Specialize range operations for better performance

src/internal/partial_solution.rs

@@ -501,7 +501,7 @@ impl<P: Package, VS: VersionSet> PackageAssignments<P, VS> {
         let idx = self
             .dated_derivations
             .as_slice()
-            .partition_point(|dd| dd.accumulated_intersection.intersection(start_term) != empty);
+            .partition_point(|dd| !dd.accumulated_intersection.is_disjoint(start_term));
         if let Some(dd) = self.dated_derivations.get(idx) {
             debug_assert_eq!(dd.accumulated_intersection.intersection(start_term), empty);
             return (Some(dd.cause), dd.global_index, dd.decision_level);

src/range.rs

@@ -328,8 +328,10 @@ fn within_bounds<V: PartialOrd>(v: &V, segment: &Interval<V>) -> Ordering {
     Ordering::Greater
 }
 
+/// A valid segment is one where at least one version fits between start and end
 fn valid_segment<T: PartialOrd>(start: &Bound<T>, end: &Bound<T>) -> bool {
     match (start, end) {
+        // Singleton interval are allowed
         (Included(s), Included(e)) => s <= e,
         (Included(s), Excluded(e)) => s < e,
         (Excluded(s), Included(e)) => s < e,
@@ -338,6 +340,47 @@ fn valid_segment<T: PartialOrd>(start: &Bound<T>, end: &Bound<T>) -> bool {
     }
 }
 
+/// The end of one interval is before the start of the next one, so they can't be concatenated
+/// into a single interval. The `union` method calling with both intervals and then the intervals
+/// switched. If either is true, the intervals are separate in the union and if both are false, they
+/// are merged.
+/// ```text
+/// True for these two:
+///  |----|
+///                |-----|
+///       ^ end    ^ start
+/// False for these two:
+///  |----|
+///     |-----|
+/// Here it depends: If they both exclude the position they share, there is a version in between
+/// them that blocks concatenation
+///  |----|
+///       |-----|
+/// ```
+fn end_before_start_with_gap<V: PartialOrd>(end: &Bound<V>, start: &Bound<V>) -> bool {
+    match (end, start) {
+        (_, Unbounded) => false,
+        (Unbounded, _) => false,
+        (Included(left), Included(right)) => left < right,
+        (Included(left), Excluded(right)) => left < right,
+        (Excluded(left), Included(right)) => left < right,
+        (Excluded(left), Excluded(right)) => left <= right,
+    }
+}
+
+/// The end of one interval is before the start of the next one, meaning they don't share any
+/// version.
+fn end_before_start<V: PartialOrd>(end: Bound<&V>, start: Bound<&V>) -> bool {
+    match (end, start) {
+        (_, Unbounded) => false,
+        (Unbounded, _) => false,
+        (Included(left), Included(right)) => left < right,
+        (Included(left), Excluded(right)) => left <= right,
+        (Excluded(left), Included(right)) => left <= right,
+        (Excluded(left), Excluded(right)) => left <= right,
+    }
+}
+
 /// Group adjacent versions locations.
 ///
 /// ```text
@@ -371,10 +414,70 @@ fn group_adjacent_locations(
 impl<V: Ord + Clone> Range<V> {
     /// Computes the union of this `Range` and another.
     pub fn union(&self, other: &Self) -> Self {
-        self.complement()
-            .intersection(&other.complement())
-            .complement()
-            .check_invariants()
+        let mut output: SmallVec<Interval<V>> = SmallVec::empty();
+        let mut accumulator: Option<(&Bound<_>, &Bound<_>)> = None;
+        let mut left_iter = self.segments.iter().peekable();
+        let mut right_iter = other.segments.iter().peekable();
+        loop {
+            let smaller_interval = match (left_iter.peek(), right_iter.peek()) {
+                (Some((left_start, left_end)), Some((right_start, right_end))) => {
+                    let left_start_is_smaller = match (left_start, right_start) {
+                        (Unbounded, _) => true,
+                        (_, Unbounded) => false,
+                        (Included(l), Included(r)) => l <= r,
+                        (Excluded(l), Excluded(r)) => l <= r,
+                        (Included(l), Excluded(r)) => l <= r,
+                        (Excluded(l), Included(r)) => l < r,
+                    };
+
+                    if left_start_is_smaller {
+                        left_iter.next();
+                        (left_start, left_end)
+                    } else {
+                        right_iter.next();
+                        (right_start, right_end)
+                    }
+                }
+                (Some((left_start, left_end)), None) => {
+                    left_iter.next();
+                    (left_start, left_end)
+                }
+                (None, Some((right_start, right_end))) => {
+                    right_iter.next();
+                    (right_start, right_end)
+                }
+                (None, None) => break,
+            };
+
+            if let Some(accumulator_) = accumulator {
+                if end_before_start_with_gap(accumulator_.1, smaller_interval.0) {
+                    output.push((accumulator_.0.clone(), accumulator_.1.clone()));
+                    accumulator = Some(smaller_interval);
+                } else {
+                    let accumulator_end = match (accumulator_.1, smaller_interval.1) {
+                        (_, Unbounded) | (Unbounded, _) => &Unbounded,
+                        (Included(l), Excluded(r) | Included(r)) if l == r => accumulator_.1,
+                        (Excluded(l) | Included(l), Included(r)) if l == r => smaller_interval.1,
+                        (Included(l) | Excluded(l), Included(r) | Excluded(r)) => {
+                            if l > r {
+                                accumulator_.1
+                            } else {
+                                smaller_interval.1
+                            }
+                        }
+                    };
+                    accumulator = Some((accumulator_.0, accumulator_end));
+                }
+            } else {
+                accumulator = Some(smaller_interval)
+            }
+        }
+
+        if let Some(accumulator) = accumulator {
+            output.push((accumulator.0.clone(), accumulator.1.clone()));
+        }
+
+        Self { segments: output }.check_invariants()
     }
 
     /// Computes the intersection of two sets of versions.
@@ -444,6 +547,92 @@ impl<V: Ord + Clone> Range<V> {
         Self { segments: output }.check_invariants()
     }
 
+    /// Return true is there can be no `V` so that `V` is contained in both `self` and `other`.
+    ///
+    /// Note that we don't know that set of all existing `V`s here, so we only check if the segments
+    /// are disjoint, not if no version is contained in both.
+    pub fn is_disjoint(&self, other: &Self) -> bool {
+        let end_before_start = |end, start| match (end, start) {
+            (_, Unbounded) => false,
+            (Unbounded, _) => false,
+            (Included(left), Included(right)) => left < right,
+            (Included(left), Excluded(right)) => left <= right,
+            (Excluded(left), Included(right)) => left <= right,
+            (Excluded(left), Excluded(right)) => left <= right,
+        };
+
+        // The operation is symmetric
+        let mut left_iter = self.segments.iter().peekable();
+        let mut right_iter = other.segments.iter().peekable();
+
+        while let (Some(left), Some(right)) = (left_iter.peek(), right_iter.peek()) {
+            if end_before_start(left.end_bound(), right.start_bound()) {
+                left_iter.next();
+            } else if end_before_start(right.end_bound(), left.start_bound()) {
+                right_iter.next();
+            } else {
+                return false;
+            }
+        }
+
+        // The remaining element(s) can't intersect anymore
+        true
+    }
+
+    /// Return true if any `V` that is contained in `self` is also contained in `other`.
+    ///
+    /// Note that we don't know that set of all existing `V`s here, so we only check if all
+    /// segments `self` are contained in a segment of `other`.
+    fn subset_of(&self, other: &Self) -> bool {
+        let mut containing_iter = other.segments.iter();
+        let mut subset_iter = self.segments.iter();
+        let Some(mut containing_elem) = containing_iter.next() else {
+            // As long as we have subset elements, we need containing elements
+            return subset_iter.next().is_none();
+        };
+
+        for subset_elem in subset_iter {
+            // Check if the current containing element ends before the subset element.
+            // There needs to be another containing element for our subset element in this case.
+            while end_before_start(containing_elem.end_bound(), subset_elem.start_bound()) {
+                if let Some(containing_elem_) = containing_iter.next() {
+                    containing_elem = containing_elem_;
+                } else {
+                    return false;
+                };
+            }
+
+            let start_contained = match (containing_elem.start_bound(), subset_elem.start_bound()) {
+                (Unbounded, _) => true,
+                (Included(_) | Excluded(_), Unbounded) => false,
+                // This is the only case where the subset is bound is "wider" than containing bound ...
+                (Excluded(left), Included(right)) => left < right,
+                // ... while in the other cases they can share the point
+                (Included(left), Included(right)) => left <= right,
+                (Included(left), Excluded(right)) => left <= right,
+                (Excluded(left), Excluded(right)) => left <= right,
+            };
+
+            let end_contained = match (subset_elem.end_bound(), containing_elem.end_bound()) {
+                (_, Unbounded) => true,
+                (Unbounded, Included(_) | Excluded(_)) => false,
+                // This is the only case where the subset is bound is "wider" than containing bound ...
+                (Included(left), Excluded(right)) => left < right,
+                // ... while in the other cases they can share the point
+                (Included(left), Included(right)) => left <= right,
+                (Excluded(left), Included(right)) => left <= right,
+                (Excluded(left), Excluded(right)) => left <= right,
+            };
+
+            if !(start_contained && end_contained) {
+                // This subset element is not contained
+                return false;
+            }
+        }
+
+        true
+    }
+
     /// Returns a simpler Range that contains the same versions
     ///
     /// For every one of the Versions provided in versions the existing range and
@@ -538,6 +727,14 @@ impl<T: Debug + Display + Clone + Eq + Ord> VersionSet for Range<T> {
     fn union(&self, other: &Self) -> Self {
         Range::union(self, other)
     }
+
+    fn is_disjoint(&self, other: &Self) -> bool {
+        Range::is_disjoint(self, other)
+    }
+
+    fn subset_of(&self, other: &Self) -> bool {
+        Range::subset_of(self, other)
+    }
 }
 
 // REPORT ######################################################################
@@ -749,6 +946,16 @@ pub mod tests {
             assert_eq!(r1.union(&r2).contains(&version), r1.contains(&version) || r2.contains(&version));
         }
 
+        #[test]
+        fn union_through_intersection(r1 in strategy(), r2 in strategy()) {
+            let union_def = r1
+                .complement()
+                .intersection(&r2.complement())
+                .complement()
+                .check_invariants();
+            assert_eq!(r1.union(&r2), union_def);
+        }
+
         // Testing contains --------------------------------
 
         #[test]

src/term.rs

@@ -6,7 +6,14 @@
 use crate::version_set::VersionSet;
 use std::fmt::{self, Display};
 
-///  A positive or negative expression regarding a set of versions.
+/// A positive or negative expression regarding a set of versions.
+///
+/// If a version is selected then `Positive(r)` and `Negative(r.complement())` are equivalent, but
+/// they have different semantics when no version is selected. A `Positive` term in the partial
+/// solution requires a version to be selected. But a `Negative` term allows for a solution that
+/// does not have that package selected. Specifically, `Positive(VS::empty())` means that there was
+/// a conflict, we need to select a version for the package but can't pick any, while
+/// `Negative(VS::full())` would mean it is fine as long as we don't select the package.
 #[derive(Debug, Clone, Eq, PartialEq)]
 pub enum Term<VS: VersionSet> {
     /// For example, "1.0.0 <= v < 2.0.0" is a positive expression
@@ -84,7 +91,8 @@ impl<VS: VersionSet> Term<VS> {
 /// Set operations with terms.
 impl<VS: VersionSet> Term<VS> {
     /// Compute the intersection of two terms.
-    /// If at least one term is positive, the intersection is also positive.
+    ///
+    /// The intersection is positive is at least one of the two terms is positive.
     pub(crate) fn intersection(&self, other: &Self) -> Self {
         match (self, other) {
             (Self::Positive(r1), Self::Positive(r2)) => Self::Positive(r1.intersection(r2)),
@@ -98,10 +106,34 @@ impl<VS: VersionSet> Term<VS> {
         }
     }
 
+    /// Check whether two terms are mutually exclusive.
+    ///
+    /// An optimization for the native implementation of checking whether the intersection of two sets is empty.
+    pub(crate) fn is_disjoint(&self, other: &Self) -> bool {
+        match (self, other) {
+            (Self::Positive(r1), Self::Positive(r2)) => r1.is_disjoint(r2),
+            (Self::Negative(r1), Self::Negative(r2)) => r1 == &VS::empty() && r2 == &VS::empty(),
+            // If the positive term is a subset of the negative term, it lies fully in the region that the negative
+            // term excludes.
+            (Self::Positive(r1), Self::Negative(r2)) => r1.subset_of(r2),
+            // Inversely, if there is a region outside the negative, they overlap in this region.
+            (Self::Negative(r1), Self::Positive(r2)) => r2.subset_of(r1),
+        }
+    }
+
     /// Compute the union of two terms.
     /// If at least one term is negative, the union is also negative.
     pub(crate) fn union(&self, other: &Self) -> Self {
-        (self.negate().intersection(&other.negate())).negate()
+        match (self, other) {
+            (Self::Positive(r1), Self::Positive(r2)) => Self::Positive(r1.union(r2)),
+            (Self::Positive(r1), Self::Negative(r2)) => {
+                Self::Negative(r1.union(&r2.complement()).complement())
+            }
+            (Self::Negative(r1), Self::Positive(r2)) => {
+                Self::Negative(r1.complement().union(r2).complement())
+            }
+            (Self::Negative(r1), Self::Negative(r2)) => Self::Negative(r1.intersection(r2)),
+        }
     }
 
     /// Indicate if this term is a subset of another term.
@@ -200,7 +232,6 @@ pub mod tests {
             crate::range::tests::strategy().prop_map(Term::Negative),
         ]
     }
-
     proptest! {
 
         // Testing relation --------------------------------
@@ -217,5 +248,13 @@ pub mod tests {
             }
         }
 
+        /// Ensure that we don't wrongly convert between positive and negative ranges
+        #[test]
+        fn positive_negative(term1 in strategy(), term2 in strategy()) {
+            let intersection_positive = term1.is_positive() || term2.is_positive();
+            let union_positive = term1.is_positive() && term2.is_positive();
+            assert_eq!(term1.intersection(&term2).is_positive(), intersection_positive);
+            assert_eq!(term1.union(&term2).is_positive(), union_positive);
+        }
     }
 }

src/version_set.rs

@@ -57,4 +57,14 @@ pub trait VersionSet: Debug + Display + Clone + Eq {
             .intersection(&other.complement())
             .complement()
     }
+
+    /// Whether the range have no overlapping segmets
+    fn is_disjoint(&self, other: &Self) -> bool {
+        self.intersection(other) == Self::empty()
+    }
+
+    /// Whether all range of `self` are contained in `other`
+    fn subset_of(&self, other: &Self) -> bool {
+        self == &self.intersection(other)
+    }
 }