ENH: Added multicolumn/multirow support for latex #14184

sgsaenger · 2016-09-08T12:14:10Z

closes Use \multirow \multicolumn in latex output #13508
tests added / passed
passes git diff upstream/master | flake8 --diff
whatsnew entry

Print names of MultiIndex columns.
Added "multicol" and "multirow" flags to to_latex
which trigger the corresponding feature.
Multirow adds clines to visually separate sections.

TomAugspurger · 2016-09-08T17:17:22Z

pandas/core/frame.py

@@ -1613,7 +1613,7 @@ def to_latex(self, buf=None, columns=None, col_space=None, header=True,
                 index=True, na_rep='NaN', formatters=None, float_format=None,
                 sparsify=None, index_names=True, bold_rows=True,
                 column_format=None, longtable=None, escape=None,
-                 encoding=None, decimal='.'):
+                 encoding=None, decimal='.',multicol=None,multirow=None):


You'll need spaces after the ,s here (check git diff upstream/master | flake8 --diff)

I see!
Should have started using this sooner, quite nice.

TomAugspurger · 2016-09-28T15:04:47Z

You'll need to rebase your branch on master. Just post here if you need help with the git commands.

sgsaenger · 2016-09-28T15:15:33Z

Oh my, what a mess...
I did git rebase upstream/master as in the documentation, which seemed to work correctly.
Before pushing, though, git performed an automatic merge of my local and remote branch.
Any way to clean this up? Thanks for helping!

TomAugspurger · 2016-09-28T15:35:30Z

You'll want to start by

git fetch upstream
git rebase -i upstream/master

This will start applying changes, but you'll have some merge conflicts. You'll need to edit the files (see git status for "Unmerged Paths") and look for the conflict markers git auto inserts (<<<<<<<, ======== and >>>>>>>). Fixup those files to look like how the should (make sure to remove the markers too), and the files you just edited, and then

git rebase --continue

Side-issue, typically you'll develop on a feature branch, no master. So before you make any changes you would git checkout -b <branch-name>. Not a big deal now though.

We can also jump into the chat room if that's easier.

sgsaenger · 2016-09-28T17:03:23Z

pandas/formats/format.py

+                                    .format(ncol, coltext.strip()))
+                    else:
+                        row2.append(coltext)
+                for c in crow[ilevels:]:


Without the multicolumn feature, long names in higher levels break the layout severely:

compared to:

Thus, i would propose this feature to be enabled by default.

This means, that empty columns won't be printed, which may alter the code of the line with the row-names.
Not the result though, as latex ignores these trailing empty parts anyways.
Should i rather set the default to False and maintain compatibility in code, or fix the rendered output by rewriting fixing tests?
Or has somebody an idea to satisfy both?
Never mind, fixed.

codecov-io · 2016-09-28T19:39:51Z

Codecov Report

Merging #14184 into master will decrease coverage by -0.02%.
The diff coverage is 98.55%.

@@            Coverage Diff             @@
##           master   #14184      +/-   ##
==========================================
- Coverage   91.04%   91.03%   -0.02%     
==========================================
  Files         136      136              
  Lines       49088    49152      +64     
==========================================
+ Hits        44694    44744      +50     
- Misses       4394     4408      +14

Impacted Files	Coverage Δ
pandas/core/config_init.py	`95.27% <100%> (+0.23%)`	✅
pandas/core/frame.py	`97.83% <100%> (-0.1%)`	❌
pandas/formats/format.py	`95.4% <98.24%> (+0.09%)`	✅
pandas/io/gbq.py	`25% <0%> (-58.34%)`	❌
pandas/tools/merge.py	`91.78% <0%> (-0.35%)`	❌
pandas/util/testing.py	`81.87% <0%> (-0.19%)`	❌

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 211ecd5...ce0a0c9. Read the comment docs.

jorisvandenbossche

Some other comments:

Can we use multicolumn instead of multicol as keyword name? I know it is a bit shorter, but since latex also uses multicolumn I would use that
Now you center the multi-columns. I think left aligning as the other column may be better as a default.
Is there a way to specify this?
You added a bunch of code lines in the write_result function. Can you see whether it makes sense to refactor part of it into a separate function to have smaller more specific functions?

jorisvandenbossche · 2016-10-27T14:33:16Z

pandas/core/frame.py


-            .. versionadded:: 0.18.0


Can you leave this one (I think it belonged to the decimal option)

jorisvandenbossche · 2016-10-27T14:33:24Z

pandas/core/frame.py


-            .. versionadded:: 0.18.0
+            .. versionadded:: 0.19.0


0.19 ->0.20

jorisvandenbossche · 2016-10-27T14:35:47Z

pandas/formats/format.py

+                ncol = 1
+                coltext = ''
+
+                def appendCol():


it's an internal function, so not that important, but let's be consistent in PEP 8 naming, so appendCol -> append_col

sgsaenger · 2016-10-27T17:01:40Z

Thanks, i've incorporated your renaming suggestions.
Also added an additional multicolumn_format parameter where one can specify the alignment similar to the existing column_format.
Unfortunately, I don't really see how one could separate the different parts of this routine without rewriting it from scratch, even before the routine was basically "take the result of _to_str_columns and completely change it".

jreback · 2016-12-05T23:47:47Z

lgtm. @jorisvandenbossche @TomAugspurger
(conflict in the whatsnew), but can fix on merge if all other good.

jorisvandenbossche

Some more comments.

As I also said in my previous comment, you added a bunch of code lines in the write_result function. Can you see whether it makes sense to put some of that in separate helper functions, or can you also add some code comments to make it a bit easier to follow what happens?

jorisvandenbossche · 2016-12-14T11:53:39Z

pandas/core/frame.py

+            Use \multirow to enhance MultiIndex rows. Requires adding a
+            \usepackage{multirow} to your LaTeX preamble.
+
+            .. versionadded:: 0.20.0


Can you add this for the other keywords as well

the ..versionadded:: 0.20.0 part?

yes (so like you did!)

jorisvandenbossche · 2016-12-14T11:54:15Z

pandas/core/frame.py

@@ -1644,12 +1645,31 @@ def to_latex(self, buf=None, columns=None, col_space=None, header=True,

            .. versionadded:: 0.18.0

+        multicolumn : boolean, default will be read from the config module


Can you put "default will be read from the config module" in the explanation, and here just "default True"

jorisvandenbossche · 2016-12-14T13:05:32Z

pandas/formats/format.py

@@ -873,6 +886,9 @@ def get_col_type(dtype):
                            compat.string_types):  # pragma: no cover
            raise AssertionError('column_format must be str or unicode, not %s'
                                 % type(column_format))
+        multicolumn_format = self.multicolumn_format
+        if multicolumn_format is None:
+            multicolumn_format = get_option("display.latex.multicolumn_format")


Isn't this already done in DataFrame.to_latex?

It is.
I was just unsure whether this function can ever be called from another context, in which case it could fail with the default value for multicolumn_format.
Both the encoding (in DataFrameFormatter.to_latex) and column_format (write_result) argument are checked for being None.
Is there a better way to do this?
If this is kept, i should probably also add a typecheck for multicolumn_format?

I think it is enough that it is only checked in DataFrame.to_latex. A user should in principle never call the Formatter.to_latex directly

(it seems that we are a bit inconsistent here, as eg longtable and escate are only checked in DataFrame.to_latex)

will remove the check :)

jorisvandenbossche · 2016-12-14T13:06:57Z

pandas/formats/format.py

-        nlevels = self.frame.columns.nlevels
+        ilevels = self.frame.index.nlevels
+        clevels = self.frame.columns.nlevels
+        nlevels = clevels


Is there a reason to have both nlevels and clevels

No
Kinda.
I used clevels to keep from fiddling with the handling of the index-names-line.
On the compiled side there's no difference if i just use the nlevels, but the resulting code differs, and i'd rather not change the behaviour in parts that don't need to be modified.

jorisvandenbossche · 2016-12-14T13:19:56Z

pandas/tests/formats/test_format.py

+        self.assertEqual(result, expected)
+
+        df.index = df.T.index
+        result = df.T.to_latex(multirow=True,multicolumn=True,multicolumn_format='c')


PEP8: spaces after comma. But it is strange that Travis does not fail because of this

Will fix it in the next iteration. Missed it because for the test, all pep8 checks are disabled at the file-level (line 4).

jorisvandenbossche · 2016-12-14T13:22:54Z

pandas/tests/formats/test_format.py

+\cline{1-7}
+\multirow{2}{*}{c2} & 0 &  0 &  1 &  2 &  3 &  4 \\
+   & 1 &  5 &  6 &  7 &  8 &  9 \\
+\cline{1-7}


Do we want the cline?
Can you maybe show an example of how this looks like?

I definitely felt it necessary:

The multirows are centered, especially on deeper hierarchies it's rather confusing without them:

It's not as bad without multirows there the items are aligned on the top of the data-block.
I guess in this regard the multirow switch is more of a taste-decision akin to multicolum_format, less so than the multicolumn switch. Thus i had it default to False :)

Another question for the default: would we rather want to have it top-aligned instead of center? (here it may look better centered, but once you don't have exactly the same number of rows in each subpart, then I think top-aligned will look better)

Yes, that's why I defaulted it to False.
The multirow feature is specifically for centered labels, top-alignment can be made without it (as it is done already in the code).
Rather, should the cline be untied from the multirow?

Yes, that's why I defaulted it to False.

Ah, OK, I understand now.
Can you clarify this in the docstring explanation of multirow? (the fact that it gives centered labels instead of top-aligned, which is done by default)

jorisvandenbossche · 2017-01-02T10:02:16Z

doc/source/whatsnew/v0.20.0.txt

@@ -108,6 +108,9 @@ Other enhancements

 - ``.select_dtypes()`` now allows `datetimetz` to generically select datetimes with tz (:issue:`14910`)

+- The ``.to_latex()`` method will now accept ``multicolumn`` and ``multirow`` arguments to use the accompanying LaTeX enhancements
+
+- ``pd.cut`` and ``pd.qcut`` now support datetime64 and timedelta64 dtypes (issue:`14714`)


I suppose this is an error from rebasing

yes...should have really stuck to the instructions...

jorisvandenbossche · 2017-01-02T10:05:03Z

pandas/formats/format.py

@@ -873,6 +886,9 @@ def get_col_type(dtype):
                            compat.string_types):  # pragma: no cover
            raise AssertionError('column_format must be str or unicode, not %s'
                                 % type(column_format))
+        multicolumn_format = self.multicolumn_format
+        if multicolumn_format is None:
+            multicolumn_format = get_option("display.latex.multicolumn_format")


I think it is enough that it is only checked in DataFrame.to_latex. A user should in principle never call the Formatter.to_latex directly

(it seems that we are a bit inconsistent here, as eg longtable and escate are only checked in DataFrame.to_latex)

jorisvandenbossche · 2017-01-02T10:06:43Z

pandas/tests/formats/test_format.py

+\cline{1-7}
+\multirow{2}{*}{c2} & 0 &  0 &  1 &  2 &  3 &  4 \\
+   & 1 &  5 &  6 &  7 &  8 &  9 \\
+\cline{1-7}


Another question for the default: would we rather want to have it top-aligned instead of center? (here it may look better centered, but once you don't have exactly the same number of rows in each subpart, then I think top-aligned will look better)

jreback · 2017-01-21T23:17:09Z

status of this?

sgsaenger · 2017-02-28T10:17:40Z

Is there anything I should still improve?

jorisvandenbossche

Sorry it look so long to review again! Added some comments regarding docs, will take a final look at the code this evening.

Additional comment:

Can you add the new options to the overview table of available options in the docs (options.rst, it is a bit duplication of content, but that is how it is done right now)

jorisvandenbossche · 2017-02-28T13:15:45Z

pandas/core/config_init.py

+: bool
+    This specifies if the to_latex method of a Dataframe uses multicolumns
+    to pretty-print MultiIndex columns.
+    method. Valid values: False,True


I think this 'method. ' here is a left-over from copying it from somewhere else?

Ah, yes it is. Copied it from longtable or escape which are both still ill-formatted in the current master. I'll fix them too while i'm at it.

jorisvandenbossche · 2017-02-28T13:15:54Z

pandas/core/config_init.py

+: bool
+    This specifies if the to_latex method of a Dataframe uses multirows
+    to pretty-print MultiIndex rows.
+    method. Valid values: False,True


jorisvandenbossche · 2017-02-28T13:17:07Z

pandas/core/frame.py

+
+        multicolumn_format : str, default 'l'
+            default will be read from the config module
+            The alignment for multicolumns, similar to column_format


single backticks around column_format (`column_format`)

jorisvandenbossche · 2017-02-28T13:19:11Z

pandas/core/frame.py

@@ -1606,12 +1607,37 @@ def to_latex(self, buf=None, columns=None, col_space=None, header=True,

            .. versionadded:: 0.18.0

+        multicolumn : boolean, default True
+            default will be read from the config module
+            Use \multicolumn to enhance MultiIndex columns.


Can you use a proper sentence (Capital, punctuation) for the first line of 'default ..' (the thing is, here this looks OK in plain text, but in the html docs, the line breaks are not preserved, so will look line one long sentence without '.' in the middle)

Also, would start with the actual explanation of the keyword, and put "Default will be ..." at the end.

and the same for the other ones

This is again also true for longtable and escape, so i'll do the same to them.

jorisvandenbossche · 2017-02-28T13:21:23Z

pandas/tests/formats/test_format.py

+\cline{1-7}
+\multirow{2}{*}{c2} & 0 &  0 &  1 &  2 &  3 &  4 \\
+   & 1 &  5 &  6 &  7 &  8 &  9 \\
+\cline{1-7}


Yes, that's why I defaulted it to False.

Ah, OK, I understand now.
Can you clarify this in the docstring explanation of multirow? (the fact that it gives centered labels instead of top-aligned, which is done by default)

jorisvandenbossche

Thanks for the doc updates!

I looked in some more detail at the code, and added a few comments on how I think it can be made a bit more maintainable (it is now, as I mentioned earlier, a big chunk of code added in the for loop).
(to be clear, it is only about some restructuring, not fundamental concerns!)

Travis is doing a bit strange at the moment, but based on the other ones, it seems one of your tests is failing.

jorisvandenbossche · 2017-02-28T22:30:42Z

pandas/tests/formats/test_format.py

+                    ('c2',1):dict((x,x+5) for x in range(5)),
+                    ('c3',0):dict((x,x) for x in range(5))
+                    })
+        result = df.to_latex(multicolumn=True)


Can you add a test for multicolumn=False as well?

Further, can you add test cases with and without index/column level names? (I think now all are without?) Or is this already tested somewhere else?

You're right, seems like all tests have unnamed indices. Will do.

jorisvandenbossche · 2017-02-28T22:45:47Z

pandas/formats/format.py

@@ -906,8 +927,59 @@ def get_col_type(dtype):
                         if x else '{}') for x in row]
            else:
                crow = [x if x else '{}' for x in row]
+            if i < clevels and self.fmt.header and self.multicolumn:
+                # sum up columns to multicolumns


Can you put the content contained in this if statement in a separate method on the class? Eg call it something like _format_multicolumn(..) ? Then you can give it a docstring with some explanation what exactly happens. And that will make the big for loop here a bit more comprehensible.

Are class-methods preferred over nested-functions?

jorisvandenbossche · 2017-02-28T22:46:57Z

pandas/formats/format.py

+                if coltext:  # write last column name
+                    append_col()
+                crow = row2
+            if i >= nlevels and self.fmt.index and self.multirow:


Should this only happen when ilevels > 1 ?

I guess, yeah. I'll add this check.
Just for curiosity, is it possible to create some clashes in a single index, e.g. via hiding parts of a MultiIndex? This thought never crossed my mind when i wrote this.

is it possible to create some clashes in a single index, e.g. via hiding parts of a MultiIndex?

What do you mean exactly?

jorisvandenbossche · 2017-02-28T22:52:19Z

pandas/formats/format.py

+                crow = row2
+            if i >= nlevels and self.fmt.index and self.multirow:
+                # perform look-ahead to determine which rows can be joined
+                for j in range(ilevels):


Can you do the same here? (Put this in a method)
This one is shorter, so less needed, but then it is a bit more consistent

jorisvandenbossche · 2017-02-28T23:02:08Z

pandas/formats/format.py

+                for cl in clinebuf:
+                    cl[0] -= 1
+                    if cl[0] == 0:
+                        buf.write('\cline{{{0:d}-{1:d}}}\n'.format(cl[1],


I find this part very hard to understand? (not the buf.write itself, but the lines above regarding clinebuf, and also the line below this). Can you see if you can simplify it?
Or maybe try to explain it in a few words to me, then maybe I see another way or what would be a good comment so I would understand it more easily.

In the above block, i iterate over the index levels and check if a multirow should be used. If i find a multirow, i save the number of rows it spans, and it's level (indented by 1) in clinebuf (list, because multiple multirows can occur at the same time). When writing this line and the following, i decrease the first entry in the buffered pairs. If this value reaches zero, it marks the point where the multirow ends, and i place the cline. When all levels have been checked, i flush the buffer.

I guess it would be more clear to just save the target-row (as i've done with the target-col :D ) and drop the list-entry when cl[0] matches the loopcounter.

sgsaenger · 2017-03-02T16:26:17Z

Thanks for the feedback, i hope the changes go in the direction you intended.

The test that failed previously was caused by the to_latex_multiindex test expecting no multicolumns, obviously. This has been fixed.
I don't know what causes the new failures, looks more like a technical issue to me, locally it just fails by trying to load pyqt4 and 5 simultaneously, everything i've touched works as inteded ;)

jorisvandenbossche · 2017-03-02T18:25:44Z

I don't know what causes the new failures, looks more like a technical issue to me,

Yes, there are currently some issues with travis, nothing to do with this PR.

jorisvandenbossche · 2017-03-02T18:30:06Z

pandas/formats/format.py

+                # sum up columns to multicolumns
+                crow = self._format_multicolumn(crow, ilevels)
+            if i >= nlevels and self.fmt.index and self.multirow and\
+                    ilevels > 1:


small style issue: in pandas we try to not use \, but rather use (...) for line continuation

jorisvandenbossche · 2017-03-02T18:34:48Z

pandas/formats/format.py

+                if coltext:  # write last column name
+                    append_col()
+                crow = row2
+            if i >= nlevels and self.fmt.index and self.multirow:


is it possible to create some clashes in a single index, e.g. via hiding parts of a MultiIndex?

What do you mean exactly?

jorisvandenbossche

Thanks! I think that makes it much clearer (certainly with the explanation in the docstrings)

Some small follow-up comments

jorisvandenbossche · 2017-03-02T20:13:15Z

pandas/formats/format.py

+        for cl in self.clinebuf:
+            if cl[0] == i:
+                buf.write('\cline{{{0:d}-{1:d}}}\n'.format(cl[1], l))
+        self.clinebuf = [x for x in self.clinebuf if x[0] != i]


I find this already clearer, only the rewriting of clinebuf on the line above I don't fully understand (is that just to not have to iterate over all clinebuf elements that are not needed anymore because of previous lines? so not essential to have the code working? No problem with having it, just to be sure I understand?)
Maybe add a small comment above it

Yes, as soon as they're written they get deleted. Not necessary with the new mechanism I guess, I didn't like polluting the list with past entries.

jorisvandenbossche · 2017-03-02T20:13:47Z

pandas/formats/format.py

+                    self.clinebuf.append([i + nrow - 1, j + 1])
+        return row
+
+    def _print_cline(self, buf, i, l):


l -> ncols ? (as more self-explanatory name, if ncols is correct of course)

- [X] closes pandas-dev#13508 - [X] tests added / passed - [X] passes `git diff upstream/master | flake8 --diff` - [X] whatsnew entry Print names of MultiIndex columns. Added "multicolumn" and "multirow" flags to to_latex which trigger the corresponding feature. "multicolumn_format" is used to select alignment. Multirow adds clines to visually separate sections.

jorisvandenbossche · 2017-03-03T09:17:17Z

@hein09 Thanks a lot!
(the travis failure was a pep8 issue, fixed that before merging)

closes pandas-dev#13508 Print names of MultiIndex columns. Added "multicolumn" and "multirow" flags to to_latex which trigger the corresponding feature. "multicolumn_format" is used to select alignment. Multirow adds clines to visually separate sections.

TomAugspurger reviewed Sep 8, 2016
View reviewed changes

jreback added Enhancement IO LaTeX to_latex labels Sep 9, 2016

sgsaenger force-pushed the master branch from 9960993 to 23ab680 Compare September 28, 2016 16:04

sgsaenger commented Sep 28, 2016

View reviewed changes

sgsaenger force-pushed the master branch from 9366e5c to 0656829 Compare September 29, 2016 13:22

sgsaenger force-pushed the master branch 2 times, most recently from 9c81088 to 69f05c4 Compare October 12, 2016 13:51

jorisvandenbossche reviewed Oct 27, 2016

View reviewed changes

sgsaenger force-pushed the master branch from 69f05c4 to 0d6a4b7 Compare October 27, 2016 16:17

sgsaenger force-pushed the master branch from 0d6a4b7 to 1ecd65a Compare December 5, 2016 14:25

jorisvandenbossche reviewed Dec 14, 2016

View reviewed changes

sgsaenger force-pushed the master branch 2 times, most recently from bef053f to 840321a Compare December 21, 2016 00:40

jorisvandenbossche reviewed Jan 2, 2017

View reviewed changes

sgsaenger force-pushed the master branch from 840321a to 6bb865a Compare January 24, 2017 09:47

jorisvandenbossche reviewed Feb 28, 2017

View reviewed changes

sgsaenger force-pushed the master branch from 6bb865a to e23f0f8 Compare February 28, 2017 14:02

jorisvandenbossche reviewed Feb 28, 2017

View reviewed changes

sgsaenger force-pushed the master branch from e23f0f8 to 65d7778 Compare March 1, 2017 15:04

sgsaenger force-pushed the master branch from 65d7778 to ced00bc Compare March 2, 2017 13:16

jorisvandenbossche reviewed Mar 2, 2017

View reviewed changes

sgsaenger force-pushed the master branch 2 times, most recently from 21063df to d2ae82d Compare March 3, 2017 07:54

sgsaenger and others added 2 commits March 3, 2017 08:55

pep8

ce0a0c9

jorisvandenbossche merged commit 24a2155 into pandas-dev:master Mar 3, 2017

jorisvandenbossche added this to the 0.20.0 milestone Mar 3, 2017

		@@ -1644,12 +1645,31 @@ def to_latex(self, buf=None, columns=None, col_space=None, header=True,

		.. versionadded:: 0.18.0

		multicolumn : boolean, default will be read from the config module

Uh oh!

ENH: Added multicolumn/multirow support for latex #14184

ENH: Added multicolumn/multirow support for latex #14184

Uh oh!

Conversation

sgsaenger commented Sep 8, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

TomAugspurger commented Sep 28, 2016

Uh oh!

sgsaenger commented Sep 28, 2016

Uh oh!

TomAugspurger commented Sep 28, 2016

Uh oh!

sgsaenger Sep 28, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

codecov-io commented Sep 28, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

jorisvandenbossche left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sgsaenger commented Oct 27, 2016

Uh oh!

jreback commented Dec 5, 2016

Uh oh!

jorisvandenbossche left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sgsaenger Dec 20, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sgsaenger commented Sep 8, 2016 •

edited

Loading

sgsaenger Sep 28, 2016 •

edited

Loading

codecov-io commented Sep 28, 2016 •

edited

Loading

sgsaenger Dec 20, 2016 •

edited

Loading