COMPAT: mpl 2.0.0

[jreback]

seems @tacaswell / @mdboom have been busy

https://github.com/matplotlib/matplotlib/releases/tag/v2.0.0b1

we should test with this.

is this going up on a conda-forge dev channel? (is there such a thing)?

[jreback]

@sinhrks @jorisvandenbossche @TomAugspurger

[tacaswell]

conda install -c conda-forge/label/rc -c conda-forge matplotlib

[jreback]

on OSX.

[Tue May 31 09:45:42 ~/pandas]$ nosetests  pandas/tests/test_graphics.py pandas/tests/test_graphics_others.py
EE
======================================================================
ERROR: Failure: AttributeError ('Cycler' object has no attribute 'change_key')
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/nose/loader.py", line 418, in loadTestsFromName
    addr.filename, addr.module)
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/nose/importer.py", line 47, in importFromPath
    return self.importFromDir(dir_path, fqname)
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/nose/importer.py", line 94, in importFromDir
    mod = load_module(part_fqname, fh, filename, desc)
  File "/Users/jreback/pandas/pandas/__init__.py", line 39, in <module>
    from pandas.core.api import *
  File "/Users/jreback/pandas/pandas/core/api.py", line 10, in <module>
    from pandas.core.groupby import Grouper
  File "/Users/jreback/pandas/pandas/core/groupby.py", line 19, in <module>
    from pandas.core.frame import DataFrame
  File "/Users/jreback/pandas/pandas/core/frame.py", line 39, in <module>
    from pandas.core.series import Series
  File "/Users/jreback/pandas/pandas/core/series.py", line 2933, in <module>
    import pandas.tools.plotting as _gfx  # noqa
  File "/Users/jreback/pandas/pandas/tools/plotting.py", line 27, in <module>
    import pandas.tseries.converter as conv
  File "/Users/jreback/pandas/pandas/tseries/converter.py", line 7, in <module>
    import matplotlib.units as units
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/matplotlib/__init__.py", line 1162, in <module>
    rcParams = rc_params()
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/matplotlib/__init__.py", line 1005, in rc_params
    return rc_params_from_file(fname, fail_on_error)
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/matplotlib/__init__.py", line 1138, in rc_params_from_file
    if key not in _all_deprecated])
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/matplotlib/__init__.py", line 889, in __init__
    self[k] = v
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/matplotlib/__init__.py", line 906, in __setitem__
    cval = self.validate[key](val)
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/matplotlib/rcsetup.py", line 840, in validate_cycler
    cycler_inst.change_key(prop, norm_prop)
AttributeError: 'Cycler' object has no attribute 'change_key'

======================================================================
ERROR: Failure: ImportError (C extension: hashtable not built. If you want to import pandas from the source directory, you may need to run 'python setup.py build_ext --inplace --force' to build the C extensions first.)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/nose/loader.py", line 418, in loadTestsFromName
    addr.filename, addr.module)
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/nose/importer.py", line 47, in importFromPath
    return self.importFromDir(dir_path, fqname)
  File "/Users/jreback/miniconda/lib/python2.7/site-packages/nose/importer.py", line 94, in importFromDir
    mod = load_module(part_fqname, fh, filename, desc)
  File "/Users/jreback/pandas/pandas/__init__.py", line 31, in <module>
    "the C extensions first.".format(module))
ImportError: C extension: hashtable not built. If you want to import pandas from the source directory, you may need to run 'python setup.py build_ext --inplace --force' to build the C extensions first.

----------------------------------------------------------------------
Ran 2 tests in 0.001s

[jreback]

pandas master

[tacaswell]

You need cycler >= 0.10. Should have pulled that from conda-forge

[jreback]

hmm, maybe needs to be part of your recipe?

[tacaswell]

xref: conda-forge/matplotlib-feedstock#27

[sinhrks]

There looks some errors / failures (most of them are caused by figure size check). Will dig in.

Can we add one more SLOW build using mpl2.0 on Travis with allowed failure?

ERROR: test_errorbar_asymmetrical (pandas.tests.test_graphics.TestDataFramePlots)
ERROR: test_errorbar_scatter (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_bar_linewidth (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_boxplot (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_boxplot_colors (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_boxplot_vertical (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_hexbin_basic (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_hist_df (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_kde_df (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_line_colors_and_styles_subplots (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_line_lim (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_plot (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_plot_scatter (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_scatter_colors (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_subplots (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_subplots_layout (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_subplots_multiple_axes (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_subplots_timeseries (pandas.tests.test_graphics.TestDataFramePlots)
FAIL: test_bar_log (pandas.tests.test_graphics.TestSeriesPlots)
FAIL: test_hist_layout_with_by (pandas.tests.test_graphics.TestSeriesPlots)
FAIL: test_plot (pandas.tests.test_graphics.TestSeriesPlots)

[jreback]

@sinhrks sure. is 2.0 rc available on a channel somewhere?

[sinhrks]

Not yet. Will try to built from source if it's not available until I need.

[jorisvandenbossche]

I think it is available on conda-forge: https://anaconda.org/conda-forge/matplotlib/labels (edit: at least the beta, not rc)

[tacaswell]

There is no rc yet, should be one 'soon'

[TomAugspurger]

@sinhrks for the figsize failures, I think the new default in MPL is 6.4 x 4.8. We had the old 8 x 6 as the default for _check_axes_shape. Will just need to make that default None, and determine it based on the MPL version.

[TomAugspurger]

@tacaswell does this make sense to you? Regarding errorbar

import matplotlib.pyplot as plt
fig, ax = plt.subplots()
r = ax.errorbar(range(5), range(5), xerr=.2)
print(ax.lines)

In MPL 1.5.x

[<matplotlib.lines.Line2D object at 0x1193411d0>, <matplotlib.lines.Line2D object at 0x1193414a8>, <matplotlib.lines.Line2D object at 0x119341da0>]

in MPL 2.0 beta

[<matplotlib.lines.Line2D object at 0x117a79128>]

I found matplotlib/matplotlib#5047 (don't draw caps by default) but not sure if that's what's going on here.

[tacaswell]

We changed the default dpi so reduced the figure size (in inches) to keep the same figure size (in pixels).

re errorbar: that does look correct, the caps are not created which does not even add the artists. I would look at the contents of r rather than ax.lines for testing purposes.

[TomAugspurger]

@sinhrks I can work on this if you want.

[sinhrks]

@TomAugspurger It would be very helpful!