Backport PR #47327 on branch 1.4.x (REGR: Fix fillna making a copy when dict was given as fill value and inplace is set) #47448
Conversation
…t was given as fill value and inplace is set
|
@phofl XPASS on |
|
This is interesting, do you know if this was an intended change? Otherwise we should open an issue about it |
not sure. As this is 1.4.x, I'm not sure array manager changes are a concern. @jbrockmendel ? |
|
Sorry that wasn't clear. I meant: If this is passing on 1.4.x and failing on main, this looks like a regression on main compared to the 1.4 release |
|
yeah. can open an issue for main. for this branch, i'm sure we just skip any failing array manager tests to get ci green. |
|
if there is a problem and a fix on main, we could then maybe consider backporting the fix for 1.4.4 |
| @@ -146,3 +148,14 @@ def test_update_with_different_dtype(self): | |||
|
|
|||
| expected = DataFrame({"a": [1, 3], "b": [np.nan, 2], "c": ["foo", np.nan]}) | |||
| tm.assert_frame_equal(df, expected) | |||
|
|
|||
| @td.skip_array_manager_invalid_test | |||
There was a problem hiding this comment.
should this be skip_array_manager_not_implemented? same above?
There was a problem hiding this comment.
maybe for the two added tests. but this is the backport. so should change on main for the change to be relevant.
for test_fillna_dictlike_value_duplicate_colnames it is XPASSing so we need to skip as skip_array_manager_not_implemented xfails.
i wont lose any sleep over array manager tests failing |
Backport PR #47327