ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev.html
(Generate patch)

Comparing libev/ev.html (file contents):
Revision 1.1 by root, Mon Nov 12 07:58:13 2007 UTC vs.
Revision 1.64 by root, Fri Dec 7 16:49:49 2007 UTC

4<head> 4<head>
5 <title>libev</title> 5 <title>libev</title>
6 <meta name="description" content="Pod documentation for libev" /> 6 <meta name="description" content="Pod documentation for libev" />
7 <meta name="inputfile" content="&lt;standard input&gt;" /> 7 <meta name="inputfile" content="&lt;standard input&gt;" />
8 <meta name="outputfile" content="&lt;standard output&gt;" /> 8 <meta name="outputfile" content="&lt;standard output&gt;" />
9 <meta name="created" content="Mon Nov 12 08:58:02 2007" /> 9 <meta name="created" content="Fri Dec 7 17:49:47 2007" />
10 <meta name="generator" content="Pod::Xhtml 1.57" /> 10 <meta name="generator" content="Pod::Xhtml 1.57" />
11<link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head> 11<link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
12<body> 12<body>
13<div class="pod"> 13<div class="pod">
14<!-- INDEX START --> 14<!-- INDEX START -->
15<h3 id="TOP">Index</h3> 15<h3 id="TOP">Index</h3>
16 16
17<ul><li><a href="#NAME">NAME</a></li> 17<ul><li><a href="#NAME">NAME</a></li>
18<li><a href="#SYNOPSIS">SYNOPSIS</a></li> 18<li><a href="#SYNOPSIS">SYNOPSIS</a></li>
19<li><a href="#EXAMPLE_PROGRAM">EXAMPLE PROGRAM</a></li>
19<li><a href="#DESCRIPTION">DESCRIPTION</a></li> 20<li><a href="#DESCRIPTION">DESCRIPTION</a></li>
20<li><a href="#FEATURES">FEATURES</a></li> 21<li><a href="#FEATURES">FEATURES</a></li>
21<li><a href="#CONVENTIONS">CONVENTIONS</a></li> 22<li><a href="#CONVENTIONS">CONVENTIONS</a></li>
23<li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
22<li><a href="#TIME_AND_OTHER_GLOBAL_FUNCTIONS">TIME AND OTHER GLOBAL FUNCTIONS</a></li> 24<li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
23<li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li> 25<li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
24<li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a> 26<li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
27<ul><li><a href="#GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</a></li>
25<ul><li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li> 28<li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
26</ul> 29</ul>
27</li> 30</li>
28<li><a href="#WATCHER_TYPES">WATCHER TYPES</a> 31<li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
29<ul><li><a href="#struct_ev_io_is_my_file_descriptor_r">struct ev_io - is my file descriptor readable or writable</a></li> 32<ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</a></li>
30<li><a href="#struct_ev_timer_relative_and_optiona">struct ev_timer - relative and optionally recurring timeouts</a></li> 33<li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
31<li><a href="#ev_periodic">ev_periodic</a></li> 34<li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
32<li><a href="#ev_signal_signal_me_when_a_signal_ge">ev_signal - signal me when a signal gets signalled</a></li> 35<li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
33<li><a href="#ev_child_wait_for_pid_status_changes">ev_child - wait for pid status changes</a></li> 36<li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
37<li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
34<li><a href="#ev_idle_when_you_ve_got_nothing_bett">ev_idle - when you've got nothing better to do</a></li> 38<li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</a></li>
35<li><a href="#prepare_and_check_your_hooks_into_th">prepare and check - your hooks into the event loop</a></li> 39<li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</a></li>
40<li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
41<li><a href="#code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</a></li>
36</ul> 42</ul>
37</li> 43</li>
38<li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li> 44<li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
45<li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
46<li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
47<li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
48<li><a href="#EMBEDDING">EMBEDDING</a>
49<ul><li><a href="#FILESETS">FILESETS</a>
50<ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
51<li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
52<li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
53</ul>
54</li>
55<li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
56<li><a href="#EXAMPLES">EXAMPLES</a></li>
57</ul>
58</li>
59<li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
39<li><a href="#AUTHOR">AUTHOR</a> 60<li><a href="#AUTHOR">AUTHOR</a>
40</li> 61</li>
41</ul><hr /> 62</ul><hr />
42<!-- INDEX END --> 63<!-- INDEX END -->
43 64
44<h1 id="NAME">NAME</h1><p><a href="#TOP" class="toplink">Top</a></p> 65<h1 id="NAME">NAME</h1>
45<div id="NAME_CONTENT"> 66<div id="NAME_CONTENT">
46<p>libev - a high performance full-featured event loop written in C</p> 67<p>libev - a high performance full-featured event loop written in C</p>
47 68
48</div> 69</div>
49<h1 id="SYNOPSIS">SYNOPSIS</h1><p><a href="#TOP" class="toplink">Top</a></p> 70<h1 id="SYNOPSIS">SYNOPSIS</h1>
50<div id="SYNOPSIS_CONTENT"> 71<div id="SYNOPSIS_CONTENT">
51<pre> #include &lt;ev.h&gt; 72<pre> #include &lt;ev.h&gt;
52 73
53</pre> 74</pre>
54 75
55</div> 76</div>
56<h1 id="DESCRIPTION">DESCRIPTION</h1><p><a href="#TOP" class="toplink">Top</a></p> 77<h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
78<div id="EXAMPLE_PROGRAM_CONTENT">
79<pre> #include &lt;ev.h&gt;
80
81 ev_io stdin_watcher;
82 ev_timer timeout_watcher;
83
84 /* called when data readable on stdin */
85 static void
86 stdin_cb (EV_P_ struct ev_io *w, int revents)
87 {
88 /* puts (&quot;stdin ready&quot;); */
89 ev_io_stop (EV_A_ w); /* just a syntax example */
90 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
91 }
92
93 static void
94 timeout_cb (EV_P_ struct ev_timer *w, int revents)
95 {
96 /* puts (&quot;timeout&quot;); */
97 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
98 }
99
100 int
101 main (void)
102 {
103 struct ev_loop *loop = ev_default_loop (0);
104
105 /* initialise an io watcher, then start it */
106 ev_io_init (&amp;stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
107 ev_io_start (loop, &amp;stdin_watcher);
108
109 /* simple non-repeating 5.5 second timeout */
110 ev_timer_init (&amp;timeout_watcher, timeout_cb, 5.5, 0.);
111 ev_timer_start (loop, &amp;timeout_watcher);
112
113 /* loop till timeout or data ready */
114 ev_loop (loop, 0);
115
116 return 0;
117 }
118
119</pre>
120
121</div>
122<h1 id="DESCRIPTION">DESCRIPTION</h1>
57<div id="DESCRIPTION_CONTENT"> 123<div id="DESCRIPTION_CONTENT">
58<p>Libev is an event loop: you register interest in certain events (such as a 124<p>Libev is an event loop: you register interest in certain events (such as a
59file descriptor being readable or a timeout occuring), and it will manage 125file descriptor being readable or a timeout occuring), and it will manage
60these event sources and provide your program events.</p> 126these event sources and provide your program with events.</p>
61<p>To do this, it must take more or less complete control over your process 127<p>To do this, it must take more or less complete control over your process
62(or thread) by executing the <i>event loop</i> handler, and will then 128(or thread) by executing the <i>event loop</i> handler, and will then
63communicate events via a callback mechanism.</p> 129communicate events via a callback mechanism.</p>
64<p>You register interest in certain events by registering so-called <i>event 130<p>You register interest in certain events by registering so-called <i>event
65watchers</i>, which are relatively small C structures you initialise with the 131watchers</i>, which are relatively small C structures you initialise with the
66details of the event, and then hand it over to libev by <i>starting</i> the 132details of the event, and then hand it over to libev by <i>starting</i> the
67watcher.</p> 133watcher.</p>
68 134
69</div> 135</div>
70<h1 id="FEATURES">FEATURES</h1><p><a href="#TOP" class="toplink">Top</a></p> 136<h1 id="FEATURES">FEATURES</h1>
71<div id="FEATURES_CONTENT"> 137<div id="FEATURES_CONTENT">
72<p>Libev supports select, poll, the linux-specific epoll and the bsd-specific 138<p>Libev supports <code>select</code>, <code>poll</code>, the Linux-specific <code>epoll</code>, the
73kqueue mechanisms for file descriptor events, relative timers, absolute 139BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
74timers with customised rescheduling, signal events, process status change 140for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
75events (related to SIGCHLD), and event watchers dealing with the event 141(for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
76loop mechanism itself (idle, prepare and check watchers).</p> 142with customised rescheduling (<code>ev_periodic</code>), synchronous signals
143(<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
144watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
145<code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
146file watchers (<code>ev_stat</code>) and even limited support for fork events
147(<code>ev_fork</code>).</p>
148<p>It also is quite fast (see this
149<a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
150for example).</p>
77 151
78</div> 152</div>
79<h1 id="CONVENTIONS">CONVENTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 153<h1 id="CONVENTIONS">CONVENTIONS</h1>
80<div id="CONVENTIONS_CONTENT"> 154<div id="CONVENTIONS_CONTENT">
81<p>Libev is very configurable. In this manual the default configuration 155<p>Libev is very configurable. In this manual the default configuration will
82will be described, which supports multiple event loops. For more info 156be described, which supports multiple event loops. For more info about
83about various configuraiton options please have a look at the file 157various configuration options please have a look at <strong>EMBED</strong> section in
84<cite>README.embed</cite> in the libev distribution. If libev was configured without 158this manual. If libev was configured without support for multiple event
85support for multiple event loops, then all functions taking an initial 159loops, then all functions taking an initial argument of name <code>loop</code>
86argument of name <code>loop</code> (which is always of type <code>struct ev_loop *</code>) 160(which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
87will not have this argument.</p>
88 161
89</div> 162</div>
90<h1 id="TIME_AND_OTHER_GLOBAL_FUNCTIONS">TIME AND OTHER GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 163<h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
91<div id="TIME_AND_OTHER_GLOBAL_FUNCTIONS_CONT"> 164<div id="TIME_REPRESENTATION_CONTENT">
92<p>Libev represents time as a single floating point number. This type is 165<p>Libev represents time as a single floating point number, representing the
166(fractional) number of seconds since the (POSIX) epoch (somewhere near
167the beginning of 1970, details are complicated, don't ask). This type is
93called <code>ev_tstamp</code>, which is what you should use too. It usually aliases 168called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
94to the double type in C.</p> 169to the <code>double</code> type in C, and when you need to do any calculations on
170it, you should treat it as such.</p>
171
172</div>
173<h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
174<div id="GLOBAL_FUNCTIONS_CONTENT">
175<p>These functions can be called anytime, even before initialising the
176library in any way.</p>
95<dl> 177<dl>
96 <dt>ev_tstamp ev_time ()</dt> 178 <dt>ev_tstamp ev_time ()</dt>
97 <dd> 179 <dd>
98 <p>Returns the current time as libev would use it.</p> 180 <p>Returns the current time as libev would use it. Please note that the
181<code>ev_now</code> function is usually faster and also often returns the timestamp
182you actually want to know.</p>
99 </dd> 183 </dd>
100 <dt>int ev_version_major ()</dt> 184 <dt>int ev_version_major ()</dt>
101 <dt>int ev_version_minor ()</dt> 185 <dt>int ev_version_minor ()</dt>
102 <dd> 186 <dd>
103 <p>You can find out the major and minor version numbers of the library 187 <p>You can find out the major and minor version numbers of the library
104you linked against by calling the functions <code>ev_version_major</code> and 188you linked against by calling the functions <code>ev_version_major</code> and
105<code>ev_version_minor</code>. If you want, you can compare against the global 189<code>ev_version_minor</code>. If you want, you can compare against the global
106symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the 190symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
107version of the library your program was compiled against.</p> 191version of the library your program was compiled against.</p>
108 <p>Usually, its a good idea to terminate if the major versions mismatch, 192 <p>Usually, it's a good idea to terminate if the major versions mismatch,
109as this indicates an incompatible change. Minor versions are usually 193as this indicates an incompatible change. Minor versions are usually
110compatible to older versions, so a larger minor version alone is usually 194compatible to older versions, so a larger minor version alone is usually
111not a problem.</p> 195not a problem.</p>
196 <p>Example: Make sure we haven't accidentally been linked against the wrong
197version.</p>
198<pre> assert ((&quot;libev version mismatch&quot;,
199 ev_version_major () == EV_VERSION_MAJOR
200 &amp;&amp; ev_version_minor () &gt;= EV_VERSION_MINOR));
201
202</pre>
203 </dd>
204 <dt>unsigned int ev_supported_backends ()</dt>
205 <dd>
206 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
207value) compiled into this binary of libev (independent of their
208availability on the system you are running on). See <code>ev_default_loop</code> for
209a description of the set values.</p>
210 <p>Example: make sure we have the epoll method, because yeah this is cool and
211a must have and can we have a torrent of it please!!!11</p>
212<pre> assert ((&quot;sorry, no epoll, no sex&quot;,
213 ev_supported_backends () &amp; EVBACKEND_EPOLL));
214
215</pre>
216 </dd>
217 <dt>unsigned int ev_recommended_backends ()</dt>
218 <dd>
219 <p>Return the set of all backends compiled into this binary of libev and also
220recommended for this platform. This set is often smaller than the one
221returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
222most BSDs and will not be autodetected unless you explicitly request it
223(assuming you know what you are doing). This is the set of backends that
224libev will probe for if you specify no backends explicitly.</p>
225 </dd>
226 <dt>unsigned int ev_embeddable_backends ()</dt>
227 <dd>
228 <p>Returns the set of backends that are embeddable in other event loops. This
229is the theoretical, all-platform, value. To find which backends
230might be supported on the current system, you would need to look at
231<code>ev_embeddable_backends () &amp; ev_supported_backends ()</code>, likewise for
232recommended ones.</p>
233 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
112 </dd> 234 </dd>
113 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt> 235 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
114 <dd> 236 <dd>
115 <p>Sets the allocation function to use (the prototype is similar to the 237 <p>Sets the allocation function to use (the prototype is similar - the
116realloc function). It is used to allocate and free memory (no surprises 238semantics is identical - to the realloc C function). It is used to
117here). If it returns zero when memory needs to be allocated, the library 239allocate and free memory (no surprises here). If it returns zero when
118might abort or take some potentially destructive action. The default is 240memory needs to be allocated, the library might abort or take some
119your system realloc function.</p> 241potentially destructive action. The default is your system realloc
242function.</p>
120 <p>You could override this function in high-availability programs to, say, 243 <p>You could override this function in high-availability programs to, say,
121free some memory if it cannot allocate memory, to use a special allocator, 244free some memory if it cannot allocate memory, to use a special allocator,
122or even to sleep a while and retry until some memory is available.</p> 245or even to sleep a while and retry until some memory is available.</p>
246 <p>Example: Replace the libev allocator with one that waits a bit and then
247retries).</p>
248<pre> static void *
249 persistent_realloc (void *ptr, size_t size)
250 {
251 for (;;)
252 {
253 void *newptr = realloc (ptr, size);
254
255 if (newptr)
256 return newptr;
257
258 sleep (60);
259 }
260 }
261
262 ...
263 ev_set_allocator (persistent_realloc);
264
265</pre>
123 </dd> 266 </dd>
124 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt> 267 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
125 <dd> 268 <dd>
126 <p>Set the callback function to call on a retryable syscall error (such 269 <p>Set the callback function to call on a retryable syscall error (such
127as failed select, poll, epoll_wait). The message is a printable string 270as failed select, poll, epoll_wait). The message is a printable string
128indicating the system call or subsystem causing the problem. If this 271indicating the system call or subsystem causing the problem. If this
129callback is set, then libev will expect it to remedy the sitution, no 272callback is set, then libev will expect it to remedy the sitution, no
130matter what, when it returns. That is, libev will geenrally retry the 273matter what, when it returns. That is, libev will generally retry the
131requested operation, or, if the condition doesn't go away, do bad stuff 274requested operation, or, if the condition doesn't go away, do bad stuff
132(such as abort).</p> 275(such as abort).</p>
276 <p>Example: This is basically the same thing that libev does internally, too.</p>
277<pre> static void
278 fatal_error (const char *msg)
279 {
280 perror (msg);
281 abort ();
282 }
283
284 ...
285 ev_set_syserr_cb (fatal_error);
286
287</pre>
133 </dd> 288 </dd>
134</dl> 289</dl>
135 290
136</div> 291</div>
137<h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p> 292<h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
138<div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2"> 293<div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
139<p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two 294<p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
140types of such loops, the <i>default</i> loop, which supports signals and child 295types of such loops, the <i>default</i> loop, which supports signals and child
141events, and dynamically created loops which do not.</p> 296events, and dynamically created loops which do not.</p>
142<p>If you use threads, a common model is to run the default event loop 297<p>If you use threads, a common model is to run the default event loop
143in your main thread (or in a separate thrad) and for each thread you 298in your main thread (or in a separate thread) and for each thread you
144create, you also create another event loop. Libev itself does no lockign 299create, you also create another event loop. Libev itself does no locking
145whatsoever, so if you mix calls to different event loops, make sure you 300whatsoever, so if you mix calls to the same event loop in different
146lock (this is usually a bad idea, though, even if done right).</p> 301threads, make sure you lock (this is usually a bad idea, though, even if
302done correctly, because it's hideous and inefficient).</p>
147<dl> 303<dl>
148 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt> 304 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
149 <dd> 305 <dd>
150 <p>This will initialise the default event loop if it hasn't been initialised 306 <p>This will initialise the default event loop if it hasn't been initialised
151yet and return it. If the default loop could not be initialised, returns 307yet and return it. If the default loop could not be initialised, returns
152false. If it already was initialised it simply returns it (and ignores the 308false. If it already was initialised it simply returns it (and ignores the
153flags).</p> 309flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
154 <p>If you don't know what event loop to use, use the one returned from this 310 <p>If you don't know what event loop to use, use the one returned from this
155function.</p> 311function.</p>
156 <p>The flags argument can be used to specify special behaviour or specific 312 <p>The flags argument can be used to specify special behaviour or specific
157backends to use, and is usually specified as 0 (or EVFLAG_AUTO)</p> 313backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
158 <p>It supports the following flags:</p> 314 <p>The following flags are supported:</p>
159 <p> 315 <p>
160 <dl> 316 <dl>
161 <dt>EVFLAG_AUTO</dt> 317 <dt><code>EVFLAG_AUTO</code></dt>
162 <dd> 318 <dd>
163 <p>The default flags value. Use this if you have no clue (its the right 319 <p>The default flags value. Use this if you have no clue (it's the right
164thing, believe me).</p> 320thing, believe me).</p>
165 </dd> 321 </dd>
166 <dt>EVFLAG_NOENV</dt> 322 <dt><code>EVFLAG_NOENV</code></dt>
167 <dd> 323 <dd>
168 <p>If this flag bit is ored into the flag value then libev will <i>not</i> look 324 <p>If this flag bit is ored into the flag value (or the program runs setuid
169at the environment variable <code>LIBEV_FLAGS</code>. Otherwise (the default), this 325or setgid) then libev will <i>not</i> look at the environment variable
170environment variable will override the flags completely. This is useful 326<code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
327override the flags completely if it is found in the environment. This is
171to try out specific backends to tets their performance, or to work around 328useful to try out specific backends to test their performance, or to work
172bugs.</p> 329around bugs.</p>
173 </dd> 330 </dd>
174 <dt>EVMETHOD_SELECT portable select backend</dt> 331 <dt><code>EVFLAG_FORKCHECK</code></dt>
175 <dt>EVMETHOD_POLL poll backend (everywhere except windows)</dt>
176 <dt>EVMETHOD_EPOLL linux only</dt>
177 <dt>EVMETHOD_KQUEUE some bsds only</dt>
178 <dt>EVMETHOD_DEVPOLL solaris 8 only</dt>
179 <dt>EVMETHOD_PORT solaris 10 only</dt>
180 <dd> 332 <dd>
181 <p>If one or more of these are ored into the flags value, then only these 333 <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
182backends will be tried (in the reverse order as given here). If one are 334a fork, you can also make libev check for a fork in each iteration by
183specified, any backend will do.</p> 335enabling this flag.</p>
336 <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
337and thus this might slow down your event loop if you do a lot of loop
338iterations and little real work, but is usually not noticeable (on my
339Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
340without a syscall and thus <i>very</i> fast, but my Linux system also has
341<code>pthread_atfork</code> which is even faster).</p>
342 <p>The big advantage of this flag is that you can forget about fork (and
343forget about forgetting to tell libev about forking) when you use this
344flag.</p>
345 <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
346environment variable.</p>
347 </dd>
348 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
349 <dd>
350 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
351libev tries to roll its own fd_set with no limits on the number of fds,
352but if that fails, expect a fairly low limit on the number of fds when
353using this backend. It doesn't scale too well (O(highest_fd)), but its usually
354the fastest backend for a low number of fds.</p>
355 </dd>
356 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
357 <dd>
358 <p>And this is your standard poll(2) backend. It's more complicated than
359select, but handles sparse fds better and has no artificial limit on the
360number of fds you can use (except it will slow down considerably with a
361lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
362 </dd>
363 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
364 <dd>
365 <p>For few fds, this backend is a bit little slower than poll and select,
366but it scales phenomenally better. While poll and select usually scale like
367O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
368either O(1) or O(active_fds).</p>
369 <p>While stopping and starting an I/O watcher in the same iteration will
370result in some caching, there is still a syscall per such incident
371(because the fd could point to a different file description now), so its
372best to avoid that. Also, dup()ed file descriptors might not work very
373well if you register events for both fds.</p>
374 <p>Please note that epoll sometimes generates spurious notifications, so you
375need to use non-blocking I/O or other means to avoid blocking when no data
376(or space) is available.</p>
377 </dd>
378 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
379 <dd>
380 <p>Kqueue deserves special mention, as at the time of this writing, it
381was broken on all BSDs except NetBSD (usually it doesn't work with
382anything but sockets and pipes, except on Darwin, where of course its
383completely useless). For this reason its not being &quot;autodetected&quot;
384unless you explicitly specify it explicitly in the flags (i.e. using
385<code>EVBACKEND_KQUEUE</code>).</p>
386 <p>It scales in the same way as the epoll backend, but the interface to the
387kernel is more efficient (which says nothing about its actual speed, of
388course). While starting and stopping an I/O watcher does not cause an
389extra syscall as with epoll, it still adds up to four event changes per
390incident, so its best to avoid that.</p>
391 </dd>
392 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
393 <dd>
394 <p>This is not implemented yet (and might never be).</p>
395 </dd>
396 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
397 <dd>
398 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
399it's really slow, but it still scales very well (O(active_fds)).</p>
400 <p>Please note that solaris ports can result in a lot of spurious
401notifications, so you need to use non-blocking I/O or other means to avoid
402blocking when no data (or space) is available.</p>
403 </dd>
404 <dt><code>EVBACKEND_ALL</code></dt>
405 <dd>
406 <p>Try all backends (even potentially broken ones that wouldn't be tried
407with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
408<code>EVBACKEND_ALL &amp; ~EVBACKEND_KQUEUE</code>.</p>
184 </dd> 409 </dd>
185 </dl> 410 </dl>
186 </p> 411 </p>
412 <p>If one or more of these are ored into the flags value, then only these
413backends will be tried (in the reverse order as given here). If none are
414specified, most compiled-in backend will be tried, usually in reverse
415order of their flag values :)</p>
416 <p>The most typical usage is like this:</p>
417<pre> if (!ev_default_loop (0))
418 fatal (&quot;could not initialise libev, bad $LIBEV_FLAGS in environment?&quot;);
419
420</pre>
421 <p>Restrict libev to the select and poll backends, and do not allow
422environment settings to be taken into account:</p>
423<pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
424
425</pre>
426 <p>Use whatever libev has to offer, but make sure that kqueue is used if
427available (warning, breaks stuff, best use only with your own private
428event loop and only if you know the OS supports your types of fds):</p>
429<pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
430
431</pre>
187 </dd> 432 </dd>
188 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt> 433 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
189 <dd> 434 <dd>
190 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is 435 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
191always distinct from the default loop. Unlike the default loop, it cannot 436always distinct from the default loop. Unlike the default loop, it cannot
192handle signal and child watchers, and attempts to do so will be greeted by 437handle signal and child watchers, and attempts to do so will be greeted by
193undefined behaviour (or a failed assertion if assertions are enabled).</p> 438undefined behaviour (or a failed assertion if assertions are enabled).</p>
439 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
440<pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
441 if (!epoller)
442 fatal (&quot;no epoll found here, maybe it hides under your chair&quot;);
443
444</pre>
194 </dd> 445 </dd>
195 <dt>ev_default_destroy ()</dt> 446 <dt>ev_default_destroy ()</dt>
196 <dd> 447 <dd>
197 <p>Destroys the default loop again (frees all memory and kernel state 448 <p>Destroys the default loop again (frees all memory and kernel state
198etc.). This stops all registered event watchers (by not touching them in 449etc.). None of the active event watchers will be stopped in the normal
199any way whatsoever, although you cnanot rely on this :).</p> 450sense, so e.g. <code>ev_is_active</code> might still return true. It is your
451responsibility to either stop all watchers cleanly yoursef <i>before</i>
452calling this function, or cope with the fact afterwards (which is usually
453the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
454for example).</p>
200 </dd> 455 </dd>
201 <dt>ev_loop_destroy (loop)</dt> 456 <dt>ev_loop_destroy (loop)</dt>
202 <dd> 457 <dd>
203 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an 458 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
204earlier call to <code>ev_loop_new</code>.</p> 459earlier call to <code>ev_loop_new</code>.</p>
207 <dd> 462 <dd>
208 <p>This function reinitialises the kernel state for backends that have 463 <p>This function reinitialises the kernel state for backends that have
209one. Despite the name, you can call it anytime, but it makes most sense 464one. Despite the name, you can call it anytime, but it makes most sense
210after forking, in either the parent or child process (or both, but that 465after forking, in either the parent or child process (or both, but that
211again makes little sense).</p> 466again makes little sense).</p>
212 <p>You <i>must</i> call this function after forking if and only if you want to 467 <p>You <i>must</i> call this function in the child process after forking if and
213use the event library in both processes. If you just fork+exec, you don't 468only if you want to use the event library in both processes. If you just
214have to call it.</p> 469fork+exec, you don't have to call it.</p>
215 <p>The function itself is quite fast and its usually not a problem to call 470 <p>The function itself is quite fast and it's usually not a problem to call
216it just in case after a fork. To make this easy, the function will fit in 471it just in case after a fork. To make this easy, the function will fit in
217quite nicely into a call to <code>pthread_atfork</code>:</p> 472quite nicely into a call to <code>pthread_atfork</code>:</p>
218<pre> pthread_atfork (0, 0, ev_default_fork); 473<pre> pthread_atfork (0, 0, ev_default_fork);
219 474
220</pre> 475</pre>
476 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
477without calling this function, so if you force one of those backends you
478do not need to care.</p>
221 </dd> 479 </dd>
222 <dt>ev_loop_fork (loop)</dt> 480 <dt>ev_loop_fork (loop)</dt>
223 <dd> 481 <dd>
224 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by 482 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
225<code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop 483<code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
226after fork, and how you do this is entirely your own problem.</p> 484after fork, and how you do this is entirely your own problem.</p>
227 </dd> 485 </dd>
228 <dt>unsigned int ev_method (loop)</dt> 486 <dt>unsigned int ev_loop_count (loop)</dt>
487 <dd>
488 <p>Returns the count of loop iterations for the loop, which is identical to
489the number of times libev did poll for new events. It starts at <code>0</code> and
490happily wraps around with enough iterations.</p>
491 <p>This value can sometimes be useful as a generation counter of sorts (it
492&quot;ticks&quot; the number of loop iterations), as it roughly corresponds with
493<code>ev_prepare</code> and <code>ev_check</code> calls.</p>
229 <dd> 494 </dd>
495 <dt>unsigned int ev_backend (loop)</dt>
496 <dd>
230 <p>Returns one of the <code>EVMETHOD_*</code> flags indicating the event backend in 497 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
231use.</p> 498use.</p>
232 </dd> 499 </dd>
233 <dt>ev_tstamp = ev_now (loop)</dt> 500 <dt>ev_tstamp ev_now (loop)</dt>
234 <dd> 501 <dd>
235 <p>Returns the current &quot;event loop time&quot;, which is the time the event loop 502 <p>Returns the current &quot;event loop time&quot;, which is the time the event loop
236got events and started processing them. This timestamp does not change 503received events and started processing them. This timestamp does not
237as long as callbacks are being processed, and this is also the base time 504change as long as callbacks are being processed, and this is also the base
238used for relative timers. You can treat it as the timestamp of the event 505time used for relative timers. You can treat it as the timestamp of the
239occuring (or more correctly, the mainloop finding out about it).</p> 506event occuring (or more correctly, libev finding out about it).</p>
240 </dd> 507 </dd>
241 <dt>ev_loop (loop, int flags)</dt> 508 <dt>ev_loop (loop, int flags)</dt>
242 <dd> 509 <dd>
243 <p>Finally, this is it, the event handler. This function usually is called 510 <p>Finally, this is it, the event handler. This function usually is called
244after you initialised all your watchers and you want to start handling 511after you initialised all your watchers and you want to start handling
245events.</p> 512events.</p>
246 <p>If the flags argument is specified as 0, it will not return until either 513 <p>If the flags argument is specified as <code>0</code>, it will not return until
247no event watchers are active anymore or <code>ev_unloop</code> was called.</p> 514either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
515 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
516relying on all watchers to be stopped when deciding when a program has
517finished (especially in interactive programs), but having a program that
518automatically loops as long as it has to and no longer by virtue of
519relying on its watchers stopping correctly is a thing of beauty.</p>
248 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle 520 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
249those events and any outstanding ones, but will not block your process in 521those events and any outstanding ones, but will not block your process in
250case there are no events.</p> 522case there are no events and will return after one iteration of the loop.</p>
251 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if 523 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
252neccessary) and will handle those and any outstanding ones. It will block 524neccessary) and will handle those and any outstanding ones. It will block
253your process until at least one new event arrives.</p> 525your process until at least one new event arrives, and will return after
254 <p>This flags value could be used to implement alternative looping 526one iteration of the loop. This is useful if you are waiting for some
255constructs, but the <code>prepare</code> and <code>check</code> watchers provide a better and 527external event in conjunction with something not expressible using other
256more generic mechanism.</p> 528libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
529usually a better approach for this kind of thing.</p>
530 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
531<pre> * If there are no active watchers (reference count is zero), return.
532 - Queue prepare watchers and then call all outstanding watchers.
533 - If we have been forked, recreate the kernel state.
534 - Update the kernel state with all outstanding changes.
535 - Update the &quot;event loop time&quot;.
536 - Calculate for how long to block.
537 - Block the process, waiting for any events.
538 - Queue all outstanding I/O (fd) events.
539 - Update the &quot;event loop time&quot; and do time jump handling.
540 - Queue all outstanding timers.
541 - Queue all outstanding periodics.
542 - If no events are pending now, queue all idle watchers.
543 - Queue all check watchers.
544 - Call all queued watchers in reverse order (i.e. check watchers first).
545 Signals and child watchers are implemented as I/O watchers, and will
546 be handled here by queueing them when their watcher gets executed.
547 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
548 were used, return, otherwise continue with step *.
549
550</pre>
551 <p>Example: Queue some jobs and then loop until no events are outsanding
552anymore.</p>
553<pre> ... queue jobs here, make sure they register event watchers as long
554 ... as they still have work to do (even an idle watcher will do..)
555 ev_loop (my_loop, 0);
556 ... jobs done. yeah!
557
558</pre>
257 </dd> 559 </dd>
258 <dt>ev_unloop (loop, how)</dt> 560 <dt>ev_unloop (loop, how)</dt>
259 <dd> 561 <dd>
260 <p>Can be used to make a call to <code>ev_loop</code> return early. The <code>how</code> argument 562 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
563has processed all outstanding events). The <code>how</code> argument must be either
261must be either <code>EVUNLOOP_ONCE</code>, which will make the innermost <code>ev_loop</code> 564<code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
262call return, or <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> 565<code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
263calls return.</p>
264 </dd> 566 </dd>
265 <dt>ev_ref (loop)</dt> 567 <dt>ev_ref (loop)</dt>
266 <dt>ev_unref (loop)</dt> 568 <dt>ev_unref (loop)</dt>
267 <dd> 569 <dd>
268 <p>Ref/unref can be used to add or remove a refcount on the event loop: Every 570 <p>Ref/unref can be used to add or remove a reference count on the event
269watcher keeps one reference. If you have a long-runing watcher you never 571loop: Every watcher keeps one reference, and as long as the reference
270unregister that should not keep ev_loop from running, ev_unref() after 572count is nonzero, <code>ev_loop</code> will not return on its own. If you have
271starting, and ev_ref() before stopping it. Libev itself uses this for 573a watcher you never unregister that should not keep <code>ev_loop</code> from
272example for its internal signal pipe: It is not visible to you as a user 574returning, ev_unref() after starting, and ev_ref() before stopping it. For
273and should not keep <code>ev_loop</code> from exiting if the work is done. It is 575example, libev itself uses this for its internal signal pipe: It is not
274also an excellent way to do this for generic recurring timers or from 576visible to the libev user and should not keep <code>ev_loop</code> from exiting if
275within third-party libraries. Just remember to unref after start and ref 577no event watchers registered by it are active. It is also an excellent
276before stop.</p> 578way to do this for generic recurring timers or from within third-party
579libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
580 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
581running when nothing else is active.</p>
582<pre> struct ev_signal exitsig;
583 ev_signal_init (&amp;exitsig, sig_cb, SIGINT);
584 ev_signal_start (loop, &amp;exitsig);
585 evf_unref (loop);
586
587</pre>
588 <p>Example: For some weird reason, unregister the above signal handler again.</p>
589<pre> ev_ref (loop);
590 ev_signal_stop (loop, &amp;exitsig);
591
592</pre>
277 </dd> 593 </dd>
278</dl> 594</dl>
279 595
596
597
598
599
280</div> 600</div>
281<h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p> 601<h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
282<div id="ANATOMY_OF_A_WATCHER_CONTENT"> 602<div id="ANATOMY_OF_A_WATCHER_CONTENT">
283<p>A watcher is a structure that you create and register to record your 603<p>A watcher is a structure that you create and register to record your
284interest in some event. For instance, if you want to wait for STDIN to 604interest in some event. For instance, if you want to wait for STDIN to
285become readable, you would create an ev_io watcher for that:</p> 605become readable, you would create an <code>ev_io</code> watcher for that:</p>
286<pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents) 606<pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
287 { 607 {
288 ev_io_stop (w); 608 ev_io_stop (w);
289 ev_unloop (loop, EVUNLOOP_ALL); 609 ev_unloop (loop, EVUNLOOP_ALL);
290 } 610 }
313with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher 633with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher
314*)</code>), and you can stop watching for events at any time by calling the 634*)</code>), and you can stop watching for events at any time by calling the
315corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p> 635corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p>
316<p>As long as your watcher is active (has been started but not stopped) you 636<p>As long as your watcher is active (has been started but not stopped) you
317must not touch the values stored in it. Most specifically you must never 637must not touch the values stored in it. Most specifically you must never
318reinitialise it or call its set method.</p> 638reinitialise it or call its <code>set</code> macro.</p>
319<p>You cna check wether an event is active by calling the <code>ev_is_active
320(watcher *)</code> macro. To see wether an event is outstanding (but the
321callback for it has not been called yet) you cna use the <code>ev_is_pending
322(watcher *)</code> macro.</p>
323<p>Each and every callback receives the event loop pointer as first, the 639<p>Each and every callback receives the event loop pointer as first, the
324registered watcher structure as second, and a bitset of received events as 640registered watcher structure as second, and a bitset of received events as
325third argument.</p> 641third argument.</p>
326<p>The rceeived events usually include a single bit per event type received 642<p>The received events usually include a single bit per event type received
327(you can receive multiple events at the same time). The possible bit masks 643(you can receive multiple events at the same time). The possible bit masks
328are:</p> 644are:</p>
329<dl> 645<dl>
330 <dt>EV_READ</dt> 646 <dt><code>EV_READ</code></dt>
331 <dt>EV_WRITE</dt> 647 <dt><code>EV_WRITE</code></dt>
332 <dd> 648 <dd>
333 <p>The file descriptor in the ev_io watcher has become readable and/or 649 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
334writable.</p> 650writable.</p>
335 </dd> 651 </dd>
336 <dt>EV_TIMEOUT</dt> 652 <dt><code>EV_TIMEOUT</code></dt>
337 <dd>
338 <p>The ev_timer watcher has timed out.</p>
339 </dd> 653 <dd>
340 <dt>EV_PERIODIC</dt> 654 <p>The <code>ev_timer</code> watcher has timed out.</p>
341 <dd> 655 </dd>
342 <p>The ev_periodic watcher has timed out.</p> 656 <dt><code>EV_PERIODIC</code></dt>
343 </dd> 657 <dd>
344 <dt>EV_SIGNAL</dt> 658 <p>The <code>ev_periodic</code> watcher has timed out.</p>
345 <dd> 659 </dd>
660 <dt><code>EV_SIGNAL</code></dt>
661 <dd>
346 <p>The signal specified in the ev_signal watcher has been received by a thread.</p> 662 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
347 </dd>
348 <dt>EV_CHILD</dt>
349 <dd> 663 </dd>
664 <dt><code>EV_CHILD</code></dt>
665 <dd>
350 <p>The pid specified in the ev_child watcher has received a status change.</p> 666 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
351 </dd>
352 <dt>EV_IDLE</dt>
353 <dd> 667 </dd>
668 <dt><code>EV_STAT</code></dt>
669 <dd>
670 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
671 </dd>
672 <dt><code>EV_IDLE</code></dt>
673 <dd>
354 <p>The ev_idle watcher has determined that you have nothing better to do.</p> 674 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
355 </dd>
356 <dt>EV_PREPARE</dt>
357 <dt>EV_CHECK</dt>
358 <dd> 675 </dd>
676 <dt><code>EV_PREPARE</code></dt>
677 <dt><code>EV_CHECK</code></dt>
678 <dd>
359 <p>All ev_prepare watchers are invoked just <i>before</i> <code>ev_loop</code> starts 679 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
360to gather new events, and all ev_check watchers are invoked just after 680to gather new events, and all <code>ev_check</code> watchers are invoked just after
361<code>ev_loop</code> has gathered them, but before it invokes any callbacks for any 681<code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
362received events. Callbacks of both watcher types can start and stop as 682received events. Callbacks of both watcher types can start and stop as
363many watchers as they want, and all of them will be taken into account 683many watchers as they want, and all of them will be taken into account
364(for example, a ev_prepare watcher might start an idle watcher to keep 684(for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
365<code>ev_loop</code> from blocking).</p> 685<code>ev_loop</code> from blocking).</p>
366 </dd> 686 </dd>
367 <dt>EV_ERROR</dt> 687 <dt><code>EV_EMBED</code></dt>
688 <dd>
689 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
690 </dd>
691 <dt><code>EV_FORK</code></dt>
692 <dd>
693 <p>The event loop has been resumed in the child process after fork (see
694<code>ev_fork</code>).</p>
695 </dd>
696 <dt><code>EV_ERROR</code></dt>
368 <dd> 697 <dd>
369 <p>An unspecified error has occured, the watcher has been stopped. This might 698 <p>An unspecified error has occured, the watcher has been stopped. This might
370happen because the watcher could not be properly started because libev 699happen because the watcher could not be properly started because libev
371ran out of memory, a file descriptor was found to be closed or any other 700ran out of memory, a file descriptor was found to be closed or any other
372problem. You best act on it by reporting the problem and somehow coping 701problem. You best act on it by reporting the problem and somehow coping
378programs, though, so beware.</p> 707programs, though, so beware.</p>
379 </dd> 708 </dd>
380</dl> 709</dl>
381 710
382</div> 711</div>
712<h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
713<div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
714<p>In the following description, <code>TYPE</code> stands for the watcher type,
715e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
716<dl>
717 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
718 <dd>
719 <p>This macro initialises the generic portion of a watcher. The contents
720of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
721the generic parts of the watcher are initialised, you <i>need</i> to call
722the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
723type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
724which rolls both calls into one.</p>
725 <p>You can reinitialise a watcher at any time as long as it has been stopped
726(or never started) and there are no pending events outstanding.</p>
727 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
728int revents)</code>.</p>
729 </dd>
730 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
731 <dd>
732 <p>This macro initialises the type-specific parts of a watcher. You need to
733call <code>ev_init</code> at least once before you call this macro, but you can
734call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
735macro on a watcher that is active (it can be pending, however, which is a
736difference to the <code>ev_init</code> macro).</p>
737 <p>Although some watcher types do not have type-specific arguments
738(e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
739 </dd>
740 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
741 <dd>
742 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
743calls into a single call. This is the most convinient method to initialise
744a watcher. The same limitations apply, of course.</p>
745 </dd>
746 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
747 <dd>
748 <p>Starts (activates) the given watcher. Only active watchers will receive
749events. If the watcher is already active nothing will happen.</p>
750 </dd>
751 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
752 <dd>
753 <p>Stops the given watcher again (if active) and clears the pending
754status. It is possible that stopped watchers are pending (for example,
755non-repeating timers are being stopped when they become pending), but
756<code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
757you want to free or reuse the memory used by the watcher it is therefore a
758good idea to always call its <code>ev_TYPE_stop</code> function.</p>
759 </dd>
760 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
761 <dd>
762 <p>Returns a true value iff the watcher is active (i.e. it has been started
763and not yet been stopped). As long as a watcher is active you must not modify
764it.</p>
765 </dd>
766 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
767 <dd>
768 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
769events but its callback has not yet been invoked). As long as a watcher
770is pending (but not active) you must not call an init function on it (but
771<code>ev_TYPE_set</code> is safe) and you must make sure the watcher is available to
772libev (e.g. you cnanot <code>free ()</code> it).</p>
773 </dd>
774 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
775 <dd>
776 <p>Returns the callback currently set on the watcher.</p>
777 </dd>
778 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
779 <dd>
780 <p>Change the callback. You can change the callback at virtually any time
781(modulo threads).</p>
782 </dd>
783 <dt>ev_set_priority (ev_TYPE *watcher, priority)</dt>
784 <dt>int ev_priority (ev_TYPE *watcher)</dt>
785 <dd>
786 <p>Set and query the priority of the watcher. The priority is a small
787integer between <code>EV_MAXPRI</code> (default: <code>2</code>) and <code>EV_MINPRI</code>
788(default: <code>-2</code>). Pending watchers with higher priority will be invoked
789before watchers with lower priority, but priority will not keep watchers
790from being executed (except for <code>ev_idle</code> watchers).</p>
791 <p>This means that priorities are <i>only</i> used for ordering callback
792invocation after new events have been received. This is useful, for
793example, to reduce latency after idling, or more often, to bind two
794watchers on the same event and make sure one is called first.</p>
795 <p>If you need to suppress invocation when higher priority events are pending
796you need to look at <code>ev_idle</code> watchers, which provide this functionality.</p>
797 <p>The default priority used by watchers when no priority has been set is
798always <code>0</code>, which is supposed to not be too high and not be too low :).</p>
799 <p>Setting a priority outside the range of <code>EV_MINPRI</code> to <code>EV_MAXPRI</code> is
800fine, as long as you do not mind that the priority value you query might
801or might not have been adjusted to be within valid range.</p>
802 </dd>
803</dl>
804
805
806
807
808
809</div>
383<h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2> 810<h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
384<div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2"> 811<div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
385<p>Each watcher has, by default, a member <code>void *data</code> that you can change 812<p>Each watcher has, by default, a member <code>void *data</code> that you can change
386and read at any time, libev will completely ignore it. This cna be used 813and read at any time, libev will completely ignore it. This can be used
387to associate arbitrary data with your watcher. If you need more data and 814to associate arbitrary data with your watcher. If you need more data and
388don't want to allocate memory and store a pointer to it in that data 815don't want to allocate memory and store a pointer to it in that data
389member, you can also &quot;subclass&quot; the watcher type and provide your own 816member, you can also &quot;subclass&quot; the watcher type and provide your own
390data:</p> 817data:</p>
391<pre> struct my_io 818<pre> struct my_io
404 struct my_io *w = (struct my_io *)w_; 831 struct my_io *w = (struct my_io *)w_;
405 ... 832 ...
406 } 833 }
407 834
408</pre> 835</pre>
409<p>More interesting and less C-conformant ways of catsing your callback type 836<p>More interesting and less C-conformant ways of casting your callback type
410have been omitted....</p> 837instead have been omitted.</p>
838<p>Another common scenario is having some data structure with multiple
839watchers:</p>
840<pre> struct my_biggy
841 {
842 int some_data;
843 ev_timer t1;
844 ev_timer t2;
845 }
411 846
847</pre>
848<p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
849you need to use <code>offsetof</code>:</p>
850<pre> #include &lt;stddef.h&gt;
412 851
852 static void
853 t1_cb (EV_P_ struct ev_timer *w, int revents)
854 {
855 struct my_biggy big = (struct my_biggy *
856 (((char *)w) - offsetof (struct my_biggy, t1));
857 }
413 858
859 static void
860 t2_cb (EV_P_ struct ev_timer *w, int revents)
861 {
862 struct my_biggy big = (struct my_biggy *
863 (((char *)w) - offsetof (struct my_biggy, t2));
864 }
414 865
415 866
867
868
869</pre>
870
416</div> 871</div>
417<h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p> 872<h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
418<div id="WATCHER_TYPES_CONTENT"> 873<div id="WATCHER_TYPES_CONTENT">
419<p>This section describes each watcher in detail, but will not repeat 874<p>This section describes each watcher in detail, but will not repeat
420information given in the last section.</p> 875information given in the last section. Any initialisation/set macros,
876functions and members specific to the watcher type are explained.</p>
877<p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
878while the watcher is active, you can look at the member and expect some
879sensible content, but you must not modify it (you can modify it while the
880watcher is stopped to your hearts content), or <i>[read-write]</i>, which
881means you can expect it to have some sensible content while the watcher
882is active, but you can also modify it. Modifying it may not do something
883sensible or take immediate effect (or do anything at all), but libev will
884not crash or malfunction in any way.</p>
421 885
886
887
888
889
422</div> 890</div>
423<h2 id="struct_ev_io_is_my_file_descriptor_r">struct ev_io - is my file descriptor readable or writable</h2> 891<h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
424<div id="struct_ev_io_is_my_file_descriptor_r-2"> 892<div id="code_ev_io_code_is_this_file_descrip-2">
425<p>I/O watchers check wether a file descriptor is readable or writable 893<p>I/O watchers check whether a file descriptor is readable or writable
426in each iteration of the event loop (This behaviour is called 894in each iteration of the event loop, or, more precisely, when reading
427level-triggering because you keep receiving events as long as the 895would not block the process and writing would at least be able to write
428condition persists. Remember you cna stop the watcher if you don't want to 896some data. This behaviour is called level-triggering because you keep
429act on the event and neither want to receive future events).</p> 897receiving events as long as the condition persists. Remember you can stop
898the watcher if you don't want to act on the event and neither want to
899receive future events.</p>
900<p>In general you can register as many read and/or write event watchers per
901fd as you want (as long as you don't confuse yourself). Setting all file
902descriptors to non-blocking mode is also usually a good idea (but not
903required if you know what you are doing).</p>
904<p>You have to be careful with dup'ed file descriptors, though. Some backends
905(the linux epoll backend is a notable example) cannot handle dup'ed file
906descriptors correctly if you register interest in two or more fds pointing
907to the same underlying file/socket/etc. description (that is, they share
908the same underlying &quot;file open&quot;).</p>
909<p>If you must do this, then force the use of a known-to-be-good backend
910(at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
911<code>EVBACKEND_POLL</code>).</p>
912<p>Another thing you have to watch out for is that it is quite easy to
913receive &quot;spurious&quot; readyness notifications, that is your callback might
914be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
915because there is no data. Not only are some backends known to create a
916lot of those (for example solaris ports), it is very easy to get into
917this situation even with a relatively standard program structure. Thus
918it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
919<code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
920<p>If you cannot run the fd in non-blocking mode (for example you should not
921play around with an Xlib connection), then you have to seperately re-test
922wether a file descriptor is really ready with a known-to-be good interface
923such as poll (fortunately in our Xlib example, Xlib already does this on
924its own, so its quite safe to use).</p>
430<dl> 925<dl>
431 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt> 926 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
432 <dt>ev_io_set (ev_io *, int fd, int events)</dt> 927 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
433 <dd> 928 <dd>
434 <p>Configures an ev_io watcher. The fd is the file descriptor to rceeive 929 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
435events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ | 930rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
436EV_WRITE</code> to receive the given events.</p> 931<code>EV_READ | EV_WRITE</code> to receive the given events.</p>
932 </dd>
933 <dt>int fd [read-only]</dt>
934 <dd>
935 <p>The file descriptor being watched.</p>
936 </dd>
937 <dt>int events [read-only]</dt>
938 <dd>
939 <p>The events being watched.</p>
437 </dd> 940 </dd>
438</dl> 941</dl>
942<p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
943readable, but only once. Since it is likely line-buffered, you could
944attempt to read a whole line in the callback.</p>
945<pre> static void
946 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
947 {
948 ev_io_stop (loop, w);
949 .. read from stdin here (or from w-&gt;fd) and haqndle any I/O errors
950 }
439 951
952 ...
953 struct ev_loop *loop = ev_default_init (0);
954 struct ev_io stdin_readable;
955 ev_io_init (&amp;stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
956 ev_io_start (loop, &amp;stdin_readable);
957 ev_loop (loop, 0);
958
959
960
961
962</pre>
963
440</div> 964</div>
441<h2 id="struct_ev_timer_relative_and_optiona">struct ev_timer - relative and optionally recurring timeouts</h2> 965<h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
442<div id="struct_ev_timer_relative_and_optiona-2"> 966<div id="code_ev_timer_code_relative_and_opti-2">
443<p>Timer watchers are simple relative timers that generate an event after a 967<p>Timer watchers are simple relative timers that generate an event after a
444given time, and optionally repeating in regular intervals after that.</p> 968given time, and optionally repeating in regular intervals after that.</p>
445<p>The timers are based on real time, that is, if you register an event that 969<p>The timers are based on real time, that is, if you register an event that
446times out after an hour and youreset your system clock to last years 970times out after an hour and you reset your system clock to last years
447time, it will still time out after (roughly) and hour. &quot;Roughly&quot; because 971time, it will still time out after (roughly) and hour. &quot;Roughly&quot; because
448detecting time jumps is hard, and soem inaccuracies are unavoidable (the 972detecting time jumps is hard, and some inaccuracies are unavoidable (the
449monotonic clock option helps a lot here).</p> 973monotonic clock option helps a lot here).</p>
974<p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
975time. This is usually the right thing as this timestamp refers to the time
976of the event triggering whatever timeout you are modifying/starting. If
977you suspect event processing to be delayed and you <i>need</i> to base the timeout
978on the current time, use something like this to adjust for this:</p>
979<pre> ev_timer_set (&amp;timer, after + ev_now () - ev_time (), 0.);
980
981</pre>
982<p>The callback is guarenteed to be invoked only when its timeout has passed,
983but if multiple timers become ready during the same loop iteration then
984order of execution is undefined.</p>
450<dl> 985<dl>
451 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt> 986 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
452 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt> 987 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
453 <dd> 988 <dd>
454 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is 989 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
456timer will automatically be configured to trigger again <code>repeat</code> seconds 991timer will automatically be configured to trigger again <code>repeat</code> seconds
457later, again, and again, until stopped manually.</p> 992later, again, and again, until stopped manually.</p>
458 <p>The timer itself will do a best-effort at avoiding drift, that is, if you 993 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
459configure a timer to trigger every 10 seconds, then it will trigger at 994configure a timer to trigger every 10 seconds, then it will trigger at
460exactly 10 second intervals. If, however, your program cannot keep up with 995exactly 10 second intervals. If, however, your program cannot keep up with
461the timer (ecause it takes longer than those 10 seconds to do stuff) the 996the timer (because it takes longer than those 10 seconds to do stuff) the
462timer will not fire more than once per event loop iteration.</p> 997timer will not fire more than once per event loop iteration.</p>
463 </dd> 998 </dd>
464 <dt>ev_timer_again (loop)</dt> 999 <dt>ev_timer_again (loop)</dt>
465 <dd> 1000 <dd>
466 <p>This will act as if the timer timed out and restart it again if it is 1001 <p>This will act as if the timer timed out and restart it again if it is
467repeating. The exact semantics are:</p> 1002repeating. The exact semantics are:</p>
1003 <p>If the timer is pending, its pending status is cleared.</p>
468 <p>If the timer is started but nonrepeating, stop it.</p> 1004 <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
469 <p>If the timer is repeating, either start it if necessary (with the repeat 1005 <p>If the timer is repeating, either start it if necessary (with the
470value), or reset the running timer to the repeat value.</p> 1006<code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
471 <p>This sounds a bit complicated, but here is a useful and typical 1007 <p>This sounds a bit complicated, but here is a useful and typical
472example: Imagine you have a tcp connection and you want a so-called idle 1008example: Imagine you have a tcp connection and you want a so-called idle
473timeout, that is, you want to be called when there have been, say, 60 1009timeout, that is, you want to be called when there have been, say, 60
474seconds of inactivity on the socket. The easiest way to do this is to 1010seconds of inactivity on the socket. The easiest way to do this is to
475configure an ev_timer with after=repeat=60 and calling ev_timer_again each 1011configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
476time you successfully read or write some data. If you go into an idle 1012<code>ev_timer_again</code> each time you successfully read or write some data. If
477state where you do not expect data to travel on the socket, you can stop 1013you go into an idle state where you do not expect data to travel on the
1014socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
478the timer, and again will automatically restart it if need be.</p> 1015automatically restart it if need be.</p>
1016 <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
1017altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
1018<pre> ev_timer_init (timer, callback, 0., 5.);
1019 ev_timer_again (loop, timer);
1020 ...
1021 timer-&gt;again = 17.;
1022 ev_timer_again (loop, timer);
1023 ...
1024 timer-&gt;again = 10.;
1025 ev_timer_again (loop, timer);
1026
1027</pre>
1028 <p>This is more slightly efficient then stopping/starting the timer each time
1029you want to modify its timeout value.</p>
1030 </dd>
1031 <dt>ev_tstamp repeat [read-write]</dt>
1032 <dd>
1033 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1034or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1035which is also when any modifications are taken into account.</p>
479 </dd> 1036 </dd>
480</dl> 1037</dl>
1038<p>Example: Create a timer that fires after 60 seconds.</p>
1039<pre> static void
1040 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1041 {
1042 .. one minute over, w is actually stopped right here
1043 }
481 1044
1045 struct ev_timer mytimer;
1046 ev_timer_init (&amp;mytimer, one_minute_cb, 60., 0.);
1047 ev_timer_start (loop, &amp;mytimer);
1048
1049</pre>
1050<p>Example: Create a timeout timer that times out after 10 seconds of
1051inactivity.</p>
1052<pre> static void
1053 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1054 {
1055 .. ten seconds without any activity
1056 }
1057
1058 struct ev_timer mytimer;
1059 ev_timer_init (&amp;mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1060 ev_timer_again (&amp;mytimer); /* start timer */
1061 ev_loop (loop, 0);
1062
1063 // and in some piece of code that gets executed on any &quot;activity&quot;:
1064 // reset the timeout to start ticking again at 10 seconds
1065 ev_timer_again (&amp;mytimer);
1066
1067
1068
1069
1070</pre>
1071
482</div> 1072</div>
483<h2 id="ev_periodic">ev_periodic</h2> 1073<h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
484<div id="ev_periodic_CONTENT"> 1074<div id="code_ev_periodic_code_to_cron_or_not-2">
485<p>Periodic watchers are also timers of a kind, but they are very versatile 1075<p>Periodic watchers are also timers of a kind, but they are very versatile
486(and unfortunately a bit complex).</p> 1076(and unfortunately a bit complex).</p>
487<p>Unlike ev_timer's, they are not based on real time (or relative time) 1077<p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
488but on wallclock time (absolute time). You can tell a periodic watcher 1078but on wallclock time (absolute time). You can tell a periodic watcher
489to trigger &quot;at&quot; some specific point in time. For example, if you tell a 1079to trigger &quot;at&quot; some specific point in time. For example, if you tell a
490periodic watcher to trigger in 10 seconds (by specifiying e.g. c&lt;ev_now () 1080periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
491+ 10.&gt;) and then reset your system clock to the last year, then it will 1081+ 10.</code>) and then reset your system clock to the last year, then it will
492take a year to trigger the event (unlike an ev_timer, which would trigger 1082take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
493roughly 10 seconds later and of course not if you reset your system time 1083roughly 10 seconds later and of course not if you reset your system time
494again).</p> 1084again).</p>
495<p>They can also be used to implement vastly more complex timers, such as 1085<p>They can also be used to implement vastly more complex timers, such as
496triggering an event on eahc midnight, local time.</p> 1086triggering an event on eahc midnight, local time.</p>
1087<p>As with timers, the callback is guarenteed to be invoked only when the
1088time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1089during the same loop iteration then order of execution is undefined.</p>
497<dl> 1090<dl>
498 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt> 1091 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
499 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt> 1092 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
500 <dd> 1093 <dd>
501 <p>Lots of arguments, lets sort it out... There are basically three modes of 1094 <p>Lots of arguments, lets sort it out... There are basically three modes of
502operation, and we will explain them from simplest to complex:</p> 1095operation, and we will explain them from simplest to complex:</p>
503
504
505
506
507 <p> 1096 <p>
508 <dl> 1097 <dl>
509 <dt>* absolute timer (interval = reschedule_cb = 0)</dt> 1098 <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
510 <dd> 1099 <dd>
511 <p>In this configuration the watcher triggers an event at the wallclock time 1100 <p>In this configuration the watcher triggers an event at the wallclock time
523<pre> ev_periodic_set (&amp;periodic, 0., 3600., 0); 1112<pre> ev_periodic_set (&amp;periodic, 0., 3600., 0);
524 1113
525</pre> 1114</pre>
526 <p>This doesn't mean there will always be 3600 seconds in between triggers, 1115 <p>This doesn't mean there will always be 3600 seconds in between triggers,
527but only that the the callback will be called when the system time shows a 1116but only that the the callback will be called when the system time shows a
528full hour (UTC), or more correct, when the system time is evenly divisible 1117full hour (UTC), or more correctly, when the system time is evenly divisible
529by 3600.</p> 1118by 3600.</p>
530 <p>Another way to think about it (for the mathematically inclined) is that 1119 <p>Another way to think about it (for the mathematically inclined) is that
531ev_periodic will try to run the callback in this mode at the next possible 1120<code>ev_periodic</code> will try to run the callback in this mode at the next possible
532time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p> 1121time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
533 </dd> 1122 </dd>
534 <dt>* manual reschedule mode (reschedule_cb = callback)</dt> 1123 <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
535 <dd> 1124 <dd>
536 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being 1125 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
537ignored. Instead, each time the periodic watcher gets scheduled, the 1126ignored. Instead, each time the periodic watcher gets scheduled, the
538reschedule callback will be called with the watcher as first, and the 1127reschedule callback will be called with the watcher as first, and the
539current time as second argument.</p> 1128current time as second argument.</p>
540 <p>NOTE: <i>This callback MUST NOT stop or destroy the periodic or any other 1129 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
541periodic watcher, ever, or make any event loop modificstions</i>. If you need 1130ever, or make any event loop modifications</i>. If you need to stop it,
542to stop it, return 1e30 (or so, fudge fudge) and stop it afterwards.</p> 1131return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1132starting a prepare watcher).</p>
543 <p>Its prototype is c&lt;ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1133 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
544ev_tstamp now)&gt;, e.g.:</p> 1134ev_tstamp now)</code>, e.g.:</p>
545<pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now) 1135<pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
546 { 1136 {
547 return now + 60.; 1137 return now + 60.;
548 } 1138 }
549 1139
550</pre> 1140</pre>
551 <p>It must return the next time to trigger, based on the passed time value 1141 <p>It must return the next time to trigger, based on the passed time value
552(that is, the lowest time value larger than to the second argument). It 1142(that is, the lowest time value larger than to the second argument). It
553will usually be called just before the callback will be triggered, but 1143will usually be called just before the callback will be triggered, but
554might be called at other times, too.</p> 1144might be called at other times, too.</p>
1145 <p>NOTE: <i>This callback must always return a time that is later than the
1146passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
555 <p>This can be used to create very complex timers, such as a timer that 1147 <p>This can be used to create very complex timers, such as a timer that
556triggers on each midnight, local time. To do this, you would calculate the 1148triggers on each midnight, local time. To do this, you would calculate the
557next midnight after <code>now</code> and return the timestamp value for this. How you do this 1149next midnight after <code>now</code> and return the timestamp value for this. How
558is, again, up to you (but it is not trivial).</p> 1150you do this is, again, up to you (but it is not trivial, which is the main
1151reason I omitted it as an example).</p>
559 </dd> 1152 </dd>
560 </dl> 1153 </dl>
561 </p> 1154 </p>
562 </dd> 1155 </dd>
563 <dt>ev_periodic_again (loop, ev_periodic *)</dt> 1156 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
565 <p>Simply stops and restarts the periodic watcher again. This is only useful 1158 <p>Simply stops and restarts the periodic watcher again. This is only useful
566when you changed some parameters or the reschedule callback would return 1159when you changed some parameters or the reschedule callback would return
567a different time than the last time it was called (e.g. in a crond like 1160a different time than the last time it was called (e.g. in a crond like
568program when the crontabs have changed).</p> 1161program when the crontabs have changed).</p>
569 </dd> 1162 </dd>
1163 <dt>ev_tstamp interval [read-write]</dt>
1164 <dd>
1165 <p>The current interval value. Can be modified any time, but changes only
1166take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1167called.</p>
1168 </dd>
1169 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1170 <dd>
1171 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1172switched off. Can be changed any time, but changes only take effect when
1173the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1174 </dd>
570</dl> 1175</dl>
1176<p>Example: Call a callback every hour, or, more precisely, whenever the
1177system clock is divisible by 3600. The callback invocation times have
1178potentially a lot of jittering, but good long-term stability.</p>
1179<pre> static void
1180 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1181 {
1182 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1183 }
571 1184
1185 struct ev_periodic hourly_tick;
1186 ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 3600., 0);
1187 ev_periodic_start (loop, &amp;hourly_tick);
1188
1189</pre>
1190<p>Example: The same as above, but use a reschedule callback to do it:</p>
1191<pre> #include &lt;math.h&gt;
1192
1193 static ev_tstamp
1194 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1195 {
1196 return fmod (now, 3600.) + 3600.;
1197 }
1198
1199 ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1200
1201</pre>
1202<p>Example: Call a callback every hour, starting now:</p>
1203<pre> struct ev_periodic hourly_tick;
1204 ev_periodic_init (&amp;hourly_tick, clock_cb,
1205 fmod (ev_now (loop), 3600.), 3600., 0);
1206 ev_periodic_start (loop, &amp;hourly_tick);
1207
1208
1209
1210
1211</pre>
1212
572</div> 1213</div>
573<h2 id="ev_signal_signal_me_when_a_signal_ge">ev_signal - signal me when a signal gets signalled</h2> 1214<h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
574<div id="ev_signal_signal_me_when_a_signal_ge-2"> 1215<div id="code_ev_signal_code_signal_me_when_a-2">
575<p>Signal watchers will trigger an event when the process receives a specific 1216<p>Signal watchers will trigger an event when the process receives a specific
576signal one or more times. Even though signals are very asynchronous, libev 1217signal one or more times. Even though signals are very asynchronous, libev
577will try its best to deliver signals synchronously, i.e. as part of the 1218will try it's best to deliver signals synchronously, i.e. as part of the
578normal event processing, like any other event.</p> 1219normal event processing, like any other event.</p>
579<p>You cna configure as many watchers as you like per signal. Only when the 1220<p>You can configure as many watchers as you like per signal. Only when the
580first watcher gets started will libev actually register a signal watcher 1221first watcher gets started will libev actually register a signal watcher
581with the kernel (thus it coexists with your own signal handlers as long 1222with the kernel (thus it coexists with your own signal handlers as long
582as you don't register any with libev). Similarly, when the last signal 1223as you don't register any with libev). Similarly, when the last signal
583watcher for a signal is stopped libev will reset the signal handler to 1224watcher for a signal is stopped libev will reset the signal handler to
584SIG_DFL (regardless of what it was set to before).</p> 1225SIG_DFL (regardless of what it was set to before).</p>
587 <dt>ev_signal_set (ev_signal *, int signum)</dt> 1228 <dt>ev_signal_set (ev_signal *, int signum)</dt>
588 <dd> 1229 <dd>
589 <p>Configures the watcher to trigger on the given signal number (usually one 1230 <p>Configures the watcher to trigger on the given signal number (usually one
590of the <code>SIGxxx</code> constants).</p> 1231of the <code>SIGxxx</code> constants).</p>
591 </dd> 1232 </dd>
1233 <dt>int signum [read-only]</dt>
1234 <dd>
1235 <p>The signal the watcher watches out for.</p>
1236 </dd>
592</dl> 1237</dl>
593 1238
1239
1240
1241
1242
594</div> 1243</div>
595<h2 id="ev_child_wait_for_pid_status_changes">ev_child - wait for pid status changes</h2> 1244<h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
596<div id="ev_child_wait_for_pid_status_changes-2"> 1245<div id="code_ev_child_code_watch_out_for_pro-2">
597<p>Child watchers trigger when your process receives a SIGCHLD in response to 1246<p>Child watchers trigger when your process receives a SIGCHLD in response to
598some child status changes (most typically when a child of yours dies).</p> 1247some child status changes (most typically when a child of yours dies).</p>
599<dl> 1248<dl>
600 <dt>ev_child_init (ev_child *, callback, int pid)</dt> 1249 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
601 <dt>ev_child_set (ev_child *, int pid)</dt> 1250 <dt>ev_child_set (ev_child *, int pid)</dt>
602 <dd> 1251 <dd>
603 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or 1252 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
604<i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look 1253<i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
605at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see 1254at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
606the status word (use the macros from <code>sys/wait.h</code>). The <code>rpid</code> member 1255the status word (use the macros from <code>sys/wait.h</code> and see your systems
1256<code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
607contains the pid of the process causing the status change.</p> 1257process causing the status change.</p>
1258 </dd>
1259 <dt>int pid [read-only]</dt>
1260 <dd>
1261 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1262 </dd>
1263 <dt>int rpid [read-write]</dt>
1264 <dd>
1265 <p>The process id that detected a status change.</p>
1266 </dd>
1267 <dt>int rstatus [read-write]</dt>
1268 <dd>
1269 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1270<code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
608 </dd> 1271 </dd>
609</dl> 1272</dl>
1273<p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1274<pre> static void
1275 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1276 {
1277 ev_unloop (loop, EVUNLOOP_ALL);
1278 }
610 1279
1280 struct ev_signal signal_watcher;
1281 ev_signal_init (&amp;signal_watcher, sigint_cb, SIGINT);
1282 ev_signal_start (loop, &amp;sigint_cb);
1283
1284
1285
1286
1287</pre>
1288
611</div> 1289</div>
1290<h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1291<div id="code_ev_stat_code_did_the_file_attri-2">
1292<p>This watches a filesystem path for attribute changes. That is, it calls
1293<code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1294compared to the last time, invoking the callback if it did.</p>
1295<p>The path does not need to exist: changing from &quot;path exists&quot; to &quot;path does
1296not exist&quot; is a status change like any other. The condition &quot;path does
1297not exist&quot; is signified by the <code>st_nlink</code> field being zero (which is
1298otherwise always forced to be at least one) and all the other fields of
1299the stat buffer having unspecified contents.</p>
1300<p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1301relative and your working directory changes, the behaviour is undefined.</p>
1302<p>Since there is no standard to do this, the portable implementation simply
1303calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1304can specify a recommended polling interval for this case. If you specify
1305a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1306unspecified default</i> value will be used (which you can expect to be around
1307five seconds, although this might change dynamically). Libev will also
1308impose a minimum interval which is currently around <code>0.1</code>, but thats
1309usually overkill.</p>
1310<p>This watcher type is not meant for massive numbers of stat watchers,
1311as even with OS-supported change notifications, this can be
1312resource-intensive.</p>
1313<p>At the time of this writing, only the Linux inotify interface is
1314implemented (implementing kqueue support is left as an exercise for the
1315reader). Inotify will be used to give hints only and should not change the
1316semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1317to fall back to regular polling again even with inotify, but changes are
1318usually detected immediately, and if the file exists there will be no
1319polling.</p>
1320<dl>
1321 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1322 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1323 <dd>
1324 <p>Configures the watcher to wait for status changes of the given
1325<code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1326be detected and should normally be specified as <code>0</code> to let libev choose
1327a suitable value. The memory pointed to by <code>path</code> must point to the same
1328path for as long as the watcher is active.</p>
1329 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1330relative to the attributes at the time the watcher was started (or the
1331last change was detected).</p>
1332 </dd>
1333 <dt>ev_stat_stat (ev_stat *)</dt>
1334 <dd>
1335 <p>Updates the stat buffer immediately with new values. If you change the
1336watched path in your callback, you could call this fucntion to avoid
1337detecting this change (while introducing a race condition). Can also be
1338useful simply to find out the new values.</p>
1339 </dd>
1340 <dt>ev_statdata attr [read-only]</dt>
1341 <dd>
1342 <p>The most-recently detected attributes of the file. Although the type is of
1343<code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1344suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1345was some error while <code>stat</code>ing the file.</p>
1346 </dd>
1347 <dt>ev_statdata prev [read-only]</dt>
1348 <dd>
1349 <p>The previous attributes of the file. The callback gets invoked whenever
1350<code>prev</code> != <code>attr</code>.</p>
1351 </dd>
1352 <dt>ev_tstamp interval [read-only]</dt>
1353 <dd>
1354 <p>The specified interval.</p>
1355 </dd>
1356 <dt>const char *path [read-only]</dt>
1357 <dd>
1358 <p>The filesystem path that is being watched.</p>
1359 </dd>
1360</dl>
1361<p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1362<pre> static void
1363 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1364 {
1365 /* /etc/passwd changed in some way */
1366 if (w-&gt;attr.st_nlink)
1367 {
1368 printf (&quot;passwd current size %ld\n&quot;, (long)w-&gt;attr.st_size);
1369 printf (&quot;passwd current atime %ld\n&quot;, (long)w-&gt;attr.st_mtime);
1370 printf (&quot;passwd current mtime %ld\n&quot;, (long)w-&gt;attr.st_mtime);
1371 }
1372 else
1373 /* you shalt not abuse printf for puts */
1374 puts (&quot;wow, /etc/passwd is not there, expect problems. &quot;
1375 &quot;if this is windows, they already arrived\n&quot;);
1376 }
1377
1378 ...
1379 ev_stat passwd;
1380
1381 ev_stat_init (&amp;passwd, passwd_cb, &quot;/etc/passwd&quot;);
1382 ev_stat_start (loop, &amp;passwd);
1383
1384
1385
1386
1387</pre>
1388
1389</div>
612<h2 id="ev_idle_when_you_ve_got_nothing_bett">ev_idle - when you've got nothing better to do</h2> 1390<h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
613<div id="ev_idle_when_you_ve_got_nothing_bett-2"> 1391<div id="code_ev_idle_code_when_you_ve_got_no-2">
614<p>Idle watchers trigger events when there are no other I/O or timer (or 1392<p>Idle watchers trigger events when no other events of the same or higher
615periodic) events pending. That is, as long as your process is busy 1393priority are pending (prepare, check and other idle watchers do not
616handling sockets or timeouts it will not be called. But when your process 1394count).</p>
617is idle all idle watchers are being called again and again - until 1395<p>That is, as long as your process is busy handling sockets or timeouts
1396(or even signals, imagine) of the same or higher priority it will not be
1397triggered. But when your process is idle (or only lower-priority watchers
1398are pending), the idle watchers are being called once per event loop
618stopped, that is, or your process receives more events.</p> 1399iteration - until stopped, that is, or your process receives more events
1400and becomes busy again with higher priority stuff.</p>
619<p>The most noteworthy effect is that as long as any idle watchers are 1401<p>The most noteworthy effect is that as long as any idle watchers are
620active, the process will not block when waiting for new events.</p> 1402active, the process will not block when waiting for new events.</p>
621<p>Apart from keeping your process non-blocking (which is a useful 1403<p>Apart from keeping your process non-blocking (which is a useful
622effect on its own sometimes), idle watchers are a good place to do 1404effect on its own sometimes), idle watchers are a good place to do
623&quot;pseudo-background processing&quot;, or delay processing stuff to after the 1405&quot;pseudo-background processing&quot;, or delay processing stuff to after the
628 <p>Initialises and configures the idle watcher - it has no parameters of any 1410 <p>Initialises and configures the idle watcher - it has no parameters of any
629kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless, 1411kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
630believe me.</p> 1412believe me.</p>
631 </dd> 1413 </dd>
632</dl> 1414</dl>
1415<p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1416callback, free it. Also, use no error checking, as usual.</p>
1417<pre> static void
1418 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1419 {
1420 free (w);
1421 // now do something you wanted to do when the program has
1422 // no longer asnything immediate to do.
1423 }
633 1424
1425 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1426 ev_idle_init (idle_watcher, idle_cb);
1427 ev_idle_start (loop, idle_cb);
1428
1429
1430
1431
1432</pre>
1433
634</div> 1434</div>
635<h2 id="prepare_and_check_your_hooks_into_th">prepare and check - your hooks into the event loop</h2> 1435<h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
636<div id="prepare_and_check_your_hooks_into_th-2"> 1436<div id="code_ev_prepare_code_and_code_ev_che-2">
637<p>Prepare and check watchers usually (but not always) are used in 1437<p>Prepare and check watchers are usually (but not always) used in tandem:
638tandom. Prepare watchers get invoked before the process blocks and check 1438prepare watchers get invoked before the process blocks and check watchers
639watchers afterwards.</p> 1439afterwards.</p>
1440<p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1441the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1442watchers. Other loops than the current one are fine, however. The
1443rationale behind this is that you do not need to check for recursion in
1444those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1445<code>ev_check</code> so if you have one watcher of each kind they will always be
1446called in pairs bracketing the blocking call.</p>
640<p>Their main purpose is to integrate other event mechanisms into libev. This 1447<p>Their main purpose is to integrate other event mechanisms into libev and
641could be used, for example, to track variable changes, implement your own 1448their use is somewhat advanced. This could be used, for example, to track
642watchers, integrate net-snmp or a coroutine library and lots more.</p> 1449variable changes, implement your own watchers, integrate net-snmp or a
1450coroutine library and lots more. They are also occasionally useful if
1451you cache some data and want to flush it before blocking (for example,
1452in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1453watcher).</p>
643<p>This is done by examining in each prepare call which file descriptors need 1454<p>This is done by examining in each prepare call which file descriptors need
644to be watched by the other library, registering ev_io watchers for them 1455to be watched by the other library, registering <code>ev_io</code> watchers for
645and starting an ev_timer watcher for any timeouts (many libraries provide 1456them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
646just this functionality). Then, in the check watcher you check for any 1457provide just this functionality). Then, in the check watcher you check for
647events that occured (by making your callbacks set soem flags for example) 1458any events that occured (by checking the pending status of all watchers
648and call back into the library.</p> 1459and stopping them) and call back into the library. The I/O and timer
1460callbacks will never actually be called (but must be valid nevertheless,
1461because you never know, you know?).</p>
649<p>As another example, the perl Coro module uses these hooks to integrate 1462<p>As another example, the Perl Coro module uses these hooks to integrate
650coroutines into libev programs, by yielding to other active coroutines 1463coroutines into libev programs, by yielding to other active coroutines
651during each prepare and only letting the process block if no coroutines 1464during each prepare and only letting the process block if no coroutines
652are ready to run.</p> 1465are ready to run (it's actually more complicated: it only runs coroutines
1466with priority higher than or equal to the event loop and one coroutine
1467of lower priority, but only once, using idle watchers to keep the event
1468loop from blocking if lower-priority coroutines are active, thus mapping
1469low-priority coroutines to idle/background tasks).</p>
653<dl> 1470<dl>
654 <dt>ev_prepare_init (ev_prepare *, callback)</dt> 1471 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
655 <dt>ev_check_init (ev_check *, callback)</dt> 1472 <dt>ev_check_init (ev_check *, callback)</dt>
656 <dd> 1473 <dd>
657 <p>Initialises and configures the prepare or check watcher - they have no 1474 <p>Initialises and configures the prepare or check watcher - they have no
658parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code> 1475parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
659macros, but using them is utterly, utterly pointless.</p> 1476macros, but using them is utterly, utterly and completely pointless.</p>
660 </dd> 1477 </dd>
661</dl> 1478</dl>
1479<p>Example: To include a library such as adns, you would add IO watchers
1480and a timeout watcher in a prepare handler, as required by libadns, and
1481in a check watcher, destroy them and call into libadns. What follows is
1482pseudo-code only of course:</p>
1483<pre> static ev_io iow [nfd];
1484 static ev_timer tw;
662 1485
1486 static void
1487 io_cb (ev_loop *loop, ev_io *w, int revents)
1488 {
1489 // set the relevant poll flags
1490 // could also call adns_processreadable etc. here
1491 struct pollfd *fd = (struct pollfd *)w-&gt;data;
1492 if (revents &amp; EV_READ ) fd-&gt;revents |= fd-&gt;events &amp; POLLIN;
1493 if (revents &amp; EV_WRITE) fd-&gt;revents |= fd-&gt;events &amp; POLLOUT;
1494 }
1495
1496 // create io watchers for each fd and a timer before blocking
1497 static void
1498 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1499 {
1500 int timeout = 3600000;
1501 struct pollfd fds [nfd];
1502 // actual code will need to loop here and realloc etc.
1503 adns_beforepoll (ads, fds, &amp;nfd, &amp;timeout, timeval_from (ev_time ()));
1504
1505 /* the callback is illegal, but won't be called as we stop during check */
1506 ev_timer_init (&amp;tw, 0, timeout * 1e-3);
1507 ev_timer_start (loop, &amp;tw);
1508
1509 // create on ev_io per pollfd
1510 for (int i = 0; i &lt; nfd; ++i)
1511 {
1512 ev_io_init (iow + i, io_cb, fds [i].fd,
1513 ((fds [i].events &amp; POLLIN ? EV_READ : 0)
1514 | (fds [i].events &amp; POLLOUT ? EV_WRITE : 0)));
1515
1516 fds [i].revents = 0;
1517 iow [i].data = fds + i;
1518 ev_io_start (loop, iow + i);
1519 }
1520 }
1521
1522 // stop all watchers after blocking
1523 static void
1524 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1525 {
1526 ev_timer_stop (loop, &amp;tw);
1527
1528 for (int i = 0; i &lt; nfd; ++i)
1529 ev_io_stop (loop, iow + i);
1530
1531 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1532 }
1533
1534
1535
1536
1537</pre>
1538
663</div> 1539</div>
664<h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 1540<h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1541<div id="code_ev_embed_code_when_one_backend_-2">
1542<p>This is a rather advanced watcher type that lets you embed one event loop
1543into another (currently only <code>ev_io</code> events are supported in the embedded
1544loop, other types of watchers might be handled in a delayed or incorrect
1545fashion and must not be used).</p>
1546<p>There are primarily two reasons you would want that: work around bugs and
1547prioritise I/O.</p>
1548<p>As an example for a bug workaround, the kqueue backend might only support
1549sockets on some platform, so it is unusable as generic backend, but you
1550still want to make use of it because you have many sockets and it scales
1551so nicely. In this case, you would create a kqueue-based loop and embed it
1552into your default loop (which might use e.g. poll). Overall operation will
1553be a bit slower because first libev has to poll and then call kevent, but
1554at least you can use both at what they are best.</p>
1555<p>As for prioritising I/O: rarely you have the case where some fds have
1556to be watched and handled very quickly (with low latency), and even
1557priorities and idle watchers might have too much overhead. In this case
1558you would put all the high priority stuff in one loop and all the rest in
1559a second one, and embed the second one in the first.</p>
1560<p>As long as the watcher is active, the callback will be invoked every time
1561there might be events pending in the embedded loop. The callback must then
1562call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1563their callbacks (you could also start an idle watcher to give the embedded
1564loop strictly lower priority for example). You can also set the callback
1565to <code>0</code>, in which case the embed watcher will automatically execute the
1566embedded loop sweep.</p>
1567<p>As long as the watcher is started it will automatically handle events. The
1568callback will be invoked whenever some events have been handled. You can
1569set the callback to <code>0</code> to avoid having to specify one if you are not
1570interested in that.</p>
1571<p>Also, there have not currently been made special provisions for forking:
1572when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1573but you will also have to stop and restart any <code>ev_embed</code> watchers
1574yourself.</p>
1575<p>Unfortunately, not all backends are embeddable, only the ones returned by
1576<code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1577portable one.</p>
1578<p>So when you want to use this feature you will always have to be prepared
1579that you cannot get an embeddable loop. The recommended way to get around
1580this is to have a separate variables for your embeddable loop, try to
1581create it, and if that fails, use the normal loop for everything:</p>
1582<pre> struct ev_loop *loop_hi = ev_default_init (0);
1583 struct ev_loop *loop_lo = 0;
1584 struct ev_embed embed;
1585
1586 // see if there is a chance of getting one that works
1587 // (remember that a flags value of 0 means autodetection)
1588 loop_lo = ev_embeddable_backends () &amp; ev_recommended_backends ()
1589 ? ev_loop_new (ev_embeddable_backends () &amp; ev_recommended_backends ())
1590 : 0;
1591
1592 // if we got one, then embed it, otherwise default to loop_hi
1593 if (loop_lo)
1594 {
1595 ev_embed_init (&amp;embed, 0, loop_lo);
1596 ev_embed_start (loop_hi, &amp;embed);
1597 }
1598 else
1599 loop_lo = loop_hi;
1600
1601</pre>
1602<dl>
1603 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1604 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1605 <dd>
1606 <p>Configures the watcher to embed the given loop, which must be
1607embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1608invoked automatically, otherwise it is the responsibility of the callback
1609to invoke it (it will continue to be called until the sweep has been done,
1610if you do not want thta, you need to temporarily stop the embed watcher).</p>
1611 </dd>
1612 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1613 <dd>
1614 <p>Make a single, non-blocking sweep over the embedded loop. This works
1615similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1616apropriate way for embedded loops.</p>
1617 </dd>
1618 <dt>struct ev_loop *loop [read-only]</dt>
1619 <dd>
1620 <p>The embedded event loop.</p>
1621 </dd>
1622</dl>
1623
1624
1625
1626
1627
1628</div>
1629<h2 id="code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</h2>
1630<div id="code_ev_fork_code_the_audacity_to_re-2">
1631<p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1632whoever is a good citizen cared to tell libev about it by calling
1633<code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1634event loop blocks next and before <code>ev_check</code> watchers are being called,
1635and only in the child after the fork. If whoever good citizen calling
1636<code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1637handlers will be invoked, too, of course.</p>
1638<dl>
1639 <dt>ev_fork_init (ev_signal *, callback)</dt>
1640 <dd>
1641 <p>Initialises and configures the fork watcher - it has no parameters of any
1642kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1643believe me.</p>
1644 </dd>
1645</dl>
1646
1647
1648
1649
1650
1651</div>
1652<h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
665<div id="OTHER_FUNCTIONS_CONTENT"> 1653<div id="OTHER_FUNCTIONS_CONTENT">
666<p>There are some other fucntions of possible interest. Described. Here. Now.</p> 1654<p>There are some other functions of possible interest. Described. Here. Now.</p>
667<dl> 1655<dl>
668 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt> 1656 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
669 <dd> 1657 <dd>
670 <p>This function combines a simple timer and an I/O watcher, calls your 1658 <p>This function combines a simple timer and an I/O watcher, calls your
671callback on whichever event happens first and automatically stop both 1659callback on whichever event happens first and automatically stop both
672watchers. This is useful if you want to wait for a single event on an fd 1660watchers. This is useful if you want to wait for a single event on an fd
673or timeout without havign to allocate/configure/start/stop/free one or 1661or timeout without having to allocate/configure/start/stop/free one or
674more watchers yourself.</p> 1662more watchers yourself.</p>
675 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events is 1663 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
676ignored. Otherwise, an ev_io watcher for the given <code>fd</code> and <code>events</code> set 1664is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
677will be craeted and started.</p> 1665<code>events</code> set will be craeted and started.</p>
678 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be 1666 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
679started. Otherwise an ev_timer watcher with after = <code>timeout</code> (and repeat 1667started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
680= 0) will be started.</p> 1668repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1669dubious value.</p>
681 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and 1670 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
682gets passed an events set (normally a combination of EV_ERROR, EV_READ, 1671passed an <code>revents</code> set like normal event callbacks (a combination of
683EV_WRITE or EV_TIMEOUT) and the <code>arg</code> value passed to <code>ev_once</code>:</p> 1672<code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1673value passed to <code>ev_once</code>:</p>
684<pre> static void stdin_ready (int revents, void *arg) 1674<pre> static void stdin_ready (int revents, void *arg)
685 { 1675 {
686 if (revents &amp; EV_TIMEOUT) 1676 if (revents &amp; EV_TIMEOUT)
687 /* doh, nothing entered */ 1677 /* doh, nothing entered */;
688 else if (revents &amp; EV_READ) 1678 else if (revents &amp; EV_READ)
689 /* stdin might have data for us, joy! */ 1679 /* stdin might have data for us, joy! */;
690 } 1680 }
691 1681
692 ev_once (STDIN_FILENO, EV_READm 10., stdin_ready, 0); 1682 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
693 1683
694</pre> 1684</pre>
695 </dd> 1685 </dd>
696 <dt>ev_feed_event (loop, watcher, int events)</dt> 1686 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
697 <dd> 1687 <dd>
698 <p>Feeds the given event set into the event loop, as if the specified event 1688 <p>Feeds the given event set into the event loop, as if the specified event
699has happened for the specified watcher (which must be a pointer to an 1689had happened for the specified watcher (which must be a pointer to an
700initialised but not necessarily active event watcher).</p> 1690initialised but not necessarily started event watcher).</p>
701 </dd> 1691 </dd>
702 <dt>ev_feed_fd_event (loop, int fd, int revents)</dt> 1692 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
703 <dd> 1693 <dd>
704 <p>Feed an event on the given fd, as if a file descriptor backend detected it.</p> 1694 <p>Feed an event on the given fd, as if a file descriptor backend detected
1695the given events it.</p>
705 </dd> 1696 </dd>
706 <dt>ev_feed_signal_event (loop, int signum)</dt> 1697 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
707 <dd> 1698 <dd>
708 <p>Feed an event as if the given signal occured (loop must be the default loop!).</p> 1699 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1700loop!).</p>
709 </dd> 1701 </dd>
710</dl> 1702</dl>
711 1703
1704
1705
1706
1707
712</div> 1708</div>
713<h1 id="AUTHOR">AUTHOR</h1><p><a href="#TOP" class="toplink">Top</a></p> 1709<h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1710<div id="LIBEVENT_EMULATION_CONTENT">
1711<p>Libev offers a compatibility emulation layer for libevent. It cannot
1712emulate the internals of libevent, so here are some usage hints:</p>
1713<dl>
1714 <dt>* Use it by including &lt;event.h&gt;, as usual.</dt>
1715 <dt>* The following members are fully supported: ev_base, ev_callback,
1716ev_arg, ev_fd, ev_res, ev_events.</dt>
1717 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1718maintained by libev, it does not work exactly the same way as in libevent (consider
1719it a private API).</dt>
1720 <dt>* Priorities are not currently supported. Initialising priorities
1721will fail and all watchers will have the same priority, even though there
1722is an ev_pri field.</dt>
1723 <dt>* Other members are not supported.</dt>
1724 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1725to use the libev header file and library.</dt>
1726</dl>
1727
1728</div>
1729<h1 id="C_SUPPORT">C++ SUPPORT</h1>
1730<div id="C_SUPPORT_CONTENT">
1731<p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1732you to use some convinience methods to start/stop watchers and also change
1733the callback model to a model using method callbacks on objects.</p>
1734<p>To use it,</p>
1735<pre> #include &lt;ev++.h&gt;
1736
1737</pre>
1738<p>(it is not installed by default). This automatically includes <cite>ev.h</cite>
1739and puts all of its definitions (many of them macros) into the global
1740namespace. All C++ specific things are put into the <code>ev</code> namespace.</p>
1741<p>It should support all the same embedding options as <cite>ev.h</cite>, most notably
1742<code>EV_MULTIPLICITY</code>.</p>
1743<p>Here is a list of things available in the <code>ev</code> namespace:</p>
1744<dl>
1745 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1746 <dd>
1747 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1748macros from <cite>ev.h</cite>.</p>
1749 </dd>
1750 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1751 <dd>
1752 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1753 </dd>
1754 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1755 <dd>
1756 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1757the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1758which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1759defines by many implementations.</p>
1760 <p>All of those classes have these methods:</p>
1761 <p>
1762 <dl>
1763 <dt>ev::TYPE::TYPE (object *, object::method *)</dt>
1764 <dt>ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)</dt>
1765 <dt>ev::TYPE::~TYPE</dt>
1766 <dd>
1767 <p>The constructor takes a pointer to an object and a method pointer to
1768the event handler callback to call in this class. The constructor calls
1769<code>ev_init</code> for you, which means you have to call the <code>set</code> method
1770before starting it. If you do not specify a loop then the constructor
1771automatically associates the default loop with this watcher.</p>
1772 <p>The destructor automatically stops the watcher if it is active.</p>
1773 </dd>
1774 <dt>w-&gt;set (struct ev_loop *)</dt>
1775 <dd>
1776 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1777do this when the watcher is inactive (and not pending either).</p>
1778 </dd>
1779 <dt>w-&gt;set ([args])</dt>
1780 <dd>
1781 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1782called at least once. Unlike the C counterpart, an active watcher gets
1783automatically stopped and restarted.</p>
1784 </dd>
1785 <dt>w-&gt;start ()</dt>
1786 <dd>
1787 <p>Starts the watcher. Note that there is no <code>loop</code> argument as the
1788constructor already takes the loop.</p>
1789 </dd>
1790 <dt>w-&gt;stop ()</dt>
1791 <dd>
1792 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1793 </dd>
1794 <dt>w-&gt;again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1795 <dd>
1796 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1797<code>ev_TYPE_again</code> function.</p>
1798 </dd>
1799 <dt>w-&gt;sweep () <code>ev::embed</code> only</dt>
1800 <dd>
1801 <p>Invokes <code>ev_embed_sweep</code>.</p>
1802 </dd>
1803 <dt>w-&gt;update () <code>ev::stat</code> only</dt>
1804 <dd>
1805 <p>Invokes <code>ev_stat_stat</code>.</p>
1806 </dd>
1807 </dl>
1808 </p>
1809 </dd>
1810</dl>
1811<p>Example: Define a class with an IO and idle watcher, start one of them in
1812the constructor.</p>
1813<pre> class myclass
1814 {
1815 ev_io io; void io_cb (ev::io &amp;w, int revents);
1816 ev_idle idle void idle_cb (ev::idle &amp;w, int revents);
1817
1818 myclass ();
1819 }
1820
1821 myclass::myclass (int fd)
1822 : io (this, &amp;myclass::io_cb),
1823 idle (this, &amp;myclass::idle_cb)
1824 {
1825 io.start (fd, ev::READ);
1826 }
1827
1828
1829
1830
1831</pre>
1832
1833</div>
1834<h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
1835<div id="MACRO_MAGIC_CONTENT">
1836<p>Libev can be compiled with a variety of options, the most fundemantal is
1837<code>EV_MULTIPLICITY</code>. This option determines wether (most) functions and
1838callbacks have an initial <code>struct ev_loop *</code> argument.</p>
1839<p>To make it easier to write programs that cope with either variant, the
1840following macros are defined:</p>
1841<dl>
1842 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
1843 <dd>
1844 <p>This provides the loop <i>argument</i> for functions, if one is required (&quot;ev
1845loop argument&quot;). The <code>EV_A</code> form is used when this is the sole argument,
1846<code>EV_A_</code> is used when other arguments are following. Example:</p>
1847<pre> ev_unref (EV_A);
1848 ev_timer_add (EV_A_ watcher);
1849 ev_loop (EV_A_ 0);
1850
1851</pre>
1852 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
1853which is often provided by the following macro.</p>
1854 </dd>
1855 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
1856 <dd>
1857 <p>This provides the loop <i>parameter</i> for functions, if one is required (&quot;ev
1858loop parameter&quot;). The <code>EV_P</code> form is used when this is the sole parameter,
1859<code>EV_P_</code> is used when other parameters are following. Example:</p>
1860<pre> // this is how ev_unref is being declared
1861 static void ev_unref (EV_P);
1862
1863 // this is how you can declare your typical callback
1864 static void cb (EV_P_ ev_timer *w, int revents)
1865
1866</pre>
1867 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
1868suitable for use with <code>EV_A</code>.</p>
1869 </dd>
1870 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
1871 <dd>
1872 <p>Similar to the other two macros, this gives you the value of the default
1873loop, if multiple loops are supported (&quot;ev loop default&quot;).</p>
1874 </dd>
1875</dl>
1876<p>Example: Declare and initialise a check watcher, utilising the above
1877macros so it will work regardless of wether multiple loops are supported
1878or not.</p>
1879<pre> static void
1880 check_cb (EV_P_ ev_timer *w, int revents)
1881 {
1882 ev_check_stop (EV_A_ w);
1883 }
1884
1885 ev_check check;
1886 ev_check_init (&amp;check, check_cb);
1887 ev_check_start (EV_DEFAULT_ &amp;check);
1888 ev_loop (EV_DEFAULT_ 0);
1889
1890</pre>
1891
1892</div>
1893<h1 id="EMBEDDING">EMBEDDING</h1>
1894<div id="EMBEDDING_CONTENT">
1895<p>Libev can (and often is) directly embedded into host
1896applications. Examples of applications that embed it include the Deliantra
1897Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1898and rxvt-unicode.</p>
1899<p>The goal is to enable you to just copy the neecssary files into your
1900source directory without having to change even a single line in them, so
1901you can easily upgrade by simply copying (or having a checked-out copy of
1902libev somewhere in your source tree).</p>
1903
1904</div>
1905<h2 id="FILESETS">FILESETS</h2>
1906<div id="FILESETS_CONTENT">
1907<p>Depending on what features you need you need to include one or more sets of files
1908in your app.</p>
1909
1910</div>
1911<h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
1912<div id="CORE_EVENT_LOOP_CONTENT">
1913<p>To include only the libev core (all the <code>ev_*</code> functions), with manual
1914configuration (no autoconf):</p>
1915<pre> #define EV_STANDALONE 1
1916 #include &quot;ev.c&quot;
1917
1918</pre>
1919<p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
1920single C source file only to provide the function implementations. To use
1921it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
1922done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
1923where you can put other configuration options):</p>
1924<pre> #define EV_STANDALONE 1
1925 #include &quot;ev.h&quot;
1926
1927</pre>
1928<p>Both header files and implementation files can be compiled with a C++
1929compiler (at least, thats a stated goal, and breakage will be treated
1930as a bug).</p>
1931<p>You need the following files in your source tree, or in a directory
1932in your include path (e.g. in libev/ when using -Ilibev):</p>
1933<pre> ev.h
1934 ev.c
1935 ev_vars.h
1936 ev_wrap.h
1937
1938 ev_win32.c required on win32 platforms only
1939
1940 ev_select.c only when select backend is enabled (which is enabled by default)
1941 ev_poll.c only when poll backend is enabled (disabled by default)
1942 ev_epoll.c only when the epoll backend is enabled (disabled by default)
1943 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
1944 ev_port.c only when the solaris port backend is enabled (disabled by default)
1945
1946</pre>
1947<p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
1948to compile this single file.</p>
1949
1950</div>
1951<h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
1952<div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
1953<p>To include the libevent compatibility API, also include:</p>
1954<pre> #include &quot;event.c&quot;
1955
1956</pre>
1957<p>in the file including <cite>ev.c</cite>, and:</p>
1958<pre> #include &quot;event.h&quot;
1959
1960</pre>
1961<p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
1962<p>You need the following additional files for this:</p>
1963<pre> event.h
1964 event.c
1965
1966</pre>
1967
1968</div>
1969<h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
1970<div id="AUTOCONF_SUPPORT_CONTENT">
1971<p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
1972whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
1973<cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
1974include <cite>config.h</cite> and configure itself accordingly.</p>
1975<p>For this of course you need the m4 file:</p>
1976<pre> libev.m4
1977
1978</pre>
1979
1980</div>
1981<h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
1982<div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
1983<p>Libev can be configured via a variety of preprocessor symbols you have to define
1984before including any of its files. The default is not to build for multiplicity
1985and only include the select backend.</p>
1986<dl>
1987 <dt>EV_STANDALONE</dt>
1988 <dd>
1989 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
1990keeps libev from including <cite>config.h</cite>, and it also defines dummy
1991implementations for some libevent functions (such as logging, which is not
1992supported). It will also not define any of the structs usually found in
1993<cite>event.h</cite> that are not directly supported by the libev core alone.</p>
1994 </dd>
1995 <dt>EV_USE_MONOTONIC</dt>
1996 <dd>
1997 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
1998monotonic clock option at both compiletime and runtime. Otherwise no use
1999of the monotonic clock option will be attempted. If you enable this, you
2000usually have to link against librt or something similar. Enabling it when
2001the functionality isn't available is safe, though, althoguh you have
2002to make sure you link against any libraries where the <code>clock_gettime</code>
2003function is hiding in (often <cite>-lrt</cite>).</p>
2004 </dd>
2005 <dt>EV_USE_REALTIME</dt>
2006 <dd>
2007 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2008realtime clock option at compiletime (and assume its availability at
2009runtime if successful). Otherwise no use of the realtime clock option will
2010be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
2011(CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
2012in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
2013 </dd>
2014 <dt>EV_USE_SELECT</dt>
2015 <dd>
2016 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
2017<code>select</code>(2) backend. No attempt at autodetection will be done: if no
2018other method takes over, select will be it. Otherwise the select backend
2019will not be compiled in.</p>
2020 </dd>
2021 <dt>EV_SELECT_USE_FD_SET</dt>
2022 <dd>
2023 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
2024structure. This is useful if libev doesn't compile due to a missing
2025<code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
2026exotic systems. This usually limits the range of file descriptors to some
2027low limit such as 1024 or might have other limitations (winsocket only
2028allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
2029influence the size of the <code>fd_set</code> used.</p>
2030 </dd>
2031 <dt>EV_SELECT_IS_WINSOCKET</dt>
2032 <dd>
2033 <p>When defined to <code>1</code>, the select backend will assume that
2034select/socket/connect etc. don't understand file descriptors but
2035wants osf handles on win32 (this is the case when the select to
2036be used is the winsock select). This means that it will call
2037<code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2038it is assumed that all these functions actually work on fds, even
2039on win32. Should not be defined on non-win32 platforms.</p>
2040 </dd>
2041 <dt>EV_USE_POLL</dt>
2042 <dd>
2043 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2044backend. Otherwise it will be enabled on non-win32 platforms. It
2045takes precedence over select.</p>
2046 </dd>
2047 <dt>EV_USE_EPOLL</dt>
2048 <dd>
2049 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2050<code>epoll</code>(7) backend. Its availability will be detected at runtime,
2051otherwise another method will be used as fallback. This is the
2052preferred backend for GNU/Linux systems.</p>
2053 </dd>
2054 <dt>EV_USE_KQUEUE</dt>
2055 <dd>
2056 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2057<code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2058otherwise another method will be used as fallback. This is the preferred
2059backend for BSD and BSD-like systems, although on most BSDs kqueue only
2060supports some types of fds correctly (the only platform we found that
2061supports ptys for example was NetBSD), so kqueue might be compiled in, but
2062not be used unless explicitly requested. The best way to use it is to find
2063out whether kqueue supports your type of fd properly and use an embedded
2064kqueue loop.</p>
2065 </dd>
2066 <dt>EV_USE_PORT</dt>
2067 <dd>
2068 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
206910 port style backend. Its availability will be detected at runtime,
2070otherwise another method will be used as fallback. This is the preferred
2071backend for Solaris 10 systems.</p>
2072 </dd>
2073 <dt>EV_USE_DEVPOLL</dt>
2074 <dd>
2075 <p>reserved for future expansion, works like the USE symbols above.</p>
2076 </dd>
2077 <dt>EV_USE_INOTIFY</dt>
2078 <dd>
2079 <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2080interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2081be detected at runtime.</p>
2082 </dd>
2083 <dt>EV_H</dt>
2084 <dd>
2085 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2086undefined is <code>&lt;ev.h&gt;</code> in <cite>event.h</cite> and <code>&quot;ev.h&quot;</code> in <cite>ev.c</cite>. This
2087can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2088 </dd>
2089 <dt>EV_CONFIG_H</dt>
2090 <dd>
2091 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2092<cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2093<code>EV_H</code>, above.</p>
2094 </dd>
2095 <dt>EV_EVENT_H</dt>
2096 <dd>
2097 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2098of how the <cite>event.h</cite> header can be found.</p>
2099 </dd>
2100 <dt>EV_PROTOTYPES</dt>
2101 <dd>
2102 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2103prototypes, but still define all the structs and other symbols. This is
2104occasionally useful if you want to provide your own wrapper functions
2105around libev functions.</p>
2106 </dd>
2107 <dt>EV_MULTIPLICITY</dt>
2108 <dd>
2109 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2110will have the <code>struct ev_loop *</code> as first argument, and you can create
2111additional independent event loops. Otherwise there will be no support
2112for multiple event loops and there is no first event loop pointer
2113argument. Instead, all functions act on the single default loop.</p>
2114 </dd>
2115 <dt>EV_PERIODIC_ENABLE</dt>
2116 <dd>
2117 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2118defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2119code.</p>
2120 </dd>
2121 <dt>EV_IDLE_ENABLE</dt>
2122 <dd>
2123 <p>If undefined or defined to be <code>1</code>, then idle watchers are supported. If
2124defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2125code.</p>
2126 </dd>
2127 <dt>EV_EMBED_ENABLE</dt>
2128 <dd>
2129 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2130defined to be <code>0</code>, then they are not.</p>
2131 </dd>
2132 <dt>EV_STAT_ENABLE</dt>
2133 <dd>
2134 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2135defined to be <code>0</code>, then they are not.</p>
2136 </dd>
2137 <dt>EV_FORK_ENABLE</dt>
2138 <dd>
2139 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2140defined to be <code>0</code>, then they are not.</p>
2141 </dd>
2142 <dt>EV_MINIMAL</dt>
2143 <dd>
2144 <p>If you need to shave off some kilobytes of code at the expense of some
2145speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2146some inlining decisions, saves roughly 30% codesize of amd64.</p>
2147 </dd>
2148 <dt>EV_PID_HASHSIZE</dt>
2149 <dd>
2150 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2151pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2152than enough. If you need to manage thousands of children you might want to
2153increase this value (<i>must</i> be a power of two).</p>
2154 </dd>
2155 <dt>EV_INOTIFY_HASHSIZE</dt>
2156 <dd>
2157 <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2158inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2159usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2160watchers you might want to increase this value (<i>must</i> be a power of
2161two).</p>
2162 </dd>
2163 <dt>EV_COMMON</dt>
2164 <dd>
2165 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2166this macro to a something else you can include more and other types of
2167members. You have to define it each time you include one of the files,
2168though, and it must be identical each time.</p>
2169 <p>For example, the perl EV module uses something like this:</p>
2170<pre> #define EV_COMMON \
2171 SV *self; /* contains this struct */ \
2172 SV *cb_sv, *fh /* note no trailing &quot;;&quot; */
2173
2174</pre>
2175 </dd>
2176 <dt>EV_CB_DECLARE (type)</dt>
2177 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2178 <dt>ev_set_cb (ev, cb)</dt>
2179 <dd>
2180 <p>Can be used to change the callback member declaration in each watcher,
2181and the way callbacks are invoked and set. Must expand to a struct member
2182definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2183their default definitions. One possible use for overriding these is to
2184avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2185method calls instead of plain function calls in C++.</p>
2186
2187</div>
2188<h2 id="EXAMPLES">EXAMPLES</h2>
2189<div id="EXAMPLES_CONTENT">
2190 <p>For a real-world example of a program the includes libev
2191verbatim, you can have a look at the EV perl module
2192(<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2193the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2194interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2195will be compiled. It is pretty complex because it provides its own header
2196file.</p>
2197 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2198that everybody includes and which overrides some configure choices:</p>
2199<pre> #define EV_MINIMAL 1
2200 #define EV_USE_POLL 0
2201 #define EV_MULTIPLICITY 0
2202 #define EV_PERIODIC_ENABLE 0
2203 #define EV_STAT_ENABLE 0
2204 #define EV_FORK_ENABLE 0
2205 #define EV_CONFIG_H &lt;config.h&gt;
2206 #define EV_MINPRI 0
2207 #define EV_MAXPRI 0
2208
2209 #include &quot;ev++.h&quot;
2210
2211</pre>
2212 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2213<pre> #include &quot;ev_cpp.h&quot;
2214 #include &quot;ev.c&quot;
2215
2216
2217
2218
2219</pre>
2220
2221</div>
2222<h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2223<div id="COMPLEXITIES_CONTENT">
2224 <p>In this section the complexities of (many of) the algorithms used inside
2225libev will be explained. For complexity discussions about backends see the
2226documentation for <code>ev_default_init</code>.</p>
2227 <p>
2228 <dl>
2229 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2230 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2231 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2232 <dt>Stopping check/prepare/idle watchers: O(1)</dt>
2233 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2234 <dt>Finding the next timer per loop iteration: O(1)</dt>
2235 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2236 <dt>Activating one watcher: O(1)</dt>
2237 </dl>
2238 </p>
2239
2240
2241
2242
2243
2244</div>
2245<h1 id="AUTHOR">AUTHOR</h1>
714<div id="AUTHOR_CONTENT"> 2246<div id="AUTHOR_CONTENT">
715<p>Marc Lehmann &lt;libev@schmorp.de&gt;.</p> 2247 <p>Marc Lehmann &lt;libev@schmorp.de&gt;.</p>
716 2248
717</div> 2249</div>
718</div></body> 2250</div></body>
719</html> 2251</html>

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines