Gnome's Computers - High Life...
This software house started out with just 2 people! I've been asked by my fans to expand a bit on my experiences in 'the business'. Hope it's not too boring.
As I wrote previously, my days at Bedford Computer Service lasted only 6 months and the company was taken over and 'asset-stripped'. The next job I had came to an end when the funding ceased; and I was invited to join a 1-man band to program a time-sheet system for a local accountancy firm.
This was programmed in Cobol, a good language for text-handling but very tedious to write because of all the pencil-pushing. After 3 coding sheets you might have the initial declarations done and start the actual program.
After writing the program, it was passed to a 2-girl sub-contracting firm (who later came to work in the same building as us) where it was punched onto cards - perhaps 500 to 1000 of them. Accuracy was very good as they were all verified by re-typing it all on a checking machine. But bad handwriting would cause copying errors.
The accounting system was a success, and the firm's several branches sent in weekly handwritten time-sheets for us to process and produce an analysis of man-hours spent by each accountant on clients.
This phase of our work has been covered in 'Molecular 18': so I'll pass on to our CBM days with the ledgers packages. Our original trio were together again.
I had to write the main utility subroutine library for the 3032 and later the 8032. There were 3 main sections to this library. Disk access, printing and 'general'. In the disk section all our variables began with the letter X, which was not allowed anywhere else.
If you wanted to open a file, you made XN$=<file name>, XL=<record length>, XD=<drive number>, XN=<record number> and so on. Each subroutine was thus "parameter driven". A table was kept in memory of which files were open at what record.
Similarly, for printing you used the letter Y for the variables. YP%=current page number, YL%=line no. etc. If you wanted 2 blank lines, you wrote YA%=2: GOSUB 50010. A table of preset headings was maintained, for printing on each page. YP defined the printer-type used.
General utilities used letter Z, which was also used for work-variables and not allowed elsewhere in programs.
It took about 4 months to complete the first version of this library, while my colleagues designed and started writing the main programs. They couldn't begin much coding until the library was at least designed.
Our leader was John L. and he attended to company business and most designing. Mike and I (John M.) did the coding. One day Mike showed me a page of coding and asked, "What does this do?" I looked at it slightly irritably and said, "I don't know. Looks like a load of rubbish." He seemed surprised. "But you wrote it only last month" he said. Mike had a wonderful memory for details, while for me it was an assembly-line job. Finished work was passed on, and a new job went on the stocks.
The 3032 had a 40-column screen. When the 8032 (80 column screen) came along we had to do a lot of redesigning of screen layouts. We also got an 8050 twin drive with full support for relative files. And in this new Basic (4.0 I think?) you could ask for disk status by DS (error no.) and DS$ (error message). Up to now we had been using direct-access files which, though quite easy to use, had two snags. First, a V0 command deletes all files not in the disk dir. Not all users read our dire warnings not to do this. Also there was a problem with expanding a file if it got nearly full. I don't think relative files could be used on our earlier 3040 drive (though they could on a 4040). Anyway we never heard of relative files until our 8050 came along.
Relative files made a big difference to the speed and efficiency of our programs. Not many people seem to use them. People seem to think that they're difficult and unreliable. This is only because they don't follow the rules, which are vital. Some rules are not well explained except in CBM supplements not generally released. You use fixed-length records: and you can request any record by its number, and read from any byte. All you need is some algorithm to convert things like customer or part number to a record number. Access is very quick.
After about 6 months of this, we merged our company with a small Commodore dealer; a man who had been selling PETs from his garage, with considerable success. We left our nice office on the ground-floor of our accountants and moved into tiny premises in the middle of Bedford. You've all seen the sort of place:- you go up some wooden stairs and try to find which door to enter. We had 2 rooms. Mike and I were in one above an alleyway through to a back yard. The floor was perishing cold but there was a powerful gas fire which roasted us.
Happily, business was good - more for the hardware than the software at that time. Later the situation reversed. We moved again into a complete floor above a finance-house. We took on another programmer and an office-boy who did the post, despatched disks, answered the phone etc. You could have held a dance in the programmers' room.
We had no end of applications for the programming job and several came for an interview. John L. was showing one round and was asked "Who's the little old man in the corner?" It was about then that 'old gnome' started to be used to refer to me. My father, who lived to 93, was also rather puckish.
I've mentioned the CBM 8096 and 700 in 'Onwards to CBM'. It was the cockup of the 700 which turned many dealers off CBM altogether, and I dare not say much about it for fear of libel actions. But it's well-known that some dealers were sent empty 700 cases to display. And at least one of these was sold to a customer because the dealer hadn't realised it was a dummy! We had one of the first working models, direct from CBM. We discovered that some of the operating system was missing, and spent a morning at Commodore, at Slough, where a rather agitated engineer tried out dozens of chips to try and find one without bugs. More were arriving daily from the U.S.
We had a stand at the Commodore Show that year, and I met a Commodore manager on the basement stairs. I was told that another, upgraded chip had just arrived from the States. A few minutes later I attended the 'Programmers Workshop' meeting, chaired by Jim Butterfield. During discussion, I said that a new 700 chip had just come. There was a sudden silence. Who was that gnome who seemed to know the latest CBM info? It made my day and I smiled gnomishly.
The dealership part of our company went through a dodgy time, but managed to get IBM and DEC approval, which enabled them spread out. The software part of the company, now about 6 people, split off - for various reasons - and are still doing well as FACTSSOFTWARE. They sell MICROFACTS, EASYFACTS etc. but you won't find them on CBM, except perhaps on the PC.
The dealership were bought up by 'The Computer Centre' of Peterborough, who have opened more branches in the East Midlands. I'm delighted to see both these businesses flourishing still.