[stdlib] Set, Dictionary: Enable per-instance hash seeds

The performance of operations merging two similar-sized sets or dictionaries is currently quadratic. Enabling per-instance seeds makes most of these linear, which is what most people expect.

This affects Set.union, Set.symmetricDifference and similar operations, including the for-loop based forms that people write manually.

The only remaining quadratic cases are when two sets/dictionaries are merged that are mutated copy-on-write copies of the same original instance. (We can’t change the hash seed without a full rehash,

Author: lorentey

stdlib/public/core/HashTable.swift

@@ -94,7 +94,6 @@ extension _HashTable {
     for object: AnyObject,
     scale: Int8
   ) -> Int {
-#if false // FIXME: Enable per-instance seeding
     // We generate a new hash seed whenever a new hash table is allocated and
     // whenever an existing table is resized, so that we avoid certain copy
     // operations becoming quadratic.  (For background details, see
@@ -115,10 +114,6 @@ extension _HashTable {
     // guarantee that no two tables with the same seed can coexist at the same
     // time (apart from copy-on-write derivatives of the same table).
     return unsafeBitCast(object, to: Int.self)
-#else
-    // Use per-capacity seeding for now.
-    return Int(scale)
-#endif
   }
 }