Concatenating two localized datetimes returns NaTs

[andyljones]

Issue:

import pandas as pd
from datetime import datetime

### Concat'ing two UTC times works
first = pd.DataFrame([[datetime(2016, 1, 1)]])
first[0] = first[0].dt.tz_localize('UTC')

second = pd.DataFrame([[datetime(2016, 1, 2)]])
second[0] = second[0].dt.tz_localize('UTC')

print pd.concat([first, second]) # Works


### Concat'ing two London times doesn't work
first = pd.DataFrame([[datetime(2016, 1, 1)]])
first[0] = first[0].dt.tz_localize('Europe/London')

second = pd.DataFrame([[datetime(2016, 1, 2)]])
second[0] = second[0].dt.tz_localize('Europe/London')

print pd.concat([first, second]) # Doesn't work - returns a DF full of NaTs


### Concat'ing 2+1 London times works
first = pd.DataFrame([[datetime(2016, 1, 1)], [datetime(2016, 1, 2)]])
first[0] = first[0].dt.tz_localize('Europe/London')

second = pd.DataFrame([[datetime(2016, 1, 3)]])
second[0] = second[0].dt.tz_localize('Europe/London')

print pd.concat([first, second]) # Works


### Concat'ing 1+2 London times doesn't work
first = pd.DataFrame([[datetime(2016, 1, 1)]])
first[0] = first[0].dt.tz_localize('Europe/London')

second = pd.DataFrame([[datetime(2016, 1, 2)], [datetime(2016, 1, 3)]])
second[0] = second[0].dt.tz_localize('Europe/London')

print pd.concat([first, second]) # Doesn't work - first row is NaT

Dependencies:

INSTALLED VERSIONS
------------------
commit: None
python: 2.7.11.final.0
python-bits: 64
OS: Linux
OS-release: 3.17.7-300.fc21.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.UTF-8

pandas: 0.17.1
nose: None
pip: 8.0.2
setuptools: 19.4
Cython: None
numpy: 1.10.2
scipy: 0.16.0
statsmodels: None
IPython: 4.0.1
sphinx: 1.3.1
patsy: None
dateutil: 2.4.2
pytz: 2015.7
blosc: None
bottleneck: None
tables: None
numexpr: 2.4.4
matplotlib: 1.5.0
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: 3.5.0
bs4: 4.4.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None
Jinja2: None

[pag]

I can replicate with python 2.7.10, windows 10, pandas 0.17.1, numpy 1.9.3.

[jreback]

hmm, partially was addressed in #12195

but still looks buggy