Wouldn't the type of k (if it's not a primitive) be derived from the ss13 syntax of var/list/datum/foo/my_list ? Aka, the var type definition of the rhs? I don't see how there's any relation to v.

I agree with the var_type thing - however DM will not let you give a var type to v in the for(var/k, v.

Thinking about it I should create a var_type corresponding to an empty typepath and use that instead but I am not really sure how I can actually do that.

I didnt notice the parser had unit tests. I am more than happy to include the for(var/k, v cases. I included them because I wanted to show maintainers the cases that I had tested in the event that they had questions about it :)

I've added a basic unit test that shows a for(var/k, v in A) is parsed. I can add more specific error messages checking with some help -- I have no idea how to make rust check exactly that a specific error message got printed.

I've also added a block :

                // the "v" in a DM for (var/k, v) statement is essentially typeless.
                // There is currently no way to change that.
                let var_type_value = VarType {
                    flags: VarTypeFlags::from_bits_truncate(0),
                    type_path: Box::new([]),
                    input_type: InputType::from_bits_truncate(0),
                };

DM lets you type k but values are always essentially typeless and there's no way to change it as the moment. This has all the cases I lint covered in unit tests now.

this is fine for now