Allow count worfklow executions by different filters #18

phishermaan · 2016-11-15T02:22:46Z

add unit tests and fake acceptance tests.

…ons, reformat for flake8

coveralls · 2016-11-15T19:07:21Z

Coverage increased (+1.03%) to 89.384% when pulling ef7a7ca on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

movefast1

Looks good overall :) It'll be great if you can also make the naming & docstrings of functions easier for users to understand and choose from since we've got plenty of apis.

movefast1 · 2016-11-29T19:38:57Z

py_swf/clients/workflow.py

+            latest_start_date=latest_start_date,
+        )
+
+    def count_open_workflow_by_type(self, name, oldest_start_date=None, latest_start_date=None, version=None):


Since this function takes in a time filter in addition to the type filter, maybe we could also reflect this behavior in its name to distinguish from count_open_workflow_by_time?

Ah you mean like count_open_workflow_by_type_and_time() ? I think it is fine. Well I think the function name reflects the purpose or args we really need. Since time range is an optional argument, do you think it is OK to keep it and thus have a shorter name? I think the doc string may help?

You are right, since they are optional params a clear docs string should suffice.

movefast1 · 2016-11-29T21:21:13Z

py_swf/clients/workflow.py

+            )
+
+
+def _build_time_filter_dict(oldest_start_date=None, latest_start_date=None, oldest_close_date=None, latest_close_date=None):


We can probably make it much simpler by providing separate apis when filtering by start time vs close time since they are mutually exclusive. In that case we won't need the validation or the exception .

Yeah I think separate them may be more reasonable.

What's the value for breaking these into separate APIs? And what's the value for providing a custom validation exception instead of utilizing the existing boto3 exceptions?

The goal of this library is to function as a thin wrapper around boto3 and providing the Amazon API in a pythonic manner. One way is to provide as close to a one-to-one mapping of the APIs provided by boto3, and to present them as pythonic as possible. This makes it easier for people to use this library as much of the documentation provided by Amazon also applies here. Adding custom validation increases cognitive load since not only is there validation by boto3, there is validation in this library too.

Hmm I think the your point on validation part is reasonable. boto client should warn us when calling it wrongly. I will remove the new exception class.
For separating APIs by start date & close date, I think it makes more sense because start_date and close_date are mutually exclusive. A date arg is required for these boto API calls. Separating them makes users easy to understand what date args is required, like a start date is always needed.

Just in case that the diff is not updated, I have already stopped date validation in _build_time_filter_dict after talking with @movefast1 . Currently I only use date_tuple to get oldest_date and latest_date

For separating APIs by start date & close date, I think it makes more sense because start_date and close_date are mutually exclusive. A date arg is required for these boto API calls. Separating them makes users easy to understand what date args is required, like a start date is always needed.

Multiplexing the API based on the inputs increases the number of functions this library exposes and complicates the library. Having lots of functions that perform the same operation make it harder for a developer to figure out which function to use, and harder for us to maintain since it increases the surface area of behavior division when we want to change the behavior of counting workflows. Again, since boto3 does validation, it'll complain if both start date and close dates are used at the same time. We could have one time input, and a flag specifying whether it's a start date, or a close date, and the rest of the filters as optional kwargs. So the function signature could look like:

def count_executions(self, latest_date, oldest_date, count_open=True, filter_by_start=True, workflow_id=None, tag=None, type=None)

Thoughts?

Hmm I think it is possible to combine all the filter type in this count_executions. In that case my question is do I have to document the usage in great detail so that the user won't get confused when using count_executions? Or it is up to the user to learn by having exceptions? Like to know which filters can be used together while others cannot?

The reason that I separate count open workflow from count close workflow is their filter parameters don't really mirror each other(open workflow only take start date, while close workflow can take start & close date, and it can take a dedicated close_status filter). I could combine APIs in count_close_workflows to take one date paramter to reduce API numbers, while keep the API separated by filter type. How do you think of that?

Having high quality documentation I think is a good idea.

You bring up a good point that Amazon separates the API into count count_open_workflow_executions and count_closed_workflow_executions, though, I still think we should minimize having to multiplex these two calls into every permutation of arguments puissant to the arguments defined by the SWF API. Looking at one function and reading a single place for documentation seems easier than having to read each function's name, then reading each docstring for each function to determine what it does.

Hmm I think maybe we can reduce the API into 2, which is count_open_workflow, count_closed_workflow. Then with proper document it should be easy to understand.

movefast1 · 2016-11-29T21:30:10Z

py_swf/clients/workflow.py

+
+
+def _build_tag_filter_dict(tag):
+    filter_dict = {'tag': tag}


we don't have to define this var since it's used in the next line immediately after the definition.

Oh yeah you are right.

coveralls · 2016-12-01T01:05:50Z

Coverage increased (+0.8%) to 89.199% when pulling 344bdb2 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-01T02:13:28Z

Coverage increased (+1.2%) to 89.539% when pulling 991fc96 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-01T02:19:04Z

Coverage increased (+1.2%) to 89.539% when pulling 53609ba on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

movefast1 · 2016-12-02T02:16:18Z

py_swf/clients/workflow.py

@@ -348,7 +362,7 @@ def _count_closed_workflow_executions(

 def _build_time_filter_dict(oldest_start_date=None, latest_start_date=None, oldest_close_date=None, latest_close_date=None):
    """
-    Build time_filter_dict for count open/closed workflow executions.
+    Build time_filter_dict for calls to _count_closed_workflow_executions and _count_open_workflow_executions.
    Return result is a dict: {time_filter_type : filter_dict}
    filter_dict must contains key 'oldestDate'.


It seems like we only removed the validations here but didn't break it into two separate functions? Did you just forget or it was intentional?

Ah I did separate the count by start_date and close_date. Take count_closed_workflow_by_id_and_start_time and count_closed_workflow_by_id_and_close_time by example. Since the function args make sure we only get start_date or close_date, that is why I remove the validation.

quantsini · 2016-12-07T19:21:36Z

HISTORY.rst

+1.2.0 (2016-11-14)
++++++++++++++++++
+
+* WorkflowClient now supports count open or closed workflows by time range and other filter 


Let's be more explicit on what "other filter" means.

Sure thing. I will be more specific on the change.

quantsini · 2016-12-07T19:32:03Z

py_swf/clients/workflow.py

+            )
+
+
+def _build_time_filter_dict(oldest_start_date=None, latest_start_date=None, oldest_close_date=None, latest_close_date=None):


What's the value for breaking these into separate APIs? And what's the value for providing a custom validation exception instead of utilizing the existing boto3 exceptions?

The goal of this library is to function as a thin wrapper around boto3 and providing the Amazon API in a pythonic manner. One way is to provide as close to a one-to-one mapping of the APIs provided by boto3, and to present them as pythonic as possible. This makes it easier for people to use this library as much of the documentation provided by Amazon also applies here. Adding custom validation increases cognitive load since not only is there validation by boto3, there is validation in this library too.

…o exception raising

coveralls · 2016-12-07T20:19:02Z

Coverage increased (+1.3%) to 89.691% when pulling 8f1e029 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

quantsini · 2016-12-08T23:09:15Z

py_swf/clients/workflow.py

+        :param workflow_filter_dict: dict. filter type as key and filter criteria as value. filter criteria is a dictionary
+        :return: total number of closed workflows that meet the filter criteria
+        """
+        if not workflow_filter_dict:


nit: explicitly check for is None. Also, better to inline these as kwargs, instead of having to instantiate a dictionary that will be passed in as kwargs anyway to the underlying count function.

coveralls · 2016-12-09T22:12:03Z

Coverage decreased (-0.6%) to 87.755% when pulling 0c5607e on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-09T22:33:04Z

Coverage decreased (-0.6%) to 87.755% when pulling 269a39d on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-09T22:47:14Z

Coverage decreased (-2.2%) to 86.163% when pulling c8a276d on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-09T23:16:26Z

Coverage decreased (-0.04%) to 88.323% when pulling 43810ca on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-09T23:30:29Z

Coverage increased (+0.4%) to 88.788% when pulling 6efafc0 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

quantsini · 2016-12-12T21:25:31Z

py_swf/clients/workflow.py

+    elif close_status:
+        workflow_filter_dict = _build_close_status_filter_dict(close_status)
+    else:
+        workflow_filter_dict = None


if this returned an empty dict, the caller doesn't have to special case None.

coveralls · 2016-12-12T21:43:22Z

Coverage increased (+0.4%) to 88.72% when pulling 6affa00 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

quantsini · 2016-12-12T21:28:51Z

py_swf/clients/workflow.py

+            close_status=close_status,
+        )
+
+        if workflow_filter_dict is None:


allowing _build_time_filter_dict to only return dictionaries would make this is None check not necessary

quantsini · 2016-12-12T21:29:16Z

py_swf/clients/workflow.py

+        """
+        Count the number of open workflows for a domain. You can pass in filtering criteria.
+        The results are best effort and may not exactly reflect recent updates and changes.
+        http://docs.aws.amazon.com/amazonswf/latest/apireference/API_CountOpenWorkflowExecutions.html


protip: our sphinx configuration is set up so that you can link to the external SWF docs. Example:
Passthrough to :meth:~SWF.Client.count_open_workflow_executions``

quantsini · 2016-12-12T21:29:36Z

py_swf/clients/workflow.py

+        """
+        Count the number of closed workflows for a domain. You can pass in filtering criteria.
+        The results are best effort and may not exactly reflect recent updates and changes.
+        http://docs.aws.amazon.com/amazonswf/latest/apireference/API_CountClosedWorkflowExecutions.html#SWF-CountClosedWorkflowExecutions-request-executionFilter


Ditto the comment on built in sphinx configuration

quantsini · 2016-12-12T21:31:20Z

py_swf/clients/workflow.py

+        date_tuple = (oldest_close_date, latest_close_date)
+
+    if not date_tuple:
+        return None


Returning an empty dictionary would make it easier for callers since they wouldn't need to special case a None check.

quantsini · 2016-12-12T21:31:35Z

py_swf/clients/workflow.py

+
+def _build_type_filter_dict(name, version):
+    filter_dict = {'name': name}
+    if version:


It's unclear whether this should be checking for Noneness. Should '' (empty string) be a valid input?

Yeah good catch. empty string is allowed.

quantsini · 2016-12-12T21:38:04Z

py_swf/clients/workflow.py

+    if oldest_start_date:
+        filter_type = 'startTimeFilter'
+        date_tuple = (oldest_start_date, latest_start_date)
+    elif oldest_close_date:


Let's explicitly check for None. The default argument is None and I assume we want to change behavior based on its Noneness.

quantsini · 2016-12-12T21:39:25Z

py_swf/clients/workflow.py

+    Build time_filter_dict for calls to _count_closed_workflow_executions and _count_open_workflow_executions.
+    Return result is a dict: {time_filter_type : filter_dict}
+    filter_dict must contains key 'oldestDate'.
+    return example:


Could you also include the input for this example? That would help people make the connection between the inputs and the example return. Bonus points for more examples showcasing the behavior.

quantsini · 2016-12-12T21:47:38Z

py_swf/clients/workflow.py

+        return None
+
+    oldest_date, latest_date = date_tuple
+    filter_dict = dict(oldestDate=oldest_date)


nit: Let's consistently use {} or dict(). There's a lot of resources online about which one is better. http://stackoverflow.com/questions/2972212/creating-an-empty-list-in-python is one example.

quantsini · 2016-12-12T21:48:36Z

py_swf/clients/workflow.py

+        workflow_id=None,
+        close_status=None,
+):
+    if workflow_name:


Why are these mutually exclusive? (for example, I wouldn't be able to search on the conjunction of both tag and close_status.

In the docstring of count open/closed workflows there is a description saying that these filters are mutually exclusive.

Gotcha. I see it here in the SWF documentation too: http://docs.aws.amazon.com/amazonswf/latest/apireference/API_CountClosedWorkflowExecutions.html

quantsini · 2016-12-12T22:00:56Z

py_swf/clients/workflow.py

+        return count_closed_workflow_within_time_range(**workflow_filter_dict)
+
+
+def _build_time_filter_dict(oldest_start_date=None, latest_start_date=None, oldest_close_date=None, latest_close_date=None):


I think this function would be smaller if it dumbly built the dictionary allowing for "invalid" configurations. We can rely on the boto3 library to raise an appropriate error.

This current implementation can have silent errors since the caller can provide startTimeFilter and closeTimeFilter and this library would assume startTimeFilter (when the caller may only meant to do closeTimeFilter). boto3 would otherwise error with invalid inputs.

If that is what you are worried about, I will try building both startTimeFilter dict and closeTimeFilter and put them in the result. In this case if user passes in both start date and close date then boto3 will get them both. This will cause exceptions and prevent silent errors.

…n with invalidate inputs

coveralls · 2016-12-12T23:51:37Z

Coverage increased (+0.3%) to 88.668% when pulling ad6e450 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-13T00:00:38Z

Coverage increased (+0.3%) to 88.668% when pulling 5fd5e76 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

quantsini · 2016-12-12T23:51:34Z

py_swf/__init__.py

@@ -4,4 +4,4 @@

 __author__ = 'Henri Bai'
 __email__ = 'hbai@yelp.com'
-__version__ = '1.2.0'
+__version__ = '1.2.0-a1-abcde'


What's up with this?

quantsini · 2016-12-12T23:58:12Z

py_swf/clients/workflow.py

+            latest_close_date=latest_close_date,
+        )
+
+        count_closed_workflow_within_time_range = partial(


I think this may read better if we don't use the partial since we're not re-using the partial, and we're just passing in **workflow_filter_dict immediately after.

I don't think we can unpack workflow_filter_dict while having the time_filter_dict unpacked as key word dict. Or am I understanding it wrong?

Is this specifically combining dicts? You can do this with dict.update: https://docs.python.org/2/library/stdtypes.html#dict.update
Using partial for combining **kwargs seems overkill.

quantsini · 2016-12-13T00:13:15Z

tests/acceptance/workflow_test.py

@@ -112,3 +113,71 @@ def test_walk_execution_history(workflow_client, decision_client):
    decision_task = poll_decision_and_respond(decision_client)
    terminate_workflow(workflow_client, workflow_id)
    walk_execution_history(decision_client, workflow_id, decision_task.workflow_run_id)
+
+
+def test_count_open_workflow_executions(workflow_client):


Can you add tests that assert validation errors are raised for invalid input?

Hmm I think the boto_client we are using is a mock thus cannot raise exceptions, since we want the validation to happen on boto_client only? The integration test has covered it though if you want to have a look.

The line of code this comment is tagged is for the acceptence tests. This boto client is not a mock and actually calls out to SWF. Although our current travis ci integration doesn't run them properly due to #19

https://github.com/Yelp/pyswf/blob/master/tests/acceptance/conftest.py#L42 shows how the boto client is created in a pytest conftest under the acceptance folder.

quantsini · 2016-12-13T00:13:56Z

tests/unit/clients/workflow_test.py

@@ -54,3 +77,239 @@ def test_terminate_workflow(workflow_config, workflow_client, boto_client):
        workflowId='workflow_id',
        reason='reason',
    )
+
+
+def test_build_time_filter_dict(oldest_start_date, latest_start_date):


Since you're already unit testing with different inputs, let's break this up into individual tests.

quantsini · 2016-12-13T00:14:33Z

tests/unit/clients/workflow_test.py

+    assert time_filter_dict['closeTimeFilter'] == {'oldestDate': start_date, 'latestDate': end_date}
+
+
+def test_count_open_workflow_executions(workflow_config, workflow_client, boto_client, oldest_start_date, latest_start_date):


Ditto breaking up the big test into smaller unit tests.

quantsini · 2016-12-13T00:14:50Z

tests/unit/clients/workflow_test.py

+    )
+
+
+def test_count_closed_workflow_executions(


Ditto breaking up the big test into smaller tests.

quantsini · 2016-12-13T00:15:47Z

tests/unit/clients/workflow_test.py

+    _assert_boto_method_called_with(boto_client.count_open_workflow_executions, workflow_config.domain, **kwargs)
+
+
+def _assert_count_closed_workflow_executions_called_with(boto_client, workflow_config, **kwargs):


Let's remove these calls since the level of indirection isn't necessary and would make things more explicitly when asserting calls.

coveralls · 2016-12-13T01:30:09Z

Coverage increased (+0.5%) to 88.855% when pulling c581995 on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

coveralls · 2016-12-13T01:47:24Z

Coverage increased (+0.5%) to 88.839% when pulling 8329e6c on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

quantsini · 2016-12-13T01:50:36Z

py_swf/clients/workflow.py

+
+        return self.boto_client.count_open_workflow_executions(
+            domain=self.workflow_client_config.domain,
+            startTimeFilter=start_time_filter_dict['startTimeFilter'],


Should this also mimic the count_closed_workflow_executions so that we combine the dicts before unpacking?

I think since there is only one type of time filter, there is no benefits of packing them together. Passing them separately makes users be clear on what args they are expecting I think.

…tion to boto

coveralls · 2016-12-13T02:39:16Z

Coverage decreased (-0.4%) to 88.0% when pulling 3bf180f on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

…ing boto_client

coveralls · 2016-12-14T20:00:01Z

Coverage increased (+11.09%) to 99.448% when pulling 8b51a5e on phishermaan:allow-count-worfklow-executions-by-state into 296c47c on Yelp:master.

quantsini · 2016-12-14T21:11:22Z

LGTM!!

phishermaan added 5 commits November 14, 2016 16:09

add count open/closed workflow, finish writing unit tests

e7808e1

add fake accpetance test because they are set to xfail

b6ffeb6

add test version

ac71808

update version & history

29bb42f

move _build_time_filter_dict to _count_<open, close>_workflow_executi…

ef7a7ca

…ons, reformat for flake8

movefast1 reviewed Nov 29, 2016

View reviewed changes

seperate start date and close date query

344bdb2

any count query need a time filter

991fc96

correct version

53609ba

movefast1 reviewed Dec 2, 2016

View reviewed changes

quantsini suggested changes Dec 7, 2016

View reviewed changes

remove custom exception type InvalidQueryToCountWorkflow, rely on bot…

8f1e029

…o exception raising

quantsini reviewed Dec 8, 2016

View reviewed changes

only have count open/closed workflow APIs, combine filters into the API

0c5607e

correct test version to be formal

269a39d

update acceptance test

c8a276d

delete unnecessary acceptance tests that lower test coverage

43810ca

remove unnecessary part to improve coverall

6efafc0

quantsini reviewed Dec 12, 2016

View reviewed changes

use sphinx external link

6affa00

quantsini suggested changes Dec 12, 2016

View reviewed changes

phishermaan added 2 commits December 12, 2016 15:49

make sure _build_time_filter generate time filter as user inputs, eve…

8b40e21

…n with invalidate inputs

alpha version number for test

ad6e450

correct version

5fd5e76

quantsini suggested changes Dec 13, 2016

View reviewed changes

separate unit tests

c581995

combine work_filter with time_dict

8329e6c

quantsini reviewed Dec 13, 2016

View reviewed changes

just let workflow_filter_dict build whatever user inputs, pass valida…

3bf180f

…tion to boto

test coverage does not count acceptance test because failure at build…

8b51a5e

…ing boto_client

quantsini approved these changes Dec 14, 2016

View reviewed changes

quantsini merged commit 1f073e5 into Yelp:master Dec 14, 2016

		)


		def _build_time_filter_dict(oldest_start_date=None, latest_start_date=None, oldest_close_date=None, latest_close_date=None):

		return count_closed_workflow_within_time_range(**workflow_filter_dict)


		def _build_time_filter_dict(oldest_start_date=None, latest_start_date=None, oldest_close_date=None, latest_close_date=None):

		assert time_filter_dict['closeTimeFilter'] == {'oldestDate': start_date, 'latestDate': end_date}


		def test_count_open_workflow_executions(workflow_config, workflow_client, boto_client, oldest_start_date, latest_start_date):

		_assert_boto_method_called_with(boto_client.count_open_workflow_executions, workflow_config.domain, **kwargs)


		def _assert_count_closed_workflow_executions_called_with(boto_client, workflow_config, **kwargs):

Allow count worfklow executions by different filters #18

Allow count worfklow executions by different filters #18

Conversation

phishermaan commented Nov 15, 2016

coveralls commented Nov 15, 2016 • edited Loading

movefast1 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

phishermaan Dec 8, 2016 • edited Loading

Choose a reason for hiding this comment

quantsini Dec 9, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

movefast1 Nov 29, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coveralls commented Dec 1, 2016 • edited Loading

coveralls commented Dec 1, 2016 • edited Loading

coveralls commented Dec 1, 2016 • edited Loading

movefast1 Dec 2, 2016 • 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

coveralls commented Dec 7, 2016 • edited Loading

Choose a reason for hiding this comment

coveralls commented Dec 9, 2016 • edited Loading

coveralls commented Dec 9, 2016 • edited Loading

coveralls commented Dec 9, 2016 • edited Loading

coveralls commented Dec 9, 2016 • edited Loading

coveralls commented Dec 9, 2016 • edited Loading

Choose a reason for hiding this comment

coveralls commented Dec 12, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coveralls commented Dec 12, 2016

coveralls commented Dec 13, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coveralls commented Dec 13, 2016 • edited Loading

coveralls commented Dec 13, 2016 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coveralls commented Dec 13, 2016 • edited Loading

coveralls commented Dec 14, 2016 • edited Loading

quantsini commented Dec 14, 2016

coveralls commented Nov 15, 2016 •

edited

Loading

phishermaan Dec 8, 2016 •

edited

Loading

quantsini Dec 9, 2016 •

edited

Loading

movefast1 Nov 29, 2016 •

edited

Loading

coveralls commented Dec 1, 2016 •

edited

Loading

coveralls commented Dec 1, 2016 •

edited

Loading

coveralls commented Dec 1, 2016 •

edited

Loading

movefast1 Dec 2, 2016 •

edited

Loading

coveralls commented Dec 7, 2016 •

edited

Loading

coveralls commented Dec 9, 2016 •

edited

Loading

coveralls commented Dec 9, 2016 •

edited

Loading

coveralls commented Dec 9, 2016 •

edited

Loading

coveralls commented Dec 9, 2016 •

edited

Loading

coveralls commented Dec 9, 2016 •

edited

Loading

coveralls commented Dec 12, 2016 •

edited

Loading

coveralls commented Dec 13, 2016 •

edited

Loading

coveralls commented Dec 13, 2016 •

edited

Loading

coveralls commented Dec 13, 2016 •

edited

Loading

coveralls commented Dec 13, 2016 •

edited

Loading

coveralls commented Dec 14, 2016 •

edited

Loading