Incorrect behaviour when merging dfs using on and left_index or right_index with outer

[alan-wong]

data

df1 = pd.DataFrame({'A': ['A0', 'A1', 'A2', 'A3'],
                'B': ['B0', 'B1', 'B2', 'B3'],
                'C': ['C0', 'C1', 'C2', 'C3'],
                'D': ['D0', 'D1', 'D2', 'D3']},index=[0, 1, 2, 3])
df2 = pd.DataFrame({'A': ['A0', 'A5', 'A6', 'A7'],
                'B': ['B1', 'B5', 'B6', 'B7'],
                'C': ['A1', 'C5', 'C6', 'C7'],
                'D': ['B1', 'D5', 'D6', 'D7']},index=[4, 5, 6, 7])

With merge using left_index=True with outer:

In [68]:
pd.merge(df1, df2, how='outer', left_index=True, left_on='A', right_on='A')

Out[68]:
    A  B_x  C_x  D_x  B_y  C_y  D_y
4  A0   B0   C0   D0   B1   A1   B1
7  A1   B1   C1   D1  NaN  NaN  NaN
7  A2   B2   C2   D2  NaN  NaN  NaN
7  A3   B3   C3   D3  NaN  NaN  NaN
5  A5  NaN  NaN  NaN   B5   C5   D5
6  A6  NaN  NaN  NaN   B6   C6   D6
7  A7  NaN  NaN  NaN   B7   C7   D7

When we try the opposite:

In [69]:
pd.merge(df1, df2, how='outer', right_index=True, left_on='A', right_on='A')

Out[69]:
    A  A_x  B_x  C_x  D_x  A_y  B_y  C_y  D_y
0  A0   A0   B0   C0   D0  NaN  NaN  NaN  NaN
1  A1   A1   B1   C1   D1  NaN  NaN  NaN  NaN
2  A2   A2   B2   C2   D2  NaN  NaN  NaN  NaN
3  A3   A3   B3   C3   D3  NaN  NaN  NaN  NaN
3   4  NaN  NaN  NaN  NaN   A0   B1   A1   B1
3   5  NaN  NaN  NaN  NaN   A5   B5   C5   D5
3   6  NaN  NaN  NaN  NaN   A6   B6   C6   D6
3   7  NaN  NaN  NaN  NaN   A7   B7   C7   D7

I'd expect the left_index version to contain the index of the left df or is my understanding incorrect here and the result should be the opposite of the right_index version?

Version:

In [71]:

pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 3.4.3.final.0
python-bits: 64
OS: Windows
OS-release: 7
machine: AMD64
processor: Intel64 Family 6 Model 42 Stepping 7, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None

pandas: 0.19.2
nose: 1.3.7
pip: 9.0.1
setuptools: 17.1.1
Cython: 0.22.1
numpy: 1.11.0
scipy: 0.17.0
statsmodels: None
xarray: None
IPython: 3.2.1
sphinx: 1.3.1
patsy: 0.4.0
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: None
tables: None
numexpr: None
matplotlib: 1.4.3
openpyxl: None
xlrd: 0.9.4
xlwt: None
xlsxwriter: 0.7.3
lxml: None
bs4: 4.3.2
html5lib: None
httplib2: 0.9.2
apiclient: None
sqlalchemy: 1.0.8
pymysql: None
psycopg2: None
jinja2: 2.7.3
boto: None
pandas_datareader: 0.2.1

There seems to be an inconsistency here and bug with left_index version and possibly also the right_index

[jreback]

dupe of #16228