Remove Index.__inv__

[toobaz]

Code Sample, a copy-pastable example if possible

In [1]: import pandas as pd

In [2]: idx = pd.Index([2, 4, 6])

In [3]: idx.__inv__()
Out[3]: Int64Index([-2, -4, -6], dtype='int64')

Problem description

idx.__inv__ does what idx.__neg__ does, and not do what its name suggests (that is, operator.inv). By the way, ~idx raises TypeError anyway, and idx.__invert__ does not exist, differently from what the Python docs suggest.

Notice that instead

In [4]: s = pd.Series([2, 4, 6])

In [5]: ~s
Out[5]: 
0   -3
1   -5
2   -7
dtype: int64

In [6]: s.__invert__()
Out[6]: 
0   -3
1   -5
2   -7
dtype: int64

which is correct (and Series.__inv__ does not exist).

Expected Output

Out[3]: Int64Index([-3, -5, -7], dtype='int64')

Given how wrong, undocumented and not exposed this is, I suggest to immediately remove Index.__inv__. Then we can later decide to introduce Index.__invert__ which does the right thing, and also enable the ~ operator.

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.5.3.final.0
python-bits: 64
OS: Linux
OS-release: 4.9.0-6-amd64
machine: x86_64
processor:
byteorder: little
LC_ALL: None
LANG: it_IT.UTF-8
LOCALE: it_IT.UTF-8

pandas: 0.24.0.dev0+460.geb0ac5437
pytest: 3.5.0
pip: 9.0.1
setuptools: 39.2.0
Cython: 0.28.4
numpy: 1.14.3
scipy: 0.19.0
pyarrow: None
xarray: None
IPython: 6.2.1
sphinx: 1.5.6
patsy: 0.5.0
dateutil: 2.7.3
pytz: 2018.4
blosc: None
bottleneck: 1.2.0dev
tables: 3.3.0
numexpr: 2.6.1
feather: 0.3.1
matplotlib: 2.2.2.post1634.dev0+ge8120cf6d
openpyxl: 2.3.0
xlrd: 1.0.0
xlwt: 1.3.0
xlsxwriter: 0.9.6
lxml: 4.1.1
bs4: 4.5.3
html5lib: 0.999999999
sqlalchemy: 1.0.15
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: 0.2.1
gcsfs: None

[jorisvandenbossche]

By the way, ~idx raises TypeError anyway

Do you know why that is? Because in principle, if __inv__() works, so should ~ (that might indicate another bug in our code)

The TypeError is raised by the Index.__inv__ implementation, but Int64Index overrides that with a working implementation.

[toobaz]

Do you know why that is? Because in principle, if inv() works, so should ~ (that might indicate another bug in our code)

To the best of my understanding, and at least from Python 2.7 onwards, ~ is overridden just by using __invert__, not __inv__ (although the latter is exposed by the operators module).

[jbrockmendel]

We define __invert__ in 5 places, __inv__ in one. Is one of these wrong?

[toobaz]

Given that __inv__ has the wrong name (for what it does), and that the name is of an obscure method (not supported by any other pandas object, not sure it even exists in numpy, and at the same time trivial to compute as 1/idx), I think this bug should be fixed by:

• making a copy of the method at https://github.com/pandas-dev/pandas/blob/master/pandas/core/indexes/base.py#L5429 with the name changed to __invert__, so that at least the right method works
• insert a deprecation warning in the original __inv__ telling users to switch to __invert__ (or ~), and later delete it

(unless the first point was done already - can't test at the moment what ~idx does in master)

[toobaz]

By the way, I'm not even sure a method which only exists as __inv__ requires a deprecation cycle...

[jbrockmendel]

closed by #45006