programming an os

AlephZero

In the beginning (and before my time) ... http://www.bbc.co.uk/news/technology-20395212

The first computer I ever programmed was an IBM 1620, which also worked using decimal arithemtic not binary, and was way ahead of its time in doing arbitrary-length decimal arithmetic in hardware. IIRC, multiplying two 100-digit numbers (with one machine code instruction) took about 2 minutes elapsed time.

jtbell

We had one of those, too. I wrote a program to calculate factorials on it. I still have a printout somewhere of the result for 3000!, with all of its digits. I think it took something like six or seven hours to calculate. The display lights on the console cycled through a repeating pattern for each multiplication, taking more and more time for each cycle as the numbers became longer and longer.

As a bonus, I could put a transistor radio on top of the cabinet, tune it to an empty channel, and listen to the "music" broadcast via the RF radiation that the computer emitted!

harborsparrow

This article can give you a quick overview of that an OS is, and how people had to program before operating systems existed: http://en.citizendium.org/wiki/Operating_system

rcgldr

From that article:
Batch job systems in the 1960s - IBM 360 series, and JCL (Job Control Language) -
Batch operating systems could only execute one program at a time.

1960's batch operating systems such as OS/MFT and OS/MVT could run multiple programs at a time, mainly as multiple parallel batch job streams, but most IBM 360's (all but the 360/67) didn't have virtual memory (no address translation), which limited the functionality. There was also SABRE, a multi-terminal application used for airline reservation systems.

wiki_ibm_os_history.htm

The CDC 6000 series pre-dated the IBM 360 by about a year, and were mostly used for scientific type environments. These had multiple peripheral processors. I only saw them at a data center, and don't remember much about the operating system, The CRT monitors were unusual in that the images were effectively "plotted" on the screen (sort of like an oscilloscope).

http://en.wikipedia.org/wiki/CDC_6000_series

http://en.wikipedia.org/wiki/CDC_6600

Other mainframes of the 1960s: Burroughs, Honeywell, Sperry, UNIVAC, ...

Mark44

With regard to the class I took in '72, where we had to use a keypunch machine, you might not have been all that far behind. I took a class in Fortran in the summer of '80 at the Univ. of Washington, and we used keypunch machines again. The next time I took a programming class a couple of years later, they had Digital Equipment Corp VAX minicomputers with terminals, a big improvement over keypunch machines.

ImaLooser

At Harvard they had/have an Aiken computer in the lobby of one of the labs. No one but me ever seemed to pay any attention to it. It was about 50 feet long and built like a tractor. It even had a big drive shaft with a universal joint for all those mechanical relays.

http://history-computer.com/ModernComputer/Relays/Aiken.html

Programmed with paper tape and rheostats.

TylerH

If you're interested in learning about programming modern kernels, the OSDev wiki is the best free place to start.