Proposed resolution:

This wording is relative to N5046.

1. Modify [mdspan.sub.overview] as indicated:

   -8- Given a type `E` that is a specialization of `extents`, a type `S` is a valid submdspan slice type for the k^th extent of `E` if `S` is a canonical slice type for `E::index_type`, and for `x` equal to `E::static_extent(k)`, either `x` is equal to `dynamic_extent`; or

   1. (8.?) — `S` is `E::index_type` or `full_extent_t`, or
   2. (8.1) — if `S` is a specialization of `extent_slice`, then […]

There is an oversight in paragraph 8 of [mdspan.sub.overview] `dynamic_extent`. This is plainly an oversight and can be corrected by introducing a new 8.1 paragraph:

— `S` is `E::index_type` or `full_extent_t`, otherwise

If `S` is `index_type` (i.e. a runtime value) no further condition on the validity of the type exists, and if it is `full_extent_t`, it is trivially a valid slice specifier type.

Note that a constant integral argument (`constant_wrapper`) is handled at the end in current paragraph 8.2.

Note furthermore that, `S` is already a slice specifier after canonicalization in this paragraph (or rather in the context where "valid slice specifier type" is used), hence we don't need to consider types convertible to things like `full_extent_t` or `index_type`.