-
-
Notifications
You must be signed in to change notification settings - Fork 18.4k
Test failures on debian sid #4918
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
@avalentino try with PyTables 2.4...! |
@avalentino Any reason why you're setting |
@avalentino Can you show the output of ci/print_versions.py |
@jreback e are working to push PyTables 3.0 (and numexpr 2.2) in sid and my purpose is to check if the new version of pytables breaks some of the dependent packages.
@cpcloud no, mainly a question of habit |
well, I'm running tests from a checkout built with
|
|
Okay now can you show the output of If you do |
Unfortunately now I'm no more able to run the test suite:
|
@avalentino Would you be willing to set up a |
@cpcloud OK, I managed to setup a virtual environment with:
and then I installed pandas from the check-out directory (updated this mourning).
|
@avalentino can you show the latest entry in the |
cc @Komnomnomnom, this is on mac, so maybe something new for the json issues? |
are u sure it's on a mac? mac paths usually don't look like that |
@avalentino can you run the tests from the cloned directory (instead of |
@avalentino so to recap: # assuming you're in the virtualenv
git clone git://github.com/pydata/pandas.git
python setup.py develop
nosetests -w pandas |
nvm not a mac...was reading the |
Hi @jreback, hi @cpcloud,
With this setup if I run the commands suggested by @cpcloud
the result is the one reported in the comment above. |
I suspect (at least some) of the JSON issues are due to the locale setting (it_IT), will try to test and fix |
@avalentino can you run in a virtualenv without system site packages? would make it easier to debug... |
@cpcloud I can reproduce most of these on my Arch64 setup by adding import locale
locale.setlocale(locale.LC_ALL, 'it_IT.UTF-8') to
|
i believe the google stuff is time zone sensitive so that should be fixed if possible. not sure about the json stuff. i would bet u have some ideas though 😄 |
@cpcloud we have 1 build with an encoding |
or maybe add a separate new build |
problem is that only 5 builds can run at a time i believe so going to be slowest build + 6th build time == 😪 |
i'll take a look and assign myself to keep track of things |
JSON fix would be to force locale to en_GB (or en_US) during serialisation or deserialisation. Decimal separator is always '.' according to JSON spec. Will send through a PR. It's weird that @avalentino is seeing these failures though because I didn't think Python used the system's locale unless it was forcibly set using |
@cpcloud why don't we just set it the main build then - or is that weird? (or 2.6 build) just to try to catch stuff like this |
@jreback i think 2.6 build would be fine, i'll try and get something up this week @Komnomnomnom thanks! very much appreciated |
I can confirm that using the C locale for numbers and dates things go better:
|
@avalentino Can you pull down my PR #5073 and see if anything is still failing? Much appreciated! |
|
those tests betray my english centric mind :) good that the others are not failing anymore these are easy "fix" ... just skip when the locale is not |
@avalentino pls test with #5073 and lmk if anything comes up. thx! |
if nothing comes up i'll close |
@cpcloud all tests pass now, thanks. |
closed by #5073 |
Hi, I'm testing the current pandas master on debian sid with updated numexpr and PyTables from experimental and I'm experimenting some test failures:
Some cfg info:
The text was updated successfully, but these errors were encountered: