Talk:Onyx Systems

CTS-300 Unisis on the CSRG DVD #1
In CSRG Archive Historic DVDs, all of the diskettes mentioned in the LABEL file are present. $ cat LABEL Here is the information on the UniSoft V7 system that was shipped with a Codata 68000 computer:

Distribution on 16 floppies: 1 boot, 15 volumes of dump.

Box label:

UNISIS O.S.	System Diskettes S/N: 220 .7f  2/28/83

Label on boot disk:

0.0.1.7.0.1.2		(7 is in pen, replacing a 5) CODATA SYSTEMS U33md.7f8b46665 System Bootstrap .7f Disk 1 of 1, 2/17/83

The dump floppies are all labeled in the same way except for the disk number:

DUMP/RESTOR CODATA SYSTEMS Disk 1 of 15 .7f, 2/25/83

The following notice appears in some of the header files:

/* * Copyright 1982 UniSoft Corporation * * Use of this code is subject to your disclosure agreement with AT&T, * Western Electric, and UniSoft Corporation */

In a header file it says: Constants and definitions for the CTS-300 CPU board while in a banner it says: Codata CTW-300 system with UNISIS not CTS-300. CTW-300 is written in a ad, it's not a typo.

Jamplevia (talk) 21:43, 19 June 2023 (UTC)

Some inside history regarding Onyx Systems by John L. Bass, Lead UNIX Systems Programmer at Onyx during 1980
Hired by Bob Marsh and reporting to Craig Forney, I joined Onyx Systems early in 1980. Bob McClure's consulting company failed to deliver a completed UNIX C Compiler and UNIX V7 port for the Zilog Z8000 based C8002 product. Bob McClure had won two fixed price software development projects and didn't have the staff to complete both on time, so they focused on the much larger oil field project, and defaulted on the Onyx Systems Unix port. Onyx Systems faced a crisis as all the founders and employees stock options were contingent on shipping production ready C8002 systems by end of 2Q80.

I came with five years of extensive UNIX Kernel and System development experience at CalPoly San Luis Obispo and SRI International (formally Stanford Research Institute). Plus another six years of extensive systems programming experience on IBM, CDC, NCR, Data General, Varian, and other minicomputer systems. The skill set and experience I brought Onyx Systems was rare at the time, with maybe a couple dozen other seasoned UNIX Kernel systems programmers world wide. With SRI lead investigators help, we had hosted several West Coast UNIX User Group meeting at SRI in Menlo Park, with strong US and World Wide attendance. This gave me the visibility for Bob Marsh to approach and hire me.

As I came into Onyx Systems, I was tasked by Bob Marsh to assess Bob McClure's progress and ability to deliver. At the time Onyx System's offices were in the shipping area of IMI. Carl Berg, a large real estate developer, had funded the 8" hard drive company IMI, and needed to create several computer companies to provide a market for the new hard drive. I was given a key to Bob McClure's office and was tasked with several "do nothing projects" by Bob McClure to keep me busy. For nearly all that first two weeks, all of McClures staff was in Texas trying to complete the oil field project which was also late. When I shared with Bob Marsh that McClure and staff were in Texas for the last two weeks, he was panicked by the Onyx UNIX default when I didn't have the answers he needed.

I went to McClure's office that afternoon to get Bob Marsh the answers he needed about progress on the Onyx Port. I powered up McClure's PDP11, and booted UNIX V7 from a 9 track magnetic tape I kept for hard disk repair at CalPoly and SRI. I mounted McClure's hard drive, and went to the Onyx port source tree's directory. I did a "find" command for the key kernel sources, and there was a single copy, that was nearly virgin AT&T original unmodified sources, with a few trivial and insignificant changes. I did a full backup of the disk to tape, and left McClure's office knowing I had be lied by McClure about the progress of the UNIX port. Nothing was done, just in progress work on the C compiler and assembler for the Z8000 being tested on standalone boot code for the Z80 based disk/tape system. PDP11 device registers were still in both the UNIX kernel assembly and C code that would cause a hard exception running on the Z8000.

I went back to Onyx Systems, and reported this to Craig Forney who was in total disbelief that McClure had lied to them that the project was nearly complete, and only required a few final finishing touches. So I took Craig back to McClures office, booted the PDP11 from my tape again, and let Craig search the hard drive for "golden code" that I presumably had missed. Two hours later, it was clear to Criag there wasn't any "golden code" waiting to be delivered to Onyx Systems. I left Craig to deliver the bad news to Bob Marsh, where "all hell broke loose". This was the week Onyx Systems was moving into it's own building. The meeting the next morning with Craig and Bob in the new office was tense (a gross understatement). The call to McClure in Texas by Marsh wasn't that great either, and prompted McClure returning to Calif late the next night (Friday night) with a meeting scheduled for Monday morning with McClure at the new Onyx Systems office. McClure stuck to the story that his team was just a few days, maybe a week or two from delivering the project, and that Craig and I didn't find the correct source tree.

At the time I only lived a couple miles from McClure's office, so I checked it frequently. Shortly after dark Friday night the lights at McClure's office came on, and several cars showed up at the office as part of McClure's team took on the task of making as much progress as possible by Monday morning. On Monday morning McClure showed up with a listing under his arm of UNIX sources, and stood by the story his team was only a few days, or weeks from delivering the project completed. I walked up to the table with the listing, and took a yellow highlighter marking the PDP11 register accesses that would fault the Z8000 in the kernel initialization code. I also outlined over a dozen critical C compiler bugs that would also cause the kernel to fault on boot, that I discovered while working on the "do nothing" projects that McClure had assigned me. McClure didn't bring the expected boot tape for the C8002 to show that it was almost functional. There was a very "heated" discussion and McClure left.

I had spent the weekend doing a project plan to rescue the ONYX UNIX port project, assuming McClure would completely default. At that time, McClure was one of several people in the world that had done significant work inside the UNIX C compiler ... that experience set was not replaceable. It was difficult for Bob Marsh to accept that he still needed Bob McClure to meet the 2Q80 deadline in four months. I recommended that McClure be given all or most of the original porting contract value to complete the C compiler and tools, and that ultimately is what saved Onyx Systems from it's 2Q80 default. I setup in a remote office with a PDP11 running UNIX, and my own C8002 system. Long 7x24 weeks leaving site only a few hours a week for supermarket and laundry runs. For the next 3 and a half months I debugged the C compiler, tools and library at several bugs per day. I would give McClure a new set of bug reports every day, and get a new compiler and tools with fixes every few days. For every bug I found, I would "@ifdef bug" around the incorrectly compiled code segment, and replace it with a functionally equivalent C or inline asm code segment that was functional and keep working ahead. For each fixed compiler bug, I would turn off the "#ifdef bug" to verify the fixes. I wrote and debugged all the kernel support and drivers, and about two weeks to deadline, it came up to a shell prompt and would execute simple commands. The next two weeks were debugging the last of the C compiler and tools problems, generating install tapes, getting several other machines running for others to test. We shipped a dozen machines to customers, and met the 2Q80 deadline.

Not long after, Carl Berg bet on CPM/MPM based Z80 systems, and pulled the plug on the Z8000 UNIX team and founding management. Bob Marsh, and the others, started over founding Plexus Systems a few months later. I stuck around till Christmas, then helped found Fortune Systems. The entire industry jumped on 5" hard drives, the 8" IMI drive became history as IMI tried to catch up with a 5" product.

IBM released it's PC with MSDOS not long after, and Onyx systems failed a couple years later trying to build a new UNIX team after the CPM/MPM bet failed.

I've told most of this story a number of times in various forums over the last 30 years ... so with a little digging, it can most likely be found elsewhere too.

John L.Bass offworld@ymail.com

I gave up my dmsd.com domain after selling businesses a couple years ago. Now located on a small farm in SW Kansas. Still doing engineering, and having fun with Ardunio products. 172.58.180.13 (talk) 02:43, 17 May 2024 (UTC)