TYP: simple return types #54786
TYP: simple return types #54786
Conversation
| @@ -110,7 +110,7 @@ | |||
| ) | |||
|
|
|||
|
|
|||
| IntervalSideT = Union[TimeArrayLike, np.ndarray] | |||
| IntervalSide = Union[TimeArrayLike, np.ndarray] | |||
There was a problem hiding this comment.
...T is often used for TypeVars
pandas/core/computation/expr.py
Outdated
| @@ -349,8 +350,8 @@ class BaseExprVisitor(ast.NodeVisitor): | |||
| preparser : callable | |||
| """ | |||
|
|
|||
| const_type: type[Term] = Constant | |||
| term_type = Term | |||
| const_type: ClassVar[type[Constant]] = Constant | |||
There was a problem hiding this comment.
Should probably have one PR just focused on making ClassVars more explicit (type checkers have difficulties determining whether they are class or instance variables).
| @@ -770,6 +772,7 @@ def value_counts( | |||
| mask = ids != -1 | |||
| ids, val = ids[mask], val[mask] | |||
|
|
|||
| lab: Index | np.ndarray | |||
There was a problem hiding this comment.
This pattern happens quite often because of mypy's early binding assumption (first declaration, even if only in one branch, determines the type). I prefer pyright's behavior:
if foo():
a = 1 # mypy: a is of type int (in this branch, in all other branches, and after the if/else)
# pyright: a is of type int in this branch
else:
a = "1" # mypy error!
# pyright: a is of type str in this branch
# pyright: a is of type int | str
There was a problem hiding this comment.
No real opinion here - mypy forces us to be more explicit for readers prior to branching which I think is a plus.
| @@ -770,6 +772,7 @@ def value_counts( | |||
| mask = ids != -1 | |||
| ids, val = ids[mask], val[mask] | |||
|
|
|||
| lab: Index | np.ndarray | |||
There was a problem hiding this comment.
No real opinion here - mypy forces us to be more explicit for readers prior to branching which I think is a plus.
|
Thank you @rhshadrach for your thorough review! I think I addressed your comments as best as I can. |
|
Thanks @twoertwein |
|
hm, the docs timed out on main as well - did the annotation cause the docs to fail? |
|
I think that's unrelated (my running theory is that something processing the whatsnew files is not parallel safe and hanging) |
* Return None * Return simple types * ruff false positive * isort+mypy * typo, use " for cast * SingleArrayManager.dtype can also be a numpy dtype * comments + test assert on CI * wider return types at the cost of one fewer mypy ignore * DatetimeArray reaches IntervalArray._combined * avoid some ignores * remove assert False
"simple" meaning, non-unions and non-overload return types (at least in most cases).
xref #47521
edit: sorry for changes in many files - I grep'ed for missing return types inferred by pyright and then fixed the resulting type errors.