[MRG] Tests for no sparse y support in RandomForests by dalmia · Pull Request #7996 · scikit-learn/scikit-learn

dalmia · 2016-12-07T03:21:44Z

Reference Issue

What does this implement/fix? Explain your changes.

This adds the test that we don't support sparse y as of yet for multilabel estimators.

Any other comments?

Since RandomTreesEmbedding doesn't make use of y, so I was not really clear as to how should it be included in the tests and hence, have left it out currently. Also, I did cross the maximum characters limit for the sake of clarity by a very small margin and noticed that there are other such lines (bigger) present, hence, chose to stick to it.

jnothman · 2016-12-07T12:01:24Z

We should be testing for a consistent error message text. use assert_raises_message

jnothman · 2016-12-07T12:02:28Z

But note that the issue only really concerns classifiers.

jnothman · 2016-12-07T12:03:00Z

I.e. classifiers and not embeddings, regressors or outlier detectors.

dalmia · 2016-12-08T03:32:34Z

Just noticed something that the assert_raise_message is not mentioned in __all__ in sklearn.utils.testing. Is that intentional?

jnothman · 2016-12-08T04:24:06Z

no, it is not. it's possible that a lot of things are missing from __all__ there.

…

On 8 December 2016 at 14:32, Aman Dalmia ***@***.***> wrote: Just noticed something that the assert_raise_message is not mentioned in __all__ . Is that intentional? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#7996 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz69Huw92_SxnFV1LQX1RZLXEC1B1Zks5rF3pTgaJpZM4LGJ4_> .

dalmia · 2016-12-08T04:52:30Z

I found something weird. When I make a multilabel dataset:

X_multilabel, y_multilabel = make_multilabel_classification(random_state=0, n_samples=100)
print(sp.issparse(X_multilabel))

False

But while testing the check_no_sparse_y_support (modified for convenience):

def check_no_sparse_y_support(name):
    X, y = X_multilabel, sp.csr_matrix(y_multilabel)
    TreeEstimator = ALL_TREES[name]
    TreeEstimator(random_state=0).fit(X, y)


def test_no_sparse_y_support():
    # Currently we don't support sparse y
    for name in ALL_TREES:
        check_no_sparse_y_support(name)

I get the error that X is sparse:

TypeError                                 Traceback (most recent call last)
<ipython-input-37-296b45b2dfcc> in <module>()
----> 1 test_no_sparse_y_support()

<ipython-input-36-105d4a9cae8c> in test_no_sparse_y_support()
     10     # Currently we don't support sparse y
     11     for name in ALL_TREES:
---> 12         check_no_sparse_y_support(name)

<ipython-input-36-105d4a9cae8c> in check_no_sparse_y_support(name)
      4     X, y = X_multilabel, sp.csr_matrix(y_multilabel)
      5     TreeEstimator = ALL_TREES[name]
----> 6     TreeEstimator(random_state=0).fit(X, y)
      7 
      8 

/media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/tree/tree.pyc in fit(self, X, y, sample_weight, check_input, X_idx_sorted)
   1027             sample_weight=sample_weight,
   1028             check_input=check_input,
-> 1029             X_idx_sorted=X_idx_sorted)
   1030         return self
   1031 

/media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/tree/tree.pyc in fit(self, X, y, sample_weight, check_input, X_idx_sorted)
    121         if check_input:
    122             X = check_array(X, dtype=DTYPE, accept_sparse="csc")
--> 123             y = check_array(y, ensure_2d=False, dtype=None)
    124             if issparse(X):
    125                 X.sort_indices()

/media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/utils/validation.pyc in check_array(array, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, warn_on_dtype, estimator)
    378     if sp.issparse(array):
    379         array = _ensure_sparse_format(array, accept_sparse, dtype, copy,
--> 380                                       force_all_finite)
    381     else:
    382         array = np.array(array, dtype=dtype, order=order, copy=copy)

/media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/utils/validation.pyc in _ensure_sparse_format(spmatrix, accept_sparse, dtype, copy, force_all_finite)
    241     """
    242     if accept_sparse in [None, False]:
--> 243         raise TypeError('A sparse matrix was passed, but dense '
    244                         'data is required. Use X.toarray() to '
    245                         'convert to a dense numpy array.')

TypeError: A sparse matrix was passed, but dense data is required. Use X.toarray() to convert to a dense numpy array.

This doesn't seem right. Please let me know what you think. @jnothman

jnothman · 2016-12-08T04:56:39Z

no, it's talking about y, not about X. look at that traceback closely.

…

On 8 December 2016 at 15:52, Aman Dalmia ***@***.***> wrote: I found something weird. When I make a multilabel dataset: X_multilabel, y_multilabel = make_multilabel_classification(random_state=0, n_samples=100)print(sp.issparse(X_multilabel)) False But while testing the check_no_sparse_y_support (modified for convenience): def check_no_sparse_y_support(name): X, y = X_multilabel, sp.csr_matrix(y_multilabel) TreeEstimator = ALL_TREES[name] TreeEstimator(random_state=0).fit(X, y) def test_no_sparse_y_support(): # Currently we don't support sparse y for name in ALL_TREES: check_no_sparse_y_support(name) I get the error that X is sparse: TypeError Traceback (most recent call last)<ipython-input-37-296b45b2dfcc> in <module>()----> 1 test_no_sparse_y_support() <ipython-input-36-105d4a9cae8c> in test_no_sparse_y_support() 10 # Currently we don't support sparse y 11 for name in ALL_TREES:---> 12 check_no_sparse_y_support(name) <ipython-input-36-105d4a9cae8c> in check_no_sparse_y_support(name) 4 X, y = X_multilabel, sp.csr_matrix(y_multilabel) 5 TreeEstimator = ALL_TREES[name]----> 6 TreeEstimator(random_state=0).fit(X, y) 7 8 /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/tree/tree.pyc in fit(self, X, y, sample_weight, check_input, X_idx_sorted) 1027 sample_weight=sample_weight, 1028 check_input=check_input,-> 1029 X_idx_sorted=X_idx_sorted) 1030 return self 1031 /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/tree/tree.pyc in fit(self, X, y, sample_weight, check_input, X_idx_sorted) 121 if check_input: 122 X = check_array(X, dtype=DTYPE, accept_sparse="csc")--> 123 y = check_array(y, ensure_2d=False, dtype=None) 124 if issparse(X): 125 X.sort_indices() /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/utils/validation.pyc in check_array(array, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, warn_on_dtype, estimator) 378 if sp.issparse(array): 379 array = _ensure_sparse_format(array, accept_sparse, dtype, copy,--> 380 force_all_finite) 381 else: 382 array = np.array(array, dtype=dtype, order=order, copy=copy) /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/utils/validation.pyc in _ensure_sparse_format(spmatrix, accept_sparse, dtype, copy, force_all_finite) 241 """ 242 if accept_sparse in [None, False]:--> 243 raise TypeError('A sparse matrix was passed, but dense ' 244 'data is required. Use X.toarray() to ' 245 'convert to a dense numpy array.')TypeError: A sparse matrix was passed, but dense data is required. Use X.toarray() to convert to a dense numpy array. This doesn't seem right. Please let me know what you think. @jnothman <https://github.com/jnothman> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#7996 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz66cALuC3swYmn9pYnIm4QIbuCmIXks5rF40QgaJpZM4LGJ4_> .

jnothman · 2016-12-08T04:56:52Z

but yes, perhaps the error message could be improved

…

On 8 December 2016 at 15:56, Joel Nothman ***@***.***> wrote: no, it's talking about y, not about X. look at that traceback closely. On 8 December 2016 at 15:52, Aman Dalmia ***@***.***> wrote: > I found something weird. When I make a multilabel dataset: > > X_multilabel, y_multilabel = make_multilabel_classification(random_state=0, n_samples=100)print(sp.issparse(X_multilabel)) > > False > > But while testing the check_no_sparse_y_support (modified for > convenience): > > def check_no_sparse_y_support(name): > X, y = X_multilabel, sp.csr_matrix(y_multilabel) > TreeEstimator = ALL_TREES[name] > TreeEstimator(random_state=0).fit(X, y) > > def test_no_sparse_y_support(): > # Currently we don't support sparse y > for name in ALL_TREES: > check_no_sparse_y_support(name) > > I get the error that X is sparse: > > TypeError Traceback (most recent call last)<ipython-input-37-296b45b2dfcc> in <module>()----> 1 test_no_sparse_y_support() > <ipython-input-36-105d4a9cae8c> in test_no_sparse_y_support() > 10 # Currently we don't support sparse y > 11 for name in ALL_TREES:---> 12 check_no_sparse_y_support(name) > <ipython-input-36-105d4a9cae8c> in check_no_sparse_y_support(name) > 4 X, y = X_multilabel, sp.csr_matrix(y_multilabel) > 5 TreeEstimator = ALL_TREES[name]----> 6 TreeEstimator(random_state=0).fit(X, y) > 7 > 8 > /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/tree/tree.pyc in fit(self, X, y, sample_weight, check_input, X_idx_sorted) > 1027 sample_weight=sample_weight, > 1028 check_input=check_input,-> 1029 X_idx_sorted=X_idx_sorted) > 1030 return self > 1031 > /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/tree/tree.pyc in fit(self, X, y, sample_weight, check_input, X_idx_sorted) > 121 if check_input: > 122 X = check_array(X, dtype=DTYPE, accept_sparse="csc")--> 123 y = check_array(y, ensure_2d=False, dtype=None) > 124 if issparse(X): > 125 X.sort_indices() > /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/utils/validation.pyc in check_array(array, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, warn_on_dtype, estimator) > 378 if sp.issparse(array): > 379 array = _ensure_sparse_format(array, accept_sparse, dtype, copy,--> 380 force_all_finite) > 381 else: > 382 array = np.array(array, dtype=dtype, order=order, copy=copy) > /media/aman/BE66ECBA66EC7515/Open Source/scikit-learn/sklearn/utils/validation.pyc in _ensure_sparse_format(spmatrix, accept_sparse, dtype, copy, force_all_finite) > 241 """ 242 if accept_sparse in [None, False]:--> 243 raise TypeError('A sparse matrix was passed, but dense ' 244 'data is required. Use X.toarray() to ' 245 'convert to a dense numpy array.')TypeError: A sparse matrix was passed, but dense data is required. Use X.toarray() to convert to a dense numpy array. > > This doesn't seem right. Please let me know what you think. @jnothman > <https://github.com/jnothman> > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > <#7996 (comment)>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/AAEz66cALuC3swYmn9pYnIm4QIbuCmIXks5rF40QgaJpZM4LGJ4_> > . >

dalmia · 2016-12-08T05:11:16Z

Sorry, yes I meant to say that the error message includes X which makes things a bit confusing. I found the error message for RandomForests as the most precise one. Because of sparse matrices not being allowed in y = check_array(y, ensure_2d=False, dtype=None), we weren't getting a consistent error message. Thought of a workaround, please review.

dalmia · 2016-12-08T10:23:17Z

Also, corrected a few PEP8 errors already existing in the files.

dalmia · 2016-12-10T06:32:23Z

ping @jnothman @amueller @lesteve

jnothman · 2016-12-11T09:59:37Z

It's fair enough to ping, but please wait a little more than two days! there's far too much to get through

…

On 10 December 2016 at 17:32, Aman Dalmia ***@***.***> wrote: ping @jnothman <https://github.com/jnothman> @amueller <https://github.com/amueller> @lesteve <https://github.com/lesteve> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#7996 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz6yP7qagQ0fojSFh_JKTcy134Q6lIks5rGkd4gaJpZM4LGJ4_> .

jnothman · 2016-12-11T10:00:01Z

I assure you: my silence is not because I intend to ignore this issue.

…

On 11 December 2016 at 20:59, Joel Nothman ***@***.***> wrote: It's fair enough to ping, but please wait a little more than two days! there's far too much to get through On 10 December 2016 at 17:32, Aman Dalmia ***@***.***> wrote: > ping @jnothman <https://github.com/jnothman> @amueller > <https://github.com/amueller> @lesteve <https://github.com/lesteve> > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > <#7996 (comment)>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/AAEz6yP7qagQ0fojSFh_JKTcy134Q6lIks5rGkd4gaJpZM4LGJ4_> > . >

dalmia · 2016-12-11T12:02:31Z

@jnothman I was not really sure as to when is it appropriate to ping. Thank you for clearing it up.

jnothman · 2016-12-27T03:48:28Z

sklearn/ensemble/tests/test_forest.py

+                                                   n_samples=50)
+    y_sparse = csr_matrix(y)
+    ForestClassifier = FOREST_CLASSIFIERS[name]
+    assert_raise_message(ValueError, "Unknown label type: 'unknown'",


I don't think that's a useful error message. For sparse X not supported we raise TypeError (not that I'm sure this decision was correct) with message "A sparse matrix was passed, but dense data is required. Use X.toarray() to convert to a dense numpy array."

One option is to use check_X_y and extend upon its multioutput parameter with an allow_sparse_y parameter to standardise checking and error message.

jnothman · 2016-12-27T03:49:11Z

sklearn/tree/tree.py

            X = check_array(X, dtype=DTYPE, accept_sparse="csc")
-            y = check_array(y, ensure_2d=False, dtype=None)
+            y = check_array(y, ensure_2d=False, dtype=None,
+                            accept_sparse=accept_sparse)


But surely accept_sparse should be False?

I added this to ensure a consistent error message for the classification case. Since we are to change the message, I'll revert this.

Conflicts: sklearn/ensemble/forest.py

jnothman · 2017-01-06T05:21:48Z

sklearn/ensemble/tests/test_forest.py

+    y_sparse = csr_matrix(y)
+    ForestClassifier = FOREST_CLASSIFIERS[name]
+    assert_raise_message(TypeError, 'A sparse matrix was passed, but '
+                         'dense data is required. Use X.toarray() to '


I think referring to X.toarray() will only confuse the user.

Yes, I agree that is an issue. But is there any other alternative to this because at the end we are relying on check_array to raise the error message?

dalmia · 2017-01-06T16:08:42Z

Adding separately a check for sparse multilabel y seems the only option to produce a custom message if the current one seems confusing.

jnothman · 2017-01-07T10:38:04Z

check_array isn't set in stone. You could add a parameter for a variable name placeholder to be used in error messages. Or make a modification to check_X_y. This issue is best solved by modifying something in utils.validation so that then estimators can be consistent in their error messages.

…

On 7 January 2017 at 03:08, Aman Dalmia ***@***.***> wrote: Adding separately a check for sparse multilabel y seems the only option to produce a custom message if the current one seems confusing. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#7996 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz66feC5G_X7YoVKathrHpSb2Iwbftks5rPmcMgaJpZM4LGJ4_> .

dalmia · 2017-01-08T02:13:13Z

Something like this?

dalmia · 2017-01-08T05:27:27Z

Oh my bad. Please check now.

jnothman

Please also use the variable_name in the ensure_2d check.

FWIW, git grep 'check_array([^X]' sklearn/ suggests there are many more places in the codebase that variable_name should be used.

jnothman · 2017-01-08T07:45:32Z

I'm thinking that perhaps the addition of variable_name to check_array might benefit from being its own separate PR. WDYT?

dalmia · 2017-01-08T08:23:24Z

Yes, I agree. This seems an independent feature addition and seems cleaner to keep it in a separate PR. If I may have an overview as to at what places it might be needed, I can make that PR with the suitable change.

dalmia · 2017-01-08T09:02:23Z

I think this needs a rebuild:

======================================================================
FAIL: sklearn.decomposition.tests.test_pca.test_singular_values
----------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Python35\lib\site-packages\nose\case.py", line 198, in runTest
    self.test(*self.arg)
  File "C:\Python35\lib\site-packages\sklearn\decomposition\tests\test_pca.py", line 266, in test_singular_values
    np.linalg.norm(X_apca, "fro")**2.0, 12)
  File "C:\Python35\lib\site-packages\numpy\testing\utils.py", line 842, in assert_array_almost_equal
    precision=decimal)
  File "C:\Python35\lib\site-packages\numpy\testing\utils.py", line 665, in assert_array_compare
    raise AssertionError(msg)
AssertionError: 
Arrays are not almost equal to 12 decimals
(mismatch 100.0%)
 x: array(613.0583877392909)
 y: array(613.0583877392893)
sklearn.decomposition.tests.test_dict_learning.test_dict_learning_reconstruction_parallel: 4.4997s

jnothman · 2017-01-08T09:59:28Z

Use git grep to find possible places for variable_name

…

On 8 January 2017 at 20:02, Aman Dalmia ***@***.***> wrote: I think this needs a rebuild: ======================================================================FAIL: sklearn.decomposition.tests.test_pca.test_singular_values---------------------------------------------------------------------- Traceback (most recent call last): File "C:\Python35\lib\site-packages\nose\case.py", line 198, in runTest self.test(*self.arg) File "C:\Python35\lib\site-packages\sklearn\decomposition\tests\test_pca.py", line 266, in test_singular_values np.linalg.norm(X_apca, "fro")**2.0, 12) File "C:\Python35\lib\site-packages\numpy\testing\utils.py", line 842, in assert_array_almost_equal precision=decimal) File "C:\Python35\lib\site-packages\numpy\testing\utils.py", line 665, in assert_array_compare raise AssertionError(msg)AssertionError: Arrays are not almost equal to 12 decimals (mismatch 100.0%) x: array(613.0583877392909) y: array(613.0583877392893) sklearn.decomposition.tests.test_dict_learning.test_dict_learning_reconstruction_parallel: 4.4997s — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#7996 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz67FI0xYsnKnRrIR2fbuFNBHiR50Vks5rQKYhgaJpZM4LGJ4_> .

dalmia · 2017-01-09T05:37:47Z

Would you prefer passing the variable_name for checks like:
data = check_array(data, accept_sparse='csr') or should I restrict myself to changing it wherever y appears?

jnothman · 2017-01-09T06:55:26Z

if data is a parameter name within the api, then data too.

…

On 9 Jan 2017 4:37 pm, "Aman Dalmia" ***@***.***> wrote: Would you prefer passing the variable_name for checks like: data = check_array(data, accept_sparse='csr') or should I restrict myself to changing it wherever y appears? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#7996 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz6xBcBR-LDiA6CF1wT2RktnUo4ynlks5rQcesgaJpZM4LGJ4_> .

jnothman · 2017-06-08T05:17:16Z

I assume @dalmia, you'd like someone else to complete this?

jnothman · 2017-06-14T13:56:42Z

Since #9068, check_array no longer mentions X

adrinjalali · 2025-09-25T13:41:58Z

Closing due to lack of activity from the OP. Happy to reopen if you get to work on this further.

TST: tests for no sparse y support in RandomForests

f2f1420

dalmia changed the title ~~[MRG] Tests for no sparse y support in RandomForests~~ [WIP] Tests for no sparse y support in RandomForests Dec 8, 2016

TST: added tests to ensure consistent error message

0455e5f

dalmia changed the title ~~[WIP] Tests for no sparse y support in RandomForests~~ [MRG] Tests for no sparse y support in RandomForests Dec 8, 2016

FIX: removed indentation error

64e1cab

jnothman reviewed Dec 27, 2016

View reviewed changes

dalmia added 2 commits January 5, 2017 09:29

FIX: changes for getting better error message

fb68c44

Merge branch 'master' into 7886

9ed5088

Conflicts: sklearn/ensemble/forest.py

jnothman reviewed Jan 6, 2017

View reviewed changes

ENH: added placeholder for custom error messages

ac77578

ENH: added variable_name for error messages

bb4c1a0

jnothman reviewed Jan 8, 2017

View reviewed changes

FIX: remove flake8 errors

7de6c47

dalmia mentioned this pull request Jan 9, 2017

[MRG] Added variable_name as a parameter for check_array #8178

Closed

dalmia mentioned this pull request Feb 22, 2017

[WIP] support sparse y in GridSearchCV #8438

Closed

6 tasks

jnothman added the Need Contributor label Jun 8, 2017

lesteve added help wanted and removed Need Contributor labels Oct 18, 2017

jnothman added the Stalled label Feb 6, 2018

github-actions bot added module:ensemble module:tree module:utils labels Mar 2, 2020

Base automatically changed from master to main January 22, 2021 10:49

lucyleeow added this to Quansight's scikit-learn Project Board Feb 27, 2025

github-project-automation bot moved this to Todo📬 in Quansight's scikit-learn Project Board Feb 27, 2025

adrinjalali closed this Sep 25, 2025

github-project-automation bot moved this from Todo📬 to Done🚀 in Quansight's scikit-learn Project Board Sep 25, 2025

Uh oh!

Conversation

dalmia commented Dec 7, 2016

Reference Issue

What does this implement/fix? Explain your changes.

Any other comments?

Uh oh!

jnothman commented Dec 7, 2016

Uh oh!

jnothman commented Dec 7, 2016

Uh oh!

jnothman commented Dec 7, 2016

Uh oh!

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

Uh oh!

Uh oh!

jnothman commented Dec 8, 2016 via email

Uh oh!

dalmia commented Dec 8, 2016

Uh oh!

jnothman commented Dec 8, 2016 via email

Uh oh!

jnothman commented Dec 8, 2016 via email

Uh oh!

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

Uh oh!

Uh oh!

dalmia commented Dec 8, 2016

Uh oh!

dalmia commented Dec 10, 2016

Uh oh!

jnothman commented Dec 11, 2016 via email

Uh oh!

jnothman commented Dec 11, 2016 via email

Uh oh!

dalmia commented Dec 11, 2016

Uh oh!

jnothman Dec 27, 2016

Choose a reason for hiding this comment

Uh oh!

jnothman Dec 27, 2016

Choose a reason for hiding this comment

Uh oh!

dalmia Jan 4, 2017

Choose a reason for hiding this comment

Uh oh!

jnothman Jan 6, 2017

Choose a reason for hiding this comment

Uh oh!

dalmia Jan 6, 2017

Choose a reason for hiding this comment

Uh oh!

dalmia commented Jan 6, 2017

Uh oh!

jnothman commented Jan 7, 2017 via email

Uh oh!

dalmia commented Jan 8, 2017

Uh oh!

dalmia commented Jan 8, 2017

Uh oh!

jnothman left a comment • 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!

jnothman commented Jan 8, 2017

Uh oh!

dalmia commented Jan 8, 2017

Uh oh!

dalmia commented Jan 8, 2017

Uh oh!

jnothman commented Jan 8, 2017 via email

Uh oh!

dalmia commented Jan 9, 2017

Uh oh!

jnothman commented Jan 9, 2017 via email

Uh oh!

jnothman commented Jun 8, 2017

Uh oh!

jnothman commented Jun 14, 2017

Uh oh!

dalmia commented Dec 8, 2016 •

edited

Loading

dalmia commented Dec 8, 2016 •

edited

Loading

jnothman left a comment •

edited

Loading