CLN/DEPS: Clean up post numpy bump to 1.12 #23796

h-vetinari · 2018-11-19T21:32:18Z

follow-up to DEPS: drop numpy < 1.12 #23062
tests added / passed
passes git diff upstream/master -u -- "*.py" | flake8 --diff

I stumbled over a place I had forgotten to change in #23062 - README.md. And since I recently found some grep-like powers on windows with findstr, I took this opportunity to have a look through the code base for other things that got missed in that PR.

Turns out there was quite a bit of crusty old code left from before the last bump to numpy 1.9 (the oldest I found was some commented out code due to segfaults for numpy < 1.5.1).

Also, since #23062 bumped matplotlib to 2.0.0, I looked for old code there as well.

There were a few places where it was not 100% clear what the right way to remove them was - I'm relying on a passing test suite to be a good enough yardstick. In any case, this should have a thorough review to make sure I didn't eff something up.

I'll add a more detailed review right away to explain my reasoning behind each one of the changes.

h-vetinari

Some more details on the how/why and open questions.

h-vetinari · 2018-11-19T21:33:46Z

pandas/core/dtypes/cast.py

+        except Exception:
+            # upcast to object
+            dtype = np.object_
+            fill_value = np.nan


This follows the comments in the compat code above. Not sure if this is correct, but the tests pass.

h-vetinari · 2018-11-19T21:34:47Z

pandas/core/dtypes/missing.py

@@ -408,6 +408,7 @@ def array_equivalent(left, right, strict_nan=False):

    # Object arrays can contain None, NaN and NaT.
    # string dtypes must be come to this path for NumPy 1.7.1 compat
+    # TODO: remove old numpy compat code (or comment)


In this case, I'm not sure what's to be done here (expand below; no numpy code to be found)

remove this comment

I don't think that "for NumPy 1.7.1 compat" should be left standing, but ok.

h-vetinari · 2018-11-19T21:35:32Z

pandas/core/internals/blocks.py

@@ -3145,17 +3138,6 @@ def _merge_blocks(blocks, dtype=None, _can_consolidate=True):
    return blocks


-def _vstack(to_stack, dtype):


This was only used once, and doesn't need to be defined at all once the shim isn't necessary anymore

h-vetinari · 2018-11-19T21:36:07Z

pandas/core/internals/managers.py

@@ -1178,8 +1163,7 @@ def insert(self, loc, item, value, allow_duplicates=False):
                blk.mgr_locs = new_mgr_locs

        if loc == self._blklocs.shape[0]:
-            # np.append is a lot faster (at least in numpy 1.7.1), let's use it
-            # if we can.
+            # np.append is a lot faster, let's use it if we can.


No code changed here, just the comment

h-vetinari · 2018-11-19T21:37:21Z

pandas/io/packers.py

@@ -251,7 +251,6 @@ def dtype_for(t):
            'complex128': np.float64,
            'complex64': np.float32}

-# numpy 1.6.1 compat
 if hasattr(np, 'float128'):
    c2f_dict['complex256'] = np.float128


This is not just numpy compat (hence the removed comment) - including 'complex256': np.float128 into c2f_dict does not work on my windows machine, for example

maybe a comment to that effect? "windows compat"or "windows 32 bit compat" or "[whatever would be accurate] compat"?

h-vetinari · 2018-11-19T21:39:02Z

pandas/tests/arrays/categorical/test_repr.py

@@ -37,7 +37,6 @@ def test_big_print(self):
    def test_empty_print(self):
        factor = Categorical([], ["a", "b", "c"])
        expected = ("[], Categories (3, object): [a, b, c]")
-        # hack because array_repr changed in numpy > 1.6.x
        actual = repr(factor)
        assert actual == expected


Fair enough to say that the repr of np.array has arrived, no? no need to call getting a list from it a hack, IMO

h-vetinari · 2018-11-19T21:40:01Z

pandas/tests/indexing/test_chaining_and_caching.py

@@ -94,7 +94,6 @@ class TestChaining(object):
    def test_setitem_chained_setfault(self):

        # GH6026
-        # setfaults under numpy 1.7.1 (ok on 1.8)


Comment not relevant any more, IMO (issue reference is still there)

h-vetinari · 2018-11-19T21:40:19Z

pandas/tests/indexing/test_iloc.py

@@ -337,7 +337,7 @@ def test_iloc_setitem_list(self):
        tm.assert_frame_equal(df, expected)

    def test_iloc_setitem_pandas_object(self):
-        # GH 17193, affecting old numpy (1.7 and 1.8)
+        # GH 17193


h-vetinari · 2018-11-19T21:40:31Z

pandas/tests/io/test_pytables.py

@@ -199,8 +199,6 @@ def roundtrip(key, obj, **kwargs):
    def test_long_strings(self):

        # GH6166
-        # unconversion of long strings was being chopped in earlier
-        # versions of numpy < 1.7.2


h-vetinari · 2018-11-19T21:41:15Z

pandas/tests/test_base.py

@@ -292,11 +292,10 @@ def test_none_comparison(self):
                assert not result.iat[0]
                assert not result.iat[1]

-                # this fails for numpy < 1.9
-                # and oddly for *some* platforms


Works for me locally, let's see what the CI says...

jbrockmendel · 2018-11-19T21:51:35Z

pandas/core/internals/blocks.py

-        if values.dtype != _NS_DTYPE:
-            # Workaround for numpy 1.6 bug
-            values = conversion.ensure_datetime64ns(values)
+        values = conversion.ensure_datetime64ns(values)


possibly pass copy=False?

No problem.

jbrockmendel · 2018-11-19T21:52:27Z

pandas/core/internals/managers.py

-        return np.c_[nz, counts[nz]]
+    counts = np.bincount(arr.astype(np.int_))
+    nz = counts.nonzero()[0]
+    return np.c_[nz, counts[nz]]


is this used often? worth keeping as a standalone function vs inlining?

Got no strong opinion on that.

its called twice as an operand for a loop, so ok to keep

jbrockmendel · 2018-11-19T21:54:41Z

pandas/tests/indexes/common.py

@@ -850,12 +850,9 @@ def test_equals_op(self):
            tm.assert_series_equal(series_a == item, Series(expected3))

    def test_numpy_ufuncs(self):
-        # test ufuncs of numpy 1.9.2. see:
+        # test ufuncs of numpy, see:


should the version number be removed or updated?

Ufuncs are here to stay now, no? That's why I removed it.

Unless you mean that the set of ufuncs is related to the version number?

h-vetinari

Of note: I don't know many of the touched code paths, so I'm just trying to make an educated guess what should be removed. Didn't try to go 2 steps beyond and worry about further simplifications that might now be possible

h-vetinari · 2018-11-19T21:54:37Z

pandas/core/internals/blocks.py

-        if values.dtype != _NS_DTYPE:
-            # Workaround for numpy 1.6 bug
-            values = conversion.ensure_datetime64ns(values)
+        values = conversion.ensure_datetime64ns(values)


No problem.

h-vetinari · 2018-11-19T21:55:05Z

pandas/core/internals/managers.py

-        return np.c_[nz, counts[nz]]
+    counts = np.bincount(arr.astype(np.int_))
+    nz = counts.nonzero()[0]
+    return np.c_[nz, counts[nz]]


Got no strong opinion on that.

jbrockmendel · 2018-11-19T21:59:39Z

pandas/_libs/lib.pyx

@@ -485,8 +485,7 @@ def astype_intsafe(ndarray[object] arr, new_dtype):
        bint is_datelike
        ndarray result

-    # on 32-bit, 1.6.2 numpy M8[ns] is a subdtype of integer, which is weird
-    is_datelike = new_dtype in ['M8[ns]', 'm8[ns]']
+    is_datelike = new_dtype == 'm8[ns]'



core.dtypes.cast is the only place where this gets called, and it has a comment # work around for NumPy brokenness, #1987, might have more things worth cleaning up

Removed this and the call in dtypes.cast. pandas/tests/dtypes/test_cast.py still passes, so let's try how this goes here.

h-vetinari

Thanks for checking this out

h-vetinari · 2018-11-19T22:04:20Z

pandas/_libs/lib.pyx

@@ -485,8 +485,7 @@ def astype_intsafe(ndarray[object] arr, new_dtype):
        bint is_datelike
        ndarray result

-    # on 32-bit, 1.6.2 numpy M8[ns] is a subdtype of integer, which is weird
-    is_datelike = new_dtype in ['M8[ns]', 'm8[ns]']
+    is_datelike = new_dtype == 'm8[ns]'



Removed this and the call in dtypes.cast. pandas/tests/dtypes/test_cast.py still passes, so let's try how this goes here.

h-vetinari · 2018-11-19T22:05:31Z

pandas/io/packers.py

@@ -251,7 +251,6 @@ def dtype_for(t):
            'complex128': np.float64,
            'complex64': np.float32}

-# numpy 1.6.1 compat
 if hasattr(np, 'float128'):
    c2f_dict['complex256'] = np.float128


h-vetinari · 2018-11-19T22:06:19Z

pandas/tests/indexes/common.py

@@ -850,12 +850,9 @@ def test_equals_op(self):
            tm.assert_series_equal(series_a == item, Series(expected3))

    def test_numpy_ufuncs(self):
-        # test ufuncs of numpy 1.9.2. see:
+        # test ufuncs of numpy, see:


Ufuncs are here to stay now, no? That's why I removed it.

Unless you mean that the set of ufuncs is related to the version number?

h-vetinari

Reverted a change that was causing a segfault. Solution TBD

h-vetinari · 2018-11-19T22:23:46Z

pandas/core/dtypes/cast.py

@@ -267,6 +267,8 @@ def maybe_promote(dtype, fill_value=np.nan):
        # for now: refuse to upcast datetime64


replacing this compat code with

if issubclass(dtype.type, (np.datetime64, np.timedelta64)): try: fill_value = tslibs.Timedelta(fill_value).value except Exception: # upcast to object dtype = np.object_ fill_value = np.nan

leads to a segfault in tests/test_take.py::TestTake::test_2d_datetime64. have not investigated further

jreback · 2018-11-20T01:22:30Z

looks fine. needs a rebase and everything passing. ping when green. if something is not working after trying, can just mark it and come back with a followup.

h-vetinari · 2018-11-21T17:27:03Z

pandas/core/dtypes/cast.py

        if np.issubdtype(dtype.type, np.integer):
+            # TODO: this is an old numpy compat branch that is not necessary


@jbrockmendel Could you have a look at this. Couldn't fix it easily, not sure how relevant that test is.

Yikes. Not sure off the top whats going on here.

Why the comment saying it is no longer necessary? np.timedelta64 is still a subdtype of np.integer

To the extent that this branch is just for timedelta64 dtypes, I would make that explicit instead of checking if np.issubdtype(dtype.type, np.integer)

The original comment (and issue) was about unsafe casting from object to int - that's not the case for timedelta types, and even more so, there's a dedicated branch for such dtypes below that is never hit (and actually leads to a test failure in tests/reshape/merge/test_merge.py::TestMerge:;test_other_timedelta_unit). That's why I wanted to ask if you have an idea about it. :)

jreback · 2018-11-21T17:28:27Z

pandas/core/dtypes/cast.py

@@ -723,8 +725,19 @@ def astype_nansafe(arr, dtype, copy=True, skipna=False):

    elif is_object_dtype(arr):

-        # work around NumPy brokenness, #1987
        if np.issubdtype(dtype.type, np.integer):


so change this to a test on is_timedelta_dtype then

Reverting back to original now contradicts this other review comment of yours. But I agree this whole thing is something for a follow-up

@h-vetinari as the code evolves, I re-evaluate if this is a good change or not. This got messy real fast and is not worth doing here.

@jreback
That's fair enough, but - by that time - I had already reverted the code changes and only added an explanatory comment. I'd say this would be value added for anyone stumbling across these lines, but have removed this now as well.

jreback

small comment

h-vetinari · 2018-11-21T21:21:10Z

pandas/core/dtypes/cast.py

        if np.issubdtype(dtype.type, np.integer):
+            # TODO: this is an old numpy compat branch that is not necessary


The original comment (and issue) was about unsafe casting from object to int - that's not the case for timedelta types, and even more so, there's a dedicated branch for such dtypes below that is never hit (and actually leads to a test failure in tests/reshape/merge/test_merge.py::TestMerge:;test_other_timedelta_unit). That's why I wanted to ask if you have an idea about it. :)

h-vetinari · 2018-11-21T21:22:35Z

pandas/tests/frame/test_constructors.py

-        # rec.dtype.names = list(rec.dtype.names)[::-1]
+        if PY3:
+            # unicode error under PY2
+            rec.dtype.names = list(rec.dtype.names)[::-1]


Not sure if this ancient code path (that has been commented out since forever) is worth keeping. It's failing on PY2, passing on PY3

h-vetinari · 2018-11-21T21:23:14Z

pandas/tests/plotting/test_series.py

-        # in mpl 1.5+ this is a TypeError
-        with pytest.raises((ValueError, TypeError)):
+        # MPL > 2.0.0 will most likely use TypeError here
+        with pytest.raises((TypeError, ValueError)):


This was still raising ValueErrors in the 2.0.0 builds

codecov · 2018-11-22T00:15:40Z

Codecov Report

Merging #23796 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #23796      +/-   ##
==========================================
+ Coverage   92.29%   92.31%   +0.01%     
==========================================
  Files         161      161              
  Lines       51498    51464      -34     
==========================================
- Hits        47530    47507      -23     
+ Misses       3968     3957      -11

Flag	Coverage Δ
#multiple	`90.7% <100%> (+0.01%)`	⬆️
#single	`42.42% <90.9%> (-0.01%)`	⬇️

Impacted Files	Coverage Δ
pandas/io/packers.py	`88.08% <ø> (ø)`	⬆️
pandas/core/arrays/categorical.py	`95.34% <ø> (-0.02%)`	⬇️
pandas/plotting/_misc.py	`38.68% <ø> (-0.31%)`	⬇️
pandas/core/arrays/datetimelike.py	`95.96% <ø> (ø)`	⬆️
pandas/core/arrays/sparse.py	`91.93% <ø> (-0.02%)`	⬇️
pandas/core/indexes/datetimes.py	`96.48% <ø> (+0.28%)`	⬆️
pandas/core/dtypes/dtypes.py	`95.59% <100%> (-0.03%)`	⬇️
pandas/core/reshape/tile.py	`94.73% <100%> (ø)`	⬆️
pandas/core/dtypes/cast.py	`89.62% <100%> (+0.63%)`	⬆️
pandas/core/internals/blocks.py	`93.71% <100%> (+0.03%)`	⬆️
... and 1 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a7b187a...ce5bd94. Read the comment docs.

pep8speaks · 2018-11-22T18:54:05Z

Hello @h-vetinari! Thanks for updating the PR.

There are no PEP8 issues in the file pandas/core/arrays/categorical.py !
There are no PEP8 issues in the file pandas/core/arrays/datetimelike.py !
There are no PEP8 issues in the file pandas/core/arrays/sparse.py !
There are no PEP8 issues in the file pandas/core/dtypes/cast.py !
There are no PEP8 issues in the file pandas/core/dtypes/dtypes.py !
There are no PEP8 issues in the file pandas/core/dtypes/missing.py !
There are no PEP8 issues in the file pandas/core/indexes/datetimes.py !
There are no PEP8 issues in the file pandas/core/internals/blocks.py !
There are no PEP8 issues in the file pandas/core/internals/managers.py !
There are no PEP8 issues in the file pandas/core/reshape/tile.py !
There are no PEP8 issues in the file pandas/io/packers.py !
There are no PEP8 issues in the file pandas/io/pickle.py !
There are no PEP8 issues in the file pandas/plotting/_misc.py !
There are no PEP8 issues in the file pandas/tests/arrays/categorical/test_repr.py !
There are no PEP8 issues in the file pandas/tests/dtypes/test_inference.py !
There are no PEP8 issues in the file pandas/tests/frame/test_constructors.py !
There are no PEP8 issues in the file pandas/tests/frame/test_operators.py !
There are no PEP8 issues in the file pandas/tests/indexes/common.py !
There are no PEP8 issues in the file pandas/tests/indexes/multi/test_analytics.py !
There are no PEP8 issues in the file pandas/tests/indexing/test_chaining_and_caching.py !
There are no PEP8 issues in the file pandas/tests/indexing/test_iloc.py !
There are no PEP8 issues in the file pandas/tests/io/test_pytables.py !
There are no PEP8 issues in the file pandas/tests/plotting/test_datetimelike.py !
There are no PEP8 issues in the file pandas/tests/plotting/test_frame.py !
There are no PEP8 issues in the file pandas/tests/plotting/test_series.py !
There are no PEP8 issues in the file pandas/tests/series/test_analytics.py !
There are no PEP8 issues in the file pandas/tests/test_base.py !
There are no PEP8 issues in the file pandas/tests/test_nanops.py !
There are no PEP8 issues in the file pandas/tests/test_panel.py !
There are no PEP8 issues in the file pandas/tests/test_sorting.py !

jreback

except for all of the changes in the integer casting this is fine. remove those and can merge.

jreback · 2018-11-23T02:47:16Z

pandas/core/dtypes/cast.py

@@ -264,27 +264,22 @@ def maybe_promote(dtype, fill_value=np.nan):

    # returns tuple of (dtype, fill_value)
    if issubclass(dtype.type, (np.datetime64, np.timedelta64)):
-        # for now: refuse to upcast datetime64
-        # (this is because datetime64 will not implicitly upconvert


so rather than do it this way, just make the top level 2 checks (1 for timedelta, 1 for datetime) then you don't need to check twice.

jreback · 2018-11-23T02:47:55Z

pandas/core/dtypes/cast.py

+            try:
+                fill_value = tslibs.Timedelta(fill_value).value
+            except Exception:
+                dtype = np.object_


are these exceptions actually hit in any tests?

The thing is that maybe_promote is not tested directly (see also #23833), I'm working on a PR for that currently.

Also, I've had a segfault in pandas/tests/test_take.py::TestTake::test_2d_datetime64 on an earlier attempt, but let's see.

jreback · 2018-11-23T02:49:38Z

pandas/core/dtypes/cast.py

+            # anymore for its original purpose (unsafe casting from object to
+            # int, see GH 1987).
+            # Currently, timedelta dtypes get routed through here; whereas
+            # uncommenting them would re-call (see below)


these comments are way too long. what exactly are you trying to do here? let's leave this entire casting as is for now w/o any changes. this needs a separate followup.

I'm trying to leave some breadcrumbs as I spent 2-3 hours trying to investigate this.

I've shortened the comment now.

i would rather you not and just revert back to what it was. if you want to make an issue ok.

jreback · 2018-11-23T02:50:22Z

pandas/core/dtypes/missing.py

@@ -408,6 +408,7 @@ def array_equivalent(left, right, strict_nan=False):

    # Object arrays can contain None, NaN and NaT.
    # string dtypes must be come to this path for NumPy 1.7.1 compat
+    # TODO: remove old numpy compat code (or comment)


remove this comment

jreback · 2018-11-27T02:50:28Z

thanks @h-vetinari nice cleanup

h-vetinari added 2 commits November 19, 2018 22:24

CLN/DEPS: Clean up post numpy bump to 1.12

70a8d0e

Clean up post MPL bump to 2.0

8e00645

h-vetinari commented Nov 19, 2018

View reviewed changes

jbrockmendel reviewed Nov 19, 2018

View reviewed changes

h-vetinari commented Nov 19, 2018

View reviewed changes

jbrockmendel reviewed Nov 19, 2018

View reviewed changes

h-vetinari added 2 commits November 19, 2018 23:02

Revert change that causes segfault in take

91e9710

Review (jbrockmendel)

17d971d

h-vetinari commented Nov 19, 2018

View reviewed changes

jreback added Clean Dependencies Required and optional dependencies labels Nov 20, 2018

h-vetinari mentioned this pull request Nov 21, 2018

BUG/Internals: maybe_promote #23833

Open

h-vetinari added 2 commits November 21, 2018 17:34

Merge remote-tracking branch 'upstream/master' into post_numpy_bump

aef3654

Revert removal that breaks a test

7ef44b6

h-vetinari commented Nov 21, 2018

View reviewed changes

jreback reviewed Nov 21, 2018

View reviewed changes

jreback requested changes Nov 21, 2018

View reviewed changes

h-vetinari added 8 commits November 21, 2018 21:47

Fix isort

83813d7

Add if branch for old code failing under PY2

f8a68a8

Partly revert MPL compat fix

aef0fe2

revert fix in arrays/categorical.py

1fdaecd

Correctly rename azure container

d2cb333

Change timedelta-branch in cast.py (review jreback)

d854b29

Merge remote-tracking branch 'upstream/master' into post_numpy_bump

8995d94

Re-add missing empty lines

d2b1f78

h-vetinari commented Nov 21, 2018

View reviewed changes

Re-add integer subtype branch due to failed IntegerArray tests

9c2cfbd

h-vetinari added 2 commits November 22, 2018 19:52

Merge remote-tracking branch 'upstream/master' into post_numpy_bump

9bac669

Second attempt at removing compat code from maybe_promote

ecfe824

jreback requested changes Nov 23, 2018

View reviewed changes

h-vetinari added 4 commits November 23, 2018 08:40

Merge remote-tracking branch 'upstream/master' into post_numpy_bump

91d0501

Review (jreback)

844c2aa

Merge remote-tracking branch 'upstream/master' into post_numpy_bump

65f06e8

Review (jreback)

ce5bd94

h-vetinari mentioned this pull request Nov 26, 2018

CLN: Remove/deprecate unused/misleading dtype functions #23917

Merged

jreback added this to the 0.24.0 milestone Nov 27, 2018

jreback approved these changes Nov 27, 2018

View reviewed changes

jreback merged commit 6af9c30 into pandas-dev:master Nov 27, 2018

h-vetinari deleted the post_numpy_bump branch November 27, 2018 06:43

h-vetinari mentioned this pull request Nov 28, 2018

TST: add test coverage for maybe_promote #23982

Merged

4 tasks

h-vetinari mentioned this pull request Dec 19, 2018

CLN: post-post numpy bump #24365

Merged

Pingviinituutti pushed a commit to Pingviinituutti/pandas that referenced this pull request Feb 28, 2019

CLN/DEPS: Clean up post numpy bump to 1.12 (pandas-dev#23796)

385a95d

Pingviinituutti pushed a commit to Pingviinituutti/pandas that referenced this pull request Feb 28, 2019

CLN/DEPS: Clean up post numpy bump to 1.12 (pandas-dev#23796)

6f11ca6

		@@ -3145,17 +3138,6 @@ def _merge_blocks(blocks, dtype=None, _can_consolidate=True):
		return blocks


		def _vstack(to_stack, dtype):

		@@ -267,6 +267,8 @@ def maybe_promote(dtype, fill_value=np.nan):
		# for now: refuse to upcast datetime64

		if np.issubdtype(dtype.type, np.integer):
		# TODO: this is an old numpy compat branch that is not necessary

CLN/DEPS: Clean up post numpy bump to 1.12 #23796

CLN/DEPS: Clean up post numpy bump to 1.12 #23796

Conversation

h-vetinari commented Nov 19, 2018 • edited Loading

h-vetinari left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

h-vetinari Nov 19, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

h-vetinari left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

h-vetinari left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

h-vetinari left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Nov 20, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov bot commented Nov 22, 2018 • edited Loading

Codecov Report

pep8speaks commented Nov 22, 2018

jreback left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Nov 27, 2018

h-vetinari commented Nov 19, 2018 •

edited

Loading

h-vetinari Nov 19, 2018 •

edited

Loading

codecov bot commented Nov 22, 2018 •

edited

Loading