Sebastian Rittau

Results 518 comments of Sebastian Rittau

This works when testing locally, but for unknown reasons fails in CI. #11575 doesn't help either.

I'm generally not a fan of diverging from the implementation when it can be avoided, but I see some merit in this case. An alternative would be to define a...

Dependencies are updated automatically by renovate each UTC night.

Looking at the primer output: pydantic and streamlit are true positives. They use `is_dataclass()` and immediately call `asdict()` if true. This can lead to errors if `is_dataclass()` is a dataclass...

I'm fine with merging this, despite the false positive, but I'll leave it open for a day or two if someone has an idea how to fix/work around it.

This is a trial to judge the impact of such a change that could prevent runtime errors due to mixing timezone-aware and unaware datetimes.

While I think there's some promise in this approach, I struggle with `Any` annotated datetimes. The `__sub__` overloads (for example), derive a return type of `Any`, probably because mypy is...

@bluetech > * Why default TzInfoT to `Any` instead of `tzinfo | None`? It tried both, but both didn't work. `tzinfo | None` is also my preferred default, but I...

It sounds as if we should use a `TypeVar` here, but the existing overload makes me a bit anxious. Any fix should include tests to check a few use cases....

One of the ideas behind #5952 was to actually test third-party stubs using the vendored stdlib of the respective type checkers. Maybe we should resurrect that idea?