Series append function drops timezone information

[dirkdevriendt]

Except when one of the two series is empty; in that case a tz-aware Series is returned

Code to reproduce:

import numpy as np
import pandas as pd
rng1 = pd.date_range('1/1/2011 01:00', periods=1, freq='H')
rng2 = pd.date_range('1/1/2011 02:00', periods=1, freq='H')
ts1 = pd.Series(np.random.randn(len(rng1)), index=rng1).tz_localize('UTC')
ts2 = pd.Series(np.random.randn(len(rng2)), index=rng2).tz_localize('UTC')
ts_result = ts1.append(ts2)

Result in IPython:

In [1]: import numpy as np

In [2]: import pandas as pd

In [3]: rng1 = pd.date_range('1/1/2011 01:00', periods=1, freq='H')

In [4]: rng2 = pd.date_range('1/1/2011 02:00', periods=1, freq='H')

In [5]: ts1 = pd.Series(np.random.randn(len(rng1)), index=rng1).tz_localize('UTC')

In [6]: ts1
Out[6]:
2011-01-01 01:00:00+00:00 1.870471
Freq: H

In [7]: ts2 = pd.Series(np.random.randn(len(rng2)), index=rng2).tz_localize('UTC')

In [8]: ts2
Out[8]:
2011-01-01 02:00:00+00:00 1.266536
Freq: H

In [9]: ts_result = ts1.append(ts2)

In [10]: ts_result
Out[10]:
2011-01-01 01:00:00 1.870471
2011-01-01 02:00:00 1.266536