inf comparison doesn't emulate NaN with use_inf_as_null

[wcbeard]

Code Sample, a copy-pastable example if possible

import numpy as np
import pandas as pd

with pd.option_context('mode.use_inf_as_null', True):
    s = pd.Series([np.nan, np.inf, 0])
    d = DataFrame(OrderedDict([
        ('s', s),
        ('repr(s)', s.map(repr)),
        ('s == s', s == s),
        ('s == np.inf', s == np.inf),
        ('s.values == np.inf', s.values == np.inf),
    ]))

d
=>

     s repr(s)  s == s  s == np.inf  s.values == np.inf
0  NaN     nan   False        False               False
1  NaN     inf    True        False                True
2  0.0     0.0    True        False               False

Problem description

I expected s == s to return False for inf values with use_inf_as_null mode on, or at least for s == np.inf to agree with s.values == np.inf.

Expected Output

     s repr(s)  s == s  s == np.inf  s.values == np.inf
0  NaN     nan   False        False               False
1  NaN     inf   False         True                True
2  0.0     0.0    True        False               False

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.5.3.final.0 python-bits: 64 OS: Darwin OS-release: 15.5.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8

pandas: 0.20.2
pytest: 3.1.2
pip: 9.0.1
setuptools: 33.1.1.post20170320
Cython: 0.25.2
numpy: 1.12.1
scipy: 0.19.0
xarray: None
IPython: 5.0.0
sphinx: None
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2017.2
blosc: None
bottleneck: None
tables: 3.4.2
numexpr: 2.6.2
feather: 0.4.0
matplotlib: 2.0.2
openpyxl: 2.4.8
xlrd: 1.0.0
xlwt: None
xlsxwriter: None
lxml: 3.8.0
bs4: 4.5.3
html5lib: 0.999999999
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.9.5
s3fs: None
pandas_gbq: None
pandas_datareader: None

[chris-b1]

If we're fully copying NaN semantics, then the output for s = np.inf is correct.

In [19]: s = pd.Series([np.nan, np.inf, 0])

In [20]: s == np.nan
Out[20]: 
0    False
1    False
2    False
dtype: bool

I do agree on the s == s output - I'm not sure this mode is that widely tested, PRs welcome.

[wcbeard]

This also seems a bit weird

s = pd.Series([np.nan, np.inf, 0])
s.isin([np.inf])
=>
0     True
1     True
2    False
dtype: bool

[jbrockmendel]

use_inf_as_null has been gone for a while. closing.