True, but I think what I don't like is that this PEP essentially creates an entire new way of "type definitions" that is separate from the type definitions we a

True, but I think what I don't like is that this PEP essentially creates an entire new way of "type definitions" that is separate from the type definitions we already have. I get the rationale for "anonymous strict" return types, but then I think a better way would be to think up some way to accomplish that for dataclasses.