Christian Woltering

@FObermaier I'll do a pull request.

@ondrasvoboda thanks, I'll add it to the pull request.

Sure, this should be documented. The `writenodes` method is responsible for numbering the nodes. The original Triangle code uses the `nonodewritten` field of the `behavior ` struct to check if...

Additionally, numbering the nodes is done by using `setvertexmark`. So, if this field of the vertex data structure is actually used as intended (as an integer boundary marker), there should...

A while back, I looked at some indirect approaches to quad meshing (indirect = converting a triangular mesh to quads): * [gmsh: frontal delaunay quad](https://duckduckgo.com/?q=frontal+delaunay+quad): advancing front algorithm * [gmsh:...

Triangle ignores duplicate input vertices. This will cause a problem, if you reference such a vertex in the segmentlist. The .NET version is doing the same thing. It doesn't crash...

It's actually not what I thought. The original Triangle executable works fine with the following poly file: ``` 8 2 0 1 1 0 0 1 2 10 0 1...

Well, it's not really a fix, just a workaround. I think to fix the issue, you should look where `UNDEADVERTEX` gets set (by `setvertextype`), and then add ``` if (b->poly)...

## Creating a mesh Using the new API, you'd first create a context: ```c context* triangle_context_create(); ``` The context contains pointers to `behavior` and `mesh` objects. You can pass in...

Please read the original documentation at http://www.cs.cmu.edu/~quake/triangle.html Here are some hints: * Triangulation of point sets will always include the convex hull. * Adding the c switch will return boundary...