[Insight-developers] New valgrind defect in itkBSplineSyNImageRegistrationTest

[Nicholas Tustison]

My guess is that this patch will probably fix the current
three valgrind errors.

http://review.source.kitware.com/#/c/4833/

Basically I initialized the size of the metric derivative
outside the call of the metric.

Nick


On Mar 26, 2012, at 10:17 AM, Nicholas Tustison wrote:

> Hi Michael and Brian,
> 
> If you take a look at the valgrind error that Brad linked
> to, it's basically the same as the two other registration
> method valgrind errors where we perform optimization 
> within the method class itself because they have to be 
> handled individually.  What's interesting is that the only 
> other similar class where we don't get a valgrind error is
> the itkTimeVaryingBSplineVelocityFieldImageRegistrationMethod
> class but I just looked and there was a bug in which we
> were calling the metric GetValueAndDerivative() function
> twice.  I fixed that here
> 
> http://review.source.kitware.com/#/c/4823/
> 
> but I bet you that by fixing it we'll add another valgrind 
> error.  
> 
> What I can't figure out is why.  For standard registration
> in which we're using the base class, 
> ImageToImageRegistrationMethod, the optimizer calls
> the GetValueAndDerivative() function and there's no
> valgrind error.  What I'm wondering if is if the valgrind 
> checking sees the repeated resizing of the metric 
> derivative as a problem.
> 
> Nick
> 
> 
> 
> On Mar 26, 2012, at 9:20 AM, Bradley Lowekamp wrote:
> 
>> Hello,
>> 
>> I was just closely looking at the dashboard, and noticed that the valgrind number of defect increased to 3. This one looks like the new one.
>> 
>> http://open.cdash.org/viewDynamicAnalysisFile.php?id=2726550
>> 
>> 
>> If there other two defects are not going to be addressed and they are considered false positives, they really should be suppressed, so that new defects can more easily be seen.
>> 
>> Brad
>> 
>> ========================================================
>> Bradley Lowekamp  
>> Medical Science and Computing for
>> Office of High Performance Computing and Communications
>> National Library of Medicine 
>> blowekamp at mail.nih.gov
>> 
>> 
>> 
>> _______________________________________________
>> Powered by www.kitware.com
>> 
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>> 
>> Kitware offers ITK Training Courses, for more information visit:
>> http://kitware.com/products/protraining.php
>> 
>> Please keep messages on-topic and check the ITK FAQ at:
>> http://www.itk.org/Wiki/ITK_FAQ
>> 
>> Follow this link to subscribe/unsubscribe:
>> http://www.itk.org/mailman/listinfo/insight-developers
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-developers/attachments/20120326/51cb0a92/attachment.htm>