Annotations in typing.Annotated are not checked

[NiklasRosenstein]

Bug Report

To Reproduce

import typing_extensions as te
x: te.Annotated[int, Foobar] = 42

Expected Behavior

Mypy complains about Foobar not being defined. In other cases, e.g. when a function is called in an Annotated argument, it checks if the arguments passed to the function are valid (or number of arguments match, etc. etc.).

Actual Behavior

Mypy does not complain about Foobar not being defined.

Your Environment

• Mypy version used: 0.910
• Mypy command-line flags: n/a
• Mypy configuration options from mypy.ini (and other config files): n/a
• Python version used: 3.8.2
• Operating system and version: OSX

[hauntsaninja]

I think this is a result of #9625
People even want this non-checking to be extended to all usages of Annotated, e.g., see #10777

Fwiw, my reading of PEP 593 is that mypy should complain about your example, but not if from __future__ import annotations, since then it won't fail at runtime and falls more under the category of "Unknown annotations" which PEP 593 says type checkers should ignore.

[Dreamsorcerer]

This will become particularly useful with Pydantic's pipeline API:

v = validate_as(int).transform(str.lower)  # Type error here

class MyDict(TypedDict):
    a: Annotated[str, validate_as(int).transform(str.lower)]  # No type error here