🚀 go-pugleaf

The 'blackjack program that cheated' brings to mind another (true,
dammit) story from when I was just a pup.

A large eastern university had a 7094 (1 mip, 1 megaflop, 32k 36-bit
words, the true 'power machine' of its day) that had a 1410 used as a
peripheral spooler -- the 1410 read the cards and printed the output,
and the 7094 could crank away.  The system started out using tape as
the medium of transfer, but by the time I showed up, had migrated to
a more elegant solution - an disk drive cabled to both machines. No vendor
software -- this used a homegrown monitor that allowed the 1410 to print
on three printers, read from two card readers, and read and write on the
disk, all at once.  Pushing the envelope for the early Sixties.

This homegrown nightmare was maintained by 'Hal'. 'Hal' showed up
at night, patched the monitor, handled modification requests, chatted
with the operators, and left every morning when the machine was given to
the CE's for the daily PM. Not seen by the rest of us mortals, but
productive and happy.

The university had a 'use it or lose it' policy on vacations for the non-
academic staff, so 'Hal' took off the week of his birthday and Christmas
week every year.  He explained that it was not right to work on his
birthday.

One morning the CE's finished PM and turned the machine back to the
operators.  They hit the load button, typed in the date and time
(machines then did not remember such things), lights blinked, fans
whirred and the machine stopped.  The operators turned the machine back
to the CE's, who reran all of their diagnostics.  Every diagnostic passed.
Load, type date and time, whirr, blink, and stop. Back to the CE's.
Pass. Back to the operators. Fail. By noon, someone suspected that
there might be a problem in the monitor. Reload the monitor from tape.
Load, type date and time, whirr, blink, stop. Call 'Hal'. No answer.
Someone realized that he was on vacation. No one else knew beans about
the monitor.

By late afternoon, a card reader and a printer are cabled to the 7094,
and the day's backlog is being handled, slowly. (The 7094 handled
the ideosyncrasies of the printer by means of timing loops). All of the
previous day's output that had been spooled to the disk is inaccessable:
the 7094 can write print images to the disk, it can't read them.

Six am the following day, the machines are given to the CE's. Eight am,
all diagnostics passed, the CE's give the machines back to the operators.
On a whim, one of the operators tries the 1410. Load, type date and time,
whirr, blink, READY. We now power down again to attach the card reader
and the printer back to the 1410. Life continues, albeit with a day out.
No one thinks to do more than note the problem in the monitor problem log.
'Hal' returns from vacation the following week, on his usual schedule.

A year later, 'Hal' is gone, and my mentor was given the task of
supporting the monitor.  She inventoried the disk, and found a
module in the startup sequence that was not listed in the documentation.
She could not find source for the module, so she disassembled it by hand.
If the date was 'Hal's' birthday, halt. It wasn't right for the
machine to be working on 'Hal's' birthday, either.
-- 
Elliott Frank      ...!{hplabs,ames,sun}!amdahl!esf00     (408) 746-6384
               or ....!{bnrmtv,drivax,hoptoad}!amdahl!esf00

[the above opinions are strictly mine, if anyone's.]
[the above signature may or may not be repeated, depending upon some
inscrutable property of the mailer-of-the-week.]