posted by Matthias Felleisen and Matthew Flatt
The First Thoughts [Matthias]
Wednesday 25 January 1995
If functional programming is that good, why is nobody using it?
Thursday 26 January 1995
We all met in Corky Cartwright’s office because it was bigger than mine. I announced that I wanted to leave theory behind and build a curriculum, a programming language, and support software to use functional programming for teaching algebra, more math, and programming in K-12 schools.
Shriram embraced the idea on the spot. “I have always wanted to be involved in such a project.” Bruce loved the chance to design a new language. Cormac made clear he’d help but he really wanted a dissertation. Matthew was quiet. Corky later dismissed the idea. “Make sure you have path back when it fails.”
We discussed for quite a while and came to two conclusions: the PhD students needed to focus on software issues for dissertations in case we’d fail. Bruce and I would focus on the language and the curriculum.
Friday 27 January 1995
Bruce and I started designing a language called Jam. It was supposed to be simple, Scheme-ish, and without parentheses because we were convinced that teachers would not like the parenthetical notation. At the same time, we were sure that the syntax should be basically “parenthetically simple.” But most importantly, we wanted to have image-like parts in this language so that Sum, Integral, Indexing, and so on would look as much as possible like the math text books.
In parallel, all of us discussed that we’d use Scheme to build this software ecosystem.
Saturday 28 January 1995
A few weeks later I found out that over this weekend Matthew had started to “cobbled together MrEd.” I am sure he didn’t use “cobble” but something close. And it probably took a bit more than this weekend. What I am pretty sure about is that the first key strokes of conception probably took place on that day (cross-producting my memory with my understanding of Matthew).
The First Keystrokes [Matthew]
I don’t remember the different meetings. Maybe it was the second
meeting where I suggested the current user interface—
In any case, I set out to build a user-friendly environment. The main problem, it seemed at the time, was to build a GUI text editor that could handle modern entities like pictures and live objects. I picked wxWindows as a starting point, because it seemed like the most promising cross-platform GUI library, and libscheme as the Scheme implementation, because it was easy to embed. I figured that the hard part of a text editor was making it run fast enough, so I wrote that part in C++.
The program was called “MrEd” because it was mostly an editor. The name “ed” was already taken. Among the words you get by adding a letter to the front, only “red” was appealing for whatever reason, but that name was also taken (as “restricted ed”). Adding one more letter arrived at “mred.” I liked the “Mr” part and didn’t mind that it was already the name of a talking horse.
Although the editor core was in C++, I at least knew enough or had enough direction from the group to aim for Scheme for the rest of the environment’s implementation. So, libscheme became not only the vehicle for running student programs, but also the language for implementing much of the GUI itself. To make that practical, I started changing libscheme: adding a built-in object system that supports extension of C++ classes, adding compilation to an AST instead of interpreting S-expressions directly, and so on. After a few months, I became attached enough to this part of the implementation to break it out as a separate piece. I called it "Ms. Scheme", but since "Ms" seemed too likely to suggest "Microsoft", spelled it "MzScheme". (Around 2001, we started using the umbrella name "PLT Scheme" and moved away from adopting cute names for subsystems. "PLT Scheme" became "Racket" in 2010.)
That’s why the oldest entry for HISTORY.txt in the gui-lib package is “Version 0.7: May 10, 1995,” while the oldest entry in HISTORY.txt for the racket collection is “Verion 0.27: September 26, 1995.” The editor came first, and the language was something of an afterthought.
Of course, the language quickly came to dominate the idea—
The First Users [Matthias]
By summer ’95, I had designed the basics of what is now known as the design recipe and a few days worth of teaching material. We recruited four teachers and one of Corky’s PhD candidates to test this material with a week-long workshop.
I started with Chez Scheme and Emacs.
During the week, Matthew demoed a rudimentary IDE based on MrEd. I was suitably impressed and tried MrEd on the four participants, with simple images and all. It didn’t go too well, but we had the first four innocent users try out our very first teaching environment.
And that’s how Racket launched.