New Issue: dyno: should 'nil' and 'none' have their own AST nodes?

danilafe · April 15, 2024, 4:23pm

24851, "DanilaFe", "dyno: should 'nil' and 'none' have their own AST nodes?", "2024-04-15T16:22:54Z"

dyno: should `nil` and `none` have their own AST nodes?

opened 04:22PM - 15 Apr 24 UTC

area: Dyno

While working to implement resolution for `none`, I noticed that it would be tre…ated distinctly similarly from `true` and `false`, despite being quite similar. Specifically, there's special if-else logic to handle looking up the identifier 'nil'. However, for 'true' and 'false' we just have "literal" nodes. Since `none` and `nil` are both reserved, it seems like it might be cleaner (and more declarative) to represent them as their own AST literals, and add AST traversal logic for such literals instead of going through identifier machinery.

While working to implement resolution for none, I noticed that it would be treated distinctly similarly from true and false, despite being quite similar. Specifically, there's special if-else logic to handle looking up the identifier 'nil'. However, for 'true' and 'false' we just have "literal" nodes. Since none and nil are both reserved, it seems like it might be cleaner (and more declarative) to represent them as their own AST literals, and add AST traversal logic for such literals instead of going through identifier machinery.