[Clam-devel] We are in RED!

[Natanael Olaiz]

El 06/13/2008 03:14 PM, David García Garzón escribió:
> On Divendres 13 Juny 2008, David García Garzón wrote:
>   
>> On Divendres 13 Juny 2008, David García Garzón wrote:
>>     
>>> On Divendres 13 Juny 2008, JunJun wrote:
>>>       
>>>>>> I just commited a patch from Jun and it has turned testfarm in red.
>>>>>> It's a new unit test Jun added which seems correct and I am passing
>>>>>> it in green at my laptop and at my home pc, with no local
>>>>>> modifications. Can anyone reproduce the red or providing any clue?
>>>>>>             
>>>>> At my home PC I DO reproduce it. Have you double-checked everything?
>>>>> (installing the libs, etc).
>>>>> Should fix or revert...
>>>>>
>>>>> PMay it concern the tests that we temporarily comment out? The
>>>>> attachment is the Diffs of the SConsript.
>>>>>           
>>> Commited. But just the options part as the patch also  contained the
>>> disabled tests for fast compilation.
>>>       
>> Sorry, the reply levels confused me. i thought it was just a pacth to
>> apply. Anyway, the removed vc options are well applied.
>>
>> Answering your question, those test removed should not be a problem unless
>> previous tests do any memory corruption or race condition. I will comment
>> out the new test until i can talk with Pau.
>>
>> Jun, when ill do the commit, uncommit it locally since you'll need to run
>> it. The commit is just to lie testfarm.
>>     
>
> What a mess of sentences.
>
> I mean that i did a commit that comments out the failing test just for the 
> testfarm to be green and happy. Because we are still working on that, Jun, 
> you will need that test uncommented locally. Just uncomment the 
> CPPUNIT_TEST(testStoreOn) declaration.
>

I don't understand... what line are you talking about? #82 on
CLAM/test/UnitTests/VisualizationTests/DiscontinuousSegmentationTest?
Now is uncommented, and it pass the tests here (it didn't when you asked).

> Pau, please, could you do a valgrind on the UnitTests just to see whether 
> there is any memory corruption that could lead to such different results.
>
>
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> Clam-devel mailing list
> Clam-devel at llistes.projectes.lafarga.org
> https://llistes.projectes.lafarga.org/cgi-bin/mailman/listinfo/clam-devel
>