DEPR: Deprecate ravel #56053
DEPR: Deprecate ravel #56053
Conversation
pandas/core/series.py
Outdated
| array([1, 2, 3]) | ||
| """ | ||
| warnings.warn( | ||
| "Series.ravel is deprecated. Use numpy.ravel directly.", |
There was a problem hiding this comment.
Or maybe we should also point to to_numpy() in case people use this to essentially get the numpy series (the "ravel" aspect never does something, since Series is already 1D)
There was a problem hiding this comment.
or just "dont do it bc it isnt necessary"? (also should we deprecate Index.ravel?)
There was a problem hiding this comment.
or just "dont do it bc it isnt necessary"?
That's not necessarily correct. If you rely on the current return value, i.e. if your code requires that the object is a 1D numpy array, and you currently use ravel for that, you do need to replace ravel with a conversion to numpy.
(and that's actually how we were implicitly using it internally)
pandas/core/series.py
Outdated
| warnings.warn( | ||
| "Series.ravel is deprecated. Use numpy.ravel directly.", | ||
| FutureWarning, | ||
| stacklevel=find_stack_level(), |
There was a problem hiding this comment.
| stacklevel=find_stack_level(), | |
| stacklevel=2, |
For a simple case like this where the warning is directly raised in the top-level method, I prefer setting it explicitly, which avoids potential hickups with the function (I think we had perf issues in certain cases, and this also avoids getting a wrong stacklevel in case we forget a use case of ravel in our own code base)
pandas/core/reshape/pivot.py
Outdated
| @@ -309,15 +309,17 @@ def _add_margins( | |||
|
|
|||
| row_names = result.index.names | |||
| # check the result column and leave floats | |||
|
|
|||
| def _cast_types(x, dtype): | |||
| return maybe_downcast_to_dtype(x._values, dtype) | |||
There was a problem hiding this comment.
Should we do that inside maybe_downcast_to_dtype instead? That function is also used in some other places (I don't know if it is guaranteed to be called with a non-Series in those other places, but our test coverage might also be lacking on that aspect)
pandas/core/series.py
Outdated
| array([1, 2, 3]) | ||
| """ | ||
| warnings.warn( | ||
| "Series.ravel is deprecated. The underlying array is already 1D, so " | ||
| "ravel is not necessary.", |
There was a problem hiding this comment.
| "ravel is not necessary.", | |
| "ravel is not necessary. Use `to_numpy()` for conversion to a numpy array instead.", |
| @@ -846,6 +846,8 @@ def ravel(self, order: str = "C") -> ArrayLike: | |||
| """ | |||
| Return the flattened underlying data as an ndarray or ExtensionArray. | |||
|
|
|||
| .. deprecated:: 2.2.0 | |||
There was a problem hiding this comment.
Can you add the content of the warning message here as well?
|
Thanks! |
doc/source/whatsnew/vX.X.X.rstfile if fixing a bug or adding a new feature.