Parser generator performance

[teofr]

The Makam parser generator takes too much memory. This started as an issue for Circle, but is now an issue for local building as well.

Even if some quick fixes could help, I think the ideal solution would be to generate it on Rust (or even JS) and do some interop. It's also nice to only mantain one grammar.

I'd need to:

• Understand how JS <-> Makam interop works.
• Understand how Rust <-> JS interop works.
• Make the parser on the Rust side (and the AST) an independent library (this is also nice for the project)

cc @aspiwack @astampoulis

[teofr]

I'll leave this for now, but just documenting how the Makam <-> JS interop seems to work (as far as I could tell).

There's a builtin js.eval : string -> string -> prop that evaluates the first string (JS code) and returns the result as a string. Then, you would use refl.fromstring : [A] string -> A -> prop that relates a string (a Makam expression) with its actual expression.

[yannham]

Closing old stale issues about Makam.