Hacker News Reader

Comment by Blammar | Hacker News Reader

My First Computer Was A LGP-30 at age 14.

5 bit Friden Flexowriter punching paper tape for coding in machine language. Coding sheets. Weird hexadecimal digits 0-9fgjkqw (it took me a while to adapt to 0-9a-f when I first saw them!) Ordering instructions so they'd appear under the drum reader when it was time to execute them. Spending hours debugging code to the point where I was flagged in the weekly publication as the person most likely to be on the computer.

The most amazing thing I ever saw on the LGP-30 was that someone had managed to implement floating point on it -- we got a chance to try it out near the end of our summer course. The second most amazing thing was a program to compute e to several thousand decimal places.

View on HN

FGJKQW makes sense because almost all the rest of the alphabet was used for the instruction set (one letter per instruction) so these were left over.

For the life of me, I can't figure out why ORDVAC use KSNJFL. If you Google around you'll find the claim that this was because in figure shift mode the input teletype produced 01010 to 01111 in binary on the tape for those letters. Except, I can't find a 5 level code that actually conforms to that. It's certainly not Baudot as some claim. And if you look at the manual the claim about figure shift isn't even true for the actual code used. https://www.ideals.illinois.edu/items/55048

> https://www.ideals.illinois.edu/items/55048

On p4 the code has KSNJFL punch binary 10–15. It starts with a permutation of Baudot ITA2 (notice the QWERTYUIOP/1234567890) to make 0–9 punch their values.

Yes. But why that permutation? And although there is a relationship (like Q and 1 being on the same key) there isn't a relationship with Baudot or ITA2 (e.g. Q/1 isn't 00001 it's 11101).

In case you're not already aware (and/or for readers who aren't, because I wasn't yesterday), they made substantial modifications to Model 15 Teletypes to get their chosen coding and key layout (numeric-pad-like rather than QWERTY), including modifying and permuting levers, type bars, and keycaps. It could be that KSNJFL was somehow most convenient in that respect, though it's hard to imagine.

https://archive.org/details/modificationofte56mill

Yeah. It's pretty interesting because they clearly did this super intentionally. As you say they moved everything around. Since they are using the existing keycaps which have Q/1, W/2 etc. they lose QWERTYUIOP but they've still got ASDFGHJKLZXCVBNM from which to get the hex digits.

The one clue I haven't explored deeply is that there are references to ORDVAC and ILLIAC I using the same teletype equipment as prepared for the Institute of Advanced Study (IAS) Machine (i.e. von Neumann's machine). These machines all have the same principles and it's possible that these teletypes (prepared by the US Bureau of Standards) are the source of KSNJFL.

My late father used to be a service technician with Royal McBee who sold the LGP-30 in Europe. Before I got my ZX-81, he gave me the LGP-30 manual as an introduction to programming. He had some stories about programmers optimizing drum memory access and what headaches this caused when the timing of the motor was slightly off.

Reply on news.ycombinator.com