<xstd/ints/concepts/integer.hpp>

xstd::integer

template<class T>
concept integer =
  integer_class<T> and
  requires {
    typename make_signed_t<T>;
    typename make_unsigned_t<T>;
  } and
  same_as<make_signed_t<make_unsigned_t<T>>, make_signed_t<T>> and
  same_as<make_unsigned_t<make_signed_t<T>>, make_unsigned_t<T>> and
  (same_as<T, make_signed_t<T>> or same_as<T, make_unsigned_t<T>>);

An integer, built-in or integer-class, that is one half of a signed and unsigned pair. It is P3701R0's arithmetic boundary over the built-ins, extended to integer-class types that a specialization of make_signed and make_unsigned pairs.

Models

• Every exact-width type models it, the 128-bit, Boost, Abseil and bit-precise ones included, under every cv-qualification.
• signed char, unsigned char, int8_t and uint8_t model it; char, wchar_t, char8_t, char16_t, char32_t and bool do not under any cv-qualification, and neither do double, void and int*.
• unsigned _BitInt(1), which has no signed counterpart, does not model it, and asking is not an error.
• An integer-class type that no specialization pairs does not model it.

Notes

• The last clause is what turns the character types away, none being named: a character type is neither half of its own pair, make_signed_t<char> being signed char.
• bool fails integer_class and never reaches the pair.
• A trip through the other signedness and back is the identity.
• A pair whose round trip lands on a third type is turned away by the two same_as clauses that compose the transformations.

See also

Entity	Relation
integer_class	the concept it refines
signed_integer, unsigned_integer	its two halves
make_signed, make_unsigned	the pairing it asks for
std::integral	the standard's concept, which admits bool and the character types