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="Wed Dec 12 05:53:55 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>
33 <ul><li><a href="#The_special_problem_of_disappearing_">The special problem of disappearing file descriptors</a></li>
36 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
37 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
38 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
39 <li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
40 <li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
41 <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>
42 <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>
43 <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
44 <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>
47 <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
48 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
49 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
50 <li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
51 <li><a href="#EMBEDDING">EMBEDDING</a>
52 <ul><li><a href="#FILESETS">FILESETS</a>
53 <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
54 <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
55 <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
58 <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
59 <li><a href="#EXAMPLES">EXAMPLES</a></li>
62 <li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
63 <li><a href="#AUTHOR">AUTHOR</a>
68 <h1 id="NAME">NAME</h1>
69 <div id="NAME_CONTENT">
70 <p>libev - a high performance full-featured event loop written in C</p>
73 <h1 id="SYNOPSIS">SYNOPSIS</h1>
74 <div id="SYNOPSIS_CONTENT">
75 <pre> #include <ev.h>
80 <h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
81 <div id="EXAMPLE_PROGRAM_CONTENT">
82 <pre> #include <ev.h>
85 ev_timer timeout_watcher;
87 /* called when data readable on stdin */
89 stdin_cb (EV_P_ struct ev_io *w, int revents)
91 /* puts ("stdin ready"); */
92 ev_io_stop (EV_A_ w); /* just a syntax example */
93 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
97 timeout_cb (EV_P_ struct ev_timer *w, int revents)
99 /* puts ("timeout"); */
100 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
106 struct ev_loop *loop = ev_default_loop (0);
108 /* initialise an io watcher, then start it */
109 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
110 ev_io_start (loop, &stdin_watcher);
112 /* simple non-repeating 5.5 second timeout */
113 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
114 ev_timer_start (loop, &timeout_watcher);
116 /* loop till timeout or data ready */
125 <h1 id="DESCRIPTION">DESCRIPTION</h1>
126 <div id="DESCRIPTION_CONTENT">
127 <p>The newest version of this document is also available as a html-formatted
128 web page you might find easier to navigate when reading it for the first
129 time: <a href="http://cvs.schmorp.de/libev/ev.html">http://cvs.schmorp.de/libev/ev.html</a>.</p>
130 <p>Libev is an event loop: you register interest in certain events (such as a
131 file descriptor being readable or a timeout occuring), and it will manage
132 these event sources and provide your program with events.</p>
133 <p>To do this, it must take more or less complete control over your process
134 (or thread) by executing the <i>event loop</i> handler, and will then
135 communicate events via a callback mechanism.</p>
136 <p>You register interest in certain events by registering so-called <i>event
137 watchers</i>, which are relatively small C structures you initialise with the
138 details of the event, and then hand it over to libev by <i>starting</i> the
142 <h1 id="FEATURES">FEATURES</h1>
143 <div id="FEATURES_CONTENT">
144 <p>Libev supports <code>select</code>, <code>poll</code>, the Linux-specific <code>epoll</code>, the
145 BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
146 for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
147 (for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
148 with customised rescheduling (<code>ev_periodic</code>), synchronous signals
149 (<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
150 watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
151 <code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
152 file watchers (<code>ev_stat</code>) and even limited support for fork events
153 (<code>ev_fork</code>).</p>
154 <p>It also is quite fast (see this
155 <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
159 <h1 id="CONVENTIONS">CONVENTIONS</h1>
160 <div id="CONVENTIONS_CONTENT">
161 <p>Libev is very configurable. In this manual the default configuration will
162 be described, which supports multiple event loops. For more info about
163 various configuration options please have a look at <strong>EMBED</strong> section in
164 this manual. If libev was configured without support for multiple event
165 loops, then all functions taking an initial argument of name <code>loop</code>
166 (which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
169 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
170 <div id="TIME_REPRESENTATION_CONTENT">
171 <p>Libev represents time as a single floating point number, representing the
172 (fractional) number of seconds since the (POSIX) epoch (somewhere near
173 the beginning of 1970, details are complicated, don't ask). This type is
174 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
175 to the <code>double</code> type in C, and when you need to do any calculations on
176 it, you should treat it as such.</p>
179 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
180 <div id="GLOBAL_FUNCTIONS_CONTENT">
181 <p>These functions can be called anytime, even before initialising the
182 library in any way.</p>
184 <dt>ev_tstamp ev_time ()</dt>
186 <p>Returns the current time as libev would use it. Please note that the
187 <code>ev_now</code> function is usually faster and also often returns the timestamp
188 you actually want to know.</p>
190 <dt>int ev_version_major ()</dt>
191 <dt>int ev_version_minor ()</dt>
193 <p>You can find out the major and minor ABI version numbers of the library
194 you linked against by calling the functions <code>ev_version_major</code> and
195 <code>ev_version_minor</code>. If you want, you can compare against the global
196 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
197 version of the library your program was compiled against.</p>
198 <p>These version numbers refer to the ABI version of the library, not the
200 <p>Usually, it's a good idea to terminate if the major versions mismatch,
201 as this indicates an incompatible change. Minor versions are usually
202 compatible to older versions, so a larger minor version alone is usually
204 <p>Example: Make sure we haven't accidentally been linked against the wrong
206 <pre> assert (("libev version mismatch",
207 ev_version_major () == EV_VERSION_MAJOR
208 && ev_version_minor () >= EV_VERSION_MINOR));
212 <dt>unsigned int ev_supported_backends ()</dt>
214 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
215 value) compiled into this binary of libev (independent of their
216 availability on the system you are running on). See <code>ev_default_loop</code> for
217 a description of the set values.</p>
218 <p>Example: make sure we have the epoll method, because yeah this is cool and
219 a must have and can we have a torrent of it please!!!11</p>
220 <pre> assert (("sorry, no epoll, no sex",
221 ev_supported_backends () & EVBACKEND_EPOLL));
225 <dt>unsigned int ev_recommended_backends ()</dt>
227 <p>Return the set of all backends compiled into this binary of libev and also
228 recommended for this platform. This set is often smaller than the one
229 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
230 most BSDs and will not be autodetected unless you explicitly request it
231 (assuming you know what you are doing). This is the set of backends that
232 libev will probe for if you specify no backends explicitly.</p>
234 <dt>unsigned int ev_embeddable_backends ()</dt>
236 <p>Returns the set of backends that are embeddable in other event loops. This
237 is the theoretical, all-platform, value. To find which backends
238 might be supported on the current system, you would need to look at
239 <code>ev_embeddable_backends () & ev_supported_backends ()</code>, likewise for
240 recommended ones.</p>
241 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
243 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
245 <p>Sets the allocation function to use (the prototype is similar - the
246 semantics is identical - to the realloc C function). It is used to
247 allocate and free memory (no surprises here). If it returns zero when
248 memory needs to be allocated, the library might abort or take some
249 potentially destructive action. The default is your system realloc
251 <p>You could override this function in high-availability programs to, say,
252 free some memory if it cannot allocate memory, to use a special allocator,
253 or even to sleep a while and retry until some memory is available.</p>
254 <p>Example: Replace the libev allocator with one that waits a bit and then
257 persistent_realloc (void *ptr, size_t size)
261 void *newptr = realloc (ptr, size);
271 ev_set_allocator (persistent_realloc);
275 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
277 <p>Set the callback function to call on a retryable syscall error (such
278 as failed select, poll, epoll_wait). The message is a printable string
279 indicating the system call or subsystem causing the problem. If this
280 callback is set, then libev will expect it to remedy the sitution, no
281 matter what, when it returns. That is, libev will generally retry the
282 requested operation, or, if the condition doesn't go away, do bad stuff
284 <p>Example: This is basically the same thing that libev does internally, too.</p>
286 fatal_error (const char *msg)
293 ev_set_syserr_cb (fatal_error);
300 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
301 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
302 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
303 types of such loops, the <i>default</i> loop, which supports signals and child
304 events, and dynamically created loops which do not.</p>
305 <p>If you use threads, a common model is to run the default event loop
306 in your main thread (or in a separate thread) and for each thread you
307 create, you also create another event loop. Libev itself does no locking
308 whatsoever, so if you mix calls to the same event loop in different
309 threads, make sure you lock (this is usually a bad idea, though, even if
310 done correctly, because it's hideous and inefficient).</p>
312 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
314 <p>This will initialise the default event loop if it hasn't been initialised
315 yet and return it. If the default loop could not be initialised, returns
316 false. If it already was initialised it simply returns it (and ignores the
317 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
318 <p>If you don't know what event loop to use, use the one returned from this
320 <p>The flags argument can be used to specify special behaviour or specific
321 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
322 <p>The following flags are supported:</p>
325 <dt><code>EVFLAG_AUTO</code></dt>
327 <p>The default flags value. Use this if you have no clue (it's the right
328 thing, believe me).</p>
330 <dt><code>EVFLAG_NOENV</code></dt>
332 <p>If this flag bit is ored into the flag value (or the program runs setuid
333 or setgid) then libev will <i>not</i> look at the environment variable
334 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
335 override the flags completely if it is found in the environment. This is
336 useful to try out specific backends to test their performance, or to work
339 <dt><code>EVFLAG_FORKCHECK</code></dt>
341 <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
342 a fork, you can also make libev check for a fork in each iteration by
343 enabling this flag.</p>
344 <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
345 and thus this might slow down your event loop if you do a lot of loop
346 iterations and little real work, but is usually not noticeable (on my
347 Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
348 without a syscall and thus <i>very</i> fast, but my Linux system also has
349 <code>pthread_atfork</code> which is even faster).</p>
350 <p>The big advantage of this flag is that you can forget about fork (and
351 forget about forgetting to tell libev about forking) when you use this
353 <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
354 environment variable.</p>
356 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
358 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
359 libev tries to roll its own fd_set with no limits on the number of fds,
360 but if that fails, expect a fairly low limit on the number of fds when
361 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
362 the fastest backend for a low number of fds.</p>
364 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
366 <p>And this is your standard poll(2) backend. It's more complicated than
367 select, but handles sparse fds better and has no artificial limit on the
368 number of fds you can use (except it will slow down considerably with a
369 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
371 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
373 <p>For few fds, this backend is a bit little slower than poll and select,
374 but it scales phenomenally better. While poll and select usually scale like
375 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
376 either O(1) or O(active_fds).</p>
377 <p>While stopping and starting an I/O watcher in the same iteration will
378 result in some caching, there is still a syscall per such incident
379 (because the fd could point to a different file description now), so its
380 best to avoid that. Also, dup()ed file descriptors might not work very
381 well if you register events for both fds.</p>
382 <p>Please note that epoll sometimes generates spurious notifications, so you
383 need to use non-blocking I/O or other means to avoid blocking when no data
384 (or space) is available.</p>
386 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
388 <p>Kqueue deserves special mention, as at the time of this writing, it
389 was broken on all BSDs except NetBSD (usually it doesn't work with
390 anything but sockets and pipes, except on Darwin, where of course its
391 completely useless). For this reason its not being "autodetected"
392 unless you explicitly specify it explicitly in the flags (i.e. using
393 <code>EVBACKEND_KQUEUE</code>).</p>
394 <p>It scales in the same way as the epoll backend, but the interface to the
395 kernel is more efficient (which says nothing about its actual speed, of
396 course). While starting and stopping an I/O watcher does not cause an
397 extra syscall as with epoll, it still adds up to four event changes per
398 incident, so its best to avoid that.</p>
400 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
402 <p>This is not implemented yet (and might never be).</p>
404 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
406 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
407 it's really slow, but it still scales very well (O(active_fds)).</p>
408 <p>Please note that solaris ports can result in a lot of spurious
409 notifications, so you need to use non-blocking I/O or other means to avoid
410 blocking when no data (or space) is available.</p>
412 <dt><code>EVBACKEND_ALL</code></dt>
414 <p>Try all backends (even potentially broken ones that wouldn't be tried
415 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
416 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
420 <p>If one or more of these are ored into the flags value, then only these
421 backends will be tried (in the reverse order as given here). If none are
422 specified, most compiled-in backend will be tried, usually in reverse
423 order of their flag values :)</p>
424 <p>The most typical usage is like this:</p>
425 <pre> if (!ev_default_loop (0))
426 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
429 <p>Restrict libev to the select and poll backends, and do not allow
430 environment settings to be taken into account:</p>
431 <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
434 <p>Use whatever libev has to offer, but make sure that kqueue is used if
435 available (warning, breaks stuff, best use only with your own private
436 event loop and only if you know the OS supports your types of fds):</p>
437 <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
441 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
443 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
444 always distinct from the default loop. Unlike the default loop, it cannot
445 handle signal and child watchers, and attempts to do so will be greeted by
446 undefined behaviour (or a failed assertion if assertions are enabled).</p>
447 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
448 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
450 fatal ("no epoll found here, maybe it hides under your chair");
454 <dt>ev_default_destroy ()</dt>
456 <p>Destroys the default loop again (frees all memory and kernel state
457 etc.). None of the active event watchers will be stopped in the normal
458 sense, so e.g. <code>ev_is_active</code> might still return true. It is your
459 responsibility to either stop all watchers cleanly yoursef <i>before</i>
460 calling this function, or cope with the fact afterwards (which is usually
461 the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
464 <dt>ev_loop_destroy (loop)</dt>
466 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
467 earlier call to <code>ev_loop_new</code>.</p>
469 <dt>ev_default_fork ()</dt>
471 <p>This function reinitialises the kernel state for backends that have
472 one. Despite the name, you can call it anytime, but it makes most sense
473 after forking, in either the parent or child process (or both, but that
474 again makes little sense).</p>
475 <p>You <i>must</i> call this function in the child process after forking if and
476 only if you want to use the event library in both processes. If you just
477 fork+exec, you don't have to call it.</p>
478 <p>The function itself is quite fast and it's usually not a problem to call
479 it just in case after a fork. To make this easy, the function will fit in
480 quite nicely into a call to <code>pthread_atfork</code>:</p>
481 <pre> pthread_atfork (0, 0, ev_default_fork);
484 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
485 without calling this function, so if you force one of those backends you
486 do not need to care.</p>
488 <dt>ev_loop_fork (loop)</dt>
490 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
491 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
492 after fork, and how you do this is entirely your own problem.</p>
494 <dt>unsigned int ev_loop_count (loop)</dt>
496 <p>Returns the count of loop iterations for the loop, which is identical to
497 the number of times libev did poll for new events. It starts at <code>0</code> and
498 happily wraps around with enough iterations.</p>
499 <p>This value can sometimes be useful as a generation counter of sorts (it
500 "ticks" the number of loop iterations), as it roughly corresponds with
501 <code>ev_prepare</code> and <code>ev_check</code> calls.</p>
503 <dt>unsigned int ev_backend (loop)</dt>
505 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
508 <dt>ev_tstamp ev_now (loop)</dt>
510 <p>Returns the current "event loop time", which is the time the event loop
511 received events and started processing them. This timestamp does not
512 change as long as callbacks are being processed, and this is also the base
513 time used for relative timers. You can treat it as the timestamp of the
514 event occuring (or more correctly, libev finding out about it).</p>
516 <dt>ev_loop (loop, int flags)</dt>
518 <p>Finally, this is it, the event handler. This function usually is called
519 after you initialised all your watchers and you want to start handling
521 <p>If the flags argument is specified as <code>0</code>, it will not return until
522 either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
523 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
524 relying on all watchers to be stopped when deciding when a program has
525 finished (especially in interactive programs), but having a program that
526 automatically loops as long as it has to and no longer by virtue of
527 relying on its watchers stopping correctly is a thing of beauty.</p>
528 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
529 those events and any outstanding ones, but will not block your process in
530 case there are no events and will return after one iteration of the loop.</p>
531 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
532 neccessary) and will handle those and any outstanding ones. It will block
533 your process until at least one new event arrives, and will return after
534 one iteration of the loop. This is useful if you are waiting for some
535 external event in conjunction with something not expressible using other
536 libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
537 usually a better approach for this kind of thing.</p>
538 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
539 <pre> - Before the first iteration, call any pending watchers.
540 * If there are no active watchers (reference count is zero), return.
541 - Queue all prepare watchers and then call all outstanding watchers.
542 - If we have been forked, recreate the kernel state.
543 - Update the kernel state with all outstanding changes.
544 - Update the "event loop time".
545 - Calculate for how long to block.
546 - Block the process, waiting for any events.
547 - Queue all outstanding I/O (fd) events.
548 - Update the "event loop time" and do time jump handling.
549 - Queue all outstanding timers.
550 - Queue all outstanding periodics.
551 - If no events are pending now, queue all idle watchers.
552 - Queue all check watchers.
553 - Call all queued watchers in reverse order (i.e. check watchers first).
554 Signals and child watchers are implemented as I/O watchers, and will
555 be handled here by queueing them when their watcher gets executed.
556 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
557 were used, return, otherwise continue with step *.
560 <p>Example: Queue some jobs and then loop until no events are outsanding
562 <pre> ... queue jobs here, make sure they register event watchers as long
563 ... as they still have work to do (even an idle watcher will do..)
564 ev_loop (my_loop, 0);
569 <dt>ev_unloop (loop, how)</dt>
571 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
572 has processed all outstanding events). The <code>how</code> argument must be either
573 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
574 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
576 <dt>ev_ref (loop)</dt>
577 <dt>ev_unref (loop)</dt>
579 <p>Ref/unref can be used to add or remove a reference count on the event
580 loop: Every watcher keeps one reference, and as long as the reference
581 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
582 a watcher you never unregister that should not keep <code>ev_loop</code> from
583 returning, ev_unref() after starting, and ev_ref() before stopping it. For
584 example, libev itself uses this for its internal signal pipe: It is not
585 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
586 no event watchers registered by it are active. It is also an excellent
587 way to do this for generic recurring timers or from within third-party
588 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
589 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
590 running when nothing else is active.</p>
591 <pre> struct ev_signal exitsig;
592 ev_signal_init (&exitsig, sig_cb, SIGINT);
593 ev_signal_start (loop, &exitsig);
597 <p>Example: For some weird reason, unregister the above signal handler again.</p>
599 ev_signal_stop (loop, &exitsig);
610 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
611 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
612 <p>A watcher is a structure that you create and register to record your
613 interest in some event. For instance, if you want to wait for STDIN to
614 become readable, you would create an <code>ev_io</code> watcher for that:</p>
615 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
618 ev_unloop (loop, EVUNLOOP_ALL);
621 struct ev_loop *loop = ev_default_loop (0);
622 struct ev_io stdin_watcher;
623 ev_init (&stdin_watcher, my_cb);
624 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
625 ev_io_start (loop, &stdin_watcher);
629 <p>As you can see, you are responsible for allocating the memory for your
630 watcher structures (and it is usually a bad idea to do this on the stack,
631 although this can sometimes be quite valid).</p>
632 <p>Each watcher structure must be initialised by a call to <code>ev_init
633 (watcher *, callback)</code>, which expects a callback to be provided. This
634 callback gets invoked each time the event occurs (or, in the case of io
635 watchers, each time the event loop detects that the file descriptor given
636 is readable and/or writable).</p>
637 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
638 with arguments specific to this watcher type. There is also a macro
639 to combine initialisation and setting in one call: <code>ev_<type>_init
640 (watcher *, callback, ...)</code>.</p>
641 <p>To make the watcher actually watch out for events, you have to start it
642 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
643 *)</code>), and you can stop watching for events at any time by calling the
644 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
645 <p>As long as your watcher is active (has been started but not stopped) you
646 must not touch the values stored in it. Most specifically you must never
647 reinitialise it or call its <code>set</code> macro.</p>
648 <p>Each and every callback receives the event loop pointer as first, the
649 registered watcher structure as second, and a bitset of received events as
651 <p>The received events usually include a single bit per event type received
652 (you can receive multiple events at the same time). The possible bit masks
655 <dt><code>EV_READ</code></dt>
656 <dt><code>EV_WRITE</code></dt>
658 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
661 <dt><code>EV_TIMEOUT</code></dt>
663 <p>The <code>ev_timer</code> watcher has timed out.</p>
665 <dt><code>EV_PERIODIC</code></dt>
667 <p>The <code>ev_periodic</code> watcher has timed out.</p>
669 <dt><code>EV_SIGNAL</code></dt>
671 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
673 <dt><code>EV_CHILD</code></dt>
675 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
677 <dt><code>EV_STAT</code></dt>
679 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
681 <dt><code>EV_IDLE</code></dt>
683 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
685 <dt><code>EV_PREPARE</code></dt>
686 <dt><code>EV_CHECK</code></dt>
688 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
689 to gather new events, and all <code>ev_check</code> watchers are invoked just after
690 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
691 received events. Callbacks of both watcher types can start and stop as
692 many watchers as they want, and all of them will be taken into account
693 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
694 <code>ev_loop</code> from blocking).</p>
696 <dt><code>EV_EMBED</code></dt>
698 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
700 <dt><code>EV_FORK</code></dt>
702 <p>The event loop has been resumed in the child process after fork (see
703 <code>ev_fork</code>).</p>
705 <dt><code>EV_ERROR</code></dt>
707 <p>An unspecified error has occured, the watcher has been stopped. This might
708 happen because the watcher could not be properly started because libev
709 ran out of memory, a file descriptor was found to be closed or any other
710 problem. You best act on it by reporting the problem and somehow coping
711 with the watcher being stopped.</p>
712 <p>Libev will usually signal a few "dummy" events together with an error,
713 for example it might indicate that a fd is readable or writable, and if
714 your callbacks is well-written it can just attempt the operation and cope
715 with the error from read() or write(). This will not work in multithreaded
716 programs, though, so beware.</p>
721 <h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
722 <div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
723 <p>In the following description, <code>TYPE</code> stands for the watcher type,
724 e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
726 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
728 <p>This macro initialises the generic portion of a watcher. The contents
729 of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
730 the generic parts of the watcher are initialised, you <i>need</i> to call
731 the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
732 type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
733 which rolls both calls into one.</p>
734 <p>You can reinitialise a watcher at any time as long as it has been stopped
735 (or never started) and there are no pending events outstanding.</p>
736 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
737 int revents)</code>.</p>
739 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
741 <p>This macro initialises the type-specific parts of a watcher. You need to
742 call <code>ev_init</code> at least once before you call this macro, but you can
743 call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
744 macro on a watcher that is active (it can be pending, however, which is a
745 difference to the <code>ev_init</code> macro).</p>
746 <p>Although some watcher types do not have type-specific arguments
747 (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
749 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
751 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
752 calls into a single call. This is the most convinient method to initialise
753 a watcher. The same limitations apply, of course.</p>
755 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
757 <p>Starts (activates) the given watcher. Only active watchers will receive
758 events. If the watcher is already active nothing will happen.</p>
760 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
762 <p>Stops the given watcher again (if active) and clears the pending
763 status. It is possible that stopped watchers are pending (for example,
764 non-repeating timers are being stopped when they become pending), but
765 <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
766 you want to free or reuse the memory used by the watcher it is therefore a
767 good idea to always call its <code>ev_TYPE_stop</code> function.</p>
769 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
771 <p>Returns a true value iff the watcher is active (i.e. it has been started
772 and not yet been stopped). As long as a watcher is active you must not modify
775 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
777 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
778 events but its callback has not yet been invoked). As long as a watcher
779 is pending (but not active) you must not call an init function on it (but
780 <code>ev_TYPE_set</code> is safe), you must not change its priority, and you must
781 make sure the watcher is available to libev (e.g. you cannot <code>free ()</code>
784 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
786 <p>Returns the callback currently set on the watcher.</p>
788 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
790 <p>Change the callback. You can change the callback at virtually any time
791 (modulo threads).</p>
793 <dt>ev_set_priority (ev_TYPE *watcher, priority)</dt>
794 <dt>int ev_priority (ev_TYPE *watcher)</dt>
796 <p>Set and query the priority of the watcher. The priority is a small
797 integer between <code>EV_MAXPRI</code> (default: <code>2</code>) and <code>EV_MINPRI</code>
798 (default: <code>-2</code>). Pending watchers with higher priority will be invoked
799 before watchers with lower priority, but priority will not keep watchers
800 from being executed (except for <code>ev_idle</code> watchers).</p>
801 <p>This means that priorities are <i>only</i> used for ordering callback
802 invocation after new events have been received. This is useful, for
803 example, to reduce latency after idling, or more often, to bind two
804 watchers on the same event and make sure one is called first.</p>
805 <p>If you need to suppress invocation when higher priority events are pending
806 you need to look at <code>ev_idle</code> watchers, which provide this functionality.</p>
807 <p>You <i>must not</i> change the priority of a watcher as long as it is active or
809 <p>The default priority used by watchers when no priority has been set is
810 always <code>0</code>, which is supposed to not be too high and not be too low :).</p>
811 <p>Setting a priority outside the range of <code>EV_MINPRI</code> to <code>EV_MAXPRI</code> is
812 fine, as long as you do not mind that the priority value you query might
813 or might not have been adjusted to be within valid range.</p>
815 <dt>ev_invoke (loop, ev_TYPE *watcher, int revents)</dt>
817 <p>Invoke the <code>watcher</code> with the given <code>loop</code> and <code>revents</code>. Neither
818 <code>loop</code> nor <code>revents</code> need to be valid as long as the watcher callback
819 can deal with that fact.</p>
821 <dt>int ev_clear_pending (loop, ev_TYPE *watcher)</dt>
823 <p>If the watcher is pending, this function returns clears its pending status
824 and returns its <code>revents</code> bitset (as if its callback was invoked). If the
825 watcher isn't pending it does nothing and returns <code>0</code>.</p>
834 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
835 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
836 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
837 and read at any time, libev will completely ignore it. This can be used
838 to associate arbitrary data with your watcher. If you need more data and
839 don't want to allocate memory and store a pointer to it in that data
840 member, you can also "subclass" the watcher type and provide your own
847 struct whatever *mostinteresting;
851 <p>And since your callback will be called with a pointer to the watcher, you
852 can cast it back to your own type:</p>
853 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
855 struct my_io *w = (struct my_io *)w_;
860 <p>More interesting and less C-conformant ways of casting your callback type
861 instead have been omitted.</p>
862 <p>Another common scenario is having some data structure with multiple
864 <pre> struct my_biggy
872 <p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
873 you need to use <code>offsetof</code>:</p>
874 <pre> #include <stddef.h>
877 t1_cb (EV_P_ struct ev_timer *w, int revents)
879 struct my_biggy big = (struct my_biggy *
880 (((char *)w) - offsetof (struct my_biggy, t1));
884 t2_cb (EV_P_ struct ev_timer *w, int revents)
886 struct my_biggy big = (struct my_biggy *
887 (((char *)w) - offsetof (struct my_biggy, t2));
896 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
897 <div id="WATCHER_TYPES_CONTENT">
898 <p>This section describes each watcher in detail, but will not repeat
899 information given in the last section. Any initialisation/set macros,
900 functions and members specific to the watcher type are explained.</p>
901 <p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
902 while the watcher is active, you can look at the member and expect some
903 sensible content, but you must not modify it (you can modify it while the
904 watcher is stopped to your hearts content), or <i>[read-write]</i>, which
905 means you can expect it to have some sensible content while the watcher
906 is active, but you can also modify it. Modifying it may not do something
907 sensible or take immediate effect (or do anything at all), but libev will
908 not crash or malfunction in any way.</p>
915 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
916 <div id="code_ev_io_code_is_this_file_descrip-2">
917 <p>I/O watchers check whether a file descriptor is readable or writable
918 in each iteration of the event loop, or, more precisely, when reading
919 would not block the process and writing would at least be able to write
920 some data. This behaviour is called level-triggering because you keep
921 receiving events as long as the condition persists. Remember you can stop
922 the watcher if you don't want to act on the event and neither want to
923 receive future events.</p>
924 <p>In general you can register as many read and/or write event watchers per
925 fd as you want (as long as you don't confuse yourself). Setting all file
926 descriptors to non-blocking mode is also usually a good idea (but not
927 required if you know what you are doing).</p>
928 <p>You have to be careful with dup'ed file descriptors, though. Some backends
929 (the linux epoll backend is a notable example) cannot handle dup'ed file
930 descriptors correctly if you register interest in two or more fds pointing
931 to the same underlying file/socket/etc. description (that is, they share
932 the same underlying "file open").</p>
933 <p>If you must do this, then force the use of a known-to-be-good backend
934 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
935 <code>EVBACKEND_POLL</code>).</p>
936 <p>Another thing you have to watch out for is that it is quite easy to
937 receive "spurious" readyness notifications, that is your callback might
938 be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
939 because there is no data. Not only are some backends known to create a
940 lot of those (for example solaris ports), it is very easy to get into
941 this situation even with a relatively standard program structure. Thus
942 it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
943 <code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
944 <p>If you cannot run the fd in non-blocking mode (for example you should not
945 play around with an Xlib connection), then you have to seperately re-test
946 whether a file descriptor is really ready with a known-to-be good interface
947 such as poll (fortunately in our Xlib example, Xlib already does this on
948 its own, so its quite safe to use).</p>
951 <h3 id="The_special_problem_of_disappearing_">The special problem of disappearing file descriptors</h3>
952 <div id="The_special_problem_of_disappearing_-2">
953 <p>Some backends (e.g kqueue, epoll) need to be told about closing a file
954 descriptor (either by calling <code>close</code> explicitly or by any other means,
955 such as <code>dup</code>). The reason is that you register interest in some file
956 descriptor, but when it goes away, the operating system will silently drop
957 this interest. If another file descriptor with the same number then is
958 registered with libev, there is no efficient way to see that this is, in
959 fact, a different file descriptor.</p>
960 <p>To avoid having to explicitly tell libev about such cases, libev follows
961 the following policy: Each time <code>ev_io_set</code> is being called, libev
962 will assume that this is potentially a new file descriptor, otherwise
963 it is assumed that the file descriptor stays the same. That means that
964 you <i>have</i> to call <code>ev_io_set</code> (or <code>ev_io_init</code>) when you change the
965 descriptor even if the file descriptor number itself did not change.</p>
966 <p>This is how one would do it normally anyway, the important point is that
967 the libev application should not optimise around libev but should leave
968 optimisations to libev.</p>
974 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
975 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
977 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
978 rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
979 <code>EV_READ | EV_WRITE</code> to receive the given events.</p>
981 <dt>int fd [read-only]</dt>
983 <p>The file descriptor being watched.</p>
985 <dt>int events [read-only]</dt>
987 <p>The events being watched.</p>
990 <p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
991 readable, but only once. Since it is likely line-buffered, you could
992 attempt to read a whole line in the callback.</p>
994 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
996 ev_io_stop (loop, w);
997 .. read from stdin here (or from w->fd) and haqndle any I/O errors
1001 struct ev_loop *loop = ev_default_init (0);
1002 struct ev_io stdin_readable;
1003 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1004 ev_io_start (loop, &stdin_readable);
1013 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
1014 <div id="code_ev_timer_code_relative_and_opti-2">
1015 <p>Timer watchers are simple relative timers that generate an event after a
1016 given time, and optionally repeating in regular intervals after that.</p>
1017 <p>The timers are based on real time, that is, if you register an event that
1018 times out after an hour and you reset your system clock to last years
1019 time, it will still time out after (roughly) and hour. "Roughly" because
1020 detecting time jumps is hard, and some inaccuracies are unavoidable (the
1021 monotonic clock option helps a lot here).</p>
1022 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
1023 time. This is usually the right thing as this timestamp refers to the time
1024 of the event triggering whatever timeout you are modifying/starting. If
1025 you suspect event processing to be delayed and you <i>need</i> to base the timeout
1026 on the current time, use something like this to adjust for this:</p>
1027 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1030 <p>The callback is guarenteed to be invoked only when its timeout has passed,
1031 but if multiple timers become ready during the same loop iteration then
1032 order of execution is undefined.</p>
1034 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
1035 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
1037 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
1038 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
1039 timer will automatically be configured to trigger again <code>repeat</code> seconds
1040 later, again, and again, until stopped manually.</p>
1041 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
1042 configure a timer to trigger every 10 seconds, then it will trigger at
1043 exactly 10 second intervals. If, however, your program cannot keep up with
1044 the timer (because it takes longer than those 10 seconds to do stuff) the
1045 timer will not fire more than once per event loop iteration.</p>
1047 <dt>ev_timer_again (loop)</dt>
1049 <p>This will act as if the timer timed out and restart it again if it is
1050 repeating. The exact semantics are:</p>
1051 <p>If the timer is pending, its pending status is cleared.</p>
1052 <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
1053 <p>If the timer is repeating, either start it if necessary (with the
1054 <code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
1055 <p>This sounds a bit complicated, but here is a useful and typical
1056 example: Imagine you have a tcp connection and you want a so-called idle
1057 timeout, that is, you want to be called when there have been, say, 60
1058 seconds of inactivity on the socket. The easiest way to do this is to
1059 configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
1060 <code>ev_timer_again</code> each time you successfully read or write some data. If
1061 you go into an idle state where you do not expect data to travel on the
1062 socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
1063 automatically restart it if need be.</p>
1064 <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
1065 altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
1066 <pre> ev_timer_init (timer, callback, 0., 5.);
1067 ev_timer_again (loop, timer);
1069 timer->again = 17.;
1070 ev_timer_again (loop, timer);
1072 timer->again = 10.;
1073 ev_timer_again (loop, timer);
1076 <p>This is more slightly efficient then stopping/starting the timer each time
1077 you want to modify its timeout value.</p>
1079 <dt>ev_tstamp repeat [read-write]</dt>
1081 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1082 or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1083 which is also when any modifications are taken into account.</p>
1086 <p>Example: Create a timer that fires after 60 seconds.</p>
1088 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1090 .. one minute over, w is actually stopped right here
1093 struct ev_timer mytimer;
1094 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1095 ev_timer_start (loop, &mytimer);
1098 <p>Example: Create a timeout timer that times out after 10 seconds of
1101 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1103 .. ten seconds without any activity
1106 struct ev_timer mytimer;
1107 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1108 ev_timer_again (&mytimer); /* start timer */
1111 // and in some piece of code that gets executed on any "activity":
1112 // reset the timeout to start ticking again at 10 seconds
1113 ev_timer_again (&mytimer);
1121 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
1122 <div id="code_ev_periodic_code_to_cron_or_not-2">
1123 <p>Periodic watchers are also timers of a kind, but they are very versatile
1124 (and unfortunately a bit complex).</p>
1125 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
1126 but on wallclock time (absolute time). You can tell a periodic watcher
1127 to trigger "at" some specific point in time. For example, if you tell a
1128 periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
1129 + 10.</code>) and then reset your system clock to the last year, then it will
1130 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
1131 roughly 10 seconds later).</p>
1132 <p>They can also be used to implement vastly more complex timers, such as
1133 triggering an event on each midnight, local time or other, complicated,
1135 <p>As with timers, the callback is guarenteed to be invoked only when the
1136 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1137 during the same loop iteration then order of execution is undefined.</p>
1139 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
1140 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
1142 <p>Lots of arguments, lets sort it out... There are basically three modes of
1143 operation, and we will explain them from simplest to complex:</p>
1146 <dt>* absolute timer (at = time, interval = reschedule_cb = 0)</dt>
1148 <p>In this configuration the watcher triggers an event at the wallclock time
1149 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
1150 that is, if it is to be run at January 1st 2011 then it will run when the
1151 system time reaches or surpasses this time.</p>
1153 <dt>* non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)</dt>
1155 <p>In this mode the watcher will always be scheduled to time out at the next
1156 <code>at + N * interval</code> time (for some integer N, which can also be negative)
1157 and then repeat, regardless of any time jumps.</p>
1158 <p>This can be used to create timers that do not drift with respect to system
1160 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
1163 <p>This doesn't mean there will always be 3600 seconds in between triggers,
1164 but only that the the callback will be called when the system time shows a
1165 full hour (UTC), or more correctly, when the system time is evenly divisible
1167 <p>Another way to think about it (for the mathematically inclined) is that
1168 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
1169 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1170 <p>For numerical stability it is preferable that the <code>at</code> value is near
1171 <code>ev_now ()</code> (the current time), but there is no range requirement for
1174 <dt>* manual reschedule mode (at and interval ignored, reschedule_cb = callback)</dt>
1176 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
1177 ignored. Instead, each time the periodic watcher gets scheduled, the
1178 reschedule callback will be called with the watcher as first, and the
1179 current time as second argument.</p>
1180 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
1181 ever, or make any event loop modifications</i>. If you need to stop it,
1182 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1183 starting an <code>ev_prepare</code> watcher, which is legal).</p>
1184 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1185 ev_tstamp now)</code>, e.g.:</p>
1186 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1192 <p>It must return the next time to trigger, based on the passed time value
1193 (that is, the lowest time value larger than to the second argument). It
1194 will usually be called just before the callback will be triggered, but
1195 might be called at other times, too.</p>
1196 <p>NOTE: <i>This callback must always return a time that is later than the
1197 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
1198 <p>This can be used to create very complex timers, such as a timer that
1199 triggers on each midnight, local time. To do this, you would calculate the
1200 next midnight after <code>now</code> and return the timestamp value for this. How
1201 you do this is, again, up to you (but it is not trivial, which is the main
1202 reason I omitted it as an example).</p>
1207 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
1209 <p>Simply stops and restarts the periodic watcher again. This is only useful
1210 when you changed some parameters or the reschedule callback would return
1211 a different time than the last time it was called (e.g. in a crond like
1212 program when the crontabs have changed).</p>
1214 <dt>ev_tstamp offset [read-write]</dt>
1216 <p>When repeating, this contains the offset value, otherwise this is the
1217 absolute point in time (the <code>at</code> value passed to <code>ev_periodic_set</code>).</p>
1218 <p>Can be modified any time, but changes only take effect when the periodic
1219 timer fires or <code>ev_periodic_again</code> is being called.</p>
1221 <dt>ev_tstamp interval [read-write]</dt>
1223 <p>The current interval value. Can be modified any time, but changes only
1224 take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1227 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1229 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1230 switched off. Can be changed any time, but changes only take effect when
1231 the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1234 <p>Example: Call a callback every hour, or, more precisely, whenever the
1235 system clock is divisible by 3600. The callback invocation times have
1236 potentially a lot of jittering, but good long-term stability.</p>
1238 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1240 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1243 struct ev_periodic hourly_tick;
1244 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1245 ev_periodic_start (loop, &hourly_tick);
1248 <p>Example: The same as above, but use a reschedule callback to do it:</p>
1249 <pre> #include <math.h>
1252 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1254 return fmod (now, 3600.) + 3600.;
1257 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1260 <p>Example: Call a callback every hour, starting now:</p>
1261 <pre> struct ev_periodic hourly_tick;
1262 ev_periodic_init (&hourly_tick, clock_cb,
1263 fmod (ev_now (loop), 3600.), 3600., 0);
1264 ev_periodic_start (loop, &hourly_tick);
1272 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
1273 <div id="code_ev_signal_code_signal_me_when_a-2">
1274 <p>Signal watchers will trigger an event when the process receives a specific
1275 signal one or more times. Even though signals are very asynchronous, libev
1276 will try it's best to deliver signals synchronously, i.e. as part of the
1277 normal event processing, like any other event.</p>
1278 <p>You can configure as many watchers as you like per signal. Only when the
1279 first watcher gets started will libev actually register a signal watcher
1280 with the kernel (thus it coexists with your own signal handlers as long
1281 as you don't register any with libev). Similarly, when the last signal
1282 watcher for a signal is stopped libev will reset the signal handler to
1283 SIG_DFL (regardless of what it was set to before).</p>
1285 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1286 <dt>ev_signal_set (ev_signal *, int signum)</dt>
1288 <p>Configures the watcher to trigger on the given signal number (usually one
1289 of the <code>SIGxxx</code> constants).</p>
1291 <dt>int signum [read-only]</dt>
1293 <p>The signal the watcher watches out for.</p>
1302 <h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
1303 <div id="code_ev_child_code_watch_out_for_pro-2">
1304 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1305 some child status changes (most typically when a child of yours dies).</p>
1307 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1308 <dt>ev_child_set (ev_child *, int pid)</dt>
1310 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1311 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1312 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1313 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1314 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1315 process causing the status change.</p>
1317 <dt>int pid [read-only]</dt>
1319 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1321 <dt>int rpid [read-write]</dt>
1323 <p>The process id that detected a status change.</p>
1325 <dt>int rstatus [read-write]</dt>
1327 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1328 <code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
1331 <p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1333 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1335 ev_unloop (loop, EVUNLOOP_ALL);
1338 struct ev_signal signal_watcher;
1339 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1340 ev_signal_start (loop, &sigint_cb);
1348 <h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1349 <div id="code_ev_stat_code_did_the_file_attri-2">
1350 <p>This watches a filesystem path for attribute changes. That is, it calls
1351 <code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1352 compared to the last time, invoking the callback if it did.</p>
1353 <p>The path does not need to exist: changing from "path exists" to "path does
1354 not exist" is a status change like any other. The condition "path does
1355 not exist" is signified by the <code>st_nlink</code> field being zero (which is
1356 otherwise always forced to be at least one) and all the other fields of
1357 the stat buffer having unspecified contents.</p>
1358 <p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1359 relative and your working directory changes, the behaviour is undefined.</p>
1360 <p>Since there is no standard to do this, the portable implementation simply
1361 calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1362 can specify a recommended polling interval for this case. If you specify
1363 a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1364 unspecified default</i> value will be used (which you can expect to be around
1365 five seconds, although this might change dynamically). Libev will also
1366 impose a minimum interval which is currently around <code>0.1</code>, but thats
1367 usually overkill.</p>
1368 <p>This watcher type is not meant for massive numbers of stat watchers,
1369 as even with OS-supported change notifications, this can be
1370 resource-intensive.</p>
1371 <p>At the time of this writing, only the Linux inotify interface is
1372 implemented (implementing kqueue support is left as an exercise for the
1373 reader). Inotify will be used to give hints only and should not change the
1374 semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1375 to fall back to regular polling again even with inotify, but changes are
1376 usually detected immediately, and if the file exists there will be no
1379 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1380 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1382 <p>Configures the watcher to wait for status changes of the given
1383 <code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1384 be detected and should normally be specified as <code>0</code> to let libev choose
1385 a suitable value. The memory pointed to by <code>path</code> must point to the same
1386 path for as long as the watcher is active.</p>
1387 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1388 relative to the attributes at the time the watcher was started (or the
1389 last change was detected).</p>
1391 <dt>ev_stat_stat (ev_stat *)</dt>
1393 <p>Updates the stat buffer immediately with new values. If you change the
1394 watched path in your callback, you could call this fucntion to avoid
1395 detecting this change (while introducing a race condition). Can also be
1396 useful simply to find out the new values.</p>
1398 <dt>ev_statdata attr [read-only]</dt>
1400 <p>The most-recently detected attributes of the file. Although the type is of
1401 <code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1402 suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1403 was some error while <code>stat</code>ing the file.</p>
1405 <dt>ev_statdata prev [read-only]</dt>
1407 <p>The previous attributes of the file. The callback gets invoked whenever
1408 <code>prev</code> != <code>attr</code>.</p>
1410 <dt>ev_tstamp interval [read-only]</dt>
1412 <p>The specified interval.</p>
1414 <dt>const char *path [read-only]</dt>
1416 <p>The filesystem path that is being watched.</p>
1419 <p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1421 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1423 /* /etc/passwd changed in some way */
1424 if (w->attr.st_nlink)
1426 printf ("passwd current size %ld\n", (long)w->attr.st_size);
1427 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1428 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1431 /* you shalt not abuse printf for puts */
1432 puts ("wow, /etc/passwd is not there, expect problems. "
1433 "if this is windows, they already arrived\n");
1439 ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1440 ev_stat_start (loop, &passwd);
1448 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
1449 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1450 <p>Idle watchers trigger events when no other events of the same or higher
1451 priority are pending (prepare, check and other idle watchers do not
1453 <p>That is, as long as your process is busy handling sockets or timeouts
1454 (or even signals, imagine) of the same or higher priority it will not be
1455 triggered. But when your process is idle (or only lower-priority watchers
1456 are pending), the idle watchers are being called once per event loop
1457 iteration - until stopped, that is, or your process receives more events
1458 and becomes busy again with higher priority stuff.</p>
1459 <p>The most noteworthy effect is that as long as any idle watchers are
1460 active, the process will not block when waiting for new events.</p>
1461 <p>Apart from keeping your process non-blocking (which is a useful
1462 effect on its own sometimes), idle watchers are a good place to do
1463 "pseudo-background processing", or delay processing stuff to after the
1464 event loop has handled all outstanding events.</p>
1466 <dt>ev_idle_init (ev_signal *, callback)</dt>
1468 <p>Initialises and configures the idle watcher - it has no parameters of any
1469 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1473 <p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1474 callback, free it. Also, use no error checking, as usual.</p>
1476 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1479 // now do something you wanted to do when the program has
1480 // no longer asnything immediate to do.
1483 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1484 ev_idle_init (idle_watcher, idle_cb);
1485 ev_idle_start (loop, idle_cb);
1493 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
1494 <div id="code_ev_prepare_code_and_code_ev_che-2">
1495 <p>Prepare and check watchers are usually (but not always) used in tandem:
1496 prepare watchers get invoked before the process blocks and check watchers
1498 <p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1499 the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1500 watchers. Other loops than the current one are fine, however. The
1501 rationale behind this is that you do not need to check for recursion in
1502 those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1503 <code>ev_check</code> so if you have one watcher of each kind they will always be
1504 called in pairs bracketing the blocking call.</p>
1505 <p>Their main purpose is to integrate other event mechanisms into libev and
1506 their use is somewhat advanced. This could be used, for example, to track
1507 variable changes, implement your own watchers, integrate net-snmp or a
1508 coroutine library and lots more. They are also occasionally useful if
1509 you cache some data and want to flush it before blocking (for example,
1510 in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1512 <p>This is done by examining in each prepare call which file descriptors need
1513 to be watched by the other library, registering <code>ev_io</code> watchers for
1514 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1515 provide just this functionality). Then, in the check watcher you check for
1516 any events that occured (by checking the pending status of all watchers
1517 and stopping them) and call back into the library. The I/O and timer
1518 callbacks will never actually be called (but must be valid nevertheless,
1519 because you never know, you know?).</p>
1520 <p>As another example, the Perl Coro module uses these hooks to integrate
1521 coroutines into libev programs, by yielding to other active coroutines
1522 during each prepare and only letting the process block if no coroutines
1523 are ready to run (it's actually more complicated: it only runs coroutines
1524 with priority higher than or equal to the event loop and one coroutine
1525 of lower priority, but only once, using idle watchers to keep the event
1526 loop from blocking if lower-priority coroutines are active, thus mapping
1527 low-priority coroutines to idle/background tasks).</p>
1528 <p>It is recommended to give <code>ev_check</code> watchers highest (<code>EV_MAXPRI</code>)
1529 priority, to ensure that they are being run before any other watchers
1530 after the poll. Also, <code>ev_check</code> watchers (and <code>ev_prepare</code> watchers,
1531 too) should not activate ("feed") events into libev. While libev fully
1532 supports this, they will be called before other <code>ev_check</code> watchers did
1533 their job. As <code>ev_check</code> watchers are often used to embed other event
1534 loops those other event loops might be in an unusable state until their
1535 <code>ev_check</code> watcher ran (always remind yourself to coexist peacefully with
1538 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1539 <dt>ev_check_init (ev_check *, callback)</dt>
1541 <p>Initialises and configures the prepare or check watcher - they have no
1542 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1543 macros, but using them is utterly, utterly and completely pointless.</p>
1546 <p>There are a number of principal ways to embed other event loops or modules
1547 into libev. Here are some ideas on how to include libadns into libev
1548 (there is a Perl module named <code>EV::ADNS</code> that does this, which you could
1549 use for an actually working example. Another Perl module named <code>EV::Glib</code>
1550 embeds a Glib main context into libev, and finally, <code>Glib::EV</code> embeds EV
1551 into the Glib event loop).</p>
1552 <p>Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1553 and in a check watcher, destroy them and call into libadns. What follows
1554 is pseudo-code only of course. This requires you to either use a low
1555 priority for the check watcher or use <code>ev_clear_pending</code> explicitly, as
1556 the callbacks for the IO/timeout watchers might not have been called yet.</p>
1557 <pre> static ev_io iow [nfd];
1561 io_cb (ev_loop *loop, ev_io *w, int revents)
1565 // create io watchers for each fd and a timer before blocking
1567 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1569 int timeout = 3600000;
1570 struct pollfd fds [nfd];
1571 // actual code will need to loop here and realloc etc.
1572 adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1574 /* the callback is illegal, but won't be called as we stop during check */
1575 ev_timer_init (&tw, 0, timeout * 1e-3);
1576 ev_timer_start (loop, &tw);
1578 // create one ev_io per pollfd
1579 for (int i = 0; i < nfd; ++i)
1581 ev_io_init (iow + i, io_cb, fds [i].fd,
1582 ((fds [i].events & POLLIN ? EV_READ : 0)
1583 | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1585 fds [i].revents = 0;
1586 ev_io_start (loop, iow + i);
1590 // stop all watchers after blocking
1592 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1594 ev_timer_stop (loop, &tw);
1596 for (int i = 0; i < nfd; ++i)
1598 // set the relevant poll flags
1599 // could also call adns_processreadable etc. here
1600 struct pollfd *fd = fds + i;
1601 int revents = ev_clear_pending (iow + i);
1602 if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1603 if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1605 // now stop the watcher
1606 ev_io_stop (loop, iow + i);
1609 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1613 <p>Method 2: This would be just like method 1, but you run <code>adns_afterpoll</code>
1614 in the prepare watcher and would dispose of the check watcher.</p>
1615 <p>Method 3: If the module to be embedded supports explicit event
1616 notification (adns does), you can also make use of the actual watcher
1617 callbacks, and only destroy/create the watchers in the prepare watcher.</p>
1619 timer_cb (EV_P_ ev_timer *w, int revents)
1621 adns_state ads = (adns_state)w->data;
1624 adns_processtimeouts (ads, &tv_now);
1628 io_cb (EV_P_ ev_io *w, int revents)
1630 adns_state ads = (adns_state)w->data;
1633 if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1634 if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1637 // do not ever call adns_afterpoll
1640 <p>Method 4: Do not use a prepare or check watcher because the module you
1641 want to embed is too inflexible to support it. Instead, youc na override
1642 their poll function. The drawback with this solution is that the main
1643 loop is now no longer controllable by EV. The <code>Glib::EV</code> module does
1646 event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1650 for (n = 0; n < nfds; ++n)
1651 // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1653 if (timeout >= 0)
1654 // create/start timer
1660 if (timeout >= 0)
1661 ev_timer_stop (EV_A_ &to);
1663 // stop io watchers again - their callbacks should have set
1664 for (n = 0; n < nfds; ++n)
1665 ev_io_stop (EV_A_ iow [n]);
1676 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1677 <div id="code_ev_embed_code_when_one_backend_-2">
1678 <p>This is a rather advanced watcher type that lets you embed one event loop
1679 into another (currently only <code>ev_io</code> events are supported in the embedded
1680 loop, other types of watchers might be handled in a delayed or incorrect
1681 fashion and must not be used).</p>
1682 <p>There are primarily two reasons you would want that: work around bugs and
1684 <p>As an example for a bug workaround, the kqueue backend might only support
1685 sockets on some platform, so it is unusable as generic backend, but you
1686 still want to make use of it because you have many sockets and it scales
1687 so nicely. In this case, you would create a kqueue-based loop and embed it
1688 into your default loop (which might use e.g. poll). Overall operation will
1689 be a bit slower because first libev has to poll and then call kevent, but
1690 at least you can use both at what they are best.</p>
1691 <p>As for prioritising I/O: rarely you have the case where some fds have
1692 to be watched and handled very quickly (with low latency), and even
1693 priorities and idle watchers might have too much overhead. In this case
1694 you would put all the high priority stuff in one loop and all the rest in
1695 a second one, and embed the second one in the first.</p>
1696 <p>As long as the watcher is active, the callback will be invoked every time
1697 there might be events pending in the embedded loop. The callback must then
1698 call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1699 their callbacks (you could also start an idle watcher to give the embedded
1700 loop strictly lower priority for example). You can also set the callback
1701 to <code>0</code>, in which case the embed watcher will automatically execute the
1702 embedded loop sweep.</p>
1703 <p>As long as the watcher is started it will automatically handle events. The
1704 callback will be invoked whenever some events have been handled. You can
1705 set the callback to <code>0</code> to avoid having to specify one if you are not
1706 interested in that.</p>
1707 <p>Also, there have not currently been made special provisions for forking:
1708 when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1709 but you will also have to stop and restart any <code>ev_embed</code> watchers
1711 <p>Unfortunately, not all backends are embeddable, only the ones returned by
1712 <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1714 <p>So when you want to use this feature you will always have to be prepared
1715 that you cannot get an embeddable loop. The recommended way to get around
1716 this is to have a separate variables for your embeddable loop, try to
1717 create it, and if that fails, use the normal loop for everything:</p>
1718 <pre> struct ev_loop *loop_hi = ev_default_init (0);
1719 struct ev_loop *loop_lo = 0;
1720 struct ev_embed embed;
1722 // see if there is a chance of getting one that works
1723 // (remember that a flags value of 0 means autodetection)
1724 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1725 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1728 // if we got one, then embed it, otherwise default to loop_hi
1731 ev_embed_init (&embed, 0, loop_lo);
1732 ev_embed_start (loop_hi, &embed);
1739 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1740 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1742 <p>Configures the watcher to embed the given loop, which must be
1743 embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1744 invoked automatically, otherwise it is the responsibility of the callback
1745 to invoke it (it will continue to be called until the sweep has been done,
1746 if you do not want thta, you need to temporarily stop the embed watcher).</p>
1748 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1750 <p>Make a single, non-blocking sweep over the embedded loop. This works
1751 similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1752 apropriate way for embedded loops.</p>
1754 <dt>struct ev_loop *loop [read-only]</dt>
1756 <p>The embedded event loop.</p>
1765 <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>
1766 <div id="code_ev_fork_code_the_audacity_to_re-2">
1767 <p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1768 whoever is a good citizen cared to tell libev about it by calling
1769 <code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1770 event loop blocks next and before <code>ev_check</code> watchers are being called,
1771 and only in the child after the fork. If whoever good citizen calling
1772 <code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1773 handlers will be invoked, too, of course.</p>
1775 <dt>ev_fork_init (ev_signal *, callback)</dt>
1777 <p>Initialises and configures the fork watcher - it has no parameters of any
1778 kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1788 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
1789 <div id="OTHER_FUNCTIONS_CONTENT">
1790 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1792 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1794 <p>This function combines a simple timer and an I/O watcher, calls your
1795 callback on whichever event happens first and automatically stop both
1796 watchers. This is useful if you want to wait for a single event on an fd
1797 or timeout without having to allocate/configure/start/stop/free one or
1798 more watchers yourself.</p>
1799 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1800 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1801 <code>events</code> set will be craeted and started.</p>
1802 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1803 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1804 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1806 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1807 passed an <code>revents</code> set like normal event callbacks (a combination of
1808 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1809 value passed to <code>ev_once</code>:</p>
1810 <pre> static void stdin_ready (int revents, void *arg)
1812 if (revents & EV_TIMEOUT)
1813 /* doh, nothing entered */;
1814 else if (revents & EV_READ)
1815 /* stdin might have data for us, joy! */;
1818 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1822 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1824 <p>Feeds the given event set into the event loop, as if the specified event
1825 had happened for the specified watcher (which must be a pointer to an
1826 initialised but not necessarily started event watcher).</p>
1828 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1830 <p>Feed an event on the given fd, as if a file descriptor backend detected
1831 the given events it.</p>
1833 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1835 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1845 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1846 <div id="LIBEVENT_EMULATION_CONTENT">
1847 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1848 emulate the internals of libevent, so here are some usage hints:</p>
1850 <dt>* Use it by including <event.h>, as usual.</dt>
1851 <dt>* The following members are fully supported: ev_base, ev_callback,
1852 ev_arg, ev_fd, ev_res, ev_events.</dt>
1853 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1854 maintained by libev, it does not work exactly the same way as in libevent (consider
1855 it a private API).</dt>
1856 <dt>* Priorities are not currently supported. Initialising priorities
1857 will fail and all watchers will have the same priority, even though there
1858 is an ev_pri field.</dt>
1859 <dt>* Other members are not supported.</dt>
1860 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1861 to use the libev header file and library.</dt>
1865 <h1 id="C_SUPPORT">C++ SUPPORT</h1>
1866 <div id="C_SUPPORT_CONTENT">
1867 <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1868 you to use some convinience methods to start/stop watchers and also change
1869 the callback model to a model using method callbacks on objects.</p>
1871 <pre> #include <ev++.h>
1874 <p>This automatically includes <cite>ev.h</cite> and puts all of its definitions (many
1875 of them macros) into the global namespace. All C++ specific things are
1876 put into the <code>ev</code> namespace. It should support all the same embedding
1877 options as <cite>ev.h</cite>, most notably <code>EV_MULTIPLICITY</code>.</p>
1878 <p>Care has been taken to keep the overhead low. The only data member the C++
1879 classes add (compared to plain C-style watchers) is the event loop pointer
1880 that the watcher is associated with (or no additional members at all if
1881 you disable <code>EV_MULTIPLICITY</code> when embedding libev).</p>
1882 <p>Currently, functions, and static and non-static member functions can be
1883 used as callbacks. Other types should be easy to add as long as they only
1884 need one additional pointer for context. If you need support for other
1885 types of functors please contact the author (preferably after implementing
1887 <p>Here is a list of things available in the <code>ev</code> namespace:</p>
1889 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1891 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1892 macros from <cite>ev.h</cite>.</p>
1894 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1896 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1898 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1900 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1901 the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1902 which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1903 defines by many implementations.</p>
1904 <p>All of those classes have these methods:</p>
1907 <dt>ev::TYPE::TYPE ()</dt>
1908 <dt>ev::TYPE::TYPE (struct ev_loop *)</dt>
1909 <dt>ev::TYPE::~TYPE</dt>
1911 <p>The constructor (optionally) takes an event loop to associate the watcher
1912 with. If it is omitted, it will use <code>EV_DEFAULT</code>.</p>
1913 <p>The constructor calls <code>ev_init</code> for you, which means you have to call the
1914 <code>set</code> method before starting it.</p>
1915 <p>It will not set a callback, however: You have to call the templated <code>set</code>
1916 method to set a callback before you can start the watcher.</p>
1917 <p>(The reason why you have to use a method is a limitation in C++ which does
1918 not allow explicit template arguments for constructors).</p>
1919 <p>The destructor automatically stops the watcher if it is active.</p>
1921 <dt>w->set<class, &class::method> (object *)</dt>
1923 <p>This method sets the callback method to call. The method has to have a
1924 signature of <code>void (*)(ev_TYPE &, int)</code>, it receives the watcher as
1925 first argument and the <code>revents</code> as second. The object must be given as
1926 parameter and is stored in the <code>data</code> member of the watcher.</p>
1927 <p>This method synthesizes efficient thunking code to call your method from
1928 the C callback that libev requires. If your compiler can inline your
1929 callback (i.e. it is visible to it at the place of the <code>set</code> call and
1930 your compiler is good :), then the method will be fully inlined into the
1931 thunking function, making it as fast as a direct C callback.</p>
1932 <p>Example: simple class declaration and watcher initialisation</p>
1933 <pre> struct myclass
1935 void io_cb (ev::io &w, int revents) { }
1940 iow.set <myclass, &myclass::io_cb> (&obj);
1944 <dt>w->set<function> (void *data = 0)</dt>
1946 <p>Also sets a callback, but uses a static method or plain function as
1947 callback. The optional <code>data</code> argument will be stored in the watcher's
1948 <code>data</code> member and is free for you to use.</p>
1949 <p>The prototype of the <code>function</code> must be <code>void (*)(ev::TYPE &w, int)</code>.</p>
1950 <p>See the method-<code>set</code> above for more details.</p>
1952 <pre> static void io_cb (ev::io &w, int revents) { }
1953 iow.set <io_cb> ();
1957 <dt>w->set (struct ev_loop *)</dt>
1959 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1960 do this when the watcher is inactive (and not pending either).</p>
1962 <dt>w->set ([args])</dt>
1964 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1965 called at least once. Unlike the C counterpart, an active watcher gets
1966 automatically stopped and restarted when reconfiguring it with this
1969 <dt>w->start ()</dt>
1971 <p>Starts the watcher. Note that there is no <code>loop</code> argument, as the
1972 constructor already stores the event loop.</p>
1974 <dt>w->stop ()</dt>
1976 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1978 <dt>w->again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1980 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1981 <code>ev_TYPE_again</code> function.</p>
1983 <dt>w->sweep () <code>ev::embed</code> only</dt>
1985 <p>Invokes <code>ev_embed_sweep</code>.</p>
1987 <dt>w->update () <code>ev::stat</code> only</dt>
1989 <p>Invokes <code>ev_stat_stat</code>.</p>
1995 <p>Example: Define a class with an IO and idle watcher, start one of them in
1996 the constructor.</p>
1999 ev_io io; void io_cb (ev::io &w, int revents);
2000 ev_idle idle void idle_cb (ev::idle &w, int revents);
2005 myclass::myclass (int fd)
2007 io .set <myclass, &myclass::io_cb > (this);
2008 idle.set <myclass, &myclass::idle_cb> (this);
2010 io.start (fd, ev::READ);
2019 <h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
2020 <div id="MACRO_MAGIC_CONTENT">
2021 <p>Libev can be compiled with a variety of options, the most fundemantal is
2022 <code>EV_MULTIPLICITY</code>. This option determines whether (most) functions and
2023 callbacks have an initial <code>struct ev_loop *</code> argument.</p>
2024 <p>To make it easier to write programs that cope with either variant, the
2025 following macros are defined:</p>
2027 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
2029 <p>This provides the loop <i>argument</i> for functions, if one is required ("ev
2030 loop argument"). The <code>EV_A</code> form is used when this is the sole argument,
2031 <code>EV_A_</code> is used when other arguments are following. Example:</p>
2032 <pre> ev_unref (EV_A);
2033 ev_timer_add (EV_A_ watcher);
2037 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
2038 which is often provided by the following macro.</p>
2040 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
2042 <p>This provides the loop <i>parameter</i> for functions, if one is required ("ev
2043 loop parameter"). The <code>EV_P</code> form is used when this is the sole parameter,
2044 <code>EV_P_</code> is used when other parameters are following. Example:</p>
2045 <pre> // this is how ev_unref is being declared
2046 static void ev_unref (EV_P);
2048 // this is how you can declare your typical callback
2049 static void cb (EV_P_ ev_timer *w, int revents)
2052 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
2053 suitable for use with <code>EV_A</code>.</p>
2055 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
2057 <p>Similar to the other two macros, this gives you the value of the default
2058 loop, if multiple loops are supported ("ev loop default").</p>
2061 <p>Example: Declare and initialise a check watcher, utilising the above
2062 macros so it will work regardless of whether multiple loops are supported
2065 check_cb (EV_P_ ev_timer *w, int revents)
2067 ev_check_stop (EV_A_ w);
2071 ev_check_init (&check, check_cb);
2072 ev_check_start (EV_DEFAULT_ &check);
2073 ev_loop (EV_DEFAULT_ 0);
2078 <h1 id="EMBEDDING">EMBEDDING</h1>
2079 <div id="EMBEDDING_CONTENT">
2080 <p>Libev can (and often is) directly embedded into host
2081 applications. Examples of applications that embed it include the Deliantra
2082 Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2083 and rxvt-unicode.</p>
2084 <p>The goal is to enable you to just copy the neecssary files into your
2085 source directory without having to change even a single line in them, so
2086 you can easily upgrade by simply copying (or having a checked-out copy of
2087 libev somewhere in your source tree).</p>
2090 <h2 id="FILESETS">FILESETS</h2>
2091 <div id="FILESETS_CONTENT">
2092 <p>Depending on what features you need you need to include one or more sets of files
2096 <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
2097 <div id="CORE_EVENT_LOOP_CONTENT">
2098 <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
2099 configuration (no autoconf):</p>
2100 <pre> #define EV_STANDALONE 1
2101 #include "ev.c"
2104 <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
2105 single C source file only to provide the function implementations. To use
2106 it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
2107 done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
2108 where you can put other configuration options):</p>
2109 <pre> #define EV_STANDALONE 1
2110 #include "ev.h"
2113 <p>Both header files and implementation files can be compiled with a C++
2114 compiler (at least, thats a stated goal, and breakage will be treated
2116 <p>You need the following files in your source tree, or in a directory
2117 in your include path (e.g. in libev/ when using -Ilibev):</p>
2123 ev_win32.c required on win32 platforms only
2125 ev_select.c only when select backend is enabled (which is enabled by default)
2126 ev_poll.c only when poll backend is enabled (disabled by default)
2127 ev_epoll.c only when the epoll backend is enabled (disabled by default)
2128 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2129 ev_port.c only when the solaris port backend is enabled (disabled by default)
2132 <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
2133 to compile this single file.</p>
2136 <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
2137 <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
2138 <p>To include the libevent compatibility API, also include:</p>
2139 <pre> #include "event.c"
2142 <p>in the file including <cite>ev.c</cite>, and:</p>
2143 <pre> #include "event.h"
2146 <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
2147 <p>You need the following additional files for this:</p>
2154 <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
2155 <div id="AUTOCONF_SUPPORT_CONTENT">
2156 <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
2157 whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
2158 <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
2159 include <cite>config.h</cite> and configure itself accordingly.</p>
2160 <p>For this of course you need the m4 file:</p>
2166 <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
2167 <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
2168 <p>Libev can be configured via a variety of preprocessor symbols you have to define
2169 before including any of its files. The default is not to build for multiplicity
2170 and only include the select backend.</p>
2172 <dt>EV_STANDALONE</dt>
2174 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
2175 keeps libev from including <cite>config.h</cite>, and it also defines dummy
2176 implementations for some libevent functions (such as logging, which is not
2177 supported). It will also not define any of the structs usually found in
2178 <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
2180 <dt>EV_USE_MONOTONIC</dt>
2182 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2183 monotonic clock option at both compiletime and runtime. Otherwise no use
2184 of the monotonic clock option will be attempted. If you enable this, you
2185 usually have to link against librt or something similar. Enabling it when
2186 the functionality isn't available is safe, though, althoguh you have
2187 to make sure you link against any libraries where the <code>clock_gettime</code>
2188 function is hiding in (often <cite>-lrt</cite>).</p>
2190 <dt>EV_USE_REALTIME</dt>
2192 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2193 realtime clock option at compiletime (and assume its availability at
2194 runtime if successful). Otherwise no use of the realtime clock option will
2195 be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
2196 (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
2197 in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
2199 <dt>EV_USE_SELECT</dt>
2201 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
2202 <code>select</code>(2) backend. No attempt at autodetection will be done: if no
2203 other method takes over, select will be it. Otherwise the select backend
2204 will not be compiled in.</p>
2206 <dt>EV_SELECT_USE_FD_SET</dt>
2208 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
2209 structure. This is useful if libev doesn't compile due to a missing
2210 <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
2211 exotic systems. This usually limits the range of file descriptors to some
2212 low limit such as 1024 or might have other limitations (winsocket only
2213 allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
2214 influence the size of the <code>fd_set</code> used.</p>
2216 <dt>EV_SELECT_IS_WINSOCKET</dt>
2218 <p>When defined to <code>1</code>, the select backend will assume that
2219 select/socket/connect etc. don't understand file descriptors but
2220 wants osf handles on win32 (this is the case when the select to
2221 be used is the winsock select). This means that it will call
2222 <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2223 it is assumed that all these functions actually work on fds, even
2224 on win32. Should not be defined on non-win32 platforms.</p>
2226 <dt>EV_USE_POLL</dt>
2228 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2229 backend. Otherwise it will be enabled on non-win32 platforms. It
2230 takes precedence over select.</p>
2232 <dt>EV_USE_EPOLL</dt>
2234 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2235 <code>epoll</code>(7) backend. Its availability will be detected at runtime,
2236 otherwise another method will be used as fallback. This is the
2237 preferred backend for GNU/Linux systems.</p>
2239 <dt>EV_USE_KQUEUE</dt>
2241 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2242 <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2243 otherwise another method will be used as fallback. This is the preferred
2244 backend for BSD and BSD-like systems, although on most BSDs kqueue only
2245 supports some types of fds correctly (the only platform we found that
2246 supports ptys for example was NetBSD), so kqueue might be compiled in, but
2247 not be used unless explicitly requested. The best way to use it is to find
2248 out whether kqueue supports your type of fd properly and use an embedded
2251 <dt>EV_USE_PORT</dt>
2253 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
2254 10 port style backend. Its availability will be detected at runtime,
2255 otherwise another method will be used as fallback. This is the preferred
2256 backend for Solaris 10 systems.</p>
2258 <dt>EV_USE_DEVPOLL</dt>
2260 <p>reserved for future expansion, works like the USE symbols above.</p>
2262 <dt>EV_USE_INOTIFY</dt>
2264 <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2265 interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2266 be detected at runtime.</p>
2270 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2271 undefined is <code><ev.h></code> in <cite>event.h</cite> and <code>"ev.h"</code> in <cite>ev.c</cite>. This
2272 can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2274 <dt>EV_CONFIG_H</dt>
2276 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2277 <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2278 <code>EV_H</code>, above.</p>
2282 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2283 of how the <cite>event.h</cite> header can be found.</p>
2285 <dt>EV_PROTOTYPES</dt>
2287 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2288 prototypes, but still define all the structs and other symbols. This is
2289 occasionally useful if you want to provide your own wrapper functions
2290 around libev functions.</p>
2292 <dt>EV_MULTIPLICITY</dt>
2294 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2295 will have the <code>struct ev_loop *</code> as first argument, and you can create
2296 additional independent event loops. Otherwise there will be no support
2297 for multiple event loops and there is no first event loop pointer
2298 argument. Instead, all functions act on the single default loop.</p>
2303 <p>The range of allowed priorities. <code>EV_MINPRI</code> must be smaller or equal to
2304 <code>EV_MAXPRI</code>, but otherwise there are no non-obvious limitations. You can
2305 provide for more priorities by overriding those symbols (usually defined
2306 to be <code>-2</code> and <code>2</code>, respectively).</p>
2307 <p>When doing priority-based operations, libev usually has to linearly search
2308 all the priorities, so having many of them (hundreds) uses a lot of space
2309 and time, so using the defaults of five priorities (-2 .. +2) is usually
2311 <p>If your embedding app does not need any priorities, defining these both to
2312 <code>0</code> will save some memory and cpu.</p>
2314 <dt>EV_PERIODIC_ENABLE</dt>
2316 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2317 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2320 <dt>EV_IDLE_ENABLE</dt>
2322 <p>If undefined or defined to be <code>1</code>, then idle watchers are supported. If
2323 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2326 <dt>EV_EMBED_ENABLE</dt>
2328 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2329 defined to be <code>0</code>, then they are not.</p>
2331 <dt>EV_STAT_ENABLE</dt>
2333 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2334 defined to be <code>0</code>, then they are not.</p>
2336 <dt>EV_FORK_ENABLE</dt>
2338 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2339 defined to be <code>0</code>, then they are not.</p>
2343 <p>If you need to shave off some kilobytes of code at the expense of some
2344 speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2345 some inlining decisions, saves roughly 30% codesize of amd64.</p>
2347 <dt>EV_PID_HASHSIZE</dt>
2349 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2350 pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2351 than enough. If you need to manage thousands of children you might want to
2352 increase this value (<i>must</i> be a power of two).</p>
2354 <dt>EV_INOTIFY_HASHSIZE</dt>
2356 <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2357 inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2358 usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2359 watchers you might want to increase this value (<i>must</i> be a power of
2364 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2365 this macro to a something else you can include more and other types of
2366 members. You have to define it each time you include one of the files,
2367 though, and it must be identical each time.</p>
2368 <p>For example, the perl EV module uses something like this:</p>
2369 <pre> #define EV_COMMON \
2370 SV *self; /* contains this struct */ \
2371 SV *cb_sv, *fh /* note no trailing ";" */
2375 <dt>EV_CB_DECLARE (type)</dt>
2376 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2377 <dt>ev_set_cb (ev, cb)</dt>
2379 <p>Can be used to change the callback member declaration in each watcher,
2380 and the way callbacks are invoked and set. Must expand to a struct member
2381 definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2382 their default definitions. One possible use for overriding these is to
2383 avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2384 method calls instead of plain function calls in C++.</p>
2387 <h2 id="EXAMPLES">EXAMPLES</h2>
2388 <div id="EXAMPLES_CONTENT">
2389 <p>For a real-world example of a program the includes libev
2390 verbatim, you can have a look at the EV perl module
2391 (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2392 the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2393 interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2394 will be compiled. It is pretty complex because it provides its own header
2396 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2397 that everybody includes and which overrides some configure choices:</p>
2398 <pre> #define EV_MINIMAL 1
2399 #define EV_USE_POLL 0
2400 #define EV_MULTIPLICITY 0
2401 #define EV_PERIODIC_ENABLE 0
2402 #define EV_STAT_ENABLE 0
2403 #define EV_FORK_ENABLE 0
2404 #define EV_CONFIG_H <config.h>
2408 #include "ev++.h"
2411 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2412 <pre> #include "ev_cpp.h"
2413 #include "ev.c"
2421 <h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2422 <div id="COMPLEXITIES_CONTENT">
2423 <p>In this section the complexities of (many of) the algorithms used inside
2424 libev will be explained. For complexity discussions about backends see the
2425 documentation for <code>ev_default_init</code>.</p>
2426 <p>All of the following are about amortised time: If an array needs to be
2427 extended, libev needs to realloc and move the whole array, but this
2428 happens asymptotically never with higher number of elements, so O(1) might
2429 mean it might do a lengthy realloc operation in rare cases, but on average
2430 it is much faster and asymptotically approaches constant time.</p>
2433 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2435 <p>This means that, when you have a watcher that triggers in one hour and
2436 there are 100 watchers that would trigger before that then inserting will
2437 have to skip those 100 watchers.</p>
2439 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2441 <p>That means that for changing a timer costs less than removing/adding them
2442 as only the relative motion in the event queue has to be paid for.</p>
2444 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2446 <p>These just add the watcher into an array or at the head of a list.
2447 =item Stopping check/prepare/idle watchers: O(1)</p>
2449 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2451 <p>These watchers are stored in lists then need to be walked to find the
2452 correct watcher to remove. The lists are usually short (you don't usually
2453 have many watchers waiting for the same fd or signal).</p>
2455 <dt>Finding the next timer per loop iteration: O(1)</dt>
2456 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2458 <p>A change means an I/O watcher gets started or stopped, which requires
2459 libev to recalculate its status (and possibly tell the kernel).</p>
2461 <dt>Activating one watcher: O(1)</dt>
2462 <dt>Priority handling: O(number_of_priorities)</dt>
2464 <p>Priorities are implemented by allocating some space for each
2465 priority. When doing priority-based operations, libev usually has to
2466 linearly search all the priorities.</p>
2476 <h1 id="AUTHOR">AUTHOR</h1>
2477 <div id="AUTHOR_CONTENT">
2478 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob