1 <?xml version="1.0" encoding="UTF-8"?>
2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
3 <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
6 <meta name="description" content="Pod documentation for libev" />
7 <meta name="inputfile" content="<standard input>" />
8 <meta name="outputfile" content="<standard output>" />
9 <meta name="created" content="Tue Nov 27 21:14:27 2007" />
10 <meta name="generator" content="Pod::Xhtml 1.57" />
11 <link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
15 <h3 id="TOP">Index</h3>
17 <ul><li><a href="#NAME">NAME</a></li>
18 <li><a href="#SYNOPSIS">SYNOPSIS</a></li>
19 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
20 <li><a href="#FEATURES">FEATURES</a></li>
21 <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
22 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
23 <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
24 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
25 <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
26 <ul><li><a href="#GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</a></li>
27 <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
30 <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
31 <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</a></li>
32 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
33 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
34 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
35 <li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
36 <li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
37 <li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</a></li>
38 <li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</a></li>
39 <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
40 <li><a href="#code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</a></li>
43 <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
44 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
45 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
46 <li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
47 <li><a href="#EMBEDDING">EMBEDDING</a>
48 <ul><li><a href="#FILESETS">FILESETS</a>
49 <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
50 <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
51 <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
54 <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
55 <li><a href="#EXAMPLES">EXAMPLES</a></li>
58 <li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
59 <li><a href="#AUTHOR">AUTHOR</a>
64 <h1 id="NAME">NAME</h1><p><a href="#TOP" class="toplink">Top</a></p>
65 <div id="NAME_CONTENT">
66 <p>libev - a high performance full-featured event loop written in C</p>
69 <h1 id="SYNOPSIS">SYNOPSIS</h1><p><a href="#TOP" class="toplink">Top</a></p>
70 <div id="SYNOPSIS_CONTENT">
71 <pre> /* this is the only header you need */
74 /* what follows is a fully working example program */
76 ev_timer timeout_watcher;
78 /* called when data readable on stdin */
80 stdin_cb (EV_P_ struct ev_io *w, int revents)
82 /* puts ("stdin ready"); */
83 ev_io_stop (EV_A_ w); /* just a syntax example */
84 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
88 timeout_cb (EV_P_ struct ev_timer *w, int revents)
90 /* puts ("timeout"); */
91 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
97 struct ev_loop *loop = ev_default_loop (0);
99 /* initialise an io watcher, then start it */
100 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
101 ev_io_start (loop, &stdin_watcher);
103 /* simple non-repeating 5.5 second timeout */
104 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
105 ev_timer_start (loop, &timeout_watcher);
107 /* loop till timeout or data ready */
116 <h1 id="DESCRIPTION">DESCRIPTION</h1><p><a href="#TOP" class="toplink">Top</a></p>
117 <div id="DESCRIPTION_CONTENT">
118 <p>Libev is an event loop: you register interest in certain events (such as a
119 file descriptor being readable or a timeout occuring), and it will manage
120 these event sources and provide your program with events.</p>
121 <p>To do this, it must take more or less complete control over your process
122 (or thread) by executing the <i>event loop</i> handler, and will then
123 communicate events via a callback mechanism.</p>
124 <p>You register interest in certain events by registering so-called <i>event
125 watchers</i>, which are relatively small C structures you initialise with the
126 details of the event, and then hand it over to libev by <i>starting</i> the
130 <h1 id="FEATURES">FEATURES</h1><p><a href="#TOP" class="toplink">Top</a></p>
131 <div id="FEATURES_CONTENT">
132 <p>Libev supports select, poll, the linux-specific epoll and the bsd-specific
133 kqueue mechanisms for file descriptor events, relative timers, absolute
134 timers with customised rescheduling, signal events, process status change
135 events (related to SIGCHLD), and event watchers dealing with the event
136 loop mechanism itself (idle, prepare and check watchers). It also is quite
137 fast (see this <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing
138 it to libevent for example).</p>
141 <h1 id="CONVENTIONS">CONVENTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
142 <div id="CONVENTIONS_CONTENT">
143 <p>Libev is very configurable. In this manual the default configuration
144 will be described, which supports multiple event loops. For more info
145 about various configuration options please have a look at the file
146 <cite>README.embed</cite> in the libev distribution. If libev was configured without
147 support for multiple event loops, then all functions taking an initial
148 argument of name <code>loop</code> (which is always of type <code>struct ev_loop *</code>)
149 will not have this argument.</p>
152 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
153 <div id="TIME_REPRESENTATION_CONTENT">
154 <p>Libev represents time as a single floating point number, representing the
155 (fractional) number of seconds since the (POSIX) epoch (somewhere near
156 the beginning of 1970, details are complicated, don't ask). This type is
157 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
158 to the <code>double</code> type in C, and when you need to do any calculations on
159 it, you should treat it as such.</p>
162 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
163 <div id="GLOBAL_FUNCTIONS_CONTENT">
164 <p>These functions can be called anytime, even before initialising the
165 library in any way.</p>
167 <dt>ev_tstamp ev_time ()</dt>
169 <p>Returns the current time as libev would use it. Please note that the
170 <code>ev_now</code> function is usually faster and also often returns the timestamp
171 you actually want to know.</p>
173 <dt>int ev_version_major ()</dt>
174 <dt>int ev_version_minor ()</dt>
176 <p>You can find out the major and minor version numbers of the library
177 you linked against by calling the functions <code>ev_version_major</code> and
178 <code>ev_version_minor</code>. If you want, you can compare against the global
179 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
180 version of the library your program was compiled against.</p>
181 <p>Usually, it's a good idea to terminate if the major versions mismatch,
182 as this indicates an incompatible change. Minor versions are usually
183 compatible to older versions, so a larger minor version alone is usually
185 <p>Example: make sure we haven't accidentally been linked against the wrong
187 <pre> assert (("libev version mismatch",
188 ev_version_major () == EV_VERSION_MAJOR
189 && ev_version_minor () >= EV_VERSION_MINOR));
193 <dt>unsigned int ev_supported_backends ()</dt>
195 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
196 value) compiled into this binary of libev (independent of their
197 availability on the system you are running on). See <code>ev_default_loop</code> for
198 a description of the set values.</p>
199 <p>Example: make sure we have the epoll method, because yeah this is cool and
200 a must have and can we have a torrent of it please!!!11</p>
201 <pre> assert (("sorry, no epoll, no sex",
202 ev_supported_backends () & EVBACKEND_EPOLL));
206 <dt>unsigned int ev_recommended_backends ()</dt>
208 <p>Return the set of all backends compiled into this binary of libev and also
209 recommended for this platform. This set is often smaller than the one
210 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
211 most BSDs and will not be autodetected unless you explicitly request it
212 (assuming you know what you are doing). This is the set of backends that
213 libev will probe for if you specify no backends explicitly.</p>
215 <dt>unsigned int ev_embeddable_backends ()</dt>
217 <p>Returns the set of backends that are embeddable in other event loops. This
218 is the theoretical, all-platform, value. To find which backends
219 might be supported on the current system, you would need to look at
220 <code>ev_embeddable_backends () & ev_supported_backends ()</code>, likewise for
221 recommended ones.</p>
222 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
224 <dt>ev_set_allocator (void *(*cb)(void *ptr, size_t size))</dt>
226 <p>Sets the allocation function to use (the prototype and semantics are
227 identical to the realloc C function). It is used to allocate and free
228 memory (no surprises here). If it returns zero when memory needs to be
229 allocated, the library might abort or take some potentially destructive
230 action. The default is your system realloc function.</p>
231 <p>You could override this function in high-availability programs to, say,
232 free some memory if it cannot allocate memory, to use a special allocator,
233 or even to sleep a while and retry until some memory is available.</p>
234 <p>Example: replace the libev allocator with one that waits a bit and then
235 retries: better than mine).</p>
237 persistent_realloc (void *ptr, size_t size)
241 void *newptr = realloc (ptr, size);
251 ev_set_allocator (persistent_realloc);
255 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
257 <p>Set the callback function to call on a retryable syscall error (such
258 as failed select, poll, epoll_wait). The message is a printable string
259 indicating the system call or subsystem causing the problem. If this
260 callback is set, then libev will expect it to remedy the sitution, no
261 matter what, when it returns. That is, libev will generally retry the
262 requested operation, or, if the condition doesn't go away, do bad stuff
264 <p>Example: do the same thing as libev does internally:</p>
266 fatal_error (const char *msg)
273 ev_set_syserr_cb (fatal_error);
280 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p>
281 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
282 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
283 types of such loops, the <i>default</i> loop, which supports signals and child
284 events, and dynamically created loops which do not.</p>
285 <p>If you use threads, a common model is to run the default event loop
286 in your main thread (or in a separate thread) and for each thread you
287 create, you also create another event loop. Libev itself does no locking
288 whatsoever, so if you mix calls to the same event loop in different
289 threads, make sure you lock (this is usually a bad idea, though, even if
290 done correctly, because it's hideous and inefficient).</p>
292 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
294 <p>This will initialise the default event loop if it hasn't been initialised
295 yet and return it. If the default loop could not be initialised, returns
296 false. If it already was initialised it simply returns it (and ignores the
297 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
298 <p>If you don't know what event loop to use, use the one returned from this
300 <p>The flags argument can be used to specify special behaviour or specific
301 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
302 <p>The following flags are supported:</p>
305 <dt><code>EVFLAG_AUTO</code></dt>
307 <p>The default flags value. Use this if you have no clue (it's the right
308 thing, believe me).</p>
310 <dt><code>EVFLAG_NOENV</code></dt>
312 <p>If this flag bit is ored into the flag value (or the program runs setuid
313 or setgid) then libev will <i>not</i> look at the environment variable
314 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
315 override the flags completely if it is found in the environment. This is
316 useful to try out specific backends to test their performance, or to work
319 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
321 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
322 libev tries to roll its own fd_set with no limits on the number of fds,
323 but if that fails, expect a fairly low limit on the number of fds when
324 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
325 the fastest backend for a low number of fds.</p>
327 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
329 <p>And this is your standard poll(2) backend. It's more complicated than
330 select, but handles sparse fds better and has no artificial limit on the
331 number of fds you can use (except it will slow down considerably with a
332 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
334 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
336 <p>For few fds, this backend is a bit little slower than poll and select,
337 but it scales phenomenally better. While poll and select usually scale like
338 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
339 either O(1) or O(active_fds).</p>
340 <p>While stopping and starting an I/O watcher in the same iteration will
341 result in some caching, there is still a syscall per such incident
342 (because the fd could point to a different file description now), so its
343 best to avoid that. Also, dup()ed file descriptors might not work very
344 well if you register events for both fds.</p>
345 <p>Please note that epoll sometimes generates spurious notifications, so you
346 need to use non-blocking I/O or other means to avoid blocking when no data
347 (or space) is available.</p>
349 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
351 <p>Kqueue deserves special mention, as at the time of this writing, it
352 was broken on all BSDs except NetBSD (usually it doesn't work with
353 anything but sockets and pipes, except on Darwin, where of course its
354 completely useless). For this reason its not being "autodetected"
355 unless you explicitly specify it explicitly in the flags (i.e. using
356 <code>EVBACKEND_KQUEUE</code>).</p>
357 <p>It scales in the same way as the epoll backend, but the interface to the
358 kernel is more efficient (which says nothing about its actual speed, of
359 course). While starting and stopping an I/O watcher does not cause an
360 extra syscall as with epoll, it still adds up to four event changes per
361 incident, so its best to avoid that.</p>
363 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
365 <p>This is not implemented yet (and might never be).</p>
367 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
369 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
370 it's really slow, but it still scales very well (O(active_fds)).</p>
371 <p>Please note that solaris ports can result in a lot of spurious
372 notifications, so you need to use non-blocking I/O or other means to avoid
373 blocking when no data (or space) is available.</p>
375 <dt><code>EVBACKEND_ALL</code></dt>
377 <p>Try all backends (even potentially broken ones that wouldn't be tried
378 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
379 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
383 <p>If one or more of these are ored into the flags value, then only these
384 backends will be tried (in the reverse order as given here). If none are
385 specified, most compiled-in backend will be tried, usually in reverse
386 order of their flag values :)</p>
387 <p>The most typical usage is like this:</p>
388 <pre> if (!ev_default_loop (0))
389 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
392 <p>Restrict libev to the select and poll backends, and do not allow
393 environment settings to be taken into account:</p>
394 <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
397 <p>Use whatever libev has to offer, but make sure that kqueue is used if
398 available (warning, breaks stuff, best use only with your own private
399 event loop and only if you know the OS supports your types of fds):</p>
400 <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
404 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
406 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
407 always distinct from the default loop. Unlike the default loop, it cannot
408 handle signal and child watchers, and attempts to do so will be greeted by
409 undefined behaviour (or a failed assertion if assertions are enabled).</p>
410 <p>Example: try to create a event loop that uses epoll and nothing else.</p>
411 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
413 fatal ("no epoll found here, maybe it hides under your chair");
417 <dt>ev_default_destroy ()</dt>
419 <p>Destroys the default loop again (frees all memory and kernel state
420 etc.). None of the active event watchers will be stopped in the normal
421 sense, so e.g. <code>ev_is_active</code> might still return true. It is your
422 responsibility to either stop all watchers cleanly yoursef <i>before</i>
423 calling this function, or cope with the fact afterwards (which is usually
424 the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
427 <dt>ev_loop_destroy (loop)</dt>
429 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
430 earlier call to <code>ev_loop_new</code>.</p>
432 <dt>ev_default_fork ()</dt>
434 <p>This function reinitialises the kernel state for backends that have
435 one. Despite the name, you can call it anytime, but it makes most sense
436 after forking, in either the parent or child process (or both, but that
437 again makes little sense).</p>
438 <p>You <i>must</i> call this function in the child process after forking if and
439 only if you want to use the event library in both processes. If you just
440 fork+exec, you don't have to call it.</p>
441 <p>The function itself is quite fast and it's usually not a problem to call
442 it just in case after a fork. To make this easy, the function will fit in
443 quite nicely into a call to <code>pthread_atfork</code>:</p>
444 <pre> pthread_atfork (0, 0, ev_default_fork);
447 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
448 without calling this function, so if you force one of those backends you
449 do not need to care.</p>
451 <dt>ev_loop_fork (loop)</dt>
453 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
454 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
455 after fork, and how you do this is entirely your own problem.</p>
457 <dt>unsigned int ev_backend (loop)</dt>
459 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
462 <dt>ev_tstamp ev_now (loop)</dt>
464 <p>Returns the current "event loop time", which is the time the event loop
465 received events and started processing them. This timestamp does not
466 change as long as callbacks are being processed, and this is also the base
467 time used for relative timers. You can treat it as the timestamp of the
468 event occuring (or more correctly, libev finding out about it).</p>
470 <dt>ev_loop (loop, int flags)</dt>
472 <p>Finally, this is it, the event handler. This function usually is called
473 after you initialised all your watchers and you want to start handling
475 <p>If the flags argument is specified as <code>0</code>, it will not return until
476 either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
477 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
478 relying on all watchers to be stopped when deciding when a program has
479 finished (especially in interactive programs), but having a program that
480 automatically loops as long as it has to and no longer by virtue of
481 relying on its watchers stopping correctly is a thing of beauty.</p>
482 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
483 those events and any outstanding ones, but will not block your process in
484 case there are no events and will return after one iteration of the loop.</p>
485 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
486 neccessary) and will handle those and any outstanding ones. It will block
487 your process until at least one new event arrives, and will return after
488 one iteration of the loop. This is useful if you are waiting for some
489 external event in conjunction with something not expressible using other
490 libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
491 usually a better approach for this kind of thing.</p>
492 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
493 <pre> * If there are no active watchers (reference count is zero), return.
494 - Queue prepare watchers and then call all outstanding watchers.
495 - If we have been forked, recreate the kernel state.
496 - Update the kernel state with all outstanding changes.
497 - Update the "event loop time".
498 - Calculate for how long to block.
499 - Block the process, waiting for any events.
500 - Queue all outstanding I/O (fd) events.
501 - Update the "event loop time" and do time jump handling.
502 - Queue all outstanding timers.
503 - Queue all outstanding periodics.
504 - If no events are pending now, queue all idle watchers.
505 - Queue all check watchers.
506 - Call all queued watchers in reverse order (i.e. check watchers first).
507 Signals and child watchers are implemented as I/O watchers, and will
508 be handled here by queueing them when their watcher gets executed.
509 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
510 were used, return, otherwise continue with step *.
513 <p>Example: queue some jobs and then loop until no events are outsanding
515 <pre> ... queue jobs here, make sure they register event watchers as long
516 ... as they still have work to do (even an idle watcher will do..)
517 ev_loop (my_loop, 0);
522 <dt>ev_unloop (loop, how)</dt>
524 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
525 has processed all outstanding events). The <code>how</code> argument must be either
526 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
527 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
529 <dt>ev_ref (loop)</dt>
530 <dt>ev_unref (loop)</dt>
532 <p>Ref/unref can be used to add or remove a reference count on the event
533 loop: Every watcher keeps one reference, and as long as the reference
534 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
535 a watcher you never unregister that should not keep <code>ev_loop</code> from
536 returning, ev_unref() after starting, and ev_ref() before stopping it. For
537 example, libev itself uses this for its internal signal pipe: It is not
538 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
539 no event watchers registered by it are active. It is also an excellent
540 way to do this for generic recurring timers or from within third-party
541 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
542 <p>Example: create a signal watcher, but keep it from keeping <code>ev_loop</code>
543 running when nothing else is active.</p>
544 <pre> struct dv_signal exitsig;
545 ev_signal_init (&exitsig, sig_cb, SIGINT);
546 ev_signal_start (myloop, &exitsig);
550 <p>Example: for some weird reason, unregister the above signal handler again.</p>
551 <pre> ev_ref (myloop);
552 ev_signal_stop (myloop, &exitsig);
563 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p>
564 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
565 <p>A watcher is a structure that you create and register to record your
566 interest in some event. For instance, if you want to wait for STDIN to
567 become readable, you would create an <code>ev_io</code> watcher for that:</p>
568 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
571 ev_unloop (loop, EVUNLOOP_ALL);
574 struct ev_loop *loop = ev_default_loop (0);
575 struct ev_io stdin_watcher;
576 ev_init (&stdin_watcher, my_cb);
577 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
578 ev_io_start (loop, &stdin_watcher);
582 <p>As you can see, you are responsible for allocating the memory for your
583 watcher structures (and it is usually a bad idea to do this on the stack,
584 although this can sometimes be quite valid).</p>
585 <p>Each watcher structure must be initialised by a call to <code>ev_init
586 (watcher *, callback)</code>, which expects a callback to be provided. This
587 callback gets invoked each time the event occurs (or, in the case of io
588 watchers, each time the event loop detects that the file descriptor given
589 is readable and/or writable).</p>
590 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
591 with arguments specific to this watcher type. There is also a macro
592 to combine initialisation and setting in one call: <code>ev_<type>_init
593 (watcher *, callback, ...)</code>.</p>
594 <p>To make the watcher actually watch out for events, you have to start it
595 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
596 *)</code>), and you can stop watching for events at any time by calling the
597 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
598 <p>As long as your watcher is active (has been started but not stopped) you
599 must not touch the values stored in it. Most specifically you must never
600 reinitialise it or call its <code>set</code> macro.</p>
601 <p>Each and every callback receives the event loop pointer as first, the
602 registered watcher structure as second, and a bitset of received events as
604 <p>The received events usually include a single bit per event type received
605 (you can receive multiple events at the same time). The possible bit masks
608 <dt><code>EV_READ</code></dt>
609 <dt><code>EV_WRITE</code></dt>
611 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
614 <dt><code>EV_TIMEOUT</code></dt>
616 <p>The <code>ev_timer</code> watcher has timed out.</p>
618 <dt><code>EV_PERIODIC</code></dt>
620 <p>The <code>ev_periodic</code> watcher has timed out.</p>
622 <dt><code>EV_SIGNAL</code></dt>
624 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
626 <dt><code>EV_CHILD</code></dt>
628 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
630 <dt><code>EV_STAT</code></dt>
632 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
634 <dt><code>EV_IDLE</code></dt>
636 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
638 <dt><code>EV_PREPARE</code></dt>
639 <dt><code>EV_CHECK</code></dt>
641 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
642 to gather new events, and all <code>ev_check</code> watchers are invoked just after
643 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
644 received events. Callbacks of both watcher types can start and stop as
645 many watchers as they want, and all of them will be taken into account
646 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
647 <code>ev_loop</code> from blocking).</p>
649 <dt><code>EV_EMBED</code></dt>
651 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
653 <dt><code>EV_FORK</code></dt>
655 <p>The event loop has been resumed in the child process after fork (see
656 <code>ev_fork</code>).</p>
658 <dt><code>EV_ERROR</code></dt>
660 <p>An unspecified error has occured, the watcher has been stopped. This might
661 happen because the watcher could not be properly started because libev
662 ran out of memory, a file descriptor was found to be closed or any other
663 problem. You best act on it by reporting the problem and somehow coping
664 with the watcher being stopped.</p>
665 <p>Libev will usually signal a few "dummy" events together with an error,
666 for example it might indicate that a fd is readable or writable, and if
667 your callbacks is well-written it can just attempt the operation and cope
668 with the error from read() or write(). This will not work in multithreaded
669 programs, though, so beware.</p>
674 <h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
675 <div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
676 <p>In the following description, <code>TYPE</code> stands for the watcher type,
677 e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
679 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
681 <p>This macro initialises the generic portion of a watcher. The contents
682 of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
683 the generic parts of the watcher are initialised, you <i>need</i> to call
684 the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
685 type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
686 which rolls both calls into one.</p>
687 <p>You can reinitialise a watcher at any time as long as it has been stopped
688 (or never started) and there are no pending events outstanding.</p>
689 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
690 int revents)</code>.</p>
692 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
694 <p>This macro initialises the type-specific parts of a watcher. You need to
695 call <code>ev_init</code> at least once before you call this macro, but you can
696 call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
697 macro on a watcher that is active (it can be pending, however, which is a
698 difference to the <code>ev_init</code> macro).</p>
699 <p>Although some watcher types do not have type-specific arguments
700 (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
702 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
704 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
705 calls into a single call. This is the most convinient method to initialise
706 a watcher. The same limitations apply, of course.</p>
708 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
710 <p>Starts (activates) the given watcher. Only active watchers will receive
711 events. If the watcher is already active nothing will happen.</p>
713 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
715 <p>Stops the given watcher again (if active) and clears the pending
716 status. It is possible that stopped watchers are pending (for example,
717 non-repeating timers are being stopped when they become pending), but
718 <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
719 you want to free or reuse the memory used by the watcher it is therefore a
720 good idea to always call its <code>ev_TYPE_stop</code> function.</p>
722 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
724 <p>Returns a true value iff the watcher is active (i.e. it has been started
725 and not yet been stopped). As long as a watcher is active you must not modify
728 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
730 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
731 events but its callback has not yet been invoked). As long as a watcher
732 is pending (but not active) you must not call an init function on it (but
733 <code>ev_TYPE_set</code> is safe) and you must make sure the watcher is available to
734 libev (e.g. you cnanot <code>free ()</code> it).</p>
736 <dt>callback = ev_cb (ev_TYPE *watcher)</dt>
738 <p>Returns the callback currently set on the watcher.</p>
740 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
742 <p>Change the callback. You can change the callback at virtually any time
743 (modulo threads).</p>
752 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
753 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
754 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
755 and read at any time, libev will completely ignore it. This can be used
756 to associate arbitrary data with your watcher. If you need more data and
757 don't want to allocate memory and store a pointer to it in that data
758 member, you can also "subclass" the watcher type and provide your own
765 struct whatever *mostinteresting;
769 <p>And since your callback will be called with a pointer to the watcher, you
770 can cast it back to your own type:</p>
771 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
773 struct my_io *w = (struct my_io *)w_;
778 <p>More interesting and less C-conformant ways of catsing your callback type
779 have been omitted....</p>
786 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p>
787 <div id="WATCHER_TYPES_CONTENT">
788 <p>This section describes each watcher in detail, but will not repeat
789 information given in the last section. Any initialisation/set macros,
790 functions and members specific to the watcher type are explained.</p>
791 <p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
792 while the watcher is active, you can look at the member and expect some
793 sensible content, but you must not modify it (you can modify it while the
794 watcher is stopped to your hearts content), or <i>[read-write]</i>, which
795 means you can expect it to have some sensible content while the watcher
796 is active, but you can also modify it. Modifying it may not do something
797 sensible or take immediate effect (or do anything at all), but libev will
798 not crash or malfunction in any way.</p>
805 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
806 <div id="code_ev_io_code_is_this_file_descrip-2">
807 <p>I/O watchers check whether a file descriptor is readable or writable
808 in each iteration of the event loop, or, more precisely, when reading
809 would not block the process and writing would at least be able to write
810 some data. This behaviour is called level-triggering because you keep
811 receiving events as long as the condition persists. Remember you can stop
812 the watcher if you don't want to act on the event and neither want to
813 receive future events.</p>
814 <p>In general you can register as many read and/or write event watchers per
815 fd as you want (as long as you don't confuse yourself). Setting all file
816 descriptors to non-blocking mode is also usually a good idea (but not
817 required if you know what you are doing).</p>
818 <p>You have to be careful with dup'ed file descriptors, though. Some backends
819 (the linux epoll backend is a notable example) cannot handle dup'ed file
820 descriptors correctly if you register interest in two or more fds pointing
821 to the same underlying file/socket/etc. description (that is, they share
822 the same underlying "file open").</p>
823 <p>If you must do this, then force the use of a known-to-be-good backend
824 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
825 <code>EVBACKEND_POLL</code>).</p>
826 <p>Another thing you have to watch out for is that it is quite easy to
827 receive "spurious" readyness notifications, that is your callback might
828 be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
829 because there is no data. Not only are some backends known to create a
830 lot of those (for example solaris ports), it is very easy to get into
831 this situation even with a relatively standard program structure. Thus
832 it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
833 <code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
834 <p>If you cannot run the fd in non-blocking mode (for example you should not
835 play around with an Xlib connection), then you have to seperately re-test
836 wether a file descriptor is really ready with a known-to-be good interface
837 such as poll (fortunately in our Xlib example, Xlib already does this on
838 its own, so its quite safe to use).</p>
840 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
841 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
843 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
844 rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
845 <code>EV_READ | EV_WRITE</code> to receive the given events.</p>
847 <dt>int fd [read-only]</dt>
849 <p>The file descriptor being watched.</p>
851 <dt>int events [read-only]</dt>
853 <p>The events being watched.</p>
856 <p>Example: call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
857 readable, but only once. Since it is likely line-buffered, you could
858 attempt to read a whole line in the callback:</p>
860 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
862 ev_io_stop (loop, w);
863 .. read from stdin here (or from w->fd) and haqndle any I/O errors
867 struct ev_loop *loop = ev_default_init (0);
868 struct ev_io stdin_readable;
869 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
870 ev_io_start (loop, &stdin_readable);
879 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
880 <div id="code_ev_timer_code_relative_and_opti-2">
881 <p>Timer watchers are simple relative timers that generate an event after a
882 given time, and optionally repeating in regular intervals after that.</p>
883 <p>The timers are based on real time, that is, if you register an event that
884 times out after an hour and you reset your system clock to last years
885 time, it will still time out after (roughly) and hour. "Roughly" because
886 detecting time jumps is hard, and some inaccuracies are unavoidable (the
887 monotonic clock option helps a lot here).</p>
888 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
889 time. This is usually the right thing as this timestamp refers to the time
890 of the event triggering whatever timeout you are modifying/starting. If
891 you suspect event processing to be delayed and you <i>need</i> to base the timeout
892 on the current time, use something like this to adjust for this:</p>
893 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
896 <p>The callback is guarenteed to be invoked only when its timeout has passed,
897 but if multiple timers become ready during the same loop iteration then
898 order of execution is undefined.</p>
900 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
901 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
903 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
904 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
905 timer will automatically be configured to trigger again <code>repeat</code> seconds
906 later, again, and again, until stopped manually.</p>
907 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
908 configure a timer to trigger every 10 seconds, then it will trigger at
909 exactly 10 second intervals. If, however, your program cannot keep up with
910 the timer (because it takes longer than those 10 seconds to do stuff) the
911 timer will not fire more than once per event loop iteration.</p>
913 <dt>ev_timer_again (loop)</dt>
915 <p>This will act as if the timer timed out and restart it again if it is
916 repeating. The exact semantics are:</p>
917 <p>If the timer is started but nonrepeating, stop it.</p>
918 <p>If the timer is repeating, either start it if necessary (with the repeat
919 value), or reset the running timer to the repeat value.</p>
920 <p>This sounds a bit complicated, but here is a useful and typical
921 example: Imagine you have a tcp connection and you want a so-called
922 idle timeout, that is, you want to be called when there have been,
923 say, 60 seconds of inactivity on the socket. The easiest way to do
924 this is to configure an <code>ev_timer</code> with <code>after</code>=<code>repeat</code>=<code>60</code> and calling
925 <code>ev_timer_again</code> each time you successfully read or write some data. If
926 you go into an idle state where you do not expect data to travel on the
927 socket, you can stop the timer, and again will automatically restart it if
929 <p>You can also ignore the <code>after</code> value and <code>ev_timer_start</code> altogether
930 and only ever use the <code>repeat</code> value:</p>
931 <pre> ev_timer_init (timer, callback, 0., 5.);
932 ev_timer_again (loop, timer);
934 timer->again = 17.;
935 ev_timer_again (loop, timer);
937 timer->again = 10.;
938 ev_timer_again (loop, timer);
941 <p>This is more efficient then stopping/starting the timer eahc time you want
942 to modify its timeout value.</p>
944 <dt>ev_tstamp repeat [read-write]</dt>
946 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
947 or <code>ev_timer_again</code> is called and determines the next timeout (if any),
948 which is also when any modifications are taken into account.</p>
951 <p>Example: create a timer that fires after 60 seconds.</p>
953 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
955 .. one minute over, w is actually stopped right here
958 struct ev_timer mytimer;
959 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
960 ev_timer_start (loop, &mytimer);
963 <p>Example: create a timeout timer that times out after 10 seconds of
966 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
968 .. ten seconds without any activity
971 struct ev_timer mytimer;
972 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
973 ev_timer_again (&mytimer); /* start timer */
976 // and in some piece of code that gets executed on any "activity":
977 // reset the timeout to start ticking again at 10 seconds
978 ev_timer_again (&mytimer);
986 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
987 <div id="code_ev_periodic_code_to_cron_or_not-2">
988 <p>Periodic watchers are also timers of a kind, but they are very versatile
989 (and unfortunately a bit complex).</p>
990 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
991 but on wallclock time (absolute time). You can tell a periodic watcher
992 to trigger "at" some specific point in time. For example, if you tell a
993 periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
994 + 10.</code>) and then reset your system clock to the last year, then it will
995 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
996 roughly 10 seconds later and of course not if you reset your system time
998 <p>They can also be used to implement vastly more complex timers, such as
999 triggering an event on eahc midnight, local time.</p>
1000 <p>As with timers, the callback is guarenteed to be invoked only when the
1001 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1002 during the same loop iteration then order of execution is undefined.</p>
1004 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
1005 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
1007 <p>Lots of arguments, lets sort it out... There are basically three modes of
1008 operation, and we will explain them from simplest to complex:</p>
1011 <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
1013 <p>In this configuration the watcher triggers an event at the wallclock time
1014 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
1015 that is, if it is to be run at January 1st 2011 then it will run when the
1016 system time reaches or surpasses this time.</p>
1018 <dt>* non-repeating interval timer (interval > 0, reschedule_cb = 0)</dt>
1020 <p>In this mode the watcher will always be scheduled to time out at the next
1021 <code>at + N * interval</code> time (for some integer N) and then repeat, regardless
1022 of any time jumps.</p>
1023 <p>This can be used to create timers that do not drift with respect to system
1025 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
1028 <p>This doesn't mean there will always be 3600 seconds in between triggers,
1029 but only that the the callback will be called when the system time shows a
1030 full hour (UTC), or more correctly, when the system time is evenly divisible
1032 <p>Another way to think about it (for the mathematically inclined) is that
1033 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
1034 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1036 <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
1038 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
1039 ignored. Instead, each time the periodic watcher gets scheduled, the
1040 reschedule callback will be called with the watcher as first, and the
1041 current time as second argument.</p>
1042 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
1043 ever, or make any event loop modifications</i>. If you need to stop it,
1044 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1045 starting a prepare watcher).</p>
1046 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1047 ev_tstamp now)</code>, e.g.:</p>
1048 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1054 <p>It must return the next time to trigger, based on the passed time value
1055 (that is, the lowest time value larger than to the second argument). It
1056 will usually be called just before the callback will be triggered, but
1057 might be called at other times, too.</p>
1058 <p>NOTE: <i>This callback must always return a time that is later than the
1059 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
1060 <p>This can be used to create very complex timers, such as a timer that
1061 triggers on each midnight, local time. To do this, you would calculate the
1062 next midnight after <code>now</code> and return the timestamp value for this. How
1063 you do this is, again, up to you (but it is not trivial, which is the main
1064 reason I omitted it as an example).</p>
1069 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
1071 <p>Simply stops and restarts the periodic watcher again. This is only useful
1072 when you changed some parameters or the reschedule callback would return
1073 a different time than the last time it was called (e.g. in a crond like
1074 program when the crontabs have changed).</p>
1076 <dt>ev_tstamp interval [read-write]</dt>
1078 <p>The current interval value. Can be modified any time, but changes only
1079 take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1082 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1084 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1085 switched off. Can be changed any time, but changes only take effect when
1086 the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1089 <p>Example: call a callback every hour, or, more precisely, whenever the
1090 system clock is divisible by 3600. The callback invocation times have
1091 potentially a lot of jittering, but good long-term stability.</p>
1093 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1095 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1098 struct ev_periodic hourly_tick;
1099 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1100 ev_periodic_start (loop, &hourly_tick);
1103 <p>Example: the same as above, but use a reschedule callback to do it:</p>
1104 <pre> #include <math.h>
1107 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1109 return fmod (now, 3600.) + 3600.;
1112 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1115 <p>Example: call a callback every hour, starting now:</p>
1116 <pre> struct ev_periodic hourly_tick;
1117 ev_periodic_init (&hourly_tick, clock_cb,
1118 fmod (ev_now (loop), 3600.), 3600., 0);
1119 ev_periodic_start (loop, &hourly_tick);
1127 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
1128 <div id="code_ev_signal_code_signal_me_when_a-2">
1129 <p>Signal watchers will trigger an event when the process receives a specific
1130 signal one or more times. Even though signals are very asynchronous, libev
1131 will try it's best to deliver signals synchronously, i.e. as part of the
1132 normal event processing, like any other event.</p>
1133 <p>You can configure as many watchers as you like per signal. Only when the
1134 first watcher gets started will libev actually register a signal watcher
1135 with the kernel (thus it coexists with your own signal handlers as long
1136 as you don't register any with libev). Similarly, when the last signal
1137 watcher for a signal is stopped libev will reset the signal handler to
1138 SIG_DFL (regardless of what it was set to before).</p>
1140 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1141 <dt>ev_signal_set (ev_signal *, int signum)</dt>
1143 <p>Configures the watcher to trigger on the given signal number (usually one
1144 of the <code>SIGxxx</code> constants).</p>
1146 <dt>int signum [read-only]</dt>
1148 <p>The signal the watcher watches out for.</p>
1157 <h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
1158 <div id="code_ev_child_code_watch_out_for_pro-2">
1159 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1160 some child status changes (most typically when a child of yours dies).</p>
1162 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1163 <dt>ev_child_set (ev_child *, int pid)</dt>
1165 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1166 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1167 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1168 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1169 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1170 process causing the status change.</p>
1172 <dt>int pid [read-only]</dt>
1174 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1176 <dt>int rpid [read-write]</dt>
1178 <p>The process id that detected a status change.</p>
1180 <dt>int rstatus [read-write]</dt>
1182 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1183 <code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
1186 <p>Example: try to exit cleanly on SIGINT and SIGTERM.</p>
1188 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1190 ev_unloop (loop, EVUNLOOP_ALL);
1193 struct ev_signal signal_watcher;
1194 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1195 ev_signal_start (loop, &sigint_cb);
1203 <h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1204 <div id="code_ev_stat_code_did_the_file_attri-2">
1205 <p>This watches a filesystem path for attribute changes. That is, it calls
1206 <code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1207 compared to the last time, invoking the callback if it did.</p>
1208 <p>The path does not need to exist: changing from "path exists" to "path does
1209 not exist" is a status change like any other. The condition "path does
1210 not exist" is signified by the <code>st_nlink</code> field being zero (which is
1211 otherwise always forced to be at least one) and all the other fields of
1212 the stat buffer having unspecified contents.</p>
1213 <p>Since there is no standard to do this, the portable implementation simply
1214 calls <code>stat (2)</code> regulalry on the path to see if it changed somehow. You
1215 can specify a recommended polling interval for this case. If you specify
1216 a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1217 unspecified default</i> value will be used (which you can expect to be around
1218 five seconds, although this might change dynamically). Libev will also
1219 impose a minimum interval which is currently around <code>0.1</code>, but thats
1220 usually overkill.</p>
1221 <p>This watcher type is not meant for massive numbers of stat watchers,
1222 as even with OS-supported change notifications, this can be
1223 resource-intensive.</p>
1224 <p>At the time of this writing, no specific OS backends are implemented, but
1225 if demand increases, at least a kqueue and inotify backend will be added.</p>
1227 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1228 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1230 <p>Configures the watcher to wait for status changes of the given
1231 <code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1232 be detected and should normally be specified as <code>0</code> to let libev choose
1233 a suitable value. The memory pointed to by <code>path</code> must point to the same
1234 path for as long as the watcher is active.</p>
1235 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1236 relative to the attributes at the time the watcher was started (or the
1237 last change was detected).</p>
1239 <dt>ev_stat_stat (ev_stat *)</dt>
1241 <p>Updates the stat buffer immediately with new values. If you change the
1242 watched path in your callback, you could call this fucntion to avoid
1243 detecting this change (while introducing a race condition). Can also be
1244 useful simply to find out the new values.</p>
1246 <dt>ev_statdata attr [read-only]</dt>
1248 <p>The most-recently detected attributes of the file. Although the type is of
1249 <code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1250 suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1251 was some error while <code>stat</code>ing the file.</p>
1253 <dt>ev_statdata prev [read-only]</dt>
1255 <p>The previous attributes of the file. The callback gets invoked whenever
1256 <code>prev</code> != <code>attr</code>.</p>
1258 <dt>ev_tstamp interval [read-only]</dt>
1260 <p>The specified interval.</p>
1262 <dt>const char *path [read-only]</dt>
1264 <p>The filesystem path that is being watched.</p>
1267 <p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1269 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1271 /* /etc/passwd changed in some way */
1272 if (w->attr.st_nlink)
1274 printf ("passwd current size %ld\n", (long)w->attr.st_size);
1275 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1276 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1279 /* you shalt not abuse printf for puts */
1280 puts ("wow, /etc/passwd is not there, expect problems. "
1281 "if this is windows, they already arrived\n");
1287 ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1288 ev_stat_start (loop, &passwd);
1296 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
1297 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1298 <p>Idle watchers trigger events when there are no other events are pending
1299 (prepare, check and other idle watchers do not count). That is, as long
1300 as your process is busy handling sockets or timeouts (or even signals,
1301 imagine) it will not be triggered. But when your process is idle all idle
1302 watchers are being called again and again, once per event loop iteration -
1303 until stopped, that is, or your process receives more events and becomes
1305 <p>The most noteworthy effect is that as long as any idle watchers are
1306 active, the process will not block when waiting for new events.</p>
1307 <p>Apart from keeping your process non-blocking (which is a useful
1308 effect on its own sometimes), idle watchers are a good place to do
1309 "pseudo-background processing", or delay processing stuff to after the
1310 event loop has handled all outstanding events.</p>
1312 <dt>ev_idle_init (ev_signal *, callback)</dt>
1314 <p>Initialises and configures the idle watcher - it has no parameters of any
1315 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1319 <p>Example: dynamically allocate an <code>ev_idle</code>, start it, and in the
1320 callback, free it. Alos, use no error checking, as usual.</p>
1322 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1325 // now do something you wanted to do when the program has
1326 // no longer asnything immediate to do.
1329 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1330 ev_idle_init (idle_watcher, idle_cb);
1331 ev_idle_start (loop, idle_cb);
1339 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
1340 <div id="code_ev_prepare_code_and_code_ev_che-2">
1341 <p>Prepare and check watchers are usually (but not always) used in tandem:
1342 prepare watchers get invoked before the process blocks and check watchers
1344 <p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1345 the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1346 watchers. Other loops than the current one are fine, however. The
1347 rationale behind this is that you do not need to check for recursion in
1348 those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1349 <code>ev_check</code> so if you have one watcher of each kind they will always be
1350 called in pairs bracketing the blocking call.</p>
1351 <p>Their main purpose is to integrate other event mechanisms into libev and
1352 their use is somewhat advanced. This could be used, for example, to track
1353 variable changes, implement your own watchers, integrate net-snmp or a
1354 coroutine library and lots more. They are also occasionally useful if
1355 you cache some data and want to flush it before blocking (for example,
1356 in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1358 <p>This is done by examining in each prepare call which file descriptors need
1359 to be watched by the other library, registering <code>ev_io</code> watchers for
1360 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1361 provide just this functionality). Then, in the check watcher you check for
1362 any events that occured (by checking the pending status of all watchers
1363 and stopping them) and call back into the library. The I/O and timer
1364 callbacks will never actually be called (but must be valid nevertheless,
1365 because you never know, you know?).</p>
1366 <p>As another example, the Perl Coro module uses these hooks to integrate
1367 coroutines into libev programs, by yielding to other active coroutines
1368 during each prepare and only letting the process block if no coroutines
1369 are ready to run (it's actually more complicated: it only runs coroutines
1370 with priority higher than or equal to the event loop and one coroutine
1371 of lower priority, but only once, using idle watchers to keep the event
1372 loop from blocking if lower-priority coroutines are active, thus mapping
1373 low-priority coroutines to idle/background tasks).</p>
1375 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1376 <dt>ev_check_init (ev_check *, callback)</dt>
1378 <p>Initialises and configures the prepare or check watcher - they have no
1379 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1380 macros, but using them is utterly, utterly and completely pointless.</p>
1383 <p>Example: To include a library such as adns, you would add IO watchers
1384 and a timeout watcher in a prepare handler, as required by libadns, and
1385 in a check watcher, destroy them and call into libadns. What follows is
1386 pseudo-code only of course:</p>
1387 <pre> static ev_io iow [nfd];
1391 io_cb (ev_loop *loop, ev_io *w, int revents)
1393 // set the relevant poll flags
1394 // could also call adns_processreadable etc. here
1395 struct pollfd *fd = (struct pollfd *)w->data;
1396 if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1397 if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1400 // create io watchers for each fd and a timer before blocking
1402 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1404 int timeout = 3600000;truct pollfd fds [nfd];
1405 // actual code will need to loop here and realloc etc.
1406 adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1408 /* the callback is illegal, but won't be called as we stop during check */
1409 ev_timer_init (&tw, 0, timeout * 1e-3);
1410 ev_timer_start (loop, &tw);
1412 // create on ev_io per pollfd
1413 for (int i = 0; i < nfd; ++i)
1415 ev_io_init (iow + i, io_cb, fds [i].fd,
1416 ((fds [i].events & POLLIN ? EV_READ : 0)
1417 | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1419 fds [i].revents = 0;
1420 iow [i].data = fds + i;
1421 ev_io_start (loop, iow + i);
1425 // stop all watchers after blocking
1427 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1429 ev_timer_stop (loop, &tw);
1431 for (int i = 0; i < nfd; ++i)
1432 ev_io_stop (loop, iow + i);
1434 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1443 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1444 <div id="code_ev_embed_code_when_one_backend_-2">
1445 <p>This is a rather advanced watcher type that lets you embed one event loop
1446 into another (currently only <code>ev_io</code> events are supported in the embedded
1447 loop, other types of watchers might be handled in a delayed or incorrect
1448 fashion and must not be used).</p>
1449 <p>There are primarily two reasons you would want that: work around bugs and
1451 <p>As an example for a bug workaround, the kqueue backend might only support
1452 sockets on some platform, so it is unusable as generic backend, but you
1453 still want to make use of it because you have many sockets and it scales
1454 so nicely. In this case, you would create a kqueue-based loop and embed it
1455 into your default loop (which might use e.g. poll). Overall operation will
1456 be a bit slower because first libev has to poll and then call kevent, but
1457 at least you can use both at what they are best.</p>
1458 <p>As for prioritising I/O: rarely you have the case where some fds have
1459 to be watched and handled very quickly (with low latency), and even
1460 priorities and idle watchers might have too much overhead. In this case
1461 you would put all the high priority stuff in one loop and all the rest in
1462 a second one, and embed the second one in the first.</p>
1463 <p>As long as the watcher is active, the callback will be invoked every time
1464 there might be events pending in the embedded loop. The callback must then
1465 call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1466 their callbacks (you could also start an idle watcher to give the embedded
1467 loop strictly lower priority for example). You can also set the callback
1468 to <code>0</code>, in which case the embed watcher will automatically execute the
1469 embedded loop sweep.</p>
1470 <p>As long as the watcher is started it will automatically handle events. The
1471 callback will be invoked whenever some events have been handled. You can
1472 set the callback to <code>0</code> to avoid having to specify one if you are not
1473 interested in that.</p>
1474 <p>Also, there have not currently been made special provisions for forking:
1475 when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1476 but you will also have to stop and restart any <code>ev_embed</code> watchers
1478 <p>Unfortunately, not all backends are embeddable, only the ones returned by
1479 <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1481 <p>So when you want to use this feature you will always have to be prepared
1482 that you cannot get an embeddable loop. The recommended way to get around
1483 this is to have a separate variables for your embeddable loop, try to
1484 create it, and if that fails, use the normal loop for everything:</p>
1485 <pre> struct ev_loop *loop_hi = ev_default_init (0);
1486 struct ev_loop *loop_lo = 0;
1487 struct ev_embed embed;
1489 // see if there is a chance of getting one that works
1490 // (remember that a flags value of 0 means autodetection)
1491 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1492 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1495 // if we got one, then embed it, otherwise default to loop_hi
1498 ev_embed_init (&embed, 0, loop_lo);
1499 ev_embed_start (loop_hi, &embed);
1506 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1507 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1509 <p>Configures the watcher to embed the given loop, which must be
1510 embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1511 invoked automatically, otherwise it is the responsibility of the callback
1512 to invoke it (it will continue to be called until the sweep has been done,
1513 if you do not want thta, you need to temporarily stop the embed watcher).</p>
1515 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1517 <p>Make a single, non-blocking sweep over the embedded loop. This works
1518 similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1519 apropriate way for embedded loops.</p>
1521 <dt>struct ev_loop *loop [read-only]</dt>
1523 <p>The embedded event loop.</p>
1532 <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>
1533 <div id="code_ev_fork_code_the_audacity_to_re-2">
1534 <p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1535 whoever is a good citizen cared to tell libev about it by calling
1536 <code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1537 event loop blocks next and before <code>ev_check</code> watchers are being called,
1538 and only in the child after the fork. If whoever good citizen calling
1539 <code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1540 handlers will be invoked, too, of course.</p>
1542 <dt>ev_fork_init (ev_signal *, callback)</dt>
1544 <p>Initialises and configures the fork watcher - it has no parameters of any
1545 kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1555 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
1556 <div id="OTHER_FUNCTIONS_CONTENT">
1557 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1559 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1561 <p>This function combines a simple timer and an I/O watcher, calls your
1562 callback on whichever event happens first and automatically stop both
1563 watchers. This is useful if you want to wait for a single event on an fd
1564 or timeout without having to allocate/configure/start/stop/free one or
1565 more watchers yourself.</p>
1566 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1567 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1568 <code>events</code> set will be craeted and started.</p>
1569 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1570 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1571 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1573 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1574 passed an <code>revents</code> set like normal event callbacks (a combination of
1575 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1576 value passed to <code>ev_once</code>:</p>
1577 <pre> static void stdin_ready (int revents, void *arg)
1579 if (revents & EV_TIMEOUT)
1580 /* doh, nothing entered */;
1581 else if (revents & EV_READ)
1582 /* stdin might have data for us, joy! */;
1585 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1589 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1591 <p>Feeds the given event set into the event loop, as if the specified event
1592 had happened for the specified watcher (which must be a pointer to an
1593 initialised but not necessarily started event watcher).</p>
1595 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1597 <p>Feed an event on the given fd, as if a file descriptor backend detected
1598 the given events it.</p>
1600 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1602 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1612 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
1613 <div id="LIBEVENT_EMULATION_CONTENT">
1614 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1615 emulate the internals of libevent, so here are some usage hints:</p>
1617 <dt>* Use it by including <event.h>, as usual.</dt>
1618 <dt>* The following members are fully supported: ev_base, ev_callback,
1619 ev_arg, ev_fd, ev_res, ev_events.</dt>
1620 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1621 maintained by libev, it does not work exactly the same way as in libevent (consider
1622 it a private API).</dt>
1623 <dt>* Priorities are not currently supported. Initialising priorities
1624 will fail and all watchers will have the same priority, even though there
1625 is an ev_pri field.</dt>
1626 <dt>* Other members are not supported.</dt>
1627 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1628 to use the libev header file and library.</dt>
1632 <h1 id="C_SUPPORT">C++ SUPPORT</h1><p><a href="#TOP" class="toplink">Top</a></p>
1633 <div id="C_SUPPORT_CONTENT">
1634 <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1635 you to use some convinience methods to start/stop watchers and also change
1636 the callback model to a model using method callbacks on objects.</p>
1638 <pre> #include <ev++.h>
1641 <p>(it is not installed by default). This automatically includes <cite>ev.h</cite>
1642 and puts all of its definitions (many of them macros) into the global
1643 namespace. All C++ specific things are put into the <code>ev</code> namespace.</p>
1644 <p>It should support all the same embedding options as <cite>ev.h</cite>, most notably
1645 <code>EV_MULTIPLICITY</code>.</p>
1646 <p>Here is a list of things available in the <code>ev</code> namespace:</p>
1648 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1650 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1651 macros from <cite>ev.h</cite>.</p>
1653 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1655 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1657 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1659 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1660 the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1661 which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1662 defines by many implementations.</p>
1663 <p>All of those classes have these methods:</p>
1666 <dt>ev::TYPE::TYPE (object *, object::method *)</dt>
1667 <dt>ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)</dt>
1668 <dt>ev::TYPE::~TYPE</dt>
1670 <p>The constructor takes a pointer to an object and a method pointer to
1671 the event handler callback to call in this class. The constructor calls
1672 <code>ev_init</code> for you, which means you have to call the <code>set</code> method
1673 before starting it. If you do not specify a loop then the constructor
1674 automatically associates the default loop with this watcher.</p>
1675 <p>The destructor automatically stops the watcher if it is active.</p>
1677 <dt>w->set (struct ev_loop *)</dt>
1679 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1680 do this when the watcher is inactive (and not pending either).</p>
1682 <dt>w->set ([args])</dt>
1684 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1685 called at least once. Unlike the C counterpart, an active watcher gets
1686 automatically stopped and restarted.</p>
1688 <dt>w->start ()</dt>
1690 <p>Starts the watcher. Note that there is no <code>loop</code> argument as the
1691 constructor already takes the loop.</p>
1693 <dt>w->stop ()</dt>
1695 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1697 <dt>w->again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1699 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1700 <code>ev_TYPE_again</code> function.</p>
1702 <dt>w->sweep () <code>ev::embed</code> only</dt>
1704 <p>Invokes <code>ev_embed_sweep</code>.</p>
1706 <dt>w->update () <code>ev::stat</code> only</dt>
1708 <p>Invokes <code>ev_stat_stat</code>.</p>
1714 <p>Example: Define a class with an IO and idle watcher, start one of them in
1715 the constructor.</p>
1718 ev_io io; void io_cb (ev::io &w, int revents);
1719 ev_idle idle void idle_cb (ev::idle &w, int revents);
1724 myclass::myclass (int fd)
1725 : io (this, &myclass::io_cb),
1726 idle (this, &myclass::idle_cb)
1728 io.start (fd, ev::READ);
1737 <h1 id="MACRO_MAGIC">MACRO MAGIC</h1><p><a href="#TOP" class="toplink">Top</a></p>
1738 <div id="MACRO_MAGIC_CONTENT">
1739 <p>Libev can be compiled with a variety of options, the most fundemantal is
1740 <code>EV_MULTIPLICITY</code>. This option determines wether (most) functions and
1741 callbacks have an initial <code>struct ev_loop *</code> argument.</p>
1742 <p>To make it easier to write programs that cope with either variant, the
1743 following macros are defined:</p>
1745 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
1747 <p>This provides the loop <i>argument</i> for functions, if one is required ("ev
1748 loop argument"). The <code>EV_A</code> form is used when this is the sole argument,
1749 <code>EV_A_</code> is used when other arguments are following. Example:</p>
1750 <pre> ev_unref (EV_A);
1751 ev_timer_add (EV_A_ watcher);
1755 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
1756 which is often provided by the following macro.</p>
1758 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
1760 <p>This provides the loop <i>parameter</i> for functions, if one is required ("ev
1761 loop parameter"). The <code>EV_P</code> form is used when this is the sole parameter,
1762 <code>EV_P_</code> is used when other parameters are following. Example:</p>
1763 <pre> // this is how ev_unref is being declared
1764 static void ev_unref (EV_P);
1766 // this is how you can declare your typical callback
1767 static void cb (EV_P_ ev_timer *w, int revents)
1770 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
1771 suitable for use with <code>EV_A</code>.</p>
1773 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
1775 <p>Similar to the other two macros, this gives you the value of the default
1776 loop, if multiple loops are supported ("ev loop default").</p>
1779 <p>Example: Declare and initialise a check watcher, working regardless of
1780 wether multiple loops are supported or not.</p>
1782 check_cb (EV_P_ ev_timer *w, int revents)
1784 ev_check_stop (EV_A_ w);
1788 ev_check_init (&check, check_cb);
1789 ev_check_start (EV_DEFAULT_ &check);
1790 ev_loop (EV_DEFAULT_ 0);
1798 <h1 id="EMBEDDING">EMBEDDING</h1><p><a href="#TOP" class="toplink">Top</a></p>
1799 <div id="EMBEDDING_CONTENT">
1800 <p>Libev can (and often is) directly embedded into host
1801 applications. Examples of applications that embed it include the Deliantra
1802 Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1803 and rxvt-unicode.</p>
1804 <p>The goal is to enable you to just copy the neecssary files into your
1805 source directory without having to change even a single line in them, so
1806 you can easily upgrade by simply copying (or having a checked-out copy of
1807 libev somewhere in your source tree).</p>
1810 <h2 id="FILESETS">FILESETS</h2>
1811 <div id="FILESETS_CONTENT">
1812 <p>Depending on what features you need you need to include one or more sets of files
1816 <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
1817 <div id="CORE_EVENT_LOOP_CONTENT">
1818 <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
1819 configuration (no autoconf):</p>
1820 <pre> #define EV_STANDALONE 1
1821 #include "ev.c"
1824 <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
1825 single C source file only to provide the function implementations. To use
1826 it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
1827 done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
1828 where you can put other configuration options):</p>
1829 <pre> #define EV_STANDALONE 1
1830 #include "ev.h"
1833 <p>Both header files and implementation files can be compiled with a C++
1834 compiler (at least, thats a stated goal, and breakage will be treated
1836 <p>You need the following files in your source tree, or in a directory
1837 in your include path (e.g. in libev/ when using -Ilibev):</p>
1843 ev_win32.c required on win32 platforms only
1845 ev_select.c only when select backend is enabled (which is by default)
1846 ev_poll.c only when poll backend is enabled (disabled by default)
1847 ev_epoll.c only when the epoll backend is enabled (disabled by default)
1848 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
1849 ev_port.c only when the solaris port backend is enabled (disabled by default)
1852 <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
1853 to compile this single file.</p>
1856 <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
1857 <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
1858 <p>To include the libevent compatibility API, also include:</p>
1859 <pre> #include "event.c"
1862 <p>in the file including <cite>ev.c</cite>, and:</p>
1863 <pre> #include "event.h"
1866 <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
1867 <p>You need the following additional files for this:</p>
1874 <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
1875 <div id="AUTOCONF_SUPPORT_CONTENT">
1876 <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
1877 whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
1878 <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
1879 include <cite>config.h</cite> and configure itself accordingly.</p>
1880 <p>For this of course you need the m4 file:</p>
1886 <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
1887 <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
1888 <p>Libev can be configured via a variety of preprocessor symbols you have to define
1889 before including any of its files. The default is not to build for multiplicity
1890 and only include the select backend.</p>
1892 <dt>EV_STANDALONE</dt>
1894 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
1895 keeps libev from including <cite>config.h</cite>, and it also defines dummy
1896 implementations for some libevent functions (such as logging, which is not
1897 supported). It will also not define any of the structs usually found in
1898 <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
1900 <dt>EV_USE_MONOTONIC</dt>
1902 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
1903 monotonic clock option at both compiletime and runtime. Otherwise no use
1904 of the monotonic clock option will be attempted. If you enable this, you
1905 usually have to link against librt or something similar. Enabling it when
1906 the functionality isn't available is safe, though, althoguh you have
1907 to make sure you link against any libraries where the <code>clock_gettime</code>
1908 function is hiding in (often <cite>-lrt</cite>).</p>
1910 <dt>EV_USE_REALTIME</dt>
1912 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
1913 realtime clock option at compiletime (and assume its availability at
1914 runtime if successful). Otherwise no use of the realtime clock option will
1915 be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
1916 (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
1917 in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
1919 <dt>EV_USE_SELECT</dt>
1921 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
1922 <code>select</code>(2) backend. No attempt at autodetection will be done: if no
1923 other method takes over, select will be it. Otherwise the select backend
1924 will not be compiled in.</p>
1926 <dt>EV_SELECT_USE_FD_SET</dt>
1928 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
1929 structure. This is useful if libev doesn't compile due to a missing
1930 <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
1931 exotic systems. This usually limits the range of file descriptors to some
1932 low limit such as 1024 or might have other limitations (winsocket only
1933 allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
1934 influence the size of the <code>fd_set</code> used.</p>
1936 <dt>EV_SELECT_IS_WINSOCKET</dt>
1938 <p>When defined to <code>1</code>, the select backend will assume that
1939 select/socket/connect etc. don't understand file descriptors but
1940 wants osf handles on win32 (this is the case when the select to
1941 be used is the winsock select). This means that it will call
1942 <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
1943 it is assumed that all these functions actually work on fds, even
1944 on win32. Should not be defined on non-win32 platforms.</p>
1946 <dt>EV_USE_POLL</dt>
1948 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
1949 backend. Otherwise it will be enabled on non-win32 platforms. It
1950 takes precedence over select.</p>
1952 <dt>EV_USE_EPOLL</dt>
1954 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
1955 <code>epoll</code>(7) backend. Its availability will be detected at runtime,
1956 otherwise another method will be used as fallback. This is the
1957 preferred backend for GNU/Linux systems.</p>
1959 <dt>EV_USE_KQUEUE</dt>
1961 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
1962 <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
1963 otherwise another method will be used as fallback. This is the preferred
1964 backend for BSD and BSD-like systems, although on most BSDs kqueue only
1965 supports some types of fds correctly (the only platform we found that
1966 supports ptys for example was NetBSD), so kqueue might be compiled in, but
1967 not be used unless explicitly requested. The best way to use it is to find
1968 out whether kqueue supports your type of fd properly and use an embedded
1971 <dt>EV_USE_PORT</dt>
1973 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
1974 10 port style backend. Its availability will be detected at runtime,
1975 otherwise another method will be used as fallback. This is the preferred
1976 backend for Solaris 10 systems.</p>
1978 <dt>EV_USE_DEVPOLL</dt>
1980 <p>reserved for future expansion, works like the USE symbols above.</p>
1984 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
1985 undefined is <code><ev.h></code> in <cite>event.h</cite> and <code>"ev.h"</code> in <cite>ev.c</cite>. This
1986 can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
1988 <dt>EV_CONFIG_H</dt>
1990 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
1991 <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
1992 <code>EV_H</code>, above.</p>
1996 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
1997 of how the <cite>event.h</cite> header can be found.</p>
1999 <dt>EV_PROTOTYPES</dt>
2001 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2002 prototypes, but still define all the structs and other symbols. This is
2003 occasionally useful if you want to provide your own wrapper functions
2004 around libev functions.</p>
2006 <dt>EV_MULTIPLICITY</dt>
2008 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2009 will have the <code>struct ev_loop *</code> as first argument, and you can create
2010 additional independent event loops. Otherwise there will be no support
2011 for multiple event loops and there is no first event loop pointer
2012 argument. Instead, all functions act on the single default loop.</p>
2014 <dt>EV_PERIODIC_ENABLE</dt>
2016 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2017 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2020 <dt>EV_EMBED_ENABLE</dt>
2022 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2023 defined to be <code>0</code>, then they are not.</p>
2025 <dt>EV_STAT_ENABLE</dt>
2027 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2028 defined to be <code>0</code>, then they are not.</p>
2030 <dt>EV_FORK_ENABLE</dt>
2032 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2033 defined to be <code>0</code>, then they are not.</p>
2037 <p>If you need to shave off some kilobytes of code at the expense of some
2038 speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2039 some inlining decisions, saves roughly 30% codesize of amd64.</p>
2041 <dt>EV_PID_HASHSIZE</dt>
2043 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2044 pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2045 than enough. If you need to manage thousands of children you might want to
2046 increase this value.</p>
2050 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2051 this macro to a something else you can include more and other types of
2052 members. You have to define it each time you include one of the files,
2053 though, and it must be identical each time.</p>
2054 <p>For example, the perl EV module uses something like this:</p>
2055 <pre> #define EV_COMMON \
2056 SV *self; /* contains this struct */ \
2057 SV *cb_sv, *fh /* note no trailing ";" */
2061 <dt>EV_CB_DECLARE (type)</dt>
2062 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2063 <dt>ev_set_cb (ev, cb)</dt>
2065 <p>Can be used to change the callback member declaration in each watcher,
2066 and the way callbacks are invoked and set. Must expand to a struct member
2067 definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2068 their default definitions. One possible use for overriding these is to
2069 avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2070 method calls instead of plain function calls in C++.</p>
2073 <h2 id="EXAMPLES">EXAMPLES</h2>
2074 <div id="EXAMPLES_CONTENT">
2075 <p>For a real-world example of a program the includes libev
2076 verbatim, you can have a look at the EV perl module
2077 (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2078 the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2079 interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2080 will be compiled. It is pretty complex because it provides its own header
2082 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2083 that everybody includes and which overrides some autoconf choices:</p>
2084 <pre> #define EV_USE_POLL 0
2085 #define EV_MULTIPLICITY 0
2086 #define EV_PERIODICS 0
2087 #define EV_CONFIG_H <config.h>
2089 #include "ev++.h"
2092 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2093 <pre> #include "ev_cpp.h"
2094 #include "ev.c"
2102 <h1 id="COMPLEXITIES">COMPLEXITIES</h1><p><a href="#TOP" class="toplink">Top</a></p>
2103 <div id="COMPLEXITIES_CONTENT">
2104 <p>In this section the complexities of (many of) the algorithms used inside
2105 libev will be explained. For complexity discussions about backends see the
2106 documentation for <code>ev_default_init</code>.</p>
2109 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2110 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2111 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2112 <dt>Stopping check/prepare/idle watchers: O(1)</dt>
2113 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % 16))</dt>
2114 <dt>Finding the next timer per loop iteration: O(1)</dt>
2115 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2116 <dt>Activating one watcher: O(1)</dt>
2125 <h1 id="AUTHOR">AUTHOR</h1><p><a href="#TOP" class="toplink">Top</a></p>
2126 <div id="AUTHOR_CONTENT">
2127 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob