Absolutely it is a good approach, and I imagine it is quite a bit cleaner than the good ol fashioned switch block. Personally I havent approached it that way yet :)

@@Logan-mj3wxIt certainly looks nice, and I'm using a macro to instantiate my token trie and enum. Although the latter process of selecting between operators/keywords, number literals and identifiers has a bit to go before i could call it a pretty sight.

I'm actually working on a grammar for this, but chances are I will use dedentation for nested constructs :( The important part is leading every statement with a keyword/operator

If you want to have scope you would have to define it somehow, be it by keyword, file, spacing, etc. Otherwise the program would have to be in a global scope.

@@Logan-mj3wxIt's a simple ML/OCaml inpired language, and every statement is prefixed by either a keyword or an operator, which means the compiler could understand scope based on expansion rules. Still there are issues with nested structures as if you were saying that every time a `let` keyword is encountered it is a new statement, then a nested function would not be possible. This particular example can be overcome with lambdas, but other constructs like cases may have to have dedentation.

@@James-zn8nzAwesome stuff, you are braver than I to create something that is closer to lambda calculus. I love OCaml though, every couple months I apply to Jane Street I want to work with it 😂

@@Logan-mj3wxFunctional languages actually seem quite easy to build. No mutability means simple functions and composition means simple production rules. I'm sure it'll be more difficult than that though. Initially there will be no complex typing system, just the built-ins, however I do plan on building some dependant type magic at some point. Also, I have written a lot more Haskell than OCaml, but I prefer some of the OCaml syntax to Haskell's. In particular, it's the indentation-sensitivity that I don't like. Also, not having to repeat the name of my function tens of times is quite nice.