About me, the Mainframe years.

I've been using computers since mid 1969, when I was hired by the Commonwealth Bureau o Census and Statistics in Perth, Western Australia, as a programming assistant.

In those days, our CDC 3200 with 16K 24-bit words was a mainframe. Data entry (including programs) was by punched card and paper tape, the computer could run one program at a time. Data storage was on 7-track magnetic tape, not very reliable even then. It was then and there I wrote my first computer programs, in FORTRAN and COMPASS (COMPrehensive ASSembly).

Over time, I moved to Canberra, undertook a little study, and was promoted to the Department of Social Security where I encountered IBM mainframes - System 360 models 40 and 65, OS/360 PCP, MVTII and MVT (mostly remotely, our computers were everywhere but in Canberra, though we did use Health's computers too for some time).

Time passed, the Whitlam Government was elected in 1972, and one of Labor's ambitions was an improved national health system, which they called "Medibank."

More staff, more computers and much fun. At last, we had computers in Canberra, two of these. By then I was officially a programmer, part of the Database Administration team, charged with managing the department's IMS configuration and database. The computers were running OS/VS2 release 1, also known as SVS for Single Virtual Storage. A problem we had was occasional corruption of our database. I proposed a program to read the database outside of IMS control, using VSAM (from assembler) to check for and report on errors. I was assigned the task of writing the program, but

time passed, the Whitlam Government made way for the Fraser Government in '75, and the public service was reorganised. Medibank (and the computers) passed to Health, and I elected to stay with Social Security.

I left the program in the hands of one Don Ryan who continued with it, and inherited a communications program written in assembler (which was fine), using a communications protocol, BTAM, which I didn't understand, using a protocol unique to the Australian Public Service which hardly anyone understood, and which to this day nobody else than DSS has implemented. I got the program working well enough to determine that the basic design of the protocol wasn't going to do what we wanted without major work on the minicomputers at the other end of the wire, and those were now owned by Health who had no interest in the application.

Life settled down after that to our group's maintaining the software (OS/VS1) on our remaining computers, but I'd become the communications specialist. I implemented Remote Job Entry (I didn't write the software, merely installed it and talked with those needing to be talked to about progress) across the Bass Strait - DSS Tasmania still didn't have a computer. The above communications application resurfaced, and I got to reimplement it, somewhat better if I say so, using EXCP - BTAM was too inflexible, and I had to justify EXCP to my boss, and remote 3270 terminals.

I also provided technical support for the department's evaluation of ADABAS and COM-PLETE, undertaken on the department's System/370 model 135 in Brisbane.

Time passed, The Boss was promoted to another department, neither of those we (the team) considered the best replacement got his job, so we left too.

I worked for SPL (Australia) for a while - Geoff Holloway, whom I had met briefly years before, was the boss, and I was supporting ADABAS and Natural (a programming language) in Canberra. My duties included installing the software, delivering instruction, writing and maintaining an interface between Natural and AIM/DC on Fujitsu mainframe, and adapting the new release of ADABAS (4.1) to run on ISIV/F4 E40.

It was around this time that the first home computers appeared, often self-built, sporting such CPUs as the 6800, 6502, i8080, i8080A, i8085, Z80.

Time passed, SPL and I parted and I went contracting. I did discuss with SPL more work, but the relevant project didn't eventuate. I did some work for Fujitsu (a tender for DSS) , for CSIRO related to my ADABAS/Natural work, and after the tenders were announced, at DSS for some years. There, I was again part of the team responsible for mainframe software, but by now the team was larger (18 people), the Amdahl computers physically smaller, but much more powerful, running OS/VS2 SP1 (later SP2), an SNA network, CICS, Model 204 and a large shopping basket of other software. I was responsible for maintaining third-party software relevant to the programmers - not ISPF because that was an IBM product, but Panvalet, Panexec, easytrieve and so on, and developing and maintaining other tools such as ISPF dialogs to combine the tools. I also got the tasks of writing JES2 exits, one a job separator writer, another to examine the internal format of JCL so as to enforce departmental standards.

It's my view that, where rules punish users (and remember, my users were other employees), for example by cancelling a job for non-conformance, they also punish the employer, and that makes no sense to me. I recommended, and this was accepted, that as far as possible users would describe the resources they require - how much CPU time, how much printed output, and we'd change the job specification accordingly. About the only prescription I can recall was, that if they wanted tape drives, then they had to code "unit=tape" because otherwise we couldn't tell whether a tape unit was required. This limitation arose from the way JES2 worked, looking up a catalog entry caused JES2 to abend. As part of that project, for a time I ran a secondary copy of JES2, so there were two copies of JES2 running at the same time.

It was during this time the first personal computers (under various monikers) appeared, and I bought an Osborne 1 and then two APCs.

Time passed, my time at DSS ended and I moved to Melbourne where I worked for a time with Amdahl, for Fujitsu, and for Clem Clarke, author of JOL. The main highlight from that period was porting JOL to OSIV/X8. OSIV/F4 F4 was basically Fujitsu's version of MVS, close enough that it shared some bugs. X8, also called FSP, internally was a bit like F4 (and so MVS), but its JCL and many internal control blocks were very different. There was a requirement to port JOL from IBM's OS and Fujitsu's OSIV/F4, where it had worked for years, to OSIV/F8. The documentation was not bad, as far as Jinglish documentation goes - at least it was available, but a lot of the porting went, "I wonder what happens if we change that..." referring to system information in memory.

About here, my main involvement with computers changed to personal computers.