BUG: Regression in sort_index with ascending=None

[YarShev]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

# Your code here
import pandas
import numpy as np
df = pandas.DataFrame(np.arange(64))
length = len(df.index)
df.index = [(i - length / 2) % length for i in range(length)]
df.sort_index(axis=0, ascending=None, na_position="first")
ValueError: the 'axis' parameter is not supported in the pandas implementation of argsort()

Problem description

There was no such issue in pandas version 1.1.5

Output of pd.show_versions()

pandas : 1.2.1
numpy : 1.19.5
pytz : 2020.5
dateutil : 2.8.1
pip : 20.3.3
setuptools : 51.1.2.post20210112
Cython : None
pytest : 6.2.1
hypothesis : None
sphinx : None
blosc : None
feather : 0.4.1
xlsxwriter : None
lxml.etree : 4.6.2
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.12.0
pandas_datareader: None
bs4 : 4.9.3
bottleneck : None
fsspec : 0.8.3
fastparquet : None
gcsfs : None
matplotlib : 3.2.2
numexpr : 2.7.2
odfpy : None
openpyxl : 3.0.6
pandas_gbq : None
pyarrow : 2.0.0
pyxlsb : None
s3fs : 0.5.2
scipy : 1.6.0
sqlalchemy : 1.3.22
tables : 3.6.1
tabulate : None
xarray : 0.16.2
xlrd : 2.0.1
xlwt : None
numba : None

[YarShev]

The same is true for Series.sort_index(...).

[attack68]

ascending is an argument which accepts boolean or list of booleans. I do not believe it should be considered a bug or a regression if the execution has changed from one version to another with an erroneous input argument.

I would not know what output to expect with ascending=None?

[jreback]

we should be validating the ascending arg type as boolean - it's possible we r not

[YarShev]

I would not know what output to expect with ascending=None?

As I see, ascending=None and ascending=False behaviors were identical in 1.1.5 version. So, I think it would be good to add a check for ascending.

[simonjayhawkins]

moving to 1.2.3

[simonjayhawkins]

There was no such issue in pandas version 1.1.5

first bad commit: [aa43f17] REF: nargsort incorrectly calling _values_for_argsort (#37266)

As I see, ascending=None and ascending=False behaviors were identical in 1.1.5 version. So, I think it would be good to add a check for ascending.

@ivanovmg an alternative could be to restore the old behaviour for 1.2.x and repurpose #39451 for 1.3 and then can advertise the 'breaking change', which is permissible since the usage is incorrect.

[jreback]

no i think we should just fix this - the. new is wrong

[ivanovmg]

@jreback @simonjayhawkins
Can you please clarify what will be the preferable way?
Should I restore the old behavior in nargsort as @simonjayhawkins recommends?
@jreback it is not clear for my what you suggest here.