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="Sun Dec 9 20:47:27 2007" />
10 <meta name="generator" content="Pod::Xhtml 1.57" />
11 <link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
15 <h3 id="TOP">Index</h3>
17 <ul><li><a href="#NAME">NAME</a></li>
18 <li><a href="#SYNOPSIS">SYNOPSIS</a></li>
19 <li><a href="#EXAMPLE_PROGRAM">EXAMPLE PROGRAM</a></li>
20 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
21 <li><a href="#FEATURES">FEATURES</a></li>
22 <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
23 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
24 <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
25 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
26 <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
27 <ul><li><a href="#GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</a></li>
28 <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
31 <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
32 <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</a></li>
33 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
34 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
35 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
36 <li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
37 <li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
38 <li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</a></li>
39 <li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</a></li>
40 <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
41 <li><a href="#code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</a></li>
44 <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
45 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
46 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
47 <li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
48 <li><a href="#EMBEDDING">EMBEDDING</a>
49 <ul><li><a href="#FILESETS">FILESETS</a>
50 <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
51 <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
52 <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
55 <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
56 <li><a href="#EXAMPLES">EXAMPLES</a></li>
59 <li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
60 <li><a href="#AUTHOR">AUTHOR</a>
65 <h1 id="NAME">NAME</h1>
66 <div id="NAME_CONTENT">
67 <p>libev - a high performance full-featured event loop written in C</p>
70 <h1 id="SYNOPSIS">SYNOPSIS</h1>
71 <div id="SYNOPSIS_CONTENT">
72 <pre> #include <ev.h>
77 <h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
78 <div id="EXAMPLE_PROGRAM_CONTENT">
79 <pre> #include <ev.h>
82 ev_timer timeout_watcher;
84 /* called when data readable on stdin */
86 stdin_cb (EV_P_ struct ev_io *w, int revents)
88 /* puts ("stdin ready"); */
89 ev_io_stop (EV_A_ w); /* just a syntax example */
90 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
94 timeout_cb (EV_P_ struct ev_timer *w, int revents)
96 /* puts ("timeout"); */
97 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
103 struct ev_loop *loop = ev_default_loop (0);
105 /* initialise an io watcher, then start it */
106 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
107 ev_io_start (loop, &stdin_watcher);
109 /* simple non-repeating 5.5 second timeout */
110 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
111 ev_timer_start (loop, &timeout_watcher);
113 /* loop till timeout or data ready */
122 <h1 id="DESCRIPTION">DESCRIPTION</h1>
123 <div id="DESCRIPTION_CONTENT">
124 <p>The newest version of this document is also available as a html-formatted
125 web page you might find easier to navigate when reading it for the first
126 time: <a href="http://cvs.schmorp.de/libev/ev.html">http://cvs.schmorp.de/libev/ev.html</a>.</p>
127 <p>Libev is an event loop: you register interest in certain events (such as a
128 file descriptor being readable or a timeout occuring), and it will manage
129 these event sources and provide your program with events.</p>
130 <p>To do this, it must take more or less complete control over your process
131 (or thread) by executing the <i>event loop</i> handler, and will then
132 communicate events via a callback mechanism.</p>
133 <p>You register interest in certain events by registering so-called <i>event
134 watchers</i>, which are relatively small C structures you initialise with the
135 details of the event, and then hand it over to libev by <i>starting</i> the
139 <h1 id="FEATURES">FEATURES</h1>
140 <div id="FEATURES_CONTENT">
141 <p>Libev supports <code>select</code>, <code>poll</code>, the Linux-specific <code>epoll</code>, the
142 BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
143 for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
144 (for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
145 with customised rescheduling (<code>ev_periodic</code>), synchronous signals
146 (<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
147 watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
148 <code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
149 file watchers (<code>ev_stat</code>) and even limited support for fork events
150 (<code>ev_fork</code>).</p>
151 <p>It also is quite fast (see this
152 <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
156 <h1 id="CONVENTIONS">CONVENTIONS</h1>
157 <div id="CONVENTIONS_CONTENT">
158 <p>Libev is very configurable. In this manual the default configuration will
159 be described, which supports multiple event loops. For more info about
160 various configuration options please have a look at <strong>EMBED</strong> section in
161 this manual. If libev was configured without support for multiple event
162 loops, then all functions taking an initial argument of name <code>loop</code>
163 (which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
166 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
167 <div id="TIME_REPRESENTATION_CONTENT">
168 <p>Libev represents time as a single floating point number, representing the
169 (fractional) number of seconds since the (POSIX) epoch (somewhere near
170 the beginning of 1970, details are complicated, don't ask). This type is
171 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
172 to the <code>double</code> type in C, and when you need to do any calculations on
173 it, you should treat it as such.</p>
176 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
177 <div id="GLOBAL_FUNCTIONS_CONTENT">
178 <p>These functions can be called anytime, even before initialising the
179 library in any way.</p>
181 <dt>ev_tstamp ev_time ()</dt>
183 <p>Returns the current time as libev would use it. Please note that the
184 <code>ev_now</code> function is usually faster and also often returns the timestamp
185 you actually want to know.</p>
187 <dt>int ev_version_major ()</dt>
188 <dt>int ev_version_minor ()</dt>
190 <p>You can find out the major and minor ABI version numbers of the library
191 you linked against by calling the functions <code>ev_version_major</code> and
192 <code>ev_version_minor</code>. If you want, you can compare against the global
193 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
194 version of the library your program was compiled against.</p>
195 <p>These version numbers refer to the ABI version of the library, not the
197 <p>Usually, it's a good idea to terminate if the major versions mismatch,
198 as this indicates an incompatible change. Minor versions are usually
199 compatible to older versions, so a larger minor version alone is usually
201 <p>Example: Make sure we haven't accidentally been linked against the wrong
203 <pre> assert (("libev version mismatch",
204 ev_version_major () == EV_VERSION_MAJOR
205 && ev_version_minor () >= EV_VERSION_MINOR));
209 <dt>unsigned int ev_supported_backends ()</dt>
211 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
212 value) compiled into this binary of libev (independent of their
213 availability on the system you are running on). See <code>ev_default_loop</code> for
214 a description of the set values.</p>
215 <p>Example: make sure we have the epoll method, because yeah this is cool and
216 a must have and can we have a torrent of it please!!!11</p>
217 <pre> assert (("sorry, no epoll, no sex",
218 ev_supported_backends () & EVBACKEND_EPOLL));
222 <dt>unsigned int ev_recommended_backends ()</dt>
224 <p>Return the set of all backends compiled into this binary of libev and also
225 recommended for this platform. This set is often smaller than the one
226 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
227 most BSDs and will not be autodetected unless you explicitly request it
228 (assuming you know what you are doing). This is the set of backends that
229 libev will probe for if you specify no backends explicitly.</p>
231 <dt>unsigned int ev_embeddable_backends ()</dt>
233 <p>Returns the set of backends that are embeddable in other event loops. This
234 is the theoretical, all-platform, value. To find which backends
235 might be supported on the current system, you would need to look at
236 <code>ev_embeddable_backends () & ev_supported_backends ()</code>, likewise for
237 recommended ones.</p>
238 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
240 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
242 <p>Sets the allocation function to use (the prototype is similar - the
243 semantics is identical - to the realloc C function). It is used to
244 allocate and free memory (no surprises here). If it returns zero when
245 memory needs to be allocated, the library might abort or take some
246 potentially destructive action. The default is your system realloc
248 <p>You could override this function in high-availability programs to, say,
249 free some memory if it cannot allocate memory, to use a special allocator,
250 or even to sleep a while and retry until some memory is available.</p>
251 <p>Example: Replace the libev allocator with one that waits a bit and then
254 persistent_realloc (void *ptr, size_t size)
258 void *newptr = realloc (ptr, size);
268 ev_set_allocator (persistent_realloc);
272 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
274 <p>Set the callback function to call on a retryable syscall error (such
275 as failed select, poll, epoll_wait). The message is a printable string
276 indicating the system call or subsystem causing the problem. If this
277 callback is set, then libev will expect it to remedy the sitution, no
278 matter what, when it returns. That is, libev will generally retry the
279 requested operation, or, if the condition doesn't go away, do bad stuff
281 <p>Example: This is basically the same thing that libev does internally, too.</p>
283 fatal_error (const char *msg)
290 ev_set_syserr_cb (fatal_error);
297 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
298 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
299 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
300 types of such loops, the <i>default</i> loop, which supports signals and child
301 events, and dynamically created loops which do not.</p>
302 <p>If you use threads, a common model is to run the default event loop
303 in your main thread (or in a separate thread) and for each thread you
304 create, you also create another event loop. Libev itself does no locking
305 whatsoever, so if you mix calls to the same event loop in different
306 threads, make sure you lock (this is usually a bad idea, though, even if
307 done correctly, because it's hideous and inefficient).</p>
309 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
311 <p>This will initialise the default event loop if it hasn't been initialised
312 yet and return it. If the default loop could not be initialised, returns
313 false. If it already was initialised it simply returns it (and ignores the
314 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
315 <p>If you don't know what event loop to use, use the one returned from this
317 <p>The flags argument can be used to specify special behaviour or specific
318 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
319 <p>The following flags are supported:</p>
322 <dt><code>EVFLAG_AUTO</code></dt>
324 <p>The default flags value. Use this if you have no clue (it's the right
325 thing, believe me).</p>
327 <dt><code>EVFLAG_NOENV</code></dt>
329 <p>If this flag bit is ored into the flag value (or the program runs setuid
330 or setgid) then libev will <i>not</i> look at the environment variable
331 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
332 override the flags completely if it is found in the environment. This is
333 useful to try out specific backends to test their performance, or to work
336 <dt><code>EVFLAG_FORKCHECK</code></dt>
338 <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
339 a fork, you can also make libev check for a fork in each iteration by
340 enabling this flag.</p>
341 <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
342 and thus this might slow down your event loop if you do a lot of loop
343 iterations and little real work, but is usually not noticeable (on my
344 Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
345 without a syscall and thus <i>very</i> fast, but my Linux system also has
346 <code>pthread_atfork</code> which is even faster).</p>
347 <p>The big advantage of this flag is that you can forget about fork (and
348 forget about forgetting to tell libev about forking) when you use this
350 <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
351 environment variable.</p>
353 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
355 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
356 libev tries to roll its own fd_set with no limits on the number of fds,
357 but if that fails, expect a fairly low limit on the number of fds when
358 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
359 the fastest backend for a low number of fds.</p>
361 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
363 <p>And this is your standard poll(2) backend. It's more complicated than
364 select, but handles sparse fds better and has no artificial limit on the
365 number of fds you can use (except it will slow down considerably with a
366 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
368 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
370 <p>For few fds, this backend is a bit little slower than poll and select,
371 but it scales phenomenally better. While poll and select usually scale like
372 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
373 either O(1) or O(active_fds).</p>
374 <p>While stopping and starting an I/O watcher in the same iteration will
375 result in some caching, there is still a syscall per such incident
376 (because the fd could point to a different file description now), so its
377 best to avoid that. Also, dup()ed file descriptors might not work very
378 well if you register events for both fds.</p>
379 <p>Please note that epoll sometimes generates spurious notifications, so you
380 need to use non-blocking I/O or other means to avoid blocking when no data
381 (or space) is available.</p>
383 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
385 <p>Kqueue deserves special mention, as at the time of this writing, it
386 was broken on all BSDs except NetBSD (usually it doesn't work with
387 anything but sockets and pipes, except on Darwin, where of course its
388 completely useless). For this reason its not being "autodetected"
389 unless you explicitly specify it explicitly in the flags (i.e. using
390 <code>EVBACKEND_KQUEUE</code>).</p>
391 <p>It scales in the same way as the epoll backend, but the interface to the
392 kernel is more efficient (which says nothing about its actual speed, of
393 course). While starting and stopping an I/O watcher does not cause an
394 extra syscall as with epoll, it still adds up to four event changes per
395 incident, so its best to avoid that.</p>
397 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
399 <p>This is not implemented yet (and might never be).</p>
401 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
403 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
404 it's really slow, but it still scales very well (O(active_fds)).</p>
405 <p>Please note that solaris ports can result in a lot of spurious
406 notifications, so you need to use non-blocking I/O or other means to avoid
407 blocking when no data (or space) is available.</p>
409 <dt><code>EVBACKEND_ALL</code></dt>
411 <p>Try all backends (even potentially broken ones that wouldn't be tried
412 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
413 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
417 <p>If one or more of these are ored into the flags value, then only these
418 backends will be tried (in the reverse order as given here). If none are
419 specified, most compiled-in backend will be tried, usually in reverse
420 order of their flag values :)</p>
421 <p>The most typical usage is like this:</p>
422 <pre> if (!ev_default_loop (0))
423 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
426 <p>Restrict libev to the select and poll backends, and do not allow
427 environment settings to be taken into account:</p>
428 <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
431 <p>Use whatever libev has to offer, but make sure that kqueue is used if
432 available (warning, breaks stuff, best use only with your own private
433 event loop and only if you know the OS supports your types of fds):</p>
434 <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
438 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
440 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
441 always distinct from the default loop. Unlike the default loop, it cannot
442 handle signal and child watchers, and attempts to do so will be greeted by
443 undefined behaviour (or a failed assertion if assertions are enabled).</p>
444 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
445 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
447 fatal ("no epoll found here, maybe it hides under your chair");
451 <dt>ev_default_destroy ()</dt>
453 <p>Destroys the default loop again (frees all memory and kernel state
454 etc.). None of the active event watchers will be stopped in the normal
455 sense, so e.g. <code>ev_is_active</code> might still return true. It is your
456 responsibility to either stop all watchers cleanly yoursef <i>before</i>
457 calling this function, or cope with the fact afterwards (which is usually
458 the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
461 <dt>ev_loop_destroy (loop)</dt>
463 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
464 earlier call to <code>ev_loop_new</code>.</p>
466 <dt>ev_default_fork ()</dt>
468 <p>This function reinitialises the kernel state for backends that have
469 one. Despite the name, you can call it anytime, but it makes most sense
470 after forking, in either the parent or child process (or both, but that
471 again makes little sense).</p>
472 <p>You <i>must</i> call this function in the child process after forking if and
473 only if you want to use the event library in both processes. If you just
474 fork+exec, you don't have to call it.</p>
475 <p>The function itself is quite fast and it's usually not a problem to call
476 it just in case after a fork. To make this easy, the function will fit in
477 quite nicely into a call to <code>pthread_atfork</code>:</p>
478 <pre> pthread_atfork (0, 0, ev_default_fork);
481 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
482 without calling this function, so if you force one of those backends you
483 do not need to care.</p>
485 <dt>ev_loop_fork (loop)</dt>
487 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
488 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
489 after fork, and how you do this is entirely your own problem.</p>
491 <dt>unsigned int ev_loop_count (loop)</dt>
493 <p>Returns the count of loop iterations for the loop, which is identical to
494 the number of times libev did poll for new events. It starts at <code>0</code> and
495 happily wraps around with enough iterations.</p>
496 <p>This value can sometimes be useful as a generation counter of sorts (it
497 "ticks" the number of loop iterations), as it roughly corresponds with
498 <code>ev_prepare</code> and <code>ev_check</code> calls.</p>
500 <dt>unsigned int ev_backend (loop)</dt>
502 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
505 <dt>ev_tstamp ev_now (loop)</dt>
507 <p>Returns the current "event loop time", which is the time the event loop
508 received events and started processing them. This timestamp does not
509 change as long as callbacks are being processed, and this is also the base
510 time used for relative timers. You can treat it as the timestamp of the
511 event occuring (or more correctly, libev finding out about it).</p>
513 <dt>ev_loop (loop, int flags)</dt>
515 <p>Finally, this is it, the event handler. This function usually is called
516 after you initialised all your watchers and you want to start handling
518 <p>If the flags argument is specified as <code>0</code>, it will not return until
519 either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
520 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
521 relying on all watchers to be stopped when deciding when a program has
522 finished (especially in interactive programs), but having a program that
523 automatically loops as long as it has to and no longer by virtue of
524 relying on its watchers stopping correctly is a thing of beauty.</p>
525 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
526 those events and any outstanding ones, but will not block your process in
527 case there are no events and will return after one iteration of the loop.</p>
528 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
529 neccessary) and will handle those and any outstanding ones. It will block
530 your process until at least one new event arrives, and will return after
531 one iteration of the loop. This is useful if you are waiting for some
532 external event in conjunction with something not expressible using other
533 libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
534 usually a better approach for this kind of thing.</p>
535 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
536 <pre> - Before the first iteration, call any pending watchers.
537 * If there are no active watchers (reference count is zero), return.
538 - Queue all prepare watchers and then call all outstanding watchers.
539 - If we have been forked, recreate the kernel state.
540 - Update the kernel state with all outstanding changes.
541 - Update the "event loop time".
542 - Calculate for how long to block.
543 - Block the process, waiting for any events.
544 - Queue all outstanding I/O (fd) events.
545 - Update the "event loop time" and do time jump handling.
546 - Queue all outstanding timers.
547 - Queue all outstanding periodics.
548 - If no events are pending now, queue all idle watchers.
549 - Queue all check watchers.
550 - Call all queued watchers in reverse order (i.e. check watchers first).
551 Signals and child watchers are implemented as I/O watchers, and will
552 be handled here by queueing them when their watcher gets executed.
553 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
554 were used, return, otherwise continue with step *.
557 <p>Example: Queue some jobs and then loop until no events are outsanding
559 <pre> ... queue jobs here, make sure they register event watchers as long
560 ... as they still have work to do (even an idle watcher will do..)
561 ev_loop (my_loop, 0);
566 <dt>ev_unloop (loop, how)</dt>
568 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
569 has processed all outstanding events). The <code>how</code> argument must be either
570 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
571 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
573 <dt>ev_ref (loop)</dt>
574 <dt>ev_unref (loop)</dt>
576 <p>Ref/unref can be used to add or remove a reference count on the event
577 loop: Every watcher keeps one reference, and as long as the reference
578 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
579 a watcher you never unregister that should not keep <code>ev_loop</code> from
580 returning, ev_unref() after starting, and ev_ref() before stopping it. For
581 example, libev itself uses this for its internal signal pipe: It is not
582 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
583 no event watchers registered by it are active. It is also an excellent
584 way to do this for generic recurring timers or from within third-party
585 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
586 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
587 running when nothing else is active.</p>
588 <pre> struct ev_signal exitsig;
589 ev_signal_init (&exitsig, sig_cb, SIGINT);
590 ev_signal_start (loop, &exitsig);
594 <p>Example: For some weird reason, unregister the above signal handler again.</p>
596 ev_signal_stop (loop, &exitsig);
607 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
608 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
609 <p>A watcher is a structure that you create and register to record your
610 interest in some event. For instance, if you want to wait for STDIN to
611 become readable, you would create an <code>ev_io</code> watcher for that:</p>
612 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
615 ev_unloop (loop, EVUNLOOP_ALL);
618 struct ev_loop *loop = ev_default_loop (0);
619 struct ev_io stdin_watcher;
620 ev_init (&stdin_watcher, my_cb);
621 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
622 ev_io_start (loop, &stdin_watcher);
626 <p>As you can see, you are responsible for allocating the memory for your
627 watcher structures (and it is usually a bad idea to do this on the stack,
628 although this can sometimes be quite valid).</p>
629 <p>Each watcher structure must be initialised by a call to <code>ev_init
630 (watcher *, callback)</code>, which expects a callback to be provided. This
631 callback gets invoked each time the event occurs (or, in the case of io
632 watchers, each time the event loop detects that the file descriptor given
633 is readable and/or writable).</p>
634 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
635 with arguments specific to this watcher type. There is also a macro
636 to combine initialisation and setting in one call: <code>ev_<type>_init
637 (watcher *, callback, ...)</code>.</p>
638 <p>To make the watcher actually watch out for events, you have to start it
639 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
640 *)</code>), and you can stop watching for events at any time by calling the
641 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
642 <p>As long as your watcher is active (has been started but not stopped) you
643 must not touch the values stored in it. Most specifically you must never
644 reinitialise it or call its <code>set</code> macro.</p>
645 <p>Each and every callback receives the event loop pointer as first, the
646 registered watcher structure as second, and a bitset of received events as
648 <p>The received events usually include a single bit per event type received
649 (you can receive multiple events at the same time). The possible bit masks
652 <dt><code>EV_READ</code></dt>
653 <dt><code>EV_WRITE</code></dt>
655 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
658 <dt><code>EV_TIMEOUT</code></dt>
660 <p>The <code>ev_timer</code> watcher has timed out.</p>
662 <dt><code>EV_PERIODIC</code></dt>
664 <p>The <code>ev_periodic</code> watcher has timed out.</p>
666 <dt><code>EV_SIGNAL</code></dt>
668 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
670 <dt><code>EV_CHILD</code></dt>
672 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
674 <dt><code>EV_STAT</code></dt>
676 <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
678 <dt><code>EV_IDLE</code></dt>
680 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
682 <dt><code>EV_PREPARE</code></dt>
683 <dt><code>EV_CHECK</code></dt>
685 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
686 to gather new events, and all <code>ev_check</code> watchers are invoked just after
687 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
688 received events. Callbacks of both watcher types can start and stop as
689 many watchers as they want, and all of them will be taken into account
690 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
691 <code>ev_loop</code> from blocking).</p>
693 <dt><code>EV_EMBED</code></dt>
695 <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
697 <dt><code>EV_FORK</code></dt>
699 <p>The event loop has been resumed in the child process after fork (see
700 <code>ev_fork</code>).</p>
702 <dt><code>EV_ERROR</code></dt>
704 <p>An unspecified error has occured, the watcher has been stopped. This might
705 happen because the watcher could not be properly started because libev
706 ran out of memory, a file descriptor was found to be closed or any other
707 problem. You best act on it by reporting the problem and somehow coping
708 with the watcher being stopped.</p>
709 <p>Libev will usually signal a few "dummy" events together with an error,
710 for example it might indicate that a fd is readable or writable, and if
711 your callbacks is well-written it can just attempt the operation and cope
712 with the error from read() or write(). This will not work in multithreaded
713 programs, though, so beware.</p>
718 <h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
719 <div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
720 <p>In the following description, <code>TYPE</code> stands for the watcher type,
721 e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
723 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
725 <p>This macro initialises the generic portion of a watcher. The contents
726 of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
727 the generic parts of the watcher are initialised, you <i>need</i> to call
728 the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
729 type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
730 which rolls both calls into one.</p>
731 <p>You can reinitialise a watcher at any time as long as it has been stopped
732 (or never started) and there are no pending events outstanding.</p>
733 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
734 int revents)</code>.</p>
736 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
738 <p>This macro initialises the type-specific parts of a watcher. You need to
739 call <code>ev_init</code> at least once before you call this macro, but you can
740 call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
741 macro on a watcher that is active (it can be pending, however, which is a
742 difference to the <code>ev_init</code> macro).</p>
743 <p>Although some watcher types do not have type-specific arguments
744 (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
746 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
748 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
749 calls into a single call. This is the most convinient method to initialise
750 a watcher. The same limitations apply, of course.</p>
752 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
754 <p>Starts (activates) the given watcher. Only active watchers will receive
755 events. If the watcher is already active nothing will happen.</p>
757 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
759 <p>Stops the given watcher again (if active) and clears the pending
760 status. It is possible that stopped watchers are pending (for example,
761 non-repeating timers are being stopped when they become pending), but
762 <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
763 you want to free or reuse the memory used by the watcher it is therefore a
764 good idea to always call its <code>ev_TYPE_stop</code> function.</p>
766 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
768 <p>Returns a true value iff the watcher is active (i.e. it has been started
769 and not yet been stopped). As long as a watcher is active you must not modify
772 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
774 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
775 events but its callback has not yet been invoked). As long as a watcher
776 is pending (but not active) you must not call an init function on it (but
777 <code>ev_TYPE_set</code> is safe), you must not change its priority, and you must
778 make sure the watcher is available to libev (e.g. you cannot <code>free ()</code>
781 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
783 <p>Returns the callback currently set on the watcher.</p>
785 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
787 <p>Change the callback. You can change the callback at virtually any time
788 (modulo threads).</p>
790 <dt>ev_set_priority (ev_TYPE *watcher, priority)</dt>
791 <dt>int ev_priority (ev_TYPE *watcher)</dt>
793 <p>Set and query the priority of the watcher. The priority is a small
794 integer between <code>EV_MAXPRI</code> (default: <code>2</code>) and <code>EV_MINPRI</code>
795 (default: <code>-2</code>). Pending watchers with higher priority will be invoked
796 before watchers with lower priority, but priority will not keep watchers
797 from being executed (except for <code>ev_idle</code> watchers).</p>
798 <p>This means that priorities are <i>only</i> used for ordering callback
799 invocation after new events have been received. This is useful, for
800 example, to reduce latency after idling, or more often, to bind two
801 watchers on the same event and make sure one is called first.</p>
802 <p>If you need to suppress invocation when higher priority events are pending
803 you need to look at <code>ev_idle</code> watchers, which provide this functionality.</p>
804 <p>You <i>must not</i> change the priority of a watcher as long as it is active or
806 <p>The default priority used by watchers when no priority has been set is
807 always <code>0</code>, which is supposed to not be too high and not be too low :).</p>
808 <p>Setting a priority outside the range of <code>EV_MINPRI</code> to <code>EV_MAXPRI</code> is
809 fine, as long as you do not mind that the priority value you query might
810 or might not have been adjusted to be within valid range.</p>
812 <dt>ev_invoke (loop, ev_TYPE *watcher, int revents)</dt>
814 <p>Invoke the <code>watcher</code> with the given <code>loop</code> and <code>revents</code>. Neither
815 <code>loop</code> nor <code>revents</code> need to be valid as long as the watcher callback
816 can deal with that fact.</p>
818 <dt>int ev_clear_pending (loop, ev_TYPE *watcher)</dt>
820 <p>If the watcher is pending, this function returns clears its pending status
821 and returns its <code>revents</code> bitset (as if its callback was invoked). If the
822 watcher isn't pending it does nothing and returns <code>0</code>.</p>
831 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
832 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
833 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
834 and read at any time, libev will completely ignore it. This can be used
835 to associate arbitrary data with your watcher. If you need more data and
836 don't want to allocate memory and store a pointer to it in that data
837 member, you can also "subclass" the watcher type and provide your own
844 struct whatever *mostinteresting;
848 <p>And since your callback will be called with a pointer to the watcher, you
849 can cast it back to your own type:</p>
850 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
852 struct my_io *w = (struct my_io *)w_;
857 <p>More interesting and less C-conformant ways of casting your callback type
858 instead have been omitted.</p>
859 <p>Another common scenario is having some data structure with multiple
861 <pre> struct my_biggy
869 <p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
870 you need to use <code>offsetof</code>:</p>
871 <pre> #include <stddef.h>
874 t1_cb (EV_P_ struct ev_timer *w, int revents)
876 struct my_biggy big = (struct my_biggy *
877 (((char *)w) - offsetof (struct my_biggy, t1));
881 t2_cb (EV_P_ struct ev_timer *w, int revents)
883 struct my_biggy big = (struct my_biggy *
884 (((char *)w) - offsetof (struct my_biggy, t2));
893 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
894 <div id="WATCHER_TYPES_CONTENT">
895 <p>This section describes each watcher in detail, but will not repeat
896 information given in the last section. Any initialisation/set macros,
897 functions and members specific to the watcher type are explained.</p>
898 <p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
899 while the watcher is active, you can look at the member and expect some
900 sensible content, but you must not modify it (you can modify it while the
901 watcher is stopped to your hearts content), or <i>[read-write]</i>, which
902 means you can expect it to have some sensible content while the watcher
903 is active, but you can also modify it. Modifying it may not do something
904 sensible or take immediate effect (or do anything at all), but libev will
905 not crash or malfunction in any way.</p>
912 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
913 <div id="code_ev_io_code_is_this_file_descrip-2">
914 <p>I/O watchers check whether a file descriptor is readable or writable
915 in each iteration of the event loop, or, more precisely, when reading
916 would not block the process and writing would at least be able to write
917 some data. This behaviour is called level-triggering because you keep
918 receiving events as long as the condition persists. Remember you can stop
919 the watcher if you don't want to act on the event and neither want to
920 receive future events.</p>
921 <p>In general you can register as many read and/or write event watchers per
922 fd as you want (as long as you don't confuse yourself). Setting all file
923 descriptors to non-blocking mode is also usually a good idea (but not
924 required if you know what you are doing).</p>
925 <p>You have to be careful with dup'ed file descriptors, though. Some backends
926 (the linux epoll backend is a notable example) cannot handle dup'ed file
927 descriptors correctly if you register interest in two or more fds pointing
928 to the same underlying file/socket/etc. description (that is, they share
929 the same underlying "file open").</p>
930 <p>If you must do this, then force the use of a known-to-be-good backend
931 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
932 <code>EVBACKEND_POLL</code>).</p>
933 <p>Another thing you have to watch out for is that it is quite easy to
934 receive "spurious" readyness notifications, that is your callback might
935 be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
936 because there is no data. Not only are some backends known to create a
937 lot of those (for example solaris ports), it is very easy to get into
938 this situation even with a relatively standard program structure. Thus
939 it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
940 <code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
941 <p>If you cannot run the fd in non-blocking mode (for example you should not
942 play around with an Xlib connection), then you have to seperately re-test
943 whether a file descriptor is really ready with a known-to-be good interface
944 such as poll (fortunately in our Xlib example, Xlib already does this on
945 its own, so its quite safe to use).</p>
947 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
948 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
950 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
951 rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
952 <code>EV_READ | EV_WRITE</code> to receive the given events.</p>
954 <dt>int fd [read-only]</dt>
956 <p>The file descriptor being watched.</p>
958 <dt>int events [read-only]</dt>
960 <p>The events being watched.</p>
963 <p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
964 readable, but only once. Since it is likely line-buffered, you could
965 attempt to read a whole line in the callback.</p>
967 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
969 ev_io_stop (loop, w);
970 .. read from stdin here (or from w->fd) and haqndle any I/O errors
974 struct ev_loop *loop = ev_default_init (0);
975 struct ev_io stdin_readable;
976 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
977 ev_io_start (loop, &stdin_readable);
986 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
987 <div id="code_ev_timer_code_relative_and_opti-2">
988 <p>Timer watchers are simple relative timers that generate an event after a
989 given time, and optionally repeating in regular intervals after that.</p>
990 <p>The timers are based on real time, that is, if you register an event that
991 times out after an hour and you reset your system clock to last years
992 time, it will still time out after (roughly) and hour. "Roughly" because
993 detecting time jumps is hard, and some inaccuracies are unavoidable (the
994 monotonic clock option helps a lot here).</p>
995 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
996 time. This is usually the right thing as this timestamp refers to the time
997 of the event triggering whatever timeout you are modifying/starting. If
998 you suspect event processing to be delayed and you <i>need</i> to base the timeout
999 on the current time, use something like this to adjust for this:</p>
1000 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1003 <p>The callback is guarenteed to be invoked only when its timeout has passed,
1004 but if multiple timers become ready during the same loop iteration then
1005 order of execution is undefined.</p>
1007 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
1008 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
1010 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
1011 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
1012 timer will automatically be configured to trigger again <code>repeat</code> seconds
1013 later, again, and again, until stopped manually.</p>
1014 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
1015 configure a timer to trigger every 10 seconds, then it will trigger at
1016 exactly 10 second intervals. If, however, your program cannot keep up with
1017 the timer (because it takes longer than those 10 seconds to do stuff) the
1018 timer will not fire more than once per event loop iteration.</p>
1020 <dt>ev_timer_again (loop)</dt>
1022 <p>This will act as if the timer timed out and restart it again if it is
1023 repeating. The exact semantics are:</p>
1024 <p>If the timer is pending, its pending status is cleared.</p>
1025 <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
1026 <p>If the timer is repeating, either start it if necessary (with the
1027 <code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
1028 <p>This sounds a bit complicated, but here is a useful and typical
1029 example: Imagine you have a tcp connection and you want a so-called idle
1030 timeout, that is, you want to be called when there have been, say, 60
1031 seconds of inactivity on the socket. The easiest way to do this is to
1032 configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
1033 <code>ev_timer_again</code> each time you successfully read or write some data. If
1034 you go into an idle state where you do not expect data to travel on the
1035 socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
1036 automatically restart it if need be.</p>
1037 <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
1038 altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
1039 <pre> ev_timer_init (timer, callback, 0., 5.);
1040 ev_timer_again (loop, timer);
1042 timer->again = 17.;
1043 ev_timer_again (loop, timer);
1045 timer->again = 10.;
1046 ev_timer_again (loop, timer);
1049 <p>This is more slightly efficient then stopping/starting the timer each time
1050 you want to modify its timeout value.</p>
1052 <dt>ev_tstamp repeat [read-write]</dt>
1054 <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1055 or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1056 which is also when any modifications are taken into account.</p>
1059 <p>Example: Create a timer that fires after 60 seconds.</p>
1061 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1063 .. one minute over, w is actually stopped right here
1066 struct ev_timer mytimer;
1067 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1068 ev_timer_start (loop, &mytimer);
1071 <p>Example: Create a timeout timer that times out after 10 seconds of
1074 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1076 .. ten seconds without any activity
1079 struct ev_timer mytimer;
1080 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1081 ev_timer_again (&mytimer); /* start timer */
1084 // and in some piece of code that gets executed on any "activity":
1085 // reset the timeout to start ticking again at 10 seconds
1086 ev_timer_again (&mytimer);
1094 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
1095 <div id="code_ev_periodic_code_to_cron_or_not-2">
1096 <p>Periodic watchers are also timers of a kind, but they are very versatile
1097 (and unfortunately a bit complex).</p>
1098 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
1099 but on wallclock time (absolute time). You can tell a periodic watcher
1100 to trigger "at" some specific point in time. For example, if you tell a
1101 periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
1102 + 10.</code>) and then reset your system clock to the last year, then it will
1103 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
1104 roughly 10 seconds later).</p>
1105 <p>They can also be used to implement vastly more complex timers, such as
1106 triggering an event on each midnight, local time or other, complicated,
1108 <p>As with timers, the callback is guarenteed to be invoked only when the
1109 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1110 during the same loop iteration then order of execution is undefined.</p>
1112 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
1113 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
1115 <p>Lots of arguments, lets sort it out... There are basically three modes of
1116 operation, and we will explain them from simplest to complex:</p>
1119 <dt>* absolute timer (at = time, interval = reschedule_cb = 0)</dt>
1121 <p>In this configuration the watcher triggers an event at the wallclock time
1122 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
1123 that is, if it is to be run at January 1st 2011 then it will run when the
1124 system time reaches or surpasses this time.</p>
1126 <dt>* non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)</dt>
1128 <p>In this mode the watcher will always be scheduled to time out at the next
1129 <code>at + N * interval</code> time (for some integer N, which can also be negative)
1130 and then repeat, regardless of any time jumps.</p>
1131 <p>This can be used to create timers that do not drift with respect to system
1133 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
1136 <p>This doesn't mean there will always be 3600 seconds in between triggers,
1137 but only that the the callback will be called when the system time shows a
1138 full hour (UTC), or more correctly, when the system time is evenly divisible
1140 <p>Another way to think about it (for the mathematically inclined) is that
1141 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
1142 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1143 <p>For numerical stability it is preferable that the <code>at</code> value is near
1144 <code>ev_now ()</code> (the current time), but there is no range requirement for
1147 <dt>* manual reschedule mode (at and interval ignored, reschedule_cb = callback)</dt>
1149 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
1150 ignored. Instead, each time the periodic watcher gets scheduled, the
1151 reschedule callback will be called with the watcher as first, and the
1152 current time as second argument.</p>
1153 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
1154 ever, or make any event loop modifications</i>. If you need to stop it,
1155 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1156 starting an <code>ev_prepare</code> watcher, which is legal).</p>
1157 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1158 ev_tstamp now)</code>, e.g.:</p>
1159 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1165 <p>It must return the next time to trigger, based on the passed time value
1166 (that is, the lowest time value larger than to the second argument). It
1167 will usually be called just before the callback will be triggered, but
1168 might be called at other times, too.</p>
1169 <p>NOTE: <i>This callback must always return a time that is later than the
1170 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
1171 <p>This can be used to create very complex timers, such as a timer that
1172 triggers on each midnight, local time. To do this, you would calculate the
1173 next midnight after <code>now</code> and return the timestamp value for this. How
1174 you do this is, again, up to you (but it is not trivial, which is the main
1175 reason I omitted it as an example).</p>
1180 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
1182 <p>Simply stops and restarts the periodic watcher again. This is only useful
1183 when you changed some parameters or the reschedule callback would return
1184 a different time than the last time it was called (e.g. in a crond like
1185 program when the crontabs have changed).</p>
1187 <dt>ev_tstamp offset [read-write]</dt>
1189 <p>When repeating, this contains the offset value, otherwise this is the
1190 absolute point in time (the <code>at</code> value passed to <code>ev_periodic_set</code>).</p>
1191 <p>Can be modified any time, but changes only take effect when the periodic
1192 timer fires or <code>ev_periodic_again</code> is being called.</p>
1194 <dt>ev_tstamp interval [read-write]</dt>
1196 <p>The current interval value. Can be modified any time, but changes only
1197 take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1200 <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1202 <p>The current reschedule callback, or <code>0</code>, if this functionality is
1203 switched off. Can be changed any time, but changes only take effect when
1204 the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1207 <p>Example: Call a callback every hour, or, more precisely, whenever the
1208 system clock is divisible by 3600. The callback invocation times have
1209 potentially a lot of jittering, but good long-term stability.</p>
1211 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1213 ... its now a full hour (UTC, or TAI or whatever your clock follows)
1216 struct ev_periodic hourly_tick;
1217 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1218 ev_periodic_start (loop, &hourly_tick);
1221 <p>Example: The same as above, but use a reschedule callback to do it:</p>
1222 <pre> #include <math.h>
1225 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1227 return fmod (now, 3600.) + 3600.;
1230 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1233 <p>Example: Call a callback every hour, starting now:</p>
1234 <pre> struct ev_periodic hourly_tick;
1235 ev_periodic_init (&hourly_tick, clock_cb,
1236 fmod (ev_now (loop), 3600.), 3600., 0);
1237 ev_periodic_start (loop, &hourly_tick);
1245 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
1246 <div id="code_ev_signal_code_signal_me_when_a-2">
1247 <p>Signal watchers will trigger an event when the process receives a specific
1248 signal one or more times. Even though signals are very asynchronous, libev
1249 will try it's best to deliver signals synchronously, i.e. as part of the
1250 normal event processing, like any other event.</p>
1251 <p>You can configure as many watchers as you like per signal. Only when the
1252 first watcher gets started will libev actually register a signal watcher
1253 with the kernel (thus it coexists with your own signal handlers as long
1254 as you don't register any with libev). Similarly, when the last signal
1255 watcher for a signal is stopped libev will reset the signal handler to
1256 SIG_DFL (regardless of what it was set to before).</p>
1258 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1259 <dt>ev_signal_set (ev_signal *, int signum)</dt>
1261 <p>Configures the watcher to trigger on the given signal number (usually one
1262 of the <code>SIGxxx</code> constants).</p>
1264 <dt>int signum [read-only]</dt>
1266 <p>The signal the watcher watches out for.</p>
1275 <h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
1276 <div id="code_ev_child_code_watch_out_for_pro-2">
1277 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1278 some child status changes (most typically when a child of yours dies).</p>
1280 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1281 <dt>ev_child_set (ev_child *, int pid)</dt>
1283 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1284 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1285 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1286 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1287 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1288 process causing the status change.</p>
1290 <dt>int pid [read-only]</dt>
1292 <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1294 <dt>int rpid [read-write]</dt>
1296 <p>The process id that detected a status change.</p>
1298 <dt>int rstatus [read-write]</dt>
1300 <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1301 <code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
1304 <p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1306 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1308 ev_unloop (loop, EVUNLOOP_ALL);
1311 struct ev_signal signal_watcher;
1312 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1313 ev_signal_start (loop, &sigint_cb);
1321 <h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1322 <div id="code_ev_stat_code_did_the_file_attri-2">
1323 <p>This watches a filesystem path for attribute changes. That is, it calls
1324 <code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1325 compared to the last time, invoking the callback if it did.</p>
1326 <p>The path does not need to exist: changing from "path exists" to "path does
1327 not exist" is a status change like any other. The condition "path does
1328 not exist" is signified by the <code>st_nlink</code> field being zero (which is
1329 otherwise always forced to be at least one) and all the other fields of
1330 the stat buffer having unspecified contents.</p>
1331 <p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1332 relative and your working directory changes, the behaviour is undefined.</p>
1333 <p>Since there is no standard to do this, the portable implementation simply
1334 calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1335 can specify a recommended polling interval for this case. If you specify
1336 a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1337 unspecified default</i> value will be used (which you can expect to be around
1338 five seconds, although this might change dynamically). Libev will also
1339 impose a minimum interval which is currently around <code>0.1</code>, but thats
1340 usually overkill.</p>
1341 <p>This watcher type is not meant for massive numbers of stat watchers,
1342 as even with OS-supported change notifications, this can be
1343 resource-intensive.</p>
1344 <p>At the time of this writing, only the Linux inotify interface is
1345 implemented (implementing kqueue support is left as an exercise for the
1346 reader). Inotify will be used to give hints only and should not change the
1347 semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1348 to fall back to regular polling again even with inotify, but changes are
1349 usually detected immediately, and if the file exists there will be no
1352 <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1353 <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1355 <p>Configures the watcher to wait for status changes of the given
1356 <code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1357 be detected and should normally be specified as <code>0</code> to let libev choose
1358 a suitable value. The memory pointed to by <code>path</code> must point to the same
1359 path for as long as the watcher is active.</p>
1360 <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1361 relative to the attributes at the time the watcher was started (or the
1362 last change was detected).</p>
1364 <dt>ev_stat_stat (ev_stat *)</dt>
1366 <p>Updates the stat buffer immediately with new values. If you change the
1367 watched path in your callback, you could call this fucntion to avoid
1368 detecting this change (while introducing a race condition). Can also be
1369 useful simply to find out the new values.</p>
1371 <dt>ev_statdata attr [read-only]</dt>
1373 <p>The most-recently detected attributes of the file. Although the type is of
1374 <code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1375 suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1376 was some error while <code>stat</code>ing the file.</p>
1378 <dt>ev_statdata prev [read-only]</dt>
1380 <p>The previous attributes of the file. The callback gets invoked whenever
1381 <code>prev</code> != <code>attr</code>.</p>
1383 <dt>ev_tstamp interval [read-only]</dt>
1385 <p>The specified interval.</p>
1387 <dt>const char *path [read-only]</dt>
1389 <p>The filesystem path that is being watched.</p>
1392 <p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1394 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1396 /* /etc/passwd changed in some way */
1397 if (w->attr.st_nlink)
1399 printf ("passwd current size %ld\n", (long)w->attr.st_size);
1400 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1401 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1404 /* you shalt not abuse printf for puts */
1405 puts ("wow, /etc/passwd is not there, expect problems. "
1406 "if this is windows, they already arrived\n");
1412 ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1413 ev_stat_start (loop, &passwd);
1421 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
1422 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1423 <p>Idle watchers trigger events when no other events of the same or higher
1424 priority are pending (prepare, check and other idle watchers do not
1426 <p>That is, as long as your process is busy handling sockets or timeouts
1427 (or even signals, imagine) of the same or higher priority it will not be
1428 triggered. But when your process is idle (or only lower-priority watchers
1429 are pending), the idle watchers are being called once per event loop
1430 iteration - until stopped, that is, or your process receives more events
1431 and becomes busy again with higher priority stuff.</p>
1432 <p>The most noteworthy effect is that as long as any idle watchers are
1433 active, the process will not block when waiting for new events.</p>
1434 <p>Apart from keeping your process non-blocking (which is a useful
1435 effect on its own sometimes), idle watchers are a good place to do
1436 "pseudo-background processing", or delay processing stuff to after the
1437 event loop has handled all outstanding events.</p>
1439 <dt>ev_idle_init (ev_signal *, callback)</dt>
1441 <p>Initialises and configures the idle watcher - it has no parameters of any
1442 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1446 <p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1447 callback, free it. Also, use no error checking, as usual.</p>
1449 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1452 // now do something you wanted to do when the program has
1453 // no longer asnything immediate to do.
1456 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1457 ev_idle_init (idle_watcher, idle_cb);
1458 ev_idle_start (loop, idle_cb);
1466 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
1467 <div id="code_ev_prepare_code_and_code_ev_che-2">
1468 <p>Prepare and check watchers are usually (but not always) used in tandem:
1469 prepare watchers get invoked before the process blocks and check watchers
1471 <p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1472 the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1473 watchers. Other loops than the current one are fine, however. The
1474 rationale behind this is that you do not need to check for recursion in
1475 those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1476 <code>ev_check</code> so if you have one watcher of each kind they will always be
1477 called in pairs bracketing the blocking call.</p>
1478 <p>Their main purpose is to integrate other event mechanisms into libev and
1479 their use is somewhat advanced. This could be used, for example, to track
1480 variable changes, implement your own watchers, integrate net-snmp or a
1481 coroutine library and lots more. They are also occasionally useful if
1482 you cache some data and want to flush it before blocking (for example,
1483 in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1485 <p>This is done by examining in each prepare call which file descriptors need
1486 to be watched by the other library, registering <code>ev_io</code> watchers for
1487 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1488 provide just this functionality). Then, in the check watcher you check for
1489 any events that occured (by checking the pending status of all watchers
1490 and stopping them) and call back into the library. The I/O and timer
1491 callbacks will never actually be called (but must be valid nevertheless,
1492 because you never know, you know?).</p>
1493 <p>As another example, the Perl Coro module uses these hooks to integrate
1494 coroutines into libev programs, by yielding to other active coroutines
1495 during each prepare and only letting the process block if no coroutines
1496 are ready to run (it's actually more complicated: it only runs coroutines
1497 with priority higher than or equal to the event loop and one coroutine
1498 of lower priority, but only once, using idle watchers to keep the event
1499 loop from blocking if lower-priority coroutines are active, thus mapping
1500 low-priority coroutines to idle/background tasks).</p>
1501 <p>It is recommended to give <code>ev_check</code> watchers highest (<code>EV_MAXPRI</code>)
1502 priority, to ensure that they are being run before any other watchers
1503 after the poll. Also, <code>ev_check</code> watchers (and <code>ev_prepare</code> watchers,
1504 too) should not activate ("feed") events into libev. While libev fully
1505 supports this, they will be called before other <code>ev_check</code> watchers did
1506 their job. As <code>ev_check</code> watchers are often used to embed other event
1507 loops those other event loops might be in an unusable state until their
1508 <code>ev_check</code> watcher ran (always remind yourself to coexist peacefully with
1511 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1512 <dt>ev_check_init (ev_check *, callback)</dt>
1514 <p>Initialises and configures the prepare or check watcher - they have no
1515 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1516 macros, but using them is utterly, utterly and completely pointless.</p>
1519 <p>There are a number of principal ways to embed other event loops or modules
1520 into libev. Here are some ideas on how to include libadns into libev
1521 (there is a Perl module named <code>EV::ADNS</code> that does this, which you could
1522 use for an actually working example. Another Perl module named <code>EV::Glib</code>
1523 embeds a Glib main context into libev, and finally, <code>Glib::EV</code> embeds EV
1524 into the Glib event loop).</p>
1525 <p>Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1526 and in a check watcher, destroy them and call into libadns. What follows
1527 is pseudo-code only of course. This requires you to either use a low
1528 priority for the check watcher or use <code>ev_clear_pending</code> explicitly, as
1529 the callbacks for the IO/timeout watchers might not have been called yet.</p>
1530 <pre> static ev_io iow [nfd];
1534 io_cb (ev_loop *loop, ev_io *w, int revents)
1538 // create io watchers for each fd and a timer before blocking
1540 adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1542 int timeout = 3600000;
1543 struct pollfd fds [nfd];
1544 // actual code will need to loop here and realloc etc.
1545 adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1547 /* the callback is illegal, but won't be called as we stop during check */
1548 ev_timer_init (&tw, 0, timeout * 1e-3);
1549 ev_timer_start (loop, &tw);
1551 // create one ev_io per pollfd
1552 for (int i = 0; i < nfd; ++i)
1554 ev_io_init (iow + i, io_cb, fds [i].fd,
1555 ((fds [i].events & POLLIN ? EV_READ : 0)
1556 | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1558 fds [i].revents = 0;
1559 ev_io_start (loop, iow + i);
1563 // stop all watchers after blocking
1565 adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1567 ev_timer_stop (loop, &tw);
1569 for (int i = 0; i < nfd; ++i)
1571 // set the relevant poll flags
1572 // could also call adns_processreadable etc. here
1573 struct pollfd *fd = fds + i;
1574 int revents = ev_clear_pending (iow + i);
1575 if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1576 if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1578 // now stop the watcher
1579 ev_io_stop (loop, iow + i);
1582 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1586 <p>Method 2: This would be just like method 1, but you run <code>adns_afterpoll</code>
1587 in the prepare watcher and would dispose of the check watcher.</p>
1588 <p>Method 3: If the module to be embedded supports explicit event
1589 notification (adns does), you can also make use of the actual watcher
1590 callbacks, and only destroy/create the watchers in the prepare watcher.</p>
1592 timer_cb (EV_P_ ev_timer *w, int revents)
1594 adns_state ads = (adns_state)w->data;
1597 adns_processtimeouts (ads, &tv_now);
1601 io_cb (EV_P_ ev_io *w, int revents)
1603 adns_state ads = (adns_state)w->data;
1606 if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1607 if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1610 // do not ever call adns_afterpoll
1613 <p>Method 4: Do not use a prepare or check watcher because the module you
1614 want to embed is too inflexible to support it. Instead, youc na override
1615 their poll function. The drawback with this solution is that the main
1616 loop is now no longer controllable by EV. The <code>Glib::EV</code> module does
1619 event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1623 for (n = 0; n < nfds; ++n)
1624 // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1626 if (timeout >= 0)
1627 // create/start timer
1633 if (timeout >= 0)
1634 ev_timer_stop (EV_A_ &to);
1636 // stop io watchers again - their callbacks should have set
1637 for (n = 0; n < nfds; ++n)
1638 ev_io_stop (EV_A_ iow [n]);
1649 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1650 <div id="code_ev_embed_code_when_one_backend_-2">
1651 <p>This is a rather advanced watcher type that lets you embed one event loop
1652 into another (currently only <code>ev_io</code> events are supported in the embedded
1653 loop, other types of watchers might be handled in a delayed or incorrect
1654 fashion and must not be used).</p>
1655 <p>There are primarily two reasons you would want that: work around bugs and
1657 <p>As an example for a bug workaround, the kqueue backend might only support
1658 sockets on some platform, so it is unusable as generic backend, but you
1659 still want to make use of it because you have many sockets and it scales
1660 so nicely. In this case, you would create a kqueue-based loop and embed it
1661 into your default loop (which might use e.g. poll). Overall operation will
1662 be a bit slower because first libev has to poll and then call kevent, but
1663 at least you can use both at what they are best.</p>
1664 <p>As for prioritising I/O: rarely you have the case where some fds have
1665 to be watched and handled very quickly (with low latency), and even
1666 priorities and idle watchers might have too much overhead. In this case
1667 you would put all the high priority stuff in one loop and all the rest in
1668 a second one, and embed the second one in the first.</p>
1669 <p>As long as the watcher is active, the callback will be invoked every time
1670 there might be events pending in the embedded loop. The callback must then
1671 call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1672 their callbacks (you could also start an idle watcher to give the embedded
1673 loop strictly lower priority for example). You can also set the callback
1674 to <code>0</code>, in which case the embed watcher will automatically execute the
1675 embedded loop sweep.</p>
1676 <p>As long as the watcher is started it will automatically handle events. The
1677 callback will be invoked whenever some events have been handled. You can
1678 set the callback to <code>0</code> to avoid having to specify one if you are not
1679 interested in that.</p>
1680 <p>Also, there have not currently been made special provisions for forking:
1681 when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1682 but you will also have to stop and restart any <code>ev_embed</code> watchers
1684 <p>Unfortunately, not all backends are embeddable, only the ones returned by
1685 <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1687 <p>So when you want to use this feature you will always have to be prepared
1688 that you cannot get an embeddable loop. The recommended way to get around
1689 this is to have a separate variables for your embeddable loop, try to
1690 create it, and if that fails, use the normal loop for everything:</p>
1691 <pre> struct ev_loop *loop_hi = ev_default_init (0);
1692 struct ev_loop *loop_lo = 0;
1693 struct ev_embed embed;
1695 // see if there is a chance of getting one that works
1696 // (remember that a flags value of 0 means autodetection)
1697 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1698 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1701 // if we got one, then embed it, otherwise default to loop_hi
1704 ev_embed_init (&embed, 0, loop_lo);
1705 ev_embed_start (loop_hi, &embed);
1712 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1713 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1715 <p>Configures the watcher to embed the given loop, which must be
1716 embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1717 invoked automatically, otherwise it is the responsibility of the callback
1718 to invoke it (it will continue to be called until the sweep has been done,
1719 if you do not want thta, you need to temporarily stop the embed watcher).</p>
1721 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1723 <p>Make a single, non-blocking sweep over the embedded loop. This works
1724 similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1725 apropriate way for embedded loops.</p>
1727 <dt>struct ev_loop *loop [read-only]</dt>
1729 <p>The embedded event loop.</p>
1738 <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>
1739 <div id="code_ev_fork_code_the_audacity_to_re-2">
1740 <p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1741 whoever is a good citizen cared to tell libev about it by calling
1742 <code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1743 event loop blocks next and before <code>ev_check</code> watchers are being called,
1744 and only in the child after the fork. If whoever good citizen calling
1745 <code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1746 handlers will be invoked, too, of course.</p>
1748 <dt>ev_fork_init (ev_signal *, callback)</dt>
1750 <p>Initialises and configures the fork watcher - it has no parameters of any
1751 kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1761 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
1762 <div id="OTHER_FUNCTIONS_CONTENT">
1763 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1765 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1767 <p>This function combines a simple timer and an I/O watcher, calls your
1768 callback on whichever event happens first and automatically stop both
1769 watchers. This is useful if you want to wait for a single event on an fd
1770 or timeout without having to allocate/configure/start/stop/free one or
1771 more watchers yourself.</p>
1772 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1773 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1774 <code>events</code> set will be craeted and started.</p>
1775 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1776 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1777 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1779 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1780 passed an <code>revents</code> set like normal event callbacks (a combination of
1781 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1782 value passed to <code>ev_once</code>:</p>
1783 <pre> static void stdin_ready (int revents, void *arg)
1785 if (revents & EV_TIMEOUT)
1786 /* doh, nothing entered */;
1787 else if (revents & EV_READ)
1788 /* stdin might have data for us, joy! */;
1791 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1795 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1797 <p>Feeds the given event set into the event loop, as if the specified event
1798 had happened for the specified watcher (which must be a pointer to an
1799 initialised but not necessarily started event watcher).</p>
1801 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1803 <p>Feed an event on the given fd, as if a file descriptor backend detected
1804 the given events it.</p>
1806 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1808 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1818 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1819 <div id="LIBEVENT_EMULATION_CONTENT">
1820 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1821 emulate the internals of libevent, so here are some usage hints:</p>
1823 <dt>* Use it by including <event.h>, as usual.</dt>
1824 <dt>* The following members are fully supported: ev_base, ev_callback,
1825 ev_arg, ev_fd, ev_res, ev_events.</dt>
1826 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1827 maintained by libev, it does not work exactly the same way as in libevent (consider
1828 it a private API).</dt>
1829 <dt>* Priorities are not currently supported. Initialising priorities
1830 will fail and all watchers will have the same priority, even though there
1831 is an ev_pri field.</dt>
1832 <dt>* Other members are not supported.</dt>
1833 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1834 to use the libev header file and library.</dt>
1838 <h1 id="C_SUPPORT">C++ SUPPORT</h1>
1839 <div id="C_SUPPORT_CONTENT">
1840 <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1841 you to use some convinience methods to start/stop watchers and also change
1842 the callback model to a model using method callbacks on objects.</p>
1844 <pre> #include <ev++.h>
1847 <p>This automatically includes <cite>ev.h</cite> and puts all of its definitions (many
1848 of them macros) into the global namespace. All C++ specific things are
1849 put into the <code>ev</code> namespace. It should support all the same embedding
1850 options as <cite>ev.h</cite>, most notably <code>EV_MULTIPLICITY</code>.</p>
1851 <p>Care has been taken to keep the overhead low. The only data member the C++
1852 classes add (compared to plain C-style watchers) is the event loop pointer
1853 that the watcher is associated with (or no additional members at all if
1854 you disable <code>EV_MULTIPLICITY</code> when embedding libev).</p>
1855 <p>Currently, functions, and static and non-static member functions can be
1856 used as callbacks. Other types should be easy to add as long as they only
1857 need one additional pointer for context. If you need support for other
1858 types of functors please contact the author (preferably after implementing
1860 <p>Here is a list of things available in the <code>ev</code> namespace:</p>
1862 <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1864 <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1865 macros from <cite>ev.h</cite>.</p>
1867 <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1869 <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1871 <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1873 <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1874 the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1875 which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1876 defines by many implementations.</p>
1877 <p>All of those classes have these methods:</p>
1880 <dt>ev::TYPE::TYPE ()</dt>
1881 <dt>ev::TYPE::TYPE (struct ev_loop *)</dt>
1882 <dt>ev::TYPE::~TYPE</dt>
1884 <p>The constructor (optionally) takes an event loop to associate the watcher
1885 with. If it is omitted, it will use <code>EV_DEFAULT</code>.</p>
1886 <p>The constructor calls <code>ev_init</code> for you, which means you have to call the
1887 <code>set</code> method before starting it.</p>
1888 <p>It will not set a callback, however: You have to call the templated <code>set</code>
1889 method to set a callback before you can start the watcher.</p>
1890 <p>(The reason why you have to use a method is a limitation in C++ which does
1891 not allow explicit template arguments for constructors).</p>
1892 <p>The destructor automatically stops the watcher if it is active.</p>
1894 <dt>w->set<class, &class::method> (object *)</dt>
1896 <p>This method sets the callback method to call. The method has to have a
1897 signature of <code>void (*)(ev_TYPE &, int)</code>, it receives the watcher as
1898 first argument and the <code>revents</code> as second. The object must be given as
1899 parameter and is stored in the <code>data</code> member of the watcher.</p>
1900 <p>This method synthesizes efficient thunking code to call your method from
1901 the C callback that libev requires. If your compiler can inline your
1902 callback (i.e. it is visible to it at the place of the <code>set</code> call and
1903 your compiler is good :), then the method will be fully inlined into the
1904 thunking function, making it as fast as a direct C callback.</p>
1905 <p>Example: simple class declaration and watcher initialisation</p>
1906 <pre> struct myclass
1908 void io_cb (ev::io &w, int revents) { }
1913 iow.set <myclass, &myclass::io_cb> (&obj);
1917 <dt>w->set<function> (void *data = 0)</dt>
1919 <p>Also sets a callback, but uses a static method or plain function as
1920 callback. The optional <code>data</code> argument will be stored in the watcher's
1921 <code>data</code> member and is free for you to use.</p>
1922 <p>The prototype of the <code>function</code> must be <code>void (*)(ev::TYPE &w, int)</code>.</p>
1923 <p>See the method-<code>set</code> above for more details.</p>
1925 <pre> static void io_cb (ev::io &w, int revents) { }
1926 iow.set <io_cb> ();
1930 <dt>w->set (struct ev_loop *)</dt>
1932 <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1933 do this when the watcher is inactive (and not pending either).</p>
1935 <dt>w->set ([args])</dt>
1937 <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1938 called at least once. Unlike the C counterpart, an active watcher gets
1939 automatically stopped and restarted when reconfiguring it with this
1942 <dt>w->start ()</dt>
1944 <p>Starts the watcher. Note that there is no <code>loop</code> argument, as the
1945 constructor already stores the event loop.</p>
1947 <dt>w->stop ()</dt>
1949 <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1951 <dt>w->again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1953 <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1954 <code>ev_TYPE_again</code> function.</p>
1956 <dt>w->sweep () <code>ev::embed</code> only</dt>
1958 <p>Invokes <code>ev_embed_sweep</code>.</p>
1960 <dt>w->update () <code>ev::stat</code> only</dt>
1962 <p>Invokes <code>ev_stat_stat</code>.</p>
1968 <p>Example: Define a class with an IO and idle watcher, start one of them in
1969 the constructor.</p>
1972 ev_io io; void io_cb (ev::io &w, int revents);
1973 ev_idle idle void idle_cb (ev::idle &w, int revents);
1978 myclass::myclass (int fd)
1980 io .set <myclass, &myclass::io_cb > (this);
1981 idle.set <myclass, &myclass::idle_cb> (this);
1983 io.start (fd, ev::READ);
1992 <h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
1993 <div id="MACRO_MAGIC_CONTENT">
1994 <p>Libev can be compiled with a variety of options, the most fundemantal is
1995 <code>EV_MULTIPLICITY</code>. This option determines whether (most) functions and
1996 callbacks have an initial <code>struct ev_loop *</code> argument.</p>
1997 <p>To make it easier to write programs that cope with either variant, the
1998 following macros are defined:</p>
2000 <dt><code>EV_A</code>, <code>EV_A_</code></dt>
2002 <p>This provides the loop <i>argument</i> for functions, if one is required ("ev
2003 loop argument"). The <code>EV_A</code> form is used when this is the sole argument,
2004 <code>EV_A_</code> is used when other arguments are following. Example:</p>
2005 <pre> ev_unref (EV_A);
2006 ev_timer_add (EV_A_ watcher);
2010 <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
2011 which is often provided by the following macro.</p>
2013 <dt><code>EV_P</code>, <code>EV_P_</code></dt>
2015 <p>This provides the loop <i>parameter</i> for functions, if one is required ("ev
2016 loop parameter"). The <code>EV_P</code> form is used when this is the sole parameter,
2017 <code>EV_P_</code> is used when other parameters are following. Example:</p>
2018 <pre> // this is how ev_unref is being declared
2019 static void ev_unref (EV_P);
2021 // this is how you can declare your typical callback
2022 static void cb (EV_P_ ev_timer *w, int revents)
2025 <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
2026 suitable for use with <code>EV_A</code>.</p>
2028 <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
2030 <p>Similar to the other two macros, this gives you the value of the default
2031 loop, if multiple loops are supported ("ev loop default").</p>
2034 <p>Example: Declare and initialise a check watcher, utilising the above
2035 macros so it will work regardless of whether multiple loops are supported
2038 check_cb (EV_P_ ev_timer *w, int revents)
2040 ev_check_stop (EV_A_ w);
2044 ev_check_init (&check, check_cb);
2045 ev_check_start (EV_DEFAULT_ &check);
2046 ev_loop (EV_DEFAULT_ 0);
2051 <h1 id="EMBEDDING">EMBEDDING</h1>
2052 <div id="EMBEDDING_CONTENT">
2053 <p>Libev can (and often is) directly embedded into host
2054 applications. Examples of applications that embed it include the Deliantra
2055 Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2056 and rxvt-unicode.</p>
2057 <p>The goal is to enable you to just copy the neecssary files into your
2058 source directory without having to change even a single line in them, so
2059 you can easily upgrade by simply copying (or having a checked-out copy of
2060 libev somewhere in your source tree).</p>
2063 <h2 id="FILESETS">FILESETS</h2>
2064 <div id="FILESETS_CONTENT">
2065 <p>Depending on what features you need you need to include one or more sets of files
2069 <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
2070 <div id="CORE_EVENT_LOOP_CONTENT">
2071 <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
2072 configuration (no autoconf):</p>
2073 <pre> #define EV_STANDALONE 1
2074 #include "ev.c"
2077 <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
2078 single C source file only to provide the function implementations. To use
2079 it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
2080 done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
2081 where you can put other configuration options):</p>
2082 <pre> #define EV_STANDALONE 1
2083 #include "ev.h"
2086 <p>Both header files and implementation files can be compiled with a C++
2087 compiler (at least, thats a stated goal, and breakage will be treated
2089 <p>You need the following files in your source tree, or in a directory
2090 in your include path (e.g. in libev/ when using -Ilibev):</p>
2096 ev_win32.c required on win32 platforms only
2098 ev_select.c only when select backend is enabled (which is enabled by default)
2099 ev_poll.c only when poll backend is enabled (disabled by default)
2100 ev_epoll.c only when the epoll backend is enabled (disabled by default)
2101 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2102 ev_port.c only when the solaris port backend is enabled (disabled by default)
2105 <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
2106 to compile this single file.</p>
2109 <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
2110 <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
2111 <p>To include the libevent compatibility API, also include:</p>
2112 <pre> #include "event.c"
2115 <p>in the file including <cite>ev.c</cite>, and:</p>
2116 <pre> #include "event.h"
2119 <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
2120 <p>You need the following additional files for this:</p>
2127 <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
2128 <div id="AUTOCONF_SUPPORT_CONTENT">
2129 <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
2130 whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
2131 <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
2132 include <cite>config.h</cite> and configure itself accordingly.</p>
2133 <p>For this of course you need the m4 file:</p>
2139 <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
2140 <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
2141 <p>Libev can be configured via a variety of preprocessor symbols you have to define
2142 before including any of its files. The default is not to build for multiplicity
2143 and only include the select backend.</p>
2145 <dt>EV_STANDALONE</dt>
2147 <p>Must always be <code>1</code> if you do not use autoconf configuration, which
2148 keeps libev from including <cite>config.h</cite>, and it also defines dummy
2149 implementations for some libevent functions (such as logging, which is not
2150 supported). It will also not define any of the structs usually found in
2151 <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
2153 <dt>EV_USE_MONOTONIC</dt>
2155 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2156 monotonic clock option at both compiletime and runtime. Otherwise no use
2157 of the monotonic clock option will be attempted. If you enable this, you
2158 usually have to link against librt or something similar. Enabling it when
2159 the functionality isn't available is safe, though, althoguh you have
2160 to make sure you link against any libraries where the <code>clock_gettime</code>
2161 function is hiding in (often <cite>-lrt</cite>).</p>
2163 <dt>EV_USE_REALTIME</dt>
2165 <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2166 realtime clock option at compiletime (and assume its availability at
2167 runtime if successful). Otherwise no use of the realtime clock option will
2168 be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
2169 (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
2170 in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
2172 <dt>EV_USE_SELECT</dt>
2174 <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
2175 <code>select</code>(2) backend. No attempt at autodetection will be done: if no
2176 other method takes over, select will be it. Otherwise the select backend
2177 will not be compiled in.</p>
2179 <dt>EV_SELECT_USE_FD_SET</dt>
2181 <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
2182 structure. This is useful if libev doesn't compile due to a missing
2183 <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
2184 exotic systems. This usually limits the range of file descriptors to some
2185 low limit such as 1024 or might have other limitations (winsocket only
2186 allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
2187 influence the size of the <code>fd_set</code> used.</p>
2189 <dt>EV_SELECT_IS_WINSOCKET</dt>
2191 <p>When defined to <code>1</code>, the select backend will assume that
2192 select/socket/connect etc. don't understand file descriptors but
2193 wants osf handles on win32 (this is the case when the select to
2194 be used is the winsock select). This means that it will call
2195 <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2196 it is assumed that all these functions actually work on fds, even
2197 on win32. Should not be defined on non-win32 platforms.</p>
2199 <dt>EV_USE_POLL</dt>
2201 <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2202 backend. Otherwise it will be enabled on non-win32 platforms. It
2203 takes precedence over select.</p>
2205 <dt>EV_USE_EPOLL</dt>
2207 <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2208 <code>epoll</code>(7) backend. Its availability will be detected at runtime,
2209 otherwise another method will be used as fallback. This is the
2210 preferred backend for GNU/Linux systems.</p>
2212 <dt>EV_USE_KQUEUE</dt>
2214 <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2215 <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2216 otherwise another method will be used as fallback. This is the preferred
2217 backend for BSD and BSD-like systems, although on most BSDs kqueue only
2218 supports some types of fds correctly (the only platform we found that
2219 supports ptys for example was NetBSD), so kqueue might be compiled in, but
2220 not be used unless explicitly requested. The best way to use it is to find
2221 out whether kqueue supports your type of fd properly and use an embedded
2224 <dt>EV_USE_PORT</dt>
2226 <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
2227 10 port style backend. Its availability will be detected at runtime,
2228 otherwise another method will be used as fallback. This is the preferred
2229 backend for Solaris 10 systems.</p>
2231 <dt>EV_USE_DEVPOLL</dt>
2233 <p>reserved for future expansion, works like the USE symbols above.</p>
2235 <dt>EV_USE_INOTIFY</dt>
2237 <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2238 interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2239 be detected at runtime.</p>
2243 <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2244 undefined is <code><ev.h></code> in <cite>event.h</cite> and <code>"ev.h"</code> in <cite>ev.c</cite>. This
2245 can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2247 <dt>EV_CONFIG_H</dt>
2249 <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2250 <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2251 <code>EV_H</code>, above.</p>
2255 <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2256 of how the <cite>event.h</cite> header can be found.</p>
2258 <dt>EV_PROTOTYPES</dt>
2260 <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2261 prototypes, but still define all the structs and other symbols. This is
2262 occasionally useful if you want to provide your own wrapper functions
2263 around libev functions.</p>
2265 <dt>EV_MULTIPLICITY</dt>
2267 <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2268 will have the <code>struct ev_loop *</code> as first argument, and you can create
2269 additional independent event loops. Otherwise there will be no support
2270 for multiple event loops and there is no first event loop pointer
2271 argument. Instead, all functions act on the single default loop.</p>
2276 <p>The range of allowed priorities. <code>EV_MINPRI</code> must be smaller or equal to
2277 <code>EV_MAXPRI</code>, but otherwise there are no non-obvious limitations. You can
2278 provide for more priorities by overriding those symbols (usually defined
2279 to be <code>-2</code> and <code>2</code>, respectively).</p>
2280 <p>When doing priority-based operations, libev usually has to linearly search
2281 all the priorities, so having many of them (hundreds) uses a lot of space
2282 and time, so using the defaults of five priorities (-2 .. +2) is usually
2284 <p>If your embedding app does not need any priorities, defining these both to
2285 <code>0</code> will save some memory and cpu.</p>
2287 <dt>EV_PERIODIC_ENABLE</dt>
2289 <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2290 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2293 <dt>EV_IDLE_ENABLE</dt>
2295 <p>If undefined or defined to be <code>1</code>, then idle watchers are supported. If
2296 defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2299 <dt>EV_EMBED_ENABLE</dt>
2301 <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2302 defined to be <code>0</code>, then they are not.</p>
2304 <dt>EV_STAT_ENABLE</dt>
2306 <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2307 defined to be <code>0</code>, then they are not.</p>
2309 <dt>EV_FORK_ENABLE</dt>
2311 <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2312 defined to be <code>0</code>, then they are not.</p>
2316 <p>If you need to shave off some kilobytes of code at the expense of some
2317 speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2318 some inlining decisions, saves roughly 30% codesize of amd64.</p>
2320 <dt>EV_PID_HASHSIZE</dt>
2322 <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2323 pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2324 than enough. If you need to manage thousands of children you might want to
2325 increase this value (<i>must</i> be a power of two).</p>
2327 <dt>EV_INOTIFY_HASHSIZE</dt>
2329 <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2330 inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2331 usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2332 watchers you might want to increase this value (<i>must</i> be a power of
2337 <p>By default, all watchers have a <code>void *data</code> member. By redefining
2338 this macro to a something else you can include more and other types of
2339 members. You have to define it each time you include one of the files,
2340 though, and it must be identical each time.</p>
2341 <p>For example, the perl EV module uses something like this:</p>
2342 <pre> #define EV_COMMON \
2343 SV *self; /* contains this struct */ \
2344 SV *cb_sv, *fh /* note no trailing ";" */
2348 <dt>EV_CB_DECLARE (type)</dt>
2349 <dt>EV_CB_INVOKE (watcher, revents)</dt>
2350 <dt>ev_set_cb (ev, cb)</dt>
2352 <p>Can be used to change the callback member declaration in each watcher,
2353 and the way callbacks are invoked and set. Must expand to a struct member
2354 definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2355 their default definitions. One possible use for overriding these is to
2356 avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2357 method calls instead of plain function calls in C++.</p>
2360 <h2 id="EXAMPLES">EXAMPLES</h2>
2361 <div id="EXAMPLES_CONTENT">
2362 <p>For a real-world example of a program the includes libev
2363 verbatim, you can have a look at the EV perl module
2364 (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2365 the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2366 interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2367 will be compiled. It is pretty complex because it provides its own header
2369 <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2370 that everybody includes and which overrides some configure choices:</p>
2371 <pre> #define EV_MINIMAL 1
2372 #define EV_USE_POLL 0
2373 #define EV_MULTIPLICITY 0
2374 #define EV_PERIODIC_ENABLE 0
2375 #define EV_STAT_ENABLE 0
2376 #define EV_FORK_ENABLE 0
2377 #define EV_CONFIG_H <config.h>
2381 #include "ev++.h"
2384 <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2385 <pre> #include "ev_cpp.h"
2386 #include "ev.c"
2394 <h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2395 <div id="COMPLEXITIES_CONTENT">
2396 <p>In this section the complexities of (many of) the algorithms used inside
2397 libev will be explained. For complexity discussions about backends see the
2398 documentation for <code>ev_default_init</code>.</p>
2399 <p>All of the following are about amortised time: If an array needs to be
2400 extended, libev needs to realloc and move the whole array, but this
2401 happens asymptotically never with higher number of elements, so O(1) might
2402 mean it might do a lengthy realloc operation in rare cases, but on average
2403 it is much faster and asymptotically approaches constant time.</p>
2406 <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2408 <p>This means that, when you have a watcher that triggers in one hour and
2409 there are 100 watchers that would trigger before that then inserting will
2410 have to skip those 100 watchers.</p>
2412 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2414 <p>That means that for changing a timer costs less than removing/adding them
2415 as only the relative motion in the event queue has to be paid for.</p>
2417 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2419 <p>These just add the watcher into an array or at the head of a list.
2420 =item Stopping check/prepare/idle watchers: O(1)</p>
2422 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2424 <p>These watchers are stored in lists then need to be walked to find the
2425 correct watcher to remove. The lists are usually short (you don't usually
2426 have many watchers waiting for the same fd or signal).</p>
2428 <dt>Finding the next timer per loop iteration: O(1)</dt>
2429 <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2431 <p>A change means an I/O watcher gets started or stopped, which requires
2432 libev to recalculate its status (and possibly tell the kernel).</p>
2434 <dt>Activating one watcher: O(1)</dt>
2435 <dt>Priority handling: O(number_of_priorities)</dt>
2437 <p>Priorities are implemented by allocating some space for each
2438 priority. When doing priority-based operations, libev usually has to
2439 linearly search all the priorities.</p>
2449 <h1 id="AUTHOR">AUTHOR</h1>
2450 <div id="AUTHOR_CONTENT">
2451 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob