Who me? Greetings and greetings, dear readers, and welcome to the sunny spot on the interwebs that we like to call Who, Me? in which Reg readers share their stories of failed engineering tasks.
Apologies if that cheerful, upbeat attitude is off-putting to you. Your correspondent is experimenting with being a ‘Monday person’ and it feels strange.
Anyway, our story this week comes from a reader who we’ll regomize as “Howard” who many years ago (many, many years ago) worked with an OS kernel called MagicSix used in the labs of the Massachusetts Institute of Technology while he was a student. Howard had tricked an Interdata 7/32 system into using virtual memory “in a way its designers never intended” using the operating system – such was his skill.
After Howard graduated and completed further training, the MIT lab was upgraded to an Interdata 8/32 which, as I’m sure you recall, improved upon its predecessor by properly supporting virtual memory. Howard was hired as a consultant for the process of getting MagicSix working on the new machine.
What this entailed was installing the operating system on a disk pack on 7/32, removing the disk pack from that machine and installing it on 8/32, and seeing if it would work. Repeat as many times as necessary until it works.
Now you may be wondering what the term ‘disk pack’ is. We here at Who, me? they certainly were. Well, this was 1978, so hard drives weren’t what they are today. Disk drives of the era looked a bit like top-loading washing machines in Howard’s description, and the disks themselves were stacks of platters “resembling eight metal LPs,” he wrote—a comparison that may be no less baffling to younger readers.
Luckily he also provided a link to an image of a non-distinct system [PDF] of time so you can get the idea.
Moving a disk pack from one machine to another meant stopping the disk, screwing down the carrier (which would simultaneously unscrew the disks from the disk), and then lifting the carrier. There was a lock on the drive cover that, for obvious reasons, prevented you from opening it while the drives were in motion.
The stacks of disks were usually labeled on the top plate. The units Howard used were labeled with the MagicSix logo – an image of which he was also helpful.
Now the process of getting started on September 32 was not as smooth as Howard had hoped. Many hours of shuffling disk packs from one machine to another followed, with many failed attempts that meant the disks had to be returned to the first machine. It was very frustrating.
Around 3 a.m., after another boot failure, Howard opened the drive on the 8/32 and noted – for a moment – “Hey, I thought there was a label on this drive. Well, still a mystery.” He then placed the carrier’s lid…
Then: a hellish buzz and bits of plastic flying everywhere as the teeth on the cover hit the teeth on the disc pack and physics did the rest. It turned out that the lock on the drive was defective, and the reason Howard couldn’t see the label was because it was still spinning at 3600 rpm.
Luckily, Howard wasn’t the only night owl at MIT, and someone was there to show him where to find a vacuum cleaner to remove the bits of plastic from the driveway.
He eventually got the operating system working on the 8/32, by the way. And he told us, “I wish I could say I learned my lesson about not working while half asleep, but that would take another decade or two.” Oh yeah.
Has your tired self ever made a mistake that your alert self would never have made? Tell it with an email to Who, me? and maybe we’ll share your story on a future Monday.