UCSD Pascal

From AboveTheGarage

During my tenure as "lab manager" of the DCOS Lab I was given the task of making the UCSD p-System run on a Lockheed SUE minicomputer.

The p-System was brilliant for the time (1980?) and as a result of this project I met some of the people who created this thing down at UCSD. They were very bright.

Interestingly, a number of people from the DCOS Lab went as a group and started a parallel computer project at Sperry Univac. They chose Pascal as their implementation language - which was fairly ballsy at the time, as almost all OS work was done in assembly language. They started (I think) with the UCSD Pascal Compiler and converted it to generate machine language for a couple of machines. One machine was an existing piece of hardware from Sperry and the other machine was hardware they designed with a built-in ring network and other features they had developed on the DCOS project.

Someone at UCI - probably Julian Feldman - thought it would be cool to port the UCSD system to the Lockheed SUE hardware we (UCI) already owned. The Lockheed SUE minicomputer supposedly got its name as a dare to Digital Equipment Corporation to sue Lockheed over the theft of their instruction set. But the SUE computer had one major difference from the PDP-11: the byte addressing was swapped.

So after many months of toiling in assembly language I was able to boot a binary image of the UCSD Pascal system on a Lockheed SUE minicomputer. Everything worked! You could run the editor and the compiler and it was pretty amazing. Well, one - exactly one - thing didn't work. That was a specific bit testing instruction. Pascal had the concept of a set which made testing groups of bits very easy. Somewhere in the compiler was some code that bulk added a pile of bytes onto the stack and there was no way I could ever figure out if this pile of bits should be byte swapped. I think the fundamental issue was that bit addressing and byte addressing at that level were in conflict.

Well, nobody much cared about this except me. But one day, Greg Hopwood, one of the geniuses behind DCOS and the new Sperry system came by and saw what I had done. This was a guy who had dug deep into the compiler and was aware of the issues they had bringing up Pascal on the old and new hardware at Sperry.

And he said, "You did this without changing the compiler at all?"

And I said, "Yup".

And he said, "Amazing", and I had a great feeling of accomplishment.

There's nothing in life like recognition from a guy like that.

w00t.

-30-