Mypy regression - Missing named argument

[pykler]

Initial Checks

• I have searched GitHub for a duplicate issue and I'm sure this is something new
• I have searched Google & StackOverflow for a solution and couldn't find anything
• I have read and followed the docs and still think this is a bug
• I am confident that the issue is with pydantic (not my code, or another library in the ecosystem like FastAPI or mypy)

Description

pydantic version 1.10.6 + mypy 1.1.1 + mypy plugin causes error with previously ok code.
Note: mypy plugin is irrelevant, and does not affect the output in this usecase

Using Field with a description on an optional Field, without an explicit default=None, used to work

$ mypy sample.py 
sample.py:10: error: Missing named argument "bar" for "Sample"  [call-arg]
Found 1 error in 1 file (checked 1 source file)

Error started happening with pydantic 1.10.4, version 1.10.2 does not have this issue
versions >1.10.2 now require the explicit default=None on Optional fields.

This seems like a bug, if not please advise.

Example Code

from typing import Optional
from pydantic import BaseModel, Field


class Sample(BaseModel):
    foo: str
    bar: Optional[str] = Field(description="optional")


Sample(foo="hello world")

Python, Pydantic & OS Version

pydantic version: 1.10.6
            pydantic compiled: True
                 install path: /home/hatem/projects/inventory_service/venv/lib/python3.8/site-packages/pydantic
               python version: 3.8.10 (default, Sep 28 2021, 16:10:42)  [GCC 9.3.0]
                     platform: Linux-5.11.0-37-generic-x86_64-with-glibc2.29
     optional deps. installed: ['dotenv', 'typing-extensions']

Affected Components

• Compatibility between releases
• Data validation/parsing
• Data serialization - .dict() and .json()
• JSON Schema
• Dataclasses
• Model Config
• Field Types - adding or changing a particular data type
• Function validation decorator
• Generic Models
• Other Model behaviour - construct(), pickling, private attributes, ORM mode
• Plugins and integration with other tools - mypy, FastAPI, python-devtools, Hypothesis, VS Code, PyCharm, etc.

[samuelcolvin]

see #5192

[dmontagu]

This appears to have been introduced by using @dataclass_transform on the metaclass, and the reason that it works in 1.10.2 is because that version used the keyword argument field_descriptors instead of the correct field_specifiers.

I'm not sure how involved it will be to fix this properly while retaining the @dataclass_transform behavior; the good news is that in v2 this shouldn't be an issue because we no longer set an implicit default for Optional/Any fields.

Does anyone know why the keyword argument was field_descriptors before? Was that just a typo?

[dmontagu]

Seems there was some logic issues in the plugin that meant it wouldn't work correctly if you set a Field as the default but didn't specify default or default_factory. The dataclass_transform thing was just hiding the issue. I've got a fix coming shortly.