ENH: Provide dict object for to_dict() #16122 #16220

dwkenefick · 2017-05-03T21:05:46Z

closes Feature Request: Provide dict object for pd.Series.to_dict() #16122
tests added / passed
passes git diff upstream/master --name-only -- '*.py' | flake8 --diff
whatsnew entry

dwkenefick · 2017-05-03T21:06:22Z

Hey all,

I've implemented this for pd.Series.to_dict() and pd.DataFrame.to_dict(). A few notes:

I used the into keyword.
For the into keyword, I check if the supplied class is a subclass of collections.Mapping.
I had to do a few inelegant things to accommodate collections.defaultdict, since it takes it values differently than the other mappings in collections. We can remove this, but it felt odd to support the other collections mappings but not defaultdict. For now, if the user supplies defaultdict to into, the function returns a defaultdict with None for a default_factory. This is all in the docstring.
This is marked for the v0.21.0 milestone, but I added it to the “whats new” for v0.20.0.

Let me know if any of the above should change.

TomAugspurger · 2017-05-03T21:11:15Z

I haven't looked closely yet, but for the defaultdict stuff, do are you expecting an initialized defaultdict? i.e.

df.to_dict(into=defaultdict)

or

df.to_dict(into=defaultdict(list))

I would only support the second.

This is marked for the v0.21.0 milestone, but I added it to the “whats new” for v0.20.0.

There should be a 0.20.1 whatsnew. You can add it there since it's backwards-compatible.

TomAugspurger · 2017-05-03T21:20:13Z

pandas/core/frame.py

        """
+        # GH16122
+        if not issubclass(into, collections.Mapping):


You could do something like

import inspect if not inspect.isclass(into): t = type(into) else: t = into if not issubclass(t, collections.Mapping): raise(...)

I implemented something like this in core.common.

jreback · 2017-05-03T22:20:46Z

pandas/core/series.py

+        if issubclass(into, collections.Mapping):
+            if into == collections.defaultdict:
+                return into(None, compat.iteritems(self))
+            else:


make this and the code in frame into a helper function to avoid duplication
u can put this is core.common

jreback · 2017-05-03T22:21:17Z

pandas/tests/frame/test_convert_to.py

@@ -18,44 +19,48 @@ def test_to_dict(self):
            'A': {'1': 1, '2': 2},


this will be better as a paramterization

jreback · 2017-05-03T22:21:31Z

pandas/tests/series/test_io.py

-        tm.assert_series_equal(Series(self.ts.to_dict(), name='ts'), self.ts)
+        # GH16122
+        test_maps = (
+            dict, collections.defaultdict, collections.OrderedDict)


jreback · 2017-05-03T22:21:42Z

doc/source/whatsnew/v0.20.0.txt

 - ``DataFrame.style.bar()`` now accepts two more options to further customize the bar chart. Bar alignment is set with ``align='left'|'mid'|'zero'``, the default is "left", which is backward compatible; You can now pass a list of ``color=[color_negative, color_positive]``. (:issue:`14757`)
-
+- ``Series.to_dict()`` and ``DataFrame.to_dict()`` now support an ``into`` keyword which allows you to specify the ``collections.Mapping`` subclass that you would like returned.  The default is ``dict``, which is backwards compatible. (:issue:`16122`)


Moved to doc/source/whatsnew/v0.20.1.txt

codecov · 2017-05-03T22:29:07Z

Codecov Report

Merging #16220 into master will decrease coverage by <.01%.
The diff coverage is 88.23%.

@@            Coverage Diff             @@
##           master   #16220      +/-   ##
==========================================
- Coverage   90.86%   90.86%   -0.01%     
==========================================
  Files         162      162              
  Lines       50887    50895       +8     
==========================================
+ Hits        46240    46247       +7     
- Misses       4647     4648       +1

Flag	Coverage Δ
#multiple	`88.65% <88.23%> (-0.01%)`	⬇️
#single	`40.3% <17.64%> (-0.01%)`	⬇️

Impacted Files	Coverage Δ
pandas/core/series.py	`94.92% <85.71%> (-0.08%)`	⬇️
pandas/core/frame.py	`97.54% <90%> (-0.05%)`	⬇️
pandas/core/common.py	`91.03% <0%> (+0.34%)`	⬆️

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 02eafaf...9d5c8d2. Read the comment docs.

codecov · 2017-05-03T22:29:25Z

Codecov Report

Merging #16220 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #16220      +/-   ##
==========================================
- Coverage   90.38%   90.37%   -0.02%     
==========================================
  Files         161      161              
  Lines       50916    50930      +14     
==========================================
+ Hits        46021    46026       +5     
- Misses       4895     4904       +9

Flag	Coverage Δ
#multiple	`88.14% <100%> (ø)`	⬆️
#single	`40.21% <21.73%> (-0.12%)`	⬇️

Impacted Files	Coverage Δ
pandas/core/frame.py	`97.69% <100%> (-0.1%)`	⬇️
pandas/core/common.py	`91.39% <100%> (+0.35%)`	⬆️
pandas/core/series.py	`94.71% <100%> (ø)`	⬆️
pandas/io/gbq.py	`25% <0%> (-58.34%)`	⬇️

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 4bdbcb6...8469977. Read the comment docs.

jreback

looks pretty good. some minor changes.

jreback · 2017-05-06T23:40:26Z

doc/source/whatsnew/v0.20.0.txt

@@ -515,7 +515,6 @@ Other Enhancements
 - Options added to allow one to turn on/off using ``bottleneck`` and ``numexpr``, see :ref:`here <basics.accelerate>` (:issue:`16157`)
 - ``DataFrame.style.bar()`` now accepts two more options to further customize the bar chart. Bar alignment is set with ``align='left'|'mid'|'zero'``, the default is "left", which is backward compatible; You can now pass a list of ``color=[color_negative, color_positive]``. (:issue:`14757`)

-


reset this file

jreback · 2017-05-06T23:41:46Z

pandas/core/common.py

+        the desired Mapping.
+
+    """
+    if not inspect.isclass(into):


if its not a class, then you just look at the class itself, it doesn't matter if its empty. this should return a class only.

The reason I wrote it this way is because I imagined someone might pass a pre-populated mapping to this function and expect the series or frame to be appended to what was already in the passed mapping. The error was to make it explicit that this would not happen.

Also happy to simplify this function a little - I'll remove this logic if you don't think its worth checking an empty mapping.

you are defining the interface here; this is an internal function, so it should do one thing and ideally have a limited input set. This is just determining if its a valid mapping, no need to get complicated.

Yeah, I would remove this check. This function should just be to ensure that we have an instance that's satisfies the Mapping interface.

This all makes sense - I've reworked the function. It now only returns a class or a partial (if a defaultdict is passed)

jreback · 2017-05-06T23:42:01Z

pandas/core/common.py

+        if len(into) > 0:
+            raise ValueError(
+                "to_dict() only accepts empty mappings.")
+        elif type(into) == collections.defaultdict:


use isinstance

jreback · 2017-05-06T23:42:24Z

pandas/core/common.py

+        elif type(into) == collections.defaultdict:
+            return partial(
+                collections.defaultdict, into.default_factory)
+        else:


huh? what is this trying to do

For the return partial(...) part, I'm trying to get around the way defaultdict accepts arguments. The other out-of-the-box mappings will happily accept only an iterator, which is what both of the to_dict() functions supply to the return value of this function. defaultdict also needs a default_factory, so I use partial to include the default_factory so the to_dict() functions can supply only an iterator.

For the last else, I'm just sending the class of the supplied mapping (once its been confirmed empty) back to the function to avoid duplicated logic.

If you think a different structure works better, or it should be broken up into multiple functions, let me know.

ok the defaultdict handling is fine (use isinstance(type, collections.defaultdict) though)

ok I see. the problem is you need to accept instances to deal with defaultdict, so you can now be passed an instance of defaultdict OR a class of something else (or even a class of defaultdict which is an error).

hmm. I would right off the bat detect an instance of a default dict, else only allow a mapping ABC (ex-defaultdict). Then I think you are good, no? (IOW else you would raise).

jreback · 2017-05-06T23:43:25Z

pandas/core/common.py

@@ -479,6 +481,43 @@ def _dict_compat(d):
                for key, value in iteritems(d))


+def _standardize_mapping(into):


doesn't need to be leading underscore, this is a private module. do you have a more informative name? (longer is ok).

jreback · 2017-05-06T23:43:33Z

pandas/core/common.py

+    """
+    Helper function to standardize the supplied mapping so it can
+    be passed to the ``Series.to_dict()`` and ``DataFrame.to_dict()``
+


add a versionadded

jreback · 2017-05-06T23:43:51Z

pandas/core/frame.py

+            instance of the mapping type you want.  If you want a
+            collections.defaultdict, you must pass an initialized
+            instance.
+            .. versionadded:: 0.21.0


jreback · 2017-05-06T23:44:08Z

pandas/core/frame.py

        Returns
        -------
-        result : dict like {column -> {index -> value}}
+        result : collections.Mapping like {column -> {index -> value}}
+            If ``into`` is collections.defaultdict, the return


can you add some examples of using this.

jreback · 2017-05-06T23:44:21Z

pandas/core/series.py

+            object. Can be the actual class or an empty
+            instance of the mapping type you want.  If you want a
+            collections.defaultdict, you must pass an initialized
+            .. versionadded:: 0.21.0


jreback

more comments. but lgtm.

jreback · 2017-05-08T00:16:43Z

pandas/core/common.py

+    Parameters
+    ----------
+    into : instance or subclass of collections.Mapping
+        The argument supplied to ``to_dict``.  Must be a class, an


remove references to to_dict, this is generic about mappings.

Agreed, though you can add a See Also section to point to to_dict. You can also remove references to to_dict in the summary.

Removed the references and added a See Also section.

jreback · 2017-05-08T00:18:10Z

pandas/core/common.py

@@ -479,6 +481,45 @@ def _dict_compat(d):
                for key, value in iteritems(d))


+def prep_maping_for_to_dict(into):


maybe standarize_dict_like_mapping

Or even just standardize_mapping

Changed to standardize_mapping

jreback · 2017-05-08T00:19:00Z

pandas/core/common.py

+    elif into == collections.defaultdict:
+        raise TypeError(
+            'to_dict() only accepts initialized defaultdicts')
+    else:


you can just return into here, no need for the else.

jreback · 2017-05-08T00:19:34Z

pandas/core/common.py

+        elif isinstance(into, collections.defaultdict):
+            return partial(
+                collections.defaultdict, into.default_factory)
+        else:


don't need the else (just return the partial)

jreback · 2017-05-08T00:20:13Z

pandas/core/frame.py

+        >>> df.to_dict(into=OrderedDict)
+        OrderedDict([('col2', OrderedDict([('a', 0.5), ('b', 0.75)])),
+                     ('col1', OrderedDict([('a', 1), ('b', 2)]))])
+        >>> dd = defaultdict(list)


nice examples!

Added a blank line and a preamble to this example

jorisvandenbossche · 2017-05-08T09:58:48Z

pandas/core/frame.py

+
+        >>> df.to_dict(into=OrderedDict)
+        OrderedDict([('col2', OrderedDict([('a', 0.5), ('b', 0.75)])),
+                     ('col1', OrderedDict([('a', 1), ('b', 2)]))])


Is the output of this example correct? (I mean: is it what you actually get with the PR?)
As I think if you specify an OrderedDict, I would also expect to get the result in the correct order, which means col1, col2 (not col2, col1)

I messed this up when writing the example - I switched the column order in-between writing these tests. This should be fixed now.

TomAugspurger

This look pretty close @dwkenefick. Just a few changes.

TomAugspurger · 2017-05-08T11:36:23Z

pandas/core/common.py

@@ -479,6 +481,45 @@ def _dict_compat(d):
                for key, value in iteritems(d))


+def prep_maping_for_to_dict(into):


Or even just standardize_mapping

TomAugspurger · 2017-05-08T11:36:41Z

pandas/core/common.py

+    Parameters
+    ----------
+    into : instance or subclass of collections.Mapping
+        The argument supplied to ``to_dict``.  Must be a class, an


Agreed, though you can add a See Also section to point to to_dict. You can also remove references to to_dict in the summary.

TomAugspurger · 2017-05-08T11:39:16Z

pandas/core/common.py

+        the desired Mapping.
+
+    """
+    if not inspect.isclass(into):


Yeah, I would remove this check. This function should just be to ensure that we have an instance that's satisfies the Mapping interface.

TomAugspurger · 2017-05-08T11:50:11Z

pandas/core/frame.py

                    for row in self.values]
        elif orient.lower().startswith('i'):
-            return dict((k, v.to_dict()) for k, v in self.iterrows())
+            return into_c((k, v.to_dict(into)) for k, v in self.iterrows())


Is it correct to use into here? What if the user passed an instance rather than a class? Wouldn't the values all be written into the same object?

I think this is correct. v.to_dict(into) should call standardize_mapping again. Since standardize_mapping only returns a class, I don't think there is a danger of populating the same object twice.

Gotcha.nI forgot that standardize_mapping always returned a class

TomAugspurger · 2017-05-08T11:53:21Z

pandas/tests/frame/test_convert_to.py

+        dict,
+        collections.defaultdict(list),
+        collections.OrderedDict])
+    def test_to_dict(self, mapping):


This may be tested elsewhere, but can you add a test with a dataframe that has duplicate columns? Make sure to catch the warning.

Done - I added one at the end. Let me know if that is what you were getting at.

jreback

minor typos & another test, otherwise lgtm. ping on green.

jreback · 2017-05-11T11:26:16Z

pandas/core/common.py

+def standardize_mapping(into):
+    """
+    Helper function to standardize a supplied mapping.
+    .. versionadded:: 0.21.0


I think this needs a blank line

jreback · 2017-05-11T11:27:27Z

pandas/core/common.py

+    ----------
+    into : instance or subclass of collections.Mapping
+        Must be a class, an initialized collections.defaultdict,
+        or an empty instance of a collections.Mapping subclass.


I don't think this last statement is true, it doesn't have to be empty (nor should we require it)

Removed "empty," which should make this correct.

jreback · 2017-05-11T11:28:18Z

pandas/core/series.py

+            The collections.Mapping subclass to use as the return
+            object. Can be the actual class or an empty
+            instance of the mapping type you want.  If you want a
+            collections.defaultdict, you must pass an initialized


you must pass it initialized. (maybe elsewhere as well)

Fixed here and in frame.py

jreback · 2017-05-11T11:30:04Z

pandas/tests/frame/test_convert_to.py

+        comp_data['duped'] = comp_data[df.columns[0]]
+        for k, v in compat.iteritems(comp_data):
+            for k2, v2 in compat.iteritems(v):
+                assert (v2 == recons_data[k2][k])


I would also add a test that hits some of the errors people might encounter (you do check these in the testing of standardize_mapping), but this is an integration test. you can put a test right after this, maybe test_to_dict_errors

Added this test - made sure the TypeErrors were caught. Let me know if you think anything else should go in there.

jorisvandenbossche

Added a few more comments on the docs

jorisvandenbossche · 2017-05-12T07:54:07Z

pandas/core/frame.py

+
+        Examples
+        --------
+        >>> from pandas import DataFrame


you can remove this import, but use pd.DataFrame instead of DataFrame in the other places

jorisvandenbossche · 2017-05-12T07:54:34Z

pandas/core/frame.py

+        Examples
+        --------
+        >>> from pandas import DataFrame
+        >>> from collections import OrderedDict, defaultdict


Can you move this import to the example where it is used?

jorisvandenbossche · 2017-05-12T07:56:52Z

pandas/core/series.py

        """
-        Convert Series to {label -> value} dict
+        Convert Series to {label -> value} dict or dict-like object
+        Parameters


blank line between the first line and Parameters

jorisvandenbossche · 2017-05-12T07:57:53Z

pandas/core/series.py

+
+        Examples
+        --------
+        >>> from pandas import Series


same comment here as for DataFrame (leave out the import, and use pd.Series instead for consistency within the docstrings)

jorisvandenbossche · 2017-05-12T08:07:38Z

pandas/core/frame.py

-        result : dict like {column -> {index -> value}}
+        result : collections.Mapping like {column -> {index -> value}}
+            If ``into`` is collections.defaultdict, the return
+            value's default_factory will be None.


I don't really understand this one. It seems to contrast with what is said above about a defaultdict needing to be initialized. So in that case, the default_factory will just be what is provided in the initialized defaultdict? (and which can be None, if you didn't provide a default_factory, but that seems a bit besides the point to mention here)

This is from an earlier version, and I forgot to update the docs. This sentence is wrong, and its gone now.

TomAugspurger · 2017-05-15T18:27:52Z

@jreback look good to you? Your requested changes were around standardize_mapping, and I think they were all addressed.

jorisvandenbossche

Thanks for the updates!
Some minor extra comments

jorisvandenbossche · 2017-05-15T20:57:12Z

pandas/core/series.py

-        return dict(compat.iteritems(self))
+        value_dict : collections.Mapping
+            If ``into`` is collections.defaultdict, the return
+            value's default_factory will be None.


This should be removed as well I think (you updated it for the DataFrame docstring)

jorisvandenbossche · 2017-05-15T20:58:15Z

pandas/core/frame.py

+        {'a': {'col1': 1.0, 'col2': 0.5}, 'b': {'col1': 2.0, 'col2': 0.75}}
+
+        You can also specify the mapping type.
+        >>> from collections import OrderedDict, defaultdict


Can you put a blank line between the two lines above?
(sorry for all these small comments, but sphinx / rst is quite nitpickly when it comes to whitespace, in order to render correctly in the html online documentation)

No problem. These are easy.

jreback · 2017-05-15T23:58:37Z

lgtm. I think @jorisvandenbossche had a couple of small changes.

TomAugspurger · 2017-05-16T16:06:24Z

The appveyor build finished: https://ci.appveyor.com/project/pandas-dev/pandas/build/1.0.1881 Not sure why it hasn't reported that it's done.

Thanks @dwkenefick, nice work!

* upstream/master: (48 commits) BUG: Categorical comparison with unordered (pandas-dev#16339) ENH: Adding 'protocol' parameter to 'to_pickle'. PERF: improve MultiIndex get_loc performance (pandas-dev#16346) TST: remove pandas-datareader xfail as 0.4.0 works (pandas-dev#16374) TST: followup to pandas-dev#16364, catch errstate warnings (pandas-dev#16373) DOC: new oauth token TST: Add test for clip-na (pandas-dev#16369) ENH: Draft metadata specification doc for Apache Parquet (pandas-dev#16315) MAINT: Add .iml to .gitignore (pandas-dev#16368) BUG/API: Categorical constructor scalar categories (pandas-dev#16340) ENH: Provide dict object for to_dict() pandas-dev#16122 (pandas-dev#16220) PERF: improved clip performance (pandas-dev#16364) DOC: try new token for docs DOC: try with new secure token DOC: add developer section to the docs DEPS: Drop Python 3.4 support (pandas-dev#16303) DOC: remove credential helper DOC: force fetch on build docs DOC: redo dev docs access token DOC: add dataframe construction in merge_asof example (pandas-dev#16348) ...

…6220) * ENH: Provide dict object for to_dict() pandas-dev#16122 * ENH: Provide dict object for to_dict() pandas-dev#16122 * ENH: Provide dict object for to_dict() pandas-dev#16122 * ENH: Provide dict object for to_dict() pandas-dev#16122 * ENH: Provide dict object for to_dict() pandas-dev#16122 * ENH: Provide dict object for to_dict() pandas-dev#16122 * ENH: Provide dict object for to_dict() pandas-dev#16122

ed1688 · 2018-04-18T18:23:43Z

pandas/core/frame.py

+        >>> df
+           col1  col2
+        a     1   0.1
+        b     2   0.2


The 0.1 and 0.2 are wrong, should be 0.5 and 0.75

TomAugspurger reviewed May 3, 2017

View reviewed changes

jreback requested changes May 3, 2017

View reviewed changes

jreback added the API Design label May 3, 2017

dwkenefick added 2 commits May 5, 2017 00:51

ENH: Provide dict object for to_dict() pandas-dev#16122

38fa22b

ENH: Provide dict object for to_dict() pandas-dev#16122

3070fa3

dwkenefick force-pushed the issue_16122_fix branch from 693d7dc to 3070fa3 Compare May 5, 2017 04:54

dwkenefick added 2 commits May 5, 2017 20:55

ENH: Provide dict object for to_dict() pandas-dev#16122

ccc33dd

ENH: Provide dict object for to_dict() pandas-dev#16122

73acea0

jreback requested changes May 6, 2017

View reviewed changes

jreback added the Enhancement label May 6, 2017

ENH: Provide dict object for to_dict() pandas-dev#16122

67c57e8

jreback approved these changes May 8, 2017

View reviewed changes

jorisvandenbossche reviewed May 8, 2017

View reviewed changes

TomAugspurger reviewed May 8, 2017

View reviewed changes

ENH: Provide dict object for to_dict() pandas-dev#16122

546816b

TomAugspurger approved these changes May 11, 2017

View reviewed changes

jreback requested changes May 11, 2017

View reviewed changes

jreback added this to the 0.21.0 milestone May 11, 2017

ENH: Provide dict object for to_dict() pandas-dev#16122

086c598

jorisvandenbossche reviewed May 12, 2017

View reviewed changes

dwkenefick added 2 commits May 12, 2017 19:11

ENH: Provide dict object for to_dict() pandas-dev#16122

d6c0deb

ENH: Provide dict object for to_dict() pandas-dev#16122

f297ee8

jorisvandenbossche reviewed May 15, 2017

View reviewed changes

jreback approved these changes May 15, 2017

View reviewed changes

ENH: Provide dict object for to_dict() pandas-dev#16122

8469977

TomAugspurger merged commit f040ed2 into pandas-dev:master May 16, 2017

ed1688 reviewed Apr 18, 2018

View reviewed changes

pandas/core/frame.py

>>> df

col1 col2

a 1 0.1

b 2 0.2

Copy link

ed1688 Apr 18, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The 0.1 and 0.2 are wrong, should be 0.5 and 0.75

		@@ -18,44 +19,48 @@ def test_to_dict(self):
		'A': {'1': 1, '2': 2},

		- ``DataFrame.style.bar()`` now accepts two more options to further customize the bar chart. Bar alignment is set with ``align='left'\|'mid'\|'zero'``, the default is "left", which is backward compatible; You can now pass a list of ``color=[color_negative, color_positive]``. (:issue:`14757`)

		- ``Series.to_dict()`` and ``DataFrame.to_dict()`` now support an ``into`` keyword which allows you to specify the ``collections.Mapping`` subclass that you would like returned. The default is ``dict``, which is backwards compatible. (:issue:`16122`)

		@@ -515,7 +515,6 @@ Other Enhancements
		- Options added to allow one to turn on/off using ``bottleneck`` and ``numexpr``, see :ref:`here <basics.accelerate>` (:issue:`16157`)
		- ``DataFrame.style.bar()`` now accepts two more options to further customize the bar chart. Bar alignment is set with ``align='left'\|'mid'\|'zero'``, the default is "left", which is backward compatible; You can now pass a list of ``color=[color_negative, color_positive]``. (:issue:`14757`)

		@@ -479,6 +481,43 @@ def _dict_compat(d):
		for key, value in iteritems(d))


		def _standardize_mapping(into):

		@@ -479,6 +481,45 @@ def _dict_compat(d):
		for key, value in iteritems(d))


		def prep_maping_for_to_dict(into):

ENH: Provide dict object for to_dict() #16122 #16220

ENH: Provide dict object for to_dict() #16122 #16220

Conversation

dwkenefick commented May 3, 2017

dwkenefick commented May 3, 2017

TomAugspurger commented May 3, 2017 • edited Loading

TomAugspurger May 3, 2017 • 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

jreback May 3, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov bot commented May 3, 2017 • edited Loading

Codecov Report

codecov bot commented May 3, 2017 • edited Loading

Codecov Report

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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback May 7, 2017 • 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

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

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

TomAugspurger 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

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

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

jorisvandenbossche left a comment

Choose a reason for hiding this comment

TomAugspurger commented May 3, 2017 •

edited

Loading

TomAugspurger May 3, 2017 •

edited

Loading

jreback May 3, 2017 •

edited

Loading

codecov bot commented May 3, 2017 •

edited

Loading

codecov bot commented May 3, 2017 •

edited

Loading

jreback May 7, 2017 •

edited

Loading