[Freeipa-devel] [patch 0010] Python list comprehension leak breaking the test execution

[Milan Kubík]

On 08/11/2015 12:38 PM, Jan Cholasta wrote:
> On 11.8.2015 12:33, Milan Kubík wrote:
>> On 08/11/2015 12:33 PM, Milan Kubík wrote:
>>> On 08/11/2015 10:03 AM, Milan Kubík wrote:
>>>> On 08/11/2015 09:57 AM, Martin Babinsky wrote:
>>>>> On 08/10/2015 05:43 PM, Milan Kubík wrote:
>>>>>> Hi all,
>>>>>>
>>>>>> this patch fixes problem described in the ticket [1]
>>>>>> that caused the test run to fail completely at every other or so 
>>>>>> run.
>>>>>> I took the liberty to fix most of the pep8 issues while I was at it.
>>>>>>
>>>>>> Thanks to Jan Cholasta for help with identifying this one.
>>>>>>
>>>>>> [1]: https://fedorahosted.org/freeipa/ticket/5192
>>>>>>
>>>>>> Cheers,
>>>>>> Milan
>>>>>>
>>>>>>
>>>>> Hi Milan,
>>>>>
>>>>> I did not follow the rest of the conversation (probably should), but
>>>>> you can also workaround this python2 "feature" by using:
>>>>>
>>>>> """
>>>>> assert list((t.id, t.options) for t in doc.getKeyPackages()) == \
>>>>> """
>>>>>
>>>>> In this case a list is constructed by calling a generator expression
>>>>> which does not leak variables to the outside scope.
>>>>>
>>>>> It may or may not help you but I think it's a trick worth trying. As
>>>>> a bonus it should be py2/3 compatible.
>>>>>
>>>> Thanks for the suggestion, Martin.
>>>>
>>>> I will try that. Though, I suspect there may be other labels holding
>>>> references to nss objects. I think I will combine this with Jan's
>>>> suggestion.
>>>>
>>>> I will send an update once I'll have the results from a test run.
>>>>
>>>> Milan
>>>>
>>> After running the test with the decorator and list comprehensions
>>> replaced by a list initialization from a generator expression, it
>>> fails on this [1].
>>>
>>> Does anyone know what is this teardown method supposed to do?
>>>
>>> Milan
>> [1]: https://paste.fedoraproject.org/253792/14392890/
>>
>
> The error message means that the NSS database directory does not 
> exist, it is not readable, or the NSS database is not initialized (as 
> in certutil -N).
>
Thanks. My main question is why is it called in the teardown method (btw 
remnant from nose times it seems, but pytest executes it after every 
test case)

Milan