Proposed resolution:

This wording is relative to N4986.

1. Modify [version.syn] as indicated:

   [Drafting notes:

   • It is proposed to restore __cpp_lib_ranges to the value denoting P2494R2.

   • __cpp_lib_ranges_accessors will cover P2602R2.

   • __cpp_lib_ranges_indirect_invocation will cover P2609R3 and P2997R1.

   • None of P2602R2, P2609R3, and P2997R1 really affected utilities in <functional>, so these two proposed macros are not required to be also in <functional>.

   • Proposed macros are freestanding because utilities changed by these papers are all freestanding now.

   ]

   […]
   #define __cpp_lib_ranges                            202406L202207L
     // also in <algorithm>, <functional>, <iterator>, <memory>, <ranges>
   #define __cpp_lib_ranges_accessors                  202211L
     // freestanding, also in <algorithm>, <iterator>, <memory>, <ranges>
   […]
   #define __cpp_lib_ranges_generate_random            202403L // also in <random>
   #define __cpp_lib_ranges_indirect_invocation        202406L
     // freestanding, also in <algorithm>, <iterator>, <memory>, <ranges>
   #define __cpp_lib_ranges_iota                       202202L // also in <numeric>
   […]
   

[ 2024-07-24; Jiang An comments and provides improved wording ]

All utilities changed by P2602R2, P2609R3, and P2997R1 are now freestanding, so these proposed macros should also be freestanding.

[ 2023-05-24; Reflector poll ]

Set priority to 3 after reflector poll.

"Needs a more descriptive name than mechanism."

Currently MSVC STL implements P2602R2 and P2609R3 in C++20 mode as if they were defect reports. However, since P2387R3 and P2494R2, which are possibly considered pure functionality extensions in C++23, also bump __cpp_lib_ranges, it's impossible to detect the status of P2602R2 and P2609R3 in C++20 mode (see the discussion in MSVC STL repo).

This may be a non-defect as P2602R2 and P2609R3 are not officially DRs. However, if these papers (including upcoming P2538R1) are expected to be implemented in C++20 modes, it seems better to establish another feature-test macro (e.g., __cpp_lib_ranges_mechanism) for them.