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
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
|
<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%">
<P><FONT FACE="Arial,Helvetica">// $Id$</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica">#ifndef CLIENT_HANDLER_H</FONT>
<BR><FONT FACE="Arial,Helvetica">#define CLIENT_HANDLER_H</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica">/*</FONT>
<BR><FONT FACE="Arial,Helvetica"> Our client handler must exist
somewhere in the ACE_Event_Handler object</FONT>
<BR><FONT FACE="Arial,Helvetica"> hierarchy. This is
a requirement of the ACE_Reactor because it maintains</FONT>
<BR><FONT FACE="Arial,Helvetica"> ACE_Event_Handler pointers
for each registered event handler. You could</FONT>
<BR><FONT FACE="Arial,Helvetica"> derive our Client_Handler
directly from ACE_Event_Handler but you still have</FONT>
<BR><FONT FACE="Arial,Helvetica"> to have an ACE_SOCK_Stream
for the actually connection. With a direct</FONT>
<BR><FONT FACE="Arial,Helvetica"> derivative of ACE_Event_Handler,
you'll have to contain and maintain an</FONT>
<BR><FONT FACE="Arial,Helvetica"> ACE_SOCK_Stream instance
yourself. With ACE_Svc_Handler (which is a</FONT>
<BR><FONT FACE="Arial,Helvetica"> derivative of ACE_Event_Handler)
some of those details are handled for you.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica">#include "ace/Svc_Handler.h"</FONT>
<BR><FONT FACE="Arial,Helvetica">#include "ace/SOCK_Stream.h"</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica">/*</FONT>
<BR><FONT FACE="Arial,Helvetica"> Another feature of ACE_Svc_Handler
is it's ability to present the ACE_Task<></FONT>
<BR><FONT FACE="Arial,Helvetica"> interface as well.
That's what the ACE_NULL_SYNCH parameter below is all</FONT>
<BR><FONT FACE="Arial,Helvetica"> about. That's beyond
our scope here but we'll come back to it in the next</FONT>
<BR><FONT FACE="Arial,Helvetica"> tutorial when we start looking
at concurrency options.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica">class Client_Handler : public ACE_Svc_Handler
< ACE_SOCK_STREAM, ACE_NULL_SYNCH ></FONT>
<BR><FONT FACE="Arial,Helvetica">{</FONT>
<BR><FONT FACE="Arial,Helvetica">public:</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> // Constructor...</FONT>
<BR><FONT FACE="Arial,Helvetica"> Client_Handler (void);</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> /*</FONT>
<BR><FONT FACE="Arial,Helvetica"> The destroy()
method is our preferred method of destruction. We could</FONT>
<BR><FONT FACE="Arial,Helvetica"> have overloaded
the <i>delete</i> operator but that is neither easy nor</FONT>
<BR><FONT FACE="Arial,Helvetica"> intuitive (at
least to me). Instead, we provide a new method of</FONT>
<BR><FONT FACE="Arial,Helvetica"> destruction and
we make our destructor protected so that only ourselves,</FONT>
<BR><FONT FACE="Arial,Helvetica"> our derivatives
and our friends can <i>delete</i> us. It's a nice</FONT>
<BR><FONT FACE="Arial,Helvetica"> compromise.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica"> void destroy (void);</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> /*</FONT>
<BR><FONT FACE="Arial,Helvetica"> Most ACE objects
have an open() method. That's how you make them ready</FONT>
<BR><FONT FACE="Arial,Helvetica"> to do work.
ACE_Event_Handler has a virtual open() method which allows us</FONT>
<BR><FONT FACE="Arial,Helvetica"> to create this
overrride. ACE_Acceptor<> will invoke this method after</FONT>
<BR><FONT FACE="Arial,Helvetica"> creating a new
Client_Handler when a client connects. Notice that the</FONT>
<BR><FONT FACE="Arial,Helvetica"> parameter to
open() is a void*. It just so happens that the pointer</FONT>
<BR><FONT FACE="Arial,Helvetica"> points to the
acceptor which created us. You would like for the parameter</FONT>
<BR><FONT FACE="Arial,Helvetica"> to be an ACE_Acceptor<>*
but since ACE_Event_Handler is generic, that</FONT>
<BR><FONT FACE="Arial,Helvetica"> would tie it
too closely to the ACE_Acceptor<> set of objects. In our</FONT>
<BR><FONT FACE="Arial,Helvetica"> definition of
open() you'll see how we get around that.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica"> int open (void *_acceptor);</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> /*</FONT>
<BR><FONT FACE="Arial,Helvetica"> When there is
activity on a registered handler, the handle_input() method</FONT>
<BR><FONT FACE="Arial,Helvetica"> of the handler
will be invoked. If that method returns an error code (eg</FONT>
<BR><FONT FACE="Arial,Helvetica"> -- -1) then the
reactor will invoke handle_close() to allow the object to</FONT>
<BR><FONT FACE="Arial,Helvetica"> clean itself
up. Since an event handler can be registered for more than</FONT>
<BR><FONT FACE="Arial,Helvetica"> one type of callback,
the callback mask is provided to inform</FONT>
<BR><FONT FACE="Arial,Helvetica"> handle_close()
exactly which method failed. That way, you don't have to</FONT>
<BR><FONT FACE="Arial,Helvetica"> maintain state
information between your handle_* method calls. The _handle</FONT>
<BR><FONT FACE="Arial,Helvetica"> parameter is
explained below...</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica"> int handle_close (ACE_HANDLE _handle,
ACE_Reactor_Mask _mask);</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica">protected:</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> /*</FONT>
<BR><FONT FACE="Arial,Helvetica"> When we register
with the reactor, we're going to tell it that we want to</FONT>
<BR><FONT FACE="Arial,Helvetica"> be notified of
READ events. When the reactor sees that there is read</FONT>
<BR><FONT FACE="Arial,Helvetica"> activity for
us, our handle_input() will be invoked. The <i>_handle</i></FONT>
<BR><FONT FACE="Arial,Helvetica"> provided is the
handle (file descriptor in Unix) of the actual connection</FONT>
<BR><FONT FACE="Arial,Helvetica"> causing the activity.
Since we're derived from ACE_Svc_Handler<> and it</FONT>
<BR><FONT FACE="Arial,Helvetica"> maintains it's
own peer (ACE_SOCK_Stream) object, this is redundant for</FONT>
<BR><FONT FACE="Arial,Helvetica"> us. However,
if we had been derived directly from ACE_Event_Handler, we</FONT>
<BR><FONT FACE="Arial,Helvetica"> may have chosen
not to contain the peer. In that case, the <i>_handle</i></FONT>
<BR><FONT FACE="Arial,Helvetica"> would be important
to us for reading the client's data.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica"> int handle_input (ACE_HANDLE _handle);</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> /*</FONT>
<BR><FONT FACE="Arial,Helvetica"> This has nothing
at all to do with ACE. I've added this here as a worker</FONT>
<BR><FONT FACE="Arial,Helvetica"> function which
I will call from handle_input(). That allows me to</FONT>
<BR><FONT FACE="Arial,Helvetica"> introduce concurrencly
in later tutorials with a no changes to the worker</FONT>
<BR><FONT FACE="Arial,Helvetica"> function.
You can think of process() as application-level code and</FONT>
<BR><FONT FACE="Arial,Helvetica"> everything elase
as application-framework code.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica"> int process (char *_rdbuf, int
_rdbuf_len);</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica"> /*</FONT>
<BR><FONT FACE="Arial,Helvetica"> We don't really
do anything in our destructor but we've declared it to be</FONT>
<BR><FONT FACE="Arial,Helvetica"> protected to
prevent casual deletion of this object. As I said above, I</FONT>
<BR><FONT FACE="Arial,Helvetica"> really would
prefer that everyone goes through the destroy() method to get</FONT>
<BR><FONT FACE="Arial,Helvetica"> rid of us.</FONT>
<BR><FONT FACE="Arial,Helvetica"> */</FONT>
<BR><FONT FACE="Arial,Helvetica"> ~Client_Handler (void);</FONT>
<BR><FONT FACE="Arial,Helvetica">};</FONT><FONT FACE="Arial,Helvetica"></FONT>
<P><FONT FACE="Arial,Helvetica">#endif // CLIENT_HANDLER_H</FONT>
<P>
<HR WIDTH="100%">
<CENTER>[<A HREF="..">Tutorial
Index</A>] [<A HREF="page05.html">Continue This Tutorial</A>]</CENTER>
</BODY>
</HTML>
|
