线程池的C++11版本实现问题

因为你没有贴上全部代码，也很难看出来

看一下这个C++11的thread pool怎么样？

#ifndef THREAD_POOL_H    
#define THREAD_POOL_H    
#include <vector>    
#include <queue>    
#include <memory>    
#include <thread>    
#include <mutex>    
#include <condition_variable>    
#include <future>    
#include <functional>    
#include <stdexcept>    
class ThreadPool {    
public:    
   ThreadPool(size_t);    
   template<class F, class... Args>    
   auto enqueue(F&& f, Args&&... args)    
       -> std::future<typename std::result_of<F(Args...)>::type>;    
   ~ThreadPool();    
private:    
   // need to keep track of threads so we can join them    
   std::vector< std::thread > workers;    
   // the task queue    
   std::queue< std::function<void()> > tasks;    
       
   // synchronization    
   std::mutex queue_mutex;    
   std::condition_variable condition;    
   bool stop;    
};    
// the constructor just launches some amount of workers    
inline ThreadPool::ThreadPool(size_t threads)    
   :   stop(false)    
{    
   for(size_t i = 0;i<threads;++i)    
       workers.emplace_back(    
           [this]    
           {    
               for(;;)    
               {    
                   std::function<void()> task;    
                   {    
                       std::unique_lock<std::mutex> lock(this->queue_mutex);    
                       this->condition.wait(lock,    
                           [this]{ return this->stop || !this->tasks.empty(); });    
                       if(this->stop && this->tasks.empty())    
                           return;    
                       task = std::move(this->tasks.front());    
                       this->tasks.pop();    
                   }    
                   task();    
               }    
           }    
       );    
}    
// add new work item to the pool    
template<class F, class... Args>    
auto ThreadPool::enqueue(F&& f, Args&&... args)    
   -> std::future<typename std::result_of<F(Args...)>::type>    
{    
   using return_type = typename std::result_of<F(Args...)>::type;    
   auto task = std::make_shared< std::packaged_task<return_type()> >(    
           std::bind(std::forward<F>(f), std::forward<Args>(args)...)    
       );    
           
   std::future<return_type> res = task->get_future();    
   {    
       std::unique_lock<std::mutex> lock(queue_mutex);    
       // don't allow enqueueing after stopping the pool    
       if(stop)    
           throw std::runtime_error("enqueue on stopped ThreadPool");    
       tasks.emplace([task](){ (*task)(); });    
   }    
   condition.notify_one();    
   return res;    
}    
// the destructor joins all threads    
inline ThreadPool::~ThreadPool()    
{    
   {    
       std::unique_lock<std::mutex> lock(queue_mutex);    
       stop = true;    
   }    
   condition.notify_all();    
   for(std::thread &worker: workers)    
       worker.join();    
}    
#endif