Fix AABB::from_points which relied on implementation details of AABB:…

…:new_empty (#171)

- [x] I agree to follow the project's [code of
conduct](https://github.com/georust/geo/blob/master/CODE_OF_CONDUCT.md).
- [x] I added an entry to `rstar/CHANGELOG.md` if knowledge of this
change could be valuable to users.
---

This method does not just need an empty AABB, but actually requires that
the lower/upper corners use max/min values of the point coordinates.

This change therefore open-codes this to keep these invariants and the
code that relies on them close together.

A changelog entry should not be necessary as we luckily did not yet
release the regression.

Closes #170

rstar/src/aabb.rs

@@ -1,6 +1,6 @@
 use crate::point::{max_inline, Point, PointExt};
 use crate::{Envelope, RTreeObject};
-use num_traits::{One, Zero};
+use num_traits::{Bounded, One, Zero};
 
 #[cfg(feature = "serde")]
 use serde::{Deserialize, Serialize};
@@ -70,16 +70,16 @@ where
         I: IntoIterator<Item = &'a P> + 'a,
         P: 'a,
     {
-        i.into_iter()
-            .fold(Self::new_empty(), |aabb, p| aabb.add_point(p))
-    }
-
-    /// Returns the AABB that contains `self` and another point.
-    fn add_point(&self, point: &P) -> Self {
-        AABB {
-            lower: self.lower.min_point(point),
-            upper: self.upper.max_point(point),
-        }
+        i.into_iter().fold(
+            Self {
+                lower: P::from_value(P::Scalar::max_value()),
+                upper: P::from_value(P::Scalar::min_value()),
+            },
+            |aabb, p| Self {
+                lower: aabb.lower.min_point(p),
+                upper: aabb.upper.max_point(p),
+            },
+        )
     }
 
     /// Returns the point within this AABB closest to a given point.
@@ -246,4 +246,10 @@ mod test {
         let corner = [a[0], b[1], a[2]];
         assert_eq!(aabb.min_max_dist_2(&p), corner.distance_2(&p));
     }
+
+    #[test]
+    fn test_from_points_issue_170_regression() {
+        let aabb = AABB::from_points(&[(3., 3., 3.), (4., 4., 4.)]);
+        assert_eq!(aabb, AABB::from_corners((3., 3., 3.), (4., 4., 4.)));
+    }
 }

rstar/src/envelope.rs

@@ -37,8 +37,8 @@ pub trait Envelope: Clone + PartialEq + ::core::fmt::Debug {
     ///
     /// # Notes
     /// - While euclidean distance will be the correct choice for most use cases, any distance metric
-    /// fulfilling the [usual axioms](https://en.wikipedia.org/wiki/Metric_space)
-    /// can be used when implementing this method
+    ///   fulfilling the [usual axioms](https://en.wikipedia.org/wiki/Metric_space)
+    ///   can be used when implementing this method
     /// - Implementers **must** ensure that the distance metric used matches that of [crate::PointDistance::distance_2]
     fn distance_2(&self, point: &Self::Point) -> <Self::Point as Point>::Scalar;
 

rstar/src/object.rs

@@ -149,8 +149,8 @@ pub trait PointDistance: RTreeObject {
     ///
     /// # Notes
     /// - While euclidean distance will be the correct choice for most use cases, any distance metric
-    /// fulfilling the [usual axioms](https://en.wikipedia.org/wiki/Metric_space)
-    /// can be used when implementing this method
+    ///   fulfilling the [usual axioms](https://en.wikipedia.org/wiki/Metric_space)
+    ///   can be used when implementing this method
     /// - Implementers **must** ensure that the distance metric used matches that of [crate::Envelope::distance_2]
     fn distance_2(
         &self,

rstar/src/rtree.rs

@@ -115,7 +115,7 @@ where
 /// # Type Parameters
 /// * `T`: The type of objects stored in the r-tree.
 /// * `Params`: Compile time parameters that change the r-tree's internal layout. Refer to the
-/// [RTreeParams] trait for more information.
+///   [RTreeParams] trait for more information.
 ///
 /// # Defining methods generic over r-trees
 /// If a library defines a method that should be generic over the r-tree type signature, make