interface both with BASIC and with various devices. Somewhere in here, my memory of the sequence of events and discussions becomes a little unclear, but suffice it to say that we found ourselves making a bid on producing not only a BASIC for Atari, hut also the File Manager (disk device driver) which would change Atari OS to Atari DOS.
     Sometime in late September, 1978, the final proposal was made to Atari, and it was accepted by them shortly thereafter. In mid-October, 1978, we received the go-ahead. The project leader was Paul Laughton, author of Apple DOS. The bulk of the work ended up being done by Paul and Kathleen O’Brien. Though I was still involved in the finishing touches on Cromemco BASIC, I take credit for designing the floating point scheme used in Atari BASIC. Paul Krasno implemented the math library routines following guidelines supplied to us by Fred Ruckdeschel (author of the acclaimed text, BASIC Scientific Subroutines). And, of course, much credit must go to Mike Peters, our combination keypuncher/computer operator/junior programmer/troubleshooter.
     Since we obviously couldn’t have the Atari machines to work on (they hadn’t been built yet), the first step was to bring up an emulator for Atari’s CIO (“Central Input-Output,” the true heart of Atari’s OS) on our Apple II systems. With Paul Laughton leading the way (and doing a lion’s share of the work), the pieces fell together quickly. “Little” things had to be overcome: the cross-assembler was modified to handle the syntax table pseudo-ops, the 256-byte Apple disk sectors had to be made to look like 128-byte Atari sectors, the BASIC interpreter seemed to function, but was waiting for the floating point routines. And there are funny things to tell of, also. Like our cross-assembler, running on an IMP-16P (a 1973 vintage, 16-bit, bit-sliced PMOS microprocessor) that used keypunched cards for input, a floppy disk (with no DOS) as temporary storage, and a paper tape punch as output.

     Somehow, Kathleen and Paul guided the two programs unerringly toward completion. On December 28, 1978, Atari’s purchasing department at last delivered a signed copy of the final purchase order. It called for delivery of both products by April 6, 1979. There was a clause which provided for a $1,000 per week incentive (if we finished early) and penalty (if we finished late). What is especially humorous about that December 28th date is that the first working versions of both BASIC and FMS had already been delivered to Atari over a week before! That is fast work.

     Fortunately, then, Atari took their new Atari BASIC to CES. Unfortunately, there was a limit on the amount of incentive money collectible. Oh, well.

     In the months that followed, SMI fixed bugs, proofread manuals, and worked on other projects (including the Atari Assembler/Editor, which was mostly Kathleen’s effort). The nastiest bugs in BASIC were fixed by December, 1979, but it was too late: Atari had already ordered tens of thousands of BASIC ROMs. The FMS bugs were easier to get fixed, since DOS is distributed on disk.

     In mid-1980, Paul Laughton once again tore into FMS. This time, he modified it to handle the ill-fated 815 double-density disk drive and added “burst I/O” (and there will be much more about both these subjects in the technical discussion that follows).

     In late 1980, and early 1981, Bob Shepardson, owner of Shepardson Microsystems, Inc., decided that the pain and trouble of having employees wasn’t justified by the amount of extra income (if any) that he derived. Though we still occasionally function in a loose, cooperative arrangement, the halcyon days of SMI seem to be over.

A New Beginning
I negotiated with Bob Shepardson for his rights to the Atari products (FMS, BASIC, and the Assembler/Editor) and their Apple II counterparts. Thankfully, Atari had purchased from SMI only a non-exclusive right to distribute these products. SMI had retained the rights to license other users on a similar non-exclusive basis (and, indeed, SMI sold a version for the Apple II during most of 1980).
     So now it was frantic time again: this was February 25, 1981, and the West Coast Computer Faire was April 3rd. But our brand new company, Optimized Systems Software, arrived on time, bringing with it BASIC A +, OS/A + and EASMD. All three were enhanced, disk-based versions of the original Atari programs (and, in fact, derived some of their enhancements from the previous OSS Apple II products).
     The products have been well received by the Atari user community, in part due to the fact that they are truly compatible, yet enhanced, versions of standard Atari software.

Why This Book?
The decision to publish these listings was not an easy one to make; and it is, in its own way, an historic occasion. After all, have you ever seen anyone offering source or listings of CP/M, the most popular of all computer operating systems? Since Atari, to their credit, has honored the original agreement with SMI and not released either source or listings without permission, the responsibility for doing SO seemed to rest with OSS.
     But Atari has set a powerful precedent by publishing the listings of DUP (their portion of DOS 2.0S) and the OS ROMs. The clamor from Atari users for the source for FMS finally even reached us, so we have bowed to the inevitable, and honored the same commitment that Atari has made: to release as much information and aid as possible to the user community.
     We hope that the users will appreciate these efforts and, in turn, respect our rights and Copyrights. As long as there is a mutual respect and benefit, you, the user, can expect continued support.

About This Book
With the release of this book, the dedicated Atari enthusiast can examine all the inner workings of Atari DOS and modify his (or her) system to his heart’s delight. Rather than simply publish listings, we have chosen also to provide a complete guide to the workings of FMS.
     Although the listing itself is relatively clear and commented, all but the most expert would have trouble plowing through some of the tortuous logic necessary in such a program. The guide included here describes all aspects of the FMS, including the external view, the charts and tables, the various interfaces, and (in copious detail) the functions of the individual subroutines (including complete entry and exit parameters).
     There is much of value here even for the person who never intends to modify Atari DOS. We feel that FMS is a fairly well-structured, relatively sophisticated, system level assembly language program. We hope that most users will gain by the insights presented here.
     We would welcome any notes you would care to send pointing out errors either in the DOS or in this book.

Bill Wilkinson
Optimized Systems Software
Cupertino, California
February, 1982