vtkDataObject::DATA_EXTENT_TYPE() needs to be only set on a data
object's information object. This is usually done in the constructor
of the data object's class. Those that are implementing algorithms
do not have to worry about this key.

Algorithms should not use the executive to get the output information.
They should use the arguments to ProcessRequest instead. In this case,
vtkGenericDataObjectReader was setting (in RequestDataObject) the output
data object on the executive's output port. This caused problems when
the method was called with other information objects.

STYLE: Documentation fixes and improvements.

COMP: Add logic to define and use ${MPI_LIBRARIES} in TARGET_LINK_LIBRARIES otherwise ParallelIso has unresolved MPI symbols.

PERF: replacing SetPoint/GetPoint/etc calls with direct pointer manipulation to circumvent redundant array bounds checking

BUG:Another attempt to fix a clean rejection of an OpenGL implementation supporting FBO but not supporting the requested format.

BUG:Attempt to fix the case where FBO extension is supported by the requested FBO format is not as for SINTEF.LabMAC.OSX/VTKCvs_CMake262Carbon_gcc401_Qt443.

ENH: Before capturing the window, do two renders and disable reading the front buffer. This strategy should prevent the capture of windows that are covered. It is the same stratgegy used by vtkTesting to capture regression test images.

COMP: Trying to track down timer event failures on some systems. Edited the event log to remove some extraneous events.

BUG: set Skewness and Kurtosis to 0 when m2 is 0 to avoid divide by zero. Philippe please fix this to a more appropriate calculation ASAP.

BUG:Fixed direct call to Render on the vtkRenderer which is strictly forbidden as the vtkRenderer. Notably, vtkRenderer relies on values initialized by the vtkRenderWindow. In the case of Example Medical3.cxx, as Renderer() was called on the vtkRenderer before any internal initialization of the vtkRenderWindow it resulted in the use of uninitialized variable. Thanks valgrind for catching this.

BUG: Fix issue #10088 - call ScreenToClient on the position values that come into the WM_MOUSEWHEEL events on Windows. This ensures the render window interactor always returns values in the client window coordinate space from GetEventPosition.

BUG: We are supposed to do this scaling business when a scale mode is absent (ie the text actor does not manage it for us), not if its present.