Comparison for dataclasses doesn't work correctly

[barisione]

Checks

• I added a descriptive title to this issue
• I have searched (google, github) for similar issues and couldn't find anything
• I have read and followed the docs and still think this is a bug

Bug

Output of python -c "import pydantic.utils; print(pydantic.utils.version_info())":

             pydantic version: 1.7.2
            pydantic compiled: True
                 install path: /Users/bari/.pyenv/versions/3.6.10/envs/undo-core-venv3.6/lib/python3.6/site-packages/pydantic
               python version: 3.6.10 (default, May 19 2020, 15:12:14)  [GCC 4.2.1 Compatible Apple LLVM 11.0.0 (clang-1100.0.33.8)]
                     platform: Darwin-20.3.0-x86_64-i386-64bit
     optional deps. installed: ['typing-extensions']

I also tried with 1.7.3 with Python 3.8 and the bug still happens.

After reading https://pydantic-docs.helpmanual.io/usage/dataclasses/ I understand why dataclasses in a model get replaced with a different type which behaves (almost) the same as the original dataclass.
Unfortunately comparison between instances of the original dataclass (defined with order=True) and the derived pydantic one doesn't work.

from dataclasses import dataclass

import pydantic


@dataclass(order=True)
class DC:
    num: int = 42


class Model(pydantic.BaseModel):
    dc: DC


real_dc = DC()
model = Model(dc=real_dc)

print(f"Real dataclass: {real_dc}; model's dataclass: {model.dc}")

# This works as expected.
assert real_dc <= real_dc
assert model.dc <= model.dc

try:
    # This raises a TypeError.
    assert real_dc <= model.dc
except TypeError as exc:
    print(f"Comparison failed: {exc}")

The output is:

Real dataclass: DC(num=42); model's dataclass: DC(num=42)
Comparison failed: '<=' not supported between instances of 'DC' and 'DC'

I thought this would help:

class Model(pydantic.BaseModel):
    dc: DC

    class Config:
        arbitrary_types_allowed = True

But the result is the same. I guess that arbitrary_types_allowed only applies to types not handled by pydantic at all, is that correct?

[PrettyWood]

Hi @barisione
It's the same problem as #2162

[barisione]

Sorry, I searched existing issues but did not find that one. I guess I can close this then!

[PrettyWood]

Sorry I didn't mean it's the exact same issue but it's clearly related.
Now builtin dataclasses are converted into pydantic dataclasses when used in a model.
This is great and all but it breaks the __eq__, __gt__ and stuff that expect the exact same class.
I think it's worth keeping this issue open to think of all cases and not only __eq__

[PrettyWood]

Hi @barisione
FYI I started to work again on the whole stdlib dataclass to validated dataclass #2557. Feedback more than welcome! 🙏