wip: depr partial row-wise multi-column-wise enlargement

doc/source/whatsnew/v2.2.0.rst

@@ -657,6 +657,24 @@ Set the following option to opt into the future behavior:
 
     In [9]: pd.set_option("future.no_silent_downcasting", True)
 
+Deprecated list-like-column-wise enlargement with partial row indexer
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Enlarging a dataframe with a partial row indexer and a list-like column indexer is deprecated.
+
+Instead of (suppose ``'A'`` and ``'B'`` are not already columns in ``df``)
+
+.. code-block:: python
+
+    df.loc[idx, ['A', 'B']] = (val_1, val_2)
+
+please do
+
+.. code-block:: python
+
+    df.loc[idx, 'A'] = val_1
+    df.loc[idx, 'B'] = val_2
+
 Other Deprecations
 ^^^^^^^^^^^^^^^^^^
 - Changed :meth:`Timedelta.resolution_string` to return ``h``, ``min``, ``s``, ``ms``, ``us``, and ``ns`` instead of ``H``, ``T``, ``S``, ``L``, ``U``, and ``N``, for compatibility with respective deprecations in frequency aliases (:issue:`52536`)

pandas/core/indexing.py

@@ -877,7 +877,20 @@ def _ensure_listlike_indexer(self, key, axis=None, value=None) -> None:
                 )
                 self.obj._mgr = new_mgr
                 return
-
+            if len(diff):
+                warnings.warn(
+                    (
+                        "list-like-column-wise enlargement with partial-row-wise "
+                        "indexer is deprecated.\n"
+                        "Instead of\n"
+                        "    df.loc[idx, ['A', 'B']] = (val_1, val_2)\n"
+                        "please do\n"
+                        "    df.loc[idx, 'A'] = val_1\n"
+                        "    df.loc[idx, 'B'] = val_2\n"
+                    ),
+                    FutureWarning,
+                    stacklevel=find_stack_level(),
+                )
             self.obj._mgr = self.obj._mgr.reindex_axis(keys, axis=0, only_slice=True)
 
     @final

pandas/tests/indexing/multiindex/test_loc.py

@@ -377,7 +377,15 @@ def test_multiindex_setitem_columns_enlarging(self, indexer, exp_value):
         # GH#39147
         mi = MultiIndex.from_tuples([(1, 2), (3, 4)])
         df = DataFrame([[1, 2], [3, 4]], index=mi, columns=["a", "b"])
-        df.loc[indexer, ["c", "d"]] = 1.0
+        warn = FutureWarning if indexer != slice(None) else None
+        with tm.assert_produces_warning(
+            warn,
+            match=(
+                "list-like-column-wise enlargement with partial-row-wise indexer "
+                "is deprecated"
+            ),
+        ):
+            df.loc[indexer, ["c", "d"]] = 1.0
         expected = DataFrame(
             [[1, 2, 1.0, 1.0], [3, 4, exp_value, exp_value]],
             index=mi,

pandas/tests/indexing/test_loc.py

@@ -940,8 +940,12 @@ def test_loc_setitem_with_scalar_index(self, indexer, value):
     def test_loc_setitem_missing_columns(self, index, box, expected):
         # GH 29334
         df = DataFrame([[1, 2], [3, 4], [5, 6]], columns=["A", "B"])
-
-        df.loc[index] = box
+        warn = FutureWarning if index[0] != slice(None, None, None) else None
+        with tm.assert_produces_warning(
+            warn,
+            match=("list-like-column-wise enlargement with partial-row-wise indexer"),
+        ):
+            df.loc[index] = box
         tm.assert_frame_equal(df, expected)
 
     def test_loc_coercion(self):
@@ -3327,7 +3331,14 @@ def test_loc_setitem_dict_timedelta_multiple_set(self):
     def test_loc_set_multiple_items_in_multiple_new_columns(self):
         # GH 25594
         df = DataFrame(index=[1, 2], columns=["a"])
-        df.loc[1, ["b", "c"]] = [6, 7]
+        with tm.assert_produces_warning(
+            FutureWarning,
+            match=(
+                "list-like-column-wise enlargement with partial-row-wise indexer "
+                "is deprecated"
+            ),
+        ):
+            df.loc[1, ["b", "c"]] = [6, 7]
 
         expected = DataFrame(
             {