ERR: DatetimeIndex creation inconsistencies between list and ndarray

[sinhrks]

Setting invalid freq against ndarray(dtype=int) should raise ValueError.

import numpy as np
import pandas as pd

pd.DatetimeIndex(np.array([pd.Timestamp('2011-01-01').value, pd.Timestamp('2011-01-03').value]), freq='D')
# ValueError: Inferred frequency None from passed dates does not conform to passed frequency D

There seems to be 2 issues:

1. invalid frequency may set if data is list

pd.DatetimeIndex([pd.Timestamp('2011-01-01').value, pd.Timestamp('2011-01-03').value], freq='D')
# DatetimeIndex(['2011-01-01', '2011-01-03'], dtype='datetime64[ns]', freq='D')

2. ndarray data may raise unclear error message

pd.DatetimeIndex(np.array([pd.tslib.iNaT, pd.Timestamp('2011-01-01').value]), freq='D')
# ValueError: NaTType does not support time

[jreback]

[1] is correct.

In [8]: pd.DatetimeIndex([pd.Timestamp('2011-01-01').value, pd.Timestamp('2011-01-03').value], freq='D')
Out[8]: DatetimeIndex(['2011-01-01', '2011-01-03'], dtype='datetime64[ns]', freq='D')

In [9]: pd.DatetimeIndex([pd.Timestamp('2011-01-01').value, pd.Timestamp('2011-01-03').value], freq='D').inferred_freq

you can set any freq you want on a DatetimeIndex. It doesn't infer in this case because its too short.

[jbrockmendel]

@sinhrks the list and ndarray behavior now match. Can you confirm that part 1 of the issue is resolved. I'll take a look at the second part.