Use a custom hasher for 'seen' objects hashset… (#67)

…for about 10% of performance, speeding up these lookups just a little
bit.

git-pack/src/data/output/count/objects/mod.rs

@@ -14,6 +14,7 @@ mod util;
 
 mod types;
 pub use types::{Error, ObjectExpansion, Options, Outcome};
+
 mod tree;
 
 /// The return type used by [`objects()`].
@@ -68,7 +69,7 @@ where
         iter: objects_ids,
         size: chunk_size,
     };
-    let seen_objs = dashmap::DashSet::<ObjectId>::new();
+    let seen_objs = dashmap::DashSet::<ObjectId, types::OidState>::default();
     let progress = Arc::new(parking_lot::Mutex::new(progress));
 
     parallel::in_parallel(
@@ -124,7 +125,7 @@ where
     Oid: Into<ObjectId> + Send,
     IterErr: std::error::Error + Send,
 {
-    let seen_objs = RefCell::new(HashSet::<ObjectId>::new());
+    let seen_objs = RefCell::new(HashSet::<ObjectId, types::OidState>::default());
 
     let (mut buf1, mut buf2) = (Vec::new(), Vec::new());
     expand::this(

git-pack/src/data/output/count/objects/tree.rs

@@ -4,10 +4,9 @@ pub mod changes {
         Visit,
     };
     use git_hash::ObjectId;
-    use git_object::bstr::BStr;
+    use git_object::{bstr::BStr, tree::EntryMode};
 
     use crate::data::output::count::objects_impl::util::InsertImmutable;
-    use git_object::tree::EntryMode;
 
     pub struct AllNew<'a, H> {
         pub objects: Vec<ObjectId>,
@@ -61,11 +60,13 @@ pub mod changes {
 
 pub mod traverse {
     use git_hash::ObjectId;
-    use git_object::{bstr::BStr, tree::EntryRef};
+    use git_object::{
+        bstr::BStr,
+        tree::{EntryMode, EntryRef},
+    };
     use git_traverse::tree::{visit::Action, Visit};
 
     use crate::data::output::count::objects_impl::util::InsertImmutable;
-    use git_object::tree::EntryMode;
 
     pub struct AllUnseen<'a, H> {
         pub non_trees: Vec<ObjectId>,

git-pack/src/data/output/count/objects/types.rs

@@ -1,3 +1,8 @@
+use std::{
+    convert::TryInto,
+    hash::{BuildHasher, Hasher},
+};
+
 /// Information gathered during the run of [`iter_from_objects()`][super::objects()].
 #[derive(Default, PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone, Copy)]
 #[cfg_attr(feature = "serde1", derive(serde::Serialize, serde::Deserialize))]
@@ -14,6 +19,32 @@ pub struct Outcome {
     pub total_objects: usize,
 }
 
+#[derive(Default, Clone)]
+pub struct OidState;
+
+#[derive(Default, Clone)]
+pub struct OidHasher {
+    digest: u64,
+}
+
+impl Hasher for OidHasher {
+    fn finish(&self) -> u64 {
+        self.digest
+    }
+
+    fn write(&mut self, bytes: &[u8]) {
+        self.digest = u64::from_be_bytes(bytes[..8].try_into().expect("any git hash has more than 8 bytes"));
+    }
+}
+
+impl BuildHasher for OidState {
+    type Hasher = OidHasher;
+
+    fn build_hasher(&self) -> Self::Hasher {
+        OidHasher::default()
+    }
+}
+
 impl Outcome {
     pub(in crate::data::output::count) fn aggregate(
         &mut self,

git-pack/src/data/output/count/objects/util.rs

@@ -8,14 +8,15 @@ mod trait_impls {
     use dashmap::DashSet;
 
     use super::InsertImmutable;
+    use crate::data::output::count::objects_impl::types::OidState;
 
-    impl<T: Eq + Hash> InsertImmutable<T> for DashSet<T> {
+    impl<T: Eq + Hash> InsertImmutable<T> for DashSet<T, OidState> {
         fn insert(&self, item: T) -> bool {
             self.insert(item)
         }
     }
 
-    impl<T: Eq + Hash> InsertImmutable<T> for RefCell<HashSet<T>> {
+    impl<T: Eq + Hash> InsertImmutable<T> for RefCell<HashSet<T, OidState>> {
         fn insert(&self, item: T) -> bool {
             self.borrow_mut().insert(item)
         }