[libcxx-commits] [libcxx] [libcxx] patch for implementing ranges::find_last (PR #67270)

[via libcxx-commits]

================
@@ -0,0 +1,87 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _LIBCPP___ALGORITHM_RANGES_FIND_LAST_H
+#define _LIBCPP___ALGORITHM_RANGES_FIND_LAST_H
+
+#include <__config>
+#include <__concepts/assignable.h>
+#include <__functional/identity.h>
+#include <__functional/invoke.h>
+#include <__functional/ranges_operations.h>
+#include <__iterator/concepts.h>
+#include <__iterator/next.h>
+#include <__iterator/projected.h>
+#include <__ranges/access.h>
+#include <__ranges/concepts.h>
+#include <__ranges/dangling.h>
+#include <__ranges/range_adaptor.h>
+#include <__ranges/subrange.h>
+#include <__utility/move.h>
+
+#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
+#  pragma GCC system_header
+#endif
+
+#if _LIBCPP_STD_VER >= 23
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+namespace ranges {
+namespace __find_last {
+struct __fn {
+  template <forward_iterator _Ip, sentinel_for<_Ip> _Sp, class _Tp, class _Proj = identity>
+    requires indirect_binary_predicate<ranges::equal_to, projected<_Ip, _Proj>, const _Tp*>
+  _LIBCPP_NODISCARD_EXT _LIBCPP_HIDE_FROM_ABI constexpr subrange<_Ip>
+  operator()(_Ip __first, _Sp __last, const _Tp& __value, _Proj __proj = {}) const {
+    if constexpr ((bidirectional_range<_Ip> && common_range<_Ip>) ||
+                  (bidirectional_iterator<_Ip> && assignable_from<_Ip&, _Sp&>)) {
----------------
huixie90 wrote:

I think this line is pointless. This optimization is to prevent traverse from the beginning. If it is not a ‘common_range’ , you have to traverse to the end in order to start, which defeats the purpose. You could argue that you did not compare the elements in the first round traversal, but, by putting this condition in the same branch as the ‘common_range’ case, you are making the ‘common_range’ case worse, because common_range case does not need to call ranges::next   .  If you really want to optimise !common && bidi && assignable, please do it in a separate overload or use if constexpr to prevent wasteful operations inside the same overload

https://github.com/llvm/llvm-project/pull/67270