This page is a snapshot from the LWG issues list, see the Library Active Issues List for more information and the meaning of Ready status.

4562. layout_stride::is_exhaustive specification is problematic when there's an extent equal to 1

Section: 23.7.3.4.7.4 [mdspan.layout.stride.obs] Status: Ready Submitter: S. B. Tam Opened: 2026-03-29 Last modified: 2026-06-11

Priority: 3

View all issues with Ready status.

Discussion:

Consider 

constexpr std::extents ext{2, 1, 2};
constexpr std::array stride{1, 5, 2};
constexpr std::layout_stride::mapping lsm(ext, stride);

static_assert(lsm.is_exhaustive());

Implementations disagree on the result of lsm.is_exhaustive(): true on libstdc++ and libc++, false on MSVC STL.

github.com/microsoft/STL/#5477 claims that the mapping is exhaustive in this case, but 23.7.3.4.7.4 [mdspan.layout.stride.obs]/6.2 requires this function to return false.

Should is_exhaustive be made to return true in this case (which is more correct and also easier to implement)?

[2026-05-29; Reflector poll.]

Set priority to 3 after reflector poll.

[2026-06-10; Tomasz provides new wording]

Reflect implementation in the wording.

[Brno 2026-06-11; Change status: New → Ready.]

Proposed resolution:

This wording is relative to N5008.

  1. Modify 23.7.3.4.7.4 [mdspan.layout.stride.obs] as indicated:

    constexpr bool is_exhaustive() const noexcept;

    -5- Returns: required_span_size() == extents().fwd-prod-of-extents(rank_).

    1. (5.1) — true if rank_ or the size of the multidimensional index space m.extents() is 0.

    2. (5.2) — Otherwise, true if there is a permutation P of the integers in the range [0, rank_) such that stride(p0) equals 1, and stride(pi) equals stride(piāˆ’1) * extents().extent(piāˆ’1) for i in the range [1, rank_), where pi is the ith element of P.

    3. (5.3) — Otherwise, false.