The following points cause a panic with "Empty point set", which is clearly a bit wrong because I'm counting six points there. I have a suspicion that if the circumscribing sphere part of the algorithm used https://crates.io/crates/robust, that this might not happen.
use nalgebra::{Point3, Vector3};
use std::collections::VecDeque;
type Ball = miniball::Ball<f32, nalgebra::Const<3>>;
#[test]
fn test_miniball_failure_case() {
let mut buffer = [
[0.8155192, -0.5787301, 0.0],
[0.7634287, -0.6458921, 0.0],
[0.7634287, -0.6458921, 8.751844e-16],
[0.8993817, -0.43716416, -1.7844731e-15],
[0.8155192, -0.5787301, -2.9261717e-15],
[0.8993817, -0.43716416, 0.0],
]
.into_iter()
.map(|v| Vector3::<f32>::from_column_slice(&v[..]))
.map(From::from)
.collect::<VecDeque<Point3<f32>>>();
let _ = Ball::enclosing_points(&mut buffer);
}
The following points cause a panic with "Empty point set", which is clearly a bit wrong because I'm counting six points there. I have a suspicion that if the circumscribing sphere part of the algorithm used https://crates.io/crates/robust, that this might not happen.