[Paraview] Segfault on exit()/quit()/raise SystemExit in python script

[Sven Buijssen]

Hi,

When loading test data set can.ex2 into ParaView 3.7 (CVS as of today)
or ParaView 3.6.1 compiled myself from the official source tarball, not
altering any setting, saving a state file, loading this state file in a
python script and finally exiting this python script with either
a) quit()
b) raise SystemExit
c) exit()
a pvpython built with OpenGL segfaults. Script is attached.

Use "exit" instead and it does not segfault.
Don't load a state file, but directly load can.ex2 and it does not crash
(see N.B.)
Use official binary release of ParaView 3.6.1 and it does not segfault.

A pvpython built with Mesa 7.5.1 does not segfault, for either case.

Can anyone confirm this? Or am I simply triggering somehow a bug in my
GL driver?

I'll add the state file for completeness, but any state file should do.
gdb backtrace attached, too.

Sven


N.B.:
At first, I thought it might be related to
http://www.vtk.org/Bug/view.php?id=6645, but adding one of the exit
commands a)-c) to the script t6.py attached there, does not cause a
segfault.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: segfault-on-exit.py
URL: <http://www.paraview.org/pipermail/paraview/attachments/20091009/e1e9356e/attachment-0002.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: can.ex2.pvsm
URL: <http://www.paraview.org/pipermail/paraview/attachments/20091009/e1e9356e/attachment-0001.asc>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: bt-full.txt
URL: <http://www.paraview.org/pipermail/paraview/attachments/20091009/e1e9356e/attachment-0003.txt>