REF/API: DataFrame.setitem never operate in-place #39510

jbrockmendel · 2021-01-31T20:29:53Z

closes #xxxx
tests added / passed
Ensure all linting tests pass, see here for how to run them
whatsnew entry

…titem-frame-no-defer

jreback · 2021-02-03T13:49:14Z

pandas/core/frame.py

        else:
            if isinstance(value, DataFrame):
                check_key_length(self.columns, key, value)
                for k1, k2 in zip(key, value.columns):
                    self[k1] = value[k2]
+


really like to move this stuff out of here into another module, but i guess that's for another day

jreback · 2021-02-03T13:50:40Z

pandas/core/frame.py

+                raise NotImplementedError
+
+            if np.shape(value)[-1] != len(ilocs):
+                raise ValueError("Columns must be same length as key")


are all of the branches covered? this is a lot of new code

not the NotImplementedError on L3276; not sure what to do there

but the rest is covered? what about the previous paths this was taking, can they be removed now? (e.g. they are uncovered)

updated with a test that hits the two raise ValueError("Columns must be same length as key") lines. everything other than the NotImplementedError is reached

jbrockmendel · 2021-02-12T02:38:32Z

getting this and #39163 in will make it easier to keep ArrayManager behavior in sync with non-ArrayManager

jbrockmendel · 2021-02-18T05:20:22Z

@jreback gentle ping; some simplification in iloc.__setitem__ becomes feasible after this, some of which I think may make #39578 easier

jorisvandenbossche

Can you give some explanation what this PR is exactly changing? (it's hard to see that from the code changed in frame.py)

jorisvandenbossche · 2021-02-18T15:15:31Z

pandas/tests/frame/indexing/test_setitem.py

+        expected["c"] = expected["c"].astype(arr.dtype)
+        expected["d"] = expected["d"].astype(arr.dtype)
+        assert expected["c"].dtype == arr.dtype
+        assert expected["d"].dtype == arr.dtype


What changed here? (I wouldn't expect the above test to change behaviour related to this PR, as it's adding new columns)

IIRC this was needed for the 32bit builds

jbrockmendel · 2021-02-18T15:44:52Z

Can you give some explanation what this PR is exactly changing? (it's hard to see that from the code changed in frame.py)

There are cases where DataFrame.__setitem__ currently ends up dispatching to iloc.__setitem__ (which in turn ends up iterating over columns). This does the iterating-over-columns directly in the DataFrame method, avoiding the iloc dispatch. This resolves the trouble in iloc[:, foo] of having to tell whether to write into existing data or insert a new array.

phofl · 2021-02-20T20:12:05Z

I think we talked about this a few weeks back. I prepared a few tests back then but did not get around to push this forward.

The tests are checking, if we are really not operating inplace. Do you think it is worth adding them?

class TestDataFrameSetItemNotInplace:
    def test_setitem_duplicate_columns(self):
        cols = ["A", "B"] * 2
        df = DataFrame(0.0, index=[0], columns=cols)
        df_copy = df.copy()
        df_view = df[:]
        df["B"] = (2, 5)

        expected = DataFrame([[0.0, 2, 0.0, 5]], columns=cols)
        tm.assert_frame_equal(df_view, df_copy)
        tm.assert_frame_equal(df, expected)

    @pytest.mark.xfail(reason="Setitem with same dtype still changing inplace")
    @pytest.mark.parametrize("value", [1, np.array([[1], [1]]), [[1], [1]]])
    def test_setitem_same_dtype(self, value):
        cols = ["A", "B"]
        df = DataFrame(0, index=[0, 1], columns=cols)
        df_copy = df.copy()
        df_view = df[:]
        df[["B"]] = value

        expected = DataFrame([[0, 1], [0, 1]], columns=cols)
        tm.assert_frame_equal(df, expected)
        tm.assert_frame_equal(df_view, df_copy)

    @pytest.mark.parametrize("value", [1.0, np.array([[1.0], [1.0]]), [[1.0], [1.0]]])
    def test_setitem_listlike_key_scalar_value(self, value):
        cols = ["A", "B"]
        df = DataFrame(0, index=[0, 1], columns=cols)
        df_copy = df.copy()
        df_view = df[:]
        df[["B"]] = value

        expected = DataFrame([[0, 1.0], [0, 1.0]], columns=cols)
        tm.assert_frame_equal(df_view, df_copy)
        tm.assert_frame_equal(df, expected)

jbrockmendel · 2021-02-23T21:37:13Z

Do you think it is worth adding them?

absolutely. will check that these pass on the branch and then update. thanks

jbrockmendel · 2021-03-04T17:00:18Z

ping

jorisvandenbossche

Can you clarify the whatsnew a bit more about what actually changed / what the impact is for the user? (this is not really a bug fix ..)

jbrockmendel · 2021-03-04T17:56:14Z

Can you clarify the whatsnew a bit more about what actually changed / what the impact is for the user? (this is not really a bug fix ..)

The only part to call a bugfix is the int32 changes (see affected tests). The changed exception messages hardly seem worth noting

jorisvandenbossche · 2021-03-04T19:11:36Z

The whatsnew note now says "some cases". Eg could make that more specific?

jorisvandenbossche · 2021-03-04T19:32:29Z

To try to clarify what I mean. I didn't have time to review the code of this PR in enough detail to know what it changes, but I would still expect to get an idea about that based on the (whatsnew) description. But now neither the PR description nor the whatsnew gives me a clear idea of the (potential) changes.

jbrockmendel · 2021-03-04T23:25:50Z

To try to clarify what I mean. I didn't have time to review the code of this PR in enough detail to know what it changes, but I would still expect to get an idea about that based on the (whatsnew) description. But now neither the PR description nor the whatsnew gives me a clear idea of the (potential) changes.

ATM DataFrame.__setitem__ calls self.iloc[:, indexer] = value, which has different copy/view semantics than we want for __setitem__ (see #38896 (comment), also linked in the OP). Big picture: this PR is avoiding that self.iloc dispatch.

Luckily some of the tests suggested by @phofl fails in master:

# test_setitem_listlike_key_scalar_value_not_inplace with value = 1.0
cols = ["A", "B"]
df = DataFrame(0, index=[0, 1], columns=cols)
df_copy = df.copy()
df_view = df[:]
df[["B"]] = 1.0

expected = DataFrame([[0, 1.0], [0, 1.0]], columns=cols)
tm.assert_frame_equal(df_view, df_copy)  # <-- raises in master, b/c df["B"]._values is written _into_
tm.assert_frame_equal(df, expected)

# test_setitem_duplicate_columns_not_inplace
cols = ["A", "B"] * 2
df = DataFrame(0.0, index=[0], columns=cols)
df_copy = df.copy()
df_view = df[:]
df["B"] = (2, 5)

expected = DataFrame([[0.0, 2, 0.0, 5]], columns=cols)
tm.assert_frame_equal(df_view, df_copy)  # <-- raises in master, b/c df["B"]._values is written _into_
tm.assert_frame_equal(df, expected)

jorisvandenbossche · 2021-03-05T16:49:56Z

OK, so a consequence is that the resulting dtype can change compared to master when you are setting multiple columns. For example (using this branch):

In [4]: df = pd.DataFrame(np.random.randn(3, 5), columns=['a', 'b', 'c', 'd', 'e'])

In [5]: df[["b", "c"]] = 1

In [6]: df
Out[6]: 
          a  b  c         d         e
0  0.576266  1  1  0.115430  0.670362
1 -0.409783  1  1  1.824356  1.702370
2  1.168233  1  1 -0.708717  1.555301

In [7]: df.dtypes
Out[7]: 
a    float64
b      int64
c      int64
d    float64
e    float64
dtype: object

On master, the above preserves the all-float dtypes. Do you know if that is specific to that case? (multiple columns, and setting with a scalar) From quick testing, it seems that other cases (like setting multiple columns with a dataframe, or array, .. ) didn't try to preserve the dtype if possible.
But where there other cases also going through the self.iloc[:, indexer] = value path?

jbrockmendel · 2021-03-06T00:21:22Z

But where there other cases also going through the self.iloc[:, indexer] = value path?

The only place where we go through self.iloc[:, indexer] = value is in _setitem_array, which is reached in two ways (note: a bunch of other cases are caught before going through _setitem_array, so these conditions are not sufficient to ensure we go through iloc)

if isinstance(key, (Series, np.ndarray, list, Index)):
elif is_list_like(value) and 1 < len(self.columns.get_indexer_for([key])) == len(value):

which is pretty well summed up by "multiple columns"

jorisvandenbossche · 2021-03-08T19:49:41Z

I would personally then add a small subsection to the "Notable bug fixes" part showing the example of setting multiple columns now having different dtypes as result (there is already a subsection for iloc/loc changes (actually trying inplace instead of replacing), which is quite related (but opposite) change.

jorisvandenbossche · 2021-03-08T19:53:13Z

pandas/core/frame.py

-                self.iloc[:, indexer] = value
+                self._iset_not_inplace(key, value)
+
+    def _iset_not_inplace(self, key, value):


Can you add a comment (or informal docstring) here stating briefly which cases this covers? (eg I think only for listlike key and value?)

good idea, will do

updated, LMK if you find this informative

I think comments are addressed, so this is ready for another look (once the CI runs)

jreback

can you merge master

jreback · 2021-03-16T13:54:22Z

pandas/core/frame.py

+            #  hypothetically would return obj.iloc[:, i]
+            if isinstance(obj, np.ndarray):
+                return obj[..., i]
+            else:


in the future would be really nice to move lots of this elsewhere (e.g. indexing.py or similar)

jbrockmendel · 2021-03-16T16:48:46Z

rebased, green ex two issues on master being addressed elsewhere

jreback · 2021-03-16T16:56:13Z

thanks @jbrockmendel nice!

simonjayhawkins · 2021-06-29T10:38:58Z

There is a perf regression between 1.2.5 and 1.3.x that points to this PR that is not shown on the asv dashboard https://pandas.pydata.org/speed/pandas/#indexing.InsertColumns.time_assign_list_like_with_setitem?python=3.8&Cython=0.29.21&jinja2=

(pandas-dev) simon@T3630:~/pandas/asv_bench (master)$ asv compare v1.2.5 1.3.x |grep time_assign_list_like_with_setitem
x     2.38±0.03ms       27.4±0.4ms    11.48  indexing.InsertColumns.time_assign_list_like_with_setitem

profiling gives

(pandas-dev) simon@T3630:~/pandas/asv_bench (master)$ asv profile indexing.InsertColumns.time_assign_list_like_with_setitem fefd9991658c4717fa2e6b37328be81158be4192|head -n 30
· Discovering benchmarks
· Profile data does not already exist. Running profiler now.
·· Benchmarking conda-py3.8-Cython0.29.21-jinja2-matplotlib-numba-numexpr-numpy-odfpy-openpyxl-pytables-pytest-scipy-sqlalchemy-xlrd-xlsxwriter-xlwt
··· Running (indexing.InsertColumns.time_assign_list_like_with_setitem--).
··· ...rtColumns.time_assign_list_like_with_setitem         24.5±0.2ms

Tue Jun 29 11:28:30 2021    /tmp/tmpa23qf3wg

         71979 function calls (70571 primitive calls) in 0.039 seconds

   Ordered by: cumulative time

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.039    0.039 {built-in method builtins.exec}
        1    0.000    0.000    0.039    0.039 /home/simon/miniconda3/envs/pandas-dev/lib/python3.8/site-packages/asv/benchmark.py:540(method_caller)
        1    0.000    0.000    0.039    0.039 /home/simon/pandas/asv_bench/benchmarks/indexing.py:378(time_assign_list_like_with_setitem)
    101/1    0.000    0.000    0.038    0.038 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3516(__setitem__)
        1    0.000    0.000    0.038    0.038 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3548(_setitem_array)
        1    0.000    0.000    0.038    0.038 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3585(_iset_not_inplace)
      100    0.000    0.000    0.027    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3703(_set_item)
      100    0.000    0.000    0.025    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3678(_set_item_mgr)
      100    0.001    0.000    0.024    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/internals/managers.py:1230(insert)
  401/301    0.003    0.000    0.016    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:334(__new__)
      100    0.001    0.000    0.015    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:5864(insert)
      100    0.000    0.000    0.010    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:5220(get_indexer_for)
  101/100    0.000    0.000    0.009    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:3378(get_indexer)
      101    0.000    0.000    0.005    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:6174(ensure_index)
      101    0.000    0.000    0.005    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/construction.py:455(sanitize_array)
      100    0.000    0.000    0.004    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/internals/managers.py:1994(_fast_count_smallints)
 1200/800    0.001    0.000    0.004    0.000 {built-in method numpy.core._multiarray_umath.implement_array_function}

before

(pandas-dev) simon@T3630:~/pandas/asv_bench (master)$ asv profile indexing.InsertColumns.time_assign_list_like_with_setitem fefd9991658c4717fa2e6b37328be81158be4192~1|head -n 30
· Discovering benchmarks
·· Uninstalling from conda-py3.8-Cython0.29.21-jinja2-matplotlib-numba-numexpr-numpy-odfpy-openpyxl-pytables-pytest-scipy-sqlalchemy-xlrd-xlsxwriter-xlwt
·· Building 203f9014 <v1.3.0rc0~819> for conda-py3.8-Cython0.29.21-jinja2-matplotlib-numba-numexpr-numpy-odfpy-openpyxl-pytables-pytest-scipy-sqlalchemy-xlrd-xlsxwriter-xlwt
·· Installing 203f9014 <v1.3.0rc0~819> into conda-py3.8-Cython0.29.21-jinja2-matplotlib-numba-numexpr-numpy-odfpy-openpyxl-pytables-pytest-scipy-sqlalchemy-xlrd-xlsxwriter-xlwt
· Profile data does not already exist. Running profiler now.
·· Benchmarking conda-py3.8-Cython0.29.21-jinja2-matplotlib-numba-numexpr-numpy-odfpy-openpyxl-pytables-pytest-scipy-sqlalchemy-xlrd-xlsxwriter-xlwt
··· Running (indexing.InsertColumns.time_assign_list_like_with_setitem--).
··· ...rtColumns.time_assign_list_like_with_setitem        2.31±0.01ms

Tue Jun 29 11:33:24 2021    /tmp/tmp7ioufpog

         2162 function calls (2141 primitive calls) in 0.003 seconds

   Ordered by: cumulative time

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.003    0.003 {built-in method builtins.exec}
        1    0.000    0.000    0.003    0.003 /home/simon/miniconda3/envs/pandas-dev/lib/python3.8/site-packages/asv/benchmark.py:540(method_caller)
        1    0.000    0.000    0.003    0.003 /home/simon/pandas/asv_bench/benchmarks/indexing.py:378(time_assign_list_like_with_setitem)
        1    0.002    0.002    0.002    0.002 {method 'randn' of 'numpy.random.mtrand.RandomState' objects}
        1    0.000    0.000    0.001    0.001 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3516(__setitem__)
        1    0.000    0.000    0.001    0.001 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/frame.py:3548(_setitem_array)
        1    0.000    0.000    0.001    0.001 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexing.py:670(_ensure_listlike_indexer)
        1    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexing.py:1257(_get_listlike_indexer)
      6/5    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:334(__new__)
      2/1    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:2784(union)
      3/2    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:3378(get_indexer)
        1    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:5220(get_indexer_for)
        1    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/internals/base.py:62(reindex_axis)
      3/2    0.000    0.000    0.000    0.000 /home/simon/pandas/asv_bench/env/11a1c20ede452de2525075dc4a15eb94/lib/python3.8/site-packages/pandas/core/indexes/base.py:3405(_get_indexer)

so spends some time in _iset_not_inplace

Is this expected?

jbrockmendel · 2021-06-30T14:46:15Z

Is this expected?

With the current implementation, yes. I think a deep(ish) refactor can avoid the perf hit, but it's not a 1-day thing. cc @phofl

BTW is there an option you can pass to the asv profile command like strip_dirs? (ideally just strip everything up through "site-packages/")

jbrockmendel added 7 commits January 26, 2021 20:13

REF: DataFrame._setitem_array dont use iloc.__setitem__

a5c1f5e

REF: DataFrame._setitem_array dont use iloc.__setitem__

15f5265

Merge branch 'master' of https://github.com/pandas-dev/pandas into se…

73302c0

…titem-frame-no-defer

Merge branch 'master' into setitem-frame-no-defer

ceeeb78

Merge branch 'master' into setitem-frame-no-defer

6b304b8

Merge branch 'master' into setitem-frame-no-defer

fe9be7e

mypy fixup

1789fca

jreback added the Indexing Related to indexing on series/frames, not to indexes themselves label Feb 2, 2021

jbrockmendel added 8 commits February 1, 2021 18:48

Merge branch 'master' into setitem-frame-no-defer

6400be3

mypy fixup

a1cdd19

Merge branch 'master' into setitem-frame-no-defer

870ae37

32bit compat

c32a427

Merge branch 'master' into setitem-frame-no-defer

d9d8beb

troubleshoot windows builds

8a283d9

Merge branch 'master' of https://github.com/pandas-dev/pandas into se…

9718f6e

…titem-frame-no-defer

troubleshoot 32bit

9f6d8e7

jreback requested changes Feb 3, 2021

View reviewed changes

jbrockmendel added 3 commits February 3, 2021 07:33

Merge branch 'master' into setitem-frame-no-defer

dd53abc

Merge branch 'master' into setitem-frame-no-defer

f6b09b4

TST: missed raising cases

c107e77

jbrockmendel mentioned this pull request Feb 12, 2021

WIP [ArrayManager] API: setitem to set new columns / loc+iloc to update inplace #39578

Closed

Merge branch 'master' into setitem-frame-no-defer

ae3ae1b

jorisvandenbossche requested changes Feb 18, 2021

View reviewed changes

Merge branch 'master' into setitem-frame-no-defer

c631079

port phofl tests

4cfe863

whatsnew

7ea792f

jorisvandenbossche requested changes Mar 4, 2021

View reviewed changes

clarify whatsnew

20f6a16

Merge branch 'master' into setitem-frame-no-defer

fdbf6f3

jorisvandenbossche reviewed Mar 8, 2021

View reviewed changes

jbrockmendel added 3 commits March 8, 2021 12:36

Merge branch 'master' into setitem-frame-no-defer

2c6da5d

comment

2d0cf9e

whatsnew

e1ed083

jbrockmendel mentioned this pull request Mar 11, 2021

REF: setitem_with_indexer DataFrame always go through split_path #40380

Closed

jbrockmendel added 2 commits March 12, 2021 15:31

Merge branch 'master' into setitem-frame-no-defer

6896e86

xfail only for BM

1af8686

jreback approved these changes Mar 16, 2021

View reviewed changes

Merge branch 'master' into setitem-frame-no-defer

c8b8e23

jreback merged commit fefd999 into pandas-dev:master Mar 16, 2021

jbrockmendel deleted the setitem-frame-no-defer branch March 16, 2021 17:15

JulianWgs pushed a commit to JulianWgs/pandas that referenced this pull request Jul 3, 2021

REF/API: DataFrame.__setitem__ never operate in-place (pandas-dev#39510)

9b2a9ff

simonjayhawkins mentioned this pull request Jul 4, 2021

RLS: 1.3.1 #42343

Closed

jbrockmendel mentioned this pull request Sep 4, 2021

BUG: DataFrame.__setitem__ sometimes operating inplace #43406

Merged

4 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

REF/API: DataFrame.setitem never operate in-place #39510

REF/API: DataFrame.setitem never operate in-place #39510

jbrockmendel commented Jan 31, 2021

jreback Feb 3, 2021

jreback Feb 3, 2021

jbrockmendel Feb 3, 2021

jreback Feb 7, 2021

jbrockmendel Feb 7, 2021

jbrockmendel commented Feb 12, 2021

jbrockmendel commented Feb 18, 2021

jorisvandenbossche left a comment

jorisvandenbossche Feb 18, 2021

jbrockmendel Feb 18, 2021

jbrockmendel commented Feb 18, 2021

phofl commented Feb 20, 2021

jbrockmendel commented Feb 23, 2021

jbrockmendel commented Mar 4, 2021

jorisvandenbossche left a comment

jbrockmendel commented Mar 4, 2021

jorisvandenbossche commented Mar 4, 2021

jorisvandenbossche commented Mar 4, 2021

jbrockmendel commented Mar 4, 2021

jorisvandenbossche commented Mar 5, 2021

jbrockmendel commented Mar 6, 2021

jorisvandenbossche commented Mar 8, 2021

jorisvandenbossche Mar 8, 2021

jbrockmendel Mar 8, 2021

jbrockmendel Mar 12, 2021

jreback left a comment

jreback Mar 16, 2021

jbrockmendel commented Mar 16, 2021

jreback commented Mar 16, 2021

simonjayhawkins commented Jun 29, 2021

jbrockmendel commented Jun 30, 2021

REF/API: DataFrame.__setitem__ never operate in-place #39510

REF/API: DataFrame.__setitem__ never operate in-place #39510

Conversation

jbrockmendel commented Jan 31, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jbrockmendel commented Feb 12, 2021

jbrockmendel commented Feb 18, 2021

jorisvandenbossche left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jbrockmendel commented Feb 18, 2021

phofl commented Feb 20, 2021

jbrockmendel commented Feb 23, 2021

jbrockmendel commented Mar 4, 2021

jorisvandenbossche left a comment

Choose a reason for hiding this comment

jbrockmendel commented Mar 4, 2021

jorisvandenbossche commented Mar 4, 2021

jorisvandenbossche commented Mar 4, 2021

jbrockmendel commented Mar 4, 2021

jorisvandenbossche commented Mar 5, 2021

jbrockmendel commented Mar 6, 2021

jorisvandenbossche commented Mar 8, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jbrockmendel commented Mar 16, 2021

jreback commented Mar 16, 2021

simonjayhawkins commented Jun 29, 2021

jbrockmendel commented Jun 30, 2021

REF/API: DataFrame.setitem never operate in-place #39510

REF/API: DataFrame.setitem never operate in-place #39510