API: DateTimeIndex slicing removes name label

[diezguerra]

Shouldn't it carry all possible properties instead?

Steps to reproduce:

import pandas as pd


st = pd.Timestamp('2013-07-01 00:00:00', tz='America/Los_Angeles')
et = pd.Timestamp('2013-07-02 00:00:00', tz='America/Los_Angeles')

dr = pd.date_range(st, et, freq='H', name='timebucket')
assert dr.name == 'timebucket'
dr = dr[1:]
assert dr.name == None

[jreback]

related to #4039 I think, cc @jtratner

[cpcloud]

hm weird. looks like the overridden __getitem__ is not being called...

[cpcloud]

In [9]: paste
st = pd.Timestamp('2013-07-01 00:00:00', tz='America/Los_Angeles')
et = pd.Timestamp('2013-07-02 00:00:00', tz='America/Los_Angeles')

dr = pd.date_range(st, et, freq='H', name='timebucket')

## -- End pasted text --

In [10]: dr[1:]
Out[10]:
<class 'pandas.tseries.index.DatetimeIndex'>
[2013-07-01 01:00:00, ..., 2013-07-02 00:00:00]
Length: 24, Freq: H, Timezone: America/Los_Angeles

In [11]: dr[1:].name

In [12]: dr.__getitem__(slice(1, None)).name
Out[12]: 'timebucket'

[jreback]

I think is a short cut I put in, something like getitem_slice in tseries/index

[cpcloud]

same issue for Index(['a','b']) as well

[cpcloud]

problem is __getslice__ i htink

http://docs.python.org/2/reference/datamodel.html#additional-methods-for-emulation-of-sequence-types

[cpcloud]

passes when i define that in terms of __getitem__

[cpcloud]

pr coming