Update.

.gitignore

@@ -1,5 +1,3 @@
 /target/
 
 Cargo.lock
-
-/.criterion/

Cargo.toml

@@ -4,12 +4,13 @@ version = "0.1.0"
 authors = ["logaritmisk <anders.e.olsson@gmail.com>"]
 
 [dependencies]
+ordermap = "0.4"
 rayon = "0.9"
 serde = "1.0"
 unreachable = "1.0"
 
 [dev-dependencies]
-criterion = "0.1"
+criterion = "0.2"
 rand = "0.4"
 serde_test = "1.0"
 

benches/bench.rs

@@ -11,7 +11,7 @@ use rand::{Rng, SeedableRng, XorShiftRng};
 
 use bit_set::BitSet;
 
-fn default_set<C: FromIterator<u64>>(n: usize) -> C {
+fn random<C: FromIterator<u64>>(n: usize) -> C {
     let mut rng = XorShiftRng::from_seed([0, 1, 2, 3]);
 
     (0..n).map(|_| rng.next_u64()).collect()
@@ -33,20 +33,77 @@ fn insert(data: &[u64]) {
     }
 }
 
-fn sub(a: &BitSet, b: &BitSet) {
+fn iter(set: &BitSet) {
+    set.iter().collect::<Vec<_>>();
+}
+
+fn is_subset(a: &BitSet, b: &BitSet) {
+    a.is_subset(b);
+}
+
+fn ops_bit_or(a: &BitSet, b: &BitSet) {
+    a | b;
+}
+
+fn ops_sub(a: &BitSet, b: &BitSet) {
     a - b;
 }
 
 fn criterion_benchmark(c: &mut Criterion) {
-    let data: Vec<_> = default_set(1000);
+    // c.sample_size(200);
 
-    c.bench_function("contains", |b| b.iter(|| contains(&data)));
-    c.bench_function("insert", |b| b.iter(|| insert(&data)));
+    c.bench_function("contains", move |b| {
+        let data: Vec<_> = random(10_000);
 
-    let lhs: BitSet = default_set(1000);
-    let rhs: BitSet = default_set(1000);
+        b.iter(|| contains(&data));
+    });
 
-    c.bench_function("sub", |b| b.iter(|| sub(&lhs, &rhs)));
+    c.bench_function("insert", move |b| {
+        let data: Vec<_> = random(10_000);
+
+        b.iter(|| insert(&data));
+    });
+
+    c.bench_function("iter", move |b| {
+        let set: BitSet = random(10_000);
+
+        b.iter(|| iter(&set));
+    });
+
+    c.bench_function("is_subset (>)", |b| {
+        let lhs: BitSet = random(10_000);
+        let rhs: BitSet = random(5_000);
+
+        b.iter(|| is_subset(&lhs, &rhs));
+    });
+
+    c.bench_function("is_subset (<)", |b| {
+        let lhs: BitSet = random(5_000);
+        let rhs: BitSet = random(10_000);
+
+        b.iter(|| is_subset(&lhs, &rhs));
+    });
+
+    c.bench_function("is_subset (=)", |b| {
+        let lhs: BitSet = random(10_000);
+        let rhs: BitSet = random(10_000);
+
+        b.iter(|| is_subset(&lhs, &rhs));
+    });
+
+    c.bench_function("ops_bit_or", move |b| {
+        let lhs: BitSet = random(10_000);
+        let rhs: BitSet = random(10_000);
+
+        b.iter(|| ops_bit_or(&lhs, &rhs));
+    });
+
+    c.bench_function("ops_sub", move |b| {
+        let lhs: BitSet = random(10_000);
+        let rhs: BitSet = random(10_000);
+
+        b.iter(|| ops_sub(&lhs, &rhs));
+    });
 }
 
 criterion_group!(benches, criterion_benchmark);

src/lib.rs

@@ -1,3 +1,4 @@
+extern crate ordermap;
 extern crate rayon;
 extern crate serde;
 extern crate unreachable;
@@ -8,6 +9,7 @@ use std::iter::{FromIterator, IntoIterator, Iterator};
 use std::default::Default;
 use std::collections::{HashMap, LinkedList};
 
+use ordermap::OrderMap;
 use rayon::iter::{FromParallelIterator, IntoParallelIterator, ParallelExtend, ParallelIterator};
 
 use serde::{Deserialize, Deserializer, Serialize, Serializer};
@@ -22,6 +24,7 @@ const BITS: u64 = 64;
 
 type Block = u64;
 type Storage = HashMap<u64, Block, BitBuildHasher>;
+// type Storage = OrderMap<u64, Block, BitBuildHasher>;
 
 pub type BitHashMap<V> = HashMap<u64, V, BitBuildHasher>;
 
@@ -64,7 +67,7 @@ impl BitSet {
 
     pub fn shrink_to_fit(&mut self) {
         self.blocks.retain(|_, block| *block != 0);
-        self.blocks.shrink_to_fit()
+        // self.blocks.shrink_to_fit();
     }
 
     pub fn iter(&self) -> Iter {
@@ -246,6 +249,27 @@ impl<'a> ops::BitOr<&'a Self> for BitSet {
     }
 }
 
+impl<'a> ops::BitOr<&'a BitSet> for &'a BitSet {
+    type Output = BitSet;
+
+    #[inline]
+    fn bitor(self, rhs: &'a BitSet) -> BitSet {
+        let mut set = self.clone();
+
+        for (key, value) in &rhs.blocks {
+            let block = set.blocks.entry(*key).or_insert(0);
+
+            let n = block.count_ones();
+
+            *block |= value;
+
+            set.nbits += (block.count_ones() - n) as usize;
+        }
+
+        set
+    }
+}
+
 impl<'a> ops::Sub<&'a BitSet> for &'a BitSet {
     type Output = BitSet;
 
@@ -269,6 +293,7 @@ impl<'a> ops::Sub<&'a BitSet> for &'a BitSet {
 
 pub struct Iter<'a> {
     iter: std::collections::hash_map::Iter<'a, u64, u64>,
+    // iter: ordermap::Iter<'a, u64, u64>,
     block: u64,
     bits: u64,
     bit: u64,
@@ -276,6 +301,7 @@ pub struct Iter<'a> {
 
 pub struct IntoIter {
     iter: std::collections::hash_map::IntoIter<u64, u64>,
+    // iter: ordermap::IntoIter<u64, u64>,
     block: u64,
     bits: u64,
     bit: u64,