[vtk-developers] Rendering bugs on Mac OSX when using vtk with gtk

[Valerie Coffman]

Dear VTK Developers,

I posted this to the vtkusers mailing list, but got no responses.  I 
suppose this list is more appropriate, so I'm hoping someone here is 
able to help.

After the latest Mac OSX update, I noticed a bug with rendering vtk data 
embedded in gtk windows.  If I try to render a vtkVolume and a vtkActor, 
the vtkVolume is not displayed and the vtkActor is displayed 
incorrectly.  For instance, a cone is missing its base.  This does not 
happen if the vtk window is not embedded in a gtk window.  It also 
doesn't happen when using the same versions of vtk and gtk on a under 
Linux with a Linux display.  It does happen if I ssh to a Linux box and 
display on the Mac.  (I was unable to check whether the bug appeared 
when running on a Mac but displaying in Linux as this resulted in a bus 
error.)

I am not using gtkglext.  I posted a while back about how to put vtk and 
gtk together without gtkglext or gtkglarea: 
http://www.vtk.org/pipermail/vtkusers/2008-April/094892.html (see the 
script Wrapping/Python/vtk/gtk/noglext.py in the source for vtk 5.4).  
In fact, this bug appeared earlier when using gtkglext and went away 
when I implemented the fix described in the above link.  Now it has 
resurfaced.

Here are the details of the latest Mac updates

2009-06-09 13:14:35 -0400: Installed "Digital Camera Raw Compatibility 
Update" (2.6)
2009-06-09 13:14:50 -0400: Installed "AirPort Utility Software Update 
2009-002" (5.4.2)
2009-06-09 13:15:07 -0400: Installed "iWork '09 Update 2" (9.0.2)
2009-06-09 13:15:49 -0400: Installed "iTunes" (8.2.0)
2009-06-09 13:15:53 -0400: Installed "iLife Support" (9.0.3)
2009-06-09 13:16:07 -0400: Installed "iDVD Update" (7.0.4)
2009-06-09 13:18:54 -0400: Installed "Mac OS X Update" (10.5.7)
2009-06-09 13:19:26 -0400: Installed "QuickTime" (7.6.2)

The relevant version info is:
VTK 5.0.4.3 (fink package vtk-py25) (I've also tried VTK 5.4.2 and the 
result is the same)
GTK 2.16.2-2 (also installed with fink)
X11 2.3.3.2 (from http://xquartz.macosforge.org/)
pygtk 2.12.1-5 (also from fink)
Python 2.5 (fink package number 1:2.5.4-1)

I've posted test scripts, data, and pictures here: 
http://www.ctcms.nist.gov/oof/renderingbug/

There is another discussion of this bug here: 
http://lists.macosforge.org/pipermail/xquartz-dev/2009-June/002367.html   

Another thing I noticed is that if I try to print out the 
vtkRenderWindow, the colormap is set to 0 for the script that uses gtk, 
but seems to be set to a real value for the script that doesn't use 
gtk.  I don't know if that is the expected behavior or not.

Any insights would be greatly appreciated!

Thanks
Valerie Coffman