TST: failing test at tseries/tests/test_base.py TestDatetimeIndexOps test_nat

[MattRijk]

A small, complete example of the issue

======================================================================
FAIL: test_nat (pandas.tseries.tests.test_base.TestDatetimeIndexOps)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/matt/pandas-mattrijk/pandas/tseries/tests/test_base.py", line 878, in test_nat
    tm.assert_numpy_array_equal(idx._nan_idxs, np.array([], dtype=np.int64))
  File "/home/matt/pandas-mattrijk/pandas/util/testing.py", line 1084, in assert_numpy_array_equal
    assert_attr_equal('dtype', left, right, obj=obj)
  File "/home/matt/pandas-mattrijk/pandas/util/testing.py", line 878, in assert_attr_equal
    left_attr, right_attr)
  File "/home/matt/pandas-mattrijk/pandas/util/testing.py", line 1019, in raise_assert_detail
    raise AssertionError(msg)
AssertionError: numpy array are different

Attribute "dtype" are different
[left]:  int32
[right]: int64

----------------------------------------------------------------------
Ran 92 tests in 3.580s

FAILED (failures=1)


Process finished with exit code 0

Expected Output

failing test

idx = pd.DatetimeIndex(['2011-01-01', '2011-01-02'], tz=tz)
tm.assert_numpy_array_equal(idx._nan_idxs, np.array([], dtype=np.int64))

idx = pd.DatetimeIndex(['2011-01-01', 'NaT'], tz=tz)
tm.assert_numpy_array_equal(idx._nan_idxs, np.array([1], dtype=np.int64))

test passes

tm.assert_numpy_array_equal(idx._nan_idxs, np.array([], dtype=np.int32))

In Ipython on 32bit the values look fine.

In [3]: idx = pd.DatetimeIndex(['2016-01-01', 'NaT'], tz=None)

In [4]: idx.dtype
Out[4]: dtype('<M8[ns]')

In [5]: idx._nan_idxs
Out[5]: array([], dtype=int64)

In [6]: arry = np.array([], dtype=np.int64)

In [7]: arry
Out[7]: array([], dtype=int64)

In [8]: import pandas.util.testing as tm

In [9]: tm.assert_numpy_array_equal(idx._nan_idxs, arry))
Out[9]: True

This fixes it but I think theres more going on.

import platform
if platform.architecture()[0] == '32bit':
tm.assert_numpy_array_equal(idx._nan_idxs, np.array([], dtype=np.int32))
else:
tm.assert_numpy_array_equal(idx._nan_idxs, np.array([], dtype=np.int64))

Output of pd.show_versions()

## INSTALLED VERSIONS

commit: None
python: 2.7.12.final.0
python-bits: 32
OS: Linux
OS-release: 3.13.0-92-generic
machine: i686
processor: i686
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: None.None

pandas: 0.19.0
nose: 1.3.7
pip: 8.1.2
setuptools: 27.2.0
Cython: 0.24.1
numpy: 1.11.1
scipy: 0.18.1
statsmodels: None
xarray: None
IPython: 5.1.0
sphinx: 1.4.6
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.6.1
blosc: None
bottleneck: 1.1.0
tables: 3.2.3.1
numexpr: 2.6.1
matplotlib: 1.5.3
openpyxl: 2.4.0
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.3
lxml: 3.6.4
bs4: 4.3.2
html5lib: 0.999
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: 0.7.6.None
psycopg2: None
jinja2: 2.8
boto: None
pandas_datareader: 0.2.1

[chris-b1]

Thanks for the report, I think the test just needs written with an expectation of np.intp rather than np.int64. We don't run CI on 32 bit Linux, so some of these kind of things sneak in.