GitLab

COMP: The asm-language converted .cxx sources are not classes and should be in Kit_EXTRA_SRCS, not KitOpenGL_SRCS.

ENH: Change pipeline usage pattern from old style a.SetInput(b.GetOutput()) to a.SetInputConnection(b.GetOutputPort())...  (Now that Will fixed the LabeledDataMapper.)

ENH: Added cygcheck knowledge for --ldd option on Cygwin.  Added error message for --ldd option when no tool is available but the option was still requested.

BUG: VTK_LIBRARY_DIRS in VTKConfig.cmake must point to PREFIX/bin for shared library builds on Cygwin.  Cygwin programs link directly to the .dll files.

BUG: Fixed install test install root to not make installed executables from VS project builds think they are running in the build tree.

BUG: Fixed dependencies on Tk so that VTK packages that need it make sure it is loaded and VTK packages that do not need it do not care.  This should finish enabling tests to run when VTK_USE_TK is off.

BUG: Correct non-VTK_DATA_ROOT version of the quadric decimation test that was activated earlier today by the CMakeLists.txt change; now the test actually runs on machines that do not have VTK_DATA_ROOT defined.

BUG: a nasty yet fun bug/leak. The exit method of vtkWin32RenderWindowInteractor is called when the user exists interactively (such as pressing 'e' or 'q'). An extra reference is added to the callback command used to process it by the vtkSubjectHelper::InvokeEvent method. If the callback exits the application immediately in its Execute() method, this reference is never removed and the object leaks. Using 'after idle' allows the stack to unwind far enough. This is similar to Win32 PostQuitMessage() logic. At the same time, we move the "DeleteAllObjects" from vtktesting.tcl to vtk.tcl so that all objects are destroyed automatically for the user, as a convenience.

ENH: make it more clear which class is being leaked (we found out that vtkCommand reports itself as 'vtkCommand or subclass')

BUG: Reverting yesterday's automatic SetInput --> SetInputConnection change where it should not have been done... (actors and widgets only have SetInput and they should still use SetInput)

The widget's BuildRepresentation() serves to take its ivar vtkPlaneSource and generate a 2D polyline outline of its bounds.

This should only be called after UpdatePlane since within that method, RestrictPlaneToVolume is enforced, the vtkPlaneSource could be modified, and hence, the outline rep will be inconsistent.

This manifests itself when interacting as though the plane outline "bounces" off the widget's resliced image and then bounces back as one tries to drag past the input image data bounds.

BUG: Reverting yesterday's automatic SetInput --> SetInputConnection change where it should not have been done... (actors and widgets only have SetInput and they should still use SetInput)

COMP: Fix broken continuous dashboard... Didn't catch it on test build because file only included when VTK_USE_MPI is ON.

ENH: Updated wrapping commands to use new macros instead of loaded commands.  Also updated configuration of python wrappers to use separate module and shared library like the rest of VTK.

ENH: Updated wrapping commands to use new macros instead of CMake C++-implemented commands.  Also updated configuration of python wrappers to use separate module and shared library like the rest of VTK.