hckrnws

My little language Newt is 7 kloc. Dunno if it's worth including, it's mostly an exercise to learn how these things work and is not as polished as I'd like.

- Self-hosted

- Compiles to javascript

- Bidirectional typechecking with NbE (based on elaboration zoo)

- Dependent type checking

- type classes

- ADTs with dependent pattern matching

- TCO (trampoline for mutually tail recursive functions)

- Erasure of compile-time only values (0, ω quantities, but not linear)

- Web playground

- LSP (added this month)

- Syntax is similar to Agda / Idris / Haskell

https://github.com/dunhamsteve/newt

Another crazy one is SectorLISP, 223 lines of asm

https://justine.lol/sectorlisp2/

Fluent – 4K lines – including parser, interpreter, standard library, IDE, UI, docs, examples. Will grow though.

https://github.com/mlajtos/fluent/blob/main/client.tsx

I'll add it! Thanks.

EDIT: Actually, it's not quite "ML-family" enough for this post. But it is a remarkably cool project! :)

If you're accepting additions here is a fun one some friends and I did as experiment at the Topos Institute: https://github.com/ToposInstitute/polytt

And here is a set of single file lambda calculus implementations with a variety of extensions: https://github.com/solomon-b/lambda-calculus-hs

`polytt` is kind of an ended experiment but that lambda calculus repo i plan to extend in the near future.

The hardest part with small languages isn't the parser, it's the standard library and error messages. Getting a helpful IDE experience in that footprint is a significant engineering challenge.