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="Sat Nov 24 08:13:46 2007" />
10 <meta name="generator" content="Pod::Xhtml 1.57" />
11 <link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
15 <h3 id="TOP">Index</h3>
17 <ul><li><a href="#NAME">NAME</a></li>
18 <li><a href="#SYNOPSIS">SYNOPSIS</a></li>
19 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
20 <li><a href="#FEATURES">FEATURES</a></li>
21 <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
22 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
23 <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
24 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
25 <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
26 <ul><li><a href="#SUMMARY_OF_GENERIC_WATCHER_FUNCTIONS">SUMMARY OF GENERIC WATCHER FUNCTIONS</a></li>
27 <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
30 <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
31 <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</a></li>
32 <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</a></li>
33 <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</a></li>
34 <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</a></li>
35 <li><a href="#code_ev_child_code_wait_for_pid_stat"><code>ev_child</code> - wait for pid status changes</a></li>
36 <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>
37 <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>
38 <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="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
42 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
43 <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
44 <li><a href="#AUTHOR">AUTHOR</a>
49 <h1 id="NAME">NAME</h1><p><a href="#TOP" class="toplink">Top</a></p>
50 <div id="NAME_CONTENT">
51 <p>libev - a high performance full-featured event loop written in C</p>
54 <h1 id="SYNOPSIS">SYNOPSIS</h1><p><a href="#TOP" class="toplink">Top</a></p>
55 <div id="SYNOPSIS_CONTENT">
56 <pre> #include <ev.h>
61 <h1 id="DESCRIPTION">DESCRIPTION</h1><p><a href="#TOP" class="toplink">Top</a></p>
62 <div id="DESCRIPTION_CONTENT">
63 <p>Libev is an event loop: you register interest in certain events (such as a
64 file descriptor being readable or a timeout occuring), and it will manage
65 these event sources and provide your program with events.</p>
66 <p>To do this, it must take more or less complete control over your process
67 (or thread) by executing the <i>event loop</i> handler, and will then
68 communicate events via a callback mechanism.</p>
69 <p>You register interest in certain events by registering so-called <i>event
70 watchers</i>, which are relatively small C structures you initialise with the
71 details of the event, and then hand it over to libev by <i>starting</i> the
75 <h1 id="FEATURES">FEATURES</h1><p><a href="#TOP" class="toplink">Top</a></p>
76 <div id="FEATURES_CONTENT">
77 <p>Libev supports select, poll, the linux-specific epoll and the bsd-specific
78 kqueue mechanisms for file descriptor events, relative timers, absolute
79 timers with customised rescheduling, signal events, process status change
80 events (related to SIGCHLD), and event watchers dealing with the event
81 loop mechanism itself (idle, prepare and check watchers). It also is quite
82 fast (see this <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing
83 it to libevent for example).</p>
86 <h1 id="CONVENTIONS">CONVENTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
87 <div id="CONVENTIONS_CONTENT">
88 <p>Libev is very configurable. In this manual the default configuration
89 will be described, which supports multiple event loops. For more info
90 about various configuration options please have a look at the file
91 <cite>README.embed</cite> in the libev distribution. If libev was configured without
92 support for multiple event loops, then all functions taking an initial
93 argument of name <code>loop</code> (which is always of type <code>struct ev_loop *</code>)
94 will not have this argument.</p>
97 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
98 <div id="TIME_REPRESENTATION_CONTENT">
99 <p>Libev represents time as a single floating point number, representing the
100 (fractional) number of seconds since the (POSIX) epoch (somewhere near
101 the beginning of 1970, details are complicated, don't ask). This type is
102 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
103 to the <code>double</code> type in C, and when you need to do any calculations on
104 it, you should treat it as such.</p>
111 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
112 <div id="GLOBAL_FUNCTIONS_CONTENT">
113 <p>These functions can be called anytime, even before initialising the
114 library in any way.</p>
116 <dt>ev_tstamp ev_time ()</dt>
118 <p>Returns the current time as libev would use it. Please note that the
119 <code>ev_now</code> function is usually faster and also often returns the timestamp
120 you actually want to know.</p>
122 <dt>int ev_version_major ()</dt>
123 <dt>int ev_version_minor ()</dt>
125 <p>You can find out the major and minor version numbers of the library
126 you linked against by calling the functions <code>ev_version_major</code> and
127 <code>ev_version_minor</code>. If you want, you can compare against the global
128 symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
129 version of the library your program was compiled against.</p>
130 <p>Usually, it's a good idea to terminate if the major versions mismatch,
131 as this indicates an incompatible change. Minor versions are usually
132 compatible to older versions, so a larger minor version alone is usually
134 <p>Example: make sure we haven't accidentally been linked against the wrong
136 <pre> assert (("libev version mismatch",
137 ev_version_major () == EV_VERSION_MAJOR
138 && ev_version_minor () >= EV_VERSION_MINOR));
142 <dt>unsigned int ev_supported_backends ()</dt>
144 <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
145 value) compiled into this binary of libev (independent of their
146 availability on the system you are running on). See <code>ev_default_loop</code> for
147 a description of the set values.</p>
148 <p>Example: make sure we have the epoll method, because yeah this is cool and
149 a must have and can we have a torrent of it please!!!11</p>
150 <pre> assert (("sorry, no epoll, no sex",
151 ev_supported_backends () & EVBACKEND_EPOLL));
155 <dt>unsigned int ev_recommended_backends ()</dt>
157 <p>Return the set of all backends compiled into this binary of libev and also
158 recommended for this platform. This set is often smaller than the one
159 returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
160 most BSDs and will not be autodetected unless you explicitly request it
161 (assuming you know what you are doing). This is the set of backends that
162 libev will probe for if you specify no backends explicitly.</p>
164 <dt>unsigned int ev_embeddable_backends ()</dt>
166 <p>Returns the set of backends that are embeddable in other event loops. This
167 is the theoretical, all-platform, value. To find which backends
168 might be supported on the current system, you would need to look at
169 <code>ev_embeddable_backends () & ev_supported_backends ()</code>, likewise for
170 recommended ones.</p>
171 <p>See the description of <code>ev_embed</code> watchers for more info.</p>
173 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
175 <p>Sets the allocation function to use (the prototype is similar to the
176 realloc C function, the semantics are identical). It is used to allocate
177 and free memory (no surprises here). If it returns zero when memory
178 needs to be allocated, the library might abort or take some potentially
179 destructive action. The default is your system realloc function.</p>
180 <p>You could override this function in high-availability programs to, say,
181 free some memory if it cannot allocate memory, to use a special allocator,
182 or even to sleep a while and retry until some memory is available.</p>
183 <p>Example: replace the libev allocator with one that waits a bit and then
184 retries: better than mine).</p>
186 persistent_realloc (void *ptr, long size)
190 void *newptr = realloc (ptr, size);
200 ev_set_allocator (persistent_realloc);
204 <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
206 <p>Set the callback function to call on a retryable syscall error (such
207 as failed select, poll, epoll_wait). The message is a printable string
208 indicating the system call or subsystem causing the problem. If this
209 callback is set, then libev will expect it to remedy the sitution, no
210 matter what, when it returns. That is, libev will generally retry the
211 requested operation, or, if the condition doesn't go away, do bad stuff
213 <p>Example: do the same thing as libev does internally:</p>
215 fatal_error (const char *msg)
222 ev_set_syserr_cb (fatal_error);
229 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p>
230 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
231 <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
232 types of such loops, the <i>default</i> loop, which supports signals and child
233 events, and dynamically created loops which do not.</p>
234 <p>If you use threads, a common model is to run the default event loop
235 in your main thread (or in a separate thread) and for each thread you
236 create, you also create another event loop. Libev itself does no locking
237 whatsoever, so if you mix calls to the same event loop in different
238 threads, make sure you lock (this is usually a bad idea, though, even if
239 done correctly, because it's hideous and inefficient).</p>
241 <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
243 <p>This will initialise the default event loop if it hasn't been initialised
244 yet and return it. If the default loop could not be initialised, returns
245 false. If it already was initialised it simply returns it (and ignores the
246 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
247 <p>If you don't know what event loop to use, use the one returned from this
249 <p>The flags argument can be used to specify special behaviour or specific
250 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
251 <p>The following flags are supported:</p>
254 <dt><code>EVFLAG_AUTO</code></dt>
256 <p>The default flags value. Use this if you have no clue (it's the right
257 thing, believe me).</p>
259 <dt><code>EVFLAG_NOENV</code></dt>
261 <p>If this flag bit is ored into the flag value (or the program runs setuid
262 or setgid) then libev will <i>not</i> look at the environment variable
263 <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
264 override the flags completely if it is found in the environment. This is
265 useful to try out specific backends to test their performance, or to work
268 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
270 <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
271 libev tries to roll its own fd_set with no limits on the number of fds,
272 but if that fails, expect a fairly low limit on the number of fds when
273 using this backend. It doesn't scale too well (O(highest_fd)), but its usually
274 the fastest backend for a low number of fds.</p>
276 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
278 <p>And this is your standard poll(2) backend. It's more complicated than
279 select, but handles sparse fds better and has no artificial limit on the
280 number of fds you can use (except it will slow down considerably with a
281 lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
283 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
285 <p>For few fds, this backend is a bit little slower than poll and select,
286 but it scales phenomenally better. While poll and select usually scale like
287 O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
288 either O(1) or O(active_fds).</p>
289 <p>While stopping and starting an I/O watcher in the same iteration will
290 result in some caching, there is still a syscall per such incident
291 (because the fd could point to a different file description now), so its
292 best to avoid that. Also, dup()ed file descriptors might not work very
293 well if you register events for both fds.</p>
294 <p>Please note that epoll sometimes generates spurious notifications, so you
295 need to use non-blocking I/O or other means to avoid blocking when no data
296 (or space) is available.</p>
298 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
300 <p>Kqueue deserves special mention, as at the time of this writing, it
301 was broken on all BSDs except NetBSD (usually it doesn't work with
302 anything but sockets and pipes, except on Darwin, where of course its
303 completely useless). For this reason its not being "autodetected"
304 unless you explicitly specify it explicitly in the flags (i.e. using
305 <code>EVBACKEND_KQUEUE</code>).</p>
306 <p>It scales in the same way as the epoll backend, but the interface to the
307 kernel is more efficient (which says nothing about its actual speed, of
308 course). While starting and stopping an I/O watcher does not cause an
309 extra syscall as with epoll, it still adds up to four event changes per
310 incident, so its best to avoid that.</p>
312 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
314 <p>This is not implemented yet (and might never be).</p>
316 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
318 <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
319 it's really slow, but it still scales very well (O(active_fds)).</p>
320 <p>Please note that solaris ports can result in a lot of spurious
321 notifications, so you need to use non-blocking I/O or other means to avoid
322 blocking when no data (or space) is available.</p>
324 <dt><code>EVBACKEND_ALL</code></dt>
326 <p>Try all backends (even potentially broken ones that wouldn't be tried
327 with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
328 <code>EVBACKEND_ALL & ~EVBACKEND_KQUEUE</code>.</p>
332 <p>If one or more of these are ored into the flags value, then only these
333 backends will be tried (in the reverse order as given here). If none are
334 specified, most compiled-in backend will be tried, usually in reverse
335 order of their flag values :)</p>
336 <p>The most typical usage is like this:</p>
337 <pre> if (!ev_default_loop (0))
338 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
341 <p>Restrict libev to the select and poll backends, and do not allow
342 environment settings to be taken into account:</p>
343 <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
346 <p>Use whatever libev has to offer, but make sure that kqueue is used if
347 available (warning, breaks stuff, best use only with your own private
348 event loop and only if you know the OS supports your types of fds):</p>
349 <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
353 <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
355 <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
356 always distinct from the default loop. Unlike the default loop, it cannot
357 handle signal and child watchers, and attempts to do so will be greeted by
358 undefined behaviour (or a failed assertion if assertions are enabled).</p>
359 <p>Example: try to create a event loop that uses epoll and nothing else.</p>
360 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
362 fatal ("no epoll found here, maybe it hides under your chair");
366 <dt>ev_default_destroy ()</dt>
368 <p>Destroys the default loop again (frees all memory and kernel state
369 etc.). This stops all registered event watchers (by not touching them in
370 any way whatsoever, although you cannot rely on this :).</p>
372 <dt>ev_loop_destroy (loop)</dt>
374 <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
375 earlier call to <code>ev_loop_new</code>.</p>
377 <dt>ev_default_fork ()</dt>
379 <p>This function reinitialises the kernel state for backends that have
380 one. Despite the name, you can call it anytime, but it makes most sense
381 after forking, in either the parent or child process (or both, but that
382 again makes little sense).</p>
383 <p>You <i>must</i> call this function in the child process after forking if and
384 only if you want to use the event library in both processes. If you just
385 fork+exec, you don't have to call it.</p>
386 <p>The function itself is quite fast and it's usually not a problem to call
387 it just in case after a fork. To make this easy, the function will fit in
388 quite nicely into a call to <code>pthread_atfork</code>:</p>
389 <pre> pthread_atfork (0, 0, ev_default_fork);
392 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
393 without calling this function, so if you force one of those backends you
394 do not need to care.</p>
396 <dt>ev_loop_fork (loop)</dt>
398 <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
399 <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
400 after fork, and how you do this is entirely your own problem.</p>
402 <dt>unsigned int ev_backend (loop)</dt>
404 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
407 <dt>ev_tstamp ev_now (loop)</dt>
409 <p>Returns the current "event loop time", which is the time the event loop
410 received events and started processing them. This timestamp does not
411 change as long as callbacks are being processed, and this is also the base
412 time used for relative timers. You can treat it as the timestamp of the
413 event occuring (or more correctly, libev finding out about it).</p>
415 <dt>ev_loop (loop, int flags)</dt>
417 <p>Finally, this is it, the event handler. This function usually is called
418 after you initialised all your watchers and you want to start handling
420 <p>If the flags argument is specified as <code>0</code>, it will not return until
421 either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
422 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
423 relying on all watchers to be stopped when deciding when a program has
424 finished (especially in interactive programs), but having a program that
425 automatically loops as long as it has to and no longer by virtue of
426 relying on its watchers stopping correctly is a thing of beauty.</p>
427 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
428 those events and any outstanding ones, but will not block your process in
429 case there are no events and will return after one iteration of the loop.</p>
430 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
431 neccessary) and will handle those and any outstanding ones. It will block
432 your process until at least one new event arrives, and will return after
433 one iteration of the loop. This is useful if you are waiting for some
434 external event in conjunction with something not expressible using other
435 libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
436 usually a better approach for this kind of thing.</p>
437 <p>Here are the gory details of what <code>ev_loop</code> does:</p>
438 <pre> * If there are no active watchers (reference count is zero), return.
439 - Queue prepare watchers and then call all outstanding watchers.
440 - If we have been forked, recreate the kernel state.
441 - Update the kernel state with all outstanding changes.
442 - Update the "event loop time".
443 - Calculate for how long to block.
444 - Block the process, waiting for any events.
445 - Queue all outstanding I/O (fd) events.
446 - Update the "event loop time" and do time jump handling.
447 - Queue all outstanding timers.
448 - Queue all outstanding periodics.
449 - If no events are pending now, queue all idle watchers.
450 - Queue all check watchers.
451 - Call all queued watchers in reverse order (i.e. check watchers first).
452 Signals and child watchers are implemented as I/O watchers, and will
453 be handled here by queueing them when their watcher gets executed.
454 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
455 were used, return, otherwise continue with step *.
458 <p>Example: queue some jobs and then loop until no events are outsanding
460 <pre> ... queue jobs here, make sure they register event watchers as long
461 ... as they still have work to do (even an idle watcher will do..)
462 ev_loop (my_loop, 0);
467 <dt>ev_unloop (loop, how)</dt>
469 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
470 has processed all outstanding events). The <code>how</code> argument must be either
471 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
472 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
474 <dt>ev_ref (loop)</dt>
475 <dt>ev_unref (loop)</dt>
477 <p>Ref/unref can be used to add or remove a reference count on the event
478 loop: Every watcher keeps one reference, and as long as the reference
479 count is nonzero, <code>ev_loop</code> will not return on its own. If you have
480 a watcher you never unregister that should not keep <code>ev_loop</code> from
481 returning, ev_unref() after starting, and ev_ref() before stopping it. For
482 example, libev itself uses this for its internal signal pipe: It is not
483 visible to the libev user and should not keep <code>ev_loop</code> from exiting if
484 no event watchers registered by it are active. It is also an excellent
485 way to do this for generic recurring timers or from within third-party
486 libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
487 <p>Example: create a signal watcher, but keep it from keeping <code>ev_loop</code>
488 running when nothing else is active.</p>
489 <pre> struct dv_signal exitsig;
490 ev_signal_init (&exitsig, sig_cb, SIGINT);
491 ev_signal_start (myloop, &exitsig);
495 <p>Example: for some weird reason, unregister the above signal handler again.</p>
496 <pre> ev_ref (myloop);
497 ev_signal_stop (myloop, &exitsig);
504 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p>
505 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
506 <p>A watcher is a structure that you create and register to record your
507 interest in some event. For instance, if you want to wait for STDIN to
508 become readable, you would create an <code>ev_io</code> watcher for that:</p>
509 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
512 ev_unloop (loop, EVUNLOOP_ALL);
515 struct ev_loop *loop = ev_default_loop (0);
516 struct ev_io stdin_watcher;
517 ev_init (&stdin_watcher, my_cb);
518 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
519 ev_io_start (loop, &stdin_watcher);
523 <p>As you can see, you are responsible for allocating the memory for your
524 watcher structures (and it is usually a bad idea to do this on the stack,
525 although this can sometimes be quite valid).</p>
526 <p>Each watcher structure must be initialised by a call to <code>ev_init
527 (watcher *, callback)</code>, which expects a callback to be provided. This
528 callback gets invoked each time the event occurs (or, in the case of io
529 watchers, each time the event loop detects that the file descriptor given
530 is readable and/or writable).</p>
531 <p>Each watcher type has its own <code>ev_<type>_set (watcher *, ...)</code> macro
532 with arguments specific to this watcher type. There is also a macro
533 to combine initialisation and setting in one call: <code>ev_<type>_init
534 (watcher *, callback, ...)</code>.</p>
535 <p>To make the watcher actually watch out for events, you have to start it
536 with a watcher-specific start function (<code>ev_<type>_start (loop, watcher
537 *)</code>), and you can stop watching for events at any time by calling the
538 corresponding stop function (<code>ev_<type>_stop (loop, watcher *)</code>.</p>
539 <p>As long as your watcher is active (has been started but not stopped) you
540 must not touch the values stored in it. Most specifically you must never
541 reinitialise it or call its <code>set</code> macro.</p>
542 <p>Each and every callback receives the event loop pointer as first, the
543 registered watcher structure as second, and a bitset of received events as
545 <p>The received events usually include a single bit per event type received
546 (you can receive multiple events at the same time). The possible bit masks
549 <dt><code>EV_READ</code></dt>
550 <dt><code>EV_WRITE</code></dt>
552 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
555 <dt><code>EV_TIMEOUT</code></dt>
557 <p>The <code>ev_timer</code> watcher has timed out.</p>
559 <dt><code>EV_PERIODIC</code></dt>
561 <p>The <code>ev_periodic</code> watcher has timed out.</p>
563 <dt><code>EV_SIGNAL</code></dt>
565 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
567 <dt><code>EV_CHILD</code></dt>
569 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
571 <dt><code>EV_IDLE</code></dt>
573 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
575 <dt><code>EV_PREPARE</code></dt>
576 <dt><code>EV_CHECK</code></dt>
578 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
579 to gather new events, and all <code>ev_check</code> watchers are invoked just after
580 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
581 received events. Callbacks of both watcher types can start and stop as
582 many watchers as they want, and all of them will be taken into account
583 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
584 <code>ev_loop</code> from blocking).</p>
586 <dt><code>EV_ERROR</code></dt>
588 <p>An unspecified error has occured, the watcher has been stopped. This might
589 happen because the watcher could not be properly started because libev
590 ran out of memory, a file descriptor was found to be closed or any other
591 problem. You best act on it by reporting the problem and somehow coping
592 with the watcher being stopped.</p>
593 <p>Libev will usually signal a few "dummy" events together with an error,
594 for example it might indicate that a fd is readable or writable, and if
595 your callbacks is well-written it can just attempt the operation and cope
596 with the error from read() or write(). This will not work in multithreaded
597 programs, though, so beware.</p>
602 <h2 id="SUMMARY_OF_GENERIC_WATCHER_FUNCTIONS">SUMMARY OF GENERIC WATCHER FUNCTIONS</h2>
603 <div id="SUMMARY_OF_GENERIC_WATCHER_FUNCTIONS-2">
604 <p>In the following description, <code>TYPE</code> stands for the watcher type,
605 e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
607 <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
609 <p>This macro initialises the generic portion of a watcher. The contents
610 of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
611 the generic parts of the watcher are initialised, you <i>need</i> to call
612 the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
613 type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
614 which rolls both calls into one.</p>
615 <p>You can reinitialise a watcher at any time as long as it has been stopped
616 (or never started) and there are no pending events outstanding.</p>
617 <p>The callbakc is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
618 int revents)</code>.</p>
620 <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
622 <p>This macro initialises the type-specific parts of a watcher. You need to
623 call <code>ev_init</code> at least once before you call this macro, but you can
624 call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
625 macro on a watcher that is active (it can be pending, however, which is a
626 difference to the <code>ev_init</code> macro).</p>
627 <p>Although some watcher types do not have type-specific arguments
628 (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
630 <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
632 <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
633 calls into a single call. This is the most convinient method to initialise
634 a watcher. The same limitations apply, of course.</p>
636 <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
638 <p>Starts (activates) the given watcher. Only active watchers will receive
639 events. If the watcher is already active nothing will happen.</p>
641 <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
643 <p>Stops the given watcher again (if active) and clears the pending
644 status. It is possible that stopped watchers are pending (for example,
645 non-repeating timers are being stopped when they become pending), but
646 <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
647 you want to free or reuse the memory used by the watcher it is therefore a
648 good idea to always call its <code>ev_TYPE_stop</code> function.</p>
650 <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
652 <p>Returns a true value iff the watcher is active (i.e. it has been started
653 and not yet been stopped). As long as a watcher is active you must not modify
656 <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
658 <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
659 events but its callback has not yet been invoked). As long as a watcher
660 is pending (but not active) you must not call an init function on it (but
661 <code>ev_TYPE_set</code> is safe) and you must make sure the watcher is available to
662 libev (e.g. you cnanot <code>free ()</code> it).</p>
664 <dt>callback = ev_cb (ev_TYPE *watcher)</dt>
666 <p>Returns the callback currently set on the watcher.</p>
668 <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
670 <p>Change the callback. You can change the callback at virtually any time
671 (modulo threads).</p>
680 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
681 <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
682 <p>Each watcher has, by default, a member <code>void *data</code> that you can change
683 and read at any time, libev will completely ignore it. This can be used
684 to associate arbitrary data with your watcher. If you need more data and
685 don't want to allocate memory and store a pointer to it in that data
686 member, you can also "subclass" the watcher type and provide your own
693 struct whatever *mostinteresting;
697 <p>And since your callback will be called with a pointer to the watcher, you
698 can cast it back to your own type:</p>
699 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
701 struct my_io *w = (struct my_io *)w_;
706 <p>More interesting and less C-conformant ways of catsing your callback type
707 have been omitted....</p>
714 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p>
715 <div id="WATCHER_TYPES_CONTENT">
716 <p>This section describes each watcher in detail, but will not repeat
717 information given in the last section.</p>
724 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</h2>
725 <div id="code_ev_io_code_is_this_file_descrip-2">
726 <p>I/O watchers check whether a file descriptor is readable or writable
727 in each iteration of the event loop (This behaviour is called
728 level-triggering because you keep receiving events as long as the
729 condition persists. Remember you can stop the watcher if you don't want to
730 act on the event and neither want to receive future events).</p>
731 <p>In general you can register as many read and/or write event watchers per
732 fd as you want (as long as you don't confuse yourself). Setting all file
733 descriptors to non-blocking mode is also usually a good idea (but not
734 required if you know what you are doing).</p>
735 <p>You have to be careful with dup'ed file descriptors, though. Some backends
736 (the linux epoll backend is a notable example) cannot handle dup'ed file
737 descriptors correctly if you register interest in two or more fds pointing
738 to the same underlying file/socket etc. description (that is, they share
739 the same underlying "file open").</p>
740 <p>If you must do this, then force the use of a known-to-be-good backend
741 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
742 <code>EVBACKEND_POLL</code>).</p>
744 <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
745 <dt>ev_io_set (ev_io *, int fd, int events)</dt>
747 <p>Configures an <code>ev_io</code> watcher. The fd is the file descriptor to rceeive
748 events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ |
749 EV_WRITE</code> to receive the given events.</p>
750 <p>Please note that most of the more scalable backend mechanisms (for example
751 epoll and solaris ports) can result in spurious readyness notifications
752 for file descriptors, so you practically need to use non-blocking I/O (and
753 treat callback invocation as hint only), or retest separately with a safe
754 interface before doing I/O (XLib can do this), or force the use of either
755 <code>EVBACKEND_SELECT</code> or <code>EVBACKEND_POLL</code>, which don't suffer from this
756 problem. Also note that it is quite easy to have your callback invoked
757 when the readyness condition is no longer valid even when employing
758 typical ways of handling events, so its a good idea to use non-blocking
759 I/O unconditionally.</p>
762 <p>Example: call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
763 readable, but only once. Since it is likely line-buffered, you could
764 attempt to read a whole line in the callback:</p>
766 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
768 ev_io_stop (loop, w);
769 .. read from stdin here (or from w->fd) and haqndle any I/O errors
773 struct ev_loop *loop = ev_default_init (0);
774 struct ev_io stdin_readable;
775 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
776 ev_io_start (loop, &stdin_readable);
785 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</h2>
786 <div id="code_ev_timer_code_relative_and_opti-2">
787 <p>Timer watchers are simple relative timers that generate an event after a
788 given time, and optionally repeating in regular intervals after that.</p>
789 <p>The timers are based on real time, that is, if you register an event that
790 times out after an hour and you reset your system clock to last years
791 time, it will still time out after (roughly) and hour. "Roughly" because
792 detecting time jumps is hard, and some inaccuracies are unavoidable (the
793 monotonic clock option helps a lot here).</p>
794 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
795 time. This is usually the right thing as this timestamp refers to the time
796 of the event triggering whatever timeout you are modifying/starting. If
797 you suspect event processing to be delayed and you <i>need</i> to base the timeout
798 on the current time, use something like this to adjust for this:</p>
799 <pre> ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
802 <p>The callback is guarenteed to be invoked only when its timeout has passed,
803 but if multiple timers become ready during the same loop iteration then
804 order of execution is undefined.</p>
806 <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
807 <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
809 <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
810 <code>0.</code>, then it will automatically be stopped. If it is positive, then the
811 timer will automatically be configured to trigger again <code>repeat</code> seconds
812 later, again, and again, until stopped manually.</p>
813 <p>The timer itself will do a best-effort at avoiding drift, that is, if you
814 configure a timer to trigger every 10 seconds, then it will trigger at
815 exactly 10 second intervals. If, however, your program cannot keep up with
816 the timer (because it takes longer than those 10 seconds to do stuff) the
817 timer will not fire more than once per event loop iteration.</p>
819 <dt>ev_timer_again (loop)</dt>
821 <p>This will act as if the timer timed out and restart it again if it is
822 repeating. The exact semantics are:</p>
823 <p>If the timer is started but nonrepeating, stop it.</p>
824 <p>If the timer is repeating, either start it if necessary (with the repeat
825 value), or reset the running timer to the repeat value.</p>
826 <p>This sounds a bit complicated, but here is a useful and typical
827 example: Imagine you have a tcp connection and you want a so-called idle
828 timeout, that is, you want to be called when there have been, say, 60
829 seconds of inactivity on the socket. The easiest way to do this is to
830 configure an <code>ev_timer</code> with after=repeat=60 and calling ev_timer_again each
831 time you successfully read or write some data. If you go into an idle
832 state where you do not expect data to travel on the socket, you can stop
833 the timer, and again will automatically restart it if need be.</p>
836 <p>Example: create a timer that fires after 60 seconds.</p>
838 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
840 .. one minute over, w is actually stopped right here
843 struct ev_timer mytimer;
844 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
845 ev_timer_start (loop, &mytimer);
848 <p>Example: create a timeout timer that times out after 10 seconds of
851 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
853 .. ten seconds without any activity
856 struct ev_timer mytimer;
857 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
858 ev_timer_again (&mytimer); /* start timer */
861 // and in some piece of code that gets executed on any "activity":
862 // reset the timeout to start ticking again at 10 seconds
863 ev_timer_again (&mytimer);
871 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</h2>
872 <div id="code_ev_periodic_code_to_cron_or_not-2">
873 <p>Periodic watchers are also timers of a kind, but they are very versatile
874 (and unfortunately a bit complex).</p>
875 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
876 but on wallclock time (absolute time). You can tell a periodic watcher
877 to trigger "at" some specific point in time. For example, if you tell a
878 periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()
879 + 10.>) and then reset your system clock to the last year, then it will
880 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
881 roughly 10 seconds later and of course not if you reset your system time
883 <p>They can also be used to implement vastly more complex timers, such as
884 triggering an event on eahc midnight, local time.</p>
885 <p>As with timers, the callback is guarenteed to be invoked only when the
886 time (<code>at</code>) has been passed, but if multiple periodic timers become ready
887 during the same loop iteration then order of execution is undefined.</p>
889 <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
890 <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
892 <p>Lots of arguments, lets sort it out... There are basically three modes of
893 operation, and we will explain them from simplest to complex:</p>
896 <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
898 <p>In this configuration the watcher triggers an event at the wallclock time
899 <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
900 that is, if it is to be run at January 1st 2011 then it will run when the
901 system time reaches or surpasses this time.</p>
903 <dt>* non-repeating interval timer (interval > 0, reschedule_cb = 0)</dt>
905 <p>In this mode the watcher will always be scheduled to time out at the next
906 <code>at + N * interval</code> time (for some integer N) and then repeat, regardless
907 of any time jumps.</p>
908 <p>This can be used to create timers that do not drift with respect to system
910 <pre> ev_periodic_set (&periodic, 0., 3600., 0);
913 <p>This doesn't mean there will always be 3600 seconds in between triggers,
914 but only that the the callback will be called when the system time shows a
915 full hour (UTC), or more correctly, when the system time is evenly divisible
917 <p>Another way to think about it (for the mathematically inclined) is that
918 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
919 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
921 <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
923 <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
924 ignored. Instead, each time the periodic watcher gets scheduled, the
925 reschedule callback will be called with the watcher as first, and the
926 current time as second argument.</p>
927 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
928 ever, or make any event loop modifications</i>. If you need to stop it,
929 return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
930 starting a prepare watcher).</p>
931 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
932 ev_tstamp now)</code>, e.g.:</p>
933 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
939 <p>It must return the next time to trigger, based on the passed time value
940 (that is, the lowest time value larger than to the second argument). It
941 will usually be called just before the callback will be triggered, but
942 might be called at other times, too.</p>
943 <p>NOTE: <i>This callback must always return a time that is later than the
944 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
945 <p>This can be used to create very complex timers, such as a timer that
946 triggers on each midnight, local time. To do this, you would calculate the
947 next midnight after <code>now</code> and return the timestamp value for this. How
948 you do this is, again, up to you (but it is not trivial, which is the main
949 reason I omitted it as an example).</p>
954 <dt>ev_periodic_again (loop, ev_periodic *)</dt>
956 <p>Simply stops and restarts the periodic watcher again. This is only useful
957 when you changed some parameters or the reschedule callback would return
958 a different time than the last time it was called (e.g. in a crond like
959 program when the crontabs have changed).</p>
962 <p>Example: call a callback every hour, or, more precisely, whenever the
963 system clock is divisible by 3600. The callback invocation times have
964 potentially a lot of jittering, but good long-term stability.</p>
966 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
968 ... its now a full hour (UTC, or TAI or whatever your clock follows)
971 struct ev_periodic hourly_tick;
972 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
973 ev_periodic_start (loop, &hourly_tick);
976 <p>Example: the same as above, but use a reschedule callback to do it:</p>
977 <pre> #include <math.h>
980 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
982 return fmod (now, 3600.) + 3600.;
985 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
988 <p>Example: call a callback every hour, starting now:</p>
989 <pre> struct ev_periodic hourly_tick;
990 ev_periodic_init (&hourly_tick, clock_cb,
991 fmod (ev_now (loop), 3600.), 3600., 0);
992 ev_periodic_start (loop, &hourly_tick);
1000 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</h2>
1001 <div id="code_ev_signal_code_signal_me_when_a-2">
1002 <p>Signal watchers will trigger an event when the process receives a specific
1003 signal one or more times. Even though signals are very asynchronous, libev
1004 will try it's best to deliver signals synchronously, i.e. as part of the
1005 normal event processing, like any other event.</p>
1006 <p>You can configure as many watchers as you like per signal. Only when the
1007 first watcher gets started will libev actually register a signal watcher
1008 with the kernel (thus it coexists with your own signal handlers as long
1009 as you don't register any with libev). Similarly, when the last signal
1010 watcher for a signal is stopped libev will reset the signal handler to
1011 SIG_DFL (regardless of what it was set to before).</p>
1013 <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1014 <dt>ev_signal_set (ev_signal *, int signum)</dt>
1016 <p>Configures the watcher to trigger on the given signal number (usually one
1017 of the <code>SIGxxx</code> constants).</p>
1026 <h2 id="code_ev_child_code_wait_for_pid_stat"><code>ev_child</code> - wait for pid status changes</h2>
1027 <div id="code_ev_child_code_wait_for_pid_stat-2">
1028 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1029 some child status changes (most typically when a child of yours dies).</p>
1031 <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1032 <dt>ev_child_set (ev_child *, int pid)</dt>
1034 <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1035 <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1036 at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1037 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1038 <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1039 process causing the status change.</p>
1042 <p>Example: try to exit cleanly on SIGINT and SIGTERM.</p>
1044 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1046 ev_unloop (loop, EVUNLOOP_ALL);
1049 struct ev_signal signal_watcher;
1050 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1051 ev_signal_start (loop, &sigint_cb);
1059 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</h2>
1060 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1061 <p>Idle watchers trigger events when there are no other events are pending
1062 (prepare, check and other idle watchers do not count). That is, as long
1063 as your process is busy handling sockets or timeouts (or even signals,
1064 imagine) it will not be triggered. But when your process is idle all idle
1065 watchers are being called again and again, once per event loop iteration -
1066 until stopped, that is, or your process receives more events and becomes
1068 <p>The most noteworthy effect is that as long as any idle watchers are
1069 active, the process will not block when waiting for new events.</p>
1070 <p>Apart from keeping your process non-blocking (which is a useful
1071 effect on its own sometimes), idle watchers are a good place to do
1072 "pseudo-background processing", or delay processing stuff to after the
1073 event loop has handled all outstanding events.</p>
1075 <dt>ev_idle_init (ev_signal *, callback)</dt>
1077 <p>Initialises and configures the idle watcher - it has no parameters of any
1078 kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1082 <p>Example: dynamically allocate an <code>ev_idle</code>, start it, and in the
1083 callback, free it. Alos, use no error checking, as usual.</p>
1085 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1088 // now do something you wanted to do when the program has
1089 // no longer asnything immediate to do.
1092 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1093 ev_idle_init (idle_watcher, idle_cb);
1094 ev_idle_start (loop, idle_cb);
1102 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</h2>
1103 <div id="code_ev_prepare_code_and_code_ev_che-2">
1104 <p>Prepare and check watchers are usually (but not always) used in tandem:
1105 prepare watchers get invoked before the process blocks and check watchers
1107 <p>Their main purpose is to integrate other event mechanisms into libev and
1108 their use is somewhat advanced. This could be used, for example, to track
1109 variable changes, implement your own watchers, integrate net-snmp or a
1110 coroutine library and lots more.</p>
1111 <p>This is done by examining in each prepare call which file descriptors need
1112 to be watched by the other library, registering <code>ev_io</code> watchers for
1113 them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1114 provide just this functionality). Then, in the check watcher you check for
1115 any events that occured (by checking the pending status of all watchers
1116 and stopping them) and call back into the library. The I/O and timer
1117 callbacks will never actually be called (but must be valid nevertheless,
1118 because you never know, you know?).</p>
1119 <p>As another example, the Perl Coro module uses these hooks to integrate
1120 coroutines into libev programs, by yielding to other active coroutines
1121 during each prepare and only letting the process block if no coroutines
1122 are ready to run (it's actually more complicated: it only runs coroutines
1123 with priority higher than or equal to the event loop and one coroutine
1124 of lower priority, but only once, using idle watchers to keep the event
1125 loop from blocking if lower-priority coroutines are active, thus mapping
1126 low-priority coroutines to idle/background tasks).</p>
1128 <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1129 <dt>ev_check_init (ev_check *, callback)</dt>
1131 <p>Initialises and configures the prepare or check watcher - they have no
1132 parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1133 macros, but using them is utterly, utterly and completely pointless.</p>
1136 <p>Example: *TODO*.</p>
1143 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough</h2>
1144 <div id="code_ev_embed_code_when_one_backend_-2">
1145 <p>This is a rather advanced watcher type that lets you embed one event loop
1146 into another (currently only <code>ev_io</code> events are supported in the embedded
1147 loop, other types of watchers might be handled in a delayed or incorrect
1148 fashion and must not be used).</p>
1149 <p>There are primarily two reasons you would want that: work around bugs and
1151 <p>As an example for a bug workaround, the kqueue backend might only support
1152 sockets on some platform, so it is unusable as generic backend, but you
1153 still want to make use of it because you have many sockets and it scales
1154 so nicely. In this case, you would create a kqueue-based loop and embed it
1155 into your default loop (which might use e.g. poll). Overall operation will
1156 be a bit slower because first libev has to poll and then call kevent, but
1157 at least you can use both at what they are best.</p>
1158 <p>As for prioritising I/O: rarely you have the case where some fds have
1159 to be watched and handled very quickly (with low latency), and even
1160 priorities and idle watchers might have too much overhead. In this case
1161 you would put all the high priority stuff in one loop and all the rest in
1162 a second one, and embed the second one in the first.</p>
1163 <p>As long as the watcher is active, the callback will be invoked every time
1164 there might be events pending in the embedded loop. The callback must then
1165 call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1166 their callbacks (you could also start an idle watcher to give the embedded
1167 loop strictly lower priority for example). You can also set the callback
1168 to <code>0</code>, in which case the embed watcher will automatically execute the
1169 embedded loop sweep.</p>
1170 <p>As long as the watcher is started it will automatically handle events. The
1171 callback will be invoked whenever some events have been handled. You can
1172 set the callback to <code>0</code> to avoid having to specify one if you are not
1173 interested in that.</p>
1174 <p>Also, there have not currently been made special provisions for forking:
1175 when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1176 but you will also have to stop and restart any <code>ev_embed</code> watchers
1178 <p>Unfortunately, not all backends are embeddable, only the ones returned by
1179 <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1181 <p>So when you want to use this feature you will always have to be prepared
1182 that you cannot get an embeddable loop. The recommended way to get around
1183 this is to have a separate variables for your embeddable loop, try to
1184 create it, and if that fails, use the normal loop for everything:</p>
1185 <pre> struct ev_loop *loop_hi = ev_default_init (0);
1186 struct ev_loop *loop_lo = 0;
1187 struct ev_embed embed;
1189 // see if there is a chance of getting one that works
1190 // (remember that a flags value of 0 means autodetection)
1191 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1192 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1195 // if we got one, then embed it, otherwise default to loop_hi
1198 ev_embed_init (&embed, 0, loop_lo);
1199 ev_embed_start (loop_hi, &embed);
1206 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1207 <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1209 <p>Configures the watcher to embed the given loop, which must be
1210 embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1211 invoked automatically, otherwise it is the responsibility of the callback
1212 to invoke it (it will continue to be called until the sweep has been done,
1213 if you do not want thta, you need to temporarily stop the embed watcher).</p>
1215 <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1217 <p>Make a single, non-blocking sweep over the embedded loop. This works
1218 similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1219 apropriate way for embedded loops.</p>
1228 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
1229 <div id="OTHER_FUNCTIONS_CONTENT">
1230 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1232 <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1234 <p>This function combines a simple timer and an I/O watcher, calls your
1235 callback on whichever event happens first and automatically stop both
1236 watchers. This is useful if you want to wait for a single event on an fd
1237 or timeout without having to allocate/configure/start/stop/free one or
1238 more watchers yourself.</p>
1239 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1240 is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1241 <code>events</code> set will be craeted and started.</p>
1242 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1243 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1244 repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1246 <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1247 passed an <code>revents</code> set like normal event callbacks (a combination of
1248 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1249 value passed to <code>ev_once</code>:</p>
1250 <pre> static void stdin_ready (int revents, void *arg)
1252 if (revents & EV_TIMEOUT)
1253 /* doh, nothing entered */;
1254 else if (revents & EV_READ)
1255 /* stdin might have data for us, joy! */;
1258 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1262 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1264 <p>Feeds the given event set into the event loop, as if the specified event
1265 had happened for the specified watcher (which must be a pointer to an
1266 initialised but not necessarily started event watcher).</p>
1268 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1270 <p>Feed an event on the given fd, as if a file descriptor backend detected
1271 the given events it.</p>
1273 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1275 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1285 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
1286 <div id="LIBEVENT_EMULATION_CONTENT">
1287 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1288 emulate the internals of libevent, so here are some usage hints:</p>
1290 <dt>* Use it by including <event.h>, as usual.</dt>
1291 <dt>* The following members are fully supported: ev_base, ev_callback,
1292 ev_arg, ev_fd, ev_res, ev_events.</dt>
1293 <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1294 maintained by libev, it does not work exactly the same way as in libevent (consider
1295 it a private API).</dt>
1296 <dt>* Priorities are not currently supported. Initialising priorities
1297 will fail and all watchers will have the same priority, even though there
1298 is an ev_pri field.</dt>
1299 <dt>* Other members are not supported.</dt>
1300 <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1301 to use the libev header file and library.</dt>
1305 <h1 id="C_SUPPORT">C++ SUPPORT</h1><p><a href="#TOP" class="toplink">Top</a></p>
1306 <div id="C_SUPPORT_CONTENT">
1310 <h1 id="AUTHOR">AUTHOR</h1><p><a href="#TOP" class="toplink">Top</a></p>
1311 <div id="AUTHOR_CONTENT">
1312 <p>Marc Lehmann <libev@schmorp.de>.</p>
projects / software / libev.git / blob