Subject: Re: [geometry] Thoughts on adding 3D Delaunay triangulation and subsequent Voronoi diagram?
From: Barend Gehrels (barend_at_[hidden])
Date: 2017-07-04 17:36:04
Hi Christian,
Op 29-6-2017 om 20:23 schreef Christian Mazakas via Geometry:
> Yes, I am!
>
> That's a different, slightly tangential point. Boost.Polygon seems
> rather limited in scope and as a user of Boost, it was very
> unintuitive for me to seek out a specific polygon-based library that
> wasn't in Boost.Geometry.
It's a library with another approach, started independently, and they
came around the same time in Boost. However, there is Voronoi
functionality there:
http://www.boost.org/doc/libs/develop/libs/polygon/doc/voronoi_main.htm
Actually I don't know the current state of it, it was developed quite a
while ago during GSoC. But it, or parts of it, could be reused by us.
Besides that, both libraries are concept based.
>
> After all, the Delaunay triangulation and Voronoi tessellations are
> popular problems in the realm of computational geometry.
>
> Instead, I'd much rather see Geometry be expanded to include mesh
> generation. Or perhaps we can break this out into a separate
> Boost.Mesh library.
Both is possible (because it is concept based, code can be shared).
> I chose Boost.Geometry because it has a dedicated mailing list and
> should be full of people with a passion for computational geometry so
> I was hoping for a discussion about possible user-facing
> APIs/algorithmic design choices.
Sure, there is a core-team working hard on Boost.Geometry, doing many
commits per week. But that is not always visible on the list. We target
specific long-term goals, so (at least that is for me the case) I cannot
dive into a mesh or tesselation library right now.
Regards, Barend