提升asio优先级队列,从处理程序添加异步操作

时间:2015-06-24 07:11:25

标签: c++ c++11 boost-asio

基于boost asio的优先级队列示例,我遇到了设计问题。如果我在处理程序中添加一个包装的处理程序,它似乎迷路了:

有关示例,请参阅http://www.boost.org/doc/libs/1_55_0/doc/html/boost_asio/example/cpp03/invocation/prioritised_handlers.cpp

我已按原样使用了所有内容,并使用以下代码替换了main()函数:

//
// based on prioritised_handlers.cpp
// ~~~~~~~~~~~~~~~~~~~~~~~~
//
// Copyright (c) 2003-2015 Christopher M. Kohlhoff (chris at kohlhoff dot com)
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
//

#include <boost/asio.hpp>
#include <boost/function.hpp>
#include <iostream>
#include <queue>

using boost::asio::ip::tcp;

class handler_priority_queue
{
public:
  void add(int priority, boost::function<void()> function)
  {
    handlers_.push(queued_handler(priority, function));
  }

  void execute_all()
  {
    while (!handlers_.empty())
    {
      queued_handler handler = handlers_.top();
      handler.execute();
      handlers_.pop();
    }
  }

  // A generic wrapper class for handlers to allow the invocation to be hooked.
  template <typename Handler>
  class wrapped_handler
  {
  public:
    wrapped_handler(handler_priority_queue& q, int p, Handler h)
      : queue_(q), priority_(p), handler_(h)
    {
    }

    void operator()()
    {
      handler_();
    }

    template <typename Arg1>
    void operator()(Arg1 arg1)
    {
      handler_(arg1);
    }

    template <typename Arg1, typename Arg2>
    void operator()(Arg1 arg1, Arg2 arg2)
    {
      handler_(arg1, arg2);
    }

  //private:
    handler_priority_queue& queue_;
    int priority_;
    Handler handler_;
  };

  template <typename Handler>
  wrapped_handler<Handler> wrap(int priority, Handler handler)
  {
    return wrapped_handler<Handler>(*this, priority, handler);
  }

private:
  class queued_handler
  {
  public:
    queued_handler(int p, boost::function<void()> f)
      : priority_(p), function_(f)
    {
    }

    void execute()
    {
      function_();
    }

    friend bool operator<(const queued_handler& a,
        const queued_handler& b)
    {
      return a.priority_ < b.priority_;
    }

  private:
    int priority_;
    boost::function<void()> function_;
  };

  std::priority_queue<queued_handler> handlers_;
};

// Custom invocation hook for wrapped handlers.
template <typename Function, typename Handler>
void asio_handler_invoke(Function f,
    handler_priority_queue::wrapped_handler<Handler>* h)
{
  h->queue_.add(h->priority_, f);
}
void low_priority_handler()
{
  std::cout << "Low priority handler\n";
}

int main()
{
  //
  // BASED ON prioritised_handlers.cpp
  // ~~~~~~~~~~~~~~~~~~~~~~~~
  //
  // Copyright (c) 2003-2011 Christopher M. Kohlhoff (chris at kohlhoff dot com)
  //
  // Distributed under the Boost Software License, Version 1.0. (See accompanying
  // file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
  //



  //----------------------------------------------------------------------
  using boost::asio::ip::tcp;
  boost::asio::io_service io_service;

  handler_priority_queue pri_queue;

  // Post a completion handler to be run immediately.
  io_service.post(pri_queue.wrap(0, low_priority_handler));


  // Set a deadline timer to expire immediately.
  boost::asio::deadline_timer timer1(io_service);
  timer1.expires_at(boost::posix_time::neg_infin);
  timer1.async_wait(pri_queue.wrap(42, [](const boost::system::error_code& )
                                   {
                                     std::cout << "now" << std::endl;
                                   }));
  // Set a deadline timer to expire later.
  boost::asio::deadline_timer timer2(io_service, boost::posix_time::milliseconds(100));
  boost::asio::deadline_timer timer3(io_service, boost::posix_time::milliseconds(200));
  timer2.async_wait(pri_queue.wrap(100, [&pri_queue, &timer3](const boost::system::error_code& )
                                   {
                                     std::cout << "100ms" << std::endl;
                                     timer3.async_wait(pri_queue.wrap(100, [](const boost::system::error_code& )
                                                                      {
                                                                        std::cout << "200ms" << std::endl;
                                                                      }));

                                   }));

  while (io_service.run_one())
    {
      // The custom invocation hook adds the handlers to the priority queue
      // rather than executing them from within the poll_one() call.
      while (io_service.poll_one())
        ;

      pri_queue.execute_all();
    }
}

//g++ -std=c++14 -Wall -Werror  -rdynamic -lboost_system -lboost_thread -lboost_log -lpthread prioritised_handlers.cpp

打印:

now
Low priority handler
100ms

缺少来自timer3的200ms打印输出。基于我的printf调试方法,自定义调用挂钩asio_handler_invoke()永远不会被调用为“200ms”操作。不幸的是我看不出原因。

上述方法有什么问题?

从Technik提示

更新代码的其余部分

2 个答案:

答案 0 :(得分:1)

任何代码都没有任何问题,除了你保持(或者不是保持)应用程序的方式足够长,以便最后一个处理程序完成。

您正在运行io_service,轮询并运行每个作业。所有作业都得到处理,但它们是延迟的异步调用。您依靠io_service调用来保持应用程序活着,但就io_service而言,所有工作都已完成,因此它会停止阻塞,您的main()函数退出并且您永远不会看到最终处理程序的输出

为了向您证明这一点,只需将io_service包装在::work对象中,这将阻止io_service认为它无法完成工作,因此始终会阻止。

boost::asio::io_service::work w(io_service);

while (io_service.run_one())行之前添加此权限。

答案 1 :(得分:0)

我认为克里斯托弗的实施存在一个错误:

std::priority_queue

如果您将新元素添加到 point1 pop(),这是相同的优先级(在您的情况下为100) - 处理程序可能会添加到堆顶部。在这种情况下, point2 上的io_service将弹出我们的新处理程序而不执行。在这种情况下,您的point1循环将停止,因为没有更多的工作。

我认为你可以交换行point2select K, listagg(case when chgv1 = 1 then v1 else null end,',') within group (order by v1) as v1lst, --Only consider cases in listagg when rows have changed listagg(case when chgv2 = 1 then v2 else null end,',') within group (order by v2) as v2lst --Only consider cases in listagg when rows have changed from ( select k, v1, v2, row_number() over (partition by k,v1 order by v1 ) as chgv1, --Detect changes in v1. In this case it's 1. row_number() over (partition by k,v2 order by v1 ) as chgv2 --Detect changes in v1. In this case it's 1. from t) group by k; 来修复它。