feat: add Repository::compute_object_cache_size_for_tree_diffs().

Byron · Byron · commit 63c7a03bb709 · 2024-08-20T20:51:28.000+02:00
With it it's easier to obtain reasonable object cache sizes as optimized
for tree-diffs.
diff --git a/gix/src/object/tree/diff/mod.rs b/gix/src/object/tree/diff/mod.rs
@@ -33,7 +33,8 @@ impl<'repo> Tree<'repo> {
     ///
     /// # Performance
     ///
-    /// It's highly recommended to set an object cache to avoid extracting the same object multiple times.
+    /// It's highly recommended to [set an object cache](crate::Repository::compute_object_cache_size_for_tree_diffs)
+    /// to avoid extracting the same object multiple times.
     /// By default, similar to `git diff`, rename tracking will be enabled if it is not configured.
     ///
     /// Note that if a clone with `--filter=blob=none` was created, rename tracking may fail as it might
diff --git a/gix/src/repository/cache.rs b/gix/src/repository/cache.rs
@@ -27,4 +27,15 @@ impl crate::Repository {
             self.object_cache_size(bytes)
         }
     }
+
+    /// Return the amount of bytes the object cache [should be set to](Self::object_cache_size_if_unset) to perform
+    /// diffs between trees who are similar to `index` in a typical source code repository.
+    ///
+    /// Currently, this allocates about 10MB for every 10k files in `index`, and a minimum of 4KB.
+    #[cfg(feature = "index")]
+    pub fn compute_object_cache_size_for_tree_diffs(&self, index: &gix_index::State) -> usize {
+        let num_tracked = index.entries().len();
+        let ten_mb_for_every_10k_files = (num_tracked as f32 / 10_000.0) * (10 * 1024 * 1024) as f32;
+        (ten_mb_for_every_10k_files as usize).max(4 * 1024)
+    }
 }

Original file line number	Diff line number	Diff line change
`@@ -27,4 +27,15 @@ impl crate::Repository {`
`27`	`27`	`self.object_cache_size(bytes)`
`28`	`28`	`}`
`29`	`29`	`}`
	`30`	`+`
	`31`	`+ /// Return the amount of bytes the object cache [should be set to](Self::object_cache_size_if_unset) to perform`
	`32`	+ /// diffs between trees who are similar to `index` in a typical source code repository.
	`33`	`+ ///`
	`34`	+ /// Currently, this allocates about 10MB for every 10k files in `index`, and a minimum of 4KB.
	`35`	`+ #[cfg(feature = "index")]`
	`36`	`+ pub fn compute_object_cache_size_for_tree_diffs(&self, index: &gix_index::State) -> usize {`
	`37`	`+ let num_tracked = index.entries().len();`
	`38`	`+ let ten_mb_for_every_10k_files = (num_tracked as f32 / 10_000.0) * (10 * 1024 * 1024) as f32;`
	`39`	`+ (ten_mb_for_every_10k_files as usize).max(4 * 1024)`
	`40`	`+ }`
`30`	`41`	`}`