DEPR: Disallow dtype inference when setting Index into DataFrame #56102
Conversation
|
Thanks @phofl |
| # TODO: Remove kludge in sanitize_array for string mode when enforcing | ||
| # this deprecation | ||
| warnings.warn( | ||
| "Setting an Index with object dtype into a DataFrame will no longer " |
There was a problem hiding this comment.
is "will no longer" clear enough that this refers to a future version/deprecation?
There was a problem hiding this comment.
good point, I'll put up a PR to clarify
| return sanitize_array(value, self.index, copy=True, allow_2d=True), None | ||
| arr = sanitize_array(value, self.index, copy=True, allow_2d=True) | ||
| if ( | ||
| isinstance(value, Index) |
There was a problem hiding this comment.
might be more performant to do the Index check before the sanitize_array?
There was a problem hiding this comment.
Why? We have to sanitise anyway?
There was a problem hiding this comment.
if you know you have an Index (and no dtype), sanitize_array boils down to:
if len(value) != len(index): raise ...
return value._values
so a lot of sanitize_array may be made unnecessary in this case
There was a problem hiding this comment.
That's what I want in the future, but we go through maybe_infer_to_datetimelike at the moment which changes the dtype, so we can't shortcut this before we enforce the deprecation
doc/source/whatsnew/vX.X.X.rstfile if fixing a bug or adding a new feature.cc @jbrockmendel I think we chatted about this. Series keeps the dtype, so this is mostly for consistency and to make our lives a little easier