tm.assert_series_equal checking on Integer EA should be exact by default?

[simonjayhawkins]

Code Sample, a copy-pastable example if possible

>>>
>>> import pandas as pd
>>> pd.__version__
'0.26.0.dev0+1328.g03dffb1eb'
>>>
>>> import pandas.testing as tm
>>> s1 = pd.Series([10000000000000000], dtype="Int64")
>>> s1
0    10000000000000000
dtype: Int64
>>>
>>> s2 = pd.Series([9999999999999999], dtype="Int64")
>>> s2
0    9999999999999999
dtype: Int64
>>>
>>> tm.assert_series_equal(s1,s2)
>>>

Problem description

does not raise

Expected Output

AssertionError: ExtensionArray are different

Output of pd.show_versions()

INSTALLED VERSIONS

commit : 03dffb1
python : 3.7.3.final.0
python-bits : 64
OS : Windows
OS-release : 10
machine : AMD64
processor : Intel64 Family 6 Model 69 Stepping 1, GenuineIntel
byteorder : little
LC_ALL : None
LANG : en_GB.UTF-8
LOCALE : None.None

pandas : 0.26.0.dev0+1328.g03dffb1eb
numpy : 1.17.3
pytz : 2019.3
dateutil : 2.8.0
pip : 19.3.1
setuptools : 42.0.2.post20191201
Cython : 0.29.14
pytest : 5.3.2
hypothesis : 4.54.1
sphinx : 2.3.0
blosc : None
feather : None
xlsxwriter : 1.2.6
lxml.etree : 4.4.2
html5lib : 1.0.1
pymysql : None
psycopg2 : None
jinja2 : 2.10.3
IPython : 7.10.2
pandas_datareader: None
bs4 : 4.8.1
bottleneck : 1.3.1
fastparquet : 0.3.2
gcsfs : None
lxml.etree : 4.4.2
matplotlib : 3.1.2
numexpr : 2.7.0
odfpy : None
openpyxl : 3.0.1
pandas_gbq : None
pyarrow : 0.15.1
pytables : None
pytest : 5.3.2
s3fs : 0.4.0
scipy : 1.3.1
sqlalchemy : 1.3.11
tables : 3.6.1
xarray : 0.14.1
xlrd : 1.2.0
xlwt : 1.3.0
xlsxwriter : 1.2.6

discovered while investigating #30268

[TomAugspurger]

What do we do for a NumPy int64-dtype array?

[simonjayhawkins]

That behaves the same, perhaps less restrictive issue title? I see some previous discussion on this (e.g. #26272) and concerns of it being a breaking API change.

For Integer EA, maybe this won't apply?

IMO integers should be exact, could be identifiers, i.e. invoice number. If our tests are not checking for exactness by default then we could have latent bugs.

with check_exact works fine.

[TomAugspurger]

Agreed. +1 for sure on IntegerArray, and probably on int64, though I don't recall the details on why it's not exact right now.

[jbrockmendel]

Instead of making the default behavior dtype-specific, should we just make the default in assert_extension_array_equal check_exact=True? I think also in tm.assert_equal we hardcode check_exact=False which should change.

[mroeschke]

Closing in favor of #55882