[wxVTK] Redraw problems with wxVTK and wxMAC: perhaps a solution!

[Pierre F. <Pie...@so...>]

Hi!

First of all, thank you Mathieu for providing us with this wx interface 
to VTK!

Since I started using wxWidgets and wxVTK on a MAC, I realized (others 
might have experienced this also) that the display was sometimes not 
updated properly. I am using wxMAC 2.8.0, VTK 5.0.2 and MacOSX 10.4.8. I 
used wxVTK versions from 0.8 to 1.2, the problem remains in all versions.

The problem occurs when at least two wxVTK windows are embedded in a 
window (case of the SplitSample for example).

When one slowly resizes it, there is no problem, i.e. everything is 
repainted correctly. However, when one resizes the window very fast, 
(for instance when the window is put in full screen mode), then some 
parts of the wxVTK windows are not redrawn, even worse, the background 
is painted over the window. The only way to refresh the wxVTK windows is 
to slowly downsize the frame. I tried to overload the OnPaint event to 
force a call to Render() or simulate a resizing, nothing worked.

It was starting to drive me crazy since the redraw problem was really 
annoying when one wanted for instance, to put a view in full screen, and 
then back to normal...

I finally found a solution: one needs to call the "UpdageGLRegion()" 
method of the vtkCarbonRenderWindow (assuming that you're using the 
Carbon version of wxMAC, I haven't tried the other ports, but they seem 
to be less complete anyway) during the OnPaint() method of the 
wxVTKRenderWindowInteractor class. Here is what I put at the end of the 
OnPaint() method:

#ifdef __WXMAC__
  vtkCarbonRenderWindow* rwin = 
vtkCarbonRenderWindow::SafeDownCast(RenderWindow);
  if( rwin )
  {
  rwin->UpdateGLRegion();
  }
#endif

That might not be the most elegant way to bypass this redraw problem, 
but at least it works. And goodbye the ugly background painted over the 
VTK windows.

I hope this was helpful to some Mac users. Maybe the problem is 
completely related to my hardware / software version. In this case, I 
apologize for the inconvenience.

Pierre Fillard,
INRIA Sophia Antipolis, France.
http://www-sop.inria.fr/asclepios/software/MedINRIA.