Adds custom plot formatting for TimedeltaIndex. #15067

jgoppert · 2017-01-05T08:20:50Z

Instead of hacking the PeriodIndex plotting code to work for TimedeltaIndex, this just creates a simple formatter unique to TimedeltaIndex. It does fine for regular intervals up to a day, then the labels and run into eachothter. If this happens, you can use plt.autofmt_xdate() to fix it.

closes ENH/BUG: support TimedeltaIndex plotting #8711
tests passed, none added
passes git diff upstream/master | flake8 --diff

d.t_sensor_gyro_0__f_x.plot()
plt.grid()

index =  pd.timedelta_range('1 day 2 hr 30 min 10 s', periods=10, freq='1 d')
s = pd.Series(np.random.randn(len(index)), index)
s.plot()
plt.gcf().autofmt_xdate()
plt.grid()

index =  pd.timedelta_range('1 day 2 hr 30 min 10 s', periods=10, freq='1 ns')
s = pd.Series(np.random.randn(len(index)), index)
s.plot()
plt.gcf().autofmt_xdate()
plt.grid()

index =  pd.timedelta_range('0s', periods=10, freq='100000 ns')
s = pd.Series(np.random.randn(len(index)), index)
s.plot()
plt.gcf().autofmt_xdate()
plt.grid()
plt.xlabel('time')
plt.ylabel('y')

jgoppert · 2017-01-05T10:36:54Z

I think this is ready to merge, it is functional for me now.

codecov-io · 2017-01-05T10:46:59Z

Codecov Report

Merging #15067 into master will decrease coverage by -4.85%.
The diff coverage is 73.46%.

@@            Coverage Diff             @@
##           master   #15067      +/-   ##
==========================================
- Coverage   90.37%   85.52%   -4.85%     
==========================================
  Files         135      145      +10     
  Lines       49476    51325    +1849     
==========================================
- Hits        44713    43896     -817     
- Misses       4763     7429    +2666

Impacted Files	Coverage Δ
pandas/tseries/converter.py	`62.95% <100%> (+1.1%)`	✅
pandas/tools/plotting.py	`68.48% <100%> (-3.31%)`	❌
pandas/tseries/plotting.py	`61.78% <55.17%> (-2.96%)`	❌
pandas/io/sas/sas_constants.py	`0% <ø> (-100%)`	❌
pandas/io/sas/sas7bdat.py	`0% <ø> (-90.67%)`	❌
pandas/io/sas/sas_xport.py	`0% <ø> (-90.28%)`	❌
pandas/io/sas/sasreader.py	`7.4% <ø> (-77.78%)`	❌
pandas/computation/init.py	`76.92% <ø> (-13.99%)`	❌
pandas/util/decorators.py	`60.78% <ø> (-7.83%)`	❌
... and 93 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 f62e8f2...945ec14. Read the comment docs.

TomAugspurger

Thanks for taking this @jgoppert. A few inline comments.

Could you also

Add a note in doc/source/visualization.rst showing how to use plt.gcf().autofmt_xdate() after plotting
Add a note in doc/source/whatsnew/v0.20.0.txt

TomAugspurger · 2017-01-05T14:30:57Z

pandas/tseries/plotting.py

+        raise IOError('index type not supported')
+
+    # Note, DatetimeIndex does not use this
+    # interface. DatetimeIndex uses matplotlib.date directly

    # x and y coord info
    subplot.format_coord = lambda t, y: (


Should this code be running when index is a TimedeltaIndex? I think it's controlling the stuff in the bottom right of the attached screenshot.

I moved the format_coord section to the period index section, the draw_if_interactive I left common.

TomAugspurger · 2017-01-05T14:33:05Z

pandas/tseries/plotting.py

+        subplot.xaxis.set_major_formatter(majformatter)
+        subplot.xaxis.set_minor_formatter(minformatter)
+
+    elif isinstance(index, TimedeltaIndex):


Can you add tests for this in pandas/tests/plotting/test_datetimelike.py?

You should just need a couple of datasets

with n_decimals > 9

with n_decimals < 9

In the tests, if you could check that the formatting is correct, and maybe check the number of ticklabels.

TomAugspurger · 2017-01-05T14:33:51Z

pandas/tseries/plotting.py

@@ -278,8 +280,22 @@ def _maybe_convert_index(ax, data):
 # Patch methods for subplot. Only format_dateaxis is currently used.
 # Do we need the rest for convenience?

-
-def format_dateaxis(subplot, freq):
+def timeTicks(x, pos, n_decimals):


Can you also add tests for this.

Got it, needs more work though, see comment below.

jreback · 2017-01-05T15:00:34Z

pandas/tseries/plotting.py

@@ -278,8 +280,22 @@ def _maybe_convert_index(ax, data):
 # Patch methods for subplot. Only format_dateaxis is currently used.
 # Do we need the rest for convenience?

-
-def format_dateaxis(subplot, freq):
+def timeTicks(x, pos, n_decimals):


-> timeTicks -> format_time_ticks

switched it to format_timedelta_ticks to be very clear it isn't being used for period index.

jreback · 2017-01-05T15:02:07Z

pandas/tseries/plotting.py

+    d, h = divmod(h, 24)
+    decimals = int(ns * 10**(n_decimals - 9))
+    s = r'{:02d}:{:02d}:{:02d}'.format(int(h), int(m), int(s))
+    if n_decimals > 0:


you are pretty much reinveting the wheel here.

In [2]: pd.Timedelta(20,unit='s') Out[2]: Timedelta('0 days 00:00:20') In [3]: pd.Timedelta(20.5,unit='s') Out[3]: Timedelta('0 days 00:00:20.500000') In [4]: pd.Timedelta(120.5,unit='s') Out[4]: Timedelta('0 days 00:02:00.500000') # if you want to get the data out for specific formatting In [5]: pd.Timedelta(120.5,unit='s').components Out[5]: Components(days=0, hours=0, minutes=2, seconds=0, milliseconds=500, microseconds=0, nanoseconds=0)

you can simply call Timedelta(..)._repr_base(format='...') with a new format if you really want it

In [16]: pd.Timedelta('1 day')._repr_base() Out[16]: '1 days' In [17]: pd.Timedelta('1 day')._repr_base(format='long') Out[17]: '1 days 00:00:00' In [18]: pd.Timedelta('1 day')._repr_base(format='all') Out[18]: '1 days 00:00:00.000000000'

Unfortunately, python timedelta doesn't support nanoseconds.

In [4]: pd.Timedelta(1e-9, unit='s') Out[4]: Timedelta('0 days 00:00:00') In [5]: pd.Timedelta(1e-3, unit='s') Out[5]: Timedelta('0 days 00:00:00.001000')

of course it does. This is pandas Timedelta. (you are looking at the printing), which by-default suppres it

In [6]: pd.Timedelta(1, unit='ns')._repr_base(format='all') Out[6]: '0 days 00:00:00.000000001'

Ok great, got confused by the output. So how do I say I want n decimals, because that would be a drop-in for what I've already written.

Also, suppressing days when they are 0 would be nice.

Maybe the

Components(days=0, hours=0, minutes=2, seconds=0, milliseconds=500, microseconds=0, nanoseconds=0)

could compress the code a bit, but I'm not sure it would save significant space over what I've already got.

Still need to improve unit testing.

jgoppert · 2017-01-05T15:46:53Z

pandas/tests/plotting/test_datetimelike.py

+        for l in xaxis.get_ticklabels():
+            if len(l.get_text()) > 0:
+                self.assertEqual(l.get_rotation(), 30)
+


This test needs more work.

For this test, we basically want to check that for a narrow window (as shown below) we get nanosecond precision, and if at 0 to start without days.

rng = timedelta_range('0', periods=3, freq='ns') # rough code to follow expected_labels = [ '00:00:00.000000001', '00:00:00.000000002, '00:00:00.000000001'] for label, expected_label in zip(ax.get_xtick_labels(), expected_labels): # check that there are no days self.assertEqual(label, expected_label)

We also want to check for a wide window that we have second precision.

rng = timedelta_range('0', periods=10, freq='1 d')

jgoppert · 2017-01-05T15:47:23Z

pandas/tests/plotting/test_datetimelike.py

+        # test issue #8711
+        s = Series(range(5), timedelta_range('1day', periods=5))
+        _check_plot_works(s.plot)
+


I have no idea if check_plots_works will actually do anything useful here.

This basically just tests that the plot call does not error. So, for example, that means that it will test that it does not raise a MemoryError like it does now in some cases

jgoppert · 2017-01-05T15:48:07Z

doc/source/whatsnew/v0.20.0.txt

@@ -116,6 +116,7 @@ Other enhancements

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

+- ``pd.TimedeltaIndex`` now has a custom datetick formatter specifically designed for nanosecond level precision (:issue:`8711`)


Can we add a link to visualization.rst?

That is certainly possible, we only have not really a section about timedelta plotting currently.

jgoppert · 2017-01-05T15:51:20Z

@TomAugspurger @jreback do you guys think you can help me clean this up. I'm not familiar with your docs and unit tests and have to get back to what I wrote this for today.

I gave you push access to my branch since I can't push the branch to upstream.

jgoppert · 2017-01-05T17:23:28Z

I just pushed the full implementation of the unit test. Can we merge this now? It is definitely an improvement over master's mem allocation error and we can revisit the small details later.

jreback · 2017-01-05T18:18:53Z

@jgoppert things are merged when they pass all tests, have documentation, and unit tests to the satisfaction of the maintainers. This has not been reviewed as yet.

of course if you or really anyone wants to pull in your commits on top of master, they can certainly do this.

jgoppert · 2017-01-05T18:35:09Z

I understand, I'm just running out of development bandwidth and don't want it to stall. Let me know what the delta is and I'll try to do what I can.

jgoppert · 2017-01-05T18:50:48Z

@jreback Any idea why clang 3.5 failed to set the xticks label?

jreback · 2017-01-05T20:39:58Z

cc @tacaswell

jgoppert · 2017-01-06T01:20:35Z

So weird that osx with python 3.5 keeps failing and all of the others work. I'm running out of ideas.

jorisvandenbossche · 2017-01-06T11:35:39Z

doc/source/visualization.rst

+date tick adjustment from matplotlib for figures whose ticklabels overlap.
+
+See the :meth:`autofmt_xdate <matplotlib.figure.autofmt_xdate>` method and the
+`matplotlib documentation <http://matplotlib.org/api/figure_api.html#matplotlib.figure.Figure.autofmt_xdate>`__ for more.


I would refer here to http://matplotlib.org/users/recipes.html#fixing-common-date-annoyances instead (the link to the api docs will also be generated through the :meth:.. you did, and is also included in this recipe section.

jorisvandenbossche · 2017-01-06T11:50:09Z

doc/source/whatsnew/v0.20.0.txt

@@ -116,6 +116,7 @@ Other enhancements

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

+- ``pd.TimedeltaIndex`` now has a custom datetick formatter specifically designed for nanosecond level precision (:issue:`8711`)


That is certainly possible, we only have not really a section about timedelta plotting currently.

jorisvandenbossche · 2017-01-06T11:51:45Z

pandas/tests/plotting/test_datetimelike.py

+    def test_timedelta_plot(self):
+        # test issue #8711
+        s = Series(range(5), timedelta_range('1day', periods=5))
+        _check_plot_works(s.plot)


Can you add here the three examples you showed in the top post of the PR? (just with the same _check_plot call as you already did for 1 series). At least one of them was producing a memory error currently, so it would be good to include that here to ensure this does not happen again.

Fixed, just added the 2 easy ones above.

Still need to improve unit testing.

jgoppert · 2017-01-14T21:50:54Z

@jorisvandenbossche I think I addressed your review. Hopefully mac passes CI now.

jreback · 2017-01-14T21:56:26Z

pandas/tseries/plotting.py

+            lambda x, pos: format_timedelta_ticks(x, pos, n_decimals))
+        subplot.xaxis.set_major_formatter(formatter)
+    else:
+        raise IOError('index type not supported')


use TypeError

jreback · 2017-01-14T21:58:33Z

pandas/tseries/plotting.py

-def format_dateaxis(subplot, freq):
+def format_timedelta_ticks(x, pos, n_decimals):
+    ''' Convert seconds to 'D days HH:MM:SS.F' '''
+    s, ns = divmod(x, 1e9)


use triple double quotes not single quotes

jreback · 2017-01-14T22:01:34Z

pandas/tseries/plotting.py

+    elif isinstance(index, TimedeltaIndex):
+        from matplotlib import ticker
+        (vmin, vmax) = tuple(subplot.xaxis.get_view_interval())
+        n_decimals = int(np.ceil(np.log10(100 * 1e9 / (vmax - vmin))))


is there a reason not to make this interface similar to the above?

e.g. why not create a TimeSeries_TimedeltaFormatter ?

Addressed in latest commit.

jgoppert · 2017-01-15T20:51:40Z

@jreback Any idea how to proceed with this. Can we put a skip in for mac and file an issue?

jorisvandenbossche

For me, it would be OK to skip those two tests on travis for mac for now.
Tests that check the exact labels are not that robust anyway, as it can change with matplotlib versions.

What you could test additionaly (and what should pass on mac as well), is that the ax.xaxis.get_major_formatter is the expected formatter (the one you now coded), and pass in a known value to check the formatting.

jorisvandenbossche · 2017-01-15T22:18:02Z

pandas/tseries/converter.py

+    """
+
+    @staticmethod
+    def format_timedelta_ticks(x, pos, n_decimals):


you can remove the pos here

jorisvandenbossche · 2017-01-15T22:19:38Z

pandas/tseries/plotting.py

+def format_timedelta_ticks(x, pos, n_decimals):
+    """
+    Convert seconds to 'D days HH:MM:SS.F'
+    """


I think this function can be removed no that you moved it to a Formatter?

jorisvandenbossche · 2017-01-15T22:22:39Z

pandas/tests/plotting/test_datetimelike.py

+        df = DataFrame(np.random.randn(len(rng), 3), rng)
+        ax = df.plot(fontsize=2)
+        fig = ax.get_figure()
+        fig.canvas.draw()


I don't think those two lines above are needed?

(and the same for the test above)

jorisvandenbossche · 2017-01-15T22:33:54Z

Roughly I mean something like:

# this is part of what you have now in a test
rng = timedelta_range('0', periods=10, freq='ns')
df = DataFrame(np.random.randn(len(rng), 3), rng)
ax = df.plot(fontsize=2)

# this could also test the formatting in another way that should not be skipped on mac osx
formatter =  ax.xaxis.get_major_formatter()
assert isinstance(formatter, TimeSeries_TimedeltaFormatter)
self.assertEqual(formatter(1.), '00:00:00.000000001')

jreback · 2017-01-15T22:45:04Z

I replicated the exact travis env for 3.5 / OSX and these test passed for me. so ok to skip i guess (though I wonder if these have something to do with some output settings, e.g. LOCALE).

tacaswell · 2017-01-18T18:36:21Z

This is not unreasonable. It might be worth adding a format_data_short method to the formatter to customize how the mouse-over gets displayed (instead of format_coords)

…ion display issues xref #15067

jreback · 2017-02-22T16:07:43Z

thanks @jgoppert I merged this (and skip those 2 comparisons on mac).

They seem kind of odd, though I couldn't reproduce them with a very similar local environment.

If you want to keep looking at them great (otherwise ok too).

Similarly if you want to do a mouse-over formatter as @tacaswell suggest would be great as well.

jreback · 2017-02-22T16:13:41Z

hmm the tick locations look a bit odd, is this normal? (this is the just merged PR)

cc @tacaswell
@jgoppert

jgoppert · 2017-02-22T18:25:41Z

Thanks for getting this in:

When you do rot=45, it works pretty well, but it does't align to tics, the autofmt_xdate does. Also, internally the tics are just set by the nanoseconds, so it isn't intelligent about frequency. If you look at the plot you will notice that the tic locations are at meaningful locations for nanoseconds. Staring at 1 day with a period of 5 starts at an odd place in terms of nanoseconds. If you look at when I just do seconds, it works well because seconds are just related to nano-seconds via a power of 10.

jreback · 2017-02-22T18:39:51Z

thanks for the expl @jgoppert

Author: James Goppert <james.goppert@gmail.com> Author: James Goppert <jgoppert@users.noreply.github.com> Closes pandas-dev#8711 Closes pandas-dev#15067 from jgoppert/tdi_plot_fix and squashes the following commits: 945ec14 [James Goppert] Merge branch 'master' into tdi_plot_fix 7db61ec [James Goppert] Create TimeSeries_TimedeltaFormatter. 232efe6 [James Goppert] Fix comment format and exception type for tdi plotting. 4eff697 [James Goppert] Add more time delta series plotting tests. f5f32bc [James Goppert] Link time delta index docs to better matplotlib docs. d588c2c [James Goppert] Fixes test for tdi w/o autofmt_xdate. b6e6a81 [James Goppert] Disables autofmt_xdate testing. c7851e3 [James Goppert] Adjusts tdi test draw calls to try to fix CI issue. 7d28842 [James Goppert] Switch to draw_idle to try to fix bug on xticks update. 3abc310 [James Goppert] Try plt.draw() instead of canvas.draw() to fix issue on osx 3.5. 91954bd [James Goppert] Finished unit test for timedelta plotting. 41ebc85 [James Goppert] Fixes for review comments from pandas-dev#15067. f021cbd [James Goppert] Support nano-second level precision x-axis labels. 5ec65fa [James Goppert] Plot fix for tdi and added more comments. b967d24 [James Goppert] flake8 fixes for tdi plotting. efe5636 [James Goppert] Adds custom plot formatting for TimedeltaIndex.

…ion display issues xref pandas-dev#15067

arpheno · 2017-11-15T00:13:17Z

Is it intentional that this doesn't work for histograms?

timb82 · 2018-09-06T13:25:02Z

it works only for fixed time step between points on the plot

consider following code:
idx2[0] was changed from 00:00 to 00:01 and xlabels on the plot are missing

import pandas as pd
import matplotlib.pyplot as plt

idx1 = pd.to_timedelta(['00:00:00', '00:00:30', '00:01:00', '00:01:30', '00:02:00'])
idx2 = pd.to_timedelta(['00:00:01', '00:00:30', '00:01:00', '00:01:30', '00:02:00'])
vals = range(5)
s1= pd.Series(vals, idx1)
s2= pd.Series(vals, idx2)

# Labels on x are ok
plt.figure()
plt.gca().set_title('fixed frequency f=30s')
s1.plot()
plt.gcf().autofmt_xdate()
plt.show()

# Labels on x are messed up
plt.figure()
plt.gca().set_title('variable frequency')
s2.plot()
plt.gcf().autofmt_xdate()
plt.show()

sinhrks added Timedelta Timedelta data type Visualization plotting labels Jan 5, 2017

TomAugspurger requested changes Jan 5, 2017

View reviewed changes

jreback reviewed Jan 5, 2017

View reviewed changes

jgoppert added a commit to jgoppert/pandas that referenced this pull request Jan 5, 2017

Fixes for review comments from pandas-dev#15067.

9dfb3a1

Still need to improve unit testing.

jgoppert commented Jan 5, 2017

View reviewed changes

jorisvandenbossche reviewed Jan 6, 2017

View reviewed changes

jgoppert added 11 commits January 14, 2017 16:46

Adds custom plot formatting for TimedeltaIndex.

efe5636

flake8 fixes for tdi plotting.

b967d24

Plot fix for tdi and added more comments.

5ec65fa

Support nano-second level precision x-axis labels.

f021cbd

Fixes for review comments from pandas-dev#15067.

41ebc85

Still need to improve unit testing.

Finished unit test for timedelta plotting.

91954bd

Try plt.draw() instead of canvas.draw() to fix issue on osx 3.5.

3abc310

Switch to draw_idle to try to fix bug on xticks update.

7d28842

Adjusts tdi test draw calls to try to fix CI issue.

c7851e3

Disables autofmt_xdate testing.

b6e6a81

Fixes test for tdi w/o autofmt_xdate.

d588c2c

jgoppert added 2 commits January 14, 2017 16:46

Link time delta index docs to better matplotlib docs.

f5f32bc

Add more time delta series plotting tests.

4eff697

jgoppert force-pushed the tdi_plot_fix branch from 4d240a8 to 4eff697 Compare January 14, 2017 21:49

jreback requested changes Jan 14, 2017

View reviewed changes

jgoppert added 2 commits January 14, 2017 17:14

Fix comment format and exception type for tdi plotting.

232efe6

Create TimeSeries_TimedeltaFormatter.

7db61ec

jorisvandenbossche reviewed Jan 15, 2017

View reviewed changes

Merge branch 'master' into tdi_plot_fix

945ec14

jreback closed this in 1400305 Feb 22, 2017

jreback added a commit that referenced this pull request Feb 22, 2017

TST: skip some timedelta plotting tests on mac (on travis) for precis…

486e384

…ion display issues xref #15067

jreback added this to the 0.20.0 milestone Feb 22, 2017

jgoppert deleted the tdi_plot_fix branch February 22, 2017 18:21

AnkurDedania pushed a commit to AnkurDedania/pandas that referenced this pull request Mar 21, 2017

TST: skip some timedelta plotting tests on mac (on travis) for precis…

52a5481

…ion display issues xref pandas-dev#15067

This was referenced Sep 15, 2017

Can't plot pd.Timedelta (or numpy.timedelta64) versus time #9154

Closed

Improve plotting Timedelta on y-axis (Timedelta string representation) #17553

Open

jorisvandenbossche mentioned this pull request Feb 2, 2018

No index labels on Dataframe plot with TimedeltaIndex #18910

Open

		@@ -116,6 +116,7 @@ Other enhancements

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

		- ``pd.TimedeltaIndex`` now has a custom datetick formatter specifically designed for nanosecond level precision (:issue:`8711`)

Adds custom plot formatting for TimedeltaIndex. #15067

Adds custom plot formatting for TimedeltaIndex. #15067

Conversation

jgoppert commented Jan 5, 2017 • edited Loading

jgoppert commented Jan 5, 2017

codecov-io commented Jan 5, 2017 • edited Loading

Codecov Report

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

jreback Jan 5, 2017 • edited Loading

Choose a reason for hiding this comment

jreback Jan 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

Choose a reason for hiding this comment

jgoppert Jan 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

jgoppert commented Jan 5, 2017 • edited Loading

jgoppert commented Jan 5, 2017 • edited Loading

jreback commented Jan 5, 2017

jgoppert commented Jan 5, 2017

jgoppert commented Jan 5, 2017

jreback commented Jan 5, 2017

jgoppert commented Jan 6, 2017

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

jgoppert commented Jan 14, 2017

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

jgoppert commented Jan 15, 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

jorisvandenbossche commented Jan 15, 2017

jreback commented Jan 15, 2017 • edited Loading

tacaswell commented Jan 18, 2017 • edited Loading

jreback commented Feb 22, 2017

jreback commented Feb 22, 2017 • edited Loading

jgoppert commented Feb 22, 2017

jreback commented Feb 22, 2017

arpheno commented Nov 15, 2017

timb82 commented Sep 6, 2018

jgoppert commented Jan 5, 2017 •

edited

Loading

codecov-io commented Jan 5, 2017 •

edited

Loading

jreback Jan 5, 2017 •

edited

Loading

jreback Jan 5, 2017 •

edited

Loading

jgoppert Jan 5, 2017 •

edited

Loading

jgoppert commented Jan 5, 2017 •

edited

Loading

jgoppert commented Jan 5, 2017 •

edited

Loading

jreback commented Jan 15, 2017 •

edited

Loading

tacaswell commented Jan 18, 2017 •

edited

Loading

jreback commented Feb 22, 2017 •

edited

Loading