Emacs and Lisp

The first editor I learned how to use was TECO on a line printer. You’d print the line of code you were on, then you’d issue commands to move the cursor around. You tried to avoid printing the line because that would be wasting paper. So you’d move the cursor around blind until you thought you got it to where you wanted, and then you’d start inserting characters. When you thought you had it, you’d print out the edited line.

Another undergrad saw me struggling with this and asked why I wasn’t using vi. I had never heard of vi and I was amazed that you could view the code on the screen and move your cursor around visually before going into insert mode and adding text. With vi I was orders of magnitude more productive than with TECO.

When I came to the ’tute in the early 80s, I found that computer accounts were only routinely issued to students taking computer science courses. I hadn’t decided on a computer science major, so I didn’t have an account. However the Student Information Processing Board would give out a Multics account to interested students, so I signed up for that. The Multics terminal was in the basement and it had a dial up connection with an acoustically coupled modem: two rubber cups that the handset would cradle in.

Everyone at the AI Lab used a home grown editor called emacs, and there was a Multics port. I abandoned vi and learned emacs. The paradigm was different, but I didn’t see one as superior to the other. When I declared computer science as my major, I got an account at the AI Lab on the Decsystem 20 machine. The editor was TECO with the editor macros (emacs) package loaded.

When I took S&ICP, we had a lab with HP9836 “Chipmunks” running MIT Scheme. The Chipmunks were PCs, not time-shared, and each acted as its own Scheme machine, complete with an emacs clone for editing the code.

At the AI Lab, there were a couple of machines running ITS, the Incompatible Timesharing System. You could run Maclisp on them, but Maclisp was such a pig, using over a megabyte of RAM, that a couple of instances of Maclisp would bring the machine to its knees. The Lab had developed the Lisp Machine, a single user computer that would run ZetaLisp (the successor to Maclisp). In addition to Lisp, the Lisp machine ran the ZWEI editor. Zwei Was Eine Initially, and Eine Is Not Emacs, but rather an emacs clone written in Zetalisp.

ZWEI was integrated with the Lisp environment. You could insert Lisp objects in the ZWEI editor and their printed representation would appear in the edited text. The printed represention was mouse sensitive and had its own context menu.

If you weren’t using a Lisp Machine, your options were Unipress Emacs and Gosling Emacs which you could run on this new OS called “Unix”

Around this time (in the late 80s) a hacker named Stallman decided to write a new OS. His first task was to write an editor and he decided on a new version of Emacs written in its own Lisp dialect.

If you wanted to use Lisp, you interacted with it via Emacs.

These days, I use GNU Emacs and I load up the sly package. Sly is a slime fork and it turns GNU Emacs into an IDE for a Lisp running in another process. The interface gives you much of what you used to get when using ZWEI on the Lisp machine. You can evaluate subexpressions, macroexpand and compile programs, gather output, and run the Lisp debugger from within GNU emacs.

Emacs and Lisp co-evolved at MIT and Emacs has always been used as a front-end to Lisp. I’ve never gone back to vi, and when I’ve had to use it I’ve found it frustrating (but this is because I have forgotten everything about how to use it).

I understand that some people find Emacs impossible to use and have a strong preference for vi. Is the experience of hacking Lisp in vi any good?