1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
|
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="GENERATOR" CONTENT="Mozilla/4.04 [en] (X11; I; Linux 2.0.32 i486) [Netscape]">
<META NAME="Author" CONTENT="Billy Quinn">
<META NAME="Description" CONTENT="A first step towards using ACE productively">
<TITLE>ACE Tutorial 005</TITLE>
</HEAD>
<BODY TEXT="#000000" BGCOLOR="#FFFFFF" LINK="#000FFF" VLINK="#FF0F0F">
<CENTER><B><FONT SIZE=+2>ACE Tutorial 005</FONT></B></CENTER>
<CENTER><B><FONT SIZE=+2>On the road to a multithreaded server</FONT></B></CENTER>
<P>
<HR WIDTH="100%">
<P>Ok, so we've got a main() loop that sets up the acceptor and we've seen
how easy it is to create the acceptor object. So far, we've hardly
written any code at all. Well, that's just about to change...
<P>First, we look at <I><A HREF="client_handler.h">client_handler.h</A></I>
for the declaration of the Client_Handler object. Then we look at
the definition where all of the real work of the application takes place.
<P>
<HR WIDTH="100%">
<pre><FONT FACE="Arial,Helvetica">
#ifndef CLIENT_HANDLER_H
#define CLIENT_HANDLER_H
/*
Our client handler must exist somewhere in the ACE_Event_Handler object
hierarchy. This is a requirement of the ACE_Reactor because it maintains
ACE_Event_Handler pointers for each registered event handler. You could
derive our Client_Handler directly from ACE_Event_Handler but you still have
to have an ACE_SOCK_Stream for the actually connection. With a direct
derivative of ACE_Event_Handler, you'll have to contain and maintain an
ACE_SOCK_Stream instance yourself. With ACE_Svc_Handler (which is a
derivative of ACE_Event_Handler) some of those details are handled for you.
*/
#include "ace/Svc_Handler.h"
#if !defined (ACE_LACKS_PRAGMA_ONCE)
# pragma once
#endif /* ACE_LACKS_PRAGMA_ONCE */
#include "ace/SOCK_Stream.h"
/*
Another feature of ACE_Svc_Handler is it's ability to present the ACE_Task<>
interface as well. That's what the ACE_NULL_SYNCH parameter below is all
about. That's beyond our scope here but we'll come back to it in the next
tutorial when we start looking at concurrency options.
*/
class Client_Handler : public ACE_Svc_Handler < ACE_SOCK_STREAM, ACE_NULL_SYNCH >
{
public:
// Constructor...
Client_Handler (void);
/*
The destroy() method is our preferred method of destruction. We could
have overloaded the delete operator but that is neither easy nor
intuitive (at least to me). Instead, we provide a new method of
destruction and we make our destructor protected so that only ourselves,
our derivatives and our friends can delete us. It's a nice
compromise.
*/
void destroy (void);
/*
Most ACE objects have an open() method. That's how you make them ready
to do work. ACE_Event_Handler has a virtual open() method which allows us
to create this overrride. ACE_Acceptor<> will invoke this method after
creating a new Client_Handler when a client connects. Notice that the
parameter to open() is a void*. It just so happens that the pointer
points to the acceptor which created us. You would like for the parameter
to be an ACE_Acceptor<>* but since ACE_Event_Handler is generic, that
would tie it too closely to the ACE_Acceptor<> set of objects. In our
definition of open() you'll see how we get around that.
*/
int open (void *_acceptor);
/*
When there is activity on a registered handler, the handle_input() method
of the handler will be invoked. If that method returns an error code (eg
-- -1) then the reactor will invoke handle_close() to allow the object to
clean itself up. Since an event handler can be registered for more than
one type of callback, the callback mask is provided to inform
handle_close() exactly which method failed. That way, you don't have to
maintain state information between your handle_* method calls. The _handle
parameter is explained below...
As a side-effect, the reactor will also invoke remove_handler()
for the object on the mask that caused the -1 return. This means
that we don't have to do that ourselves!
*/
int handle_close (ACE_HANDLE _handle, ACE_Reactor_Mask _mask);
protected:
/*
When we register with the reactor, we're going to tell it that we want to
be notified of READ events. When the reactor sees that there is read
activity for us, our handle_input() will be invoked. The _handleg
provided is the handle (file descriptor in Unix) of the actual connection
causing the activity. Since we're derived from ACE_Svc_Handler<> and it
maintains it's own peer (ACE_SOCK_Stream) object, this is redundant for
us. However, if we had been derived directly from ACE_Event_Handler, we
may have chosen not to contain the peer. In that case, the _handleg
would be important to us for reading the client's data.
*/
int handle_input (ACE_HANDLE _handle);
/*
This has nothing at all to do with ACE. I've added this here as a worker
function which I will call from handle_input(). That allows me to
introduce concurrencly in later tutorials with a no changes to the worker
function. You can think of process() as application-level code and
everything elase as application-framework code.
*/
int process (char *_rdbuf, int _rdbuf_len);
/*
We don't really do anything in our destructor but we've declared it to be
protected to prevent casual deletion of this object. As I said above, I
really would prefer that everyone goes through the destroy() method to get
rid of us.
*/
~Client_Handler (void);
};
#endif // CLIENT_HANDLER_H
</pre>
<HR WIDTH="100%">
<CENTER>[<A HREF="..">Tutorial
Index</A>] [<A HREF="page05.html">Continue This Tutorial</A>]</CENTER>
</BODY>
</HTML>
|
