[libcxx-commits] [libcxx] [libcxx][algorithm] Optimize std::stable_sort via radix sort algorithm (PR #104683)

[Дмитрий Изволов via libcxx-commits]

================
@@ -0,0 +1,415 @@
+// -*- C++ -*-
+//===----------------------------------------------------------------------===//
+//
+// 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_RADIX_SORT_H
+#define _LIBCPP___ALGORITHM_RADIX_SORT_H
+
+#include <__algorithm/copy.h>
+#include <__algorithm/for_each.h>
+#include <__bit/countl.h>
+#include <__config>
+#include <__iterator/distance.h>
+#include <__iterator/iterator_traits.h>
+#include <__iterator/move_iterator.h>
+#include <__iterator/next.h>
+#include <__iterator/reverse_iterator.h>
+#include <__numeric/partial_sum.h>
+#include <__type_traits/decay.h>
+#include <__type_traits/enable_if.h>
+#include <__type_traits/invoke.h>
+#include <__type_traits/is_assignable.h>
+#include <__type_traits/is_integral.h>
+#include <__type_traits/is_unsigned.h>
+#include <__type_traits/make_unsigned.h>
+#include <__utility/forward.h>
+#include <__utility/integer_sequence.h>
+#include <__utility/move.h>
+#include <__utility/pair.h>
+#include <climits>
+#include <cstdint>
+#include <initializer_list>
+#include <limits>
+
+#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
+#  pragma GCC system_header
+#endif
+
+_LIBCPP_PUSH_MACROS
+#include <__undef_macros>
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+#if _LIBCPP_STD_VER >= 14
+
+inline _LIBCPP_HIDE_FROM_ABI void __variadic_expansion_dummy(initializer_list<int>) {}
+
+#  define _EXPAND_VARIADIC(expression) std::__variadic_expansion_dummy({(expression, 0)...})
+
+template <typename _Iterator>
+_LIBCPP_HIDE_FROM_ABI constexpr auto __move_assign_please(_Iterator __i)
----------------
izvolov wrote:

The idea here is as follows.

The algorithm uses an intermediate buffer into which objects are moved, and then they are moved back.
Since the algorithm (in fact) works not only with integers, but also with arbitrary types, then, if the object can be moved, we move it, and only if it can't be moved, then we copy it.

Is there a better way to do this?

https://github.com/llvm/llvm-project/pull/104683