[stdlib] Make MutableCollection.partition(by:) implementation inlinable

lorentey · lorentey · commit 4f016c9caf4d · 2021-02-16T13:56:17.000-08:00
The default `partition(by:)` implementation cannot currently be specialized, which makes it ~64-100x slower than it could be for common cases. (E.g., `Array&lt;Int&gt;.partition(by: { $0 &gt;= foo})` is slower than `.sort()` by a factor of 2-32.)

Make the implementation inlinable.
diff --git a/stdlib/public/core/CollectionAlgorithms.swift b/stdlib/public/core/CollectionAlgorithms.swift
@@ -330,8 +330,8 @@ extension MutableCollection where Self: BidirectionalCollection {
       return try _partitionImpl(by: belongsInSecondPartition)
     }
   }
-  
-  @usableFromInline
+
+  @inlinable
   internal mutating func _partitionImpl(
     by belongsInSecondPartition: (Element) throws -> Bool
   ) rethrows -> Index {

Original file line number	Diff line number	Diff line change
`@@ -330,8 +330,8 @@ extension MutableCollection where Self: BidirectionalCollection {`
`330`	`330`	`return try _partitionImpl(by: belongsInSecondPartition)`
`331`	`331`	`}`
`332`	`332`	`}`
`333`		`-`
`334`		`- @usableFromInline`
	`333`	`+`
	`334`	`+ @inlinable`
`335`	`335`	`internal mutating func _partitionImpl(`
`336`	`336`	`by belongsInSecondPartition: (Element) throws -> Bool`
`337`	`337`	`) rethrows -> Index {`