[libcxx-commits] [libcxx] [libc++] Optimize ranges::equal for vector<bool>::iterator (PR #121084)

[Louis Dionne via libcxx-commits]

================
@@ -94,6 +228,32 @@ __equal_impl(_Tp* __first1, _Tp* __last1, _Up* __first2, _Up*, _Pred&, _Proj1&,
   return std::__constexpr_memcmp_equal(__first1, __first2, __element_count(__last1 - __first1));
 }
 
+template <class _Cp,
+          bool _IC1,
+          bool _IC2,
+          class _Pred,
+          class _Proj1,
+          class _Proj2,
+          __enable_if_t<(is_same<_Pred, __equal_to>::value
+#  if _LIBCPP_STD_VER >= 20
+                         || is_same<_Pred, ranges::equal_to>::value
+#  endif
+                         ) &&
+                            __desugars_to_v<__equal_tag, _Pred, bool, bool> && __is_identity<_Proj1>::value &&
+                            __is_identity<_Proj2>::value,
----------------
ldionne wrote:

The following should be sufficient:

```suggestion
          __enable_if_t<__desugars_to_v<__equal_tag, _Pred, bool, bool> && __is_identity<_Proj1>::value &&
                            __is_identity<_Proj2>::value,
```

`__desugars_to_v<__equal_tag, ranges::equal_to, ...>` should be true, so the `is_same` are not required. You can confirm that by double-checking that you still get the same speedup in your benchmark after making this change.

https://github.com/llvm/llvm-project/pull/121084