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

Comparing libev/ev.html (file contents):
Revision 1.5 by root, Mon Nov 12 08:11:56 2007 UTC vs.
Revision 1.79 by root, Wed Dec 12 17:55:30 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 09:11:56 2007" /> 9 <meta name="created" content="Wed Dec 12 18:55:28 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>
33<ul><li><a href="#The_special_problem_of_disappearing_">The special problem of disappearing file descriptors</a></li>
34<li><a href="#Watcher_Specific_Functions">Watcher-Specific Functions</a></li>
35</ul>
36</li>
30<li><a href="#struct_ev_timer_relative_and_optiona">struct ev_timer - relative and optionally recurring timeouts</a></li> 37<li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a>
38<ul><li><a href="#Watcher_Specific_Functions_and_Data_">Watcher-Specific Functions and Data Members</a></li>
39</ul>
40</li>
31<li><a href="#ev_periodic_to_cron_or_not_to_cron_i">ev_periodic - to cron or not to cron it</a></li> 41<li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a>
42<ul><li><a href="#Watcher_Specific_Functions_and_Data_-3">Watcher-Specific Functions and Data Members</a></li>
43</ul>
44</li>
32<li><a href="#ev_signal_signal_me_when_a_signal_ge">ev_signal - signal me when a signal gets signalled</a></li> 45<li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a>
33<li><a href="#ev_child_wait_for_pid_status_changes">ev_child - wait for pid status changes</a></li> 46<ul><li><a href="#Watcher_Specific_Functions_and_Data_-4">Watcher-Specific Functions and Data Members</a></li>
47</ul>
48</li>
49<li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a>
50<ul><li><a href="#Watcher_Specific_Functions_and_Data_-5">Watcher-Specific Functions and Data Members</a></li>
51</ul>
52</li>
53<li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a>
54<ul><li><a href="#Watcher_Specific_Functions_and_Data_-6">Watcher-Specific Functions and Data Members</a></li>
55</ul>
56</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> 57<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>
35<li><a href="#prepare_and_check_your_hooks_into_th">prepare and check - your hooks into the event loop</a></li> 58<ul><li><a href="#Watcher_Specific_Functions_and_Data_-7">Watcher-Specific Functions and Data Members</a></li>
59</ul>
60</li>
61<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>
62<ul><li><a href="#Watcher_Specific_Functions_and_Data_-8">Watcher-Specific Functions and Data Members</a></li>
63</ul>
64</li>
65<li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a>
66<ul><li><a href="#Watcher_Specific_Functions_and_Data_-9">Watcher-Specific Functions and Data Members</a></li>
67</ul>
68</li>
69<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>
70<ul><li><a href="#Watcher_Specific_Functions_and_Data_-10">Watcher-Specific Functions and Data Members</a></li>
71</ul>
72</li>
36</ul> 73</ul>
37</li> 74</li>
38<li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li> 75<li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
76<li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
77<li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
78<li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
79<li><a href="#EMBEDDING">EMBEDDING</a>
80<ul><li><a href="#FILESETS">FILESETS</a>
81<ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
82<li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
83<li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
84</ul>
85</li>
86<li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
87<li><a href="#EXAMPLES">EXAMPLES</a></li>
88</ul>
89</li>
90<li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
39<li><a href="#AUTHOR">AUTHOR</a> 91<li><a href="#AUTHOR">AUTHOR</a>
40</li> 92</li>
41</ul><hr /> 93</ul><hr />
42<!-- INDEX END --> 94<!-- INDEX END -->
43 95
44<h1 id="NAME">NAME</h1><p><a href="#TOP" class="toplink">Top</a></p> 96<h1 id="NAME">NAME</h1>
45<div id="NAME_CONTENT"> 97<div id="NAME_CONTENT">
46<p>libev - a high performance full-featured event loop written in C</p> 98<p>libev - a high performance full-featured event loop written in C</p>
47 99
48</div> 100</div>
49<h1 id="SYNOPSIS">SYNOPSIS</h1><p><a href="#TOP" class="toplink">Top</a></p> 101<h1 id="SYNOPSIS">SYNOPSIS</h1>
50<div id="SYNOPSIS_CONTENT"> 102<div id="SYNOPSIS_CONTENT">
51<pre> #include &lt;ev.h&gt; 103<pre> #include &lt;ev.h&gt;
52 104
53</pre> 105</pre>
54 106
55</div> 107</div>
56<h1 id="DESCRIPTION">DESCRIPTION</h1><p><a href="#TOP" class="toplink">Top</a></p> 108<h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
109<div id="EXAMPLE_PROGRAM_CONTENT">
110<pre> #include &lt;ev.h&gt;
111
112 ev_io stdin_watcher;
113 ev_timer timeout_watcher;
114
115 /* called when data readable on stdin */
116 static void
117 stdin_cb (EV_P_ struct ev_io *w, int revents)
118 {
119 /* puts (&quot;stdin ready&quot;); */
120 ev_io_stop (EV_A_ w); /* just a syntax example */
121 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
122 }
123
124 static void
125 timeout_cb (EV_P_ struct ev_timer *w, int revents)
126 {
127 /* puts (&quot;timeout&quot;); */
128 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
129 }
130
131 int
132 main (void)
133 {
134 struct ev_loop *loop = ev_default_loop (0);
135
136 /* initialise an io watcher, then start it */
137 ev_io_init (&amp;stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
138 ev_io_start (loop, &amp;stdin_watcher);
139
140 /* simple non-repeating 5.5 second timeout */
141 ev_timer_init (&amp;timeout_watcher, timeout_cb, 5.5, 0.);
142 ev_timer_start (loop, &amp;timeout_watcher);
143
144 /* loop till timeout or data ready */
145 ev_loop (loop, 0);
146
147 return 0;
148 }
149
150</pre>
151
152</div>
153<h1 id="DESCRIPTION">DESCRIPTION</h1>
57<div id="DESCRIPTION_CONTENT"> 154<div id="DESCRIPTION_CONTENT">
155<p>The newest version of this document is also available as a html-formatted
156web page you might find easier to navigate when reading it for the first
157time: <a href="http://cvs.schmorp.de/libev/ev.html">http://cvs.schmorp.de/libev/ev.html</a>.</p>
58<p>Libev is an event loop: you register interest in certain events (such as a 158<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 159file descriptor being readable or a timeout occuring), and it will manage
60these event sources and provide your program with events.</p> 160these 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 161<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 162(or thread) by executing the <i>event loop</i> handler, and will then
65watchers</i>, which are relatively small C structures you initialise with the 165watchers</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 166details of the event, and then hand it over to libev by <i>starting</i> the
67watcher.</p> 167watcher.</p>
68 168
69</div> 169</div>
70<h1 id="FEATURES">FEATURES</h1><p><a href="#TOP" class="toplink">Top</a></p> 170<h1 id="FEATURES">FEATURES</h1>
71<div id="FEATURES_CONTENT"> 171<div id="FEATURES_CONTENT">
72<p>Libev supports select, poll, the linux-specific epoll and the bsd-specific 172<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 173BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
74timers with customised rescheduling, signal events, process status change 174for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
75events (related to SIGCHLD), and event watchers dealing with the event 175(for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
76loop mechanism itself (idle, prepare and check watchers). It also is quite 176with customised rescheduling (<code>ev_periodic</code>), synchronous signals
77fast (see a <b>http://libev.schmorp.de/bench.html</b> (<cite>benchmark</cite>) comparing it 177(<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
78to libevent).</p> 178watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
179<code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
180file watchers (<code>ev_stat</code>) and even limited support for fork events
181(<code>ev_fork</code>).</p>
182<p>It also is quite fast (see this
183<a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
184for example).</p>
79 185
80</div> 186</div>
81<h1 id="CONVENTIONS">CONVENTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 187<h1 id="CONVENTIONS">CONVENTIONS</h1>
82<div id="CONVENTIONS_CONTENT"> 188<div id="CONVENTIONS_CONTENT">
83<p>Libev is very configurable. In this manual the default configuration 189<p>Libev is very configurable. In this manual the default configuration will
84will be described, which supports multiple event loops. For more info 190be described, which supports multiple event loops. For more info about
85about various configuraiton options please have a look at the file 191various configuration options please have a look at <strong>EMBED</strong> section in
86<cite>README.embed</cite> in the libev distribution. If libev was configured without 192this manual. If libev was configured without support for multiple event
87support for multiple event loops, then all functions taking an initial 193loops, then all functions taking an initial argument of name <code>loop</code>
88argument of name <code>loop</code> (which is always of type <code>struct ev_loop *</code>) 194(which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
89will not have this argument.</p>
90 195
91</div> 196</div>
92<h1 id="TIME_AND_OTHER_GLOBAL_FUNCTIONS">TIME AND OTHER GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 197<h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
93<div id="TIME_AND_OTHER_GLOBAL_FUNCTIONS_CONT"> 198<div id="TIME_REPRESENTATION_CONTENT">
94<p>Libev represents time as a single floating point number, representing the 199<p>Libev represents time as a single floating point number, representing the
95(fractional) number of seconds since the (POSIX) epoch (somewhere near 200(fractional) number of seconds since the (POSIX) epoch (somewhere near
96the beginning of 1970, details are complicated, don't ask). This type is 201the beginning of 1970, details are complicated, don't ask). This type is
97called <code>ev_tstamp</code>, which is what you should use too. It usually aliases 202called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
98to the double type in C.</p> 203to the <code>double</code> type in C, and when you need to do any calculations on
204it, you should treat it as such.</p>
205
206</div>
207<h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
208<div id="GLOBAL_FUNCTIONS_CONTENT">
209<p>These functions can be called anytime, even before initialising the
210library in any way.</p>
99<dl> 211<dl>
100 <dt>ev_tstamp ev_time ()</dt> 212 <dt>ev_tstamp ev_time ()</dt>
101 <dd> 213 <dd>
102 <p>Returns the current time as libev would use it.</p> 214 <p>Returns the current time as libev would use it. Please note that the
215<code>ev_now</code> function is usually faster and also often returns the timestamp
216you actually want to know.</p>
103 </dd> 217 </dd>
104 <dt>int ev_version_major ()</dt> 218 <dt>int ev_version_major ()</dt>
105 <dt>int ev_version_minor ()</dt> 219 <dt>int ev_version_minor ()</dt>
106 <dd> 220 <dd>
107 <p>You can find out the major and minor version numbers of the library 221 <p>You can find out the major and minor ABI version numbers of the library
108you linked against by calling the functions <code>ev_version_major</code> and 222you linked against by calling the functions <code>ev_version_major</code> and
109<code>ev_version_minor</code>. If you want, you can compare against the global 223<code>ev_version_minor</code>. If you want, you can compare against the global
110symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the 224symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
111version of the library your program was compiled against.</p> 225version of the library your program was compiled against.</p>
226 <p>These version numbers refer to the ABI version of the library, not the
227release version.</p>
112 <p>Usually, its a good idea to terminate if the major versions mismatch, 228 <p>Usually, it's a good idea to terminate if the major versions mismatch,
113as this indicates an incompatible change. Minor versions are usually 229as this indicates an incompatible change. Minor versions are usually
114compatible to older versions, so a larger minor version alone is usually 230compatible to older versions, so a larger minor version alone is usually
115not a problem.</p> 231not a problem.</p>
232 <p>Example: Make sure we haven't accidentally been linked against the wrong
233version.</p>
234<pre> assert ((&quot;libev version mismatch&quot;,
235 ev_version_major () == EV_VERSION_MAJOR
236 &amp;&amp; ev_version_minor () &gt;= EV_VERSION_MINOR));
237
238</pre>
239 </dd>
240 <dt>unsigned int ev_supported_backends ()</dt>
241 <dd>
242 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
243value) compiled into this binary of libev (independent of their
244availability on the system you are running on). See <code>ev_default_loop</code> for
245a description of the set values.</p>
246 <p>Example: make sure we have the epoll method, because yeah this is cool and
247a must have and can we have a torrent of it please!!!11</p>
248<pre> assert ((&quot;sorry, no epoll, no sex&quot;,
249 ev_supported_backends () &amp; EVBACKEND_EPOLL));
250
251</pre>
252 </dd>
253 <dt>unsigned int ev_recommended_backends ()</dt>
254 <dd>
255 <p>Return the set of all backends compiled into this binary of libev and also
256recommended for this platform. This set is often smaller than the one
257returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
258most BSDs and will not be autodetected unless you explicitly request it
259(assuming you know what you are doing). This is the set of backends that
260libev will probe for if you specify no backends explicitly.</p>
261 </dd>
262 <dt>unsigned int ev_embeddable_backends ()</dt>
263 <dd>
264 <p>Returns the set of backends that are embeddable in other event loops. This
265is the theoretical, all-platform, value. To find which backends
266might be supported on the current system, you would need to look at
267<code>ev_embeddable_backends () &amp; ev_supported_backends ()</code>, likewise for
268recommended ones.</p>
269 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
116 </dd> 270 </dd>
117 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt> 271 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
118 <dd> 272 <dd>
119 <p>Sets the allocation function to use (the prototype is similar to the 273 <p>Sets the allocation function to use (the prototype is similar - the
120realloc function). It is used to allocate and free memory (no surprises 274semantics is identical - to the realloc C function). It is used to
121here). If it returns zero when memory needs to be allocated, the library 275allocate and free memory (no surprises here). If it returns zero when
122might abort or take some potentially destructive action. The default is 276memory needs to be allocated, the library might abort or take some
123your system realloc function.</p> 277potentially destructive action. The default is your system realloc
278function.</p>
124 <p>You could override this function in high-availability programs to, say, 279 <p>You could override this function in high-availability programs to, say,
125free some memory if it cannot allocate memory, to use a special allocator, 280free some memory if it cannot allocate memory, to use a special allocator,
126or even to sleep a while and retry until some memory is available.</p> 281or even to sleep a while and retry until some memory is available.</p>
282 <p>Example: Replace the libev allocator with one that waits a bit and then
283retries).</p>
284<pre> static void *
285 persistent_realloc (void *ptr, size_t size)
286 {
287 for (;;)
288 {
289 void *newptr = realloc (ptr, size);
290
291 if (newptr)
292 return newptr;
293
294 sleep (60);
295 }
296 }
297
298 ...
299 ev_set_allocator (persistent_realloc);
300
301</pre>
127 </dd> 302 </dd>
128 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt> 303 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
129 <dd> 304 <dd>
130 <p>Set the callback function to call on a retryable syscall error (such 305 <p>Set the callback function to call on a retryable syscall error (such
131as failed select, poll, epoll_wait). The message is a printable string 306as failed select, poll, epoll_wait). The message is a printable string
132indicating the system call or subsystem causing the problem. If this 307indicating the system call or subsystem causing the problem. If this
133callback is set, then libev will expect it to remedy the sitution, no 308callback is set, then libev will expect it to remedy the sitution, no
134matter what, when it returns. That is, libev will geenrally retry the 309matter what, when it returns. That is, libev will generally retry the
135requested operation, or, if the condition doesn't go away, do bad stuff 310requested operation, or, if the condition doesn't go away, do bad stuff
136(such as abort).</p> 311(such as abort).</p>
312 <p>Example: This is basically the same thing that libev does internally, too.</p>
313<pre> static void
314 fatal_error (const char *msg)
315 {
316 perror (msg);
317 abort ();
318 }
319
320 ...
321 ev_set_syserr_cb (fatal_error);
322
323</pre>
137 </dd> 324 </dd>
138</dl> 325</dl>
139 326
140</div> 327</div>
141<h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p> 328<h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
142<div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2"> 329<div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
143<p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two 330<p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
144types of such loops, the <i>default</i> loop, which supports signals and child 331types of such loops, the <i>default</i> loop, which supports signals and child
145events, and dynamically created loops which do not.</p> 332events, and dynamically created loops which do not.</p>
146<p>If you use threads, a common model is to run the default event loop 333<p>If you use threads, a common model is to run the default event loop
147in your main thread (or in a separate thrad) and for each thread you 334in your main thread (or in a separate thread) and for each thread you
148create, you also create another event loop. Libev itself does no lockign 335create, you also create another event loop. Libev itself does no locking
149whatsoever, so if you mix calls to different event loops, make sure you 336whatsoever, so if you mix calls to the same event loop in different
150lock (this is usually a bad idea, though, even if done right).</p> 337threads, make sure you lock (this is usually a bad idea, though, even if
338done correctly, because it's hideous and inefficient).</p>
151<dl> 339<dl>
152 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt> 340 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
153 <dd> 341 <dd>
154 <p>This will initialise the default event loop if it hasn't been initialised 342 <p>This will initialise the default event loop if it hasn't been initialised
155yet and return it. If the default loop could not be initialised, returns 343yet and return it. If the default loop could not be initialised, returns
156false. If it already was initialised it simply returns it (and ignores the 344false. If it already was initialised it simply returns it (and ignores the
157flags).</p> 345flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
158 <p>If you don't know what event loop to use, use the one returned from this 346 <p>If you don't know what event loop to use, use the one returned from this
159function.</p> 347function.</p>
160 <p>The flags argument can be used to specify special behaviour or specific 348 <p>The flags argument can be used to specify special behaviour or specific
161backends to use, and is usually specified as 0 (or EVFLAG_AUTO)</p> 349backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
162 <p>It supports the following flags:</p> 350 <p>The following flags are supported:</p>
163 <p> 351 <p>
164 <dl> 352 <dl>
165 <dt>EVFLAG_AUTO</dt> 353 <dt><code>EVFLAG_AUTO</code></dt>
166 <dd> 354 <dd>
167 <p>The default flags value. Use this if you have no clue (its the right 355 <p>The default flags value. Use this if you have no clue (it's the right
168thing, believe me).</p> 356thing, believe me).</p>
169 </dd> 357 </dd>
170 <dt>EVFLAG_NOENV</dt> 358 <dt><code>EVFLAG_NOENV</code></dt>
171 <dd>
172 <p>If this flag bit is ored into the flag value then libev will <i>not</i> look
173at the environment variable <code>LIBEV_FLAGS</code>. Otherwise (the default), this
174environment variable will override the flags completely. This is useful
175to try out specific backends to tets their performance, or to work around
176bugs.</p>
177 </dd> 359 <dd>
178 <dt>EVMETHOD_SELECT portable select backend</dt> 360 <p>If this flag bit is ored into the flag value (or the program runs setuid
179 <dt>EVMETHOD_POLL poll backend (everywhere except windows)</dt> 361or setgid) then libev will <i>not</i> look at the environment variable
180 <dt>EVMETHOD_EPOLL linux only</dt> 362<code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
181 <dt>EVMETHOD_KQUEUE some bsds only</dt> 363override the flags completely if it is found in the environment. This is
182 <dt>EVMETHOD_DEVPOLL solaris 8 only</dt> 364useful to try out specific backends to test their performance, or to work
183 <dt>EVMETHOD_PORT solaris 10 only</dt> 365around bugs.</p>
184 <dd> 366 </dd>
185 <p>If one or more of these are ored into the flags value, then only these 367 <dt><code>EVFLAG_FORKCHECK</code></dt>
186backends will be tried (in the reverse order as given here). If one are 368 <dd>
187specified, any backend will do.</p> 369 <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
370a fork, you can also make libev check for a fork in each iteration by
371enabling this flag.</p>
372 <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
373and thus this might slow down your event loop if you do a lot of loop
374iterations and little real work, but is usually not noticeable (on my
375Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
376without a syscall and thus <i>very</i> fast, but my Linux system also has
377<code>pthread_atfork</code> which is even faster).</p>
378 <p>The big advantage of this flag is that you can forget about fork (and
379forget about forgetting to tell libev about forking) when you use this
380flag.</p>
381 <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
382environment variable.</p>
383 </dd>
384 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
385 <dd>
386 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
387libev tries to roll its own fd_set with no limits on the number of fds,
388but if that fails, expect a fairly low limit on the number of fds when
389using this backend. It doesn't scale too well (O(highest_fd)), but its usually
390the fastest backend for a low number of fds.</p>
391 </dd>
392 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
393 <dd>
394 <p>And this is your standard poll(2) backend. It's more complicated than
395select, but handles sparse fds better and has no artificial limit on the
396number of fds you can use (except it will slow down considerably with a
397lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
398 </dd>
399 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
400 <dd>
401 <p>For few fds, this backend is a bit little slower than poll and select,
402but it scales phenomenally better. While poll and select usually scale like
403O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
404either O(1) or O(active_fds).</p>
405 <p>While stopping and starting an I/O watcher in the same iteration will
406result in some caching, there is still a syscall per such incident
407(because the fd could point to a different file description now), so its
408best to avoid that. Also, dup()ed file descriptors might not work very
409well if you register events for both fds.</p>
410 <p>Please note that epoll sometimes generates spurious notifications, so you
411need to use non-blocking I/O or other means to avoid blocking when no data
412(or space) is available.</p>
413 </dd>
414 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
415 <dd>
416 <p>Kqueue deserves special mention, as at the time of this writing, it
417was broken on all BSDs except NetBSD (usually it doesn't work with
418anything but sockets and pipes, except on Darwin, where of course its
419completely useless). For this reason its not being &quot;autodetected&quot;
420unless you explicitly specify it explicitly in the flags (i.e. using
421<code>EVBACKEND_KQUEUE</code>).</p>
422 <p>It scales in the same way as the epoll backend, but the interface to the
423kernel is more efficient (which says nothing about its actual speed, of
424course). While starting and stopping an I/O watcher does not cause an
425extra syscall as with epoll, it still adds up to four event changes per
426incident, so its best to avoid that.</p>
427 </dd>
428 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
429 <dd>
430 <p>This is not implemented yet (and might never be).</p>
431 </dd>
432 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
433 <dd>
434 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
435it's really slow, but it still scales very well (O(active_fds)).</p>
436 <p>Please note that solaris ports can result in a lot of spurious
437notifications, so you need to use non-blocking I/O or other means to avoid
438blocking when no data (or space) is available.</p>
439 </dd>
440 <dt><code>EVBACKEND_ALL</code></dt>
441 <dd>
442 <p>Try all backends (even potentially broken ones that wouldn't be tried
443with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
444<code>EVBACKEND_ALL &amp; ~EVBACKEND_KQUEUE</code>.</p>
188 </dd> 445 </dd>
189 </dl> 446 </dl>
190 </p> 447 </p>
448 <p>If one or more of these are ored into the flags value, then only these
449backends will be tried (in the reverse order as given here). If none are
450specified, most compiled-in backend will be tried, usually in reverse
451order of their flag values :)</p>
452 <p>The most typical usage is like this:</p>
453<pre> if (!ev_default_loop (0))
454 fatal (&quot;could not initialise libev, bad $LIBEV_FLAGS in environment?&quot;);
455
456</pre>
457 <p>Restrict libev to the select and poll backends, and do not allow
458environment settings to be taken into account:</p>
459<pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
460
461</pre>
462 <p>Use whatever libev has to offer, but make sure that kqueue is used if
463available (warning, breaks stuff, best use only with your own private
464event loop and only if you know the OS supports your types of fds):</p>
465<pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
466
467</pre>
191 </dd> 468 </dd>
192 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt> 469 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
193 <dd> 470 <dd>
194 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is 471 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
195always distinct from the default loop. Unlike the default loop, it cannot 472always distinct from the default loop. Unlike the default loop, it cannot
196handle signal and child watchers, and attempts to do so will be greeted by 473handle signal and child watchers, and attempts to do so will be greeted by
197undefined behaviour (or a failed assertion if assertions are enabled).</p> 474undefined behaviour (or a failed assertion if assertions are enabled).</p>
475 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
476<pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
477 if (!epoller)
478 fatal (&quot;no epoll found here, maybe it hides under your chair&quot;);
479
480</pre>
198 </dd> 481 </dd>
199 <dt>ev_default_destroy ()</dt> 482 <dt>ev_default_destroy ()</dt>
200 <dd> 483 <dd>
201 <p>Destroys the default loop again (frees all memory and kernel state 484 <p>Destroys the default loop again (frees all memory and kernel state
202etc.). This stops all registered event watchers (by not touching them in 485etc.). None of the active event watchers will be stopped in the normal
203any way whatsoever, although you cnanot rely on this :).</p> 486sense, so e.g. <code>ev_is_active</code> might still return true. It is your
487responsibility to either stop all watchers cleanly yoursef <i>before</i>
488calling this function, or cope with the fact afterwards (which is usually
489the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
490for example).</p>
204 </dd> 491 </dd>
205 <dt>ev_loop_destroy (loop)</dt> 492 <dt>ev_loop_destroy (loop)</dt>
206 <dd> 493 <dd>
207 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an 494 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
208earlier call to <code>ev_loop_new</code>.</p> 495earlier call to <code>ev_loop_new</code>.</p>
211 <dd> 498 <dd>
212 <p>This function reinitialises the kernel state for backends that have 499 <p>This function reinitialises the kernel state for backends that have
213one. Despite the name, you can call it anytime, but it makes most sense 500one. Despite the name, you can call it anytime, but it makes most sense
214after forking, in either the parent or child process (or both, but that 501after forking, in either the parent or child process (or both, but that
215again makes little sense).</p> 502again makes little sense).</p>
216 <p>You <i>must</i> call this function after forking if and only if you want to 503 <p>You <i>must</i> call this function in the child process after forking if and
217use the event library in both processes. If you just fork+exec, you don't 504only if you want to use the event library in both processes. If you just
218have to call it.</p> 505fork+exec, you don't have to call it.</p>
219 <p>The function itself is quite fast and its usually not a problem to call 506 <p>The function itself is quite fast and it's usually not a problem to call
220it just in case after a fork. To make this easy, the function will fit in 507it just in case after a fork. To make this easy, the function will fit in
221quite nicely into a call to <code>pthread_atfork</code>:</p> 508quite nicely into a call to <code>pthread_atfork</code>:</p>
222<pre> pthread_atfork (0, 0, ev_default_fork); 509<pre> pthread_atfork (0, 0, ev_default_fork);
223 510
224</pre> 511</pre>
512 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
513without calling this function, so if you force one of those backends you
514do not need to care.</p>
225 </dd> 515 </dd>
226 <dt>ev_loop_fork (loop)</dt> 516 <dt>ev_loop_fork (loop)</dt>
227 <dd> 517 <dd>
228 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by 518 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
229<code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop 519<code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
230after fork, and how you do this is entirely your own problem.</p> 520after fork, and how you do this is entirely your own problem.</p>
231 </dd> 521 </dd>
232 <dt>unsigned int ev_method (loop)</dt> 522 <dt>unsigned int ev_loop_count (loop)</dt>
523 <dd>
524 <p>Returns the count of loop iterations for the loop, which is identical to
525the number of times libev did poll for new events. It starts at <code>0</code> and
526happily wraps around with enough iterations.</p>
527 <p>This value can sometimes be useful as a generation counter of sorts (it
528&quot;ticks&quot; the number of loop iterations), as it roughly corresponds with
529<code>ev_prepare</code> and <code>ev_check</code> calls.</p>
233 <dd> 530 </dd>
531 <dt>unsigned int ev_backend (loop)</dt>
532 <dd>
234 <p>Returns one of the <code>EVMETHOD_*</code> flags indicating the event backend in 533 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
235use.</p> 534use.</p>
236 </dd> 535 </dd>
237 <dt>ev_tstamp = ev_now (loop)</dt> 536 <dt>ev_tstamp ev_now (loop)</dt>
238 <dd> 537 <dd>
239 <p>Returns the current &quot;event loop time&quot;, which is the time the event loop 538 <p>Returns the current &quot;event loop time&quot;, which is the time the event loop
240got events and started processing them. This timestamp does not change 539received events and started processing them. This timestamp does not
241as long as callbacks are being processed, and this is also the base time 540change as long as callbacks are being processed, and this is also the base
242used for relative timers. You can treat it as the timestamp of the event 541time used for relative timers. You can treat it as the timestamp of the
243occuring (or more correctly, the mainloop finding out about it).</p> 542event occuring (or more correctly, libev finding out about it).</p>
244 </dd> 543 </dd>
245 <dt>ev_loop (loop, int flags)</dt> 544 <dt>ev_loop (loop, int flags)</dt>
246 <dd> 545 <dd>
247 <p>Finally, this is it, the event handler. This function usually is called 546 <p>Finally, this is it, the event handler. This function usually is called
248after you initialised all your watchers and you want to start handling 547after you initialised all your watchers and you want to start handling
249events.</p> 548events.</p>
250 <p>If the flags argument is specified as 0, it will not return until either 549 <p>If the flags argument is specified as <code>0</code>, it will not return until
251no event watchers are active anymore or <code>ev_unloop</code> was called.</p> 550either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
551 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
552relying on all watchers to be stopped when deciding when a program has
553finished (especially in interactive programs), but having a program that
554automatically loops as long as it has to and no longer by virtue of
555relying on its watchers stopping correctly is a thing of beauty.</p>
252 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle 556 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
253those events and any outstanding ones, but will not block your process in 557those events and any outstanding ones, but will not block your process in
254case there are no events.</p> 558case there are no events and will return after one iteration of the loop.</p>
255 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if 559 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
256neccessary) and will handle those and any outstanding ones. It will block 560neccessary) and will handle those and any outstanding ones. It will block
257your process until at least one new event arrives.</p> 561your process until at least one new event arrives, and will return after
258 <p>This flags value could be used to implement alternative looping 562one iteration of the loop. This is useful if you are waiting for some
259constructs, but the <code>prepare</code> and <code>check</code> watchers provide a better and 563external event in conjunction with something not expressible using other
260more generic mechanism.</p> 564libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
565usually a better approach for this kind of thing.</p>
566 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
567<pre> - Before the first iteration, call any pending watchers.
568 * If there are no active watchers (reference count is zero), return.
569 - Queue all prepare watchers and then call all outstanding watchers.
570 - If we have been forked, recreate the kernel state.
571 - Update the kernel state with all outstanding changes.
572 - Update the &quot;event loop time&quot;.
573 - Calculate for how long to block.
574 - Block the process, waiting for any events.
575 - Queue all outstanding I/O (fd) events.
576 - Update the &quot;event loop time&quot; and do time jump handling.
577 - Queue all outstanding timers.
578 - Queue all outstanding periodics.
579 - If no events are pending now, queue all idle watchers.
580 - Queue all check watchers.
581 - Call all queued watchers in reverse order (i.e. check watchers first).
582 Signals and child watchers are implemented as I/O watchers, and will
583 be handled here by queueing them when their watcher gets executed.
584 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
585 were used, return, otherwise continue with step *.
586
587</pre>
588 <p>Example: Queue some jobs and then loop until no events are outsanding
589anymore.</p>
590<pre> ... queue jobs here, make sure they register event watchers as long
591 ... as they still have work to do (even an idle watcher will do..)
592 ev_loop (my_loop, 0);
593 ... jobs done. yeah!
594
595</pre>
261 </dd> 596 </dd>
262 <dt>ev_unloop (loop, how)</dt> 597 <dt>ev_unloop (loop, how)</dt>
263 <dd> 598 <dd>
264 <p>Can be used to make a call to <code>ev_loop</code> return early. The <code>how</code> argument 599 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
600has processed all outstanding events). The <code>how</code> argument must be either
265must be either <code>EVUNLOOP_ONCE</code>, which will make the innermost <code>ev_loop</code> 601<code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
266call return, or <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> 602<code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
267calls return.</p>
268 </dd> 603 </dd>
269 <dt>ev_ref (loop)</dt> 604 <dt>ev_ref (loop)</dt>
270 <dt>ev_unref (loop)</dt> 605 <dt>ev_unref (loop)</dt>
271 <dd> 606 <dd>
272 <p>Ref/unref can be used to add or remove a refcount on the event loop: Every 607 <p>Ref/unref can be used to add or remove a reference count on the event
273watcher keeps one reference. If you have a long-runing watcher you never 608loop: Every watcher keeps one reference, and as long as the reference
274unregister that should not keep ev_loop from running, ev_unref() after 609count is nonzero, <code>ev_loop</code> will not return on its own. If you have
275starting, and ev_ref() before stopping it. Libev itself uses this for 610a watcher you never unregister that should not keep <code>ev_loop</code> from
276example for its internal signal pipe: It is not visible to you as a user 611returning, ev_unref() after starting, and ev_ref() before stopping it. For
277and should not keep <code>ev_loop</code> from exiting if the work is done. It is 612example, libev itself uses this for its internal signal pipe: It is not
278also an excellent way to do this for generic recurring timers or from 613visible to the libev user and should not keep <code>ev_loop</code> from exiting if
279within third-party libraries. Just remember to unref after start and ref 614no event watchers registered by it are active. It is also an excellent
280before stop.</p> 615way to do this for generic recurring timers or from within third-party
616libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
617 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
618running when nothing else is active.</p>
619<pre> struct ev_signal exitsig;
620 ev_signal_init (&amp;exitsig, sig_cb, SIGINT);
621 ev_signal_start (loop, &amp;exitsig);
622 evf_unref (loop);
623
624</pre>
625 <p>Example: For some weird reason, unregister the above signal handler again.</p>
626<pre> ev_ref (loop);
627 ev_signal_stop (loop, &amp;exitsig);
628
629</pre>
281 </dd> 630 </dd>
282</dl> 631</dl>
283 632
633
634
635
636
284</div> 637</div>
285<h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p> 638<h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
286<div id="ANATOMY_OF_A_WATCHER_CONTENT"> 639<div id="ANATOMY_OF_A_WATCHER_CONTENT">
287<p>A watcher is a structure that you create and register to record your 640<p>A watcher is a structure that you create and register to record your
288interest in some event. For instance, if you want to wait for STDIN to 641interest in some event. For instance, if you want to wait for STDIN to
289become readable, you would create an ev_io watcher for that:</p> 642become readable, you would create an <code>ev_io</code> watcher for that:</p>
290<pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents) 643<pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
291 { 644 {
292 ev_io_stop (w); 645 ev_io_stop (w);
293 ev_unloop (loop, EVUNLOOP_ALL); 646 ev_unloop (loop, EVUNLOOP_ALL);
294 } 647 }
317with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher 670with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher
318*)</code>), and you can stop watching for events at any time by calling the 671*)</code>), and you can stop watching for events at any time by calling the
319corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p> 672corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p>
320<p>As long as your watcher is active (has been started but not stopped) you 673<p>As long as your watcher is active (has been started but not stopped) you
321must not touch the values stored in it. Most specifically you must never 674must not touch the values stored in it. Most specifically you must never
322reinitialise it or call its set method.</p> 675reinitialise it or call its <code>set</code> macro.</p>
323<p>You cna check whether an event is active by calling the <code>ev_is_active
324(watcher *)</code> macro. To see whether an event is outstanding (but the
325callback for it has not been called yet) you cna use the <code>ev_is_pending
326(watcher *)</code> macro.</p>
327<p>Each and every callback receives the event loop pointer as first, the 676<p>Each and every callback receives the event loop pointer as first, the
328registered watcher structure as second, and a bitset of received events as 677registered watcher structure as second, and a bitset of received events as
329third argument.</p> 678third argument.</p>
330<p>The rceeived events usually include a single bit per event type received 679<p>The received events usually include a single bit per event type received
331(you can receive multiple events at the same time). The possible bit masks 680(you can receive multiple events at the same time). The possible bit masks
332are:</p> 681are:</p>
333<dl> 682<dl>
334 <dt>EV_READ</dt> 683 <dt><code>EV_READ</code></dt>
335 <dt>EV_WRITE</dt> 684 <dt><code>EV_WRITE</code></dt>
336 <dd> 685 <dd>
337 <p>The file descriptor in the ev_io watcher has become readable and/or 686 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
338writable.</p> 687writable.</p>
339 </dd> 688 </dd>
340 <dt>EV_TIMEOUT</dt> 689 <dt><code>EV_TIMEOUT</code></dt>
341 <dd>
342 <p>The ev_timer watcher has timed out.</p>
343 </dd> 690 <dd>
344 <dt>EV_PERIODIC</dt> 691 <p>The <code>ev_timer</code> watcher has timed out.</p>
345 <dd> 692 </dd>
346 <p>The ev_periodic watcher has timed out.</p> 693 <dt><code>EV_PERIODIC</code></dt>
347 </dd> 694 <dd>
348 <dt>EV_SIGNAL</dt> 695 <p>The <code>ev_periodic</code> watcher has timed out.</p>
349 <dd> 696 </dd>
697 <dt><code>EV_SIGNAL</code></dt>
698 <dd>
350 <p>The signal specified in the ev_signal watcher has been received by a thread.</p> 699 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
351 </dd>
352 <dt>EV_CHILD</dt>
353 <dd> 700 </dd>
701 <dt><code>EV_CHILD</code></dt>
702 <dd>
354 <p>The pid specified in the ev_child watcher has received a status change.</p> 703 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
355 </dd>
356 <dt>EV_IDLE</dt>
357 <dd> 704 </dd>
705 <dt><code>EV_STAT</code></dt>
706 <dd>
707 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
708 </dd>
709 <dt><code>EV_IDLE</code></dt>
710 <dd>
358 <p>The ev_idle watcher has determined that you have nothing better to do.</p> 711 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
359 </dd>
360 <dt>EV_PREPARE</dt>
361 <dt>EV_CHECK</dt>
362 <dd> 712 </dd>
713 <dt><code>EV_PREPARE</code></dt>
714 <dt><code>EV_CHECK</code></dt>
715 <dd>
363 <p>All ev_prepare watchers are invoked just <i>before</i> <code>ev_loop</code> starts 716 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
364to gather new events, and all ev_check watchers are invoked just after 717to gather new events, and all <code>ev_check</code> watchers are invoked just after
365<code>ev_loop</code> has gathered them, but before it invokes any callbacks for any 718<code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
366received events. Callbacks of both watcher types can start and stop as 719received events. Callbacks of both watcher types can start and stop as
367many watchers as they want, and all of them will be taken into account 720many watchers as they want, and all of them will be taken into account
368(for example, a ev_prepare watcher might start an idle watcher to keep 721(for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
369<code>ev_loop</code> from blocking).</p> 722<code>ev_loop</code> from blocking).</p>
370 </dd> 723 </dd>
371 <dt>EV_ERROR</dt> 724 <dt><code>EV_EMBED</code></dt>
725 <dd>
726 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
727 </dd>
728 <dt><code>EV_FORK</code></dt>
729 <dd>
730 <p>The event loop has been resumed in the child process after fork (see
731<code>ev_fork</code>).</p>
732 </dd>
733 <dt><code>EV_ERROR</code></dt>
372 <dd> 734 <dd>
373 <p>An unspecified error has occured, the watcher has been stopped. This might 735 <p>An unspecified error has occured, the watcher has been stopped. This might
374happen because the watcher could not be properly started because libev 736happen because the watcher could not be properly started because libev
375ran out of memory, a file descriptor was found to be closed or any other 737ran out of memory, a file descriptor was found to be closed or any other
376problem. You best act on it by reporting the problem and somehow coping 738problem. You best act on it by reporting the problem and somehow coping
382programs, though, so beware.</p> 744programs, though, so beware.</p>
383 </dd> 745 </dd>
384</dl> 746</dl>
385 747
386</div> 748</div>
749<h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
750<div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
751<p>In the following description, <code>TYPE</code> stands for the watcher type,
752e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
753<dl>
754 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
755 <dd>
756 <p>This macro initialises the generic portion of a watcher. The contents
757of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
758the generic parts of the watcher are initialised, you <i>need</i> to call
759the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
760type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
761which rolls both calls into one.</p>
762 <p>You can reinitialise a watcher at any time as long as it has been stopped
763(or never started) and there are no pending events outstanding.</p>
764 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
765int revents)</code>.</p>
766 </dd>
767 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
768 <dd>
769 <p>This macro initialises the type-specific parts of a watcher. You need to
770call <code>ev_init</code> at least once before you call this macro, but you can
771call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
772macro on a watcher that is active (it can be pending, however, which is a
773difference to the <code>ev_init</code> macro).</p>
774 <p>Although some watcher types do not have type-specific arguments
775(e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
776 </dd>
777 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
778 <dd>
779 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
780calls into a single call. This is the most convinient method to initialise
781a watcher. The same limitations apply, of course.</p>
782 </dd>
783 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
784 <dd>
785 <p>Starts (activates) the given watcher. Only active watchers will receive
786events. If the watcher is already active nothing will happen.</p>
787 </dd>
788 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
789 <dd>
790 <p>Stops the given watcher again (if active) and clears the pending
791status. It is possible that stopped watchers are pending (for example,
792non-repeating timers are being stopped when they become pending), but
793<code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
794you want to free or reuse the memory used by the watcher it is therefore a
795good idea to always call its <code>ev_TYPE_stop</code> function.</p>
796 </dd>
797 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
798 <dd>
799 <p>Returns a true value iff the watcher is active (i.e. it has been started
800and not yet been stopped). As long as a watcher is active you must not modify
801it.</p>
802 </dd>
803 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
804 <dd>
805 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
806events but its callback has not yet been invoked). As long as a watcher
807is pending (but not active) you must not call an init function on it (but
808<code>ev_TYPE_set</code> is safe), you must not change its priority, and you must
809make sure the watcher is available to libev (e.g. you cannot <code>free ()</code>
810it).</p>
811 </dd>
812 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
813 <dd>
814 <p>Returns the callback currently set on the watcher.</p>
815 </dd>
816 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
817 <dd>
818 <p>Change the callback. You can change the callback at virtually any time
819(modulo threads).</p>
820 </dd>
821 <dt>ev_set_priority (ev_TYPE *watcher, priority)</dt>
822 <dt>int ev_priority (ev_TYPE *watcher)</dt>
823 <dd>
824 <p>Set and query the priority of the watcher. The priority is a small
825integer between <code>EV_MAXPRI</code> (default: <code>2</code>) and <code>EV_MINPRI</code>
826(default: <code>-2</code>). Pending watchers with higher priority will be invoked
827before watchers with lower priority, but priority will not keep watchers
828from being executed (except for <code>ev_idle</code> watchers).</p>
829 <p>This means that priorities are <i>only</i> used for ordering callback
830invocation after new events have been received. This is useful, for
831example, to reduce latency after idling, or more often, to bind two
832watchers on the same event and make sure one is called first.</p>
833 <p>If you need to suppress invocation when higher priority events are pending
834you need to look at <code>ev_idle</code> watchers, which provide this functionality.</p>
835 <p>You <i>must not</i> change the priority of a watcher as long as it is active or
836pending.</p>
837 <p>The default priority used by watchers when no priority has been set is
838always <code>0</code>, which is supposed to not be too high and not be too low :).</p>
839 <p>Setting a priority outside the range of <code>EV_MINPRI</code> to <code>EV_MAXPRI</code> is
840fine, as long as you do not mind that the priority value you query might
841or might not have been adjusted to be within valid range.</p>
842 </dd>
843 <dt>ev_invoke (loop, ev_TYPE *watcher, int revents)</dt>
844 <dd>
845 <p>Invoke the <code>watcher</code> with the given <code>loop</code> and <code>revents</code>. Neither
846<code>loop</code> nor <code>revents</code> need to be valid as long as the watcher callback
847can deal with that fact.</p>
848 </dd>
849 <dt>int ev_clear_pending (loop, ev_TYPE *watcher)</dt>
850 <dd>
851 <p>If the watcher is pending, this function returns clears its pending status
852and returns its <code>revents</code> bitset (as if its callback was invoked). If the
853watcher isn't pending it does nothing and returns <code>0</code>.</p>
854 </dd>
855</dl>
856
857
858
859
860
861</div>
387<h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2> 862<h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
388<div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2"> 863<div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
389<p>Each watcher has, by default, a member <code>void *data</code> that you can change 864<p>Each watcher has, by default, a member <code>void *data</code> that you can change
390and read at any time, libev will completely ignore it. This cna be used 865and read at any time, libev will completely ignore it. This can be used
391to associate arbitrary data with your watcher. If you need more data and 866to associate arbitrary data with your watcher. If you need more data and
392don't want to allocate memory and store a pointer to it in that data 867don't want to allocate memory and store a pointer to it in that data
393member, you can also &quot;subclass&quot; the watcher type and provide your own 868member, you can also &quot;subclass&quot; the watcher type and provide your own
394data:</p> 869data:</p>
395<pre> struct my_io 870<pre> struct my_io
408 struct my_io *w = (struct my_io *)w_; 883 struct my_io *w = (struct my_io *)w_;
409 ... 884 ...
410 } 885 }
411 886
412</pre> 887</pre>
413<p>More interesting and less C-conformant ways of catsing your callback type 888<p>More interesting and less C-conformant ways of casting your callback type
414have been omitted....</p> 889instead have been omitted.</p>
890<p>Another common scenario is having some data structure with multiple
891watchers:</p>
892<pre> struct my_biggy
893 {
894 int some_data;
895 ev_timer t1;
896 ev_timer t2;
897 }
415 898
899</pre>
900<p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
901you need to use <code>offsetof</code>:</p>
902<pre> #include &lt;stddef.h&gt;
416 903
904 static void
905 t1_cb (EV_P_ struct ev_timer *w, int revents)
906 {
907 struct my_biggy big = (struct my_biggy *
908 (((char *)w) - offsetof (struct my_biggy, t1));
909 }
417 910
911 static void
912 t2_cb (EV_P_ struct ev_timer *w, int revents)
913 {
914 struct my_biggy big = (struct my_biggy *
915 (((char *)w) - offsetof (struct my_biggy, t2));
916 }
418 917
419 918
919
920
921</pre>
922
420</div> 923</div>
421<h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p> 924<h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
422<div id="WATCHER_TYPES_CONTENT"> 925<div id="WATCHER_TYPES_CONTENT">
423<p>This section describes each watcher in detail, but will not repeat 926<p>This section describes each watcher in detail, but will not repeat
424information given in the last section.</p> 927information given in the last section. Any initialisation/set macros,
928functions and members specific to the watcher type are explained.</p>
929<p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
930while the watcher is active, you can look at the member and expect some
931sensible content, but you must not modify it (you can modify it while the
932watcher is stopped to your hearts content), or <i>[read-write]</i>, which
933means you can expect it to have some sensible content while the watcher
934is active, but you can also modify it. Modifying it may not do something
935sensible or take immediate effect (or do anything at all), but libev will
936not crash or malfunction in any way.</p>
425 937
938
939
940
941
426</div> 942</div>
427<h2 id="struct_ev_io_is_my_file_descriptor_r">struct ev_io - is my file descriptor readable or writable</h2> 943<h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
428<div id="struct_ev_io_is_my_file_descriptor_r-2"> 944<div id="code_ev_io_code_is_this_file_descrip-2">
429<p>I/O watchers check whether a file descriptor is readable or writable 945<p>I/O watchers check whether a file descriptor is readable or writable
430in each iteration of the event loop (This behaviour is called 946in each iteration of the event loop, or, more precisely, when reading
431level-triggering because you keep receiving events as long as the 947would not block the process and writing would at least be able to write
432condition persists. Remember you cna stop the watcher if you don't want to 948some data. This behaviour is called level-triggering because you keep
433act on the event and neither want to receive future events).</p> 949receiving events as long as the condition persists. Remember you can stop
950the watcher if you don't want to act on the event and neither want to
951receive future events.</p>
952<p>In general you can register as many read and/or write event watchers per
953fd as you want (as long as you don't confuse yourself). Setting all file
954descriptors to non-blocking mode is also usually a good idea (but not
955required if you know what you are doing).</p>
956<p>You have to be careful with dup'ed file descriptors, though. Some backends
957(the linux epoll backend is a notable example) cannot handle dup'ed file
958descriptors correctly if you register interest in two or more fds pointing
959to the same underlying file/socket/etc. description (that is, they share
960the same underlying &quot;file open&quot;).</p>
961<p>If you must do this, then force the use of a known-to-be-good backend
962(at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
963<code>EVBACKEND_POLL</code>).</p>
964<p>Another thing you have to watch out for is that it is quite easy to
965receive &quot;spurious&quot; readyness notifications, that is your callback might
966be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
967because there is no data. Not only are some backends known to create a
968lot of those (for example solaris ports), it is very easy to get into
969this situation even with a relatively standard program structure. Thus
970it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
971<code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
972<p>If you cannot run the fd in non-blocking mode (for example you should not
973play around with an Xlib connection), then you have to seperately re-test
974whether a file descriptor is really ready with a known-to-be good interface
975such as poll (fortunately in our Xlib example, Xlib already does this on
976its own, so its quite safe to use).</p>
977
978</div>
979<h3 id="The_special_problem_of_disappearing_">The special problem of disappearing file descriptors</h3>
980<div id="The_special_problem_of_disappearing_-2">
981<p>Some backends (e.g kqueue, epoll) need to be told about closing a file
982descriptor (either by calling <code>close</code> explicitly or by any other means,
983such as <code>dup</code>). The reason is that you register interest in some file
984descriptor, but when it goes away, the operating system will silently drop
985this interest. If another file descriptor with the same number then is
986registered with libev, there is no efficient way to see that this is, in
987fact, a different file descriptor.</p>
988<p>To avoid having to explicitly tell libev about such cases, libev follows
989the following policy: Each time <code>ev_io_set</code> is being called, libev
990will assume that this is potentially a new file descriptor, otherwise
991it is assumed that the file descriptor stays the same. That means that
992you <i>have</i> to call <code>ev_io_set</code> (or <code>ev_io_init</code>) when you change the
993descriptor even if the file descriptor number itself did not change.</p>
994<p>This is how one would do it normally anyway, the important point is that
995the libev application should not optimise around libev but should leave
996optimisations to libev.</p>
997
998
999
1000
1001
1002</div>
1003<h3 id="Watcher_Specific_Functions">Watcher-Specific Functions</h3>
1004<div id="Watcher_Specific_Functions_CONTENT">
434<dl> 1005<dl>
435 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt> 1006 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
436 <dt>ev_io_set (ev_io *, int fd, int events)</dt> 1007 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
437 <dd> 1008 <dd>
438 <p>Configures an ev_io watcher. The fd is the file descriptor to rceeive 1009 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
439events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ | 1010rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
440EV_WRITE</code> to receive the given events.</p> 1011<code>EV_READ | EV_WRITE</code> to receive the given events.</p>
1012 </dd>
1013 <dt>int fd [read-only]</dt>
1014 <dd>
1015 <p>The file descriptor being watched.</p>
1016 </dd>
1017 <dt>int events [read-only]</dt>
1018 <dd>
1019 <p>The events being watched.</p>
441 </dd> 1020 </dd>
442</dl> 1021</dl>
1022<p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
1023readable, but only once. Since it is likely line-buffered, you could
1024attempt to read a whole line in the callback.</p>
1025<pre> static void
1026 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1027 {
1028 ev_io_stop (loop, w);
1029 .. read from stdin here (or from w-&gt;fd) and haqndle any I/O errors
1030 }
443 1031
1032 ...
1033 struct ev_loop *loop = ev_default_init (0);
1034 struct ev_io stdin_readable;
1035 ev_io_init (&amp;stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1036 ev_io_start (loop, &amp;stdin_readable);
1037 ev_loop (loop, 0);
1038
1039
1040
1041
1042</pre>
1043
444</div> 1044</div>
445<h2 id="struct_ev_timer_relative_and_optiona">struct ev_timer - relative and optionally recurring timeouts</h2> 1045<h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
446<div id="struct_ev_timer_relative_and_optiona-2"> 1046<div id="code_ev_timer_code_relative_and_opti-2">
447<p>Timer watchers are simple relative timers that generate an event after a 1047<p>Timer watchers are simple relative timers that generate an event after a
448given time, and optionally repeating in regular intervals after that.</p> 1048given time, and optionally repeating in regular intervals after that.</p>
449<p>The timers are based on real time, that is, if you register an event that 1049<p>The timers are based on real time, that is, if you register an event that
450times out after an hour and youreset your system clock to last years 1050times out after an hour and you reset your system clock to last years
451time, it will still time out after (roughly) and hour. &quot;Roughly&quot; because 1051time, it will still time out after (roughly) and hour. &quot;Roughly&quot; because
452detecting time jumps is hard, and soem inaccuracies are unavoidable (the 1052detecting time jumps is hard, and some inaccuracies are unavoidable (the
453monotonic clock option helps a lot here).</p> 1053monotonic clock option helps a lot here).</p>
1054<p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
1055time. This is usually the right thing as this timestamp refers to the time
1056of the event triggering whatever timeout you are modifying/starting. If
1057you suspect event processing to be delayed and you <i>need</i> to base the timeout
1058on the current time, use something like this to adjust for this:</p>
1059<pre> ev_timer_set (&amp;timer, after + ev_now () - ev_time (), 0.);
1060
1061</pre>
1062<p>The callback is guarenteed to be invoked only when its timeout has passed,
1063but if multiple timers become ready during the same loop iteration then
1064order of execution is undefined.</p>
1065
1066</div>
1067<h3 id="Watcher_Specific_Functions_and_Data_">Watcher-Specific Functions and Data Members</h3>
1068<div id="Watcher_Specific_Functions_and_Data_-2">
454<dl> 1069<dl>
455 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt> 1070 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
456 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt> 1071 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
457 <dd> 1072 <dd>
458 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is 1073 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
460timer will automatically be configured to trigger again <code>repeat</code> seconds 1075timer will automatically be configured to trigger again <code>repeat</code> seconds
461later, again, and again, until stopped manually.</p> 1076later, again, and again, until stopped manually.</p>
462 <p>The timer itself will do a best-effort at avoiding drift, that is, if you 1077 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
463configure a timer to trigger every 10 seconds, then it will trigger at 1078configure a timer to trigger every 10 seconds, then it will trigger at
464exactly 10 second intervals. If, however, your program cannot keep up with 1079exactly 10 second intervals. If, however, your program cannot keep up with
465the timer (ecause it takes longer than those 10 seconds to do stuff) the 1080the timer (because it takes longer than those 10 seconds to do stuff) the
466timer will not fire more than once per event loop iteration.</p> 1081timer will not fire more than once per event loop iteration.</p>
467 </dd> 1082 </dd>
468 <dt>ev_timer_again (loop)</dt> 1083 <dt>ev_timer_again (loop)</dt>
469 <dd> 1084 <dd>
470 <p>This will act as if the timer timed out and restart it again if it is 1085 <p>This will act as if the timer timed out and restart it again if it is
471repeating. The exact semantics are:</p> 1086repeating. The exact semantics are:</p>
1087 <p>If the timer is pending, its pending status is cleared.</p>
472 <p>If the timer is started but nonrepeating, stop it.</p> 1088 <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
473 <p>If the timer is repeating, either start it if necessary (with the repeat 1089 <p>If the timer is repeating, either start it if necessary (with the
474value), or reset the running timer to the repeat value.</p> 1090<code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
475 <p>This sounds a bit complicated, but here is a useful and typical 1091 <p>This sounds a bit complicated, but here is a useful and typical
476example: Imagine you have a tcp connection and you want a so-called idle 1092example: Imagine you have a tcp connection and you want a so-called idle
477timeout, that is, you want to be called when there have been, say, 60 1093timeout, that is, you want to be called when there have been, say, 60
478seconds of inactivity on the socket. The easiest way to do this is to 1094seconds of inactivity on the socket. The easiest way to do this is to
479configure an ev_timer with after=repeat=60 and calling ev_timer_again each 1095configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
480time you successfully read or write some data. If you go into an idle 1096<code>ev_timer_again</code> each time you successfully read or write some data. If
481state where you do not expect data to travel on the socket, you can stop 1097you go into an idle state where you do not expect data to travel on the
1098socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
482the timer, and again will automatically restart it if need be.</p> 1099automatically restart it if need be.</p>
1100 <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
1101altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
1102<pre> ev_timer_init (timer, callback, 0., 5.);
1103 ev_timer_again (loop, timer);
1104 ...
1105 timer-&gt;again = 17.;
1106 ev_timer_again (loop, timer);
1107 ...
1108 timer-&gt;again = 10.;
1109 ev_timer_again (loop, timer);
1110
1111</pre>
1112 <p>This is more slightly efficient then stopping/starting the timer each time
1113you want to modify its timeout value.</p>
1114 </dd>
1115 <dt>ev_tstamp repeat [read-write]</dt>
1116 <dd>
1117 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1118or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1119which is also when any modifications are taken into account.</p>
483 </dd> 1120 </dd>
484</dl> 1121</dl>
1122<p>Example: Create a timer that fires after 60 seconds.</p>
1123<pre> static void
1124 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1125 {
1126 .. one minute over, w is actually stopped right here
1127 }
485 1128
1129 struct ev_timer mytimer;
1130 ev_timer_init (&amp;mytimer, one_minute_cb, 60., 0.);
1131 ev_timer_start (loop, &amp;mytimer);
1132
1133</pre>
1134<p>Example: Create a timeout timer that times out after 10 seconds of
1135inactivity.</p>
1136<pre> static void
1137 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1138 {
1139 .. ten seconds without any activity
1140 }
1141
1142 struct ev_timer mytimer;
1143 ev_timer_init (&amp;mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1144 ev_timer_again (&amp;mytimer); /* start timer */
1145 ev_loop (loop, 0);
1146
1147 // and in some piece of code that gets executed on any &quot;activity&quot;:
1148 // reset the timeout to start ticking again at 10 seconds
1149 ev_timer_again (&amp;mytimer);
1150
1151
1152
1153
1154</pre>
1155
486</div> 1156</div>
487<h2 id="ev_periodic_to_cron_or_not_to_cron_i">ev_periodic - to cron or not to cron it</h2> 1157<h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
488<div id="ev_periodic_to_cron_or_not_to_cron_i-2"> 1158<div id="code_ev_periodic_code_to_cron_or_not-2">
489<p>Periodic watchers are also timers of a kind, but they are very versatile 1159<p>Periodic watchers are also timers of a kind, but they are very versatile
490(and unfortunately a bit complex).</p> 1160(and unfortunately a bit complex).</p>
491<p>Unlike ev_timer's, they are not based on real time (or relative time) 1161<p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
492but on wallclock time (absolute time). You can tell a periodic watcher 1162but on wallclock time (absolute time). You can tell a periodic watcher
493to trigger &quot;at&quot; some specific point in time. For example, if you tell a 1163to trigger &quot;at&quot; some specific point in time. For example, if you tell a
494periodic watcher to trigger in 10 seconds (by specifiying e.g. c&lt;ev_now () 1164periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
495+ 10.&gt;) and then reset your system clock to the last year, then it will 1165+ 10.</code>) and then reset your system clock to the last year, then it will
496take a year to trigger the event (unlike an ev_timer, which would trigger 1166take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
497roughly 10 seconds later and of course not if you reset your system time 1167roughly 10 seconds later).</p>
498again).</p>
499<p>They can also be used to implement vastly more complex timers, such as 1168<p>They can also be used to implement vastly more complex timers, such as
500triggering an event on eahc midnight, local time.</p> 1169triggering an event on each midnight, local time or other, complicated,
1170rules.</p>
1171<p>As with timers, the callback is guarenteed to be invoked only when the
1172time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1173during the same loop iteration then order of execution is undefined.</p>
1174
1175</div>
1176<h3 id="Watcher_Specific_Functions_and_Data_-3">Watcher-Specific Functions and Data Members</h3>
1177<div id="Watcher_Specific_Functions_and_Data_-2">
501<dl> 1178<dl>
502 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt> 1179 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
503 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt> 1180 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
504 <dd> 1181 <dd>
505 <p>Lots of arguments, lets sort it out... There are basically three modes of 1182 <p>Lots of arguments, lets sort it out... There are basically three modes of
506operation, and we will explain them from simplest to complex:</p> 1183operation, and we will explain them from simplest to complex:</p>
507
508
509
510
511 <p> 1184 <p>
512 <dl> 1185 <dl>
513 <dt>* absolute timer (interval = reschedule_cb = 0)</dt> 1186 <dt>* absolute timer (at = time, interval = reschedule_cb = 0)</dt>
514 <dd> 1187 <dd>
515 <p>In this configuration the watcher triggers an event at the wallclock time 1188 <p>In this configuration the watcher triggers an event at the wallclock time
516<code>at</code> and doesn't repeat. It will not adjust when a time jump occurs, 1189<code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
517that is, if it is to be run at January 1st 2011 then it will run when the 1190that is, if it is to be run at January 1st 2011 then it will run when the
518system time reaches or surpasses this time.</p> 1191system time reaches or surpasses this time.</p>
519 </dd> 1192 </dd>
520 <dt>* non-repeating interval timer (interval &gt; 0, reschedule_cb = 0)</dt> 1193 <dt>* non-repeating interval timer (at = offset, interval &gt; 0, reschedule_cb = 0)</dt>
521 <dd> 1194 <dd>
522 <p>In this mode the watcher will always be scheduled to time out at the next 1195 <p>In this mode the watcher will always be scheduled to time out at the next
523<code>at + N * interval</code> time (for some integer N) and then repeat, regardless 1196<code>at + N * interval</code> time (for some integer N, which can also be negative)
524of any time jumps.</p> 1197and then repeat, regardless of any time jumps.</p>
525 <p>This can be used to create timers that do not drift with respect to system 1198 <p>This can be used to create timers that do not drift with respect to system
526time:</p> 1199time:</p>
527<pre> ev_periodic_set (&amp;periodic, 0., 3600., 0); 1200<pre> ev_periodic_set (&amp;periodic, 0., 3600., 0);
528 1201
529</pre> 1202</pre>
530 <p>This doesn't mean there will always be 3600 seconds in between triggers, 1203 <p>This doesn't mean there will always be 3600 seconds in between triggers,
531but only that the the callback will be called when the system time shows a 1204but only that the the callback will be called when the system time shows a
532full hour (UTC), or more correct, when the system time is evenly divisible 1205full hour (UTC), or more correctly, when the system time is evenly divisible
533by 3600.</p> 1206by 3600.</p>
534 <p>Another way to think about it (for the mathematically inclined) is that 1207 <p>Another way to think about it (for the mathematically inclined) is that
535ev_periodic will try to run the callback in this mode at the next possible 1208<code>ev_periodic</code> will try to run the callback in this mode at the next possible
536time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p> 1209time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1210 <p>For numerical stability it is preferable that the <code>at</code> value is near
1211<code>ev_now ()</code> (the current time), but there is no range requirement for
1212this value.</p>
537 </dd> 1213 </dd>
538 <dt>* manual reschedule mode (reschedule_cb = callback)</dt> 1214 <dt>* manual reschedule mode (at and interval ignored, reschedule_cb = callback)</dt>
539 <dd> 1215 <dd>
540 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being 1216 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
541ignored. Instead, each time the periodic watcher gets scheduled, the 1217ignored. Instead, each time the periodic watcher gets scheduled, the
542reschedule callback will be called with the watcher as first, and the 1218reschedule callback will be called with the watcher as first, and the
543current time as second argument.</p> 1219current time as second argument.</p>
544 <p>NOTE: <i>This callback MUST NOT stop or destroy the periodic or any other 1220 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
545periodic watcher, ever, or make any event loop modificstions</i>. If you need 1221ever, or make any event loop modifications</i>. If you need to stop it,
546to stop it, return 1e30 (or so, fudge fudge) and stop it afterwards.</p> 1222return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1223starting an <code>ev_prepare</code> watcher, which is legal).</p>
547 <p>Its prototype is c&lt;ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1224 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
548ev_tstamp now)&gt;, e.g.:</p> 1225ev_tstamp now)</code>, e.g.:</p>
549<pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now) 1226<pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
550 { 1227 {
551 return now + 60.; 1228 return now + 60.;
552 } 1229 }
553 1230
554</pre> 1231</pre>
555 <p>It must return the next time to trigger, based on the passed time value 1232 <p>It must return the next time to trigger, based on the passed time value
556(that is, the lowest time value larger than to the second argument). It 1233(that is, the lowest time value larger than to the second argument). It
557will usually be called just before the callback will be triggered, but 1234will usually be called just before the callback will be triggered, but
558might be called at other times, too.</p> 1235might be called at other times, too.</p>
1236 <p>NOTE: <i>This callback must always return a time that is later than the
1237passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
559 <p>This can be used to create very complex timers, such as a timer that 1238 <p>This can be used to create very complex timers, such as a timer that
560triggers on each midnight, local time. To do this, you would calculate the 1239triggers on each midnight, local time. To do this, you would calculate the
561next midnight after <code>now</code> and return the timestamp value for this. How you do this 1240next midnight after <code>now</code> and return the timestamp value for this. How
562is, again, up to you (but it is not trivial).</p> 1241you do this is, again, up to you (but it is not trivial, which is the main
1242reason I omitted it as an example).</p>
563 </dd> 1243 </dd>
564 </dl> 1244 </dl>
565 </p> 1245 </p>
566 </dd> 1246 </dd>
567 <dt>ev_periodic_again (loop, ev_periodic *)</dt> 1247 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
569 <p>Simply stops and restarts the periodic watcher again. This is only useful 1249 <p>Simply stops and restarts the periodic watcher again. This is only useful
570when you changed some parameters or the reschedule callback would return 1250when you changed some parameters or the reschedule callback would return
571a different time than the last time it was called (e.g. in a crond like 1251a different time than the last time it was called (e.g. in a crond like
572program when the crontabs have changed).</p> 1252program when the crontabs have changed).</p>
573 </dd> 1253 </dd>
1254 <dt>ev_tstamp offset [read-write]</dt>
1255 <dd>
1256 <p>When repeating, this contains the offset value, otherwise this is the
1257absolute point in time (the <code>at</code> value passed to <code>ev_periodic_set</code>).</p>
1258 <p>Can be modified any time, but changes only take effect when the periodic
1259timer fires or <code>ev_periodic_again</code> is being called.</p>
1260 </dd>
1261 <dt>ev_tstamp interval [read-write]</dt>
1262 <dd>
1263 <p>The current interval value. Can be modified any time, but changes only
1264take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1265called.</p>
1266 </dd>
1267 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1268 <dd>
1269 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1270switched off. Can be changed any time, but changes only take effect when
1271the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1272 </dd>
574</dl> 1273</dl>
1274<p>Example: Call a callback every hour, or, more precisely, whenever the
1275system clock is divisible by 3600. The callback invocation times have
1276potentially a lot of jittering, but good long-term stability.</p>
1277<pre> static void
1278 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1279 {
1280 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1281 }
575 1282
1283 struct ev_periodic hourly_tick;
1284 ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 3600., 0);
1285 ev_periodic_start (loop, &amp;hourly_tick);
1286
1287</pre>
1288<p>Example: The same as above, but use a reschedule callback to do it:</p>
1289<pre> #include &lt;math.h&gt;
1290
1291 static ev_tstamp
1292 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1293 {
1294 return fmod (now, 3600.) + 3600.;
1295 }
1296
1297 ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1298
1299</pre>
1300<p>Example: Call a callback every hour, starting now:</p>
1301<pre> struct ev_periodic hourly_tick;
1302 ev_periodic_init (&amp;hourly_tick, clock_cb,
1303 fmod (ev_now (loop), 3600.), 3600., 0);
1304 ev_periodic_start (loop, &amp;hourly_tick);
1305
1306
1307
1308
1309</pre>
1310
576</div> 1311</div>
577<h2 id="ev_signal_signal_me_when_a_signal_ge">ev_signal - signal me when a signal gets signalled</h2> 1312<h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
578<div id="ev_signal_signal_me_when_a_signal_ge-2"> 1313<div id="code_ev_signal_code_signal_me_when_a-2">
579<p>Signal watchers will trigger an event when the process receives a specific 1314<p>Signal watchers will trigger an event when the process receives a specific
580signal one or more times. Even though signals are very asynchronous, libev 1315signal one or more times. Even though signals are very asynchronous, libev
581will try its best to deliver signals synchronously, i.e. as part of the 1316will try it's best to deliver signals synchronously, i.e. as part of the
582normal event processing, like any other event.</p> 1317normal event processing, like any other event.</p>
583<p>You cna configure as many watchers as you like per signal. Only when the 1318<p>You can configure as many watchers as you like per signal. Only when the
584first watcher gets started will libev actually register a signal watcher 1319first watcher gets started will libev actually register a signal watcher
585with the kernel (thus it coexists with your own signal handlers as long 1320with the kernel (thus it coexists with your own signal handlers as long
586as you don't register any with libev). Similarly, when the last signal 1321as you don't register any with libev). Similarly, when the last signal
587watcher for a signal is stopped libev will reset the signal handler to 1322watcher for a signal is stopped libev will reset the signal handler to
588SIG_DFL (regardless of what it was set to before).</p> 1323SIG_DFL (regardless of what it was set to before).</p>
1324
1325</div>
1326<h3 id="Watcher_Specific_Functions_and_Data_-4">Watcher-Specific Functions and Data Members</h3>
1327<div id="Watcher_Specific_Functions_and_Data_-2-2">
589<dl> 1328<dl>
590 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt> 1329 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
591 <dt>ev_signal_set (ev_signal *, int signum)</dt> 1330 <dt>ev_signal_set (ev_signal *, int signum)</dt>
592 <dd> 1331 <dd>
593 <p>Configures the watcher to trigger on the given signal number (usually one 1332 <p>Configures the watcher to trigger on the given signal number (usually one
594of the <code>SIGxxx</code> constants).</p> 1333of the <code>SIGxxx</code> constants).</p>
595 </dd> 1334 </dd>
1335 <dt>int signum [read-only]</dt>
1336 <dd>
1337 <p>The signal the watcher watches out for.</p>
1338 </dd>
596</dl> 1339</dl>
597 1340
1341
1342
1343
1344
598</div> 1345</div>
599<h2 id="ev_child_wait_for_pid_status_changes">ev_child - wait for pid status changes</h2> 1346<h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
600<div id="ev_child_wait_for_pid_status_changes-2"> 1347<div id="code_ev_child_code_watch_out_for_pro-2">
601<p>Child watchers trigger when your process receives a SIGCHLD in response to 1348<p>Child watchers trigger when your process receives a SIGCHLD in response to
602some child status changes (most typically when a child of yours dies).</p> 1349some child status changes (most typically when a child of yours dies).</p>
1350
1351</div>
1352<h3 id="Watcher_Specific_Functions_and_Data_-5">Watcher-Specific Functions and Data Members</h3>
1353<div id="Watcher_Specific_Functions_and_Data_-2-3">
603<dl> 1354<dl>
604 <dt>ev_child_init (ev_child *, callback, int pid)</dt> 1355 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
605 <dt>ev_child_set (ev_child *, int pid)</dt> 1356 <dt>ev_child_set (ev_child *, int pid)</dt>
606 <dd> 1357 <dd>
607 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or 1358 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
608<i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look 1359<i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
609at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see 1360at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
610the status word (use the macros from <code>sys/wait.h</code>). The <code>rpid</code> member 1361the status word (use the macros from <code>sys/wait.h</code> and see your systems
1362<code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
611contains the pid of the process causing the status change.</p> 1363process causing the status change.</p>
1364 </dd>
1365 <dt>int pid [read-only]</dt>
1366 <dd>
1367 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1368 </dd>
1369 <dt>int rpid [read-write]</dt>
1370 <dd>
1371 <p>The process id that detected a status change.</p>
1372 </dd>
1373 <dt>int rstatus [read-write]</dt>
1374 <dd>
1375 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1376<code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
612 </dd> 1377 </dd>
613</dl> 1378</dl>
1379<p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1380<pre> static void
1381 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1382 {
1383 ev_unloop (loop, EVUNLOOP_ALL);
1384 }
614 1385
1386 struct ev_signal signal_watcher;
1387 ev_signal_init (&amp;signal_watcher, sigint_cb, SIGINT);
1388 ev_signal_start (loop, &amp;sigint_cb);
1389
1390
1391
1392
1393</pre>
1394
615</div> 1395</div>
1396<h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1397<div id="code_ev_stat_code_did_the_file_attri-2">
1398<p>This watches a filesystem path for attribute changes. That is, it calls
1399<code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1400compared to the last time, invoking the callback if it did.</p>
1401<p>The path does not need to exist: changing from &quot;path exists&quot; to &quot;path does
1402not exist&quot; is a status change like any other. The condition &quot;path does
1403not exist&quot; is signified by the <code>st_nlink</code> field being zero (which is
1404otherwise always forced to be at least one) and all the other fields of
1405the stat buffer having unspecified contents.</p>
1406<p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1407relative and your working directory changes, the behaviour is undefined.</p>
1408<p>Since there is no standard to do this, the portable implementation simply
1409calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1410can specify a recommended polling interval for this case. If you specify
1411a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1412unspecified default</i> value will be used (which you can expect to be around
1413five seconds, although this might change dynamically). Libev will also
1414impose a minimum interval which is currently around <code>0.1</code>, but thats
1415usually overkill.</p>
1416<p>This watcher type is not meant for massive numbers of stat watchers,
1417as even with OS-supported change notifications, this can be
1418resource-intensive.</p>
1419<p>At the time of this writing, only the Linux inotify interface is
1420implemented (implementing kqueue support is left as an exercise for the
1421reader). Inotify will be used to give hints only and should not change the
1422semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1423to fall back to regular polling again even with inotify, but changes are
1424usually detected immediately, and if the file exists there will be no
1425polling.</p>
1426
1427</div>
1428<h3 id="Watcher_Specific_Functions_and_Data_-6">Watcher-Specific Functions and Data Members</h3>
1429<div id="Watcher_Specific_Functions_and_Data_-2-4">
1430<dl>
1431 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1432 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1433 <dd>
1434 <p>Configures the watcher to wait for status changes of the given
1435<code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1436be detected and should normally be specified as <code>0</code> to let libev choose
1437a suitable value. The memory pointed to by <code>path</code> must point to the same
1438path for as long as the watcher is active.</p>
1439 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1440relative to the attributes at the time the watcher was started (or the
1441last change was detected).</p>
1442 </dd>
1443 <dt>ev_stat_stat (ev_stat *)</dt>
1444 <dd>
1445 <p>Updates the stat buffer immediately with new values. If you change the
1446watched path in your callback, you could call this fucntion to avoid
1447detecting this change (while introducing a race condition). Can also be
1448useful simply to find out the new values.</p>
1449 </dd>
1450 <dt>ev_statdata attr [read-only]</dt>
1451 <dd>
1452 <p>The most-recently detected attributes of the file. Although the type is of
1453<code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1454suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1455was some error while <code>stat</code>ing the file.</p>
1456 </dd>
1457 <dt>ev_statdata prev [read-only]</dt>
1458 <dd>
1459 <p>The previous attributes of the file. The callback gets invoked whenever
1460<code>prev</code> != <code>attr</code>.</p>
1461 </dd>
1462 <dt>ev_tstamp interval [read-only]</dt>
1463 <dd>
1464 <p>The specified interval.</p>
1465 </dd>
1466 <dt>const char *path [read-only]</dt>
1467 <dd>
1468 <p>The filesystem path that is being watched.</p>
1469 </dd>
1470</dl>
1471<p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1472<pre> static void
1473 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1474 {
1475 /* /etc/passwd changed in some way */
1476 if (w-&gt;attr.st_nlink)
1477 {
1478 printf (&quot;passwd current size %ld\n&quot;, (long)w-&gt;attr.st_size);
1479 printf (&quot;passwd current atime %ld\n&quot;, (long)w-&gt;attr.st_mtime);
1480 printf (&quot;passwd current mtime %ld\n&quot;, (long)w-&gt;attr.st_mtime);
1481 }
1482 else
1483 /* you shalt not abuse printf for puts */
1484 puts (&quot;wow, /etc/passwd is not there, expect problems. &quot;
1485 &quot;if this is windows, they already arrived\n&quot;);
1486 }
1487
1488 ...
1489 ev_stat passwd;
1490
1491 ev_stat_init (&amp;passwd, passwd_cb, &quot;/etc/passwd&quot;);
1492 ev_stat_start (loop, &amp;passwd);
1493
1494
1495
1496
1497</pre>
1498
1499</div>
616<h2 id="ev_idle_when_you_ve_got_nothing_bett">ev_idle - when you've got nothing better to do</h2> 1500<h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
617<div id="ev_idle_when_you_ve_got_nothing_bett-2"> 1501<div id="code_ev_idle_code_when_you_ve_got_no-2">
618<p>Idle watchers trigger events when there are no other I/O or timer (or 1502<p>Idle watchers trigger events when no other events of the same or higher
619periodic) events pending. That is, as long as your process is busy 1503priority are pending (prepare, check and other idle watchers do not
620handling sockets or timeouts it will not be called. But when your process 1504count).</p>
621is idle all idle watchers are being called again and again - until 1505<p>That is, as long as your process is busy handling sockets or timeouts
1506(or even signals, imagine) of the same or higher priority it will not be
1507triggered. But when your process is idle (or only lower-priority watchers
1508are pending), the idle watchers are being called once per event loop
622stopped, that is, or your process receives more events.</p> 1509iteration - until stopped, that is, or your process receives more events
1510and becomes busy again with higher priority stuff.</p>
623<p>The most noteworthy effect is that as long as any idle watchers are 1511<p>The most noteworthy effect is that as long as any idle watchers are
624active, the process will not block when waiting for new events.</p> 1512active, the process will not block when waiting for new events.</p>
625<p>Apart from keeping your process non-blocking (which is a useful 1513<p>Apart from keeping your process non-blocking (which is a useful
626effect on its own sometimes), idle watchers are a good place to do 1514effect on its own sometimes), idle watchers are a good place to do
627&quot;pseudo-background processing&quot;, or delay processing stuff to after the 1515&quot;pseudo-background processing&quot;, or delay processing stuff to after the
628event loop has handled all outstanding events.</p> 1516event loop has handled all outstanding events.</p>
1517
1518</div>
1519<h3 id="Watcher_Specific_Functions_and_Data_-7">Watcher-Specific Functions and Data Members</h3>
1520<div id="Watcher_Specific_Functions_and_Data_-2-5">
629<dl> 1521<dl>
630 <dt>ev_idle_init (ev_signal *, callback)</dt> 1522 <dt>ev_idle_init (ev_signal *, callback)</dt>
631 <dd> 1523 <dd>
632 <p>Initialises and configures the idle watcher - it has no parameters of any 1524 <p>Initialises and configures the idle watcher - it has no parameters of any
633kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless, 1525kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
634believe me.</p> 1526believe me.</p>
635 </dd> 1527 </dd>
636</dl> 1528</dl>
1529<p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1530callback, free it. Also, use no error checking, as usual.</p>
1531<pre> static void
1532 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1533 {
1534 free (w);
1535 // now do something you wanted to do when the program has
1536 // no longer asnything immediate to do.
1537 }
637 1538
1539 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1540 ev_idle_init (idle_watcher, idle_cb);
1541 ev_idle_start (loop, idle_cb);
1542
1543
1544
1545
1546</pre>
1547
638</div> 1548</div>
639<h2 id="prepare_and_check_your_hooks_into_th">prepare and check - your hooks into the event loop</h2> 1549<h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
640<div id="prepare_and_check_your_hooks_into_th-2"> 1550<div id="code_ev_prepare_code_and_code_ev_che-2">
641<p>Prepare and check watchers usually (but not always) are used in 1551<p>Prepare and check watchers are usually (but not always) used in tandem:
642tandom. Prepare watchers get invoked before the process blocks and check 1552prepare watchers get invoked before the process blocks and check watchers
643watchers afterwards.</p> 1553afterwards.</p>
1554<p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1555the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1556watchers. Other loops than the current one are fine, however. The
1557rationale behind this is that you do not need to check for recursion in
1558those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1559<code>ev_check</code> so if you have one watcher of each kind they will always be
1560called in pairs bracketing the blocking call.</p>
644<p>Their main purpose is to integrate other event mechanisms into libev. This 1561<p>Their main purpose is to integrate other event mechanisms into libev and
645could be used, for example, to track variable changes, implement your own 1562their use is somewhat advanced. This could be used, for example, to track
646watchers, integrate net-snmp or a coroutine library and lots more.</p> 1563variable changes, implement your own watchers, integrate net-snmp or a
1564coroutine library and lots more. They are also occasionally useful if
1565you cache some data and want to flush it before blocking (for example,
1566in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1567watcher).</p>
647<p>This is done by examining in each prepare call which file descriptors need 1568<p>This is done by examining in each prepare call which file descriptors need
648to be watched by the other library, registering ev_io watchers for them 1569to be watched by the other library, registering <code>ev_io</code> watchers for
649and starting an ev_timer watcher for any timeouts (many libraries provide 1570them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
650just this functionality). Then, in the check watcher you check for any 1571provide just this functionality). Then, in the check watcher you check for
651events that occured (by making your callbacks set soem flags for example) 1572any events that occured (by checking the pending status of all watchers
652and call back into the library.</p> 1573and stopping them) and call back into the library. The I/O and timer
1574callbacks will never actually be called (but must be valid nevertheless,
1575because you never know, you know?).</p>
653<p>As another example, the perl Coro module uses these hooks to integrate 1576<p>As another example, the Perl Coro module uses these hooks to integrate
654coroutines into libev programs, by yielding to other active coroutines 1577coroutines into libev programs, by yielding to other active coroutines
655during each prepare and only letting the process block if no coroutines 1578during each prepare and only letting the process block if no coroutines
656are ready to run.</p> 1579are ready to run (it's actually more complicated: it only runs coroutines
1580with priority higher than or equal to the event loop and one coroutine
1581of lower priority, but only once, using idle watchers to keep the event
1582loop from blocking if lower-priority coroutines are active, thus mapping
1583low-priority coroutines to idle/background tasks).</p>
1584<p>It is recommended to give <code>ev_check</code> watchers highest (<code>EV_MAXPRI</code>)
1585priority, to ensure that they are being run before any other watchers
1586after the poll. Also, <code>ev_check</code> watchers (and <code>ev_prepare</code> watchers,
1587too) should not activate (&quot;feed&quot;) events into libev. While libev fully
1588supports this, they will be called before other <code>ev_check</code> watchers did
1589their job. As <code>ev_check</code> watchers are often used to embed other event
1590loops those other event loops might be in an unusable state until their
1591<code>ev_check</code> watcher ran (always remind yourself to coexist peacefully with
1592others).</p>
1593
1594</div>
1595<h3 id="Watcher_Specific_Functions_and_Data_-8">Watcher-Specific Functions and Data Members</h3>
1596<div id="Watcher_Specific_Functions_and_Data_-2-6">
657<dl> 1597<dl>
658 <dt>ev_prepare_init (ev_prepare *, callback)</dt> 1598 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
659 <dt>ev_check_init (ev_check *, callback)</dt> 1599 <dt>ev_check_init (ev_check *, callback)</dt>
660 <dd> 1600 <dd>
661 <p>Initialises and configures the prepare or check watcher - they have no 1601 <p>Initialises and configures the prepare or check watcher - they have no
662parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code> 1602parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
663macros, but using them is utterly, utterly pointless.</p> 1603macros, but using them is utterly, utterly and completely pointless.</p>
664 </dd> 1604 </dd>
665</dl> 1605</dl>
1606<p>There are a number of principal ways to embed other event loops or modules
1607into libev. Here are some ideas on how to include libadns into libev
1608(there is a Perl module named <code>EV::ADNS</code> that does this, which you could
1609use for an actually working example. Another Perl module named <code>EV::Glib</code>
1610embeds a Glib main context into libev, and finally, <code>Glib::EV</code> embeds EV
1611into the Glib event loop).</p>
1612<p>Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1613and in a check watcher, destroy them and call into libadns. What follows
1614is pseudo-code only of course. This requires you to either use a low
1615priority for the check watcher or use <code>ev_clear_pending</code> explicitly, as
1616the callbacks for the IO/timeout watchers might not have been called yet.</p>
1617<pre> static ev_io iow [nfd];
1618 static ev_timer tw;
666 1619
1620 static void
1621 io_cb (ev_loop *loop, ev_io *w, int revents)
1622 {
1623 }
1624
1625 // create io watchers for each fd and a timer before blocking
1626 static void
1627 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1628 {
1629 int timeout = 3600000;
1630 struct pollfd fds [nfd];
1631 // actual code will need to loop here and realloc etc.
1632 adns_beforepoll (ads, fds, &amp;nfd, &amp;timeout, timeval_from (ev_time ()));
1633
1634 /* the callback is illegal, but won't be called as we stop during check */
1635 ev_timer_init (&amp;tw, 0, timeout * 1e-3);
1636 ev_timer_start (loop, &amp;tw);
1637
1638 // create one ev_io per pollfd
1639 for (int i = 0; i &lt; nfd; ++i)
1640 {
1641 ev_io_init (iow + i, io_cb, fds [i].fd,
1642 ((fds [i].events &amp; POLLIN ? EV_READ : 0)
1643 | (fds [i].events &amp; POLLOUT ? EV_WRITE : 0)));
1644
1645 fds [i].revents = 0;
1646 ev_io_start (loop, iow + i);
1647 }
1648 }
1649
1650 // stop all watchers after blocking
1651 static void
1652 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1653 {
1654 ev_timer_stop (loop, &amp;tw);
1655
1656 for (int i = 0; i &lt; nfd; ++i)
1657 {
1658 // set the relevant poll flags
1659 // could also call adns_processreadable etc. here
1660 struct pollfd *fd = fds + i;
1661 int revents = ev_clear_pending (iow + i);
1662 if (revents &amp; EV_READ ) fd-&gt;revents |= fd-&gt;events &amp; POLLIN;
1663 if (revents &amp; EV_WRITE) fd-&gt;revents |= fd-&gt;events &amp; POLLOUT;
1664
1665 // now stop the watcher
1666 ev_io_stop (loop, iow + i);
1667 }
1668
1669 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1670 }
1671
1672</pre>
1673<p>Method 2: This would be just like method 1, but you run <code>adns_afterpoll</code>
1674in the prepare watcher and would dispose of the check watcher.</p>
1675<p>Method 3: If the module to be embedded supports explicit event
1676notification (adns does), you can also make use of the actual watcher
1677callbacks, and only destroy/create the watchers in the prepare watcher.</p>
1678<pre> static void
1679 timer_cb (EV_P_ ev_timer *w, int revents)
1680 {
1681 adns_state ads = (adns_state)w-&gt;data;
1682 update_now (EV_A);
1683
1684 adns_processtimeouts (ads, &amp;tv_now);
1685 }
1686
1687 static void
1688 io_cb (EV_P_ ev_io *w, int revents)
1689 {
1690 adns_state ads = (adns_state)w-&gt;data;
1691 update_now (EV_A);
1692
1693 if (revents &amp; EV_READ ) adns_processreadable (ads, w-&gt;fd, &amp;tv_now);
1694 if (revents &amp; EV_WRITE) adns_processwriteable (ads, w-&gt;fd, &amp;tv_now);
1695 }
1696
1697 // do not ever call adns_afterpoll
1698
1699</pre>
1700<p>Method 4: Do not use a prepare or check watcher because the module you
1701want to embed is too inflexible to support it. Instead, youc na override
1702their poll function. The drawback with this solution is that the main
1703loop is now no longer controllable by EV. The <code>Glib::EV</code> module does
1704this.</p>
1705<pre> static gint
1706 event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1707 {
1708 int got_events = 0;
1709
1710 for (n = 0; n &lt; nfds; ++n)
1711 // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1712
1713 if (timeout &gt;= 0)
1714 // create/start timer
1715
1716 // poll
1717 ev_loop (EV_A_ 0);
1718
1719 // stop timer again
1720 if (timeout &gt;= 0)
1721 ev_timer_stop (EV_A_ &amp;to);
1722
1723 // stop io watchers again - their callbacks should have set
1724 for (n = 0; n &lt; nfds; ++n)
1725 ev_io_stop (EV_A_ iow [n]);
1726
1727 return got_events;
1728 }
1729
1730
1731
1732
1733</pre>
1734
667</div> 1735</div>
668<h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p> 1736<h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1737<div id="code_ev_embed_code_when_one_backend_-2">
1738<p>This is a rather advanced watcher type that lets you embed one event loop
1739into another (currently only <code>ev_io</code> events are supported in the embedded
1740loop, other types of watchers might be handled in a delayed or incorrect
1741fashion and must not be used).</p>
1742<p>There are primarily two reasons you would want that: work around bugs and
1743prioritise I/O.</p>
1744<p>As an example for a bug workaround, the kqueue backend might only support
1745sockets on some platform, so it is unusable as generic backend, but you
1746still want to make use of it because you have many sockets and it scales
1747so nicely. In this case, you would create a kqueue-based loop and embed it
1748into your default loop (which might use e.g. poll). Overall operation will
1749be a bit slower because first libev has to poll and then call kevent, but
1750at least you can use both at what they are best.</p>
1751<p>As for prioritising I/O: rarely you have the case where some fds have
1752to be watched and handled very quickly (with low latency), and even
1753priorities and idle watchers might have too much overhead. In this case
1754you would put all the high priority stuff in one loop and all the rest in
1755a second one, and embed the second one in the first.</p>
1756<p>As long as the watcher is active, the callback will be invoked every time
1757there might be events pending in the embedded loop. The callback must then
1758call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1759their callbacks (you could also start an idle watcher to give the embedded
1760loop strictly lower priority for example). You can also set the callback
1761to <code>0</code>, in which case the embed watcher will automatically execute the
1762embedded loop sweep.</p>
1763<p>As long as the watcher is started it will automatically handle events. The
1764callback will be invoked whenever some events have been handled. You can
1765set the callback to <code>0</code> to avoid having to specify one if you are not
1766interested in that.</p>
1767<p>Also, there have not currently been made special provisions for forking:
1768when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1769but you will also have to stop and restart any <code>ev_embed</code> watchers
1770yourself.</p>
1771<p>Unfortunately, not all backends are embeddable, only the ones returned by
1772<code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1773portable one.</p>
1774<p>So when you want to use this feature you will always have to be prepared
1775that you cannot get an embeddable loop. The recommended way to get around
1776this is to have a separate variables for your embeddable loop, try to
1777create it, and if that fails, use the normal loop for everything:</p>
1778<pre> struct ev_loop *loop_hi = ev_default_init (0);
1779 struct ev_loop *loop_lo = 0;
1780 struct ev_embed embed;
1781
1782 // see if there is a chance of getting one that works
1783 // (remember that a flags value of 0 means autodetection)
1784 loop_lo = ev_embeddable_backends () &amp; ev_recommended_backends ()
1785 ? ev_loop_new (ev_embeddable_backends () &amp; ev_recommended_backends ())
1786 : 0;
1787
1788 // if we got one, then embed it, otherwise default to loop_hi
1789 if (loop_lo)
1790 {
1791 ev_embed_init (&amp;embed, 0, loop_lo);
1792 ev_embed_start (loop_hi, &amp;embed);
1793 }
1794 else
1795 loop_lo = loop_hi;
1796
1797</pre>
1798
1799</div>
1800<h3 id="Watcher_Specific_Functions_and_Data_-9">Watcher-Specific Functions and Data Members</h3>
1801<div id="Watcher_Specific_Functions_and_Data_-2-7">
1802<dl>
1803 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1804 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1805 <dd>
1806 <p>Configures the watcher to embed the given loop, which must be
1807embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1808invoked automatically, otherwise it is the responsibility of the callback
1809to invoke it (it will continue to be called until the sweep has been done,
1810if you do not want thta, you need to temporarily stop the embed watcher).</p>
1811 </dd>
1812 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1813 <dd>
1814 <p>Make a single, non-blocking sweep over the embedded loop. This works
1815similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1816apropriate way for embedded loops.</p>
1817 </dd>
1818 <dt>struct ev_loop *loop [read-only]</dt>
1819 <dd>
1820 <p>The embedded event loop.</p>
1821 </dd>
1822</dl>
1823
1824
1825
1826
1827
1828</div>
1829<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>
1830<div id="code_ev_fork_code_the_audacity_to_re-2">
1831<p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1832whoever is a good citizen cared to tell libev about it by calling
1833<code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1834event loop blocks next and before <code>ev_check</code> watchers are being called,
1835and only in the child after the fork. If whoever good citizen calling
1836<code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1837handlers will be invoked, too, of course.</p>
1838
1839</div>
1840<h3 id="Watcher_Specific_Functions_and_Data_-10">Watcher-Specific Functions and Data Members</h3>
1841<div id="Watcher_Specific_Functions_and_Data_-2-8">
1842<dl>
1843 <dt>ev_fork_init (ev_signal *, callback)</dt>
1844 <dd>
1845 <p>Initialises and configures the fork watcher - it has no parameters of any
1846kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1847believe me.</p>
1848 </dd>
1849</dl>
1850
1851
1852
1853
1854
1855</div>
1856<h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
669<div id="OTHER_FUNCTIONS_CONTENT"> 1857<div id="OTHER_FUNCTIONS_CONTENT">
670<p>There are some other fucntions of possible interest. Described. Here. Now.</p> 1858<p>There are some other functions of possible interest. Described. Here. Now.</p>
671<dl> 1859<dl>
672 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt> 1860 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
673 <dd> 1861 <dd>
674 <p>This function combines a simple timer and an I/O watcher, calls your 1862 <p>This function combines a simple timer and an I/O watcher, calls your
675callback on whichever event happens first and automatically stop both 1863callback on whichever event happens first and automatically stop both
676watchers. This is useful if you want to wait for a single event on an fd 1864watchers. This is useful if you want to wait for a single event on an fd
677or timeout without havign to allocate/configure/start/stop/free one or 1865or timeout without having to allocate/configure/start/stop/free one or
678more watchers yourself.</p> 1866more watchers yourself.</p>
679 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events is 1867 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
680ignored. Otherwise, an ev_io watcher for the given <code>fd</code> and <code>events</code> set 1868is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
681will be craeted and started.</p> 1869<code>events</code> set will be craeted and started.</p>
682 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be 1870 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
683started. Otherwise an ev_timer watcher with after = <code>timeout</code> (and repeat 1871started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
684= 0) will be started.</p> 1872repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1873dubious value.</p>
685 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and 1874 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
686gets passed an events set (normally a combination of EV_ERROR, EV_READ, 1875passed an <code>revents</code> set like normal event callbacks (a combination of
687EV_WRITE or EV_TIMEOUT) and the <code>arg</code> value passed to <code>ev_once</code>:</p> 1876<code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1877value passed to <code>ev_once</code>:</p>
688<pre> static void stdin_ready (int revents, void *arg) 1878<pre> static void stdin_ready (int revents, void *arg)
689 { 1879 {
690 if (revents &amp; EV_TIMEOUT) 1880 if (revents &amp; EV_TIMEOUT)
691 /* doh, nothing entered */ 1881 /* doh, nothing entered */;
692 else if (revents &amp; EV_READ) 1882 else if (revents &amp; EV_READ)
693 /* stdin might have data for us, joy! */ 1883 /* stdin might have data for us, joy! */;
694 } 1884 }
695 1885
696 ev_once (STDIN_FILENO, EV_READm 10., stdin_ready, 0); 1886 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
697 1887
698</pre> 1888</pre>
699 </dd> 1889 </dd>
700 <dt>ev_feed_event (loop, watcher, int events)</dt> 1890 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
701 <dd> 1891 <dd>
702 <p>Feeds the given event set into the event loop, as if the specified event 1892 <p>Feeds the given event set into the event loop, as if the specified event
703has happened for the specified watcher (which must be a pointer to an 1893had happened for the specified watcher (which must be a pointer to an
704initialised but not necessarily active event watcher).</p> 1894initialised but not necessarily started event watcher).</p>
705 </dd> 1895 </dd>
706 <dt>ev_feed_fd_event (loop, int fd, int revents)</dt> 1896 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
707 <dd> 1897 <dd>
708 <p>Feed an event on the given fd, as if a file descriptor backend detected it.</p> 1898 <p>Feed an event on the given fd, as if a file descriptor backend detected
1899the given events it.</p>
709 </dd> 1900 </dd>
710 <dt>ev_feed_signal_event (loop, int signum)</dt> 1901 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
711 <dd> 1902 <dd>
712 <p>Feed an event as if the given signal occured (loop must be the default loop!).</p> 1903 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1904loop!).</p>
713 </dd> 1905 </dd>
714</dl> 1906</dl>
715 1907
1908
1909
1910
1911
716</div> 1912</div>
717<h1 id="AUTHOR">AUTHOR</h1><p><a href="#TOP" class="toplink">Top</a></p> 1913<h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1914<div id="LIBEVENT_EMULATION_CONTENT">
1915<p>Libev offers a compatibility emulation layer for libevent. It cannot
1916emulate the internals of libevent, so here are some usage hints:</p>
1917<dl>
1918 <dt>* Use it by including &lt;event.h&gt;, as usual.</dt>
1919 <dt>* The following members are fully supported: ev_base, ev_callback,
1920ev_arg, ev_fd, ev_res, ev_events.</dt>
1921 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1922maintained by libev, it does not work exactly the same way as in libevent (consider
1923it a private API).</dt>
1924 <dt>* Priorities are not currently supported. Initialising priorities
1925will fail and all watchers will have the same priority, even though there
1926is an ev_pri field.</dt>
1927 <dt>* Other members are not supported.</dt>
1928 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1929to use the libev header file and library.</dt>
1930</dl>
1931
1932</div>
1933<h1 id="C_SUPPORT">C++ SUPPORT</h1>
1934<div id="C_SUPPORT_CONTENT">
1935<p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1936you to use some convinience methods to start/stop watchers and also change
1937the callback model to a model using method callbacks on objects.</p>
1938<p>To use it,</p>
1939<pre> #include &lt;ev++.h&gt;
1940
1941</pre>
1942<p>This automatically includes <cite>ev.h</cite> and puts all of its definitions (many
1943of them macros) into the global namespace. All C++ specific things are
1944put into the <code>ev</code> namespace. It should support all the same embedding
1945options as <cite>ev.h</cite>, most notably <code>EV_MULTIPLICITY</code>.</p>
1946<p>Care has been taken to keep the overhead low. The only data member the C++
1947classes add (compared to plain C-style watchers) is the event loop pointer
1948that the watcher is associated with (or no additional members at all if
1949you disable <code>EV_MULTIPLICITY</code> when embedding libev).</p>
1950<p>Currently, functions, and static and non-static member functions can be
1951used as callbacks. Other types should be easy to add as long as they only
1952need one additional pointer for context. If you need support for other
1953types of functors please contact the author (preferably after implementing
1954it).</p>
1955<p>Here is a list of things available in the <code>ev</code> namespace:</p>
1956<dl>
1957 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1958 <dd>
1959 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1960macros from <cite>ev.h</cite>.</p>
1961 </dd>
1962 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1963 <dd>
1964 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1965 </dd>
1966 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1967 <dd>
1968 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1969the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1970which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1971defines by many implementations.</p>
1972 <p>All of those classes have these methods:</p>
1973 <p>
1974 <dl>
1975 <dt>ev::TYPE::TYPE ()</dt>
1976 <dt>ev::TYPE::TYPE (struct ev_loop *)</dt>
1977 <dt>ev::TYPE::~TYPE</dt>
1978 <dd>
1979 <p>The constructor (optionally) takes an event loop to associate the watcher
1980with. If it is omitted, it will use <code>EV_DEFAULT</code>.</p>
1981 <p>The constructor calls <code>ev_init</code> for you, which means you have to call the
1982<code>set</code> method before starting it.</p>
1983 <p>It will not set a callback, however: You have to call the templated <code>set</code>
1984method to set a callback before you can start the watcher.</p>
1985 <p>(The reason why you have to use a method is a limitation in C++ which does
1986not allow explicit template arguments for constructors).</p>
1987 <p>The destructor automatically stops the watcher if it is active.</p>
1988 </dd>
1989 <dt>w-&gt;set&lt;class, &amp;class::method&gt; (object *)</dt>
1990 <dd>
1991 <p>This method sets the callback method to call. The method has to have a
1992signature of <code>void (*)(ev_TYPE &amp;, int)</code>, it receives the watcher as
1993first argument and the <code>revents</code> as second. The object must be given as
1994parameter and is stored in the <code>data</code> member of the watcher.</p>
1995 <p>This method synthesizes efficient thunking code to call your method from
1996the C callback that libev requires. If your compiler can inline your
1997callback (i.e. it is visible to it at the place of the <code>set</code> call and
1998your compiler is good :), then the method will be fully inlined into the
1999thunking function, making it as fast as a direct C callback.</p>
2000 <p>Example: simple class declaration and watcher initialisation</p>
2001<pre> struct myclass
2002 {
2003 void io_cb (ev::io &amp;w, int revents) { }
2004 }
2005
2006 myclass obj;
2007 ev::io iow;
2008 iow.set &lt;myclass, &amp;myclass::io_cb&gt; (&amp;obj);
2009
2010</pre>
2011 </dd>
2012 <dt>w-&gt;set&lt;function&gt; (void *data = 0)</dt>
2013 <dd>
2014 <p>Also sets a callback, but uses a static method or plain function as
2015callback. The optional <code>data</code> argument will be stored in the watcher's
2016<code>data</code> member and is free for you to use.</p>
2017 <p>The prototype of the <code>function</code> must be <code>void (*)(ev::TYPE &amp;w, int)</code>.</p>
2018 <p>See the method-<code>set</code> above for more details.</p>
2019 <p>Example:</p>
2020<pre> static void io_cb (ev::io &amp;w, int revents) { }
2021 iow.set &lt;io_cb&gt; ();
2022
2023</pre>
2024 </dd>
2025 <dt>w-&gt;set (struct ev_loop *)</dt>
2026 <dd>
2027 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
2028do this when the watcher is inactive (and not pending either).</p>
2029 </dd>
2030 <dt>w-&gt;set ([args])</dt>
2031 <dd>
2032 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
2033called at least once. Unlike the C counterpart, an active watcher gets
2034automatically stopped and restarted when reconfiguring it with this
2035method.</p>
2036 </dd>
2037 <dt>w-&gt;start ()</dt>
2038 <dd>
2039 <p>Starts the watcher. Note that there is no <code>loop</code> argument, as the
2040constructor already stores the event loop.</p>
2041 </dd>
2042 <dt>w-&gt;stop ()</dt>
2043 <dd>
2044 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
2045 </dd>
2046 <dt>w-&gt;again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
2047 <dd>
2048 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
2049<code>ev_TYPE_again</code> function.</p>
2050 </dd>
2051 <dt>w-&gt;sweep () <code>ev::embed</code> only</dt>
2052 <dd>
2053 <p>Invokes <code>ev_embed_sweep</code>.</p>
2054 </dd>
2055 <dt>w-&gt;update () <code>ev::stat</code> only</dt>
2056 <dd>
2057 <p>Invokes <code>ev_stat_stat</code>.</p>
2058 </dd>
2059 </dl>
2060 </p>
2061 </dd>
2062</dl>
2063<p>Example: Define a class with an IO and idle watcher, start one of them in
2064the constructor.</p>
2065<pre> class myclass
2066 {
2067 ev_io io; void io_cb (ev::io &amp;w, int revents);
2068 ev_idle idle void idle_cb (ev::idle &amp;w, int revents);
2069
2070 myclass ();
2071 }
2072
2073 myclass::myclass (int fd)
2074 {
2075 io .set &lt;myclass, &amp;myclass::io_cb &gt; (this);
2076 idle.set &lt;myclass, &amp;myclass::idle_cb&gt; (this);
2077
2078 io.start (fd, ev::READ);
2079 }
2080
2081
2082
2083
2084</pre>
2085
2086</div>
2087<h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
2088<div id="MACRO_MAGIC_CONTENT">
2089<p>Libev can be compiled with a variety of options, the most fundemantal is
2090<code>EV_MULTIPLICITY</code>. This option determines whether (most) functions and
2091callbacks have an initial <code>struct ev_loop *</code> argument.</p>
2092<p>To make it easier to write programs that cope with either variant, the
2093following macros are defined:</p>
2094<dl>
2095 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
2096 <dd>
2097 <p>This provides the loop <i>argument</i> for functions, if one is required (&quot;ev
2098loop argument&quot;). The <code>EV_A</code> form is used when this is the sole argument,
2099<code>EV_A_</code> is used when other arguments are following. Example:</p>
2100<pre> ev_unref (EV_A);
2101 ev_timer_add (EV_A_ watcher);
2102 ev_loop (EV_A_ 0);
2103
2104</pre>
2105 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
2106which is often provided by the following macro.</p>
2107 </dd>
2108 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
2109 <dd>
2110 <p>This provides the loop <i>parameter</i> for functions, if one is required (&quot;ev
2111loop parameter&quot;). The <code>EV_P</code> form is used when this is the sole parameter,
2112<code>EV_P_</code> is used when other parameters are following. Example:</p>
2113<pre> // this is how ev_unref is being declared
2114 static void ev_unref (EV_P);
2115
2116 // this is how you can declare your typical callback
2117 static void cb (EV_P_ ev_timer *w, int revents)
2118
2119</pre>
2120 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
2121suitable for use with <code>EV_A</code>.</p>
2122 </dd>
2123 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
2124 <dd>
2125 <p>Similar to the other two macros, this gives you the value of the default
2126loop, if multiple loops are supported (&quot;ev loop default&quot;).</p>
2127 </dd>
2128</dl>
2129<p>Example: Declare and initialise a check watcher, utilising the above
2130macros so it will work regardless of whether multiple loops are supported
2131or not.</p>
2132<pre> static void
2133 check_cb (EV_P_ ev_timer *w, int revents)
2134 {
2135 ev_check_stop (EV_A_ w);
2136 }
2137
2138 ev_check check;
2139 ev_check_init (&amp;check, check_cb);
2140 ev_check_start (EV_DEFAULT_ &amp;check);
2141 ev_loop (EV_DEFAULT_ 0);
2142
2143</pre>
2144
2145</div>
2146<h1 id="EMBEDDING">EMBEDDING</h1>
2147<div id="EMBEDDING_CONTENT">
2148<p>Libev can (and often is) directly embedded into host
2149applications. Examples of applications that embed it include the Deliantra
2150Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2151and rxvt-unicode.</p>
2152<p>The goal is to enable you to just copy the neecssary files into your
2153source directory without having to change even a single line in them, so
2154you can easily upgrade by simply copying (or having a checked-out copy of
2155libev somewhere in your source tree).</p>
2156
2157</div>
2158<h2 id="FILESETS">FILESETS</h2>
2159<div id="FILESETS_CONTENT">
2160<p>Depending on what features you need you need to include one or more sets of files
2161in your app.</p>
2162
2163</div>
2164<h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
2165<div id="CORE_EVENT_LOOP_CONTENT">
2166<p>To include only the libev core (all the <code>ev_*</code> functions), with manual
2167configuration (no autoconf):</p>
2168<pre> #define EV_STANDALONE 1
2169 #include &quot;ev.c&quot;
2170
2171</pre>
2172<p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
2173single C source file only to provide the function implementations. To use
2174it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
2175done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
2176where you can put other configuration options):</p>
2177<pre> #define EV_STANDALONE 1
2178 #include &quot;ev.h&quot;
2179
2180</pre>
2181<p>Both header files and implementation files can be compiled with a C++
2182compiler (at least, thats a stated goal, and breakage will be treated
2183as a bug).</p>
2184<p>You need the following files in your source tree, or in a directory
2185in your include path (e.g. in libev/ when using -Ilibev):</p>
2186<pre> ev.h
2187 ev.c
2188 ev_vars.h
2189 ev_wrap.h
2190
2191 ev_win32.c required on win32 platforms only
2192
2193 ev_select.c only when select backend is enabled (which is enabled by default)
2194 ev_poll.c only when poll backend is enabled (disabled by default)
2195 ev_epoll.c only when the epoll backend is enabled (disabled by default)
2196 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2197 ev_port.c only when the solaris port backend is enabled (disabled by default)
2198
2199</pre>
2200<p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
2201to compile this single file.</p>
2202
2203</div>
2204<h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
2205<div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
2206<p>To include the libevent compatibility API, also include:</p>
2207<pre> #include &quot;event.c&quot;
2208
2209</pre>
2210<p>in the file including <cite>ev.c</cite>, and:</p>
2211<pre> #include &quot;event.h&quot;
2212
2213</pre>
2214<p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
2215<p>You need the following additional files for this:</p>
2216<pre> event.h
2217 event.c
2218
2219</pre>
2220
2221</div>
2222<h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
2223<div id="AUTOCONF_SUPPORT_CONTENT">
2224<p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
2225whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
2226<cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
2227include <cite>config.h</cite> and configure itself accordingly.</p>
2228<p>For this of course you need the m4 file:</p>
2229<pre> libev.m4
2230
2231</pre>
2232
2233</div>
2234<h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
2235<div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
2236<p>Libev can be configured via a variety of preprocessor symbols you have to define
2237before including any of its files. The default is not to build for multiplicity
2238and only include the select backend.</p>
2239<dl>
2240 <dt>EV_STANDALONE</dt>
2241 <dd>
2242 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
2243keeps libev from including <cite>config.h</cite>, and it also defines dummy
2244implementations for some libevent functions (such as logging, which is not
2245supported). It will also not define any of the structs usually found in
2246<cite>event.h</cite> that are not directly supported by the libev core alone.</p>
2247 </dd>
2248 <dt>EV_USE_MONOTONIC</dt>
2249 <dd>
2250 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2251monotonic clock option at both compiletime and runtime. Otherwise no use
2252of the monotonic clock option will be attempted. If you enable this, you
2253usually have to link against librt or something similar. Enabling it when
2254the functionality isn't available is safe, though, althoguh you have
2255to make sure you link against any libraries where the <code>clock_gettime</code>
2256function is hiding in (often <cite>-lrt</cite>).</p>
2257 </dd>
2258 <dt>EV_USE_REALTIME</dt>
2259 <dd>
2260 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2261realtime clock option at compiletime (and assume its availability at
2262runtime if successful). Otherwise no use of the realtime clock option will
2263be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
2264(CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
2265in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
2266 </dd>
2267 <dt>EV_USE_SELECT</dt>
2268 <dd>
2269 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
2270<code>select</code>(2) backend. No attempt at autodetection will be done: if no
2271other method takes over, select will be it. Otherwise the select backend
2272will not be compiled in.</p>
2273 </dd>
2274 <dt>EV_SELECT_USE_FD_SET</dt>
2275 <dd>
2276 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
2277structure. This is useful if libev doesn't compile due to a missing
2278<code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
2279exotic systems. This usually limits the range of file descriptors to some
2280low limit such as 1024 or might have other limitations (winsocket only
2281allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
2282influence the size of the <code>fd_set</code> used.</p>
2283 </dd>
2284 <dt>EV_SELECT_IS_WINSOCKET</dt>
2285 <dd>
2286 <p>When defined to <code>1</code>, the select backend will assume that
2287select/socket/connect etc. don't understand file descriptors but
2288wants osf handles on win32 (this is the case when the select to
2289be used is the winsock select). This means that it will call
2290<code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2291it is assumed that all these functions actually work on fds, even
2292on win32. Should not be defined on non-win32 platforms.</p>
2293 </dd>
2294 <dt>EV_USE_POLL</dt>
2295 <dd>
2296 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2297backend. Otherwise it will be enabled on non-win32 platforms. It
2298takes precedence over select.</p>
2299 </dd>
2300 <dt>EV_USE_EPOLL</dt>
2301 <dd>
2302 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2303<code>epoll</code>(7) backend. Its availability will be detected at runtime,
2304otherwise another method will be used as fallback. This is the
2305preferred backend for GNU/Linux systems.</p>
2306 </dd>
2307 <dt>EV_USE_KQUEUE</dt>
2308 <dd>
2309 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2310<code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2311otherwise another method will be used as fallback. This is the preferred
2312backend for BSD and BSD-like systems, although on most BSDs kqueue only
2313supports some types of fds correctly (the only platform we found that
2314supports ptys for example was NetBSD), so kqueue might be compiled in, but
2315not be used unless explicitly requested. The best way to use it is to find
2316out whether kqueue supports your type of fd properly and use an embedded
2317kqueue loop.</p>
2318 </dd>
2319 <dt>EV_USE_PORT</dt>
2320 <dd>
2321 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
232210 port style backend. Its availability will be detected at runtime,
2323otherwise another method will be used as fallback. This is the preferred
2324backend for Solaris 10 systems.</p>
2325 </dd>
2326 <dt>EV_USE_DEVPOLL</dt>
2327 <dd>
2328 <p>reserved for future expansion, works like the USE symbols above.</p>
2329 </dd>
2330 <dt>EV_USE_INOTIFY</dt>
2331 <dd>
2332 <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2333interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2334be detected at runtime.</p>
2335 </dd>
2336 <dt>EV_H</dt>
2337 <dd>
2338 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2339undefined 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
2340can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2341 </dd>
2342 <dt>EV_CONFIG_H</dt>
2343 <dd>
2344 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2345<cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2346<code>EV_H</code>, above.</p>
2347 </dd>
2348 <dt>EV_EVENT_H</dt>
2349 <dd>
2350 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2351of how the <cite>event.h</cite> header can be found.</p>
2352 </dd>
2353 <dt>EV_PROTOTYPES</dt>
2354 <dd>
2355 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2356prototypes, but still define all the structs and other symbols. This is
2357occasionally useful if you want to provide your own wrapper functions
2358around libev functions.</p>
2359 </dd>
2360 <dt>EV_MULTIPLICITY</dt>
2361 <dd>
2362 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2363will have the <code>struct ev_loop *</code> as first argument, and you can create
2364additional independent event loops. Otherwise there will be no support
2365for multiple event loops and there is no first event loop pointer
2366argument. Instead, all functions act on the single default loop.</p>
2367 </dd>
2368 <dt>EV_MINPRI</dt>
2369 <dt>EV_MAXPRI</dt>
2370 <dd>
2371 <p>The range of allowed priorities. <code>EV_MINPRI</code> must be smaller or equal to
2372<code>EV_MAXPRI</code>, but otherwise there are no non-obvious limitations. You can
2373provide for more priorities by overriding those symbols (usually defined
2374to be <code>-2</code> and <code>2</code>, respectively).</p>
2375 <p>When doing priority-based operations, libev usually has to linearly search
2376all the priorities, so having many of them (hundreds) uses a lot of space
2377and time, so using the defaults of five priorities (-2 .. +2) is usually
2378fine.</p>
2379 <p>If your embedding app does not need any priorities, defining these both to
2380<code>0</code> will save some memory and cpu.</p>
2381 </dd>
2382 <dt>EV_PERIODIC_ENABLE</dt>
2383 <dd>
2384 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2385defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2386code.</p>
2387 </dd>
2388 <dt>EV_IDLE_ENABLE</dt>
2389 <dd>
2390 <p>If undefined or defined to be <code>1</code>, then idle watchers are supported. If
2391defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2392code.</p>
2393 </dd>
2394 <dt>EV_EMBED_ENABLE</dt>
2395 <dd>
2396 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2397defined to be <code>0</code>, then they are not.</p>
2398 </dd>
2399 <dt>EV_STAT_ENABLE</dt>
2400 <dd>
2401 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2402defined to be <code>0</code>, then they are not.</p>
2403 </dd>
2404 <dt>EV_FORK_ENABLE</dt>
2405 <dd>
2406 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2407defined to be <code>0</code>, then they are not.</p>
2408 </dd>
2409 <dt>EV_MINIMAL</dt>
2410 <dd>
2411 <p>If you need to shave off some kilobytes of code at the expense of some
2412speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2413some inlining decisions, saves roughly 30% codesize of amd64.</p>
2414 </dd>
2415 <dt>EV_PID_HASHSIZE</dt>
2416 <dd>
2417 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2418pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2419than enough. If you need to manage thousands of children you might want to
2420increase this value (<i>must</i> be a power of two).</p>
2421 </dd>
2422 <dt>EV_INOTIFY_HASHSIZE</dt>
2423 <dd>
2424 <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2425inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2426usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2427watchers you might want to increase this value (<i>must</i> be a power of
2428two).</p>
2429 </dd>
2430 <dt>EV_COMMON</dt>
2431 <dd>
2432 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2433this macro to a something else you can include more and other types of
2434members. You have to define it each time you include one of the files,
2435though, and it must be identical each time.</p>
2436 <p>For example, the perl EV module uses something like this:</p>
2437<pre> #define EV_COMMON \
2438 SV *self; /* contains this struct */ \
2439 SV *cb_sv, *fh /* note no trailing &quot;;&quot; */
2440
2441</pre>
2442 </dd>
2443 <dt>EV_CB_DECLARE (type)</dt>
2444 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2445 <dt>ev_set_cb (ev, cb)</dt>
2446 <dd>
2447 <p>Can be used to change the callback member declaration in each watcher,
2448and the way callbacks are invoked and set. Must expand to a struct member
2449definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2450their default definitions. One possible use for overriding these is to
2451avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2452method calls instead of plain function calls in C++.</p>
2453
2454</div>
2455<h2 id="EXAMPLES">EXAMPLES</h2>
2456<div id="EXAMPLES_CONTENT">
2457 <p>For a real-world example of a program the includes libev
2458verbatim, you can have a look at the EV perl module
2459(<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2460the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2461interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2462will be compiled. It is pretty complex because it provides its own header
2463file.</p>
2464 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2465that everybody includes and which overrides some configure choices:</p>
2466<pre> #define EV_MINIMAL 1
2467 #define EV_USE_POLL 0
2468 #define EV_MULTIPLICITY 0
2469 #define EV_PERIODIC_ENABLE 0
2470 #define EV_STAT_ENABLE 0
2471 #define EV_FORK_ENABLE 0
2472 #define EV_CONFIG_H &lt;config.h&gt;
2473 #define EV_MINPRI 0
2474 #define EV_MAXPRI 0
2475
2476 #include &quot;ev++.h&quot;
2477
2478</pre>
2479 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2480<pre> #include &quot;ev_cpp.h&quot;
2481 #include &quot;ev.c&quot;
2482
2483
2484
2485
2486</pre>
2487
2488</div>
2489<h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2490<div id="COMPLEXITIES_CONTENT">
2491 <p>In this section the complexities of (many of) the algorithms used inside
2492libev will be explained. For complexity discussions about backends see the
2493documentation for <code>ev_default_init</code>.</p>
2494 <p>All of the following are about amortised time: If an array needs to be
2495extended, libev needs to realloc and move the whole array, but this
2496happens asymptotically never with higher number of elements, so O(1) might
2497mean it might do a lengthy realloc operation in rare cases, but on average
2498it is much faster and asymptotically approaches constant time.</p>
2499 <p>
2500 <dl>
2501 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2502 <dd>
2503 <p>This means that, when you have a watcher that triggers in one hour and
2504there are 100 watchers that would trigger before that then inserting will
2505have to skip those 100 watchers.</p>
2506 </dd>
2507 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2508 <dd>
2509 <p>That means that for changing a timer costs less than removing/adding them
2510as only the relative motion in the event queue has to be paid for.</p>
2511 </dd>
2512 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2513 <dd>
2514 <p>These just add the watcher into an array or at the head of a list.
2515=item Stopping check/prepare/idle watchers: O(1)</p>
2516 </dd>
2517 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2518 <dd>
2519 <p>These watchers are stored in lists then need to be walked to find the
2520correct watcher to remove. The lists are usually short (you don't usually
2521have many watchers waiting for the same fd or signal).</p>
2522 </dd>
2523 <dt>Finding the next timer per loop iteration: O(1)</dt>
2524 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2525 <dd>
2526 <p>A change means an I/O watcher gets started or stopped, which requires
2527libev to recalculate its status (and possibly tell the kernel).</p>
2528 </dd>
2529 <dt>Activating one watcher: O(1)</dt>
2530 <dt>Priority handling: O(number_of_priorities)</dt>
2531 <dd>
2532 <p>Priorities are implemented by allocating some space for each
2533priority. When doing priority-based operations, libev usually has to
2534linearly search all the priorities.</p>
2535 </dd>
2536 </dl>
2537 </p>
2538
2539
2540
2541
2542
2543</div>
2544<h1 id="AUTHOR">AUTHOR</h1>
718<div id="AUTHOR_CONTENT"> 2545<div id="AUTHOR_CONTENT">
719<p>Marc Lehmann &lt;libev@schmorp.de&gt;.</p> 2546 <p>Marc Lehmann &lt;libev@schmorp.de&gt;.</p>
720 2547
721</div> 2548</div>
722</div></body> 2549</div></body>
723</html> 2550</html>

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines