[libcxx-commits] [libcxx] [libc++] Optimize copy construction and assignment of __tree (PR #151304)

[Louis Dionne via libcxx-commits]

================
@@ -1277,11 +1351,28 @@ __tree<_Tp, _Compare, _Allocator>::_DetachedTreeCache::__detach_next(__node_poin
 
 template <class _Tp, class _Compare, class _Allocator>
 __tree<_Tp, _Compare, _Allocator>& __tree<_Tp, _Compare, _Allocator>::operator=(const __tree& __t) {
-  if (this != std::addressof(__t)) {
-    value_comp() = __t.value_comp();
-    __copy_assign_alloc(__t);
-    __assign_multi(__t.begin(), __t.end());
+  if (this == std::addressof(__t))
+    return *this;
+
+  value_comp() = __t.value_comp();
+  __copy_assign_alloc(__t);
+
+  if (__t.size() == 0) {
+    clear();
+    return *this;
+  }
----------------
ldionne wrote:

Is there a reason why we need to handle this as a special case? It's just intuition, but I'd be tempted to think that we might have a bug (or at least a landmine waiting to blow up) in `__copy_assign_tree` if this special case is required.

https://github.com/llvm/llvm-project/pull/151304