clarke@csri.toronto.edu (Jim Clarke) (06/01/90)
A couple of days ago I wrote: >Kind of like rerolling a big roll of paper whose middle has fallen out. ^insert "tape" >Basically, it's better to retype your program from scratch. >... Just ask me about Jodrell Bank! A couple of you took this bait, so I'll tell you about Jodrell Bank. Be warned you, though: you'd be better off working, if that's what you should be doing instead of reading news. I'd have posted in the first place if it were really worth reading. Jodrell Bank is a famous radio observatory -- perhaps the most famous, since non-astronomers can sometimes even remember its name without prompting, which is just about unique among scientific institutions. (Astronomy generally is *much* better appreciated by laypeople than either of the other sciences I've worked in, namely physics and computer science.) In 1976-77, when I was still a radio astronomer, I did a post-doc at Jodrell, which at the time was an odd mixture of old and new. (I have no idea what it's like now.) Their astronomy could reasonably be described as exciting and even daring, and they were building several new computers that were going to run a local version of Forth, which was very new at the time. (Forth is nifty for controlling telescopes, which is where it came from. But it's not so hot for calculations. There was an "underground" Fortran compiler produced by graduate students and postdocs who were unhappy with Forth. Imagine Fortran as subversive....) To control the main telescope and manage data acquisition, they had an ancient Ferranti computer with a drum memory that would fail frequently, sending Ferranti scurrying off all over northern England looking for cast-offs from sites that weren't quite so out of date. (Why Ferranti were still willing to maintain this dinosaur I do not know.) When this computer needed rebooting, which was frequently, you had to read in the operating system from paper tape. This may sound like a nuisance (it sure helps put in perspective the delay while a PC reboots from a floppy), but given the age of the system the method wasn't sur- prising. What *was* surprising was that there was ONE copy of the OS! And a worn, patched old one at that. Rolled up, it was about 10 inches in diameter; but if it had just been used, it lived in curls and snarls in a big plastic garbage bin. With paper tape, you of course roll from the end to the beginning, so you can't put it on a take-up reel. You have to wait until it's through the reader before you reroll it. After reading, the tail end is draped out the side of the garbage bin, so you can roll it up pretty fast if you're careful not to snarl it. (Fast rolling with a snarl leads to a snap, followed by careful splicing.) An interesting feature of the OS paper tape was that after you'd read it in, you had to read another, shorter paper tape to patch the errors in the main OS tape. The you had to toggle a few words in from the control panel to fix the patch tape. Back to managing paper tape ... A big worry when rolling up paper tape is having the middle fall out. You can't simply rewind starting at the end that's just fallen out, because the tape is all curled up. You have to get another garbage bin and UNWIND EVERYTHING, and then redo the whole winding operation. Observations were also recorded on paper tape. One night's observations would take up about one paper tape -- again, 10 or 12 inches in diameter. They were new tapes, so they were stronger than the OS tape. But for some reason the teletypes Jodrell owned didn't punch sprocket holes in the new tapes. Sprocket holes are the little ones about a millimeter in diameter that a little sprocket wheel grabs to pull the tape through the reader or punch. Since the teletypes had sprocket wheels in spite of the lack of holes, the effect was to make little bumps in the tapes where the wheel had tried its best to grab the tape. A long paper tape with bumps in it does not roll up tightly. A loosely- rolled paper tape a foot in diameter tends to lose its middle. After a week's observations, you could easily have twenty or thirty enormous tapes trying to spill their guts. A nightmare. 1977, you say? What about magnetic tape, you say? Well, it had been suggested, but the astronomers who controlled computing at Jodrell had decided to take a giant leap into Modern Computing, with disks, video displays and whatnot. But ... they were going to do it by BUILDING A COMPUTER THEMSELVES! And its architecture was going to be a copy of the old Ferranti's! I warned you this would be boring, so I guess I could legitimately go on forever. But maybe you've read enough. -- Jim Clarke -- Dept. of Computer Science, Univ. of Toronto, Canada M5S 1A4 (416) 978-4058 clarke@csri.toronto.edu or clarke@csri.utoronto.ca