df.Truncate now fails when arguments have different tzinfo

[JrtPec]

Code Sample

df = pd.DataFrame()
date1 = pd.Timestamp('20190101', tz='Europe/Brussels')
date2 = pd.Timestamp('20190201T00:00:00+01:00')
df.truncate(before=date1, after=date2)

File "/python3.7/site-packages/pandas/core/indexes/base.py", line 5244, in slice_locs
    raise ValueError("Both dates must have the " "same UTC offset")
ValueError: Both dates must have the same UTC offset

Problem description

In #25263, these lines were added (line 4881):

if not tz_compare(ts_start.tzinfo, ts_end.tzinfo):
    raise ValueError("Both dates must have the same UTC offset"

1. The timestamps in the example do have the same UTC offset. The difference is that one has an actual timezone, while the other has a fixed offset. So a simple comparison of tzinfo is insufficient here.
2. It is unclear to my why you shouldn't be able to truncate with timestamps with different offsets!?

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.7.3.final.0
python-bits : 64
OS : Darwin
OS-release : 18.6.0
machine : x86_64
processor : i386
byteorder : little
LC_ALL : en_US.UTF-8
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8
pandas : 0.25.3
numpy : 1.16.4
pytz : 2019.1
dateutil : 2.8.0
pip : 19.2.1
setuptools : 28.8.0
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.3.3
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : None
pandas_datareader: None
bs4 : 4.7.1
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : 4.3.3
matplotlib : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
s3fs : None
scipy : None
sqlalchemy : 1.3.3
tables : None
xarray : None
xlrd : 1.2.0
xlwt : None
xlsxwriter : None

[MarcoGorelli]

Europe/Brussels and FixedOffset(60) aren't the same though - the former does daylight saving:

>>> pytz.timezone('Europe/Brussels').localize(dt.datetime(2002, 9, 27, 6, 0, 0))                                                                                                                                         
datetime.datetime(2002, 9, 27, 6, 0, tzinfo=<DstTzInfo 'Europe/Brussels' CEST+2:00:00 DST>)

>>> pytz.FixedOffset(60).localize(dt.datetime(2002, 9, 27, 6, 0, 0))                                                                                                                                                     
datetime.datetime(2002, 9, 27, 6, 0, tzinfo=pytz.FixedOffset(60))

[JrtPec]

Your comment is correct.
Yet, in my example, the UTC Offsets of both timestamps are identical. The error says they aren't.

My current workaround is this:

# df = df.truncate(before=before, after=after) ERROR
df = df.truncate(before=before)
df = df.truncate(after=after)

So I'm still clueless as to why the oneliner that used to work in all previous versions of Pandas now suddenly doesn't.

[1kastner]

@JrtPec check #16785 and therein referred to for a lengthy discussion on why. It is technically very difficult and the previous behavior was unpredictable (pandas was even sometimes showing a wrong behavior) because sometimes the timezone information accidentially got ignored.

In conclusion, it is a feature, not a bug ;-)

[akhorshidi]

1. The timestamps in the example do have the same UTC offset. The difference is that one has an actual timezone, while the other has a fixed offset. So a simple comparison of `tzinfo` is insufficient here.

I've faced the same issue but with the DateTime objects:

start_date = datetime(start_date , tzinfo=timezone.utc)
end_date = datetime(end_date, tzinfo=pytz.utc)

and my workaround was to change the value of tzinfo argument inside the first datetime object from timezone.utc to pyzt.utc !