Release v1.3.1

[austind]

No description provided.

[gyandeeps-hh]

adding this actually breaks the use case i resolved in the previous PR. do we need just these 2 lines?

[austind]

Hmm, mypy and pylance didn't complain, let me look into it. Thanks for letting me know!

[gyandeeps-hh]

thanks.... i use mypy and it never complained in my project also so i was little surprised with airflow issues. I do use the latest mypy.

[austind]

vscode is showing the right return type, what are you seeing?

[gyandeeps-hh]

Have the method take a param in like this def get_example(x: int) -> Payload | None:

[austind]

Still works for me, what do you see?

[gyandeeps-hh]

Here is the complete code and you will see how lower example is getting confused. Very tricky lol

from uuid import UUID
from pydantic import BaseModel
import httpx
from retryhttp._retry import retry


class Payload(BaseModel):
    uuid: UUID
    id: int
    user_id: int
    user_uuid: UUID


class Defaults(BaseModel):
    timeout: int = 3
    backoff: bool = True


defaults1 = {
    "timeout": 5,
    "backoff": True,
}

defaults2 = Defaults().model_dump()


@retry(**defaults1)
def get_example1(x: int) -> Payload:
    response = httpx.get("https://example.com/")
    response.raise_for_status()

    return Payload(**response.json())


x = get_example1(1).user_id


@retry(**defaults2)
def get_example2(x: int) -> Payload:
    response = httpx.get("https://example.com/")
    response.raise_for_status()

    return Payload(**response.json())


x = get_example2(1).user_id

[gyandeeps-hh]

one of the solution is to change the defaults2 init to defaults2: dict = Defaults().model_dump()

[austind]

The testing I did suggests the issue is model_dump()'s return type. this is an unusual case (why are you using a pydantic model for kwargs?) and I'm not sure there's anything wrong with the decorator's type signatures. You're welcome to experiment--let me know if you find something that works! Just be sure it works with mypy too :)

[gyandeeps-hh]

Honestly, not a big deal as its a minor nuance. Maybe future versions of pylance can improve this. i think we are good for now. 😄 thanks for your support.