David Peter

I've seen multiple real world examples where missing support for loop control flow leads to an incorrect reachability analysis for large blocks of code within loops, because we observe a...

> And because we don't respect `first_part: bool` there's no (non-tedious) way to tell ty to chill. Obviously not ideal, but a `cast` always helps as a workaround in these...

Thank you for your contribution. I'm planning to review this soon.

Thank you for your contribution. I did not have time to review this in detail yet. One question upfront: would it be reasonable to use the existing `invalid-override` diagnostic here...

Thank you for looking into this! > but this may be resolved by type narrowing of the instance attributes, which I'm currently working on as a separate issue I'm assuming...

This is interesting. Inlining is one way to solve this problem, but it's probably limited? Both mypy and pyright see the problem in ```py foo(lambda x: x) ``` but they...

On https://github.com/astral-sh/ruff/pull/21363, we see 2,160 new diagnostics because we now start to understand some large unions in the pydantic code base, such as [this one](https://github.com/pydantic/pydantic/blob/27c95fd56e217f6cfc6ee96e07e1fb0646ef0d49/pydantic-core/python/pydantic_core/core_schema.py#L4106-L4159) (also see the "fun" comment...

> if we synthesize a protocol type to intersect with Is it possible to formulate this as a (normal) protocol? Both freqtrade and pydantic narrow based on a key, not...

Moving this ticket to GA, but #1387 into the Beta

Thank you for your feature request! The way we intend to solve this is through uv. In fact, there's an open feature request for a [`--with-requirements script.py`](https://github.com/astral-sh/uv/issues/6542) option (and an...