Tin Tvrtković

Definitely an interesting idea, but `Foo.bar.baz.qux` cannot work (since `Foo.bar` is unset for dict classes, and occupied for slot classes). And nested `fields()` get ugly.

Seems to work for me. The cost/benefit analysis seems in favor of adding this: it's very easy for us to do and IDE support is very useful, particularly if you're...

I don't think we do anything special apart from raising a RecursionError. Serializing recursive data is tricky and requires a completely different schema approach (you need to switch to some...

We should definitely do something to make the hashing behavior clearer. I'm currently working on a PR to "fix" slots hashing behavior, and ``attr.s(hash=False)`` is really throwing me off. The...

What functionalities are we aiming for exactly? I'm guessing: * make hashable by generating a hash function based on attributes * make hashable, but attributes are opt-in instead of opt-out...

> That would be a thing for the mypy plugin, no? Yes probably, did I file this in the wrong place? I'm not sure where the mypy plugin is developed.

This is above my head too but the potential benefits are too great, I guess I'll have to start reading 😇

Just so we're clear, I want to express my continued appreciation for your work on this @euresti . If we can get this done I think attrs could basically become...

Has there been any movement on this? If we can get this working, we would be the basis for the first Python orm/odm with type safe projections. If not, I...

> Though I don't remember why I just closed the PR. I must've been in a fugue state. As are we all from time to time