ACE Tutorial 015
Building a protocol stream
The Protocol_Task implementation takes care of the open(), close(),
put() and svc() methods so that derivatives can concentrate on the
send() and recv() methods. After a while you find that most
ACE_Task<> derivatives look very similar in the four basic methods and
only need one or two additional to do any real work.
// $Id: page13.html,v 1.4 1998/10/27 19:05:56 jcej Exp $
#include "Protocol_Task.h"
// Construct the object and remember the thread count.
Protocol_Task::Protocol_Task( int _thr_count )
: desired_thr_count_(_thr_count)
{
}
Protocol_Task::~Protocol_Task(void)
{
}
// Activate the object if necessary.
int Protocol_Task::open(void *arg)
{
ACE_UNUSED_ARG(arg);
if( desired_thr_count_ )
{
return this->activate(THR_NEW_LWP, desired_thr_count_);
}
return(0);
}
/* When we're being closed by the ACE_Stream and we've got threads to
worry about then we drop a hangup message onto the message queue so
that svc() will go away. Except for the call to is_active(), this
is lifted directly from Tutorial 14.
*/
int Protocol_Task::close(u_long flags)
{
if (flags == 1 && is_active() )
{
ACE_Message_Block *hangupBlock = new ACE_Message_Block();
hangupBlock->msg_type(ACE_Message_Block::MB_HANGUP);
if (this->putq(hangupBlock->duplicate()) == -1) {
ACE_ERROR_RETURN ((LM_ERROR, "%p\n", "Task::close() putq"), -1);
}
hangupBlock->release();
return this->wait();
}
return 0;
}
/* The put() method has to make a decision. If we've got threads then
put the unit of work onto the message queue for svc() to deal
with. If not then process() it directly.
*/
int Protocol_Task::put(ACE_Message_Block *message,ACE_Time_Value *timeout)
{
if( is_active() )
{
return this->putq(message,timeout);
}
return this->process(message,timeout);
}
/* svc() is about what you would expect. This is again lifted
directly from Tutorial 14 but with a call to process() for handling
the logic instead of doing the work right here.
*/
int Protocol_Task::svc(void)
{
ACE_Message_Block * message;
while (1)
{
// Get a message
if ( this->getq(message, 0) == -1) {
ACE_ERROR_RETURN ((LM_ERROR, "%p\n", "Protocol_Task::svc() getq"), -1);
}
ACE_DEBUG ((LM_DEBUG, "(%P|%t) Protocol_Task::svc() got message\n"));
// Check for hangup
if (message->msg_type() == ACE_Message_Block::MB_HANGUP) {
ACE_DEBUG ((LM_DEBUG, "(%P|%t) Protocol_Task::svc() -- HANGUP block received\n"));
// Hangup our thread-pool peers (if any)
if (this->putq(message->duplicate()) == -1) {
ACE_ERROR_RETURN ((LM_ERROR, "%p\n", "Protocol_Task::svc() putq"), -1);
}
// Leave svc()
break;
}
// Do some work on the data.
if( this->process(message->duplicate(),0) == -1 )
{
break;
}
// Give up the message block before we go get another.
message->release();
}
// Give up the message block that caused us to exit the
// while(1) loop.
message->release();
return(0);
}
/* There's nothing really magic about process(). We just decide if
we're moving data upstream or downstream and invoke the appropriate
virtual function to handle it.
*/
int Protocol_Task::process(ACE_Message_Block * message, ACE_Time_Value *timeout)
{
if( this->is_writer() )
{
return this->send(message,timeout);
}
return this->recv(message,timeout);
}
/* We must insist that derivatives provide a meaningful overload for
these methods. It's fairly common for ACE object methods to return
an error when an overload is expected but the method cannot be
safely made pure virtual.
*/
int Protocol_Task::send(ACE_Message_Block *message,
ACE_Time_Value *timeout)
{
return -1;
}
int Protocol_Task::recv(ACE_Message_Block * message,
ACE_Time_Value *timeout)
{
return -1;
}
[Tutorial Index] [Continue This Tutorial]