(0001356)
Luis Ibanez (manager)
2004-08-06 23:20

The destructor has now been declared virtual
(it wasn't despite the fact of having other virtual functions).

No the GeometryMap is visited and all its elements are erased with "delete".

(0001360)
Luis Ibanez (manager)
2004-08-07 19:33

After fixing the GeometryMap container,
We found that there are additional memory leaks.

In order to see the full extent of the memory leak
the following command may be used with Valgrind

valgrind -v --leak-check=yes --leak-resolution=high ./SimplexMeshTest

(0001929)
Lydia Ng (developer)
2005-01-14 13:40

triaged 1/14/2005

(0009073)
Luis Ibanez (manager)
2007-09-18 11:19

Urgent to be addressed during WA10.
Setup Valgrind builds and review memory leaks reports.

(0009074)
Luis Ibanez (manager)
2007-09-18 11:19

It is urgent to address the review of memory leaks.

(0022655)
kentwilliams (developer)
2010-10-21 17:16

I built on a machine with Valgrind and ran all the SimplexMesh tests under valgrind. Only itksimplexMeshToTriangleMeshFilterTest reported leaks. I'm investigating fixing this.

(0022656)
kentwilliams (developer)
2010-10-21 17:40

I am suspecting that the memory leaks reported in itkSimplexMeshToTriangleMeshFilterTest are false positives. They all have to do with the use of the member variable m_NewSimplexCellPointer, which is a member of itk::SimplexMeshAdaptTopologyFilter and itk::TriangleMeshToSimplexMeshFilter. In both cases, objects are allocated with new and assigned to this variable via its TakeOwnership method.

So either the AutoPointer class doesn't work or valgrind is confused.

(0022679)
kentwilliams (developer)
2010-10-25 15:11

The problem with the itkSimplexMeshToTriangleMeshFilterTest doesn't make any sense; if I add this:

originalTriangleMesh = 0;
simplexFilter = 0;

To the bottom of that test, the memory errors go away.

(0022681)
kentwilliams (developer)
2010-10-25 15:39

The particular bug reported is no longer reproducable. I logged a new bug http://public.kitware.com/Bug/view.php?id=11360 [^]

for the problem I found in a different, related test.

Issue History
Date Modified	Username	Field	Change
2007-09-18 11:19	Luis Ibanez	Note Added: 0009073
2007-09-18 11:19	Luis Ibanez	Note Added: 0009074
2007-09-18 11:19	Luis Ibanez	Priority	low => urgent
2008-05-20 11:03	Wilbur C. K. Wong	Assigned To	Luis Ibanez => Wilbur C. K. Wong
2010-10-21 13:53	Hans Johnson	Assigned To	Wilbur C. K. Wong => kentwilliams
2010-10-21 17:16	kentwilliams	Note Added: 0022655
2010-10-21 17:40	kentwilliams	Note Added: 0022656
2010-10-25 15:11	kentwilliams	Note Added: 0022679
2010-10-25 15:39	kentwilliams	Resolution Date	=> 2011-10-25
2010-10-25 15:39	kentwilliams	Sprint Status	=> backlog
2010-10-25 15:39	kentwilliams	Note Added: 0022681
2010-10-25 15:39	kentwilliams	Status	assigned => closed
2010-10-25 15:39	kentwilliams	Resolution	open => fixed
2010-10-25 15:39	kentwilliams	Product Version	=> ITK-4-A3

Notes
(0001356) Luis Ibanez (manager) 2004-08-06 23:20	The destructor has now been declared virtual (it wasn't despite the fact of having other virtual functions). No the GeometryMap is visited and all its elements are erased with "delete".

(0001360) Luis Ibanez (manager) 2004-08-07 19:33	After fixing the GeometryMap container, We found that there are additional memory leaks. In order to see the full extent of the memory leak the following command may be used with Valgrind valgrind -v --leak-check=yes --leak-resolution=high ./SimplexMeshTest

(0001929) Lydia Ng (developer) 2005-01-14 13:40	triaged 1/14/2005

(0009073) Luis Ibanez (manager) 2007-09-18 11:19	Urgent to be addressed during WA10. Setup Valgrind builds and review memory leaks reports.

(0009074) Luis Ibanez (manager) 2007-09-18 11:19	It is urgent to address the review of memory leaks.

(0022655) kentwilliams (developer) 2010-10-21 17:16	I built on a machine with Valgrind and ran all the SimplexMesh tests under valgrind. Only itksimplexMeshToTriangleMeshFilterTest reported leaks. I'm investigating fixing this.

(0022656) kentwilliams (developer) 2010-10-21 17:40	I am suspecting that the memory leaks reported in itkSimplexMeshToTriangleMeshFilterTest are false positives. They all have to do with the use of the member variable m_NewSimplexCellPointer, which is a member of itk::SimplexMeshAdaptTopologyFilter and itk::TriangleMeshToSimplexMeshFilter. In both cases, objects are allocated with new and assigned to this variable via its TakeOwnership method. So either the AutoPointer class doesn't work or valgrind is confused.

(0022679) kentwilliams (developer) 2010-10-25 15:11	The problem with the itkSimplexMeshToTriangleMeshFilterTest doesn't make any sense; if I add this: originalTriangleMesh = 0; simplexFilter = 0; To the bottom of that test, the memory errors go away.

(0022681) kentwilliams (developer) 2010-10-25 15:39	The particular bug reported is no longer reproducable. I logged a new bug http://public.kitware.com/Bug/view.php?id=11360 [^] for the problem I found in a different, related test.

Relationships