It was refreshing to work with microprocessor-built PCs when I joined MicroPro International in 1981 after six or seven years of mainframe and minicomputer programming. Those systems lived in a separate room with a lot of red tape to get things done, even if you used time-sharing terminals. In contrast, as a PC software developer you had all the problems in front of you. No more turnaround. Immediate feedback with changes, fixes and debugging.
The I company had previously worked for, Altergo, was an IBM mainframe consultancy. In 1977 it wanted to get into the minicomputer business and expanded its Dublin office to make this happen. I became the systems programmer there and was involved in the purchase of a DEC PDP-11/70 in 1978. Besides programming, I developed training courses and provided technical feasibility and assistance to customers that bought turnkey systems.
There was an initial sense that general purpose PCs weren’t made for doing important work. Word processing in the late 1970s, for example, required a Wang workstation costing several thousand dollars or an expensive DEC PDP minicomputer that supported a handful of terminals. Then along came comparatively very cheap software that ran on a desktop PC – software that was supplied on floppy disks and packaged with user-friendly manuals. What’s more, one PC could run many different kinds of applications.
MicroPro developed that sort of software. Seymour Rubinstein founded the company in the US in 1978 and was still there in 1981. Rob Barnaby was the original author of its flagship product, the WordStar word processor, but he had left by the time that I arrived. The company’s product set included MailMerge, which worked with WordStar to maintain mailing lists and produce mailing letters, and the SpellStar spellchecker that could run with or without WordStar. There were also three standalone applications: the CalcStar spreadsheet, the DataStar desktop database system and SuperSort, a very fast file sorting utility. All were written in 8080 assembler and ran on CP/M-80.
In 1981 MicroPro was growing faster than it could manage. WordStar was a remarkable product and was selling beyond expectations. Indeed, the company changed its name to WordStar International in 1989.
I was one of eight programmers that MicroPro recruited in Ireland in late summer 1981. It planned to build a European localization operation with IDA incentives and this group was the start-up team. Lyle Cowen, a development manager who had previously authored SpellStar, was responsible for setting up the Irish office and was the hands-on manager there for the first couple of years.
The original MicroPro office nameplate from Dun Laoghaire, pictured here, was removed in 1989 when the company changed its name to WordStar.
(Photograph by John Nolan)
To get things started, the company took its Irish programmers to San Rafael, California for product training and to get familiar with the codebase of the products. The training included actual work on the 8080 to 8086 conversion. Ray Breen, Gerry Byrne, Siobhan Duffy, Des McGrath, Warren McAllister, Sean McEvoy and David Taylor went over before me. I wrapped up my Altergo projects before joining them at the start of 1982.
In total there were between 200 and 300 people in MicroPro then and the number subsequently peaked at more than 400. When I was hired, I guess that there were 50 to 60 people in development. A large percentage of the company’s workforce was engaged in sales and marketing. There was definitely a feeling that we were ‘manufacturing’ software for small and large business offices and even for consumers (well, geeky ones, and consumers were ‘users’).
The executives wanted to expand MicroPro’s market beyond the US and started eyeing other markets. They needed European and Japanese versions of the MicroPro suite of products and had chosen Ireland as the place to produce them – most likely it was the IDA who proposed this – but many of the decision makers didn’t know how much software refactoring needed to be done.
It is anachronistic to talk about software globalization and localization in 1982. Those terms were not yet in use and the industry had no common description of what the process was. In MicroPro it may have been referred to as a European or Japanese version or simply ‘translation’. ‘Let’s have a product that can be sold in Germany’.
However, to make such a product, you needed to globalize it first – by enabling the software to handle more characters than the standard ASCII set used in the US – and then to localize it – by translating messages and menus into the required language, customizing the keyboard and creating default settings for a particular country. Support programs were also needed, so that translators would not have to worry about any arcane way of getting their text into each piece of software.
Globalization was the fun part of this process for us programmers, as the internals of the software had to be redesigned to handle non-ASCII extended characters.
This February 1984 recommendation for a company award highlighted John’s contributions to Kanji WordStar and to the customization of MicroPro’s software for the IBM PC.
(Photograph by John Nolan)
The first project that MicroPro Ireland was assigned was the conversion of the 8080 assembly code-base to 8086 assembly. Business machines based on the 16-bit architecture were slowly coming to market. When IBM introduced its first PC in the US at the end of the summer 1981, the conversion became more important (and urgent) in the eyes of the company for its future success. It was, after all, IBM and with a cheaper version of the 8086, the 8088, which could interface with cheaper 8-bit peripherals, the new PC was to be taken seriously. It wasn’t clear at first how successful the IBM PC would be. It had no market share, two floppy disks and just 16k of RAM, while some of the 8-bit CP/M systems had more RAM with bank-switched memory (MicroPro’s CP/M programmers used one of these, called the Performance Business Machine, that the company itself had developed.) Besides, IBM initially offered a choice of operating systems and the jury was still out as to whether CP/M-86 or PC-DOS would win out; so we had extra fun work to handle both.
The MicroPro Ireland project embraced these new technologies. MicroPro US continued to develop its products in CP/M-80 assembly language. CP/M-80 had a much higher market share. The 8086 systems were much more expensive. And the key programmers knew 8080 assembly much better.
While the Irish recruits were still in the US MicroPro leased offices in Dun Laoghaire. These were part of the same building as the DeerHunter pub in Sallynoggin (ah, many’s a pint…)
By the time that I moved back there in spring 1982, MicroPro Ireland had not only taken on the 8080-to-8086 conversion but also the translations project. In addition, I worked on FormSort – a new interface to the database products that was developed in C for the 8080 and then converted to 8086. Its testers and documentation people were in the US.
The IBM PC proved to be a game changer. It heralded a new industry of clones and the emergence of an OEM market for 8086-based computers running MS-DOS. The manufacturers of these machines included Digital Equipment, Tandy, Texas Instruments and Victor Technologies, whose Sirius Victor computer was sold in Europe as the ACT Sirius 1. All wanted WordStar to run on their systems and this became another function for MicroPro Ireland. We rearchitected parts of WordStar so that it would run on any of these machines with some straightforward customizations – display and printer support were not yet part of the operating system in those days.
I became the leader for globalization and OEM development in 1983 or so. My knowledge of the localization effort and projects was less detailed. As far as I recall, our sales offices in England, France, Germany and Sweden would have the message files and manuals translated into their languages and the Irish office integrated and tested them. Other people in Sallynoggin coordinated with the translators in each country. I suppose you could say, Ireland was the place where the product was created at the request of the European offices.
MicroPro’s Japanese office, meanwhile, produced a Kanji version of WordStar to run on the NEC PC and other Japanese PCs. This was based on the globalization work that we had done in Ireland.
More staff joined the Sallynoggin operation during 1982, including programmers Seamus Conlon and Mike Brady. At its peak in 1984 the headcount there reached 28, including people for software testing, computer maintenance and administration.
Lyle Cowen returned to the US in 1983. He hired Ed Kirkman as his replacement. John Livelie later came aboard as a development manager and was succeeded in turn by Paddy Malone. Other key people at MicroPro Ireland included programmer Brian Byrne, who worked on customizations, and Stephen Keating who was in charge of testing. Paul O’Reilly worked on globalization changes for the WordStar 2000 codebase and programmer Bob Gannon worked on just about all the products.
The MicroPro developers in California, meanwhile, expanded the product set. They wrote the all-new WordStar 2000 from scratch in C and released it in 1984. The Irish office provided localizations. The company also introduced a next generation spelling corrector – CorrectStar – that was written in C and able to offer up spelling suggestions ! TelMerge, which appeared at about the same time, was a communications program that accessed the messaging services that predated the internet. It could be run as a standalone product or in conjunction with WordStar.
In mid 1984 MicroPro wanted to integrate the European changes to the WordStar product code into the US version, so off to California I went. I initially thought that I would spend only a few months in San Rafael, but I became part of the US team officially in 1985. The European version became the main WordStar code base. I continued to add new features to this and other products.
MicroPro released the final CP/M version of WordStar in 1988. The software is shown here on the company’s own Performance Business Machine.
(Photograph by John Nolan)
By then, however, MicroPro International had lost its way and WordStar had lost market share. The software had gotten the reputation as being hard to use. In fact this was not the case; it was hard to learn. When WordPerfect came along with its function-key interface, it was marketed as easy to use and ate into WordStar’s business. The WordStar interface was a much better one for productivity as a user’s hands could be positioned in one place for typing and navigation.
WordStar 2000 was MicroPro’s response to the WordPerfect challenge. But it never caught on. WordStar 2000 was slower and bigger than the old WordStar – the hardware hadn’t caught up – and was fraught by a copy protection that nobody liked.
MicroPro then came up with a new product goal: an easy-to-use word processor called Easy. I refactored the WordStar core engine to interface to high-level languages, C and Modula-2. Easy had some success but WordStar wouldn’t go away. We had many more releases of the product – up to WordStar 7 in 1992.
While MicroPro was struggling, it went through more than one round of layoffs in the US. Further attrition followed because employees saw that their jobs were threatened. The company passed into the hands of new executives who did not really understand the packaged software business. They saw MicroPro Ireland as just another expense in the corporate accounts and were reluctant to assign new projects to Sallynoggin.
With MicroPro International fighting for its life, the workforce at MicroPro Ireland contracted. Nonetheless, OEM-related work continued there in the years after I moved to the US. One such project delivered WordStar on the Amstrad computer. And MicroPro Ireland continued to produce European translations of new MicroPro product releases – up to version 6.5 in WordStar’s case.
MicroPro Ireland finally shut down in 1990. Myles Cagney, a remaining programmer there, joined our US team.
WordStar International closed its doors in 1993. I had left some months before. I joined Autodesk as a senior software engineer in May 1993 and worked there for 20 years in the AutoCAD core code base as an individual contributor, designing and implementing new features and enhancements.
Last edit: March 2018
© John Nolan 2018