[libcxx-commits] [libcxx] [libc++] Speed up set_intersection() by fast-forwarding over ranges of non-matching elements with one-sided binary search. (PR #75230)

[Iuri Chaer via libcxx-commits]

================
@@ -85,6 +86,56 @@ struct _IterOps<_ClassicAlgPolicy> {
     std::advance(__iter, __count);
   }
 
+  // advance with sentinel, a la std::ranges::advance
+  // it's unclear whether _Iter has a difference_type and whether that's signed, so we play it safe:
+  // use the incoming type for returning and steer clear of negative overflows
+  template <class _Iter, class _Distance>
+  _LIBCPP_HIDE_FROM_ABI _LIBCPP_CONSTEXPR_SINCE_CXX14 static _Distance
+  advance(_Iter& __iter, _Distance __count, const _Iter& __sentinel) {
+    return _IterOps::__advance(__iter, __count, __sentinel, typename iterator_traits<_Iter>::iterator_category());
+  }
+
+private:
+  // advance with sentinel, a la std::ranges::advance -- InputIterator specialization
+  template <class _InputIter, class _Distance>
----------------
ichaer wrote:

There is some concept-checking in `ranges::advance()` which makes that hard... My impression is that, given how simple each method is, the result of that transformation would probably be net-negative.

https://github.com/llvm/llvm-project/pull/75230