Start a list of modules which require typing #8198
Conversation
Notes inline. Just one module so far!
xarray/core/rolling_exp.py
Outdated
| # pandas defines in terms of com (converting to alpha in the algo) | ||
| # so use its function to get a com and then convert to alpha | ||
|
|
||
| com = _get_center_of_mass(com, span, halflife, alpha) | ||
| return 1 / (1 + com) | ||
|
|
||
|
|
||
| def move_exp_nanmean(array, *, axis, alpha): | ||
| def move_exp_nanmean(array: np.ndarray, *, axis: int, alpha: float) -> np.ndarray: |
There was a problem hiding this comment.
Is it really true that array can only be np.ndarray? cupy arrays are not allowed?
There was a problem hiding this comment.
I hadn't thought of that — I think it is possible: https://docs.cupy.dev/en/stable/user_guide/interoperability.html#numba
Do you know what the correct item to use here is? ArrayLike?
There was a problem hiding this comment.
In #8075 we are testing T_DuckArray = typing.TypeVar("T_DuckArray", bound=typing.Any) for self.data, which should imply "an array that follows the array api".
Nothing is in main at the moment though, probably because it would open so many cans of worms. :D
I think the important thing here though is that we just have to put a T_DuckArray in types.py and start to use it anywhere Variable/DataArray.data is used.
There was a problem hiding this comment.
OK — what should do we do for this PR? This is just trying to start a pattern of allowing some strict modes — I hadn't really intended it to be breaking new ground on array types... :)
I could replace with Any if we would rather have something vague that slightly wrong...
There was a problem hiding this comment.
In my opinion, there's no point in strict mode if we're just going to use Any.
Add T_DuckArray in types.py and use it here, hopefully mypy won't complain too much.
There was a problem hiding this comment.
If #8075 merges soon, I'm happy to use that — otherwise I really did just open this to start an template for expanding strict mode, I would strongly prefer to keep this PR focused & generally allow PRs to be small marginal improvements rather than midas-touching lots of code.
(and while we should indeed attempt to not use Any, it's still better than nothing, because a function with some annotations still works to check types internally)
There was a problem hiding this comment.
Added T_DuckArray!
max-sixty
force-pushed
the
mypy-strict
branch
from
f91b8ef
to
184aeee
Compare
Notes inline. Just one module so far!