UpdateExtent propagation bug for vtkProbeFilter
This issue was created automatically from an original Mantis Issue. Further discussion may take place here.
There seems to be a bug in the way that vtkProbeFilter handles extents. This bug might actually in the VTK pipeline executive itself, rather than in the vtkProbeFilter. It manifests itself as described below.
The vtkProbeFilter produces an output by taking the geometry of the Input, and coloring each point with an interpolated scalar value from the Source. Therefore, the Output has the same data object type as the Input, while the Source can be any vtkDataSet type.
An error can be seen if you add "extent[1] += 10" at line 51 of TestProbeFilterImageInput.py (see attached modified file):
ERROR: In Common/ExecutionModel/vtkStreamingDemandDrivenPipeline.cxx, line 857 vtkCompositeDataPipeline (0x7fe6e2753480): The update extent specified in the information for output port 0 on algorithm vtkRTAnalyticSource(0x7fe6e25784d0) is 0 57 0 47 0 47, which is outside the whole extent 0 47 0 47 0 47.
The pipeline structure that gives rise to this error is as follows:
vtkRTAnalyticSource [Source] -v vtkDelaunay3D [Input] -> vtkProbeFilter (probe1)
probe1 produces an unstructured grid
probe1 [Source] -----------v vtkImageData [Input] -> vtkProbeFilter (probe2)
probe2 produces an image
Since probe1 is producing an unstructured grid, it should not care about the extent of the output produced by probe2. But somehow the UpdateExtent request is being propagating along the pipeline via the unstructured grid connection, resulting in a bogus error. A 3D UpdateExtent request should not propagate upstream for data that isn't vtkImageData or some other 3D structured data object type.