diff --git a/experiments/2025-03-18/src/extra/triangulate.rs b/experiments/2025-03-18/src/extra/triangulate.rs
index 4a667f876..c6ae670c7 100644
--- a/experiments/2025-03-18/src/extra/triangulate.rs
+++ b/experiments/2025-03-18/src/extra/triangulate.rs
@@ -41,34 +41,36 @@ pub fn triangulate(face: &Face) -> TriMesh {
 }
 
 fn points_from_half_edges(face: &Face) -> Vec<TriangulationPoint> {
-    let points_from_half_edges = face.half_edges.iter().map(|half_edge| {
-        let point_global = half_edge.start.point;
+    face.half_edges
+        .iter()
+        .map(|half_edge| {
+            let point_global = half_edge.start.point;
 
-        // Here, we project a 3D point (from the vertex) into the face's
-        // surface, creating a 2D point. Through the surface, this 2D point has
-        // a position in 3D space.
-        //
-        // But this position isn't necessarily going to be the same as the
-        // position of the original 3D point, due to numerical inaccuracy.
-        //
-        // This doesn't matter. Neither does the fact, that other faces might
-        // share the same vertices and project them into their own surfaces,
-        // creating more redundancy.
-        //
-        // The reason that it doesn't, is that we're using the projected 2D
-        // points _only_ for this local triangulation. Once that tells us how
-        // the different 3D points must connect, we use the original 3D points
-        // to build those triangles. We never convert the 2D points back into
-        // 3D.
-        let point_surface = face.surface.geometry.project_point(point_global);
+            // Here, we project a 3D point (from the vertex) into the face's
+            // surface, creating a 2D point. Through the surface, this 2D point
+            // has a position in 3D space.
+            //
+            // But this position isn't necessarily going to be the same as the
+            // position of the original 3D point, due to numerical inaccuracy.
+            //
+            // This doesn't matter. Neither does the fact, that other faces
+            // might share the same vertices and project them into their own
+            // surfaces, creating more redundancy.
+            //
+            // The reason that it doesn't, is that we're using the projected 2D
+            // points _only_ for this local triangulation. Once that tells us
+            // how the different 3D points must connect, we use the original 3D
+            // points to build those triangles. We never convert the 2D points
+            // back into 3D.
+            let point_surface =
+                face.surface.geometry.project_point(point_global);
 
-        TriangulationPoint {
-            point_surface,
-            point_global,
-        }
-    });
-
-    points_from_half_edges.collect()
+            TriangulationPoint {
+                point_surface,
+                point_global,
+            }
+        })
+        .collect()
 }
 
 fn triangles(points: &[TriangulationPoint]) -> Vec<[TriangulationPoint; 3]> {