LGTM, although the utility of this LWG issue seems suspect to me lol (given that it only triggers when disable_sized_range<R> && sized_sentinel_for<S, I>)

More precisely, it's !sized_range<R> && sized_sentinel_for<sentinel_t<R>, iterator_t<R>>. Since ranges::size is smart enough to call begin and end an return the difference, this boils down to covering single-pass (not forward) ranges with sized_sentinel_for. IIRC I pointed this out to LWG and suggested closing as NAD, but LWG was convinced that it was worth doing "just in case" since the cost is a couple of small if constexpr branches.

Thanks! Really appreciate the test case, especially when it's non-trivial to set up like this one.

@CaseyCarter @strega-nil-ms I pushed changes after you approved, to defend against min macroization.

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.

Thanks for taking the STL to the next level of conformance! 📊 🎉 😹