It Must Be Time for Tea

Friday December 22, 2006

30 Years of Code

Thirty years ago this month I wrote my first program. I'd first seen the IBM 1130 demoed at my high school's open house a couple months earlier. This guy put in a deck of cards, typed a month and year at the console, and out popped the corresponding calendar on the line printer. I thought "Cool! I want to learn how to do that!". But it wasn't until soccer season was over that I had time to follow up.

My first program wasn't much--a few lines of Basic. But I could see how the principles could be applied to write more interesting programs, like the calendar generator. I was hooked.

It didn't take long to bump into the limits of that Basic implementation, like 2-character variable names. So I moved to Fortran. It was pretty neat, too, but there were some limitations imposed by the Fortran runtime system. For one thing, all the I/O was synchronous. And since the operating system didn't do any spooling, you weren't just waiting for the OS, you were waiting for the device. So it was slow. For another thing, it wouldn't let you generate arbitrary patterns with the card punch.

So that led to assembly language. You could do asynchronous I/O. And you could compose 80x12 bitmap images and punch them on cards (the 12 punch locations in each card column mapped to 12 bits in each 16-bit word).

By the time I was a senior, we had a 6800-based[1] microcomputer, which a couple of the other students had built from a kit. It had a keyboard, a CRT, a simple ROM monitor program, a speaker, and an I/O port that you could hook up to an audio cassette recorder. The audio cassette was the one storage device. You stored your source code--6800 assembly language--on a cassette. After the assembler read it in, you popped out that cassette and popped in a new one, and the assembler wrote the binary image to it. You then loaded your binary back into the system using the ROM monitor. It made the IBM 1130 Fortran look blazing fast.

So one of the first things I did was write a binary patch for the assembler. The patch added an option to write your binary directly to memory. Once we had that working, I had great fun coming up with algorithms for making different sounds on the speaker. I wish I'd saved some of those programs; I remember that some of the sounds were pretty interesting.

I hope I'm still writing code 30 years from now. Software is such a great blend of functionality and plasticity.

[1] No, that's not a typo.