[vtk-developers] Debug leaks, dashboards and local behavior

[David Cole]

I would fully support removing the environment variable test for deciding
whether to fail or not here.

There's no need to pass the tests when there are memory leaks. That only
encourages memory leaks, and enables people to ignore them.

Get rid of them (mem leaks that is). If you don't like it, then build with
mem leak detection turned off entirely.


Just my opinion, of course.
David C.


On Mon, Feb 13, 2012 at 11:46 AM, Marcus D. Hanwell <
marcus.hanwell at kitware.com> wrote:

> Hi,
>
> I would like to change the default behavior of debug leaks. Currently
> when you turn it on, if you run tests locally and there are leaks a
> test will pass. You will only spot the failure if you run the tests
> with -V and scan the output. See,
>
> 976: vtkDebugLeaks has detected LEAKS!
> 976: Class "vtkBlueObeliskData" has 1 instance still around.
> 976: Class "vtkUnsignedShortArray" has 2 instances still around.
> 976: Class "vtkFloatArray" has 10 instances still around.
> 976: Class "vtkStringArray" has 7 instances still around.
> 976:
> 1/1 Test #976: Chemistry-TestBallAndStick .......   Passed    0.37 sec
>
> This has caught me out on more than one occasion, including the recent
> merge of the chemistry branch into VTK. We will work on getting these
> fixed up, but I would like to make the test fail by default locally.
> Does anyone remember the reason for doing this, it seems to go back
> quite a way as it also looks for DART_TEST_FROM_DART.
>
> Exporting either of those variables will then cause the test to fail
> locally. If there is a good reason to do this, wouldn't it be better
> to pass if the user explicitly exported a variable,
> I_KNOW_I_TURNED_ON_DEBUG_LEAKS_BUT_I_STILL_WANT_TESTS_TO_PASS or
> similar?
>
> Thanks,
>
> Marcus
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Follow this link to subscribe/unsubscribe:
> http://www.vtk.org/mailman/listinfo/vtk-developers
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20120213/509bf21a/attachment.html>