Crash in voronoi.triangles? about d3-voronoi HOT 4 CLOSED

I noticed this also, one has to do .extent(null) before calling triangles()
See https://bl.ocks.org/Fil/6c028ce0d9b66d2796bce7f8166f32cd

from d3-voronoi.

When we add an extent, the Voronoi cells on the boundary of the rectangle have dangling halfedges. Removing those enables .triangles() to work as expected (and documented).

-         if (i < s0.index && i < s1.index && triangleArea(site, s0, s1) < 0) {
+         if (s0 && s1 && i < s0.index && i < s1.index && triangleArea(site, s0, s1) < 0) {

This is consistent with what .links() already returns

The documentation could clarify that, in the case of a finite .extent(), what is returned is a subset of the Delaunay triangulation:

- Since the Delaunay triangulation is computed as the dual of the Voronoi diagram, and the Voronoi diagram is clipped by the extent, the Delaunay triangulation is likewise affected by the extent.
+ Since the triangulation is computed as the dual of the Voronoi diagram, and the Voronoi diagram is clipped by the extent, a subset of the Delaunay triangulation is returned.

I have updated my block (https://bl.ocks.org/Fil/6c028ce0d9b66d2796bce7f8166f32cd) to demonstrate this. (Colors are ugly, but could be a basis for something that we link from the documentation.)

from d3-voronoi.

An alternative is #13 (claims to return the "full" Delaunay).

from d3-voronoi.

Including @adithyaphilip in the discussion.

from d3-voronoi.