I hadn’t heard of it—that’s very interesting, thanks. :) It would be nice to use GADTs for Kitten’s AST, but last time I tried, it made inference complicated,

I hadn’t heard of it—that’s very interesting, thanks. :) It would be nice to use GADTs for Kitten’s AST, but last time I tried, it made inference complicated, and I never figured out a good way to encode Kitten’s permission system in Haskell types. Even just annotating Kitten types with their kinds (in the old compiler) was too cumbersome.