[FlingPoo]

I was an "operator" on a Tandem NonStop. First was Pitney Bowes, I preferred the night shift. In the computer room, there was a Tandem, and a UNIVAC. My job was mostly swapping disk packs on the Tandem, and swapping tapes on the UNIVAC.

One of my jobs was to shutdown the dialup system (customers mailing machines would send data about the mailings they made overnight). I had to shutdown the dialup lines according to the time zones across Canada. I decided to write a script to automate it. My script would shutdown each phone line in order. The first time I ran it, I "broke" the Tandem. My script was a basic loop. It would check the time, and shutdown one of the timezones. First time I ran it the Tandem "mainframe" froze. I had to call "Doug" in the middle of the night, I was freaking out. Doug came in, looked at my script and quietly pointed out that the time command was a high priority system. My script didn't have any "waits" in it, the loop I wrote was constantly asking the system the time, taking up 100% of the processing. "Doug" had to reboot the Tandem, and after the "wait" was put into my script, all was good.

After I left Pitney Bowes I was an operator at the OPP (Ontario Provincial Police), that system had the OMPPAC system on it (Ontario Municipal Police Automated Cooperative), The criminal database for the Province of Ontario.

[fuzzfactor]

Things are different when "mission-critical" is the highest priority.

Before PC's really got popular, I went out to a petroleum pipeline installation one time where they were just unboxing one of these Tandem rigs.

They were going to use it to further automate, control, and account for transfers like some of the other oil companies were doing with their mainframes. As part of an expected technology advance at the time.

Here it was not just a CRT terminal and a printer, but the whole thing right there in a fairly hazardous location in the blockhouse office where contractors would do their hand calculations.

I was there to take readings on the mechanical totalizers, especially on the piping section we had independently calibrated, and bring samples back to the lab for precision viscosity and density determination to more decimal places than available elsewhere. Along with all kinds of other routine and research parameters.

Turns out I was the pioneer in digital densitometry among the multinational contractors. That's another story altogether but within a decade they all had it and I was in more demand after the niche had grown than it was when I owned the niche. People still never want me to stop.

Anyway, I had a pretty good handle on floating-point error and was doing my part to reign it in with improvements in physical measurement.

It didn't take long to realize that my Atari would be basically capable of handling all of the things they were going to use the Tandem for.

The shortcomings would be the redundancy/reliability and Atari just couldn't count that high :)

When you're moving large numbers of barrels the numbers go through the roof when you convert to liters or even worse, some currencies.

If the figures didn't agree very well with manual calculation using 16-digit calculators, some big shot may very well hit the roof.

I would have had to hook two Ataris together and try to get more precision somehow at the same time as try some redundant reliability. Never did.

Although within a couple years I did hook up two TRS-80s together and they were quite adversarial . . .