[vtkusers] Possible 64 bit segfault

Jordi Gutiérrez Hermoso jordigh at gmail.com
Sat May 10 23:11:31 EDT 2008


On 10/05/2008, Mike Jackson <imikejackson at gmail.com> wrote:
> You need to run ccmake from the build directory, NOT the source directory.

I'm doing an in-source build, so the source and build directories are
the same, right? Anyways, I figured it out. Turns out that I had to
run cmake before I could tweak with ccmake.

Now it turns out that Octaviz won't build against cvs VTK, although it
builds against VTK 5.0. :-/

If I'm going to be looking for 64bit segfaults, I think I need to
check against the cvs version in case it's already been fixed there.
Is there a dedicated developers list, or is every VTK user also a
developer?

I'll try asking here.

Octaviz is an Octave wrapper of VTK classes that also provides a few
Octave scripts for a familiar and comfortable interface. The wrapping
is done by a file that is, if the copyright header speaks forsoothly,
a modification of VTK code for wrapping other languages.

The problem is with one of the files produced by this parser, so I'm
guessing that if the bug is with Octaviz, it must be in the parser.

Here's the problem (and 64bit problems are starting to show up here):
the parser produced some code that assumed some data type was an int
and an int*. The parsed code then tried to call
vtkIdTypeArray::GetTupleValue(vtkIdType, vtkIdType*). Now, as far as I
can tell, on my 64bit system vtkIdType is typedeffed to long long.
It's not possible to cast an int* to long long*, hence the compile
error.

I'm guessing this means that I need to fix the parser to produce
vtkIdType in certain places instead of int?

Thanks,
- Jordi G. H.



More information about the vtkusers mailing list