Possible Duplicate:
Weird timezone issue with pytz
This seems wrong:
>>> import pytz
>>> z1 = timezone('America/Edmonton')
>>> z2 = timezone('US/Mountain')
>>> z1
<DstTzInfo 'America/Edmonton' LMT-1 day, 16:26:00 STD>
>>> z2
<DstTzInfo 'US/Mountain' MST-1 day, 17:00:00 STD>
>>> pytz.VERSION
'2012f'
>>>
'America/Edmonton' and 'US/Eastern' should be the same time zone (17:00:00 STD). Not to mention 16:26:00 doesn't make any sense.
-- Update --
The above makes sense in context of Jon Skeet's answer. However, things get strange when I do this:
>>> d = datetime.now()
>>> d
datetime.datetime(2012, 10, 9, 15, 21, 41, 644706)
I created a naive date. Since 'America/Edmonton' is my timezone, I try to set that manually:
>>> d2 = d.replace(tzinfo=timezone('America/Edmonton'))
>>> d2
datetime.datetime(2012, 10, 9, 15, 21, 41, 644706, tzinfo=<DstTzInfo 'America/Edmonton' LMT-1 day, 16:26:00 STD>)
This should not have change anything because that is the correct TZ. However:
>>> d2.astimezone(timezone('US/Eastern'))
datetime.datetime(2012, 10, 9, 18, 55, 41, 644706, tzinfo=<DstTzInfo 'US/Eastern' EDT-1 day, 20:00:00 DST>)
This should give me an offset of 2 hours (difference between 'US/Eastern' and 'America/Edmonton') but it gives me 3 hours 26 minutes (which is 2 hours plus one hour 26 minutes :D )
inserting timezone('US/Mountain')
produces the correct result in astimezone()
. Creating an aware datetime with 'America/Edmonton' will also work correctly.