-
Notifications
You must be signed in to change notification settings - Fork 26
Conversation
(ooo whops, travis failed since this is dependant on PR 55) |
Even with #55 applied, I get
|
@dustin --- so: any idea why e.g. what ver of pytz do you have? since I'm not getting same behavior That issue sounds like: https://bugs.launchpad.net/pytz/+bug/310606 (which I found from searching ala: http://code.davidjanes.com/blog/2008/12/22/pytz-utcoffset/ |
Indeed, it does sound like the error you've identified.
so, that's the latest, and certainly after something released in 2010. |
try rebasing onto master now that #55 is merged, and let's see what happens in travis.. |
I thought maybe this would be due to a different zoneinfo installed on my (Gentoo) system. But from an strace, it looks like pytz ships its own zoneinfo:
Adding a
so, indeed, it looks like it's selecting the first one of those. The code basically builds a table of transitions from one offset to another, and the LMT offset spans from 1901 to 1918. |
Looks like this is a regression in pytz. I'm loathe to require an old pytz though, what-do-you-say here?
|
If it's broken, we shouldn't use it. Can you update the upstream ticket to indicate that this is (re-?)broken in 2014.3, and just require pytz==2014.1 in relengapi? If you add a comment on that requirement, and indicate that relengapi's tests will fail for broken versions of pytz, then someone in the future will feel comfortable upgrading pytz if the tests pass. Yay test! |
Tests rule :D |
Filed: https://bugs.launchpad.net/pytz/+bug/1324158 for the pytz issue |
now = datetime.datetime.now().replace(tzinfo=pytz.timezone("US/Pacific")) | ||
session.add(DevTable(date=now)) | ||
session.commit() | ||
instance = session.query(DevTable).all()[0] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
instance = session.query(DevTable).one() might be nicer here
Don't forget Dustin's comment: "and indicate that relengapi's tests will fail for broken versions of pytz" - I think this part is still missing. At the moment people might not realise we have a test for it. |
No description provided.