std::projected
From cppreference.com
| Defined in header <iterator>
|
||
| (1) | ||
template< std::indirectly_readable I, std::indirectly_regular_unary_invocable<I> Proj > struct projected { using value_type = std::remove_cvref_t<std::indirect_result_t<Proj&, I>>; std::indirect_result_t<Proj&, I> operator*() const; // not defined }; |
(since C++20) (until C++26) |
|
template< std::indirectly_readable I, std::indirectly_regular_unary_invocable<I> Proj > using projected = /*projected-impl*/<I, Proj>::/*__type*/; // see (3) |
(since C++26) | |
template< std::weakly_incrementable I, class Proj > struct incrementable_traits<std::projected<I, Proj>> { using difference_type = std::iter_difference_t<I>; }; |
(2) | (since C++20) (until C++26) |
template< class I, class Proj > struct /*projected-impl*/ { struct /*__type*/ { using value_type = std::remove_cvref_t<std::indirect_result_t<Proj&, I>>; using difference_type = std::iter_difference_t<I>; // conditionally present std::indirect_result_t<Proj&, I> operator*() const; // not defined }; }; |
(3) | (since C++26) (exposition only*) |
1) Class(until C++26)Alias(since C++26) template
projected combines an indirectly_readable type I and a callable object type Proj into a new indirectly_readable type whose reference type is the result of applying Proj to the std::iter_reference_t<I>.2) This specialization of std::incrementable_traits makes
std::projected<I, Proj> a weakly_incrementable type when I is also a weakly_incrementable type.3) An indirect layer used for avoiding unexpected argument-dependent lookup. The member type
difference_type exists only if I models weakly_incrementable.projected is used only to constrain algorithms that accept callable objects and projections, and hence its operator*() is not defined.
Template parameters
| I | - | an indirectly readable type |
| Proj | - | projection applied to a dereferenced I
|
Notes
The indirect layer prevents I and Proj to be associated classes of projected. When an associated class of I or Proj is an incomplete class type, the indirect layer avoids the unnecessary attempt to inspect the definition of that type that results in hard error.
Example
Run this code
#include <algorithm>
#include <cassert>
#include <functional>
#include <iterator>
template<class T>
struct Holder
{
T t;
};
struct Incomplete;
using P = Holder<Incomplete>*;
static_assert(std::equality_comparable<P>); // OK
static_assert(std::indirectly_comparable<P*, P*, std::equal_to<>>); // Error before C++26
static_assert(std::sortable<P*>); // Error before C++26
int main()
{
P a[10] = {}; // ten null pointers
assert(std::count(a, a + 10, nullptr) == 10); // OK
assert(std::ranges::count(a, a + 10, nullptr) == 10); // Error before C++26
}
See also
(C++26) |
computes the value type of an indirectly_readable type by projection(alias template) |