Description
Hi guys,
I am trying to add mypy support for dataclassy, which is a replacement for standard dataclasses.
You can see the code here:
https://github.com/biqqles/dataclassy/blob/15cef524b55df2c65b39d1619c7b8f1a3d28a813/dataclassy/mypy.py
One of the problems I've run in to is in supporting the feature of dataclassy that allows its dataclasses to be unpacked like tuples.
Firstly, I couldn't figure out how to represent tuple style unpacking in the mypy type system. If I create a class with def __iter__(self) -> Iterable[int]
, then it works, but I can't seem to do anything like Iterable[int, int, str, bool]
. When I reveal_type
for a NamedTuple.__iter__
I just see a generic iterator typing.Iterator[_T_co]
. I see that reveal_type((1, 'foo').__iter__())
is typing.Iterator[builtins.object*]
So I tried to figure out if there was special logic in semanal_namedtuple.py
but I couldn't see anything there other than adding Iterable
as a base in the _make
method(?) which doesn't seem like that I need. And, besides, I figure wherever the logic is it must also apply to regular tuples too.
What am I missing?