This repo currently contains a work-in-progress implementation of an interpreter for the Ludus programming language, using [Janet](https://janet-lang.org) as a host language.
Ludus is part of the [_Thinking with Computers_ project](https://alea.ludus.dev/twc/), run by Scott Richmond at the University of Toronto, with collaborator Matt Nish-Lapidus; Bree Lohman and Mynt Marsellus are the RAs for the project. Ludus is our research language, which aspires to be a free translation of Logo for the 2020s.
Ludus, like Logo, is meant to be a teaching language, often for students who don't think of themselves as "computer people." Our intended audience are humanities and art people at the university level (undergrads, grads, faculty). Everything is kept as simple as possible, but no simpler. Everything is consistent as possible. We aspire to the best error messages we can muster, which is important for a language to be teachable. That means being as strict as we can muster, _in order to be friendlier_.
Our current development target is Ludus on the web: https://web.ludus.dev. That wires what we do on the langauge interpreter (here in this repo) to a web frontend.
Statically typed programming languages generally give more helpful error messages than dynamic ones, but learning a type system (even one with robust type inference) requires learning two parallel systems: the type system and the expression system (well, and the pattern system). Type systems only really make sense once you've learned why they're necessary. And their benefits seem (to us, anyway) to be largely necessary when writing long-lived, maintainable, multi-author code. Ludus code is likely to be one-off, expressive, and single-authored.
To stay friendly, Ludus is dynamic. But: despite the dynamism, we aim to be as strict as possible. Certainly, we want to avoid the type conversion shenanigans of a language like JavaScript.
#### Functional
Ludus is emphatically functional: it uses functions for just about everything. This is both because your humble PI had his world reordered when he learned his first functional language (Elixir), and because the research into Logo and the programming cultures of MIT in the 1970s revolve around extremely functional Lisp code (i.e., Scheme). Logo is a weird little language, but it is a descendant of Lisp. So is Ludus.
Also, we believe that Ludus's immutable bindings and persistent or immutable data structures and careful approach to manipulating state lead to a lot of good pedagogical results. Learning a programming language involves learning how to model what's going on inside the computer; Ludus, we think, makes that both simpler and easier.
If you're looking for cognate languages, Ludus takes a _lot_ of design inspiration from Clojure and Elixir (which itself took a lot from Clojure). (The current--quick, dirty, and slow--version of Ludus is written in [Janet](https://janet-lang.org).) Clojure and Elixir are great! If you're asking why you should use Ludus instead of them, you're already at the point where you should be using them. Ludus is, maybe, for the people whom you'd like to work with in 5 years at your Pheonix shop (but even then, probably not).
Ludus is a scripting language. At current it does not have a good REPL. Our aim is to get interactive coding absolutely correct, and our efforts in [ludus-web](https://github.com/thinking-with-computers/ludus-web) are currently under way to surface the right interactivity models for Ludus.
Here, we use the `print!` function, which sends a string to a console (`stdout` on Unix, or a little console box on the web). Because `print!` returns the keyword `:ok` when it completes, that is the result of the last expression in the script--and so Ludus also prints this.