COMPAT/TST Matplotlib 2.0 compatability #13662
Conversation
TomAugspurger
added
Visualization
| @@ -613,3 +613,8 @@ Bug Fixes | |||
| - Bug in ``groupby`` with ``as_index=False`` returns all NaN's when grouping on multiple columns including a categorical one (:issue:`13204`) | |||
|
|
|||
| - Bug where ``pd.read_gbq()`` could throw ``ImportError: No module named discovery`` as a result of a naming conflict with another python package called apiclient (:issue:`13454`) | |||
|
|
|||
| - Compatibility with maptlolib 2.0 (:issue:`13333`) | |||
There was a problem hiding this comment.
As changes are only for tests, there is no bugs related to mpl2. Not sure on what section the description should be...
TomAugspurger
force-pushed
the
mpl-2-compat
branch
from
96c7b7f to
af516a8
Compare
Current coverage is 85.27% (diff: 66.66%)@@ master #13662 diff @@
==========================================
Files 139 139
Lines 50446 50452 +6
Methods 0 0
Messages 0 0
Branches 0 0
==========================================
+ Hits 43018 43022 +4
- Misses 7428 7430 +2
Partials 0 0
|
TomAugspurger
force-pushed
the
mpl-2-compat
branch
2 times, most recently
from
d4ce722 to
d85767b
Compare
|
This build tested against the mpl 2.0 beta. https://travis-ci.org/pydata/pandas/jobs/145822265#L1193 Things seem to be OK.
Under "Other enhancements". Look good? |
| @@ -346,7 +355,7 @@ def _check_axes_shape(self, axes, axes_num=None, layout=None, | |||
| self.assertEqual(result, layout) | |||
|
|
|||
| self.assert_numpy_array_equal( | |||
| np.round(visible_axes[0].figure.get_size_inches()), | |||
| visible_axes[0].figure.get_size_inches(), # why round? | |||
There was a problem hiding this comment.
Ahh, right will do.
There was a problem hiding this comment.
Forgot I removed it and everything worked.
TomAugspurger
force-pushed
the
mpl-2-compat
branch
from
d85767b to
47fd717
Compare
TomAugspurger
changed the title
|
Is this ready? (needs a rebase) Or do you want to wait until matplotlib has a rc release? (but we can also merge, and just update when there is one) |
|
Let me rebase quick and run one of the builds on the mpl pre-release at |
TomAugspurger
force-pushed
the
mpl-2-compat
branch
3 times, most recently
from
77edda7 to
e0b5663
Compare
|
There is an error in the py35 build: https://travis-ci.org/pydata/pandas/jobs/152242074, but not sure this is related to this PR |
TomAugspurger
force-pushed
the
mpl-2-compat
branch
3 times, most recently
from
80d450f to
d44753d
Compare
TomAugspurger
force-pushed
the
mpl-2-compat
branch
5 times, most recently
from
d92fb17 to
285ca66
Compare
|
Anyone have a clue what's going on in here? The tests pass, but flake8 seems to fail with
|
|
I also saw it, I was wondering if adding conda-forge as a channel could have to do something with it? (I don't see any other changes that could cause it). You could check by removing adding the channel and only specifying the channel for the one install line of matplotlib beta. |
TomAugspurger
force-pushed
the
mpl-2-compat
branch
from
285ca66 to
fdedc50
Compare
|
@jorisvandenbossche that did it 🎉 (no idea why, but oh well). https://travis-ci.org/pydata/pandas/builds/154140072 was the one run against the MPL 2.0 beta. Good to merge? |
|
@TomAugspurger Thanks! Maybe it would be good to report that issue to conda-forge somewhere. There is a flake8 feedstock, so possibly something wrong there. EDIT: done -> conda-forge/flake8-feedstock#3 |
git diff upstream/master | flake8 --diffJust a WIP for now, I want to cleanup how I handle the errorbar changes.
Will wait for a matplotlib RC before changing the Travis matrix I suppose?