[vtkusers] vtkFeatureEdges "feature"

[Kirt Schaper]

I believe there is an error in vtkFeatureEdges.cxx which results
in incorrect counts for feature and manifold edges.  I am using
vtk-3.1 (compiled on a lintel box -- RedHat 6.1).  I created a
small non-manifold polydata model (one fin) and a small tcl
program to compute and report the various edge types (boundary,
manifold and non-manifold).  I've attached both files.

What appears to happen is that the manifold edge count is
double what it should be, and also includes boundary edges,
that is:
	trueManifold = manifold/2 - boundary
		where
	manifold = reported manifold edge count
	boundary = reported boundary edge count
My example program reports both counts.

I've played with various edge topologies and the above formula
always holds.

In looking at the vtkFeatureEdges.cxx code, I can see why the
manifold count is twice what it should be, there is no check
to insure that the edge has not been counted previously (as
there is for non-manifold edges) and each manifold edge is
(by definition) part of two cells.  I don't, however,  see
why boundary edges are included in the manifold count.  I believe
the problem also exists for feature edges, however, I have not
explored them.

----------------------------------------------------------
	Don't just do something, sit there!
----------------------------------------------------------
Kirt Schaper		| Bell: (612) 725-2000 x4791
PET Center (11P)	|  net: kirt at pet.med.va.gov
VA Medical Center	|  FAX: (612) 725-2068
MPLS, MN  55417		|  URL: http://pet.med.va.gov:8080