What from __future__ import annotations does

Question

Accepted Answer

Without this import, Python evaluates annotations eagerly when a function or class is defined. That means any name used in an annotation must already exist at that point - forward references to classes defined later in the same file cause NameError. With 'from __future__ import annotations', every annotation becomes a lazy string. Python stores the text 'MyClass' instead of evaluating it. This solves forward references and also lets you use newer syntax like 'int | str' on Python 3.8 and 3.9 inside annotation positions, because the string is never evaluated by the interpreter. The catch: if you call get_type_hints() later, it evaluates those strings in the module's namespace. And some libraries (Pydantic v1, dataclasses with certain patterns) read __annotations__ directly and may break because they get strings instead of actual types. This behavior was originally planned to become the default in Python 3.12, but that was postponed. Check your framework's docs before enabling it blindly.