Dual Mesh and Clip Vertices nodes failing to work together

[Marcus-Richmond]

Problem statement

As part of a node setup, I would like to use the Dual Mesh node connected to the Clip Vertices node, like this:

The following error is from the Debug print node: [ERROR] sverchok.utils.sv_bmesh_utils:541 : faces.new(...): found the same (BMVert) used multiple times

After looking here: #3989, I tried using the following code and mesh clean node, but without any luck:

exec node

mesh clean

(I tried the various options with the mesh clean node, but didn't resolve the error)

The Clip Vertices node still works without the Dual Mesh node however:

Seems like without edges being input into the Clip Vertices node, this node returns this error. I tried another setup where the Clip Vertices node was only being passed verts/polygons and no edges:

Again, looking at #3989, I see the discussion is about duplicate faces, but none of these node setups appear to be creating duplicate faces.

Sverchok version

Sverchok 1.1.0 with Blender 3.1.2

[Durman]

This is certainly a bug because the example below should work according to the node documentation.

[Durman]

The problem is that this line is not called when edges are not provided.

sverchok/utils/sv_bmesh_utils.py

Line 108 in 6133967

bm.edges.index_update()

[zeffii]

weirdly enough that bm.edges.index_update() seems to have been always dependent on the presence of edges input, and somehow it now bugs out. (earliest version i can find https://github.com/nortikin/sverchok/blob/df942c23e73d6e093cf1d2306e3c10ee6e5bbecd/utils/sv_bmesh_utils.py)

[zeffii]

also, kind of awkward that the cross section node outputs "faces" from an "edges" socket (Seen https://user-images.githubusercontent.com/59568811/175453632-39e0c2d2-1c59-4240-981c-7a2b2169a11d.png )

[zeffii]

the error is emanating from truncate_vertices

  File "C:\Users\zeffi\Desktop\scripts\addons_contrib\sverchok\utils\sv_bmesh_utils.py", line 541, in truncate_vertices
    new_bm_add_face(new_face)
ValueError: faces.new(...): found the same (BMVert) used multiple times

[zeffii]

maybe we dedent bm.edges.index_update() in bmesh_from_pydata ? @Durman

[Durman]

I thought about adding extra property to the function because calling index_update method might be expensive (not sure how much)

def bmesh_from_pydata(..., edge_update=False):
    ...
    if has_element(edges) or edge_update:
        bm.edges.index_update()
    ...

It should not impact on existing usage of the function, especially if to catch the has_element function result.

[zeffii]

i thought about that too, for the same reasons, but i doubt the computational penalty is dramatic. i'll test and change if necessary

[zeffii]

a uvsphere 300*300 = 89700 is number of faces,

always index edges

89700
bmesh_from_pydata: 177.67 ms

index edges on demand

index_edges=True (which i think is a more descriptive name, index is also a verb)

89700
bmesh_from_pydata: 109.94 ms

[zeffii]

indexing edges is now an explicit switch in the bmesh_from_pydata function call. @Durman