Proposed resolution:

This wording is relative to N4981.

1. Modify [range.utility.conv.to] as indicated:

   template<class C, input_range R, class... Args> requires (!view<C>)
     constexpr C to(R&& r, Args&&... args);
   

   -1- Mandates: C is a cv-unqualified class type.

   -2- Returns: An object of type C constructed from the elements of r in the following manner:

   1. (2.1) — […]

      1. (2.1.1) — […]

      2. (2.1.2) — […]

      3. (2.1.3) — […]

      4. (2.1.4) — Otherwise, if

         1. (2.1.4.1) — constructible_from<C, Args...> is true, and

         2. (2.1.4.2) — container-appendable<C, range_reference_t<R>> is true:

            C c(std::forward<Args>(args)...);
            subrange s{r};
            if constexpr (sized_range<R>requires { s.size(); } && reservable-container<C>)
              c.reserve(static_cast<range_size_t<C>>(ranges::size(r)s.size()));
            ranges::for_each(rs, container-append(c));
            

[ 2025-02-27; post-Hagenberg status ]

The proposed resolution needs rebasing after P2846R6 was approved in Hagenberg. It also conflicts with LWG 3722.

ranges::to currently only reserves when r satisfies sized_range. However, we can also extract its size when r is an input_range that its sentinel-iterator pair satisfies sized_sentinel_for.

Given that we have specifically designed the decision tree, I see no reason not to reserve in this case, since we'd be calling ranges::begin(r) anyway.