Hisham Muhammad

> Anyway this issue is one of the reasons why TypeScript went with an unsound type system. There is an even better explanation [here](https://www.typescriptlang.org/docs/handbook/type-compatibility.html#optional-parameters-and-rest-parameters). Yeah, this is precisely the awkward...

That would imply reifying the enum in the generated Lua code as an array of strings, which would have the side effect of assigning a numeric index to each entry....

Both are interesting (and very different) proposals! @lenscas, what you are proposing is essentially a form of sum type, a.k.a. tagged union ([and so many other names](https://en.wikipedia.org/wiki/Tagged_union)). I always found...

> may I suggest that it may be beneficial to not care too much about backwards compatibility? Sure, its nice but teal hasn't reached 1.0.0 yet. To me that suggest...

> I wanted to create a visitor for a Teal AST by assigning a name in the AST node types enum to a callback that operates on nodes of that...

> discriminate across table types This will require metatable-equipped records (or objects (or some sort of tagged unions)). For reference, see related discussion in #97. > between enums | strings...

@virusdefender Thanks for the report! This is a tricky one. This is very much related to #71, except that that one is about input arguments to a function, and this...

> Maybe teal-lang can provide a more strict mode to check code? That's an interesting idea. The stricter check of arity could be a warning flag.

> Does this feel a little too structural for Teal? Teal is already a hybrid of nominal and structural, with function types being structural, because it's generally what people expect...

@Zaphyk one difficulty is that basically all Linux distros nowadays are x86-64, and the Linux build available is 32-bit. (Windows and Mac usually have an easier time running legacy 32-bit...