pandas-dev / pandas

Flexible and powerful data analysis / manipulation library for Python, providing labeled data structures similar to R data.frame objects, statistical functions, and much more
https://pandas.pydata.org
BSD 3-Clause "New" or "Revised" License
43.19k stars 17.77k forks source link

Possible test_plotting regression in 0.11 #3113

Closed ghost closed 11 years ago

ghost commented 11 years ago

I'm getting failures on master. Travis full_deps should have caught this, maybe something is screwy about my system. passes on 10.1.

check it out before releasing 0.11, fix travis if needs be.

nosetests ./pandas/tseries/tests/test_plotting.py
.....................EE..EEEEE........EE...........E.

======================================================================
ERROR: test_tsplot (pandas.tseries.tests.test_plotting.TestTSPlot)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/user1/src/pandas/pandas/tseries/tests/test_plotting.py", line 97, in test_tsplot
    _check_plot_works(f, s.index.freq, ax=ax, series=s)
  File "/home/user1/src/pandas/pandas/tseries/tests/test_plotting.py", line 967, in _check_plot_works
    with ensure_path() as path:
NameError: global name 'ensure_path' is not defined
jreback commented 11 years ago

my fault....I guess these are not tested locally (when I run nose)..... I'll push in a few

ghost commented 11 years ago

thanks, still need to figure out why travis didn't catch it.

jreback commented 11 years ago

it throws a NameError maybe somehow suppressed in the visualization routines?

ghost commented 11 years ago

if nose fails, travis should fail.

jreback commented 11 years ago

those tests are all marked slow....so not run

jreback commented 11 years ago

maybe have a travis run that does the slow tests (???) could be in the one that does the alternate encoding?

then maybe we should also mark really-slow tests as well

jreback commented 11 years ago

fixed up in master

ghost commented 11 years ago

yeah, maybe add a new build though. better not mix different concerns. I'd junk the vbench run, which generates useless data points if I weren't sure someday that data will make for an awesome pandas/statsmodels presentation on novelty detection in noisy data.

ghost commented 11 years ago

I thought I fixed FULL_DEPS to do slow tests. turns out it's slow even without.

ghost commented 11 years ago

jeff, took your suggestion and put it into the locale build.

60a8575