| Defined in header |
||
template< class InputIt, class UnaryFunc > |
(1) | (constexpr since C++20) |
template< class ExecutionPolicy, class ForwardIt, class UnaryFunc >
void for_each( ExecutionPolicy&& policy, |
(2) | (since C++17) |
Applies the given unary function object f to the result of dereferencing every iterator in the range [first, last). If f returns a result, the result is ignored.
1) f is applied in order starting from first.
2) f might not be applied in order. The algorithm is executed according to policy.
This overload participates in overload resolution only if all following conditions are satisfied:
If the iterator type (InputIt/ForwardIt) is mutable, f may modify the elements of the range through the dereferenced iterator.
Unlike the rest of the parallel algorithms, for_each is not allowed to make copies of the elements in the sequence even if they are TriviallyCopyable.
| first, last | - | the pair of iterators defining the range of elements to which the function object will be applied |
| policy | - | the execution policy to use |
| f | - | function object, to be applied to the result of dereferencing every iterator in the range [first, last)
The signature of the function should be equivalent to the following: void fun(const Type &a); The signature does not need to have const &. |
| Type requirements | ||
-InputIt must meet the requirements of LegacyInputIterator.
| ||
-ForwardIt must meet the requirements of LegacyForwardIterator.
| ||
1) f
2) (none)
Exactly std::distance(first, last) applications of f.
The overload with a template parameter named ExecutionPolicy reports errors as follows:
ExecutionPolicy is one of the standard policies, std::terminate is called. For any other ExecutionPolicy, the behavior is implementation-defined.
See also the implementations in libstdc++, libc++ and MSVC stdlib.
template<class InputIt, class UnaryFunc> constexpr UnaryFunc for_each(InputIt first, InputIt last, UnaryFunc f) { for (; first != last; ++first) f(*first); return f; // implicit move since C++11 }
For overload (1), f can be a stateful function object. The return value can be considered as the final state of the batch operation.
For overload (2), multiple copies of f may be created to perform parallel invocation. No value is returned because parallelization often does not permit efficient state accumulation.
The following example uses a lambda-expression to increment all of the elements of a vector and then uses an overloaded operator() in a function object (i.k.a., "functor") to compute their sum. Note that to compute the sum, it is recommended to use the dedicated algorithm std::accumulate.
Run this code
#include <algorithm> #include <iostream> #include <vector> int main() { std::vector<int> v{3, -4, 2, -8, 15, 267}; auto print = [](const int& n) { std::cout << n << ' '; }; std::cout << "before:\t"; std::for_each(v.cbegin(), v.cend(), print); std::cout << '\n'; // increment elements in-place std::for_each(v.begin(), v.end(), [](int &n) { n++; }); std::cout << "after:\t"; std::for_each(v.cbegin(), v.cend(), print); std::cout << '\n'; struct Sum { void operator()(int n) { sum += n; } int sum {0}; }; // invoke Sum::operator() for each element Sum s = std::for_each(v.cbegin(), v.cend(), Sum()); std::cout << "sum:\t" << s.sum << '\n'; }
Output:
before: 3 -4 2 -8 15 267 after: 4 -3 3 -7 16 268 sum: 281
The following behavior-changing defect reports were applied retroactively to previously published C++ standards.
| DR | Applied to | Behavior as published | Correct behavior |
|---|---|---|---|
| LWG 475 | C++98 | it was unclear whether f can modify the elements of the sequence being iterated over ( for_each isclassified as “non-modifying sequence operations”) |
made clear (allowed if the iterator type is mutable) |
| LWG 2747 | C++11 | overload (1) returned std::move(f) | returns f (which implicitly moves) |