在多线程环境中交换 c++ 映射对象

Question

我是 c++ 编码的新手，需要 swap/replace 旧的 multimap 对象与新构建的 multimap 对象，因为这个对象将被缓存我想在构建之后替换现有对象新对象，只需替换对象本身。这将在多线程环境中使用，因此使用原子加载。如本帖 Want an efficient way to swap two pointers in C++ 所述。我写了这段代码

#include<iostream>
#include<map>
#include<atomic>
#include<string>
using namespace std;

// MultiMap Object
struct mmap{
multimap<string,int> stringTointmap;
};

// Structure to swap two instances of multimap
struct swapMap{
  mmap* m1;
  mmap* m2;
};

int main(){

//create Two Objects
mmap* old = new mmap();
mmap* new2= new mmap();

// populate first object
old->stringTointmap.insert(make_pair("old",1));
//populate second object
new2->stringTointmap.insert(make_pair("new1",2));

//swap two objects
atomic<swapMap> swap;
auto refresh=swap.load();
refresh= {swap.m2,swap.m1};
}

但是我遇到了这个错误

error: expected expression
refresh= {swap.m2,swap.m1};

当然，我遗漏了什么，有人可以帮忙吗？

Answer 1

下面的示例代码展示了如何在 std::shared_ptr 上使用原子操作。

#include <memory>
#include <thread>
#include <chrono>
#include <atomic>
#include <iostream>

std::shared_ptr<std::string> the_string;

int main()
{
    std::atomic_store(&the_string, std::make_shared<std::string>("first string"));

    std::thread thread(
        [&](){
            for (int i = 0; i < 5; ++i)
            {
                {
                    // pin the current instance in memory so we can access it
                    std::shared_ptr<std::string> s = std::atomic_load(&the_string);

                    // access it
                    std::cout << *s << std::endl;
                }
                std::this_thread::sleep_for(std::chrono::seconds(1));
            }
        });

    std::this_thread::sleep_for(std::chrono::seconds(2));

    // replace the current instance with a new instance allowing the old instance
    // to be removed when all threads are done with it
    std::atomic_store (&the_string, std::make_shared<std::string>("second string"));

    thread.join();
}

输出：

first string
first string
second string
second string
second string

在多线程环境中交换 c++ 映射对象

swap c++ map objects in multithreaded environment

c++

multithreading

multimap