[vtk-developers] VTK 5.2

[David Cole]

Actually (and I know this is a really very late reply... sorry) this is not
the sort of response I was looking for. However....... if you, or anybody,
have time to make fixes for these issues in CVS HEAD, save the output of
"cvs commit" when you make the changes and forward it to me.
I am quite happy to merge any bug fix type changes from CVS HEAD into the
VTK-5-2 branch. It's very easy to do if the fix is isolated and independent
and if I have the cvs commit output to start with....


Thanks,
David Cole


On Tue, Jun 17, 2008 at 3:06 PM, John Platt <jcplatt at dsl.pipex.com> wrote:

>  Hi,
>
>
>
> A couple of crashes I had in CVS version 14th Jan 2008.
>
>
>
> If you try to render the prop assembly tree vtkActor > vtkPropAssembly >
> vtkPropAssembly (from leaf to root), vtkProp3D::PokeMatrix() crashes because
> this->CachedProp3D is NULL. The tree vtkActor > vtkPropAssembly renders OK.
>
> The same problem was reported some years ago
> http://public.kitware.com/pipermail/vtkusers/2002-April/060530.html
>
>
>
> In vtkPropAssembly::BuildPaths(), the line
>
>
>
>     path->AddNode(prop,NULL);
>
>
>
> will always add a null matrix even if the prop is a vtkProp3D. I have
> changed this to
>
>
>
>     path->AddNode( prop, prop->GetMatrix() );
>
>
>
> and my particular problem is resolved.
>
>
>
>
>
> Clipping an unstructured grid containing (theoretically) more than 8
> quadratic hex's with a plane may fail with an access violation copying the
> cell data. In practice, at least 24 cells are needed on Win XP and probably
> many more for nix.
>
>
>
> I think the problem is in vtkQuadraticHexahedron::Clip() with the call
>
>
>
>             this->Hex->Clip( , , , , , , this->CellData, cellId, , );
>
>
>
> this->CellData stores 8 identical copies of the quadratic hex's cell data
> for each subdivided linear hex but cellId has a maximum value of the number
> of cells in the grid – 1. The access violation occurs in vtkTetra::Clip() in
> the calls outCD->CopyData() where an attempt is made to copy cell data for
> cellId.
>
>
>
> I think the call should be
>
>
>
>             this->Hex->Clip( , , , , , , this->CellData, i, , );
>
>
>
> where 'i' is the index of the linear hex.
>
>
>
> BUT …
>
>
>
> I don't understand why it is necessary to copy the cell data in the first
> place. I would have thought that all subdivided cells have the same cell
> data as the parent. In this case, dispense with this->CellData and just use
>
>
>
>             this->Hex->Clip( , , , , , , inCd, cellId, , );
>
>
>
> A similar problem may also exist in vtkQuadraticHexahedron::Contour().
>
>
>
> My apologies if you were not soliciting this type of response J.
>
>
>
> John.
>
>
>
>
>
>
>
>
>
>
>
> -----Original Message-----
> *From:* vtk-developers-bounces at vtk.org [mailto:
> vtk-developers-bounces at vtk.org] *On Behalf Of *David Cole
> *Sent:* 17 June 2008 14:30
> *To:* VTK Developers
> *Subject:* [vtk-developers] VTK 5.2
>
>
>
> Hello VTK developers,
>
> Yesterday I merged in a handful of changes to the VTK-5-2 branch. Is there
> anything else anybody knows about that should be merged to the VTK-5-2
> branch before we build a 5.2.0 release...? Any changes made to CVS HEAD
> since April 28 when the branch was created that should be merged over?
>
> Let me know in the next couple of days if there is anything else.
>
> Thanks,
> David Cole
> Kitware, Inc.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20080721/fb2227b9/attachment.html>