Nico Schlömer

An MWE is always useful. SIGSEGVs are suspicious though, pointing towards something in CGAL itself.

Yeah, tests run on my system and on github-actions. CGAL behaves slightly differently from machine to machine, so perhaps you'd just have to adapt the tolerances slightly.

Closing since this is probably not a CGAL bug, but feel free to continue posting here.

Alright, I also see the segfault on my machine. The full error: ``` CGAL::Polyhedron_incremental_builder_3:: lookup_halfedge(): input error: facet 1 shares a halfedge from vertex 0 to vertex 1 with facet...

Your initial mesh is so rough that it will be hard for any algorithm to achieve something meaningful.

No idea where `node_coords` should come from. There must be an old version of something at play, as I've removed `node_coords` from all of my packages a while ago.

This is principally impossible. Points clouds are not enough information to reconstruct a geometry. You can construct the convex hull of a point cloud with tools like [qhull](http://www.qhull.org/) though.

Ah, I get it. No, those are not interfaced in pygalmesh.

Sure, I'll be happy to review a PR.

You probably mean to use the union, not the intersection, but nevertheless CGAL has problems with unions which share a common surface. The best solution would probably be to submit...