Carl Meyer

Hi Ben, At first glance these do look like bugs. CheckedDict with dynamic key or value should translate to object at runtime (though not in the compiler; object and dynamic...

Ok, this turns out to be kind of a nasty issue. The `expected type or Optional[T]` error comes from the runtime construction of a `CheckedDict`, and at runtime it is...

One other note: the second part of the issue (`d1: CheckedDict[str, object] = CheckedDict[str, Set[str]]({})`) I'm not sure we'd consider a bug. In the compiler, `dynamic` and `object` are quite...

> does the runtime need to throw an error I guess that's a judgment call, but I think it would be kind of confusing to allow just anything at all...

Yes, thanks for the report, this is another consequence of the fact that our priority has been on acquiring sound type information to optimize bytecode, not as much on ensuring...

> would our model be consistent with a future Static Python? Yes, we intend to support the relationship between operators and their double-underscore methods in the future, but we don't...

So the typed/untyped code thing is a bit subtle. The meaning of `__static__.int64` is not the same in typed vs untyped code; in untyped code it is already just a...

> what other classes have different meaning in typed & untyped code? Only the CTypes: so the various sizes of int and uint, double, cbool... soon I think we'll be...

It is possible, though, since flake8 can give access to the raw source, and then can use Python's parser module to get a parse tree and work on that.

Thanks for the careful and clear analysis! Yes, definitely it seems we don't currently generate valid stubs for nested classes.