[vtk-developers] corrected-patch: Cell Tree Locator Update

Wed Aug 10 06:10:23 EDT 2011

This patch is the correct one. My apologies.

JB

From: vtk-developers-bounces at vtk.org [mailto:vtk-developers-bounces at vtk.org] On Behalf Of Biddiscombe, John A.
Sent: 10 August 2011 09:50
To: Tharindu De Silva; Andy Bauer; VTK Developers
Subject: Re: [vtk-developers] Cell Tree Locator Update

The virtual call in the destructor was my fault. Sorry about that. I am aware of the badness of virtual calls in constructors and destructors. I think it's safe in this case, because it just frees some pointers and sets them to null, the base class does the same, but doesn't know about one pointer, so worst case scenario is a memory leak (but it's hard to see this happening with your change).

The class does not honour the Max cells per node variable, it declares it's own one and ignore the one inherited from abstract cell locator. the patch attached fixes that and also cleans up a doc entry which was nonsense.

JB
Apologies I can't use Gerrit due to inability to get an openID

From: Tharindu De Silva [mailto:tsameera1 at gmail.com]
Sent: 09 August 2011 23:44
To: Andy Bauer; VTK Developers; Biddiscombe, John A.
Subject: Re: [vtk-developers] Cell Tree Locator Update

On Tue, Aug 9, 2011 at 5:43 PM, Tharindu De Silva <tsameera1 at gmail.com<mailto:tsameera1 at gmail.com>> wrote:
Hi Andy,

          Thank you for the comments.  Sorry I didn't notice the trailing white spaces.

          By 'example' i meant the test,  I will go through the wiki page and try to make an example.

Thanks,

Tharindu
On Tue, Aug 9, 2011 at 4:44 PM, Andy Bauer <andy.bauer at kitware.com<mailto:andy.bauer at kitware.com>> wrote:
Hi Tharindu,

I cleaned up the code a bit doing things like getting rid of trailing whitespace (gerrit will highlight this as red to warn about it), bracketed some if/else conditions, and added a std::vector to take care of some dynamic memory allocation.  I also got rid of the compile warnings in the test.  These changes are at:
http://review.source.kitware.com/#change,2418

One thing to note is that I made it more apparent that a virtual function was getting called in the destructor.  In general this is bad practice (see http://www.artima.com/cppsource/nevercall.html).

I was able to run the test successfully.  In your previous email you mentioned an example.  Did you mean test?  I couldn't find anything called CellTreeLocator.cxx except for the test.  If you wanted to add in a different example, please check out: http://www.vtk.org/Wiki/VTK/Examples.  This is not required but I'd encourage you to do this to help other people better understand and utilize your code.

Andy

On Tue, Aug 9, 2011 at 11:23 AM, Tharindu De Silva <tsameera1 at gmail.com<mailto:tsameera1 at gmail.com>> wrote:
Hi,

  I revert the renaming of internal variables to their original names and made a note on the code that they were derived from avtCellLocatorBIH class in VisIT, so that the people familiar with that code can make improvements.  The current code can be found in http://review.source.kitware.com/#change,2413.

 There is an example named CellTreeLocator.cxx at Filtering/Examples/Cxx.  Please let me know if you encounter any bugs in running this.

Thank you.

On Tue, Aug 2, 2011 at 5:08 PM, Biddiscombe, John A. <biddisco at cscs.ch<mailto:biddisco at cscs.ch>> wrote:
I'm not sure that renaming all the internal variables from their original names was a good idea.
The original code came from the visit avtXXX class and there's a strong possibility that the visit developers will make changes/improvements similar to the intersectBox type methods that I'm adding.

Renaming the internals serves no purpose other than to make it harder to sync our version with theirs.

I am aware that I have radical opinions about the dreadful kitware code style but for me this is an unnecessary barrier to future collaboration.

JB

From: Tharindu De Silva [mailto:tsameera1 at gmail.com<mailto:tsameera1 at gmail.com>]
Sent: 02 August 2011 18:18
To: Biddiscombe, John A.; Andy Bauer; VTK Developers
Cc: trshash at gmail.com<mailto:trshash at gmail.com>

Subject: Re: [vtk-developers] Cell Tree Locator Update

Hi,

         I did some formatting changes to the code according to vtk guidelines and added the patch given by John.  The code can be found in http://review.source.kitware.com/#change,2362.

Some formatting still needs to be done.

Helper classes, vtkCellTree and vtkCellTreeNode, are made public in the current version to make them accessible by sub-classing to test for additional functions implemented in vtkModifiedBSPTree. This can be made protected after all the functions are available in vtkCellTreeLocator.

The standard template library functions are used in this class that might need to be changed as well.

Please let me know if you come across any other bugs or if there are  improvements that can be made to this code.

Thank you,

Tharindu

On Fri, Jul 29, 2011 at 4:02 PM, Biddiscombe, John A. <biddisco at cscs.ch<mailto:biddisco at cscs.ch>> wrote:
I've fixed the crash caused by a numeric overflow. some compilers might not mind and not cause the segfault - I believe the intent was to allow the std:::numeric_limits max() to cause a large volume result which would exclude the bucket from consideration.

I'd added a check to stop it. Might be worth adding a few other tests as this class is new and we have little experience with  it. I'll add one myself I think.

patch attached

JB

_______________________________________________
Powered by www.kitware.com<http://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/20110810/413205a1/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Honour-the-MaxCellsPerNode-option-and-clean-up-docs.patch
Type: application/octet-stream
Size: 3122 bytes
Desc: 0001-Honour-the-MaxCellsPerNode-option-and-clean-up-docs.patch
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20110810/413205a1/attachment-0001.obj>