aith VS hackett

As for uses of kinds in general, I'm using them a lot in Aith. Except that, instead of using higher kinds I'm parameterizing my Type to allow for fancier classification then just "this is a type". I'm using kinds for

Aith is another new one.

https://github.com/Superstar64/aith is also an interesting language, with substructural typing. The creator is also in the discord server if you have any questions

In Aith after doing hindley milner type checking, I want to annotate my ast with type annotations for several reasons:

It depends on how powerful you want your macros to be. In aith, my macros are just a compile time lambda calculus. My macros can only generate values or other macros, this limits them to being no more powerful then what you can normally do with functions but it (will when my language is usable) let me write code that I know will be inlined and edsls that compile into fast code. With these limitations I can completely type check macros ahead of time (because they are just like funcitons) and I can also use my kind system to prevent macro types from leaking out into normal ones.

I am not super knowledgeable about this, but I think you might find Type Systems as Macros interesting. There is also the in-development language Hackett which uses the approach described in the paper to unify Racket style macros with a Haskell style type system.

I agree. I've been searching for solutions for a while. A few choices: - Common Lisp Coalton, very similar to Haskell. - Hackett, a Haskell-like DSL implemented in Racket. Licensing would be an issue, so it would have to be ported to Guile Scheme if you want to build an Emacs out of it. This is not easy since it takes advantage of several Racket-specific language features. - Shen, which can be built on top of Common Lisp, Scheme, Racket, or even Emacs Lisp. The drawback is that it is a fairly cryptic language, and extending foreign language bindings is not well documented. You would basically have to program the entirety of Emacs from scratch - PreScheme is a statically-typed (Hindley-Milner family) subset of Scheme that compiles to C. Originally written to build the Scheme-48 compiler, it is being ported to Guile. Not production read yet. - Zile is an editor engine built on Guile 2.0. But there is no static typing or algebraic data types, it is simply a replacement for Emacs written in Scheme from the ground-up. It needs to be ported to Guile 3.0. Guile 3 has an Emacs Lisp interpreter built-in, but it needs to be developed further before it could run more popular Emacs Lisp applications like Org-Mode or Magit.

Meh. As my final part in this exchange I will leave this here, conclusion are left to the reader: https://lexi-lambda.github.io/hackett/

My dream is something like Hackett, but alas, Alexis didn't have time to continue it. Someday ...

The documentation says it does: https://lexi-lambda.github.io/hackett/

But we can always import Coalton, or Hackett, or miniKanren into our Lisp program if we need it.

This doesn't really answer your question but you may be interested in checking out https://lexi-lambda.github.io/hackett/ by u/lexilambda.

Extremely difficult, but you can do some cool stuff with macros if you do it. Have a look at Alexis King's "Hackett" language for a cool example: https://lexi-lambda.github.io/hackett/