[vtkusers] vtkExtractVOI + vtkStructuredPointsWriter breakage [FIXED]

[Charl P. Botha]

On Sun, 2004-02-29 at 23:18, Charl P. Botha wrote:
> Here's another segfault-causing script.  This time, the output of 
> vtkExtractVOI is sent to a vtkStructuredPointsWriter.  This simple 
> operation causes segmentation faults on Linux and Windows.
> 
> This time, the segfault is in vtkDataWriter::WriteFieldData() at line 
> 976.  strlen() is being requested on array->GetName().  array exists, 
> but ->GetName() returns 0x0.
> 
> It would seem that vtkExtractVOI is not correctly naming its output. 
> Even if this is the case, shouldn't subsequent filters be robust to 
> arrays not having names?

Okay, I've just committed a fix to the vtkExtractVOI class.  My log
message explains the problem:

revision 1.37

date: 2004/03/01 09:09:52;  author: cpbotha;  state: Exp;  lines: +1 -2
ERR: Fixed nasty bug.  vtkExtractVOI was calling AllocateScalars() *as
well as* CopyAllocate() (or PassThrough()) later on.  This meant that
subsequent filters saw 2 (two) arrays, one unnamed and empty, the other
representing the real data.  The unnamed array was able to crash
at least the vtkStructuredPointsWriter and any of the vtkXMLWriters.

Should I build some check into one of the ExtractVOI tests to prevent a
regression?

Thanks,
Charl

-- 
charl p. botha http://cpbotha.net/ http://visualisation.tudelft.nl/