Resample with period='AS'/'A' does not work on localized indexes for multiple methods in "how"

[jphme]

I encountered a similar issue to (closed) #3592 .

When trying to resample a dataframe with a localized Datetime Index using the how method:

rng = pd.date_range('1/1/2011', periods=20000, freq='H')
rng = rng.tz_localize('EST')
ts = pd.DataFrame(index=rng)
ts['first']=np.random.randn(len(rng))
ts['second']=np.cumsum(np.random.randn(len(rng)))
ts2=ts.resample('A', how={'first':np.sum, 'second':np.mean})

*File "tslib.pyx", line 465, in pandas.tslib._Timestamp.__richcmp__ (pandas\tslib.c:9313)
*TypeError: can't compare offset-naive and offset-aware datetimes

For other (,lower) frequencies like "M" the same code works fine and if I just use

how=np.mean

or just use 1 column in the dict

how={'first':np.mean}

it works too.

python 2.7.5
pandas 0.13.1

EDIT: Changed to include full example

[jreback]

can you provide the starting dataframe (code to recreate, doesn't have to be long), that reproduces this problem?

[jphme]

Changed the example so you can recreate it now!

If that means anything: with pandas 0.12 i get a TypeError, with pandas 0.13.1 it changed to ValueError (but the issue is the same).

[jreback]

gr8...thanks...will get to it for 0.14