ENH: Added more options for formats.style.bar #14757

jmarrec · 2016-11-28T16:02:35Z

4 tests added and passed
passes git diff upstream/master | flake8 --diff
Update documentation

You can now have the df.style.bar be centered on zero or midpoint value (in
addition to the already existing way of having the min value at the left side of the cell)

Documentation:

I have produced the following example that I think does a pretty good job of explaining the new capabilities

import pandas as pd
from IPython.display import HTML

# Case 1: s.min() < 0:
test1 = pd.Series([-100,-60,-30,-20], name='All Negative')
# Case 2: s.max() < 0:
test2 = pd.Series([10,20,50,100], name='All Positive')
# Case 3:
test3 = pd.Series([-10,-5,0,90], name='Both Pos and Neg')

head = """
<table>
    <thead>
        <th>Align</th>
        <th>All Negative</th>
        <th>All Positive</th>
        <th>Both Neg and Pos</th>
    </thead>
    </tbody>

"""

aligns = ['left','zero','mid']
for align in aligns:
    row = "<tr><th>{}</th>".format(align)
    for serie in [test1,test2,test3]:
        s = serie.copy()
        s.name=''
        row += "<td>{}</td>".format(s.to_frame().style.bar(align=align, 
                                                           color=['#d65f5f', '#5fba7d'], 
                                                           width=100).render()) #testn['width']
    row += '</tr>'
    head += row
    
head+= """
</tbody>
</table>"""
        

HTML(head)

What do you think?

codecov-io · 2016-11-28T16:47:43Z

Codecov Report

Merging #14757 into master will decrease coverage by 0.02%.
The diff coverage is 92.5%.

@@            Coverage Diff             @@
##           master   #14757      +/-   ##
==========================================
- Coverage   90.88%   90.86%   -0.03%     
==========================================
  Files         162      162              
  Lines       50821    50862      +41     
==========================================
+ Hits        46190    46215      +25     
- Misses       4631     4647      +16

Flag	Coverage Δ
#multiple	`88.64% <92.5%> (-0.03%)`	⬇️
#single	`40.3% <10%> (-0.03%)`	⬇️

Impacted Files	Coverage Δ
pandas/io/formats/style.py	`95.89% <92.5%> (-0.48%)`	⬇️
pandas/plotting/_converter.py	`63.54% <0%> (-1.82%)`	⬇️
pandas/core/common.py	`90.68% <0%> (-0.35%)`	⬇️
pandas/core/generic.py	`91.63% <0%> (ø)`	⬆️
pandas/core/config_init.py	`94.52% <0%> (+0.23%)`	⬆️

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 6d0118e...c7d41fb. Read the comment docs.

jmarrec · 2016-11-28T16:49:15Z

pandas/formats/style.py

+                   " of length 2 [ `color_negative`, `color_positive`] "
+                   "(eg: color=['#d65f5f', '#5fba7d'])\n"
+                   "Only the first two list elements will be used here.")
+            warnings.warn(msg, UserWarning)


Should I do raise TypeError(msg) instead of issuing a warning?

(removing the line saying that only two elements will be used of course...)

I prefer ValueError

sinhrks · 2016-11-29T00:04:02Z

pandas/formats/style.py

+                   " of length 2 [ `color_negative`, `color_positive`] "
+                   "(eg: color=['#d65f5f', '#5fba7d'])\n"
+                   "Only the first two list elements will be used here.")
+            warnings.warn(msg, UserWarning)


I prefer ValueError

sinhrks · 2016-11-29T00:05:47Z

pandas/formats/style.py

+
+        normed = s * 50 * width / (100 * m)
+
+        base = 'width: 10em; height: 80%;'


pls pass base from .bar because these are identical.

jreback · 2016-12-26T21:50:52Z

can you rebase / update

jorisvandenbossche · 2017-01-02T11:35:16Z

cc @TomAugspurger

jmarrec · 2017-01-02T19:08:09Z

Sorry for the delay, I'm trying to rebase right now (on 9947a99), but I'm getting a few errors

When doing just 'import pandas as pd':

AttributeError: module 'pandas' has no attribute 'compat'

When running the 'formats' nosetests:

AttributeError: module 'pandas.lib' has no attribute 'is_decimal'

Is it only me or is the master broken somehow?

jreback · 2017-01-02T19:20:43Z

you need to rebuild the extensions

python setup.py build_ext --inplace

generally you should do this anytime you pull from master (and there is an update)

jmarrec · 2017-01-02T19:21:52Z

That makes sense, thanks! After building everything works like it used to.

jreback · 2017-02-27T16:02:03Z

can you rebase / update

jmarrec · 2017-02-28T12:48:32Z

Done. When do you think this will be merged in?

jmarrec · 2017-02-28T14:13:54Z

The CircleCI has tests that fails, but they run fine on my local branch.

For example:

pandas$ nosetests pandas/tests/formats/test_style.py:TestStyler.test_bar_align_mid_all_pos
.
----------------------------------------------------------------------
Ran 1 test in 0.044s

OK

jreback · 2017-02-28T14:22:09Z

we switched to pytest recently FYI

take a look here: https://travis-ci.org/pandas-dev/pandas/jobs/206164419

jreback

pls add a whatsnew note (and you have some failing tests)

jreback · 2017-02-28T14:22:58Z

pandas/formats/style.py

+    def _bar_left(s, color, width, base):
+        """
+        The minimum value is aligned at the left of the cell
+        .. versionadded:: 0.17.1


you don't need the vesionadded tags (and they are outdated)

jreback · 2017-02-28T14:23:53Z

pandas/tests/formats/test_style.py

@@ -348,6 +348,109 @@ def test_bar_0points(self):
                             ', transparent 0%)']}
        self.assertEqual(result, expected)

+    def test_bar_align_zero_pos_and_neg(self):
+        df = pd.DataFrame({'A': [-10, 0, 20, 90]})


can you add this issue number as a comment

You mean this pull request number? #14757?

jreback · 2017-02-28T14:24:33Z

@TomAugspurger can you review

jreback · 2017-02-28T14:25:28Z

@jmarrec I would also take your above picture and add it to the style docs themselves (choose a nice place for it).

jmarrec · 2017-02-28T14:51:40Z

My failing tests are in python 2.7 only, because I expect 80.0% but I get 80% instead.

Eg:

Result in Python 2.7:

 (3, 0): ['width: 10em',
  ' height: 80%',
  'background: linear-gradient(90deg, transparent 0%, transparent 80%, #d65f5f 80%, #d65f5f 100%, transparent 100%)']}

expected:


 (3, 0): ['width: 10em',
  ' height: 80%',
  'background: linear-gradient(90deg, transparent 0%, transparent 80.0%, #d65f5f 80.0%, #d65f5f 100%, transparent 100%)']}

jreback · 2017-02-28T14:57:34Z

ok, so can do 1 of 2 things.

conform both results in py2 and py3 (e.g. you are formatting the string)
or
use a subtition for that value in the test (with different values for py2 and py3).

ideally first way is better

jreback · 2017-02-28T16:27:24Z

doc/source/html-styling.ipynb

   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
-    "Use `Styler.set_properties` when the style doesn't actually depend on the values."
+    "New in version 0.19.2 is the ability to customize further the bar chart: You can now have the `df.style.bar` be centered on zero or midpoint value (in addition to the already existing way of having the min value at the left side of the cell), and you can pass a list of `[color_negative, color_positive]`.\n",


jreback · 2017-02-28T16:28:26Z

pandas/formats/style.py

        width: float
            A number between 0 or 100. The largest value will cover ``width``
            percent of the cell's width
+        align : str, default 'left'
+            .. versionadded:: 0.19.2


-> 0.20.0
also need a blank line before/after the versionadded tag

jreback · 2017-02-28T16:29:14Z

cc @jorisvandenbossche

jorisvandenbossche

Nice enhancement! Added some comments

jorisvandenbossche · 2017-03-03T10:39:40Z

doc/source/html-styling.ipynb

   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
-    "Use `Styler.set_properties` when the style doesn't actually depend on the values."
+    "New in version 0.20.0 is the ability to customize further the bar chart: You can now have the `df.style.bar` be centered on zero or midpoint value (in addition to the already existing way of having the min value at the left side of the cell), and you can pass a list of `[color_negative, color_positive]`.\n",


The table below is a nice overview. But it is also a bit hard to interpret as a user what is the actual code he/she should use. Can you add first an example with the existing frame? Eg df.style.bar(subset=['B'], align='mid'). And then give the overview table.

jorisvandenbossche · 2017-03-03T10:41:27Z

pandas/formats/style.py

        .. versionadded:: 0.17.1

        Parameters
        ----------
        subset: IndexSlice, default None
            a valid slice for ``data`` to limit the style application to
        axis: int
-        color: str
+        color: str (for align='left') or 2-tuple/list (for align='zero', 'mid')


In the _bar_left function also a tuple is accepted, which contradicts with this

Thanks for catching that.

jorisvandenbossche · 2017-03-03T10:46:27Z

pandas/formats/style.py

+            - 'zero' : a value of zero is located at the center of the cell
+            - 'mid' : the center of the cell is at (max-min)/2, or
+                if values are all negative (positive) the zero is aligned
+                at the right (left) of the cell


I find the names a bit confusing. As 'zero' is actually more 'centered'? While I first thought 'mid' would give me centered bars, but that depends on the actual values you have.

BTW, 'mid' actually seems a more sensible default. Maybe we should change that? (cc @TomAugspurger)

I find the names a bit confusing. As 'zero' is actually more 'centered'?

I think this is only true if you have all positive values. So

pd.DataFrame(np.arange(-2, 5)).style.bar(align='zero')

Makes sense to me since zero is at the middle.

BTW, 'mid' actually seems a more sensible default.

Probably, but I'm not sure it's worth breaking API. Maybe with a deprecation cycle?

@jorisvandenbossche: @TomAugspurger is correct, it's really centered on zero. The example I tried to include in the doc and that I pasted in the first post of this PR should highlight that fact, otherwise I should revise it.

I have no idea how you handle deprecation cycles etc, so I'll leave that completely up to you guys.

jorisvandenbossche · 2017-03-03T10:47:59Z

pandas/formats/style.py

+                       color=color, width=width, base=base)
+        elif align == 'mid':
+            self.apply(self._bar_center_mid, subset=subset, axis=axis,
+                       color=color, width=width, base=base)


Can you do here an "else" clause to raise a ValueError when a not accepted value for align was passed? and also add a test for it.
(with playing a bit, I already was bitten by this by trying to pass 'right' and 'center' to align, and thinking that it was not working :-))

jmarrec · 2017-03-09T13:52:59Z

@jorisvandenbossche I haven't forgotten about your comments, it just fell on my back burner. I'll try to get to it asap in the coming days.

TomAugspurger

This is really close. I think just adding the test for raising with invalid aligns. Do you have time to update @jmarrec ?

TomAugspurger · 2017-02-28T13:14:40Z

pandas/formats/style.py

        width: float
            A number between 0 or 100. The largest value will cover ``width``
            percent of the cell's width
+        align : str, default 'left'
+            .. versionadded:: 0.19.2


Could you change these to 0.20.0

TomAugspurger · 2017-02-28T13:14:57Z

pandas/formats/style.py

+    def _bar_left(s, color, width, base):
+        """
+        The minimum value is aligned at the left of the cell
+        .. versionadded:: 0.17.1


.. versionadded:: 0.20.0

Are you sure you want me to do that? This method was added in 0.17.1, I just modified it to accept a tuple of color.

@jreback told me in an earlier review (Feb 28) that I didn't need the versionadded tags for these, so I deleted them all. Which?

TomAugspurger · 2017-02-28T13:15:03Z

pandas/formats/style.py

+    def _bar_center_zero(s, color, width, base):
+        """
+        Creates a bar chart where the zero is centered in the cell
+        .. versionadded:: 0.19.2


.. versionadded:: 0.20.0

TomAugspurger · 2017-02-28T13:15:11Z

pandas/formats/style.py

+    def _bar_center_mid(s, color, width, base):
+        """
+        Creates a bar chart where the midpoint is centered in the cell
+        .. versionadded:: 0.19.2


.. versionadded:: 0.20.0

TomAugspurger · 2017-02-28T13:16:45Z

pandas/formats/style.py

        width: float
            A number between 0 or 100. The largest value will cover ``width``
            percent of the cell's width
+        align : str, default 'left'


I think the proper format on this line is

align : {'left', 'zero',' mid'}

But the explanations below are great. Keep those too.

TomAugspurger · 2017-04-03T21:26:32Z

pandas/formats/style.py

+            - 'zero' : a value of zero is located at the center of the cell
+            - 'mid' : the center of the cell is at (max-min)/2, or
+                if values are all negative (positive) the zero is aligned
+                at the right (left) of the cell


I find the names a bit confusing. As 'zero' is actually more 'centered'?

I think this is only true if you have all positive values. So

pd.DataFrame(np.arange(-2, 5)).style.bar(align='zero')

Makes sense to me since zero is at the middle.

BTW, 'mid' actually seems a more sensible default.

Probably, but I'm not sure it's worth breaking API. Maybe with a deprecation cycle?

jmarrec · 2017-04-04T22:52:46Z

@TomAugspurger Really sorry, it had fallen on the back burner. I'll get on it this week, will try for tomorrow, so I can put this to bed.

jmarrec · 2017-04-05T10:26:50Z

Yikes, this looks really messy sorry. For some reason when I rebased on upstream/master today, my branch diverged. I merged origin/style-bar back into my local branch, and pushed... I didn't know it'd clutter this PR that much, and now that I pushed I don't want to rewrite history.

Anyways, I think I made all the changes you three requested, except the versionadded tags on each of the _bar_xxx methods since @jreback had told me to delete them and it seems @TomAugspurger reviewed an older version where they were still there but outdated (I have deleted them a while ago). If you want me to add them I guess I can. Otherwise I hope this is ready to go!

jreback · 2017-04-05T11:05:41Z

@jmarrec looks like you pushed all commits, do this

git rebase -i thisbranch origin/master
git push yourorigin thisbranch -f

to fix

jmarrec · 2017-04-05T11:47:27Z

@jreback in your example 'origin' means this 'pandas-dev' repo right?

I end up with a detached HEAD, I'm afraid of force pushing. Is that expected?

jreback · 2017-04-05T11:52:29Z

i'll fix your branch give me about 15 min

@sinhrks

Author: Julien Marrec <[email protected]> Closes pandas-dev#14757 from jmarrec/style-bar and squashes the following commits: dc3cbe8 [Julien Marrec] Added a whatsnew note af6c9bd [Julien Marrec] Added a simple example before the parametric one 80a3ce0 [Julien Marrec] Check for bad align value and raise. Wrote test for it too 5875eb9 [Julien Marrec] Change docstrings for color and align 5a22ee1 [Julien Marrec] Merge commit '673fb8f828952a4907e5659c1fcf83b771db7280' into style-bar 0e74b4d [Julien Marrec] Fix versionadded 1b7ffa2 [Julien Marrec] Added documentation on the new df.style.bar options for align and Colors in the documentation. 46bee6d [Julien Marrec] Change the tests to match new float formats. 01c200c [Julien Marrec] Format flots to avoid issue with py2.7 / py3.5 compta. 7ac2443 [Julien Marrec] Changes according to @sinhrks: Raise ValueError instead of warnings when color isn’t a str or 2-tuple/list. Passing “base” from bar(). e3f714c [Julien Marrec] Added a check on color argument that will issue a warning. Not sure if need to raise TypeError or issue a UserWarning if a list with more than two elements is passed. f12faab [Julien Marrec] Fixed line too long `git diff upstream/master | flake8 --diff now passes` 7c89137 [Julien Marrec] ENH: Added more options for formats.style.bar 673fb8f [Julien Marrec] Fix versionadded 506f3d2 [Julien Marrec] Added documentation on the new df.style.bar options for align and Colors in the documentation. e0563d5 [Julien Marrec] Change the tests to match new float formats. d210938 [Julien Marrec] Format flots to avoid issue with py2.7 / py3.5 compta. b22f639 [Julien Marrec] Changes according to @sinhrks: Raise ValueError instead of warnings when color isn’t a str or 2-tuple/list. Passing “base” from bar(). 3046626 [Julien Marrec] Added a check on color argument that will issue a warning. Not sure if need to raise TypeError or issue a UserWarning if a list with more than two elements is passed. 524a9ab [Julien Marrec] Fixed line too long `git diff upstream/master | flake8 --diff now passes` d1eafbb [Julien Marrec] ENH: Added more options for formats.style.bar

jmarrec · 2017-05-01T05:28:15Z

@jreback will this make it in (or through #15900, now closed)?

jreback · 2017-05-01T09:33:45Z

see the comments - iirc needs some more tests

jmarrec · 2017-05-01T13:57:59Z

Well, missing I'm missing something obvious, but I don't see any comments left to address, that's precisely why I'm asking.

Like I wrote a month ago, the only "open" things is what @TomAugspurger asked, that is to change the version added tags, but he was reviewing an outdated version, because before that you already told me to delete these version added tags, which I did.

TomAugspurger · 2017-05-01T14:10:17Z

I think @jorisvandenbossche's comment https://github.com/pandas-dev/pandas/pull/14757/files/b22f6392b3416cdc02138a79efb8f179751c8141#r104129905 is still outstanding. Handling cases like .bar(align='foo'). That should raise with something like ValueError('align' must be one of {'left', 'zero', 'mid'}, got 'foo' instead)

Also can you change all of the self.assertEqual(left, right)s to assert left == right?

jmarrec · 2017-05-01T14:50:53Z

Was done on April 5, see https://github.com/jmarrec/pandas/blob/style-bar/pandas/formats/style.py#L1034.

I did the change you requested (in all the file not just my changes, used a regex...).

Now, is everything ok?

TomAugspurger · 2017-05-01T14:53:50Z

Whoops, looks like some git issues. Let me see if I can rebase for you, one sec.

jmarrec · 2017-05-01T15:17:19Z

If you find how to fix it (Jreback did earlier), if you could please share which commands did the trick, that'd be great thanks

@TomAugspurger

You can now have the bar be centered on zero or midpoint value (in addition to the already existing way of having the min value at the left side of the cell) Fixed line too long `git diff upstream/master | flake8 --diff now passes` Change the tests to match new float formats. Added documentation on the new df.style.bar options for align and Colors in the documentation. Fix versionadded ENH: Added more options for formats.style.bar You can now have the bar be centered on zero or midpoint value (in addition to the already existing way of having the min value at the left side of the cell) Fixed line too long `git diff upstream/master | flake8 --diff now passes` Change the tests to match new float formats. Added documentation on the new df.style.bar options for align and Colors in the documentation. Fix versionadded Check for bad align value and raise. Wrote test for it too Added a simple example before the parametric one Added a whatsnew note Replaced 'self.assertEqual(left, rigth)' by 'assert left == right' like @TomAugspurger asked rebased

TomAugspurger · 2017-05-01T15:27:10Z

Typically it's just git fetch upstream, then git rebase upstream/master. This was a bit tricker than usual because it's been sitting out for so long (sorry about that) and we've re-arranged a lot of things now. This causes conflicts, since two commits touched the same lines so you have to go in and fix them manually with your editor. Then it's git add <files> and git rebase --continue. until you're fully rebased.

ca9c0d5 should have everything down in a single commit, if you want to give that a skim to make sure I didn't miss anything.

jorisvandenbossche

Added a commit with a small doc correction (indentation and white space). And also aligned the signature with documentation, adding the new keyword at the end, not breaking people who used axis positional

jorisvandenbossche · 2017-05-01T21:12:05Z

Let's merge this PR, but, I find we should think about changing the default value. If we all think 'mid' is the more sensible default, IMO we should just change it.
It's a while that I look into the PR in detail, but if I recall correctly, the 'mid' argument would only change things in case of negative values? (which I think is a less occurring usecase, and currently also a bid odd result?) If that is correct, I would opt to just change it.

gfyoung · 2017-05-02T01:59:27Z

@TomAugspurger : For subsequent PR's, could you check to make sure that old nosetest idiom is not being used? I found a couple in the diff that you just merged in.

TomAugspurger · 2017-05-02T02:10:47Z

My apologies. I thought I caught all those in the rebase :/

I'll submit a fix now.

gfyoung · 2017-05-02T02:12:43Z

No worries. I'm putting up a PR to remove vestigial statements like these, so that won't be necessary.

I just wanted to alert you about it for subsequent PR's.

@TomAugspurger

* ENH: Added more options for formats.style.bar You can now have the bar be centered on zero or midpoint value (in addition to the already existing way of having the min value at the left side of the cell) Fixed line too long `git diff upstream/master | flake8 --diff now passes` Change the tests to match new float formats. Added documentation on the new df.style.bar options for align and Colors in the documentation. Fix versionadded ENH: Added more options for formats.style.bar You can now have the bar be centered on zero or midpoint value (in addition to the already existing way of having the min value at the left side of the cell) Fixed line too long `git diff upstream/master | flake8 --diff now passes` Change the tests to match new float formats. Added documentation on the new df.style.bar options for align and Colors in the documentation. Fix versionadded Check for bad align value and raise. Wrote test for it too Added a simple example before the parametric one Added a whatsnew note Replaced 'self.assertEqual(left, rigth)' by 'assert left == right' like @TomAugspurger asked rebased * small doc fixes

halflings · 2017-07-04T10:20:42Z

This is breaking the previous behavior of Pandas when using .plot.bar : passing a list of colors (with 2 elements or more) would give each bar a different color, as it is the case in matplotlib.
I also think passing an array with two values which are implicitly interpreted as [negative_color, positive_color] is confusing, and we should have instead two explicit parameters for this feature.

Example of bar coloring breaking in 0.20:

import pandas as pd
df = pd.DataFrame(dict(created=[0, 1, 2, 3], amount=[6, 1, 2, 5], is_active=[0, 1, 0, 1]))
df.set_index('created').plot.bar(y='amount', color=df.is_active.apply(lambda v : ['red', 'blue'][v]

Shows this in previous verisons:

))

Everything is blue in 0.20+

jmarrec commented Nov 28, 2016

View reviewed changes

sinhrks added the Output-Formatting __repr__ of pandas objects, to_string label Nov 28, 2016

sinhrks reviewed Nov 29, 2016

View reviewed changes

sinhrks approved these changes Nov 29, 2016

View reviewed changes

chrisaycock mentioned this pull request Dec 1, 2016

ENH: merge_asof() has type specializations and can take multiple 'by' parameters (#13936) #14783

Closed

4 tasks

TomAugspurger requested review from TomAugspurger and removed request for TomAugspurger December 27, 2016 13:37

jmarrec force-pushed the style-bar branch from bef05c6 to e18cc0d Compare January 2, 2017 19:19

jmarrec force-pushed the style-bar branch from e18cc0d to b22f639 Compare February 28, 2017 12:44

jreback requested changes Feb 28, 2017

View reviewed changes

jreback reviewed Feb 28, 2017

View reviewed changes

jreback approved these changes Feb 28, 2017

View reviewed changes

jorisvandenbossche reviewed Mar 3, 2017

View reviewed changes

TomAugspurger requested changes Apr 3, 2017

View reviewed changes

TomAugspurger mentioned this pull request Apr 3, 2017

DOC: Use nbsphinx #15581

Merged

jreback mentioned this pull request Apr 5, 2017

(rebased) ENH: Added more options for formats.style.bar #15900

Closed

4 tasks

TomAugspurger force-pushed the style-bar branch from 433da4c to ca9c0d5 Compare May 1, 2017 15:24

small doc fixes

c7d41fb

jorisvandenbossche approved these changes May 1, 2017

View reviewed changes

TomAugspurger approved these changes May 2, 2017

View reviewed changes

TomAugspurger merged commit 217864e into pandas-dev:master May 2, 2017

halflings mentioned this pull request Jul 4, 2017

plotting in pandas is not picking the list of colors passed, it only gets the first value #16822

Closed


		normed = s * 50 * width / (100 * m)

		base = 'width: 10em; height: 80%;'

ENH: Added more options for formats.style.bar #14757

ENH: Added more options for formats.style.bar #14757

Conversation

jmarrec commented Nov 28, 2016 • edited Loading

Documentation:

codecov-io commented Nov 28, 2016 • edited by codecov bot Loading

Codecov Report

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 Dec 26, 2016

jorisvandenbossche commented Jan 2, 2017

jmarrec commented Jan 2, 2017

jreback commented Jan 2, 2017

jmarrec commented Jan 2, 2017 • edited Loading

jreback commented Feb 27, 2017

jmarrec commented Feb 28, 2017

jmarrec commented Feb 28, 2017

jreback commented Feb 28, 2017

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

jreback commented Feb 28, 2017

jreback commented Feb 28, 2017

jmarrec commented Feb 28, 2017

jreback commented Feb 28, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Feb 28, 2017

jorisvandenbossche 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

jmarrec commented Mar 9, 2017 • edited Loading

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

jmarrec Apr 5, 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

jmarrec commented Apr 4, 2017

jmarrec commented Apr 5, 2017

jreback commented Apr 5, 2017

jmarrec commented Apr 5, 2017

jreback commented Apr 5, 2017

jmarrec commented May 1, 2017

jreback commented May 1, 2017

jmarrec commented May 1, 2017

TomAugspurger commented May 1, 2017

jmarrec commented May 1, 2017

TomAugspurger commented May 1, 2017

jmarrec commented May 1, 2017

TomAugspurger commented May 1, 2017

jorisvandenbossche left a comment

Choose a reason for hiding this comment

jorisvandenbossche commented May 1, 2017

gfyoung commented May 2, 2017

TomAugspurger commented May 2, 2017

gfyoung commented May 2, 2017

halflings commented Jul 4, 2017

jmarrec commented Nov 28, 2016 •

edited

Loading

codecov-io commented Nov 28, 2016 •

edited by codecov bot

Loading

jmarrec commented Jan 2, 2017 •

edited

Loading

jmarrec commented Mar 9, 2017 •

edited

Loading

jmarrec Apr 5, 2017 •

edited

Loading