ACE_Future实现了单写多读模式,可以用来异步方法调用的结果返回

ACE_Future

实现了单写多读模式,可以用来异步方法调用的结果返回

 

激活对象调度者,继承的线程基类ACE_Task_Base

1)      方法请求出列svr()

2)      方法请求入列work()

3)      方法请求处理work_i()

 

 

ACE_Method_call的call函数函数建立了Prime_Scheduler类方法请求处理函数work_i关联.具体是ACE_Method_call的函数Call调用转调Prime_Scheduler函数work_i(),work_i()返回结果值返回到set函数中,然后用了ACE_Future的set函数再到ACE_Future_Rep的set函数来产生期望结果值value_并保存再ACE_Futrue_Rep中.然后ACE_Futrue的函数ready检查是否有结果值,它也是通过ACE_Future_Rep的ready函数检测set函数产生的值指针是否未空来实现,ACE_Furture函数get去获取结果值.

 

ACE_Method_Request类中包含一个call纯虚函数

class ACE_Export ACE_Method_Request

{

public:

  ACE_Method_Request (unsigned long priority = 0);

  virtual ~ACE_Method_Request (void);

  unsigned long priority (void) const;

  void priority (unsigned long prio);

  virtual int call (void) = 0;

protected:

  unsigned long priority_;

}

 

 

class Prime_Scheduler : public ACE_Task_Base

{

public:

         // 初始化

virtual int open (void *args = 0);

 

//终止

virtual int close (u_long flags = 0);

 

//激活对象的代理接口的部分方法

ACE_Future<u_long> work (u_long param, int count = 1);

ACE_Future<const ACE_TCHAR*> name (void);

void end (void);

 

protected:

         //运行事件循环

         virtual int svc (void);

        

         //上面两个方法work,name的实现

         u_long work_i (u_long, int);

const ACE_TCHAR *name_i (void);

 

private:

         //激活对象队列

         ACE_Activation_Queue activation_queue_;

        

         //激活对象指针

Prime_Scheduler *scheduler_;

}

 

1)请求出列

//开始处理激活对象队列中的方法请求调用

int Prime_Scheduler::svc (void)

{

  for (;;)

    {

// 出列下一个方法调用请求 (我们使用智能指针保证异常mo的清除)

auto_ptr<ACE_Method_Request> mo (this->activation_queue_.dequeue ());

 

      ACE_DEBUG ((LM_DEBUG,

                  ACE_TEXT ("(%t) calling method request/n")));

      // 调用它

      if (mo->call () == -1)

        break;

    }

}

 

2)请求入列

调度者的需要请求工作入列函数work,主要是将方法请求入列

ACE_Future<u_long>

Prime_Scheduler::work (u_long newparam,

                       int newcount)

{

  // 判断调度者的内部的调度者指针

  if (this->scheduler_) {

    return this->scheduler_->work (newparam, newcount);

  }

 

  // 没有的话直接使用激活对象队列

  else {

    ACE_Future<u_long> new_future;

        

         //方法请求入列.

    this->activation_queue_.enqueue

      (new Method_Request_work (this,

                                newparam,

                                newcount,

                                new_future));

    return new_future;

  }

}

 

3) 请求处理

// 这里是Work 发生的地方. 该方法在Method_Request_work类中call函数中调用

u_long

Prime_Scheduler::work_i (u_long param,

                         int count)

{

  ACE_UNUSED_ARG (count);

  return ACE::is_prime (param, 2, param / 2);

}

 

 

方法请求工作类Method_Request_work的实现,构造函数传入了激活对象调度者

class Method_Request_work : public ACE_Method_Request

{

  Method_Request_work (Prime_Scheduler *,

                       u_long,

                       int,

                       ACE_Future<u_long> &);

 

virtual int call (void)

{

// 派发servant的操作和保存结果到Futrue中,我们注意到set的参数1是

// 调度者的work_i函数

       return this->future_result_.set (this->scheduler_->work_i

                                   (this->param_,

                                    this->count_));

}

pivate:

Prime_Scheduler *scheduler_;

  u_long param_;

  int count_;

ACE_Future<u_long> future_result_;

}

 

 

int main()

{

Prime_Scheduler *andres, *peter, *helmut, *matias;

创建四个激活对象.

 

  for (int i = 0; i < n_loops; i++)

    {

      {

        ACE_Future<u_long> fresulta;

        ACE_Future<u_long> fresultb;

        ACE_Future<u_long> fresultc;

        ACE_Future<u_long> fresultd;

        ACE_Future<u_long> fresulte;

        ACE_Future<const ACE_TCHAR *> fname;

 

       //方法请求入列,等待对象激活调用,他们是在调度线程svr中异步执行

        fresulta = andres->work (9013);

        fresultb = peter->work (9013);

        fresultc = helmut->work (9013);

        fresultd = matias->work (9013);

      

       //看是否有结果

       if (fresulta.ready ())

 

        if (i % 3 == 0)

          {

            // 三次以上取消futures... 

            fresulta.cancel (10ul);

            fresultb.cancel (20ul);

            fresultc.cancel (30ul);

            fresultd.cancel (40ul);

          }

      

       // 获取结果

        u_long resulta = 0, resultb = 0, resultc = 0, resultd = 0, resulte = 0;

 

        fresulta.get (resulta);

        fresultb.get (resultb);

        fresultc.get (resultc);

        fresultd.get (resultd);

        fresulte.get (resulte);

}

}

 

  // 关闭激活对象

  andres->end ();

  peter->end ();

  helmut->end ();

  matias->end ();

 

  delete andres;

  delete peter;

  delete helmut;

  delete matias;

 

}

 

 

 

//实现了但写多读模式,可以用来异步方法调用的结果返回,ACE_Future重载了=操作符.

template <class T>

class ACE_Future

{

public:

  // 多个构造函数

  ACE_Future (void);

  ACE_Future (const ACE_Future<T> &r);

  ACE_Future (const T &r);

  ~ACE_Future (void);

 

  // 操作符重载

  void operator = (const ACE_Future<T> &r);

  int operator == (const ACE_Future<T> &r) const;

  int operator != (const ACE_Future<T> &r) const;

  operator T ();

 

  int cancel (const T &r);

  int cancel (void);

  int set (const T &r);

  int get (T &value, ACE_Time_Value *tv = 0) const;

  int ready (void) const;

  int attach (ACE_Future_Observer<T> *observer);

  int detach (ACE_Future_Observer<T> *observer);

  void dump (void) const;

 

  ACE_Future_Rep<T> *get_rep (void);

  ACE_ALLOC_HOOK_DECLARE;

 

private:

  /// 不允许new,delete,&操作符

  void *operator new (size_t nbytes);

  void operator delete (void *);

  void operator & ();

 

  /// 保护<Future>.的操作

  typedef ACE_Future_Rep<T> FUTURE_REP;

  FUTURE_REP *future_rep_;

};

 

 

template <class T>

class ACE_Future_Observer

{

public:

  virtual ~ACE_Future_Observer (void);

  virtual void update (const ACE_Future<T> &future) = 0;

  ACE_ALLOC_HOOK_DECLARE;

protected:

  ACE_Future_Observer (void);

};

 

 

分配this->future_rep_,并将计数++

template <class T> void

ACE_Future<T>::operator = (const ACE_Future<T> &rhs)

{

  ACE_Future<T> &r = (ACE_Future<T> &) rhs;

  FUTURE_REP::assign (this->future_rep_,

                      FUTURE_REP::attach (r.future_rep_));

}

 

attach函数是将引用++

template <class T> ACE_Future_Rep<T> *

ACE_Future_Rep<T>::attach (ACE_Future_Rep<T>*& rep)

{

  ACE_ASSERT (rep != 0);

  // Use value_ready_mutex_ for both condition and ref count management

  ACE_MT (ACE_Guard<ACE_Thread_Mutex> r_mon (rep->value_ready_mutex_));

  ++rep->ref_count_;

  return rep;

}

 

指派ACE_Furture的成员ACE_Furture_Rep

template <class T> void

ACE_Future_Rep<T>::assign (ACE_Future_Rep<T>*& rep, ACE_Future_Rep<T>* new_rep)

{

  ACE_ASSERT (rep != 0);

  ACE_ASSERT (new_rep != 0);

  // Use value_ready_mutex_ for both condition and ref count management

  ACE_MT (ACE_GUARD (ACE_Thread_Mutex, r_mon, rep->value_ready_mutex_));

 

  ACE_Future_Rep<T>* old = rep;

  rep = new_rep;

 

  // detached old last for exception safety

  if (old->ref_count_-- == 0)

    {

      ACE_MT (r_mon.release ());

      // We do not need the lock when deleting the representation.

      // There should be no side effects from deleting rep and we don

      // not want to release a deleted mutex.

      delete old;

    }

}

 

ACE_Future函数set间接调用成员的ACE_Future_Rep的set函数

template <class T> int

ACE_Future<T>::set (const T &r)

{

  // 给结果指针到ACE_Future_Rep.

  return this->future_rep_->set (r, *this);

}

 

 

template <class T> int

ACE_Future_Rep<T>::set (const T &r,

                        ACE_Future<T> &caller)

{

  // 如果值已经产生,忽略它.

  if (this->value_ == 0)

    {

      ACE_MT (ACE_GUARD_RETURN (ACE_Thread_Mutex,

                                ace_mon,

                                this->value_ready_mutex_,

                                -1));

      

// 不是的话则new一个结果值,注意这里使用双检查锁模式避免多次分配

      if (this->value_ == 0)  //还没有值,产生

        {

          ACE_NEW_RETURN (this->value_,

                          T (r),

                          -1);

 

          // Remove and notify all subscribed observers.

          ACE_TYPENAME OBSERVER_COLLECTION::iterator iterator =

            this->observer_collection_.begin ();

 

          ACE_TYPENAME OBSERVER_COLLECTION::iterator end =

            this->observer_collection_.end ();

 

          while (iterator != end)

            {

              OBSERVER *observer = *iterator++;

              observer->update (caller);

            }

                    // 对所以等待的线程广播信号

          return this->value_ready_.broadcast ();

        }

    }

  return 0;

}

 

检查结果值

template <class T> int

ACE_Future<T>::ready (void) const

{

  // We're ready if the ACE_Future_rep is ready...

  return this->future_rep_->ready ();

}

 

简单判断value指针是否未空返回

template <class T> int

ACE_Future_Rep<T>::ready (void) const

{

  return this->value_ != 0;

}

 

获取结果值,更新所以的ACE_Future_Observer

template <class T> int

ACE_Future_Rep<T>::get (T &value,

                        ACE_Time_Value *tv) const

{

  // If the value is already produced, return it.

  if (this->value_ == 0)

    {

      ACE_MT (ACE_GUARD_RETURN (ACE_Thread_Mutex, ace_mon,

                                ACE_const_cast (ACE_Thread_Mutex &, this->value_ready_mutex_),

                                -1));

      // If the value is not yet defined we must block until the

      // producer writes to it.

 

      while (this->value_ == 0)

        // Perform a timed wait.

        if ((ACE_const_cast (ACE_Condition_Thread_Mutex &, this->value_ready_)).wait (tv) == -1)

          return -1;

 

      // Destructor releases the lock.

    }

 

  value = *this->value_;

  return 0;

}

 

 

本文引用通告地址： http://blog.csdn.net/linweixuan/services/trackbacks/218752.aspx