Different behaviour of Config for pydantic model when Config is inherited and when Config is passed through dataclass decorator

[carmenere]

• OS: Ubuntu 19.04
• Python version import sys; print(sys.version): 3.7.3 (default, Oct 7 2019, 12:56:13) [GCC 8.3.0]
• Pydantic version import pydantic; print(pydantic.VERSION): 1.2

1. When Config is passed through dataclass decorator

from typing import (
    Optional,
    List,
    Dict,
    Tuple,
    Union,
    # Final
)

from pydantic import (
    BaseModel,
    Field,
    ValidationError,
    validator,
    root_validator,
    Extra
)

from pydantic.dataclasses import dataclass

class ModelConfig:
    extra = Extra.forbid


@dataclass(config=ModelConfig)
class ABC(BaseModel):
    a: Optional[int]
    b: Optional[int]
    c: Optional[int]

Call ABC(**{"a": 0}) returns TypeError: __init__() missing 2 required positional arguments: 'b' and 'c'
Call ABC(**{"a": 0, "b": 0, "c": 0}) returns AttributeError: __fields_set__
Call ABC(**{"a": 0, "b": 0, "c": 0, "garbage": 0}) returns TypeError: __init__() got an unexpected keyword argument 'garbage'

2. When Config is inherited

class BaseModelConfig(BaseModel):
    class Config:
        extra = Extra.forbid

class XYZ(BaseModelConfig):
    x: Optional[int]
    y: Optional[int]
    z: Optional[int]

Call XYZ(**{"x": 0}) returns right object: XYZ(x=0, y=None, z=None)
Call XYZ(**{"x": 0, "y": 0, "z": 0}) returns right object: XYZ(x=0, y=0, z=0)
Call XYZ(**{"x": 0, "y": 0, "z": 0, "garbage": 0}) returns

ValidationError: 1 validation error for XYZ 
garbage 
  extra fields not permitted (type=value_error.extra)

[samuelcolvin]

This is standard behaviour of dataclasses, not something pydantic can change, see #986 for details.