BUG: Bug in DataFrame construction with recarray and non-ns datetime dtype (GH6140)

[jreback]

closes #6140

[ghost]

===============================================================================
ERROR: test_from_records_with_datetimes (pandas.tests.test_frame.TestDataFrame)
-------------------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/buildslave/nd-bb-slave-sparc-wheezy/pandas-py3_x-wheezy-sparc/build/venv/lib/python3.2/site-packages/pandas-0.13.0_420_ga67c3cd-py3.2-linux-sparc64.egg/pandas/tests/test_frame.py", line 3970, in test_from_records_with_datetimes
    recarray = np.core.records.fromarrays(arrdata, dtype=dtypes)
  File "/usr/lib/python3/dist-packages/numpy/core/records.py", line 566, in fromarrays
    _array[_names[i]] = arrayList[i]
ValueError: Must be a datetime.date or datetime.datetime object

[ghost]

Best thing I ever did.

[jreback]

funny thing when I was doing this was thinking that their is a bug that caused all of the other sparc issues

namely that for some reason dtype comparisons don't work on date time types (but u can compare then as strings) - weird because it works on Linux in general

but getting off topic !

and I put in that exact comparison

ok 1.5 hrs to compile / test

[ghost]

2.5h average delay +1.5 run actually. But at least you know there's a problem the same day
you commited, not 3 weeks after.