Triangulation creates new points instead of reusing old ones

[JaminKoke]

I'm trying to map the output points of the triangulation to the input points, but that is not easy apparently!

A simple example I thought would work:

#[derive(Copy, Clone)]
struct PointLink {
    x: f32,
    y: f32,
    // Index to the original point in the vertex buffer that was used to create this
    original_index: u32,
}

impl FloatPointCompatible for PointLink {
    type Scalar = f32;

    fn x(&self) -> Self::Scalar {
        self.x
    }

    fn y(&self) -> Self::Scalar {
        self.y
    }

    // This gets called inside of a triangulation despite not needing to create new points?
    fn from_xy(x: Self::Scalar, y: Self::Scalar) -> Self {
        panic!("Should not be creating new points; Otherwise we don't know where the point is in the vertex array");
    }
}

My use case is that I have a 3d model I want to triangulate, and then map the triangulation to 3d. However the from_xy function gets called, meaning we loose all context as to what the original vertex is.

Is there a different way to accomplish what I'm trying to do, or is this something that needs to be fixed?

[NailxSharipov]

Hi!
This isn't supported yet: triangulation reconstructs points from their coordinates, so custom fields such as original_index are lost.

A possible implementation would use EdgeOverlay to simplify original geometry first. It can create a new points but it can be handled now.

The simple operations (like Delaunay) do not create new points but it can merge same vertices (currently we can not handle it).
We would need to add PointData support in iTriangle to carry those IDs through to the output.