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="Fri Dec 7 20:23:46 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="#EXAMPLE_PROGRAM">EXAMPLE PROGRAM</a></li>
20 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
21 <li><a href="#FEATURES">FEATURES</a></li>
22 <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
23 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
24 <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
25 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
26 <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
27 <ul><li><a href="#GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</a></li>
28 <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
31 <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
32 <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</a></li>
33 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
34 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
35 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
36 <li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
37 <li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
38 <li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</a></li>
39 <li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</a></li>
40 <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
41 <li><a href="#code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</a></li>
44 <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
45 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
46 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
47 <li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
48 <li><a href="#EMBEDDING">EMBEDDING</a>
49 <ul><li><a href="#FILESETS">FILESETS</a>
50 <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
51 <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
52 <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
55 <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
56 <li><a href="#EXAMPLES">EXAMPLES</a></li>
59 <li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
60 <li><a href="#AUTHOR">AUTHOR</a>
65 <h1 id="NAME">NAME</h1>
66 <div id="NAME_CONTENT">
67 <p>libev - a high performance full-featured event loop written in C</p>
70 <h1 id="SYNOPSIS">SYNOPSIS</h1>
71 <div id="SYNOPSIS_CONTENT">
72 <pre> #include <ev.h>
77 <h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
78 <div id="EXAMPLE_PROGRAM_CONTENT">
79 <pre> #include <ev.h>
82 ev_timer timeout_watcher;
84 /* called when data readable on stdin */
86 stdin_cb (EV_P_ struct ev_io *w, int revents)
88 /* puts ("stdin ready"); */
89 ev_io_stop (EV_A_ w); /* just a syntax example */
90 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
94 timeout_cb (EV_P_ struct ev_timer *w, int revents)
96 /* puts ("timeout"); */
97 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
103 struct ev_loop *loop = ev_default_loop (0);
105 /* initialise an io watcher, then start it */
106 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
107 ev_io_start (loop, &stdin_watcher);
109 /* simple non-repeating 5.5 second timeout */
110 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
111 ev_timer_start (loop, &timeout_watcher);
113 /* loop till timeout or data ready */
122 <h1 id="DESCRIPTION">DESCRIPTION</h1>
123 <div id="DESCRIPTION_CONTENT">
124 <p>The newest version of this document is also available as a html-formatted
125 web page you might find easier to navigate when reading it for the first
126 time: <a href="http://cvs.schmorp.de/libev/ev.html">http://cvs.schmorp.de/libev/ev.html</a>.</p>
127 <p>Libev is an event loop: you register interest in certain events (such as a
128 file descriptor being readable or a timeout occuring), and it will manage
129 these event sources and provide your program with events.</p>
130 <p>To do this, it must take more or less complete control over your process
131 (or thread) by executing the <i>event loop</i> handler, and will then
132 communicate events via a callback mechanism.</p>
133 <p>You register interest in certain events by registering so-called <i>event
134 watchers</i>, which are relatively small C structures you initialise with the
135 details of the event, and then hand it over to libev by <i>starting</i> the
139 <h1 id="FEATURES">FEATURES</h1>
140 <div id="FEATURES_CONTENT">
141 <p>Libev supports <code>select</code>, <code>poll</code>, the Linux-specific <code>epoll</code>, the
142 BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
143 for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
144 (for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
145 with customised rescheduling (<code>ev_periodic</code>), synchronous signals
146 (<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
147 watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
148 <code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
149 file watchers (<code>ev_stat</code>) and even limited support for fork events
150 (<code>ev_fork</code>).</p>
151 <p>It also is quite fast (see this
152 <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
156 <h1 id="CONVENTIONS">CONVENTIONS</h1>
157 <div id="CONVENTIONS_CONTENT">
158 <p>Libev is very configurable. In this manual the default configuration will
159 be described, which supports multiple event loops. For more info about
160 various configuration options please have a look at <strong>EMBED</strong> section in
161 this manual. If libev was configured without support for multiple event
162 loops, then all functions taking an initial argument of name <code>loop</code>
163 (which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
166 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
167 <div id="TIME_REPRESENTATION_CONTENT">
168 <p>Libev represents time as a single floating point number, representing the
169 (fractional) number of seconds since the (POSIX) epoch (somewhere near
170 the beginning of 1970, details are complicated, don't ask). This type is
171 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
172 to the <code>double</code> type in C, and when you need to do any calculations on
173 it, you should treat it as such.</p>
176 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
177 <div id="GLOBAL_FUNCTIONS_CONTENT">
178 <p>These functions can be called anytime, even before initialising the
179 library in any way.</p>
181 <dt>ev_tstamp ev_time ()</dt>
183 <p>Returns the current time as libev would use it. Please note that the
184 <code>ev_now</code> function is usually faster and also often returns the timestamp
185 you actually want to know.</p>
187 <dt>int ev_version_major ()</dt>
188 <dt>int ev_version_minor ()</dt>
190 <p>You can find out the major and minor version numbers of the library
191 you linked against by calling the functions <code>ev_version_major</code> and
192 <code>ev_version_minor</code>. If you want, you can compare against the global
193 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
194 version of the library your program was compiled against.</p>
195 <p>Usually, it's a good idea to terminate if the major versions mismatch,
196 as this indicates an incompatible change. Minor versions are usually
197 compatible to older versions, so a larger minor version alone is usually
199 <p>Example: Make sure we haven't accidentally been linked against the wrong
201 <pre> assert (("libev version mismatch",
202 ev_version_major () == EV_VERSION_MAJOR
203 && ev_version_minor () >= EV_VERSION_MINOR));
207 <dt>unsigned int ev_supported_backends ()</dt>
209 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
210 value) compiled into this binary of libev (independent of their
211 availability on the system you are running on). See <code>ev_default_loop</code> for
212 a description of the set values.</p>
213 <p>Example: make sure we have the epoll method, because yeah this is cool and
214 a must have and can we have a torrent of it please!!!11</p>
215 <pre> assert (("sorry, no epoll, no sex",
216 ev_supported_backends () & EVBACKEND_EPOLL));
220 <dt>unsigned int ev_recommended_backends ()</dt>
222 <p>Return the set of all backends compiled into this binary of libev and also
223 recommended for this platform. This set is often smaller than the one
224 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
225 most BSDs and will not be autodetected unless you explicitly request it
226 (assuming you know what you are doing). This is the set of backends that
227 libev will probe for if you specify no backends explicitly.</p>
229 <dt>unsigned int ev_embeddable_backends ()</dt>
231 <p>Returns the set of backends that are embeddable in other event loops. This
232 is the theoretical, all-platform, value. To find which backends
233 might be supported on the current system, you would need to look at
234 <code>ev_embeddable_backends () & ev_supported_backends ()</code>, likewise for
235 recommended ones.</p>
236 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
238 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
240 <p>Sets the allocation function to use (the prototype is similar - the
241 semantics is identical - to the realloc C function). It is used to
242 allocate and free memory (no surprises here). If it returns zero when
243 memory needs to be allocated, the library might abort or take some
244 potentially destructive action. The default is your system realloc
246 <p>You could override this function in high-availability programs to, say,
247 free some memory if it cannot allocate memory, to use a special allocator,
248 or even to sleep a while and retry until some memory is available.</p>
249 <p>Example: Replace the libev allocator with one that waits a bit and then
252 persistent_realloc (void *ptr, size_t size)
256 void *newptr = realloc (ptr, size);
266 ev_set_allocator (persistent_realloc);
270 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
272 <p>Set the callback function to call on a retryable syscall error (such
273 as failed select, poll, epoll_wait). The message is a printable string
274 indicating the system call or subsystem causing the problem. If this
275 callback is set, then libev will expect it to remedy the sitution, no
276 matter what, when it returns. That is, libev will generally retry the
277 requested operation, or, if the condition doesn't go away, do bad stuff
279 <p>Example: This is basically the same thing that libev does internally, too.</p>
281 fatal_error (const char *msg)
288 ev_set_syserr_cb (fatal_error);
295 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
296 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
297 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
298 types of such loops, the <i>default</i> loop, which supports signals and child
299 events, and dynamically created loops which do not.</p>
300 <p>If you use threads, a common model is to run the default event loop
301 in your main thread (or in a separate thread) and for each thread you
302 create, you also create another event loop. Libev itself does no locking
303 whatsoever, so if you mix calls to the same event loop in different
304 threads, make sure you lock (this is usually a bad idea, though, even if
305 done correctly, because it's hideous and inefficient).</p>
307 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
309 <p>This will initialise the default event loop if it hasn't been initialised
310 yet and return it. If the default loop could not be initialised, returns
311 false. If it already was initialised it simply returns it (and ignores the
312 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
313 <p>If you don't know what event loop to use, use the one returned from this
315 <p>The flags argument can be used to specify special behaviour or specific
316 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
317 <p>The following flags are supported:</p>
320 <dt><code>EVFLAG_AUTO</code></dt>
322 <p>The default flags value. Use this if you have no clue (it's the right
323 thing, believe me).</p>
325 <dt><code>EVFLAG_NOENV</code></dt>
327 <p>If this flag bit is ored into the flag value (or the program runs setuid
328 or setgid) then libev will <i>not</i> look at the environment variable
329 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
330 override the flags completely if it is found in the environment. This is
331 useful to try out specific backends to test their performance, or to work
334 <dt><code>EVFLAG_FORKCHECK</code></dt>
336 <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
337 a fork, you can also make libev check for a fork in each iteration by
338 enabling this flag.</p>
339 <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
340 and thus this might slow down your event loop if you do a lot of loop
341 iterations and little real work, but is usually not noticeable (on my
342 Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
343 without a syscall and thus <i>very</i> fast, but my Linux system also has
344 <code>pthread_atfork</code> which is even faster).</p>
345 <p>The big advantage of this flag is that you can forget about fork (and
346 forget about forgetting to tell libev about forking) when you use this
348 <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
349 environment variable.</p>
351 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
353 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
354 libev tries to roll its own fd_set with no limits on the number of fds,
355 but if that fails, expect a fairly low limit on the number of fds when
356 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
357 the fastest backend for a low number of fds.</p>
359 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
361 <p>And this is your standard poll(2) backend. It's more complicated than
362 select, but handles sparse fds better and has no artificial limit on the
363 number of fds you can use (except it will slow down considerably with a
364 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
366 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
368 <p>For few fds, this backend is a bit little slower than poll and select,
369 but it scales phenomenally better. While poll and select usually scale like
370 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
371 either O(1) or O(active_fds).</p>
372 <p>While stopping and starting an I/O watcher in the same iteration will
373 result in some caching, there is still a syscall per such incident
374 (because the fd could point to a different file description now), so its
375 best to avoid that. Also, dup()ed file descriptors might not work very
376 well if you register events for both fds.</p>
377 <p>Please note that epoll sometimes generates spurious notifications, so you
378 need to use non-blocking I/O or other means to avoid blocking when no data
379 (or space) is available.</p>
381 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
383 <p>Kqueue deserves special mention, as at the time of this writing, it
384 was broken on all BSDs except NetBSD (usually it doesn't work with
385 anything but sockets and pipes, except on Darwin, where of course its
386 completely useless). For this reason its not being "autodetected"
387 unless you explicitly specify it explicitly in the flags (i.e. using
388 <code>EVBACKEND_KQUEUE</code>).</p>
389 <p>It scales in the same way as the epoll backend, but the interface to the
390 kernel is more efficient (which says nothing about its actual speed, of
391 course). While starting and stopping an I/O watcher does not cause an
392 extra syscall as with epoll, it still adds up to four event changes per
393 incident, so its best to avoid that.</p>
395 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
397 <p>This is not implemented yet (and might never be).</p>
399 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
401 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
402 it's really slow, but it still scales very well (O(active_fds)).</p>
403 <p>Please note that solaris ports can result in a lot of spurious
404 notifications, so you need to use non-blocking I/O or other means to avoid
405 blocking when no data (or space) is available.</p>
407 <dt><code>EVBACKEND_ALL</code></dt>
409 <p>Try all backends (even potentially broken ones that wouldn't be tried
410 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
411 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
415 <p>If one or more of these are ored into the flags value, then only these
416 backends will be tried (in the reverse order as given here). If none are
417 specified, most compiled-in backend will be tried, usually in reverse
418 order of their flag values :)</p>
419 <p>The most typical usage is like this:</p>
420 <pre> if (!ev_default_loop (0))
421 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
424 <p>Restrict libev to the select and poll backends, and do not allow
425 environment settings to be taken into account:</p>
426 <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
429 <p>Use whatever libev has to offer, but make sure that kqueue is used if
430 available (warning, breaks stuff, best use only with your own private
431 event loop and only if you know the OS supports your types of fds):</p>
432 <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
436 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
438 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
439 always distinct from the default loop. Unlike the default loop, it cannot
440 handle signal and child watchers, and attempts to do so will be greeted by
441 undefined behaviour (or a failed assertion if assertions are enabled).</p>
442 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
443 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
445 fatal ("no epoll found here, maybe it hides under your chair");
449 <dt>ev_default_destroy ()</dt>
451 <p>Destroys the default loop again (frees all memory and kernel state
452 etc.). None of the active event watchers will be stopped in the normal
453 sense, so e.g. <code>ev_is_active</code> might still return true. It is your
454 responsibility to either stop all watchers cleanly yoursef <i>before</i>
455 calling this function, or cope with the fact afterwards (which is usually
456 the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
459 <dt>ev_loop_destroy (loop)</dt>
461 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
462 earlier call to <code>ev_loop_new</code>.</p>
464 <dt>ev_default_fork ()</dt>
466 <p>This function reinitialises the kernel state for backends that have
467 one. Despite the name, you can call it anytime, but it makes most sense
468 after forking, in either the parent or child process (or both, but that
469 again makes little sense).</p>
470 <p>You <i>must</i> call this function in the child process after forking if and
471 only if you want to use the event library in both processes. If you just
472 fork+exec, you don't have to call it.</p>
473 <p>The function itself is quite fast and it's usually not a problem to call
474 it just in case after a fork. To make this easy, the function will fit in
475 quite nicely into a call to <code>pthread_atfork</code>:</p>
476 <pre> pthread_atfork (0, 0, ev_default_fork);
479 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
480 without calling this function, so if you force one of those backends you
481 do not need to care.</p>
483 <dt>ev_loop_fork (loop)</dt>
485 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
486 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
487 after fork, and how you do this is entirely your own problem.</p>
489 <dt>unsigned int ev_loop_count (loop)</dt>
491 <p>Returns the count of loop iterations for the loop, which is identical to
492 the number of times libev did poll for new events. It starts at <code>0</code> and
493 happily wraps around with enough iterations.</p>
494 <p>This value can sometimes be useful as a generation counter of sorts (it
495 "ticks" the number of loop iterations), as it roughly corresponds with
496 <code>ev_prepare</code> and <code>ev_check</code> calls.</p>
498 <dt>unsigned int ev_backend (loop)</dt>
500 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
503 <dt>ev_tstamp ev_now (loop)</dt>
505 <p>Returns the current "event loop time", which is the time the event loop
506 received events and started processing them. This timestamp does not
507 change as long as callbacks are being processed, and this is also the base
508 time used for relative timers. You can treat it as the timestamp of the
509 event occuring (or more correctly, libev finding out about it).</p>
511 <dt>ev_loop (loop, int flags)</dt>
513 <p>Finally, this is it, the event handler. This function usually is called
514 after you initialised all your watchers and you want to start handling
516 <p>If the flags argument is specified as <code>0</code>, it will not return until
517 either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
518 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
519 relying on all watchers to be stopped when deciding when a program has
520 finished (especially in interactive programs), but having a program that
521 automatically loops as long as it has to and no longer by virtue of
522 relying on its watchers stopping correctly is a thing of beauty.</p>
523 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
524 those events and any outstanding ones, but will not block your process in
525 case there are no events and will return after one iteration of the loop.</p>
526 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
527 neccessary) and will handle those and any outstanding ones. It will block
528 your process until at least one new event arrives, and will return after
529 one iteration of the loop. This is useful if you are waiting for some
530 external event in conjunction with something not expressible using other
531 libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
532 usually a better approach for this kind of thing.</p>
533 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
534 <pre> * If there are no active watchers (reference count is zero), return.
535 - Queue prepare watchers and then call all outstanding watchers.
536 - If we have been forked, recreate the kernel state.
537 - Update the kernel state with all outstanding changes.
538 - Update the "event loop time".
539 - Calculate for how long to block.
540 - Block the process, waiting for any events.
541 - Queue all outstanding I/O (fd) events.
542 - Update the "event loop time" and do time jump handling.
543 - Queue all outstanding timers.
544 - Queue all outstanding periodics.
545 - If no events are pending now, queue all idle watchers.
546 - Queue all check watchers.
547 - Call all queued watchers in reverse order (i.e. check watchers first).
548 Signals and child watchers are implemented as I/O watchers, and will
549 be handled here by queueing them when their watcher gets executed.
550 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
551 were used, return, otherwise continue with step *.
554 <p>Example: Queue some jobs and then loop until no events are outsanding
556 <pre> ... queue jobs here, make sure they register event watchers as long
557 ... as they still have work to do (even an idle watcher will do..)
558 ev_loop (my_loop, 0);
563 <dt>ev_unloop (loop, how)</dt>
565 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
566 has processed all outstanding events). The <code>how</code> argument must be either
567 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
568 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
570 <dt>ev_ref (loop)</dt>
571 <dt>ev_unref (loop)</dt>
573 <p>Ref/unref can be used to add or remove a reference count on the event
574 loop: Every watcher keeps one reference, and as long as the reference
575 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
576 a watcher you never unregister that should not keep <code>ev_loop</code> from
577 returning, ev_unref() after starting, and ev_ref() before stopping it. For
578 example, libev itself uses this for its internal signal pipe: It is not
579 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
580 no event watchers registered by it are active. It is also an excellent
581 way to do this for generic recurring timers or from within third-party
582 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
583 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
584 running when nothing else is active.</p>
585 <pre> struct ev_signal exitsig;
586 ev_signal_init (&exitsig, sig_cb, SIGINT);
587 ev_signal_start (loop, &exitsig);
591 <p>Example: For some weird reason, unregister the above signal handler again.</p>
593 ev_signal_stop (loop, &exitsig);
604 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
605 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
606 <p>A watcher is a structure that you create and register to record your
607 interest in some event. For instance, if you want to wait for STDIN to
608 become readable, you would create an <code>ev_io</code> watcher for that:</p>
609 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
612 ev_unloop (loop, EVUNLOOP_ALL);
615 struct ev_loop *loop = ev_default_loop (0);
616 struct ev_io stdin_watcher;
617 ev_init (&stdin_watcher, my_cb);
618 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
619 ev_io_start (loop, &stdin_watcher);
623 <p>As you can see, you are responsible for allocating the memory for your
624 watcher structures (and it is usually a bad idea to do this on the stack,
625 although this can sometimes be quite valid).</p>
626 <p>Each watcher structure must be initialised by a call to <code>ev_init
627 (watcher *, callback)</code>, which expects a callback to be provided. This
628 callback gets invoked each time the event occurs (or, in the case of io
629 watchers, each time the event loop detects that the file descriptor given
630 is readable and/or writable).</p>
631 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
632 with arguments specific to this watcher type. There is also a macro
633 to combine initialisation and setting in one call: <code>ev_<type>_init
634 (watcher *, callback, ...)</code>.</p>
635 <p>To make the watcher actually watch out for events, you have to start it
636 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
637 *)</code>), and you can stop watching for events at any time by calling the
638 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
639 <p>As long as your watcher is active (has been started but not stopped) you
640 must not touch the values stored in it. Most specifically you must never
641 reinitialise it or call its <code>set</code> macro.</p>
642 <p>Each and every callback receives the event loop pointer as first, the
643 registered watcher structure as second, and a bitset of received events as
645 <p>The received events usually include a single bit per event type received
646 (you can receive multiple events at the same time). The possible bit masks
649 <dt><code>EV_READ</code></dt>
650 <dt><code>EV_WRITE</code></dt>
652 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
655 <dt><code>EV_TIMEOUT</code></dt>
657 <p>The <code>ev_timer</code> watcher has timed out.</p>
659 <dt><code>EV_PERIODIC</code></dt>
661 <p>The <code>ev_periodic</code> watcher has timed out.</p>
663 <dt><code>EV_SIGNAL</code></dt>
665 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
667 <dt><code>EV_CHILD</code></dt>
669 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
671 <dt><code>EV_STAT</code></dt>
673 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
675 <dt><code>EV_IDLE</code></dt>
677 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
679 <dt><code>EV_PREPARE</code></dt>
680 <dt><code>EV_CHECK</code></dt>
682 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
683 to gather new events, and all <code>ev_check</code> watchers are invoked just after
684 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
685 received events. Callbacks of both watcher types can start and stop as
686 many watchers as they want, and all of them will be taken into account
687 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
688 <code>ev_loop</code> from blocking).</p>
690 <dt><code>EV_EMBED</code></dt>
692 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
694 <dt><code>EV_FORK</code></dt>
696 <p>The event loop has been resumed in the child process after fork (see
697 <code>ev_fork</code>).</p>
699 <dt><code>EV_ERROR</code></dt>
701 <p>An unspecified error has occured, the watcher has been stopped. This might
702 happen because the watcher could not be properly started because libev
703 ran out of memory, a file descriptor was found to be closed or any other
704 problem. You best act on it by reporting the problem and somehow coping
705 with the watcher being stopped.</p>
706 <p>Libev will usually signal a few "dummy" events together with an error,
707 for example it might indicate that a fd is readable or writable, and if
708 your callbacks is well-written it can just attempt the operation and cope
709 with the error from read() or write(). This will not work in multithreaded
710 programs, though, so beware.</p>
715 <h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
716 <div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
717 <p>In the following description, <code>TYPE</code> stands for the watcher type,
718 e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
720 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
722 <p>This macro initialises the generic portion of a watcher. The contents
723 of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
724 the generic parts of the watcher are initialised, you <i>need</i> to call
725 the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
726 type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
727 which rolls both calls into one.</p>
728 <p>You can reinitialise a watcher at any time as long as it has been stopped
729 (or never started) and there are no pending events outstanding.</p>
730 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
731 int revents)</code>.</p>
733 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
735 <p>This macro initialises the type-specific parts of a watcher. You need to
736 call <code>ev_init</code> at least once before you call this macro, but you can
737 call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
738 macro on a watcher that is active (it can be pending, however, which is a
739 difference to the <code>ev_init</code> macro).</p>
740 <p>Although some watcher types do not have type-specific arguments
741 (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
743 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
745 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
746 calls into a single call. This is the most convinient method to initialise
747 a watcher. The same limitations apply, of course.</p>
749 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
751 <p>Starts (activates) the given watcher. Only active watchers will receive
752 events. If the watcher is already active nothing will happen.</p>
754 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
756 <p>Stops the given watcher again (if active) and clears the pending
757 status. It is possible that stopped watchers are pending (for example,
758 non-repeating timers are being stopped when they become pending), but
759 <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
760 you want to free or reuse the memory used by the watcher it is therefore a
761 good idea to always call its <code>ev_TYPE_stop</code> function.</p>
763 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
765 <p>Returns a true value iff the watcher is active (i.e. it has been started
766 and not yet been stopped). As long as a watcher is active you must not modify
769 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
771 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
772 events but its callback has not yet been invoked). As long as a watcher
773 is pending (but not active) you must not call an init function on it (but
774 <code>ev_TYPE_set</code> is safe) and you must make sure the watcher is available to
775 libev (e.g. you cnanot <code>free ()</code> it).</p>
777 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
779 <p>Returns the callback currently set on the watcher.</p>
781 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
783 <p>Change the callback. You can change the callback at virtually any time
784 (modulo threads).</p>
786 <dt>ev_set_priority (ev_TYPE *watcher, priority)</dt>
787 <dt>int ev_priority (ev_TYPE *watcher)</dt>
789 <p>Set and query the priority of the watcher. The priority is a small
790 integer between <code>EV_MAXPRI</code> (default: <code>2</code>) and <code>EV_MINPRI</code>
791 (default: <code>-2</code>). Pending watchers with higher priority will be invoked
792 before watchers with lower priority, but priority will not keep watchers
793 from being executed (except for <code>ev_idle</code> watchers).</p>
794 <p>This means that priorities are <i>only</i> used for ordering callback
795 invocation after new events have been received. This is useful, for
796 example, to reduce latency after idling, or more often, to bind two
797 watchers on the same event and make sure one is called first.</p>
798 <p>If you need to suppress invocation when higher priority events are pending
799 you need to look at <code>ev_idle</code> watchers, which provide this functionality.</p>
800 <p>The default priority used by watchers when no priority has been set is
801 always <code>0</code>, which is supposed to not be too high and not be too low :).</p>
802 <p>Setting a priority outside the range of <code>EV_MINPRI</code> to <code>EV_MAXPRI</code> is
803 fine, as long as you do not mind that the priority value you query might
804 or might not have been adjusted to be within valid range.</p>
813 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
814 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
815 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
816 and read at any time, libev will completely ignore it. This can be used
817 to associate arbitrary data with your watcher. If you need more data and
818 don't want to allocate memory and store a pointer to it in that data
819 member, you can also "subclass" the watcher type and provide your own
826 struct whatever *mostinteresting;
830 <p>And since your callback will be called with a pointer to the watcher, you
831 can cast it back to your own type:</p>
832 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
834 struct my_io *w = (struct my_io *)w_;
839 <p>More interesting and less C-conformant ways of casting your callback type
840 instead have been omitted.</p>
841 <p>Another common scenario is having some data structure with multiple
843 <pre> struct my_biggy
851 <p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
852 you need to use <code>offsetof</code>:</p>
853 <pre> #include <stddef.h>
856 t1_cb (EV_P_ struct ev_timer *w, int revents)
858 struct my_biggy big = (struct my_biggy *
859 (((char *)w) - offsetof (struct my_biggy, t1));
863 t2_cb (EV_P_ struct ev_timer *w, int revents)
865 struct my_biggy big = (struct my_biggy *
866 (((char *)w) - offsetof (struct my_biggy, t2));
875 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
876 <div id="WATCHER_TYPES_CONTENT">
877 <p>This section describes each watcher in detail, but will not repeat
878 information given in the last section. Any initialisation/set macros,
879 functions and members specific to the watcher type are explained.</p>
880 <p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
881 while the watcher is active, you can look at the member and expect some
882 sensible content, but you must not modify it (you can modify it while the
883 watcher is stopped to your hearts content), or <i>[read-write]</i>, which
884 means you can expect it to have some sensible content while the watcher
885 is active, but you can also modify it. Modifying it may not do something
886 sensible or take immediate effect (or do anything at all), but libev will
887 not crash or malfunction in any way.</p>
894 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
895 <div id="code_ev_io_code_is_this_file_descrip-2">
896 <p>I/O watchers check whether a file descriptor is readable or writable
897 in each iteration of the event loop, or, more precisely, when reading
898 would not block the process and writing would at least be able to write
899 some data. This behaviour is called level-triggering because you keep
900 receiving events as long as the condition persists. Remember you can stop
901 the watcher if you don't want to act on the event and neither want to
902 receive future events.</p>
903 <p>In general you can register as many read and/or write event watchers per
904 fd as you want (as long as you don't confuse yourself). Setting all file
905 descriptors to non-blocking mode is also usually a good idea (but not
906 required if you know what you are doing).</p>
907 <p>You have to be careful with dup'ed file descriptors, though. Some backends
908 (the linux epoll backend is a notable example) cannot handle dup'ed file
909 descriptors correctly if you register interest in two or more fds pointing
910 to the same underlying file/socket/etc. description (that is, they share
911 the same underlying "file open").</p>
912 <p>If you must do this, then force the use of a known-to-be-good backend
913 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
914 <code>EVBACKEND_POLL</code>).</p>
915 <p>Another thing you have to watch out for is that it is quite easy to
916 receive "spurious" readyness notifications, that is your callback might
917 be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
918 because there is no data. Not only are some backends known to create a
919 lot of those (for example solaris ports), it is very easy to get into
920 this situation even with a relatively standard program structure. Thus
921 it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
922 <code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
923 <p>If you cannot run the fd in non-blocking mode (for example you should not
924 play around with an Xlib connection), then you have to seperately re-test
925 whether a file descriptor is really ready with a known-to-be good interface
926 such as poll (fortunately in our Xlib example, Xlib already does this on
927 its own, so its quite safe to use).</p>
929 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
930 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
932 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
933 rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
934 <code>EV_READ | EV_WRITE</code> to receive the given events.</p>
936 <dt>int fd [read-only]</dt>
938 <p>The file descriptor being watched.</p>
940 <dt>int events [read-only]</dt>
942 <p>The events being watched.</p>
945 <p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
946 readable, but only once. Since it is likely line-buffered, you could
947 attempt to read a whole line in the callback.</p>
949 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
951 ev_io_stop (loop, w);
952 .. read from stdin here (or from w->fd) and haqndle any I/O errors
956 struct ev_loop *loop = ev_default_init (0);
957 struct ev_io stdin_readable;
958 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
959 ev_io_start (loop, &stdin_readable);
968 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
969 <div id="code_ev_timer_code_relative_and_opti-2">
970 <p>Timer watchers are simple relative timers that generate an event after a
971 given time, and optionally repeating in regular intervals after that.</p>
972 <p>The timers are based on real time, that is, if you register an event that
973 times out after an hour and you reset your system clock to last years
974 time, it will still time out after (roughly) and hour. "Roughly" because
975 detecting time jumps is hard, and some inaccuracies are unavoidable (the
976 monotonic clock option helps a lot here).</p>
977 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
978 time. This is usually the right thing as this timestamp refers to the time
979 of the event triggering whatever timeout you are modifying/starting. If
980 you suspect event processing to be delayed and you <i>need</i> to base the timeout
981 on the current time, use something like this to adjust for this:</p>
982 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
985 <p>The callback is guarenteed to be invoked only when its timeout has passed,
986 but if multiple timers become ready during the same loop iteration then
987 order of execution is undefined.</p>
989 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
990 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
992 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
993 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
994 timer will automatically be configured to trigger again <code>repeat</code> seconds
995 later, again, and again, until stopped manually.</p>
996 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
997 configure a timer to trigger every 10 seconds, then it will trigger at
998 exactly 10 second intervals. If, however, your program cannot keep up with
999 the timer (because it takes longer than those 10 seconds to do stuff) the
1000 timer will not fire more than once per event loop iteration.</p>
1002 <dt>ev_timer_again (loop)</dt>
1004 <p>This will act as if the timer timed out and restart it again if it is
1005 repeating. The exact semantics are:</p>
1006 <p>If the timer is pending, its pending status is cleared.</p>
1007 <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
1008 <p>If the timer is repeating, either start it if necessary (with the
1009 <code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
1010 <p>This sounds a bit complicated, but here is a useful and typical
1011 example: Imagine you have a tcp connection and you want a so-called idle
1012 timeout, that is, you want to be called when there have been, say, 60
1013 seconds of inactivity on the socket. The easiest way to do this is to
1014 configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
1015 <code>ev_timer_again</code> each time you successfully read or write some data. If
1016 you go into an idle state where you do not expect data to travel on the
1017 socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
1018 automatically restart it if need be.</p>
1019 <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
1020 altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
1021 <pre> ev_timer_init (timer, callback, 0., 5.);
1022 ev_timer_again (loop, timer);
1024 timer->again = 17.;
1025 ev_timer_again (loop, timer);
1027 timer->again = 10.;
1028 ev_timer_again (loop, timer);
1031 <p>This is more slightly efficient then stopping/starting the timer each time
1032 you want to modify its timeout value.</p>
1034 <dt>ev_tstamp repeat [read-write]</dt>
1036 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1037 or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1038 which is also when any modifications are taken into account.</p>
1041 <p>Example: Create a timer that fires after 60 seconds.</p>
1043 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1045 .. one minute over, w is actually stopped right here
1048 struct ev_timer mytimer;
1049 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1050 ev_timer_start (loop, &mytimer);
1053 <p>Example: Create a timeout timer that times out after 10 seconds of
1056 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1058 .. ten seconds without any activity
1061 struct ev_timer mytimer;
1062 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1063 ev_timer_again (&mytimer); /* start timer */
1066 // and in some piece of code that gets executed on any "activity":
1067 // reset the timeout to start ticking again at 10 seconds
1068 ev_timer_again (&mytimer);
1076 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
1077 <div id="code_ev_periodic_code_to_cron_or_not-2">
1078 <p>Periodic watchers are also timers of a kind, but they are very versatile
1079 (and unfortunately a bit complex).</p>
1080 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
1081 but on wallclock time (absolute time). You can tell a periodic watcher
1082 to trigger "at" some specific point in time. For example, if you tell a
1083 periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
1084 + 10.</code>) and then reset your system clock to the last year, then it will
1085 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
1086 roughly 10 seconds later and of course not if you reset your system time
1088 <p>They can also be used to implement vastly more complex timers, such as
1089 triggering an event on eahc midnight, local time.</p>
1090 <p>As with timers, the callback is guarenteed to be invoked only when the
1091 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1092 during the same loop iteration then order of execution is undefined.</p>
1094 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
1095 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
1097 <p>Lots of arguments, lets sort it out... There are basically three modes of
1098 operation, and we will explain them from simplest to complex:</p>
1101 <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
1103 <p>In this configuration the watcher triggers an event at the wallclock time
1104 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
1105 that is, if it is to be run at January 1st 2011 then it will run when the
1106 system time reaches or surpasses this time.</p>
1108 <dt>* non-repeating interval timer (interval > 0, reschedule_cb = 0)</dt>
1110 <p>In this mode the watcher will always be scheduled to time out at the next
1111 <code>at + N * interval</code> time (for some integer N) and then repeat, regardless
1112 of any time jumps.</p>
1113 <p>This can be used to create timers that do not drift with respect to system
1115 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
1118 <p>This doesn't mean there will always be 3600 seconds in between triggers,
1119 but only that the the callback will be called when the system time shows a
1120 full hour (UTC), or more correctly, when the system time is evenly divisible
1122 <p>Another way to think about it (for the mathematically inclined) is that
1123 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
1124 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1126 <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
1128 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
1129 ignored. Instead, each time the periodic watcher gets scheduled, the
1130 reschedule callback will be called with the watcher as first, and the
1131 current time as second argument.</p>
1132 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
1133 ever, or make any event loop modifications</i>. If you need to stop it,
1134 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1135 starting a prepare watcher).</p>
1136 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1137 ev_tstamp now)</code>, e.g.:</p>
1138 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1144 <p>It must return the next time to trigger, based on the passed time value
1145 (that is, the lowest time value larger than to the second argument). It
1146 will usually be called just before the callback will be triggered, but
1147 might be called at other times, too.</p>
1148 <p>NOTE: <i>This callback must always return a time that is later than the
1149 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
1150 <p>This can be used to create very complex timers, such as a timer that
1151 triggers on each midnight, local time. To do this, you would calculate the
1152 next midnight after <code>now</code> and return the timestamp value for this. How
1153 you do this is, again, up to you (but it is not trivial, which is the main
1154 reason I omitted it as an example).</p>
1159 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
1161 <p>Simply stops and restarts the periodic watcher again. This is only useful
1162 when you changed some parameters or the reschedule callback would return
1163 a different time than the last time it was called (e.g. in a crond like
1164 program when the crontabs have changed).</p>
1166 <dt>ev_tstamp interval [read-write]</dt>
1168 <p>The current interval value. Can be modified any time, but changes only
1169 take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1172 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1174 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1175 switched off. Can be changed any time, but changes only take effect when
1176 the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1179 <p>Example: Call a callback every hour, or, more precisely, whenever the
1180 system clock is divisible by 3600. The callback invocation times have
1181 potentially a lot of jittering, but good long-term stability.</p>
1183 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1185 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1188 struct ev_periodic hourly_tick;
1189 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1190 ev_periodic_start (loop, &hourly_tick);
1193 <p>Example: The same as above, but use a reschedule callback to do it:</p>
1194 <pre> #include <math.h>
1197 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1199 return fmod (now, 3600.) + 3600.;
1202 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1205 <p>Example: Call a callback every hour, starting now:</p>
1206 <pre> struct ev_periodic hourly_tick;
1207 ev_periodic_init (&hourly_tick, clock_cb,
1208 fmod (ev_now (loop), 3600.), 3600., 0);
1209 ev_periodic_start (loop, &hourly_tick);
1217 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
1218 <div id="code_ev_signal_code_signal_me_when_a-2">
1219 <p>Signal watchers will trigger an event when the process receives a specific
1220 signal one or more times. Even though signals are very asynchronous, libev
1221 will try it's best to deliver signals synchronously, i.e. as part of the
1222 normal event processing, like any other event.</p>
1223 <p>You can configure as many watchers as you like per signal. Only when the
1224 first watcher gets started will libev actually register a signal watcher
1225 with the kernel (thus it coexists with your own signal handlers as long
1226 as you don't register any with libev). Similarly, when the last signal
1227 watcher for a signal is stopped libev will reset the signal handler to
1228 SIG_DFL (regardless of what it was set to before).</p>
1230 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1231 <dt>ev_signal_set (ev_signal *, int signum)</dt>
1233 <p>Configures the watcher to trigger on the given signal number (usually one
1234 of the <code>SIGxxx</code> constants).</p>
1236 <dt>int signum [read-only]</dt>
1238 <p>The signal the watcher watches out for.</p>
1247 <h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
1248 <div id="code_ev_child_code_watch_out_for_pro-2">
1249 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1250 some child status changes (most typically when a child of yours dies).</p>
1252 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1253 <dt>ev_child_set (ev_child *, int pid)</dt>
1255 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1256 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1257 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1258 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1259 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1260 process causing the status change.</p>
1262 <dt>int pid [read-only]</dt>
1264 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1266 <dt>int rpid [read-write]</dt>
1268 <p>The process id that detected a status change.</p>
1270 <dt>int rstatus [read-write]</dt>
1272 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1273 <code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
1276 <p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1278 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1280 ev_unloop (loop, EVUNLOOP_ALL);
1283 struct ev_signal signal_watcher;
1284 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1285 ev_signal_start (loop, &sigint_cb);
1293 <h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1294 <div id="code_ev_stat_code_did_the_file_attri-2">
1295 <p>This watches a filesystem path for attribute changes. That is, it calls
1296 <code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1297 compared to the last time, invoking the callback if it did.</p>
1298 <p>The path does not need to exist: changing from "path exists" to "path does
1299 not exist" is a status change like any other. The condition "path does
1300 not exist" is signified by the <code>st_nlink</code> field being zero (which is
1301 otherwise always forced to be at least one) and all the other fields of
1302 the stat buffer having unspecified contents.</p>
1303 <p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1304 relative and your working directory changes, the behaviour is undefined.</p>
1305 <p>Since there is no standard to do this, the portable implementation simply
1306 calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1307 can specify a recommended polling interval for this case. If you specify
1308 a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1309 unspecified default</i> value will be used (which you can expect to be around
1310 five seconds, although this might change dynamically). Libev will also
1311 impose a minimum interval which is currently around <code>0.1</code>, but thats
1312 usually overkill.</p>
1313 <p>This watcher type is not meant for massive numbers of stat watchers,
1314 as even with OS-supported change notifications, this can be
1315 resource-intensive.</p>
1316 <p>At the time of this writing, only the Linux inotify interface is
1317 implemented (implementing kqueue support is left as an exercise for the
1318 reader). Inotify will be used to give hints only and should not change the
1319 semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1320 to fall back to regular polling again even with inotify, but changes are
1321 usually detected immediately, and if the file exists there will be no
1324 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1325 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1327 <p>Configures the watcher to wait for status changes of the given
1328 <code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1329 be detected and should normally be specified as <code>0</code> to let libev choose
1330 a suitable value. The memory pointed to by <code>path</code> must point to the same
1331 path for as long as the watcher is active.</p>
1332 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1333 relative to the attributes at the time the watcher was started (or the
1334 last change was detected).</p>
1336 <dt>ev_stat_stat (ev_stat *)</dt>
1338 <p>Updates the stat buffer immediately with new values. If you change the
1339 watched path in your callback, you could call this fucntion to avoid
1340 detecting this change (while introducing a race condition). Can also be
1341 useful simply to find out the new values.</p>
1343 <dt>ev_statdata attr [read-only]</dt>
1345 <p>The most-recently detected attributes of the file. Although the type is of
1346 <code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1347 suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1348 was some error while <code>stat</code>ing the file.</p>
1350 <dt>ev_statdata prev [read-only]</dt>
1352 <p>The previous attributes of the file. The callback gets invoked whenever
1353 <code>prev</code> != <code>attr</code>.</p>
1355 <dt>ev_tstamp interval [read-only]</dt>
1357 <p>The specified interval.</p>
1359 <dt>const char *path [read-only]</dt>
1361 <p>The filesystem path that is being watched.</p>
1364 <p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1366 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1368 /* /etc/passwd changed in some way */
1369 if (w->attr.st_nlink)
1371 printf ("passwd current size %ld\n", (long)w->attr.st_size);
1372 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1373 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1376 /* you shalt not abuse printf for puts */
1377 puts ("wow, /etc/passwd is not there, expect problems. "
1378 "if this is windows, they already arrived\n");
1384 ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1385 ev_stat_start (loop, &passwd);
1393 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
1394 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1395 <p>Idle watchers trigger events when no other events of the same or higher
1396 priority are pending (prepare, check and other idle watchers do not
1398 <p>That is, as long as your process is busy handling sockets or timeouts
1399 (or even signals, imagine) of the same or higher priority it will not be
1400 triggered. But when your process is idle (or only lower-priority watchers
1401 are pending), the idle watchers are being called once per event loop
1402 iteration - until stopped, that is, or your process receives more events
1403 and becomes busy again with higher priority stuff.</p>
1404 <p>The most noteworthy effect is that as long as any idle watchers are
1405 active, the process will not block when waiting for new events.</p>
1406 <p>Apart from keeping your process non-blocking (which is a useful
1407 effect on its own sometimes), idle watchers are a good place to do
1408 "pseudo-background processing", or delay processing stuff to after the
1409 event loop has handled all outstanding events.</p>
1411 <dt>ev_idle_init (ev_signal *, callback)</dt>
1413 <p>Initialises and configures the idle watcher - it has no parameters of any
1414 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1418 <p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1419 callback, free it. Also, use no error checking, as usual.</p>
1421 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1424 // now do something you wanted to do when the program has
1425 // no longer asnything immediate to do.
1428 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1429 ev_idle_init (idle_watcher, idle_cb);
1430 ev_idle_start (loop, idle_cb);
1438 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
1439 <div id="code_ev_prepare_code_and_code_ev_che-2">
1440 <p>Prepare and check watchers are usually (but not always) used in tandem:
1441 prepare watchers get invoked before the process blocks and check watchers
1443 <p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1444 the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1445 watchers. Other loops than the current one are fine, however. The
1446 rationale behind this is that you do not need to check for recursion in
1447 those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1448 <code>ev_check</code> so if you have one watcher of each kind they will always be
1449 called in pairs bracketing the blocking call.</p>
1450 <p>Their main purpose is to integrate other event mechanisms into libev and
1451 their use is somewhat advanced. This could be used, for example, to track
1452 variable changes, implement your own watchers, integrate net-snmp or a
1453 coroutine library and lots more. They are also occasionally useful if
1454 you cache some data and want to flush it before blocking (for example,
1455 in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1457 <p>This is done by examining in each prepare call which file descriptors need
1458 to be watched by the other library, registering <code>ev_io</code> watchers for
1459 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1460 provide just this functionality). Then, in the check watcher you check for
1461 any events that occured (by checking the pending status of all watchers
1462 and stopping them) and call back into the library. The I/O and timer
1463 callbacks will never actually be called (but must be valid nevertheless,
1464 because you never know, you know?).</p>
1465 <p>As another example, the Perl Coro module uses these hooks to integrate
1466 coroutines into libev programs, by yielding to other active coroutines
1467 during each prepare and only letting the process block if no coroutines
1468 are ready to run (it's actually more complicated: it only runs coroutines
1469 with priority higher than or equal to the event loop and one coroutine
1470 of lower priority, but only once, using idle watchers to keep the event
1471 loop from blocking if lower-priority coroutines are active, thus mapping
1472 low-priority coroutines to idle/background tasks).</p>
1474 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1475 <dt>ev_check_init (ev_check *, callback)</dt>
1477 <p>Initialises and configures the prepare or check watcher - they have no
1478 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1479 macros, but using them is utterly, utterly and completely pointless.</p>
1482 <p>Example: To include a library such as adns, you would add IO watchers
1483 and a timeout watcher in a prepare handler, as required by libadns, and
1484 in a check watcher, destroy them and call into libadns. What follows is
1485 pseudo-code only of course:</p>
1486 <pre> static ev_io iow [nfd];
1490 io_cb (ev_loop *loop, ev_io *w, int revents)
1492 // set the relevant poll flags
1493 // could also call adns_processreadable etc. here
1494 struct pollfd *fd = (struct pollfd *)w->data;
1495 if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1496 if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1499 // create io watchers for each fd and a timer before blocking
1501 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1503 int timeout = 3600000;
1504 struct pollfd fds [nfd];
1505 // actual code will need to loop here and realloc etc.
1506 adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1508 /* the callback is illegal, but won't be called as we stop during check */
1509 ev_timer_init (&tw, 0, timeout * 1e-3);
1510 ev_timer_start (loop, &tw);
1512 // create on ev_io per pollfd
1513 for (int i = 0; i < nfd; ++i)
1515 ev_io_init (iow + i, io_cb, fds [i].fd,
1516 ((fds [i].events & POLLIN ? EV_READ : 0)
1517 | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1519 fds [i].revents = 0;
1520 iow [i].data = fds + i;
1521 ev_io_start (loop, iow + i);
1525 // stop all watchers after blocking
1527 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1529 ev_timer_stop (loop, &tw);
1531 for (int i = 0; i < nfd; ++i)
1532 ev_io_stop (loop, iow + i);
1534 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1543 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1544 <div id="code_ev_embed_code_when_one_backend_-2">
1545 <p>This is a rather advanced watcher type that lets you embed one event loop
1546 into another (currently only <code>ev_io</code> events are supported in the embedded
1547 loop, other types of watchers might be handled in a delayed or incorrect
1548 fashion and must not be used).</p>
1549 <p>There are primarily two reasons you would want that: work around bugs and
1551 <p>As an example for a bug workaround, the kqueue backend might only support
1552 sockets on some platform, so it is unusable as generic backend, but you
1553 still want to make use of it because you have many sockets and it scales
1554 so nicely. In this case, you would create a kqueue-based loop and embed it
1555 into your default loop (which might use e.g. poll). Overall operation will
1556 be a bit slower because first libev has to poll and then call kevent, but
1557 at least you can use both at what they are best.</p>
1558 <p>As for prioritising I/O: rarely you have the case where some fds have
1559 to be watched and handled very quickly (with low latency), and even
1560 priorities and idle watchers might have too much overhead. In this case
1561 you would put all the high priority stuff in one loop and all the rest in
1562 a second one, and embed the second one in the first.</p>
1563 <p>As long as the watcher is active, the callback will be invoked every time
1564 there might be events pending in the embedded loop. The callback must then
1565 call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1566 their callbacks (you could also start an idle watcher to give the embedded
1567 loop strictly lower priority for example). You can also set the callback
1568 to <code>0</code>, in which case the embed watcher will automatically execute the
1569 embedded loop sweep.</p>
1570 <p>As long as the watcher is started it will automatically handle events. The
1571 callback will be invoked whenever some events have been handled. You can
1572 set the callback to <code>0</code> to avoid having to specify one if you are not
1573 interested in that.</p>
1574 <p>Also, there have not currently been made special provisions for forking:
1575 when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1576 but you will also have to stop and restart any <code>ev_embed</code> watchers
1578 <p>Unfortunately, not all backends are embeddable, only the ones returned by
1579 <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1581 <p>So when you want to use this feature you will always have to be prepared
1582 that you cannot get an embeddable loop. The recommended way to get around
1583 this is to have a separate variables for your embeddable loop, try to
1584 create it, and if that fails, use the normal loop for everything:</p>
1585 <pre> struct ev_loop *loop_hi = ev_default_init (0);
1586 struct ev_loop *loop_lo = 0;
1587 struct ev_embed embed;
1589 // see if there is a chance of getting one that works
1590 // (remember that a flags value of 0 means autodetection)
1591 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1592 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1595 // if we got one, then embed it, otherwise default to loop_hi
1598 ev_embed_init (&embed, 0, loop_lo);
1599 ev_embed_start (loop_hi, &embed);
1606 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1607 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1609 <p>Configures the watcher to embed the given loop, which must be
1610 embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1611 invoked automatically, otherwise it is the responsibility of the callback
1612 to invoke it (it will continue to be called until the sweep has been done,
1613 if you do not want thta, you need to temporarily stop the embed watcher).</p>
1615 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1617 <p>Make a single, non-blocking sweep over the embedded loop. This works
1618 similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1619 apropriate way for embedded loops.</p>
1621 <dt>struct ev_loop *loop [read-only]</dt>
1623 <p>The embedded event loop.</p>
1632 <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>
1633 <div id="code_ev_fork_code_the_audacity_to_re-2">
1634 <p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1635 whoever is a good citizen cared to tell libev about it by calling
1636 <code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1637 event loop blocks next and before <code>ev_check</code> watchers are being called,
1638 and only in the child after the fork. If whoever good citizen calling
1639 <code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1640 handlers will be invoked, too, of course.</p>
1642 <dt>ev_fork_init (ev_signal *, callback)</dt>
1644 <p>Initialises and configures the fork watcher - it has no parameters of any
1645 kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1655 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
1656 <div id="OTHER_FUNCTIONS_CONTENT">
1657 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1659 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1661 <p>This function combines a simple timer and an I/O watcher, calls your
1662 callback on whichever event happens first and automatically stop both
1663 watchers. This is useful if you want to wait for a single event on an fd
1664 or timeout without having to allocate/configure/start/stop/free one or
1665 more watchers yourself.</p>
1666 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1667 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1668 <code>events</code> set will be craeted and started.</p>
1669 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1670 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1671 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1673 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1674 passed an <code>revents</code> set like normal event callbacks (a combination of
1675 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1676 value passed to <code>ev_once</code>:</p>
1677 <pre> static void stdin_ready (int revents, void *arg)
1679 if (revents & EV_TIMEOUT)
1680 /* doh, nothing entered */;
1681 else if (revents & EV_READ)
1682 /* stdin might have data for us, joy! */;
1685 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1689 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1691 <p>Feeds the given event set into the event loop, as if the specified event
1692 had happened for the specified watcher (which must be a pointer to an
1693 initialised but not necessarily started event watcher).</p>
1695 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1697 <p>Feed an event on the given fd, as if a file descriptor backend detected
1698 the given events it.</p>
1700 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1702 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1712 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1713 <div id="LIBEVENT_EMULATION_CONTENT">
1714 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1715 emulate the internals of libevent, so here are some usage hints:</p>
1717 <dt>* Use it by including <event.h>, as usual.</dt>
1718 <dt>* The following members are fully supported: ev_base, ev_callback,
1719 ev_arg, ev_fd, ev_res, ev_events.</dt>
1720 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1721 maintained by libev, it does not work exactly the same way as in libevent (consider
1722 it a private API).</dt>
1723 <dt>* Priorities are not currently supported. Initialising priorities
1724 will fail and all watchers will have the same priority, even though there
1725 is an ev_pri field.</dt>
1726 <dt>* Other members are not supported.</dt>
1727 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1728 to use the libev header file and library.</dt>
1732 <h1 id="C_SUPPORT">C++ SUPPORT</h1>
1733 <div id="C_SUPPORT_CONTENT">
1734 <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1735 you to use some convinience methods to start/stop watchers and also change
1736 the callback model to a model using method callbacks on objects.</p>
1738 <pre> #include <ev++.h>
1741 <p>(it is not installed by default). This automatically includes <cite>ev.h</cite>
1742 and puts all of its definitions (many of them macros) into the global
1743 namespace. All C++ specific things are put into the <code>ev</code> namespace.</p>
1744 <p>It should support all the same embedding options as <cite>ev.h</cite>, most notably
1745 <code>EV_MULTIPLICITY</code>.</p>
1746 <p>Here is a list of things available in the <code>ev</code> namespace:</p>
1748 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1750 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1751 macros from <cite>ev.h</cite>.</p>
1753 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1755 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1757 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1759 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1760 the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1761 which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1762 defines by many implementations.</p>
1763 <p>All of those classes have these methods:</p>
1766 <dt>ev::TYPE::TYPE (object *, object::method *)</dt>
1767 <dt>ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)</dt>
1768 <dt>ev::TYPE::~TYPE</dt>
1770 <p>The constructor takes a pointer to an object and a method pointer to
1771 the event handler callback to call in this class. The constructor calls
1772 <code>ev_init</code> for you, which means you have to call the <code>set</code> method
1773 before starting it. If you do not specify a loop then the constructor
1774 automatically associates the default loop with this watcher.</p>
1775 <p>The destructor automatically stops the watcher if it is active.</p>
1777 <dt>w->set (struct ev_loop *)</dt>
1779 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1780 do this when the watcher is inactive (and not pending either).</p>
1782 <dt>w->set ([args])</dt>
1784 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1785 called at least once. Unlike the C counterpart, an active watcher gets
1786 automatically stopped and restarted.</p>
1788 <dt>w->start ()</dt>
1790 <p>Starts the watcher. Note that there is no <code>loop</code> argument as the
1791 constructor already takes the loop.</p>
1793 <dt>w->stop ()</dt>
1795 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1797 <dt>w->again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1799 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1800 <code>ev_TYPE_again</code> function.</p>
1802 <dt>w->sweep () <code>ev::embed</code> only</dt>
1804 <p>Invokes <code>ev_embed_sweep</code>.</p>
1806 <dt>w->update () <code>ev::stat</code> only</dt>
1808 <p>Invokes <code>ev_stat_stat</code>.</p>
1814 <p>Example: Define a class with an IO and idle watcher, start one of them in
1815 the constructor.</p>
1818 ev_io io; void io_cb (ev::io &w, int revents);
1819 ev_idle idle void idle_cb (ev::idle &w, int revents);
1824 myclass::myclass (int fd)
1825 : io (this, &myclass::io_cb),
1826 idle (this, &myclass::idle_cb)
1828 io.start (fd, ev::READ);
1837 <h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
1838 <div id="MACRO_MAGIC_CONTENT">
1839 <p>Libev can be compiled with a variety of options, the most fundemantal is
1840 <code>EV_MULTIPLICITY</code>. This option determines whether (most) functions and
1841 callbacks have an initial <code>struct ev_loop *</code> argument.</p>
1842 <p>To make it easier to write programs that cope with either variant, the
1843 following macros are defined:</p>
1845 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
1847 <p>This provides the loop <i>argument</i> for functions, if one is required ("ev
1848 loop argument"). The <code>EV_A</code> form is used when this is the sole argument,
1849 <code>EV_A_</code> is used when other arguments are following. Example:</p>
1850 <pre> ev_unref (EV_A);
1851 ev_timer_add (EV_A_ watcher);
1855 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
1856 which is often provided by the following macro.</p>
1858 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
1860 <p>This provides the loop <i>parameter</i> for functions, if one is required ("ev
1861 loop parameter"). The <code>EV_P</code> form is used when this is the sole parameter,
1862 <code>EV_P_</code> is used when other parameters are following. Example:</p>
1863 <pre> // this is how ev_unref is being declared
1864 static void ev_unref (EV_P);
1866 // this is how you can declare your typical callback
1867 static void cb (EV_P_ ev_timer *w, int revents)
1870 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
1871 suitable for use with <code>EV_A</code>.</p>
1873 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
1875 <p>Similar to the other two macros, this gives you the value of the default
1876 loop, if multiple loops are supported ("ev loop default").</p>
1879 <p>Example: Declare and initialise a check watcher, utilising the above
1880 macros so it will work regardless of whether multiple loops are supported
1883 check_cb (EV_P_ ev_timer *w, int revents)
1885 ev_check_stop (EV_A_ w);
1889 ev_check_init (&check, check_cb);
1890 ev_check_start (EV_DEFAULT_ &check);
1891 ev_loop (EV_DEFAULT_ 0);
1896 <h1 id="EMBEDDING">EMBEDDING</h1>
1897 <div id="EMBEDDING_CONTENT">
1898 <p>Libev can (and often is) directly embedded into host
1899 applications. Examples of applications that embed it include the Deliantra
1900 Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1901 and rxvt-unicode.</p>
1902 <p>The goal is to enable you to just copy the neecssary files into your
1903 source directory without having to change even a single line in them, so
1904 you can easily upgrade by simply copying (or having a checked-out copy of
1905 libev somewhere in your source tree).</p>
1908 <h2 id="FILESETS">FILESETS</h2>
1909 <div id="FILESETS_CONTENT">
1910 <p>Depending on what features you need you need to include one or more sets of files
1914 <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
1915 <div id="CORE_EVENT_LOOP_CONTENT">
1916 <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
1917 configuration (no autoconf):</p>
1918 <pre> #define EV_STANDALONE 1
1919 #include "ev.c"
1922 <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
1923 single C source file only to provide the function implementations. To use
1924 it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
1925 done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
1926 where you can put other configuration options):</p>
1927 <pre> #define EV_STANDALONE 1
1928 #include "ev.h"
1931 <p>Both header files and implementation files can be compiled with a C++
1932 compiler (at least, thats a stated goal, and breakage will be treated
1934 <p>You need the following files in your source tree, or in a directory
1935 in your include path (e.g. in libev/ when using -Ilibev):</p>
1941 ev_win32.c required on win32 platforms only
1943 ev_select.c only when select backend is enabled (which is enabled by default)
1944 ev_poll.c only when poll backend is enabled (disabled by default)
1945 ev_epoll.c only when the epoll backend is enabled (disabled by default)
1946 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
1947 ev_port.c only when the solaris port backend is enabled (disabled by default)
1950 <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
1951 to compile this single file.</p>
1954 <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
1955 <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
1956 <p>To include the libevent compatibility API, also include:</p>
1957 <pre> #include "event.c"
1960 <p>in the file including <cite>ev.c</cite>, and:</p>
1961 <pre> #include "event.h"
1964 <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
1965 <p>You need the following additional files for this:</p>
1972 <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
1973 <div id="AUTOCONF_SUPPORT_CONTENT">
1974 <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
1975 whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
1976 <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
1977 include <cite>config.h</cite> and configure itself accordingly.</p>
1978 <p>For this of course you need the m4 file:</p>
1984 <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
1985 <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
1986 <p>Libev can be configured via a variety of preprocessor symbols you have to define
1987 before including any of its files. The default is not to build for multiplicity
1988 and only include the select backend.</p>
1990 <dt>EV_STANDALONE</dt>
1992 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
1993 keeps libev from including <cite>config.h</cite>, and it also defines dummy
1994 implementations for some libevent functions (such as logging, which is not
1995 supported). It will also not define any of the structs usually found in
1996 <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
1998 <dt>EV_USE_MONOTONIC</dt>
2000 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2001 monotonic clock option at both compiletime and runtime. Otherwise no use
2002 of the monotonic clock option will be attempted. If you enable this, you
2003 usually have to link against librt or something similar. Enabling it when
2004 the functionality isn't available is safe, though, althoguh you have
2005 to make sure you link against any libraries where the <code>clock_gettime</code>
2006 function is hiding in (often <cite>-lrt</cite>).</p>
2008 <dt>EV_USE_REALTIME</dt>
2010 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2011 realtime clock option at compiletime (and assume its availability at
2012 runtime if successful). Otherwise no use of the realtime clock option will
2013 be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
2014 (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
2015 in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
2017 <dt>EV_USE_SELECT</dt>
2019 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
2020 <code>select</code>(2) backend. No attempt at autodetection will be done: if no
2021 other method takes over, select will be it. Otherwise the select backend
2022 will not be compiled in.</p>
2024 <dt>EV_SELECT_USE_FD_SET</dt>
2026 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
2027 structure. This is useful if libev doesn't compile due to a missing
2028 <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
2029 exotic systems. This usually limits the range of file descriptors to some
2030 low limit such as 1024 or might have other limitations (winsocket only
2031 allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
2032 influence the size of the <code>fd_set</code> used.</p>
2034 <dt>EV_SELECT_IS_WINSOCKET</dt>
2036 <p>When defined to <code>1</code>, the select backend will assume that
2037 select/socket/connect etc. don't understand file descriptors but
2038 wants osf handles on win32 (this is the case when the select to
2039 be used is the winsock select). This means that it will call
2040 <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2041 it is assumed that all these functions actually work on fds, even
2042 on win32. Should not be defined on non-win32 platforms.</p>
2044 <dt>EV_USE_POLL</dt>
2046 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2047 backend. Otherwise it will be enabled on non-win32 platforms. It
2048 takes precedence over select.</p>
2050 <dt>EV_USE_EPOLL</dt>
2052 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2053 <code>epoll</code>(7) backend. Its availability will be detected at runtime,
2054 otherwise another method will be used as fallback. This is the
2055 preferred backend for GNU/Linux systems.</p>
2057 <dt>EV_USE_KQUEUE</dt>
2059 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2060 <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2061 otherwise another method will be used as fallback. This is the preferred
2062 backend for BSD and BSD-like systems, although on most BSDs kqueue only
2063 supports some types of fds correctly (the only platform we found that
2064 supports ptys for example was NetBSD), so kqueue might be compiled in, but
2065 not be used unless explicitly requested. The best way to use it is to find
2066 out whether kqueue supports your type of fd properly and use an embedded
2069 <dt>EV_USE_PORT</dt>
2071 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
2072 10 port style backend. Its availability will be detected at runtime,
2073 otherwise another method will be used as fallback. This is the preferred
2074 backend for Solaris 10 systems.</p>
2076 <dt>EV_USE_DEVPOLL</dt>
2078 <p>reserved for future expansion, works like the USE symbols above.</p>
2080 <dt>EV_USE_INOTIFY</dt>
2082 <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2083 interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2084 be detected at runtime.</p>
2088 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2089 undefined is <code><ev.h></code> in <cite>event.h</cite> and <code>"ev.h"</code> in <cite>ev.c</cite>. This
2090 can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2092 <dt>EV_CONFIG_H</dt>
2094 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2095 <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2096 <code>EV_H</code>, above.</p>
2100 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2101 of how the <cite>event.h</cite> header can be found.</p>
2103 <dt>EV_PROTOTYPES</dt>
2105 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2106 prototypes, but still define all the structs and other symbols. This is
2107 occasionally useful if you want to provide your own wrapper functions
2108 around libev functions.</p>
2110 <dt>EV_MULTIPLICITY</dt>
2112 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2113 will have the <code>struct ev_loop *</code> as first argument, and you can create
2114 additional independent event loops. Otherwise there will be no support
2115 for multiple event loops and there is no first event loop pointer
2116 argument. Instead, all functions act on the single default loop.</p>
2121 <p>The range of allowed priorities. <code>EV_MINPRI</code> must be smaller or equal to
2122 <code>EV_MAXPRI</code>, but otherwise there are no non-obvious limitations. You can
2123 provide for more priorities by overriding those symbols (usually defined
2124 to be <code>-2</code> and <code>2</code>, respectively).</p>
2125 <p>When doing priority-based operations, libev usually has to linearly search
2126 all the priorities, so having many of them (hundreds) uses a lot of space
2127 and time, so using the defaults of five priorities (-2 .. +2) is usually
2129 <p>If your embedding app does not need any priorities, defining these both to
2130 <code>0</code> will save some memory and cpu.</p>
2132 <dt>EV_PERIODIC_ENABLE</dt>
2134 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2135 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2138 <dt>EV_IDLE_ENABLE</dt>
2140 <p>If undefined or defined to be <code>1</code>, then idle watchers are supported. If
2141 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2144 <dt>EV_EMBED_ENABLE</dt>
2146 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2147 defined to be <code>0</code>, then they are not.</p>
2149 <dt>EV_STAT_ENABLE</dt>
2151 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2152 defined to be <code>0</code>, then they are not.</p>
2154 <dt>EV_FORK_ENABLE</dt>
2156 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2157 defined to be <code>0</code>, then they are not.</p>
2161 <p>If you need to shave off some kilobytes of code at the expense of some
2162 speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2163 some inlining decisions, saves roughly 30% codesize of amd64.</p>
2165 <dt>EV_PID_HASHSIZE</dt>
2167 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2168 pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2169 than enough. If you need to manage thousands of children you might want to
2170 increase this value (<i>must</i> be a power of two).</p>
2172 <dt>EV_INOTIFY_HASHSIZE</dt>
2174 <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2175 inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2176 usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2177 watchers you might want to increase this value (<i>must</i> be a power of
2182 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2183 this macro to a something else you can include more and other types of
2184 members. You have to define it each time you include one of the files,
2185 though, and it must be identical each time.</p>
2186 <p>For example, the perl EV module uses something like this:</p>
2187 <pre> #define EV_COMMON \
2188 SV *self; /* contains this struct */ \
2189 SV *cb_sv, *fh /* note no trailing ";" */
2193 <dt>EV_CB_DECLARE (type)</dt>
2194 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2195 <dt>ev_set_cb (ev, cb)</dt>
2197 <p>Can be used to change the callback member declaration in each watcher,
2198 and the way callbacks are invoked and set. Must expand to a struct member
2199 definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2200 their default definitions. One possible use for overriding these is to
2201 avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2202 method calls instead of plain function calls in C++.</p>
2205 <h2 id="EXAMPLES">EXAMPLES</h2>
2206 <div id="EXAMPLES_CONTENT">
2207 <p>For a real-world example of a program the includes libev
2208 verbatim, you can have a look at the EV perl module
2209 (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2210 the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2211 interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2212 will be compiled. It is pretty complex because it provides its own header
2214 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2215 that everybody includes and which overrides some configure choices:</p>
2216 <pre> #define EV_MINIMAL 1
2217 #define EV_USE_POLL 0
2218 #define EV_MULTIPLICITY 0
2219 #define EV_PERIODIC_ENABLE 0
2220 #define EV_STAT_ENABLE 0
2221 #define EV_FORK_ENABLE 0
2222 #define EV_CONFIG_H <config.h>
2226 #include "ev++.h"
2229 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2230 <pre> #include "ev_cpp.h"
2231 #include "ev.c"
2239 <h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2240 <div id="COMPLEXITIES_CONTENT">
2241 <p>In this section the complexities of (many of) the algorithms used inside
2242 libev will be explained. For complexity discussions about backends see the
2243 documentation for <code>ev_default_init</code>.</p>
2244 <p>All of the following are about amortised time: If an array needs to be
2245 extended, libev needs to realloc and move the whole array, but this
2246 happens asymptotically never with higher number of elements, so O(1) might
2247 mean it might do a lengthy realloc operation in rare cases, but on average
2248 it is much faster and asymptotically approaches constant time.</p>
2251 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2253 <p>This means that, when you have a watcher that triggers in one hour and
2254 there are 100 watchers that would trigger before that then inserting will
2255 have to skip those 100 watchers.</p>
2257 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2259 <p>That means that for changing a timer costs less than removing/adding them
2260 as only the relative motion in the event queue has to be paid for.</p>
2262 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2264 <p>These just add the watcher into an array or at the head of a list.
2265 =item Stopping check/prepare/idle watchers: O(1)</p>
2267 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2269 <p>These watchers are stored in lists then need to be walked to find the
2270 correct watcher to remove. The lists are usually short (you don't usually
2271 have many watchers waiting for the same fd or signal).</p>
2273 <dt>Finding the next timer per loop iteration: O(1)</dt>
2274 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2276 <p>A change means an I/O watcher gets started or stopped, which requires
2277 libev to recalculate its status (and possibly tell the kernel).</p>
2279 <dt>Activating one watcher: O(1)</dt>
2280 <dt>Priority handling: O(number_of_priorities)</dt>
2282 <p>Priorities are implemented by allocating some space for each
2283 priority. When doing priority-based operations, libev usually has to
2284 linearly search all the priorities.</p>
2294 <h1 id="AUTHOR">AUTHOR</h1>
2295 <div id="AUTHOR_CONTENT">
2296 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob