ViewVC Help
View File | Revision Log | Show Annotations | Download File
Revision: 1.278
Committed: Thu Dec 31 06:59:47 2009 UTC (14 years, 6 months ago) by root
Branch: MAIN
CVS Tags: rel-3_9
Changes since 1.277: +7 -6 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 =head1 NAME
3 libev - a high performance full-featured event loop written in C
5 =head1 SYNOPSIS
7 #include <ev.h>
11 // a single header file is required
12 #include <ev.h>
14 #include <stdio.h> // for puts
16 // every watcher type has its own typedef'd struct
17 // with the name ev_TYPE
18 ev_io stdin_watcher;
19 ev_timer timeout_watcher;
21 // all watcher callbacks have a similar signature
22 // this callback is called when data is readable on stdin
23 static void
24 stdin_cb (EV_P_ ev_io *w, int revents)
25 {
26 puts ("stdin ready");
27 // for one-shot events, one must manually stop the watcher
28 // with its corresponding stop function.
29 ev_io_stop (EV_A_ w);
31 // this causes all nested ev_loop's to stop iterating
32 ev_unloop (EV_A_ EVUNLOOP_ALL);
33 }
35 // another callback, this time for a time-out
36 static void
37 timeout_cb (EV_P_ ev_timer *w, int revents)
38 {
39 puts ("timeout");
40 // this causes the innermost ev_loop to stop iterating
41 ev_unloop (EV_A_ EVUNLOOP_ONE);
42 }
44 int
45 main (void)
46 {
47 // use the default event loop unless you have special needs
48 struct ev_loop *loop = ev_default_loop (0);
50 // initialise an io watcher, then start it
51 // this one will watch for stdin to become readable
52 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
53 ev_io_start (loop, &stdin_watcher);
55 // initialise a timer watcher, then start it
56 // simple non-repeating 5.5 second timeout
57 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
58 ev_timer_start (loop, &timeout_watcher);
60 // now wait for events to arrive
61 ev_loop (loop, 0);
63 // unloop was called, so exit
64 return 0;
65 }
69 This document documents the libev software package.
71 The newest version of this document is also available as an html-formatted
72 web page you might find easier to navigate when reading it for the first
73 time: L<>.
75 While this document tries to be as complete as possible in documenting
76 libev, its usage and the rationale behind its design, it is not a tutorial
77 on event-based programming, nor will it introduce event-based programming
78 with libev.
80 Familarity with event based programming techniques in general is assumed
81 throughout this document.
83 =head1 ABOUT LIBEV
85 Libev is an event loop: you register interest in certain events (such as a
86 file descriptor being readable or a timeout occurring), and it will manage
87 these event sources and provide your program with events.
89 To do this, it must take more or less complete control over your process
90 (or thread) by executing the I<event loop> handler, and will then
91 communicate events via a callback mechanism.
93 You register interest in certain events by registering so-called I<event
94 watchers>, which are relatively small C structures you initialise with the
95 details of the event, and then hand it over to libev by I<starting> the
96 watcher.
98 =head2 FEATURES
100 Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
101 BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
102 for file descriptor events (C<ev_io>), the Linux C<inotify> interface
103 (for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
104 inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
105 timers (C<ev_timer>), absolute timers with customised rescheduling
106 (C<ev_periodic>), synchronous signals (C<ev_signal>), process status
107 change events (C<ev_child>), and event watchers dealing with the event
108 loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
109 C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
110 limited support for fork events (C<ev_fork>).
112 It also is quite fast (see this
113 L<benchmark|> comparing it to libevent
114 for example).
116 =head2 CONVENTIONS
118 Libev is very configurable. In this manual the default (and most common)
119 configuration will be described, which supports multiple event loops. For
120 more info about various configuration options please have a look at
121 B<EMBED> section in this manual. If libev was configured without support
122 for multiple event loops, then all functions taking an initial argument of
123 name C<loop> (which is always of type C<struct ev_loop *>) will not have
124 this argument.
128 Libev represents time as a single floating point number, representing
129 the (fractional) number of seconds since the (POSIX) epoch (somewhere
130 near the beginning of 1970, details are complicated, don't ask). This
131 type is called C<ev_tstamp>, which is what you should use too. It usually
132 aliases to the C<double> type in C. When you need to do any calculations
133 on it, you should treat it as some floating point value. Unlike the name
134 component C<stamp> might indicate, it is also used for time differences
135 throughout libev.
139 Libev knows three classes of errors: operating system errors, usage errors
140 and internal errors (bugs).
142 When libev catches an operating system error it cannot handle (for example
143 a system call indicating a condition libev cannot fix), it calls the callback
144 set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
145 abort. The default is to print a diagnostic message and to call C<abort
146 ()>.
148 When libev detects a usage error such as a negative timer interval, then
149 it will print a diagnostic message and abort (via the C<assert> mechanism,
150 so C<NDEBUG> will disable this checking): these are programming errors in
151 the libev caller and need to be fixed there.
153 Libev also has a few internal error-checking C<assert>ions, and also has
154 extensive consistency checking code. These do not trigger under normal
155 circumstances, as they indicate either a bug in libev or worse.
160 These functions can be called anytime, even before initialising the
161 library in any way.
163 =over 4
165 =item ev_tstamp ev_time ()
167 Returns the current time as libev would use it. Please note that the
168 C<ev_now> function is usually faster and also often returns the timestamp
169 you actually want to know.
171 =item ev_sleep (ev_tstamp interval)
173 Sleep for the given interval: The current thread will be blocked until
174 either it is interrupted or the given time interval has passed. Basically
175 this is a sub-second-resolution C<sleep ()>.
177 =item int ev_version_major ()
179 =item int ev_version_minor ()
181 You can find out the major and minor ABI version numbers of the library
182 you linked against by calling the functions C<ev_version_major> and
183 C<ev_version_minor>. If you want, you can compare against the global
184 symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
185 version of the library your program was compiled against.
187 These version numbers refer to the ABI version of the library, not the
188 release version.
190 Usually, it's a good idea to terminate if the major versions mismatch,
191 as this indicates an incompatible change. Minor versions are usually
192 compatible to older versions, so a larger minor version alone is usually
193 not a problem.
195 Example: Make sure we haven't accidentally been linked against the wrong
196 version.
198 assert (("libev version mismatch",
199 ev_version_major () == EV_VERSION_MAJOR
200 && ev_version_minor () >= EV_VERSION_MINOR));
202 =item unsigned int ev_supported_backends ()
204 Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
205 value) compiled into this binary of libev (independent of their
206 availability on the system you are running on). See C<ev_default_loop> for
207 a description of the set values.
209 Example: make sure we have the epoll method, because yeah this is cool and
210 a must have and can we have a torrent of it please!!!11
212 assert (("sorry, no epoll, no sex",
213 ev_supported_backends () & EVBACKEND_EPOLL));
215 =item unsigned int ev_recommended_backends ()
217 Return the set of all backends compiled into this binary of libev and also
218 recommended for this platform. This set is often smaller than the one
219 returned by C<ev_supported_backends>, as for example kqueue is broken on
220 most BSDs and will not be auto-detected unless you explicitly request it
221 (assuming you know what you are doing). This is the set of backends that
222 libev will probe for if you specify no backends explicitly.
224 =item unsigned int ev_embeddable_backends ()
226 Returns the set of backends that are embeddable in other event loops. This
227 is the theoretical, all-platform, value. To find which backends
228 might be supported on the current system, you would need to look at
229 C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
230 recommended ones.
232 See the description of C<ev_embed> watchers for more info.
234 =item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
236 Sets the allocation function to use (the prototype is similar - the
237 semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
238 used to allocate and free memory (no surprises here). If it returns zero
239 when memory needs to be allocated (C<size != 0>), the library might abort
240 or take some potentially destructive action.
242 Since some systems (at least OpenBSD and Darwin) fail to implement
243 correct C<realloc> semantics, libev will use a wrapper around the system
244 C<realloc> and C<free> functions by default.
246 You could override this function in high-availability programs to, say,
247 free some memory if it cannot allocate memory, to use a special allocator,
248 or even to sleep a while and retry until some memory is available.
250 Example: Replace the libev allocator with one that waits a bit and then
251 retries (example requires a standards-compliant C<realloc>).
253 static void *
254 persistent_realloc (void *ptr, size_t size)
255 {
256 for (;;)
257 {
258 void *newptr = realloc (ptr, size);
260 if (newptr)
261 return newptr;
263 sleep (60);
264 }
265 }
267 ...
268 ev_set_allocator (persistent_realloc);
270 =item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
272 Set the callback function to call on a retryable system call error (such
273 as failed select, poll, epoll_wait). The message is a printable string
274 indicating the system call or subsystem causing the problem. If this
275 callback is set, then libev will expect it to remedy the situation, no
276 matter what, when it returns. That is, libev will generally retry the
277 requested operation, or, if the condition doesn't go away, do bad stuff
278 (such as abort).
280 Example: This is basically the same thing that libev does internally, too.
282 static void
283 fatal_error (const char *msg)
284 {
285 perror (msg);
286 abort ();
287 }
289 ...
290 ev_set_syserr_cb (fatal_error);
292 =back
296 An event loop is described by a C<struct ev_loop *> (the C<struct>
297 is I<not> optional in this case, as there is also an C<ev_loop>
298 I<function>).
300 The library knows two types of such loops, the I<default> loop, which
301 supports signals and child events, and dynamically created loops which do
302 not.
304 =over 4
306 =item struct ev_loop *ev_default_loop (unsigned int flags)
308 This will initialise the default event loop if it hasn't been initialised
309 yet and return it. If the default loop could not be initialised, returns
310 false. If it already was initialised it simply returns it (and ignores the
311 flags. If that is troubling you, check C<ev_backend ()> afterwards).
313 If you don't know what event loop to use, use the one returned from this
314 function.
316 Note that this function is I<not> thread-safe, so if you want to use it
317 from multiple threads, you have to lock (note also that this is unlikely,
318 as loops cannot be shared easily between threads anyway).
320 The default loop is the only loop that can handle C<ev_signal> and
321 C<ev_child> watchers, and to do this, it always registers a handler
322 for C<SIGCHLD>. If this is a problem for your application you can either
323 create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
324 can simply overwrite the C<SIGCHLD> signal handler I<after> calling
325 C<ev_default_init>.
327 The flags argument can be used to specify special behaviour or specific
328 backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
330 The following flags are supported:
332 =over 4
334 =item C<EVFLAG_AUTO>
336 The default flags value. Use this if you have no clue (it's the right
337 thing, believe me).
339 =item C<EVFLAG_NOENV>
341 If this flag bit is or'ed into the flag value (or the program runs setuid
342 or setgid) then libev will I<not> look at the environment variable
343 C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
344 override the flags completely if it is found in the environment. This is
345 useful to try out specific backends to test their performance, or to work
346 around bugs.
350 Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after
351 a fork, you can also make libev check for a fork in each iteration by
352 enabling this flag.
354 This works by calling C<getpid ()> on every iteration of the loop,
355 and thus this might slow down your event loop if you do a lot of loop
356 iterations and little real work, but is usually not noticeable (on my
357 GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
358 without a system call and thus I<very> fast, but my GNU/Linux system also has
359 C<pthread_atfork> which is even faster).
361 The big advantage of this flag is that you can forget about fork (and
362 forget about forgetting to tell libev about forking) when you use this
363 flag.
365 This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
366 environment variable.
370 When this flag is specified, then libev will not attempt to use the
371 I<inotify> API for it's C<ev_stat> watchers. Apart from debugging and
372 testing, this flag can be useful to conserve inotify file descriptors, as
373 otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
377 When this flag is specified, then libev will attempt to use the
378 I<signalfd> API for it's C<ev_signal> (and C<ev_child>) watchers. This API
379 delivers signals synchronously, which makes it both faster and might make
380 it possible to get the queued signal data. It can also simplify signal
381 handling with threads, as long as you properly block signals in your
382 threads that are not interested in handling them.
384 Signalfd will not be used by default as this changes your signal mask, and
385 there are a lot of shoddy libraries and programs (glib's threadpool for
386 example) that can't properly initialise their signal masks.
388 =item C<EVBACKEND_SELECT> (value 1, portable select backend)
390 This is your standard select(2) backend. Not I<completely> standard, as
391 libev tries to roll its own fd_set with no limits on the number of fds,
392 but if that fails, expect a fairly low limit on the number of fds when
393 using this backend. It doesn't scale too well (O(highest_fd)), but its
394 usually the fastest backend for a low number of (low-numbered :) fds.
396 To get good performance out of this backend you need a high amount of
397 parallelism (most of the file descriptors should be busy). If you are
398 writing a server, you should C<accept ()> in a loop to accept as many
399 connections as possible during one iteration. You might also want to have
400 a look at C<ev_set_io_collect_interval ()> to increase the amount of
401 readiness notifications you get per iteration.
403 This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
404 C<writefds> set (and to work around Microsoft Windows bugs, also onto the
405 C<exceptfds> set on that platform).
407 =item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
409 And this is your standard poll(2) backend. It's more complicated
410 than select, but handles sparse fds better and has no artificial
411 limit on the number of fds you can use (except it will slow down
412 considerably with a lot of inactive fds). It scales similarly to select,
413 i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
414 performance tips.
416 This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
419 =item C<EVBACKEND_EPOLL> (value 4, Linux)
421 Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
422 kernels).
424 For few fds, this backend is a bit little slower than poll and select,
425 but it scales phenomenally better. While poll and select usually scale
426 like O(total_fds) where n is the total number of fds (or the highest fd),
427 epoll scales either O(1) or O(active_fds).
429 The epoll mechanism deserves honorable mention as the most misdesigned
430 of the more advanced event mechanisms: mere annoyances include silently
431 dropping file descriptors, requiring a system call per change per file
432 descriptor (and unnecessary guessing of parameters), problems with dup and
433 so on. The biggest issue is fork races, however - if a program forks then
434 I<both> parent and child process have to recreate the epoll set, which can
435 take considerable time (one syscall per file descriptor) and is of course
436 hard to detect.
438 Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
439 of course I<doesn't>, and epoll just loves to report events for totally
440 I<different> file descriptors (even already closed ones, so one cannot
441 even remove them from the set) than registered in the set (especially
442 on SMP systems). Libev tries to counter these spurious notifications by
443 employing an additional generation counter and comparing that against the
444 events to filter out spurious ones, recreating the set when required.
446 While stopping, setting and starting an I/O watcher in the same iteration
447 will result in some caching, there is still a system call per such
448 incident (because the same I<file descriptor> could point to a different
449 I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
450 file descriptors might not work very well if you register events for both
451 file descriptors.
453 Best performance from this backend is achieved by not unregistering all
454 watchers for a file descriptor until it has been closed, if possible,
455 i.e. keep at least one watcher active per fd at all times. Stopping and
456 starting a watcher (without re-setting it) also usually doesn't cause
457 extra overhead. A fork can both result in spurious notifications as well
458 as in libev having to destroy and recreate the epoll object, which can
459 take considerable time and thus should be avoided.
461 All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
462 faster than epoll for maybe up to a hundred file descriptors, depending on
463 the usage. So sad.
465 While nominally embeddable in other event loops, this feature is broken in
466 all kernel versions tested so far.
468 This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
471 =item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
473 Kqueue deserves special mention, as at the time of this writing, it
474 was broken on all BSDs except NetBSD (usually it doesn't work reliably
475 with anything but sockets and pipes, except on Darwin, where of course
476 it's completely useless). Unlike epoll, however, whose brokenness
477 is by design, these kqueue bugs can (and eventually will) be fixed
478 without API changes to existing programs. For this reason it's not being
479 "auto-detected" unless you explicitly specify it in the flags (i.e. using
480 C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
481 system like NetBSD.
483 You still can embed kqueue into a normal poll or select backend and use it
484 only for sockets (after having made sure that sockets work with kqueue on
485 the target platform). See C<ev_embed> watchers for more info.
487 It scales in the same way as the epoll backend, but the interface to the
488 kernel is more efficient (which says nothing about its actual speed, of
489 course). While stopping, setting and starting an I/O watcher does never
490 cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
491 two event changes per incident. Support for C<fork ()> is very bad (but
492 sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
493 cases
495 This backend usually performs well under most conditions.
497 While nominally embeddable in other event loops, this doesn't work
498 everywhere, so you might need to test for this. And since it is broken
499 almost everywhere, you should only use it when you have a lot of sockets
500 (for which it usually works), by embedding it into another event loop
501 (e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
502 also broken on OS X)) and, did I mention it, using it only for sockets.
504 This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
505 C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
506 C<NOTE_EOF>.
508 =item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
510 This is not implemented yet (and might never be, unless you send me an
511 implementation). According to reports, C</dev/poll> only supports sockets
512 and is not embeddable, which would limit the usefulness of this backend
513 immensely.
515 =item C<EVBACKEND_PORT> (value 32, Solaris 10)
517 This uses the Solaris 10 event port mechanism. As with everything on Solaris,
518 it's really slow, but it still scales very well (O(active_fds)).
520 Please note that Solaris event ports can deliver a lot of spurious
521 notifications, so you need to use non-blocking I/O or other means to avoid
522 blocking when no data (or space) is available.
524 While this backend scales well, it requires one system call per active
525 file descriptor per loop iteration. For small and medium numbers of file
526 descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
527 might perform better.
529 On the positive side, with the exception of the spurious readiness
530 notifications, this backend actually performed fully to specification
531 in all tests and is fully embeddable, which is a rare feat among the
532 OS-specific backends (I vastly prefer correctness over speed hacks).
534 This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
537 =item C<EVBACKEND_ALL>
539 Try all backends (even potentially broken ones that wouldn't be tried
540 with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
543 It is definitely not recommended to use this flag.
545 =back
547 If one or more of the backend flags are or'ed into the flags value,
548 then only these backends will be tried (in the reverse order as listed
549 here). If none are specified, all backends in C<ev_recommended_backends
550 ()> will be tried.
552 Example: This is the most typical usage.
554 if (!ev_default_loop (0))
555 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
557 Example: Restrict libev to the select and poll backends, and do not allow
558 environment settings to be taken into account:
562 Example: Use whatever libev has to offer, but make sure that kqueue is
563 used if available (warning, breaks stuff, best use only with your own
564 private event loop and only if you know the OS supports your types of
565 fds):
567 ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
569 =item struct ev_loop *ev_loop_new (unsigned int flags)
571 Similar to C<ev_default_loop>, but always creates a new event loop that is
572 always distinct from the default loop. Unlike the default loop, it cannot
573 handle signal and child watchers, and attempts to do so will be greeted by
574 undefined behaviour (or a failed assertion if assertions are enabled).
576 Note that this function I<is> thread-safe, and the recommended way to use
577 libev with threads is indeed to create one loop per thread, and using the
578 default loop in the "main" or "initial" thread.
580 Example: Try to create a event loop that uses epoll and nothing else.
582 struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
583 if (!epoller)
584 fatal ("no epoll found here, maybe it hides under your chair");
586 =item ev_default_destroy ()
588 Destroys the default loop again (frees all memory and kernel state
589 etc.). None of the active event watchers will be stopped in the normal
590 sense, so e.g. C<ev_is_active> might still return true. It is your
591 responsibility to either stop all watchers cleanly yourself I<before>
592 calling this function, or cope with the fact afterwards (which is usually
593 the easiest thing, you can just ignore the watchers and/or C<free ()> them
594 for example).
596 Note that certain global state, such as signal state (and installed signal
597 handlers), will not be freed by this function, and related watchers (such
598 as signal and child watchers) would need to be stopped manually.
600 In general it is not advisable to call this function except in the
601 rare occasion where you really need to free e.g. the signal handling
602 pipe fds. If you need dynamically allocated loops it is better to use
603 C<ev_loop_new> and C<ev_loop_destroy>.
605 =item ev_loop_destroy (loop)
607 Like C<ev_default_destroy>, but destroys an event loop created by an
608 earlier call to C<ev_loop_new>.
610 =item ev_default_fork ()
612 This function sets a flag that causes subsequent C<ev_loop> iterations
613 to reinitialise the kernel state for backends that have one. Despite the
614 name, you can call it anytime, but it makes most sense after forking, in
615 the child process (or both child and parent, but that again makes little
616 sense). You I<must> call it in the child before using any of the libev
617 functions, and it will only take effect at the next C<ev_loop> iteration.
619 On the other hand, you only need to call this function in the child
620 process if and only if you want to use the event library in the child. If
621 you just fork+exec, you don't have to call it at all.
623 The function itself is quite fast and it's usually not a problem to call
624 it just in case after a fork. To make this easy, the function will fit in
625 quite nicely into a call to C<pthread_atfork>:
627 pthread_atfork (0, 0, ev_default_fork);
629 =item ev_loop_fork (loop)
631 Like C<ev_default_fork>, but acts on an event loop created by
632 C<ev_loop_new>. Yes, you have to call this on every allocated event loop
633 after fork that you want to re-use in the child, and how you do this is
634 entirely your own problem.
636 =item int ev_is_default_loop (loop)
638 Returns true when the given loop is, in fact, the default loop, and false
639 otherwise.
641 =item unsigned int ev_loop_count (loop)
643 Returns the count of loop iterations for the loop, which is identical to
644 the number of times libev did poll for new events. It starts at C<0> and
645 happily wraps around with enough iterations.
647 This value can sometimes be useful as a generation counter of sorts (it
648 "ticks" the number of loop iterations), as it roughly corresponds with
649 C<ev_prepare> and C<ev_check> calls.
651 =item unsigned int ev_loop_depth (loop)
653 Returns the number of times C<ev_loop> was entered minus the number of
654 times C<ev_loop> was exited, in other words, the recursion depth.
656 Outside C<ev_loop>, this number is zero. In a callback, this number is
657 C<1>, unless C<ev_loop> was invoked recursively (or from another thread),
658 in which case it is higher.
660 Leaving C<ev_loop> abnormally (setjmp/longjmp, cancelling the thread
661 etc.), doesn't count as exit.
663 =item unsigned int ev_backend (loop)
665 Returns one of the C<EVBACKEND_*> flags indicating the event backend in
666 use.
668 =item ev_tstamp ev_now (loop)
670 Returns the current "event loop time", which is the time the event loop
671 received events and started processing them. This timestamp does not
672 change as long as callbacks are being processed, and this is also the base
673 time used for relative timers. You can treat it as the timestamp of the
674 event occurring (or more correctly, libev finding out about it).
676 =item ev_now_update (loop)
678 Establishes the current time by querying the kernel, updating the time
679 returned by C<ev_now ()> in the progress. This is a costly operation and
680 is usually done automatically within C<ev_loop ()>.
682 This function is rarely useful, but when some event callback runs for a
683 very long time without entering the event loop, updating libev's idea of
684 the current time is a good idea.
686 See also L<The special problem of time updates> in the C<ev_timer> section.
688 =item ev_suspend (loop)
690 =item ev_resume (loop)
692 These two functions suspend and resume a loop, for use when the loop is
693 not used for a while and timeouts should not be processed.
695 A typical use case would be an interactive program such as a game: When
696 the user presses C<^Z> to suspend the game and resumes it an hour later it
697 would be best to handle timeouts as if no time had actually passed while
698 the program was suspended. This can be achieved by calling C<ev_suspend>
699 in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
700 C<ev_resume> directly afterwards to resume timer processing.
702 Effectively, all C<ev_timer> watchers will be delayed by the time spend
703 between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
704 will be rescheduled (that is, they will lose any events that would have
705 occured while suspended).
707 After calling C<ev_suspend> you B<must not> call I<any> function on the
708 given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
709 without a previous call to C<ev_suspend>.
711 Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
712 event loop time (see C<ev_now_update>).
714 =item ev_loop (loop, int flags)
716 Finally, this is it, the event handler. This function usually is called
717 after you have initialised all your watchers and you want to start
718 handling events.
720 If the flags argument is specified as C<0>, it will not return until
721 either no event watchers are active anymore or C<ev_unloop> was called.
723 Please note that an explicit C<ev_unloop> is usually better than
724 relying on all watchers to be stopped when deciding when a program has
725 finished (especially in interactive programs), but having a program
726 that automatically loops as long as it has to and no longer by virtue
727 of relying on its watchers stopping correctly, that is truly a thing of
728 beauty.
730 A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
731 those events and any already outstanding ones, but will not block your
732 process in case there are no events and will return after one iteration of
733 the loop.
735 A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
736 necessary) and will handle those and any already outstanding ones. It
737 will block your process until at least one new event arrives (which could
738 be an event internal to libev itself, so there is no guarantee that a
739 user-registered callback will be called), and will return after one
740 iteration of the loop.
742 This is useful if you are waiting for some external event in conjunction
743 with something not expressible using other libev watchers (i.e. "roll your
744 own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
745 usually a better approach for this kind of thing.
747 Here are the gory details of what C<ev_loop> does:
749 - Before the first iteration, call any pending watchers.
750 * If EVFLAG_FORKCHECK was used, check for a fork.
751 - If a fork was detected (by any means), queue and call all fork watchers.
752 - Queue and call all prepare watchers.
753 - If we have been forked, detach and recreate the kernel state
754 as to not disturb the other process.
755 - Update the kernel state with all outstanding changes.
756 - Update the "event loop time" (ev_now ()).
757 - Calculate for how long to sleep or block, if at all
758 (active idle watchers, EVLOOP_NONBLOCK or not having
759 any active watchers at all will result in not sleeping).
760 - Sleep if the I/O and timer collect interval say so.
761 - Block the process, waiting for any events.
762 - Queue all outstanding I/O (fd) events.
763 - Update the "event loop time" (ev_now ()), and do time jump adjustments.
764 - Queue all expired timers.
765 - Queue all expired periodics.
766 - Unless any events are pending now, queue all idle watchers.
767 - Queue all check watchers.
768 - Call all queued watchers in reverse order (i.e. check watchers first).
769 Signals and child watchers are implemented as I/O watchers, and will
770 be handled here by queueing them when their watcher gets executed.
771 - If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
772 were used, or there are no active watchers, return, otherwise
773 continue with step *.
775 Example: Queue some jobs and then loop until no events are outstanding
776 anymore.
778 ... queue jobs here, make sure they register event watchers as long
779 ... as they still have work to do (even an idle watcher will do..)
780 ev_loop (my_loop, 0);
781 ... jobs done or somebody called unloop. yeah!
783 =item ev_unloop (loop, how)
785 Can be used to make a call to C<ev_loop> return early (but only after it
786 has processed all outstanding events). The C<how> argument must be either
787 C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
788 C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
790 This "unloop state" will be cleared when entering C<ev_loop> again.
792 It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
794 =item ev_ref (loop)
796 =item ev_unref (loop)
798 Ref/unref can be used to add or remove a reference count on the event
799 loop: Every watcher keeps one reference, and as long as the reference
800 count is nonzero, C<ev_loop> will not return on its own.
802 This is useful when you have a watcher that you never intend to
803 unregister, but that nevertheless should not keep C<ev_loop> from
804 returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
805 before stopping it.
807 As an example, libev itself uses this for its internal signal pipe: It
808 is not visible to the libev user and should not keep C<ev_loop> from
809 exiting if no event watchers registered by it are active. It is also an
810 excellent way to do this for generic recurring timers or from within
811 third-party libraries. Just remember to I<unref after start> and I<ref
812 before stop> (but only if the watcher wasn't active before, or was active
813 before, respectively. Note also that libev might stop watchers itself
814 (e.g. non-repeating timers) in which case you have to C<ev_ref>
815 in the callback).
817 Example: Create a signal watcher, but keep it from keeping C<ev_loop>
818 running when nothing else is active.
820 ev_signal exitsig;
821 ev_signal_init (&exitsig, sig_cb, SIGINT);
822 ev_signal_start (loop, &exitsig);
823 evf_unref (loop);
825 Example: For some weird reason, unregister the above signal handler again.
827 ev_ref (loop);
828 ev_signal_stop (loop, &exitsig);
830 =item ev_set_io_collect_interval (loop, ev_tstamp interval)
832 =item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
834 These advanced functions influence the time that libev will spend waiting
835 for events. Both time intervals are by default C<0>, meaning that libev
836 will try to invoke timer/periodic callbacks and I/O callbacks with minimum
837 latency.
839 Setting these to a higher value (the C<interval> I<must> be >= C<0>)
840 allows libev to delay invocation of I/O and timer/periodic callbacks
841 to increase efficiency of loop iterations (or to increase power-saving
842 opportunities).
844 The idea is that sometimes your program runs just fast enough to handle
845 one (or very few) event(s) per loop iteration. While this makes the
846 program responsive, it also wastes a lot of CPU time to poll for new
847 events, especially with backends like C<select ()> which have a high
848 overhead for the actual polling but can deliver many events at once.
850 By setting a higher I<io collect interval> you allow libev to spend more
851 time collecting I/O events, so you can handle more events per iteration,
852 at the cost of increasing latency. Timeouts (both C<ev_periodic> and
853 C<ev_timer>) will be not affected. Setting this to a non-null value will
854 introduce an additional C<ev_sleep ()> call into most loop iterations. The
855 sleep time ensures that libev will not poll for I/O events more often then
856 once per this interval, on average.
858 Likewise, by setting a higher I<timeout collect interval> you allow libev
859 to spend more time collecting timeouts, at the expense of increased
860 latency/jitter/inexactness (the watcher callback will be called
861 later). C<ev_io> watchers will not be affected. Setting this to a non-null
862 value will not introduce any overhead in libev.
864 Many (busy) programs can usually benefit by setting the I/O collect
865 interval to a value near C<0.1> or so, which is often enough for
866 interactive servers (of course not for games), likewise for timeouts. It
867 usually doesn't make much sense to set it to a lower value than C<0.01>,
868 as this approaches the timing granularity of most systems. Note that if
869 you do transactions with the outside world and you can't increase the
870 parallelity, then this setting will limit your transaction rate (if you
871 need to poll once per transaction and the I/O collect interval is 0.01,
872 then you can't do more than 100 transations per second).
874 Setting the I<timeout collect interval> can improve the opportunity for
875 saving power, as the program will "bundle" timer callback invocations that
876 are "near" in time together, by delaying some, thus reducing the number of
877 times the process sleeps and wakes up again. Another useful technique to
878 reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
879 they fire on, say, one-second boundaries only.
881 Example: we only need 0.1s timeout granularity, and we wish not to poll
882 more often than 100 times per second:
884 ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
885 ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
887 =item ev_invoke_pending (loop)
889 This call will simply invoke all pending watchers while resetting their
890 pending state. Normally, C<ev_loop> does this automatically when required,
891 but when overriding the invoke callback this call comes handy.
893 =item int ev_pending_count (loop)
895 Returns the number of pending watchers - zero indicates that no watchers
896 are pending.
898 =item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
900 This overrides the invoke pending functionality of the loop: Instead of
901 invoking all pending watchers when there are any, C<ev_loop> will call
902 this callback instead. This is useful, for example, when you want to
903 invoke the actual watchers inside another context (another thread etc.).
905 If you want to reset the callback, use C<ev_invoke_pending> as new
906 callback.
908 =item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
910 Sometimes you want to share the same loop between multiple threads. This
911 can be done relatively simply by putting mutex_lock/unlock calls around
912 each call to a libev function.
914 However, C<ev_loop> can run an indefinite time, so it is not feasible to
915 wait for it to return. One way around this is to wake up the loop via
916 C<ev_unloop> and C<av_async_send>, another way is to set these I<release>
917 and I<acquire> callbacks on the loop.
919 When set, then C<release> will be called just before the thread is
920 suspended waiting for new events, and C<acquire> is called just
921 afterwards.
923 Ideally, C<release> will just call your mutex_unlock function, and
924 C<acquire> will just call the mutex_lock function again.
926 While event loop modifications are allowed between invocations of
927 C<release> and C<acquire> (that's their only purpose after all), no
928 modifications done will affect the event loop, i.e. adding watchers will
929 have no effect on the set of file descriptors being watched, or the time
930 waited. Use an C<ev_async> watcher to wake up C<ev_loop> when you want it
931 to take note of any changes you made.
933 In theory, threads executing C<ev_loop> will be async-cancel safe between
934 invocations of C<release> and C<acquire>.
936 See also the locking example in the C<THREADS> section later in this
937 document.
939 =item ev_set_userdata (loop, void *data)
941 =item ev_userdata (loop)
943 Set and retrieve a single C<void *> associated with a loop. When
944 C<ev_set_userdata> has never been called, then C<ev_userdata> returns
945 C<0.>
947 These two functions can be used to associate arbitrary data with a loop,
948 and are intended solely for the C<invoke_pending_cb>, C<release> and
949 C<acquire> callbacks described above, but of course can be (ab-)used for
950 any other purpose as well.
952 =item ev_loop_verify (loop)
954 This function only does something when C<EV_VERIFY> support has been
955 compiled in, which is the default for non-minimal builds. It tries to go
956 through all internal structures and checks them for validity. If anything
957 is found to be inconsistent, it will print an error message to standard
958 error and call C<abort ()>.
960 This can be used to catch bugs inside libev itself: under normal
961 circumstances, this function will never abort as of course libev keeps its
962 data structures consistent.
964 =back
969 In the following description, uppercase C<TYPE> in names stands for the
970 watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
971 watchers and C<ev_io_start> for I/O watchers.
973 A watcher is a structure that you create and register to record your
974 interest in some event. For instance, if you want to wait for STDIN to
975 become readable, you would create an C<ev_io> watcher for that:
977 static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
978 {
979 ev_io_stop (w);
980 ev_unloop (loop, EVUNLOOP_ALL);
981 }
983 struct ev_loop *loop = ev_default_loop (0);
985 ev_io stdin_watcher;
987 ev_init (&stdin_watcher, my_cb);
988 ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
989 ev_io_start (loop, &stdin_watcher);
991 ev_loop (loop, 0);
993 As you can see, you are responsible for allocating the memory for your
994 watcher structures (and it is I<usually> a bad idea to do this on the
995 stack).
997 Each watcher has an associated watcher structure (called C<struct ev_TYPE>
998 or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
1000 Each watcher structure must be initialised by a call to C<ev_init
1001 (watcher *, callback)>, which expects a callback to be provided. This
1002 callback gets invoked each time the event occurs (or, in the case of I/O
1003 watchers, each time the event loop detects that the file descriptor given
1004 is readable and/or writable).
1006 Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
1007 macro to configure it, with arguments specific to the watcher type. There
1008 is also a macro to combine initialisation and setting in one call: C<<
1009 ev_TYPE_init (watcher *, callback, ...) >>.
1011 To make the watcher actually watch out for events, you have to start it
1012 with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
1013 *) >>), and you can stop watching for events at any time by calling the
1014 corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
1016 As long as your watcher is active (has been started but not stopped) you
1017 must not touch the values stored in it. Most specifically you must never
1018 reinitialise it or call its C<ev_TYPE_set> macro.
1020 Each and every callback receives the event loop pointer as first, the
1021 registered watcher structure as second, and a bitset of received events as
1022 third argument.
1024 The received events usually include a single bit per event type received
1025 (you can receive multiple events at the same time). The possible bit masks
1026 are:
1028 =over 4
1030 =item C<EV_READ>
1032 =item C<EV_WRITE>
1034 The file descriptor in the C<ev_io> watcher has become readable and/or
1035 writable.
1037 =item C<EV_TIMEOUT>
1039 The C<ev_timer> watcher has timed out.
1041 =item C<EV_PERIODIC>
1043 The C<ev_periodic> watcher has timed out.
1045 =item C<EV_SIGNAL>
1047 The signal specified in the C<ev_signal> watcher has been received by a thread.
1049 =item C<EV_CHILD>
1051 The pid specified in the C<ev_child> watcher has received a status change.
1053 =item C<EV_STAT>
1055 The path specified in the C<ev_stat> watcher changed its attributes somehow.
1057 =item C<EV_IDLE>
1059 The C<ev_idle> watcher has determined that you have nothing better to do.
1061 =item C<EV_PREPARE>
1063 =item C<EV_CHECK>
1065 All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts
1066 to gather new events, and all C<ev_check> watchers are invoked just after
1067 C<ev_loop> has gathered them, but before it invokes any callbacks for any
1068 received events. Callbacks of both watcher types can start and stop as
1069 many watchers as they want, and all of them will be taken into account
1070 (for example, a C<ev_prepare> watcher might start an idle watcher to keep
1071 C<ev_loop> from blocking).
1073 =item C<EV_EMBED>
1075 The embedded event loop specified in the C<ev_embed> watcher needs attention.
1077 =item C<EV_FORK>
1079 The event loop has been resumed in the child process after fork (see
1080 C<ev_fork>).
1082 =item C<EV_ASYNC>
1084 The given async watcher has been asynchronously notified (see C<ev_async>).
1086 =item C<EV_CUSTOM>
1088 Not ever sent (or otherwise used) by libev itself, but can be freely used
1089 by libev users to signal watchers (e.g. via C<ev_feed_event>).
1091 =item C<EV_ERROR>
1093 An unspecified error has occurred, the watcher has been stopped. This might
1094 happen because the watcher could not be properly started because libev
1095 ran out of memory, a file descriptor was found to be closed or any other
1096 problem. Libev considers these application bugs.
1098 You best act on it by reporting the problem and somehow coping with the
1099 watcher being stopped. Note that well-written programs should not receive
1100 an error ever, so when your watcher receives it, this usually indicates a
1101 bug in your program.
1103 Libev will usually signal a few "dummy" events together with an error, for
1104 example it might indicate that a fd is readable or writable, and if your
1105 callbacks is well-written it can just attempt the operation and cope with
1106 the error from read() or write(). This will not work in multi-threaded
1107 programs, though, as the fd could already be closed and reused for another
1108 thing, so beware.
1110 =back
1114 =over 4
1116 =item C<ev_init> (ev_TYPE *watcher, callback)
1118 This macro initialises the generic portion of a watcher. The contents
1119 of the watcher object can be arbitrary (so C<malloc> will do). Only
1120 the generic parts of the watcher are initialised, you I<need> to call
1121 the type-specific C<ev_TYPE_set> macro afterwards to initialise the
1122 type-specific parts. For each type there is also a C<ev_TYPE_init> macro
1123 which rolls both calls into one.
1125 You can reinitialise a watcher at any time as long as it has been stopped
1126 (or never started) and there are no pending events outstanding.
1128 The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
1129 int revents)>.
1131 Example: Initialise an C<ev_io> watcher in two steps.
1133 ev_io w;
1134 ev_init (&w, my_cb);
1135 ev_io_set (&w, STDIN_FILENO, EV_READ);
1137 =item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
1139 This macro initialises the type-specific parts of a watcher. You need to
1140 call C<ev_init> at least once before you call this macro, but you can
1141 call C<ev_TYPE_set> any number of times. You must not, however, call this
1142 macro on a watcher that is active (it can be pending, however, which is a
1143 difference to the C<ev_init> macro).
1145 Although some watcher types do not have type-specific arguments
1146 (e.g. C<ev_prepare>) you still need to call its C<set> macro.
1148 See C<ev_init>, above, for an example.
1150 =item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
1152 This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
1153 calls into a single call. This is the most convenient method to initialise
1154 a watcher. The same limitations apply, of course.
1156 Example: Initialise and set an C<ev_io> watcher in one step.
1158 ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
1160 =item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
1162 Starts (activates) the given watcher. Only active watchers will receive
1163 events. If the watcher is already active nothing will happen.
1165 Example: Start the C<ev_io> watcher that is being abused as example in this
1166 whole section.
1168 ev_io_start (EV_DEFAULT_UC, &w);
1170 =item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
1172 Stops the given watcher if active, and clears the pending status (whether
1173 the watcher was active or not).
1175 It is possible that stopped watchers are pending - for example,
1176 non-repeating timers are being stopped when they become pending - but
1177 calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
1178 pending. If you want to free or reuse the memory used by the watcher it is
1179 therefore a good idea to always call its C<ev_TYPE_stop> function.
1181 =item bool ev_is_active (ev_TYPE *watcher)
1183 Returns a true value iff the watcher is active (i.e. it has been started
1184 and not yet been stopped). As long as a watcher is active you must not modify
1185 it.
1187 =item bool ev_is_pending (ev_TYPE *watcher)
1189 Returns a true value iff the watcher is pending, (i.e. it has outstanding
1190 events but its callback has not yet been invoked). As long as a watcher
1191 is pending (but not active) you must not call an init function on it (but
1192 C<ev_TYPE_set> is safe), you must not change its priority, and you must
1193 make sure the watcher is available to libev (e.g. you cannot C<free ()>
1194 it).
1196 =item callback ev_cb (ev_TYPE *watcher)
1198 Returns the callback currently set on the watcher.
1200 =item ev_cb_set (ev_TYPE *watcher, callback)
1202 Change the callback. You can change the callback at virtually any time
1203 (modulo threads).
1205 =item ev_set_priority (ev_TYPE *watcher, int priority)
1207 =item int ev_priority (ev_TYPE *watcher)
1209 Set and query the priority of the watcher. The priority is a small
1210 integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
1211 (default: C<-2>). Pending watchers with higher priority will be invoked
1212 before watchers with lower priority, but priority will not keep watchers
1213 from being executed (except for C<ev_idle> watchers).
1215 If you need to suppress invocation when higher priority events are pending
1216 you need to look at C<ev_idle> watchers, which provide this functionality.
1218 You I<must not> change the priority of a watcher as long as it is active or
1219 pending.
1221 Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
1222 fine, as long as you do not mind that the priority value you query might
1223 or might not have been clamped to the valid range.
1225 The default priority used by watchers when no priority has been set is
1226 always C<0>, which is supposed to not be too high and not be too low :).
1228 See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
1229 priorities.
1231 =item ev_invoke (loop, ev_TYPE *watcher, int revents)
1233 Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
1234 C<loop> nor C<revents> need to be valid as long as the watcher callback
1235 can deal with that fact, as both are simply passed through to the
1236 callback.
1238 =item int ev_clear_pending (loop, ev_TYPE *watcher)
1240 If the watcher is pending, this function clears its pending status and
1241 returns its C<revents> bitset (as if its callback was invoked). If the
1242 watcher isn't pending it does nothing and returns C<0>.
1244 Sometimes it can be useful to "poll" a watcher instead of waiting for its
1245 callback to be invoked, which can be accomplished with this function.
1247 =item ev_feed_event (loop, ev_TYPE *watcher, int revents)
1249 Feeds the given event set into the event loop, as if the specified event
1250 had happened for the specified watcher (which must be a pointer to an
1251 initialised but not necessarily started event watcher). Obviously you must
1252 not free the watcher as long as it has pending events.
1254 Stopping the watcher, letting libev invoke it, or calling
1255 C<ev_clear_pending> will clear the pending event, even if the watcher was
1256 not started in the first place.
1258 See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1259 functions that do not need a watcher.
1261 =back
1266 Each watcher has, by default, a member C<void *data> that you can change
1267 and read at any time: libev will completely ignore it. This can be used
1268 to associate arbitrary data with your watcher. If you need more data and
1269 don't want to allocate memory and store a pointer to it in that data
1270 member, you can also "subclass" the watcher type and provide your own
1271 data:
1273 struct my_io
1274 {
1275 ev_io io;
1276 int otherfd;
1277 void *somedata;
1278 struct whatever *mostinteresting;
1279 };
1281 ...
1282 struct my_io w;
1283 ev_io_init (&, my_cb, fd, EV_READ);
1285 And since your callback will be called with a pointer to the watcher, you
1286 can cast it back to your own type:
1288 static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
1289 {
1290 struct my_io *w = (struct my_io *)w_;
1291 ...
1292 }
1294 More interesting and less C-conformant ways of casting your callback type
1295 instead have been omitted.
1297 Another common scenario is to use some data structure with multiple
1298 embedded watchers:
1300 struct my_biggy
1301 {
1302 int some_data;
1303 ev_timer t1;
1304 ev_timer t2;
1305 }
1307 In this case getting the pointer to C<my_biggy> is a bit more
1308 complicated: Either you store the address of your C<my_biggy> struct
1309 in the C<data> member of the watcher (for woozies), or you need to use
1310 some pointer arithmetic using C<offsetof> inside your watchers (for real
1311 programmers):
1313 #include <stddef.h>
1315 static void
1316 t1_cb (EV_P_ ev_timer *w, int revents)
1317 {
1318 struct my_biggy big = (struct my_biggy *)
1319 (((char *)w) - offsetof (struct my_biggy, t1));
1320 }
1322 static void
1323 t2_cb (EV_P_ ev_timer *w, int revents)
1324 {
1325 struct my_biggy big = (struct my_biggy *)
1326 (((char *)w) - offsetof (struct my_biggy, t2));
1327 }
1331 Many event loops support I<watcher priorities>, which are usually small
1332 integers that influence the ordering of event callback invocation
1333 between watchers in some way, all else being equal.
1335 In libev, Watcher priorities can be set using C<ev_set_priority>. See its
1336 description for the more technical details such as the actual priority
1337 range.
1339 There are two common ways how these these priorities are being interpreted
1340 by event loops:
1342 In the more common lock-out model, higher priorities "lock out" invocation
1343 of lower priority watchers, which means as long as higher priority
1344 watchers receive events, lower priority watchers are not being invoked.
1346 The less common only-for-ordering model uses priorities solely to order
1347 callback invocation within a single event loop iteration: Higher priority
1348 watchers are invoked before lower priority ones, but they all get invoked
1349 before polling for new events.
1351 Libev uses the second (only-for-ordering) model for all its watchers
1352 except for idle watchers (which use the lock-out model).
1354 The rationale behind this is that implementing the lock-out model for
1355 watchers is not well supported by most kernel interfaces, and most event
1356 libraries will just poll for the same events again and again as long as
1357 their callbacks have not been executed, which is very inefficient in the
1358 common case of one high-priority watcher locking out a mass of lower
1359 priority ones.
1361 Static (ordering) priorities are most useful when you have two or more
1362 watchers handling the same resource: a typical usage example is having an
1363 C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
1364 timeouts. Under load, data might be received while the program handles
1365 other jobs, but since timers normally get invoked first, the timeout
1366 handler will be executed before checking for data. In that case, giving
1367 the timer a lower priority than the I/O watcher ensures that I/O will be
1368 handled first even under adverse conditions (which is usually, but not
1369 always, what you want).
1371 Since idle watchers use the "lock-out" model, meaning that idle watchers
1372 will only be executed when no same or higher priority watchers have
1373 received events, they can be used to implement the "lock-out" model when
1374 required.
1376 For example, to emulate how many other event libraries handle priorities,
1377 you can associate an C<ev_idle> watcher to each such watcher, and in
1378 the normal watcher callback, you just start the idle watcher. The real
1379 processing is done in the idle watcher callback. This causes libev to
1380 continously poll and process kernel event data for the watcher, but when
1381 the lock-out case is known to be rare (which in turn is rare :), this is
1382 workable.
1384 Usually, however, the lock-out model implemented that way will perform
1385 miserably under the type of load it was designed to handle. In that case,
1386 it might be preferable to stop the real watcher before starting the
1387 idle watcher, so the kernel will not have to process the event in case
1388 the actual processing will be delayed for considerable time.
1390 Here is an example of an I/O watcher that should run at a strictly lower
1391 priority than the default, and which should only process data when no
1392 other events are pending:
1394 ev_idle idle; // actual processing watcher
1395 ev_io io; // actual event watcher
1397 static void
1398 io_cb (EV_P_ ev_io *w, int revents)
1399 {
1400 // stop the I/O watcher, we received the event, but
1401 // are not yet ready to handle it.
1402 ev_io_stop (EV_A_ w);
1404 // start the idle watcher to ahndle the actual event.
1405 // it will not be executed as long as other watchers
1406 // with the default priority are receiving events.
1407 ev_idle_start (EV_A_ &idle);
1408 }
1410 static void
1411 idle_cb (EV_P_ ev_idle *w, int revents)
1412 {
1413 // actual processing
1414 read (STDIN_FILENO, ...);
1416 // have to start the I/O watcher again, as
1417 // we have handled the event
1418 ev_io_start (EV_P_ &io);
1419 }
1421 // initialisation
1422 ev_idle_init (&idle, idle_cb);
1423 ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
1424 ev_io_start (EV_DEFAULT_ &io);
1426 In the "real" world, it might also be beneficial to start a timer, so that
1427 low-priority connections can not be locked out forever under load. This
1428 enables your program to keep a lower latency for important connections
1429 during short periods of high load, while not completely locking out less
1430 important ones.
1433 =head1 WATCHER TYPES
1435 This section describes each watcher in detail, but will not repeat
1436 information given in the last section. Any initialisation/set macros,
1437 functions and members specific to the watcher type are explained.
1439 Members are additionally marked with either I<[read-only]>, meaning that,
1440 while the watcher is active, you can look at the member and expect some
1441 sensible content, but you must not modify it (you can modify it while the
1442 watcher is stopped to your hearts content), or I<[read-write]>, which
1443 means you can expect it to have some sensible content while the watcher
1444 is active, but you can also modify it. Modifying it may not do something
1445 sensible or take immediate effect (or do anything at all), but libev will
1446 not crash or malfunction in any way.
1449 =head2 C<ev_io> - is this file descriptor readable or writable?
1451 I/O watchers check whether a file descriptor is readable or writable
1452 in each iteration of the event loop, or, more precisely, when reading
1453 would not block the process and writing would at least be able to write
1454 some data. This behaviour is called level-triggering because you keep
1455 receiving events as long as the condition persists. Remember you can stop
1456 the watcher if you don't want to act on the event and neither want to
1457 receive future events.
1459 In general you can register as many read and/or write event watchers per
1460 fd as you want (as long as you don't confuse yourself). Setting all file
1461 descriptors to non-blocking mode is also usually a good idea (but not
1462 required if you know what you are doing).
1464 If you cannot use non-blocking mode, then force the use of a
1465 known-to-be-good backend (at the time of this writing, this includes only
1466 C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
1467 descriptors for which non-blocking operation makes no sense (such as
1468 files) - libev doesn't guarentee any specific behaviour in that case.
1470 Another thing you have to watch out for is that it is quite easy to
1471 receive "spurious" readiness notifications, that is your callback might
1472 be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1473 because there is no data. Not only are some backends known to create a
1474 lot of those (for example Solaris ports), it is very easy to get into
1475 this situation even with a relatively standard program structure. Thus
1476 it is best to always use non-blocking I/O: An extra C<read>(2) returning
1477 C<EAGAIN> is far preferable to a program hanging until some data arrives.
1479 If you cannot run the fd in non-blocking mode (for example you should
1480 not play around with an Xlib connection), then you have to separately
1481 re-test whether a file descriptor is really ready with a known-to-be good
1482 interface such as poll (fortunately in our Xlib example, Xlib already
1483 does this on its own, so its quite safe to use). Some people additionally
1484 use C<SIGALRM> and an interval timer, just to be sure you won't block
1485 indefinitely.
1487 But really, best use non-blocking mode.
1489 =head3 The special problem of disappearing file descriptors
1491 Some backends (e.g. kqueue, epoll) need to be told about closing a file
1492 descriptor (either due to calling C<close> explicitly or any other means,
1493 such as C<dup2>). The reason is that you register interest in some file
1494 descriptor, but when it goes away, the operating system will silently drop
1495 this interest. If another file descriptor with the same number then is
1496 registered with libev, there is no efficient way to see that this is, in
1497 fact, a different file descriptor.
1499 To avoid having to explicitly tell libev about such cases, libev follows
1500 the following policy: Each time C<ev_io_set> is being called, libev
1501 will assume that this is potentially a new file descriptor, otherwise
1502 it is assumed that the file descriptor stays the same. That means that
1503 you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
1504 descriptor even if the file descriptor number itself did not change.
1506 This is how one would do it normally anyway, the important point is that
1507 the libev application should not optimise around libev but should leave
1508 optimisations to libev.
1510 =head3 The special problem of dup'ed file descriptors
1512 Some backends (e.g. epoll), cannot register events for file descriptors,
1513 but only events for the underlying file descriptions. That means when you
1514 have C<dup ()>'ed file descriptors or weirder constellations, and register
1515 events for them, only one file descriptor might actually receive events.
1517 There is no workaround possible except not registering events
1518 for potentially C<dup ()>'ed file descriptors, or to resort to
1521 =head3 The special problem of fork
1523 Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1524 useless behaviour. Libev fully supports fork, but needs to be told about
1525 it in the child.
1527 To support fork in your programs, you either have to call
1528 C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
1529 enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
1532 =head3 The special problem of SIGPIPE
1534 While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1535 when writing to a pipe whose other end has been closed, your program gets
1536 sent a SIGPIPE, which, by default, aborts your program. For most programs
1537 this is sensible behaviour, for daemons, this is usually undesirable.
1539 So when you encounter spurious, unexplained daemon exits, make sure you
1540 ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
1541 somewhere, as that would have given you a big clue).
1544 =head3 Watcher-Specific Functions
1546 =over 4
1548 =item ev_io_init (ev_io *, callback, int fd, int events)
1550 =item ev_io_set (ev_io *, int fd, int events)
1552 Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
1553 receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
1554 C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
1556 =item int fd [read-only]
1558 The file descriptor being watched.
1560 =item int events [read-only]
1562 The events being watched.
1564 =back
1566 =head3 Examples
1568 Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1569 readable, but only once. Since it is likely line-buffered, you could
1570 attempt to read a whole line in the callback.
1572 static void
1573 stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
1574 {
1575 ev_io_stop (loop, w);
1576 .. read from stdin here (or from w->fd) and handle any I/O errors
1577 }
1579 ...
1580 struct ev_loop *loop = ev_default_init (0);
1581 ev_io stdin_readable;
1582 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1583 ev_io_start (loop, &stdin_readable);
1584 ev_loop (loop, 0);
1587 =head2 C<ev_timer> - relative and optionally repeating timeouts
1589 Timer watchers are simple relative timers that generate an event after a
1590 given time, and optionally repeating in regular intervals after that.
1592 The timers are based on real time, that is, if you register an event that
1593 times out after an hour and you reset your system clock to January last
1594 year, it will still time out after (roughly) one hour. "Roughly" because
1595 detecting time jumps is hard, and some inaccuracies are unavoidable (the
1596 monotonic clock option helps a lot here).
1598 The callback is guaranteed to be invoked only I<after> its timeout has
1599 passed (not I<at>, so on systems with very low-resolution clocks this
1600 might introduce a small delay). If multiple timers become ready during the
1601 same loop iteration then the ones with earlier time-out values are invoked
1602 before ones of the same priority with later time-out values (but this is
1603 no longer true when a callback calls C<ev_loop> recursively).
1605 =head3 Be smart about timeouts
1607 Many real-world problems involve some kind of timeout, usually for error
1608 recovery. A typical example is an HTTP request - if the other side hangs,
1609 you want to raise some error after a while.
1611 What follows are some ways to handle this problem, from obvious and
1612 inefficient to smart and efficient.
1614 In the following, a 60 second activity timeout is assumed - a timeout that
1615 gets reset to 60 seconds each time there is activity (e.g. each time some
1616 data or other life sign was received).
1618 =over 4
1620 =item 1. Use a timer and stop, reinitialise and start it on activity.
1622 This is the most obvious, but not the most simple way: In the beginning,
1623 start the watcher:
1625 ev_timer_init (timer, callback, 60., 0.);
1626 ev_timer_start (loop, timer);
1628 Then, each time there is some activity, C<ev_timer_stop> it, initialise it
1629 and start it again:
1631 ev_timer_stop (loop, timer);
1632 ev_timer_set (timer, 60., 0.);
1633 ev_timer_start (loop, timer);
1635 This is relatively simple to implement, but means that each time there is
1636 some activity, libev will first have to remove the timer from its internal
1637 data structure and then add it again. Libev tries to be fast, but it's
1638 still not a constant-time operation.
1640 =item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
1642 This is the easiest way, and involves using C<ev_timer_again> instead of
1643 C<ev_timer_start>.
1645 To implement this, configure an C<ev_timer> with a C<repeat> value
1646 of C<60> and then call C<ev_timer_again> at start and each time you
1647 successfully read or write some data. If you go into an idle state where
1648 you do not expect data to travel on the socket, you can C<ev_timer_stop>
1649 the timer, and C<ev_timer_again> will automatically restart it if need be.
1651 That means you can ignore both the C<ev_timer_start> function and the
1652 C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
1653 member and C<ev_timer_again>.
1655 At start:
1657 ev_init (timer, callback);
1658 timer->repeat = 60.;
1659 ev_timer_again (loop, timer);
1661 Each time there is some activity:
1663 ev_timer_again (loop, timer);
1665 It is even possible to change the time-out on the fly, regardless of
1666 whether the watcher is active or not:
1668 timer->repeat = 30.;
1669 ev_timer_again (loop, timer);
1671 This is slightly more efficient then stopping/starting the timer each time
1672 you want to modify its timeout value, as libev does not have to completely
1673 remove and re-insert the timer from/into its internal data structure.
1675 It is, however, even simpler than the "obvious" way to do it.
1677 =item 3. Let the timer time out, but then re-arm it as required.
1679 This method is more tricky, but usually most efficient: Most timeouts are
1680 relatively long compared to the intervals between other activity - in
1681 our example, within 60 seconds, there are usually many I/O events with
1682 associated activity resets.
1684 In this case, it would be more efficient to leave the C<ev_timer> alone,
1685 but remember the time of last activity, and check for a real timeout only
1686 within the callback:
1688 ev_tstamp last_activity; // time of last activity
1690 static void
1691 callback (EV_P_ ev_timer *w, int revents)
1692 {
1693 ev_tstamp now = ev_now (EV_A);
1694 ev_tstamp timeout = last_activity + 60.;
1696 // if last_activity + 60. is older than now, we did time out
1697 if (timeout < now)
1698 {
1699 // timeout occured, take action
1700 }
1701 else
1702 {
1703 // callback was invoked, but there was some activity, re-arm
1704 // the watcher to fire in last_activity + 60, which is
1705 // guaranteed to be in the future, so "again" is positive:
1706 w->repeat = timeout - now;
1707 ev_timer_again (EV_A_ w);
1708 }
1709 }
1711 To summarise the callback: first calculate the real timeout (defined
1712 as "60 seconds after the last activity"), then check if that time has
1713 been reached, which means something I<did>, in fact, time out. Otherwise
1714 the callback was invoked too early (C<timeout> is in the future), so
1715 re-schedule the timer to fire at that future time, to see if maybe we have
1716 a timeout then.
1718 Note how C<ev_timer_again> is used, taking advantage of the
1719 C<ev_timer_again> optimisation when the timer is already running.
1721 This scheme causes more callback invocations (about one every 60 seconds
1722 minus half the average time between activity), but virtually no calls to
1723 libev to change the timeout.
1725 To start the timer, simply initialise the watcher and set C<last_activity>
1726 to the current time (meaning we just have some activity :), then call the
1727 callback, which will "do the right thing" and start the timer:
1729 ev_init (timer, callback);
1730 last_activity = ev_now (loop);
1731 callback (loop, timer, EV_TIMEOUT);
1733 And when there is some activity, simply store the current time in
1734 C<last_activity>, no libev calls at all:
1736 last_actiivty = ev_now (loop);
1738 This technique is slightly more complex, but in most cases where the
1739 time-out is unlikely to be triggered, much more efficient.
1741 Changing the timeout is trivial as well (if it isn't hard-coded in the
1742 callback :) - just change the timeout and invoke the callback, which will
1743 fix things for you.
1745 =item 4. Wee, just use a double-linked list for your timeouts.
1747 If there is not one request, but many thousands (millions...), all
1748 employing some kind of timeout with the same timeout value, then one can
1749 do even better:
1751 When starting the timeout, calculate the timeout value and put the timeout
1752 at the I<end> of the list.
1754 Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
1755 the list is expected to fire (for example, using the technique #3).
1757 When there is some activity, remove the timer from the list, recalculate
1758 the timeout, append it to the end of the list again, and make sure to
1759 update the C<ev_timer> if it was taken from the beginning of the list.
1761 This way, one can manage an unlimited number of timeouts in O(1) time for
1762 starting, stopping and updating the timers, at the expense of a major
1763 complication, and having to use a constant timeout. The constant timeout
1764 ensures that the list stays sorted.
1766 =back
1768 So which method the best?
1770 Method #2 is a simple no-brain-required solution that is adequate in most
1771 situations. Method #3 requires a bit more thinking, but handles many cases
1772 better, and isn't very complicated either. In most case, choosing either
1773 one is fine, with #3 being better in typical situations.
1775 Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
1776 rather complicated, but extremely efficient, something that really pays
1777 off after the first million or so of active timers, i.e. it's usually
1778 overkill :)
1780 =head3 The special problem of time updates
1782 Establishing the current time is a costly operation (it usually takes at
1783 least two system calls): EV therefore updates its idea of the current
1784 time only before and after C<ev_loop> collects new events, which causes a
1785 growing difference between C<ev_now ()> and C<ev_time ()> when handling
1786 lots of events in one iteration.
1788 The relative timeouts are calculated relative to the C<ev_now ()>
1789 time. This is usually the right thing as this timestamp refers to the time
1790 of the event triggering whatever timeout you are modifying/starting. If
1791 you suspect event processing to be delayed and you I<need> to base the
1792 timeout on the current time, use something like this to adjust for this:
1794 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1796 If the event loop is suspended for a long time, you can also force an
1797 update of the time returned by C<ev_now ()> by calling C<ev_now_update
1798 ()>.
1800 =head3 The special problems of suspended animation
1802 When you leave the server world it is quite customary to hit machines that
1803 can suspend/hibernate - what happens to the clocks during such a suspend?
1805 Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
1806 all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
1807 to run until the system is suspended, but they will not advance while the
1808 system is suspended. That means, on resume, it will be as if the program
1809 was frozen for a few seconds, but the suspend time will not be counted
1810 towards C<ev_timer> when a monotonic clock source is used. The real time
1811 clock advanced as expected, but if it is used as sole clocksource, then a
1812 long suspend would be detected as a time jump by libev, and timers would
1813 be adjusted accordingly.
1815 I would not be surprised to see different behaviour in different between
1816 operating systems, OS versions or even different hardware.
1818 The other form of suspend (job control, or sending a SIGSTOP) will see a
1819 time jump in the monotonic clocks and the realtime clock. If the program
1820 is suspended for a very long time, and monotonic clock sources are in use,
1821 then you can expect C<ev_timer>s to expire as the full suspension time
1822 will be counted towards the timers. When no monotonic clock source is in
1823 use, then libev will again assume a timejump and adjust accordingly.
1825 It might be beneficial for this latter case to call C<ev_suspend>
1826 and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
1827 deterministic behaviour in this case (you can do nothing against
1828 C<SIGSTOP>).
1830 =head3 Watcher-Specific Functions and Data Members
1832 =over 4
1834 =item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
1836 =item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
1838 Configure the timer to trigger after C<after> seconds. If C<repeat>
1839 is C<0.>, then it will automatically be stopped once the timeout is
1840 reached. If it is positive, then the timer will automatically be
1841 configured to trigger again C<repeat> seconds later, again, and again,
1842 until stopped manually.
1844 The timer itself will do a best-effort at avoiding drift, that is, if
1845 you configure a timer to trigger every 10 seconds, then it will normally
1846 trigger at exactly 10 second intervals. If, however, your program cannot
1847 keep up with the timer (because it takes longer than those 10 seconds to
1848 do stuff) the timer will not fire more than once per event loop iteration.
1850 =item ev_timer_again (loop, ev_timer *)
1852 This will act as if the timer timed out and restart it again if it is
1853 repeating. The exact semantics are:
1855 If the timer is pending, its pending status is cleared.
1857 If the timer is started but non-repeating, stop it (as if it timed out).
1859 If the timer is repeating, either start it if necessary (with the
1860 C<repeat> value), or reset the running timer to the C<repeat> value.
1862 This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
1863 usage example.
1865 =item ev_tstamp ev_timer_remaining (loop, ev_timer *)
1867 Returns the remaining time until a timer fires. If the timer is active,
1868 then this time is relative to the current event loop time, otherwise it's
1869 the timeout value currently configured.
1871 That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
1872 C<5>. When the timer is started and one second passes, C<ev_timer_remain>
1873 will return C<4>. When the timer expires and is restarted, it will return
1874 roughly C<7> (likely slightly less as callback invocation takes some time,
1875 too), and so on.
1877 =item ev_tstamp repeat [read-write]
1879 The current C<repeat> value. Will be used each time the watcher times out
1880 or C<ev_timer_again> is called, and determines the next timeout (if any),
1881 which is also when any modifications are taken into account.
1883 =back
1885 =head3 Examples
1887 Example: Create a timer that fires after 60 seconds.
1889 static void
1890 one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1891 {
1892 .. one minute over, w is actually stopped right here
1893 }
1895 ev_timer mytimer;
1896 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1897 ev_timer_start (loop, &mytimer);
1899 Example: Create a timeout timer that times out after 10 seconds of
1900 inactivity.
1902 static void
1903 timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1904 {
1905 .. ten seconds without any activity
1906 }
1908 ev_timer mytimer;
1909 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1910 ev_timer_again (&mytimer); /* start timer */
1911 ev_loop (loop, 0);
1913 // and in some piece of code that gets executed on any "activity":
1914 // reset the timeout to start ticking again at 10 seconds
1915 ev_timer_again (&mytimer);
1918 =head2 C<ev_periodic> - to cron or not to cron?
1920 Periodic watchers are also timers of a kind, but they are very versatile
1921 (and unfortunately a bit complex).
1923 Unlike C<ev_timer>, periodic watchers are not based on real time (or
1924 relative time, the physical time that passes) but on wall clock time
1925 (absolute time, the thing you can read on your calender or clock). The
1926 difference is that wall clock time can run faster or slower than real
1927 time, and time jumps are not uncommon (e.g. when you adjust your
1928 wrist-watch).
1930 You can tell a periodic watcher to trigger after some specific point
1931 in time: for example, if you tell a periodic watcher to trigger "in 10
1932 seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
1933 not a delay) and then reset your system clock to January of the previous
1934 year, then it will take a year or more to trigger the event (unlike an
1935 C<ev_timer>, which would still trigger roughly 10 seconds after starting
1936 it, as it uses a relative timeout).
1938 C<ev_periodic> watchers can also be used to implement vastly more complex
1939 timers, such as triggering an event on each "midnight, local time", or
1940 other complicated rules. This cannot be done with C<ev_timer> watchers, as
1941 those cannot react to time jumps.
1943 As with timers, the callback is guaranteed to be invoked only when the
1944 point in time where it is supposed to trigger has passed. If multiple
1945 timers become ready during the same loop iteration then the ones with
1946 earlier time-out values are invoked before ones with later time-out values
1947 (but this is no longer true when a callback calls C<ev_loop> recursively).
1949 =head3 Watcher-Specific Functions and Data Members
1951 =over 4
1953 =item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1955 =item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1957 Lots of arguments, let's sort it out... There are basically three modes of
1958 operation, and we will explain them from simplest to most complex:
1960 =over 4
1962 =item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
1964 In this configuration the watcher triggers an event after the wall clock
1965 time C<offset> has passed. It will not repeat and will not adjust when a
1966 time jump occurs, that is, if it is to be run at January 1st 2011 then it
1967 will be stopped and invoked when the system clock reaches or surpasses
1968 this point in time.
1970 =item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
1972 In this mode the watcher will always be scheduled to time out at the next
1973 C<offset + N * interval> time (for some integer N, which can also be
1974 negative) and then repeat, regardless of any time jumps. The C<offset>
1975 argument is merely an offset into the C<interval> periods.
1977 This can be used to create timers that do not drift with respect to the
1978 system clock, for example, here is an C<ev_periodic> that triggers each
1979 hour, on the hour (with respect to UTC):
1981 ev_periodic_set (&periodic, 0., 3600., 0);
1983 This doesn't mean there will always be 3600 seconds in between triggers,
1984 but only that the callback will be called when the system time shows a
1985 full hour (UTC), or more correctly, when the system time is evenly divisible
1986 by 3600.
1988 Another way to think about it (for the mathematically inclined) is that
1989 C<ev_periodic> will try to run the callback in this mode at the next possible
1990 time where C<time = offset (mod interval)>, regardless of any time jumps.
1992 For numerical stability it is preferable that the C<offset> value is near
1993 C<ev_now ()> (the current time), but there is no range requirement for
1994 this value, and in fact is often specified as zero.
1996 Note also that there is an upper limit to how often a timer can fire (CPU
1997 speed for example), so if C<interval> is very small then timing stability
1998 will of course deteriorate. Libev itself tries to be exact to be about one
1999 millisecond (if the OS supports it and the machine is fast enough).
2001 =item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
2003 In this mode the values for C<interval> and C<offset> are both being
2004 ignored. Instead, each time the periodic watcher gets scheduled, the
2005 reschedule callback will be called with the watcher as first, and the
2006 current time as second argument.
2008 NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
2009 or make ANY other event loop modifications whatsoever, unless explicitly
2010 allowed by documentation here>.
2012 If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
2013 it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
2014 only event loop modification you are allowed to do).
2016 The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
2017 *w, ev_tstamp now)>, e.g.:
2019 static ev_tstamp
2020 my_rescheduler (ev_periodic *w, ev_tstamp now)
2021 {
2022 return now + 60.;
2023 }
2025 It must return the next time to trigger, based on the passed time value
2026 (that is, the lowest time value larger than to the second argument). It
2027 will usually be called just before the callback will be triggered, but
2028 might be called at other times, too.
2030 NOTE: I<< This callback must always return a time that is higher than or
2031 equal to the passed C<now> value >>.
2033 This can be used to create very complex timers, such as a timer that
2034 triggers on "next midnight, local time". To do this, you would calculate the
2035 next midnight after C<now> and return the timestamp value for this. How
2036 you do this is, again, up to you (but it is not trivial, which is the main
2037 reason I omitted it as an example).
2039 =back
2041 =item ev_periodic_again (loop, ev_periodic *)
2043 Simply stops and restarts the periodic watcher again. This is only useful
2044 when you changed some parameters or the reschedule callback would return
2045 a different time than the last time it was called (e.g. in a crond like
2046 program when the crontabs have changed).
2048 =item ev_tstamp ev_periodic_at (ev_periodic *)
2050 When active, returns the absolute time that the watcher is supposed
2051 to trigger next. This is not the same as the C<offset> argument to
2052 C<ev_periodic_set>, but indeed works even in interval and manual
2053 rescheduling modes.
2055 =item ev_tstamp offset [read-write]
2057 When repeating, this contains the offset value, otherwise this is the
2058 absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
2059 although libev might modify this value for better numerical stability).
2061 Can be modified any time, but changes only take effect when the periodic
2062 timer fires or C<ev_periodic_again> is being called.
2064 =item ev_tstamp interval [read-write]
2066 The current interval value. Can be modified any time, but changes only
2067 take effect when the periodic timer fires or C<ev_periodic_again> is being
2068 called.
2070 =item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
2072 The current reschedule callback, or C<0>, if this functionality is
2073 switched off. Can be changed any time, but changes only take effect when
2074 the periodic timer fires or C<ev_periodic_again> is being called.
2076 =back
2078 =head3 Examples
2080 Example: Call a callback every hour, or, more precisely, whenever the
2081 system time is divisible by 3600. The callback invocation times have
2082 potentially a lot of jitter, but good long-term stability.
2084 static void
2085 clock_cb (struct ev_loop *loop, ev_io *w, int revents)
2086 {
2087 ... its now a full hour (UTC, or TAI or whatever your clock follows)
2088 }
2090 ev_periodic hourly_tick;
2091 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
2092 ev_periodic_start (loop, &hourly_tick);
2094 Example: The same as above, but use a reschedule callback to do it:
2096 #include <math.h>
2098 static ev_tstamp
2099 my_scheduler_cb (ev_periodic *w, ev_tstamp now)
2100 {
2101 return now + (3600. - fmod (now, 3600.));
2102 }
2104 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
2106 Example: Call a callback every hour, starting now:
2108 ev_periodic hourly_tick;
2109 ev_periodic_init (&hourly_tick, clock_cb,
2110 fmod (ev_now (loop), 3600.), 3600., 0);
2111 ev_periodic_start (loop, &hourly_tick);
2114 =head2 C<ev_signal> - signal me when a signal gets signalled!
2116 Signal watchers will trigger an event when the process receives a specific
2117 signal one or more times. Even though signals are very asynchronous, libev
2118 will try it's best to deliver signals synchronously, i.e. as part of the
2119 normal event processing, like any other event.
2121 If you want signals to be delivered truly asynchronously, just use
2122 C<sigaction> as you would do without libev and forget about sharing
2123 the signal. You can even use C<ev_async> from a signal handler to
2124 synchronously wake up an event loop.
2126 You can configure as many watchers as you like for the same signal, but
2127 only within the same loop, i.e. you can watch for C<SIGINT> in your
2128 default loop and for C<SIGIO> in another loop, but you cannot watch for
2129 C<SIGINT> in both the default loop and another loop at the same time. At
2130 the moment, C<SIGCHLD> is permanently tied to the default loop.
2132 When the first watcher gets started will libev actually register something
2133 with the kernel (thus it coexists with your own signal handlers as long as
2134 you don't register any with libev for the same signal).
2136 If possible and supported, libev will install its handlers with
2137 C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
2138 not be unduly interrupted. If you have a problem with system calls getting
2139 interrupted by signals you can block all signals in an C<ev_check> watcher
2140 and unblock them in an C<ev_prepare> watcher.
2142 =head3 The special problem of inheritance over fork/execve/pthread_create
2144 Both the signal mask (C<sigprocmask>) and the signal disposition
2145 (C<sigaction>) are unspecified after starting a signal watcher (and after
2146 stopping it again), that is, libev might or might not block the signal,
2147 and might or might not set or restore the installed signal handler.
2149 While this does not matter for the signal disposition (libev never
2150 sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
2151 C<execve>), this matters for the signal mask: many programs do not expect
2152 certain signals to be blocked.
2154 This means that before calling C<exec> (from the child) you should reset
2155 the signal mask to whatever "default" you expect (all clear is a good
2156 choice usually).
2158 The simplest way to ensure that the signal mask is reset in the child is
2159 to install a fork handler with C<pthread_atfork> that resets it. That will
2160 catch fork calls done by libraries (such as the libc) as well.
2162 In current versions of libev, the signal will not be blocked indefinitely
2163 unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
2164 the window of opportunity for problems, it will not go away, as libev
2165 I<has> to modify the signal mask, at least temporarily.
2167 So I can't stress this enough: I<If you do not reset your signal mask when
2168 you expect it to be empty, you have a race condition in your code>. This
2169 is not a libev-specific thing, this is true for most event libraries.
2171 =head3 Watcher-Specific Functions and Data Members
2173 =over 4
2175 =item ev_signal_init (ev_signal *, callback, int signum)
2177 =item ev_signal_set (ev_signal *, int signum)
2179 Configures the watcher to trigger on the given signal number (usually one
2180 of the C<SIGxxx> constants).
2182 =item int signum [read-only]
2184 The signal the watcher watches out for.
2186 =back
2188 =head3 Examples
2190 Example: Try to exit cleanly on SIGINT.
2192 static void
2193 sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
2194 {
2195 ev_unloop (loop, EVUNLOOP_ALL);
2196 }
2198 ev_signal signal_watcher;
2199 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
2200 ev_signal_start (loop, &signal_watcher);
2203 =head2 C<ev_child> - watch out for process status changes
2205 Child watchers trigger when your process receives a SIGCHLD in response to
2206 some child status changes (most typically when a child of yours dies or
2207 exits). It is permissible to install a child watcher I<after> the child
2208 has been forked (which implies it might have already exited), as long
2209 as the event loop isn't entered (or is continued from a watcher), i.e.,
2210 forking and then immediately registering a watcher for the child is fine,
2211 but forking and registering a watcher a few event loop iterations later or
2212 in the next callback invocation is not.
2214 Only the default event loop is capable of handling signals, and therefore
2215 you can only register child watchers in the default event loop.
2217 Due to some design glitches inside libev, child watchers will always be
2218 handled at maximum priority (their priority is set to C<EV_MAXPRI> by
2219 libev)
2221 =head3 Process Interaction
2223 Libev grabs C<SIGCHLD> as soon as the default event loop is
2224 initialised. This is necessary to guarantee proper behaviour even if the
2225 first child watcher is started after the child exits. The occurrence
2226 of C<SIGCHLD> is recorded asynchronously, but child reaping is done
2227 synchronously as part of the event loop processing. Libev always reaps all
2228 children, even ones not watched.
2230 =head3 Overriding the Built-In Processing
2232 Libev offers no special support for overriding the built-in child
2233 processing, but if your application collides with libev's default child
2234 handler, you can override it easily by installing your own handler for
2235 C<SIGCHLD> after initialising the default loop, and making sure the
2236 default loop never gets destroyed. You are encouraged, however, to use an
2237 event-based approach to child reaping and thus use libev's support for
2238 that, so other libev users can use C<ev_child> watchers freely.
2240 =head3 Stopping the Child Watcher
2242 Currently, the child watcher never gets stopped, even when the
2243 child terminates, so normally one needs to stop the watcher in the
2244 callback. Future versions of libev might stop the watcher automatically
2245 when a child exit is detected (calling C<ev_child_stop> twice is not a
2246 problem).
2248 =head3 Watcher-Specific Functions and Data Members
2250 =over 4
2252 =item ev_child_init (ev_child *, callback, int pid, int trace)
2254 =item ev_child_set (ev_child *, int pid, int trace)
2256 Configures the watcher to wait for status changes of process C<pid> (or
2257 I<any> process if C<pid> is specified as C<0>). The callback can look
2258 at the C<rstatus> member of the C<ev_child> watcher structure to see
2259 the status word (use the macros from C<sys/wait.h> and see your systems
2260 C<waitpid> documentation). The C<rpid> member contains the pid of the
2261 process causing the status change. C<trace> must be either C<0> (only
2262 activate the watcher when the process terminates) or C<1> (additionally
2263 activate the watcher when the process is stopped or continued).
2265 =item int pid [read-only]
2267 The process id this watcher watches out for, or C<0>, meaning any process id.
2269 =item int rpid [read-write]
2271 The process id that detected a status change.
2273 =item int rstatus [read-write]
2275 The process exit/trace status caused by C<rpid> (see your systems
2276 C<waitpid> and C<sys/wait.h> documentation for details).
2278 =back
2280 =head3 Examples
2282 Example: C<fork()> a new process and install a child handler to wait for
2283 its completion.
2285 ev_child cw;
2287 static void
2288 child_cb (EV_P_ ev_child *w, int revents)
2289 {
2290 ev_child_stop (EV_A_ w);
2291 printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
2292 }
2294 pid_t pid = fork ();
2296 if (pid < 0)
2297 // error
2298 else if (pid == 0)
2299 {
2300 // the forked child executes here
2301 exit (1);
2302 }
2303 else
2304 {
2305 ev_child_init (&cw, child_cb, pid, 0);
2306 ev_child_start (EV_DEFAULT_ &cw);
2307 }
2310 =head2 C<ev_stat> - did the file attributes just change?
2312 This watches a file system path for attribute changes. That is, it calls
2313 C<stat> on that path in regular intervals (or when the OS says it changed)
2314 and sees if it changed compared to the last time, invoking the callback if
2315 it did.
2317 The path does not need to exist: changing from "path exists" to "path does
2318 not exist" is a status change like any other. The condition "path does not
2319 exist" (or more correctly "path cannot be stat'ed") is signified by the
2320 C<st_nlink> field being zero (which is otherwise always forced to be at
2321 least one) and all the other fields of the stat buffer having unspecified
2322 contents.
2324 The path I<must not> end in a slash or contain special components such as
2325 C<.> or C<..>. The path I<should> be absolute: If it is relative and
2326 your working directory changes, then the behaviour is undefined.
2328 Since there is no portable change notification interface available, the
2329 portable implementation simply calls C<stat(2)> regularly on the path
2330 to see if it changed somehow. You can specify a recommended polling
2331 interval for this case. If you specify a polling interval of C<0> (highly
2332 recommended!) then a I<suitable, unspecified default> value will be used
2333 (which you can expect to be around five seconds, although this might
2334 change dynamically). Libev will also impose a minimum interval which is
2335 currently around C<0.1>, but that's usually overkill.
2337 This watcher type is not meant for massive numbers of stat watchers,
2338 as even with OS-supported change notifications, this can be
2339 resource-intensive.
2341 At the time of this writing, the only OS-specific interface implemented
2342 is the Linux inotify interface (implementing kqueue support is left as an
2343 exercise for the reader. Note, however, that the author sees no way of
2344 implementing C<ev_stat> semantics with kqueue, except as a hint).
2346 =head3 ABI Issues (Largefile Support)
2348 Libev by default (unless the user overrides this) uses the default
2349 compilation environment, which means that on systems with large file
2350 support disabled by default, you get the 32 bit version of the stat
2351 structure. When using the library from programs that change the ABI to
2352 use 64 bit file offsets the programs will fail. In that case you have to
2353 compile libev with the same flags to get binary compatibility. This is
2354 obviously the case with any flags that change the ABI, but the problem is
2355 most noticeably displayed with ev_stat and large file support.
2357 The solution for this is to lobby your distribution maker to make large
2358 file interfaces available by default (as e.g. FreeBSD does) and not
2359 optional. Libev cannot simply switch on large file support because it has
2360 to exchange stat structures with application programs compiled using the
2361 default compilation environment.
2363 =head3 Inotify and Kqueue
2365 When C<inotify (7)> support has been compiled into libev and present at
2366 runtime, it will be used to speed up change detection where possible. The
2367 inotify descriptor will be created lazily when the first C<ev_stat>
2368 watcher is being started.
2370 Inotify presence does not change the semantics of C<ev_stat> watchers
2371 except that changes might be detected earlier, and in some cases, to avoid
2372 making regular C<stat> calls. Even in the presence of inotify support
2373 there are many cases where libev has to resort to regular C<stat> polling,
2374 but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
2375 many bugs), the path exists (i.e. stat succeeds), and the path resides on
2376 a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
2377 xfs are fully working) libev usually gets away without polling.
2379 There is no support for kqueue, as apparently it cannot be used to
2380 implement this functionality, due to the requirement of having a file
2381 descriptor open on the object at all times, and detecting renames, unlinks
2382 etc. is difficult.
2384 =head3 C<stat ()> is a synchronous operation
2386 Libev doesn't normally do any kind of I/O itself, and so is not blocking
2387 the process. The exception are C<ev_stat> watchers - those call C<stat
2388 ()>, which is a synchronous operation.
2390 For local paths, this usually doesn't matter: unless the system is very
2391 busy or the intervals between stat's are large, a stat call will be fast,
2392 as the path data is usually in memory already (except when starting the
2393 watcher).
2395 For networked file systems, calling C<stat ()> can block an indefinite
2396 time due to network issues, and even under good conditions, a stat call
2397 often takes multiple milliseconds.
2399 Therefore, it is best to avoid using C<ev_stat> watchers on networked
2400 paths, although this is fully supported by libev.
2402 =head3 The special problem of stat time resolution
2404 The C<stat ()> system call only supports full-second resolution portably,
2405 and even on systems where the resolution is higher, most file systems
2406 still only support whole seconds.
2408 That means that, if the time is the only thing that changes, you can
2409 easily miss updates: on the first update, C<ev_stat> detects a change and
2410 calls your callback, which does something. When there is another update
2411 within the same second, C<ev_stat> will be unable to detect unless the
2412 stat data does change in other ways (e.g. file size).
2414 The solution to this is to delay acting on a change for slightly more
2415 than a second (or till slightly after the next full second boundary), using
2416 a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
2417 ev_timer_again (loop, w)>).
2419 The C<.02> offset is added to work around small timing inconsistencies
2420 of some operating systems (where the second counter of the current time
2421 might be be delayed. One such system is the Linux kernel, where a call to
2422 C<gettimeofday> might return a timestamp with a full second later than
2423 a subsequent C<time> call - if the equivalent of C<time ()> is used to
2424 update file times then there will be a small window where the kernel uses
2425 the previous second to update file times but libev might already execute
2426 the timer callback).
2428 =head3 Watcher-Specific Functions and Data Members
2430 =over 4
2432 =item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
2434 =item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
2436 Configures the watcher to wait for status changes of the given
2437 C<path>. The C<interval> is a hint on how quickly a change is expected to
2438 be detected and should normally be specified as C<0> to let libev choose
2439 a suitable value. The memory pointed to by C<path> must point to the same
2440 path for as long as the watcher is active.
2442 The callback will receive an C<EV_STAT> event when a change was detected,
2443 relative to the attributes at the time the watcher was started (or the
2444 last change was detected).
2446 =item ev_stat_stat (loop, ev_stat *)
2448 Updates the stat buffer immediately with new values. If you change the
2449 watched path in your callback, you could call this function to avoid
2450 detecting this change (while introducing a race condition if you are not
2451 the only one changing the path). Can also be useful simply to find out the
2452 new values.
2454 =item ev_statdata attr [read-only]
2456 The most-recently detected attributes of the file. Although the type is
2457 C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
2458 suitable for your system, but you can only rely on the POSIX-standardised
2459 members to be present. If the C<st_nlink> member is C<0>, then there was
2460 some error while C<stat>ing the file.
2462 =item ev_statdata prev [read-only]
2464 The previous attributes of the file. The callback gets invoked whenever
2465 C<prev> != C<attr>, or, more precisely, one or more of these members
2466 differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
2467 C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
2469 =item ev_tstamp interval [read-only]
2471 The specified interval.
2473 =item const char *path [read-only]
2475 The file system path that is being watched.
2477 =back
2479 =head3 Examples
2481 Example: Watch C</etc/passwd> for attribute changes.
2483 static void
2484 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
2485 {
2486 /* /etc/passwd changed in some way */
2487 if (w->attr.st_nlink)
2488 {
2489 printf ("passwd current size %ld\n", (long)w->attr.st_size);
2490 printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
2491 printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
2492 }
2493 else
2494 /* you shalt not abuse printf for puts */
2495 puts ("wow, /etc/passwd is not there, expect problems. "
2496 "if this is windows, they already arrived\n");
2497 }
2499 ...
2500 ev_stat passwd;
2502 ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
2503 ev_stat_start (loop, &passwd);
2505 Example: Like above, but additionally use a one-second delay so we do not
2506 miss updates (however, frequent updates will delay processing, too, so
2507 one might do the work both on C<ev_stat> callback invocation I<and> on
2508 C<ev_timer> callback invocation).
2510 static ev_stat passwd;
2511 static ev_timer timer;
2513 static void
2514 timer_cb (EV_P_ ev_timer *w, int revents)
2515 {
2516 ev_timer_stop (EV_A_ w);
2518 /* now it's one second after the most recent passwd change */
2519 }
2521 static void
2522 stat_cb (EV_P_ ev_stat *w, int revents)
2523 {
2524 /* reset the one-second timer */
2525 ev_timer_again (EV_A_ &timer);
2526 }
2528 ...
2529 ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
2530 ev_stat_start (loop, &passwd);
2531 ev_timer_init (&timer, timer_cb, 0., 1.02);
2534 =head2 C<ev_idle> - when you've got nothing better to do...
2536 Idle watchers trigger events when no other events of the same or higher
2537 priority are pending (prepare, check and other idle watchers do not count
2538 as receiving "events").
2540 That is, as long as your process is busy handling sockets or timeouts
2541 (or even signals, imagine) of the same or higher priority it will not be
2542 triggered. But when your process is idle (or only lower-priority watchers
2543 are pending), the idle watchers are being called once per event loop
2544 iteration - until stopped, that is, or your process receives more events
2545 and becomes busy again with higher priority stuff.
2547 The most noteworthy effect is that as long as any idle watchers are
2548 active, the process will not block when waiting for new events.
2550 Apart from keeping your process non-blocking (which is a useful
2551 effect on its own sometimes), idle watchers are a good place to do
2552 "pseudo-background processing", or delay processing stuff to after the
2553 event loop has handled all outstanding events.
2555 =head3 Watcher-Specific Functions and Data Members
2557 =over 4
2559 =item ev_idle_init (ev_idle *, callback)
2561 Initialises and configures the idle watcher - it has no parameters of any
2562 kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
2563 believe me.
2565 =back
2567 =head3 Examples
2569 Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
2570 callback, free it. Also, use no error checking, as usual.
2572 static void
2573 idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
2574 {
2575 free (w);
2576 // now do something you wanted to do when the program has
2577 // no longer anything immediate to do.
2578 }
2580 ev_idle *idle_watcher = malloc (sizeof (ev_idle));
2581 ev_idle_init (idle_watcher, idle_cb);
2582 ev_idle_start (loop, idle_watcher);
2585 =head2 C<ev_prepare> and C<ev_check> - customise your event loop!
2587 Prepare and check watchers are usually (but not always) used in pairs:
2588 prepare watchers get invoked before the process blocks and check watchers
2589 afterwards.
2591 You I<must not> call C<ev_loop> or similar functions that enter
2592 the current event loop from either C<ev_prepare> or C<ev_check>
2593 watchers. Other loops than the current one are fine, however. The
2594 rationale behind this is that you do not need to check for recursion in
2595 those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
2596 C<ev_check> so if you have one watcher of each kind they will always be
2597 called in pairs bracketing the blocking call.
2599 Their main purpose is to integrate other event mechanisms into libev and
2600 their use is somewhat advanced. They could be used, for example, to track
2601 variable changes, implement your own watchers, integrate net-snmp or a
2602 coroutine library and lots more. They are also occasionally useful if
2603 you cache some data and want to flush it before blocking (for example,
2604 in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
2605 watcher).
2607 This is done by examining in each prepare call which file descriptors
2608 need to be watched by the other library, registering C<ev_io> watchers
2609 for them and starting an C<ev_timer> watcher for any timeouts (many
2610 libraries provide exactly this functionality). Then, in the check watcher,
2611 you check for any events that occurred (by checking the pending status
2612 of all watchers and stopping them) and call back into the library. The
2613 I/O and timer callbacks will never actually be called (but must be valid
2614 nevertheless, because you never know, you know?).
2616 As another example, the Perl Coro module uses these hooks to integrate
2617 coroutines into libev programs, by yielding to other active coroutines
2618 during each prepare and only letting the process block if no coroutines
2619 are ready to run (it's actually more complicated: it only runs coroutines
2620 with priority higher than or equal to the event loop and one coroutine
2621 of lower priority, but only once, using idle watchers to keep the event
2622 loop from blocking if lower-priority coroutines are active, thus mapping
2623 low-priority coroutines to idle/background tasks).
2625 It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
2626 priority, to ensure that they are being run before any other watchers
2627 after the poll (this doesn't matter for C<ev_prepare> watchers).
2629 Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
2630 activate ("feed") events into libev. While libev fully supports this, they
2631 might get executed before other C<ev_check> watchers did their job. As
2632 C<ev_check> watchers are often used to embed other (non-libev) event
2633 loops those other event loops might be in an unusable state until their
2634 C<ev_check> watcher ran (always remind yourself to coexist peacefully with
2635 others).
2637 =head3 Watcher-Specific Functions and Data Members
2639 =over 4
2641 =item ev_prepare_init (ev_prepare *, callback)
2643 =item ev_check_init (ev_check *, callback)
2645 Initialises and configures the prepare or check watcher - they have no
2646 parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
2647 macros, but using them is utterly, utterly, utterly and completely
2648 pointless.
2650 =back
2652 =head3 Examples
2654 There are a number of principal ways to embed other event loops or modules
2655 into libev. Here are some ideas on how to include libadns into libev
2656 (there is a Perl module named C<EV::ADNS> that does this, which you could
2657 use as a working example. Another Perl module named C<EV::Glib> embeds a
2658 Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
2659 Glib event loop).
2661 Method 1: Add IO watchers and a timeout watcher in a prepare handler,
2662 and in a check watcher, destroy them and call into libadns. What follows
2663 is pseudo-code only of course. This requires you to either use a low
2664 priority for the check watcher or use C<ev_clear_pending> explicitly, as
2665 the callbacks for the IO/timeout watchers might not have been called yet.
2667 static ev_io iow [nfd];
2668 static ev_timer tw;
2670 static void
2671 io_cb (struct ev_loop *loop, ev_io *w, int revents)
2672 {
2673 }
2675 // create io watchers for each fd and a timer before blocking
2676 static void
2677 adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
2678 {
2679 int timeout = 3600000;
2680 struct pollfd fds [nfd];
2681 // actual code will need to loop here and realloc etc.
2682 adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
2684 /* the callback is illegal, but won't be called as we stop during check */
2685 ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
2686 ev_timer_start (loop, &tw);
2688 // create one ev_io per pollfd
2689 for (int i = 0; i < nfd; ++i)
2690 {
2691 ev_io_init (iow + i, io_cb, fds [i].fd,
2692 ((fds [i].events & POLLIN ? EV_READ : 0)
2693 | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
2695 fds [i].revents = 0;
2696 ev_io_start (loop, iow + i);
2697 }
2698 }
2700 // stop all watchers after blocking
2701 static void
2702 adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
2703 {
2704 ev_timer_stop (loop, &tw);
2706 for (int i = 0; i < nfd; ++i)
2707 {
2708 // set the relevant poll flags
2709 // could also call adns_processreadable etc. here
2710 struct pollfd *fd = fds + i;
2711 int revents = ev_clear_pending (iow + i);
2712 if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
2713 if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
2715 // now stop the watcher
2716 ev_io_stop (loop, iow + i);
2717 }
2719 adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
2720 }
2722 Method 2: This would be just like method 1, but you run C<adns_afterpoll>
2723 in the prepare watcher and would dispose of the check watcher.
2725 Method 3: If the module to be embedded supports explicit event
2726 notification (libadns does), you can also make use of the actual watcher
2727 callbacks, and only destroy/create the watchers in the prepare watcher.
2729 static void
2730 timer_cb (EV_P_ ev_timer *w, int revents)
2731 {
2732 adns_state ads = (adns_state)w->data;
2733 update_now (EV_A);
2735 adns_processtimeouts (ads, &tv_now);
2736 }
2738 static void
2739 io_cb (EV_P_ ev_io *w, int revents)
2740 {
2741 adns_state ads = (adns_state)w->data;
2742 update_now (EV_A);
2744 if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
2745 if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
2746 }
2748 // do not ever call adns_afterpoll
2750 Method 4: Do not use a prepare or check watcher because the module you
2751 want to embed is not flexible enough to support it. Instead, you can
2752 override their poll function. The drawback with this solution is that the
2753 main loop is now no longer controllable by EV. The C<Glib::EV> module uses
2754 this approach, effectively embedding EV as a client into the horrible
2755 libglib event loop.
2757 static gint
2758 event_poll_func (GPollFD *fds, guint nfds, gint timeout)
2759 {
2760 int got_events = 0;
2762 for (n = 0; n < nfds; ++n)
2763 // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
2765 if (timeout >= 0)
2766 // create/start timer
2768 // poll
2769 ev_loop (EV_A_ 0);
2771 // stop timer again
2772 if (timeout >= 0)
2773 ev_timer_stop (EV_A_ &to);
2775 // stop io watchers again - their callbacks should have set
2776 for (n = 0; n < nfds; ++n)
2777 ev_io_stop (EV_A_ iow [n]);
2779 return got_events;
2780 }
2783 =head2 C<ev_embed> - when one backend isn't enough...
2785 This is a rather advanced watcher type that lets you embed one event loop
2786 into another (currently only C<ev_io> events are supported in the embedded
2787 loop, other types of watchers might be handled in a delayed or incorrect
2788 fashion and must not be used).
2790 There are primarily two reasons you would want that: work around bugs and
2791 prioritise I/O.
2793 As an example for a bug workaround, the kqueue backend might only support
2794 sockets on some platform, so it is unusable as generic backend, but you
2795 still want to make use of it because you have many sockets and it scales
2796 so nicely. In this case, you would create a kqueue-based loop and embed
2797 it into your default loop (which might use e.g. poll). Overall operation
2798 will be a bit slower because first libev has to call C<poll> and then
2799 C<kevent>, but at least you can use both mechanisms for what they are
2800 best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
2802 As for prioritising I/O: under rare circumstances you have the case where
2803 some fds have to be watched and handled very quickly (with low latency),
2804 and even priorities and idle watchers might have too much overhead. In
2805 this case you would put all the high priority stuff in one loop and all
2806 the rest in a second one, and embed the second one in the first.
2808 As long as the watcher is active, the callback will be invoked every
2809 time there might be events pending in the embedded loop. The callback
2810 must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
2811 sweep and invoke their callbacks (the callback doesn't need to invoke the
2812 C<ev_embed_sweep> function directly, it could also start an idle watcher
2813 to give the embedded loop strictly lower priority for example).
2815 You can also set the callback to C<0>, in which case the embed watcher
2816 will automatically execute the embedded loop sweep whenever necessary.
2818 Fork detection will be handled transparently while the C<ev_embed> watcher
2819 is active, i.e., the embedded loop will automatically be forked when the
2820 embedding loop forks. In other cases, the user is responsible for calling
2821 C<ev_loop_fork> on the embedded loop.
2823 Unfortunately, not all backends are embeddable: only the ones returned by
2824 C<ev_embeddable_backends> are, which, unfortunately, does not include any
2825 portable one.
2827 So when you want to use this feature you will always have to be prepared
2828 that you cannot get an embeddable loop. The recommended way to get around
2829 this is to have a separate variables for your embeddable loop, try to
2830 create it, and if that fails, use the normal loop for everything.
2832 =head3 C<ev_embed> and fork
2834 While the C<ev_embed> watcher is running, forks in the embedding loop will
2835 automatically be applied to the embedded loop as well, so no special
2836 fork handling is required in that case. When the watcher is not running,
2837 however, it is still the task of the libev user to call C<ev_loop_fork ()>
2838 as applicable.
2840 =head3 Watcher-Specific Functions and Data Members
2842 =over 4
2844 =item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
2846 =item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
2848 Configures the watcher to embed the given loop, which must be
2849 embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
2850 invoked automatically, otherwise it is the responsibility of the callback
2851 to invoke it (it will continue to be called until the sweep has been done,
2852 if you do not want that, you need to temporarily stop the embed watcher).
2854 =item ev_embed_sweep (loop, ev_embed *)
2856 Make a single, non-blocking sweep over the embedded loop. This works
2857 similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
2858 appropriate way for embedded loops.
2860 =item struct ev_loop *other [read-only]
2862 The embedded event loop.
2864 =back
2866 =head3 Examples
2868 Example: Try to get an embeddable event loop and embed it into the default
2869 event loop. If that is not possible, use the default loop. The default
2870 loop is stored in C<loop_hi>, while the embeddable loop is stored in
2871 C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
2872 used).
2874 struct ev_loop *loop_hi = ev_default_init (0);
2875 struct ev_loop *loop_lo = 0;
2876 ev_embed embed;
2878 // see if there is a chance of getting one that works
2879 // (remember that a flags value of 0 means autodetection)
2880 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2881 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2882 : 0;
2884 // if we got one, then embed it, otherwise default to loop_hi
2885 if (loop_lo)
2886 {
2887 ev_embed_init (&embed, 0, loop_lo);
2888 ev_embed_start (loop_hi, &embed);
2889 }
2890 else
2891 loop_lo = loop_hi;
2893 Example: Check if kqueue is available but not recommended and create
2894 a kqueue backend for use with sockets (which usually work with any
2895 kqueue implementation). Store the kqueue/socket-only event loop in
2896 C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
2898 struct ev_loop *loop = ev_default_init (0);
2899 struct ev_loop *loop_socket = 0;
2900 ev_embed embed;
2902 if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2903 if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2904 {
2905 ev_embed_init (&embed, 0, loop_socket);
2906 ev_embed_start (loop, &embed);
2907 }
2909 if (!loop_socket)
2910 loop_socket = loop;
2912 // now use loop_socket for all sockets, and loop for everything else
2915 =head2 C<ev_fork> - the audacity to resume the event loop after a fork
2917 Fork watchers are called when a C<fork ()> was detected (usually because
2918 whoever is a good citizen cared to tell libev about it by calling
2919 C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
2920 event loop blocks next and before C<ev_check> watchers are being called,
2921 and only in the child after the fork. If whoever good citizen calling
2922 C<ev_default_fork> cheats and calls it in the wrong process, the fork
2923 handlers will be invoked, too, of course.
2925 =head3 The special problem of life after fork - how is it possible?
2927 Most uses of C<fork()> consist of forking, then some simple calls to ste
2928 up/change the process environment, followed by a call to C<exec()>. This
2929 sequence should be handled by libev without any problems.
2931 This changes when the application actually wants to do event handling
2932 in the child, or both parent in child, in effect "continuing" after the
2933 fork.
2935 The default mode of operation (for libev, with application help to detect
2936 forks) is to duplicate all the state in the child, as would be expected
2937 when I<either> the parent I<or> the child process continues.
2939 When both processes want to continue using libev, then this is usually the
2940 wrong result. In that case, usually one process (typically the parent) is
2941 supposed to continue with all watchers in place as before, while the other
2942 process typically wants to start fresh, i.e. without any active watchers.
2944 The cleanest and most efficient way to achieve that with libev is to
2945 simply create a new event loop, which of course will be "empty", and
2946 use that for new watchers. This has the advantage of not touching more
2947 memory than necessary, and thus avoiding the copy-on-write, and the
2948 disadvantage of having to use multiple event loops (which do not support
2949 signal watchers).
2951 When this is not possible, or you want to use the default loop for
2952 other reasons, then in the process that wants to start "fresh", call
2953 C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying
2954 the default loop will "orphan" (not stop) all registered watchers, so you
2955 have to be careful not to execute code that modifies those watchers. Note
2956 also that in that case, you have to re-register any signal watchers.
2958 =head3 Watcher-Specific Functions and Data Members
2960 =over 4
2962 =item ev_fork_init (ev_signal *, callback)
2964 Initialises and configures the fork watcher - it has no parameters of any
2965 kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
2966 believe me.
2968 =back
2971 =head2 C<ev_async> - how to wake up another event loop
2973 In general, you cannot use an C<ev_loop> from multiple threads or other
2974 asynchronous sources such as signal handlers (as opposed to multiple event
2975 loops - those are of course safe to use in different threads).
2977 Sometimes, however, you need to wake up another event loop you do not
2978 control, for example because it belongs to another thread. This is what
2979 C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
2980 can signal it by calling C<ev_async_send>, which is thread- and signal
2981 safe.
2983 This functionality is very similar to C<ev_signal> watchers, as signals,
2984 too, are asynchronous in nature, and signals, too, will be compressed
2985 (i.e. the number of callback invocations may be less than the number of
2986 C<ev_async_sent> calls).
2988 Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
2989 just the default loop.
2991 =head3 Queueing
2993 C<ev_async> does not support queueing of data in any way. The reason
2994 is that the author does not know of a simple (or any) algorithm for a
2995 multiple-writer-single-reader queue that works in all cases and doesn't
2996 need elaborate support such as pthreads or unportable memory access
2997 semantics.
2999 That means that if you want to queue data, you have to provide your own
3000 queue. But at least I can tell you how to implement locking around your
3001 queue:
3003 =over 4
3005 =item queueing from a signal handler context
3007 To implement race-free queueing, you simply add to the queue in the signal
3008 handler but you block the signal handler in the watcher callback. Here is
3009 an example that does that for some fictitious SIGUSR1 handler:
3011 static ev_async mysig;
3013 static void
3014 sigusr1_handler (void)
3015 {
3016 sometype data;
3018 // no locking etc.
3019 queue_put (data);
3020 ev_async_send (EV_DEFAULT_ &mysig);
3021 }
3023 static void
3024 mysig_cb (EV_P_ ev_async *w, int revents)
3025 {
3026 sometype data;
3027 sigset_t block, prev;
3029 sigemptyset (&block);
3030 sigaddset (&block, SIGUSR1);
3031 sigprocmask (SIG_BLOCK, &block, &prev);
3033 while (queue_get (&data))
3034 process (data);
3036 if (sigismember (&prev, SIGUSR1)
3037 sigprocmask (SIG_UNBLOCK, &block, 0);
3038 }
3040 (Note: pthreads in theory requires you to use C<pthread_setmask>
3041 instead of C<sigprocmask> when you use threads, but libev doesn't do it
3042 either...).
3044 =item queueing from a thread context
3046 The strategy for threads is different, as you cannot (easily) block
3047 threads but you can easily preempt them, so to queue safely you need to
3048 employ a traditional mutex lock, such as in this pthread example:
3050 static ev_async mysig;
3051 static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
3053 static void
3054 otherthread (void)
3055 {
3056 // only need to lock the actual queueing operation
3057 pthread_mutex_lock (&mymutex);
3058 queue_put (data);
3059 pthread_mutex_unlock (&mymutex);
3061 ev_async_send (EV_DEFAULT_ &mysig);
3062 }
3064 static void
3065 mysig_cb (EV_P_ ev_async *w, int revents)
3066 {
3067 pthread_mutex_lock (&mymutex);
3069 while (queue_get (&data))
3070 process (data);
3072 pthread_mutex_unlock (&mymutex);
3073 }
3075 =back
3078 =head3 Watcher-Specific Functions and Data Members
3080 =over 4
3082 =item ev_async_init (ev_async *, callback)
3084 Initialises and configures the async watcher - it has no parameters of any
3085 kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
3086 trust me.
3088 =item ev_async_send (loop, ev_async *)
3090 Sends/signals/activates the given C<ev_async> watcher, that is, feeds
3091 an C<EV_ASYNC> event on the watcher into the event loop. Unlike
3092 C<ev_feed_event>, this call is safe to do from other threads, signal or
3093 similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
3094 section below on what exactly this means).
3096 Note that, as with other watchers in libev, multiple events might get
3097 compressed into a single callback invocation (another way to look at this
3098 is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
3099 reset when the event loop detects that).
3101 This call incurs the overhead of a system call only once per event loop
3102 iteration, so while the overhead might be noticeable, it doesn't apply to
3103 repeated calls to C<ev_async_send> for the same event loop.
3105 =item bool = ev_async_pending (ev_async *)
3107 Returns a non-zero value when C<ev_async_send> has been called on the
3108 watcher but the event has not yet been processed (or even noted) by the
3109 event loop.
3111 C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
3112 the loop iterates next and checks for the watcher to have become active,
3113 it will reset the flag again. C<ev_async_pending> can be used to very
3114 quickly check whether invoking the loop might be a good idea.
3116 Not that this does I<not> check whether the watcher itself is pending,
3117 only whether it has been requested to make this watcher pending: there
3118 is a time window between the event loop checking and resetting the async
3119 notification, and the callback being invoked.
3121 =back
3126 There are some other functions of possible interest. Described. Here. Now.
3128 =over 4
3130 =item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
3132 This function combines a simple timer and an I/O watcher, calls your
3133 callback on whichever event happens first and automatically stops both
3134 watchers. This is useful if you want to wait for a single event on an fd
3135 or timeout without having to allocate/configure/start/stop/free one or
3136 more watchers yourself.
3138 If C<fd> is less than 0, then no I/O watcher will be started and the
3139 C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
3140 the given C<fd> and C<events> set will be created and started.
3142 If C<timeout> is less than 0, then no timeout watcher will be
3143 started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
3144 repeat = 0) will be started. C<0> is a valid timeout.
3146 The callback has the type C<void (*cb)(int revents, void *arg)> and gets
3147 passed an C<revents> set like normal event callbacks (a combination of
3148 C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
3149 value passed to C<ev_once>. Note that it is possible to receive I<both>
3150 a timeout and an io event at the same time - you probably should give io
3151 events precedence.
3153 Example: wait up to ten seconds for data to appear on STDIN_FILENO.
3155 static void stdin_ready (int revents, void *arg)
3156 {
3157 if (revents & EV_READ)
3158 /* stdin might have data for us, joy! */;
3159 else if (revents & EV_TIMEOUT)
3160 /* doh, nothing entered */;
3161 }
3163 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
3165 =item ev_feed_fd_event (loop, int fd, int revents)
3167 Feed an event on the given fd, as if a file descriptor backend detected
3168 the given events it.
3170 =item ev_feed_signal_event (loop, int signum)
3172 Feed an event as if the given signal occurred (C<loop> must be the default
3173 loop!).
3175 =back
3180 Libev offers a compatibility emulation layer for libevent. It cannot
3181 emulate the internals of libevent, so here are some usage hints:
3183 =over 4
3185 =item * Use it by including <event.h>, as usual.
3187 =item * The following members are fully supported: ev_base, ev_callback,
3188 ev_arg, ev_fd, ev_res, ev_events.
3190 =item * Avoid using ev_flags and the EVLIST_*-macros, while it is
3191 maintained by libev, it does not work exactly the same way as in libevent (consider
3192 it a private API).
3194 =item * Priorities are not currently supported. Initialising priorities
3195 will fail and all watchers will have the same priority, even though there
3196 is an ev_pri field.
3198 =item * In libevent, the last base created gets the signals, in libev, the
3199 first base created (== the default loop) gets the signals.
3201 =item * Other members are not supported.
3203 =item * The libev emulation is I<not> ABI compatible to libevent, you need
3204 to use the libev header file and library.
3206 =back
3208 =head1 C++ SUPPORT
3210 Libev comes with some simplistic wrapper classes for C++ that mainly allow
3211 you to use some convenience methods to start/stop watchers and also change
3212 the callback model to a model using method callbacks on objects.
3214 To use it,
3216 #include <ev++.h>
3218 This automatically includes F<ev.h> and puts all of its definitions (many
3219 of them macros) into the global namespace. All C++ specific things are
3220 put into the C<ev> namespace. It should support all the same embedding
3221 options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
3223 Care has been taken to keep the overhead low. The only data member the C++
3224 classes add (compared to plain C-style watchers) is the event loop pointer
3225 that the watcher is associated with (or no additional members at all if
3226 you disable C<EV_MULTIPLICITY> when embedding libev).
3228 Currently, functions, and static and non-static member functions can be
3229 used as callbacks. Other types should be easy to add as long as they only
3230 need one additional pointer for context. If you need support for other
3231 types of functors please contact the author (preferably after implementing
3232 it).
3234 Here is a list of things available in the C<ev> namespace:
3236 =over 4
3238 =item C<ev::READ>, C<ev::WRITE> etc.
3240 These are just enum values with the same values as the C<EV_READ> etc.
3241 macros from F<ev.h>.
3243 =item C<ev::tstamp>, C<ev::now>
3245 Aliases to the same types/functions as with the C<ev_> prefix.
3247 =item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
3249 For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
3250 the same name in the C<ev> namespace, with the exception of C<ev_signal>
3251 which is called C<ev::sig> to avoid clashes with the C<signal> macro
3252 defines by many implementations.
3254 All of those classes have these methods:
3256 =over 4
3258 =item ev::TYPE::TYPE ()
3260 =item ev::TYPE::TYPE (loop)
3262 =item ev::TYPE::~TYPE
3264 The constructor (optionally) takes an event loop to associate the watcher
3265 with. If it is omitted, it will use C<EV_DEFAULT>.
3267 The constructor calls C<ev_init> for you, which means you have to call the
3268 C<set> method before starting it.
3270 It will not set a callback, however: You have to call the templated C<set>
3271 method to set a callback before you can start the watcher.
3273 (The reason why you have to use a method is a limitation in C++ which does
3274 not allow explicit template arguments for constructors).
3276 The destructor automatically stops the watcher if it is active.
3278 =item w->set<class, &class::method> (object *)
3280 This method sets the callback method to call. The method has to have a
3281 signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
3282 first argument and the C<revents> as second. The object must be given as
3283 parameter and is stored in the C<data> member of the watcher.
3285 This method synthesizes efficient thunking code to call your method from
3286 the C callback that libev requires. If your compiler can inline your
3287 callback (i.e. it is visible to it at the place of the C<set> call and
3288 your compiler is good :), then the method will be fully inlined into the
3289 thunking function, making it as fast as a direct C callback.
3291 Example: simple class declaration and watcher initialisation
3293 struct myclass
3294 {
3295 void io_cb (ev::io &w, int revents) { }
3296 }
3298 myclass obj;
3299 ev::io iow;
3300 iow.set <myclass, &myclass::io_cb> (&obj);
3302 =item w->set (object *)
3304 This is an B<experimental> feature that might go away in a future version.
3306 This is a variation of a method callback - leaving out the method to call
3307 will default the method to C<operator ()>, which makes it possible to use
3308 functor objects without having to manually specify the C<operator ()> all
3309 the time. Incidentally, you can then also leave out the template argument
3310 list.
3312 The C<operator ()> method prototype must be C<void operator ()(watcher &w,
3313 int revents)>.
3315 See the method-C<set> above for more details.
3317 Example: use a functor object as callback.
3319 struct myfunctor
3320 {
3321 void operator() (ev::io &w, int revents)
3322 {
3323 ...
3324 }
3325 }
3327 myfunctor f;
3329 ev::io w;
3330 w.set (&f);
3332 =item w->set<function> (void *data = 0)
3334 Also sets a callback, but uses a static method or plain function as
3335 callback. The optional C<data> argument will be stored in the watcher's
3336 C<data> member and is free for you to use.
3338 The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
3340 See the method-C<set> above for more details.
3342 Example: Use a plain function as callback.
3344 static void io_cb (ev::io &w, int revents) { }
3345 iow.set <io_cb> ();
3347 =item w->set (loop)
3349 Associates a different C<struct ev_loop> with this watcher. You can only
3350 do this when the watcher is inactive (and not pending either).
3352 =item w->set ([arguments])
3354 Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
3355 called at least once. Unlike the C counterpart, an active watcher gets
3356 automatically stopped and restarted when reconfiguring it with this
3357 method.
3359 =item w->start ()
3361 Starts the watcher. Note that there is no C<loop> argument, as the
3362 constructor already stores the event loop.
3364 =item w->stop ()
3366 Stops the watcher if it is active. Again, no C<loop> argument.
3368 =item w->again () (C<ev::timer>, C<ev::periodic> only)
3370 For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
3371 C<ev_TYPE_again> function.
3373 =item w->sweep () (C<ev::embed> only)
3375 Invokes C<ev_embed_sweep>.
3377 =item w->update () (C<ev::stat> only)
3379 Invokes C<ev_stat_stat>.
3381 =back
3383 =back
3385 Example: Define a class with an IO and idle watcher, start one of them in
3386 the constructor.
3388 class myclass
3389 {
3390 ev::io io ; void io_cb (ev::io &w, int revents);
3391 ev::idle idle; void idle_cb (ev::idle &w, int revents);
3393 myclass (int fd)
3394 {
3395 io .set <myclass, &myclass::io_cb > (this);
3396 idle.set <myclass, &myclass::idle_cb> (this);
3398 io.start (fd, ev::READ);
3399 }
3400 };
3405 Libev does not offer other language bindings itself, but bindings for a
3406 number of languages exist in the form of third-party packages. If you know
3407 any interesting language binding in addition to the ones listed here, drop
3408 me a note.
3410 =over 4
3412 =item Perl
3414 The EV module implements the full libev API and is actually used to test
3415 libev. EV is developed together with libev. Apart from the EV core module,
3416 there are additional modules that implement libev-compatible interfaces
3417 to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
3418 C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
3419 and C<EV::Glib>).
3421 It can be found and installed via CPAN, its homepage is at
3422 L<>.
3424 =item Python
3426 Python bindings can be found at L<>. It
3427 seems to be quite complete and well-documented.
3429 =item Ruby
3431 Tony Arcieri has written a ruby extension that offers access to a subset
3432 of the libev API and adds file handle abstractions, asynchronous DNS and
3433 more on top of it. It can be found via gem servers. Its homepage is at
3434 L<>.
3436 Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
3437 makes rev work even on mingw.
3439 =item Haskell
3441 A haskell binding to libev is available at
3442 L<>.
3444 =item D
3446 Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
3447 be found at L<>.
3449 =item Ocaml
3451 Erkki Seppala has written Ocaml bindings for libev, to be found at
3452 L<>.
3454 =item Lua
3456 Brian Maher has written a partial interface to libev
3457 for lua (only C<ev_io> and C<ev_timer>), to be found at
3458 L<>.
3460 =back
3463 =head1 MACRO MAGIC
3465 Libev can be compiled with a variety of options, the most fundamental
3466 of which is C<EV_MULTIPLICITY>. This option determines whether (most)
3467 functions and callbacks have an initial C<struct ev_loop *> argument.
3469 To make it easier to write programs that cope with either variant, the
3470 following macros are defined:
3472 =over 4
3474 =item C<EV_A>, C<EV_A_>
3476 This provides the loop I<argument> for functions, if one is required ("ev
3477 loop argument"). The C<EV_A> form is used when this is the sole argument,
3478 C<EV_A_> is used when other arguments are following. Example:
3480 ev_unref (EV_A);
3481 ev_timer_add (EV_A_ watcher);
3482 ev_loop (EV_A_ 0);
3484 It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
3485 which is often provided by the following macro.
3487 =item C<EV_P>, C<EV_P_>
3489 This provides the loop I<parameter> for functions, if one is required ("ev
3490 loop parameter"). The C<EV_P> form is used when this is the sole parameter,
3491 C<EV_P_> is used when other parameters are following. Example:
3493 // this is how ev_unref is being declared
3494 static void ev_unref (EV_P);
3496 // this is how you can declare your typical callback
3497 static void cb (EV_P_ ev_timer *w, int revents)
3499 It declares a parameter C<loop> of type C<struct ev_loop *>, quite
3500 suitable for use with C<EV_A>.
3502 =item C<EV_DEFAULT>, C<EV_DEFAULT_>
3504 Similar to the other two macros, this gives you the value of the default
3505 loop, if multiple loops are supported ("ev loop default").
3509 Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
3510 default loop has been initialised (C<UC> == unchecked). Their behaviour
3511 is undefined when the default loop has not been initialised by a previous
3512 execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
3514 It is often prudent to use C<EV_DEFAULT> when initialising the first
3515 watcher in a function but use C<EV_DEFAULT_UC> afterwards.
3517 =back
3519 Example: Declare and initialise a check watcher, utilising the above
3520 macros so it will work regardless of whether multiple loops are supported
3521 or not.
3523 static void
3524 check_cb (EV_P_ ev_timer *w, int revents)
3525 {
3526 ev_check_stop (EV_A_ w);
3527 }
3529 ev_check check;
3530 ev_check_init (&check, check_cb);
3531 ev_check_start (EV_DEFAULT_ &check);
3532 ev_loop (EV_DEFAULT_ 0);
3534 =head1 EMBEDDING
3536 Libev can (and often is) directly embedded into host
3537 applications. Examples of applications that embed it include the Deliantra
3538 Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
3539 and rxvt-unicode.
3541 The goal is to enable you to just copy the necessary files into your
3542 source directory without having to change even a single line in them, so
3543 you can easily upgrade by simply copying (or having a checked-out copy of
3544 libev somewhere in your source tree).
3546 =head2 FILESETS
3548 Depending on what features you need you need to include one or more sets of files
3549 in your application.
3551 =head3 CORE EVENT LOOP
3553 To include only the libev core (all the C<ev_*> functions), with manual
3554 configuration (no autoconf):
3556 #define EV_STANDALONE 1
3557 #include "ev.c"
3559 This will automatically include F<ev.h>, too, and should be done in a
3560 single C source file only to provide the function implementations. To use
3561 it, do the same for F<ev.h> in all files wishing to use this API (best
3562 done by writing a wrapper around F<ev.h> that you can include instead and
3563 where you can put other configuration options):
3565 #define EV_STANDALONE 1
3566 #include "ev.h"
3568 Both header files and implementation files can be compiled with a C++
3569 compiler (at least, that's a stated goal, and breakage will be treated
3570 as a bug).
3572 You need the following files in your source tree, or in a directory
3573 in your include path (e.g. in libev/ when using -Ilibev):
3575 ev.h
3576 ev.c
3577 ev_vars.h
3578 ev_wrap.h
3580 ev_win32.c required on win32 platforms only
3582 ev_select.c only when select backend is enabled (which is enabled by default)
3583 ev_poll.c only when poll backend is enabled (disabled by default)
3584 ev_epoll.c only when the epoll backend is enabled (disabled by default)
3585 ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
3586 ev_port.c only when the solaris port backend is enabled (disabled by default)
3588 F<ev.c> includes the backend files directly when enabled, so you only need
3589 to compile this single file.
3593 To include the libevent compatibility API, also include:
3595 #include "event.c"
3597 in the file including F<ev.c>, and:
3599 #include "event.h"
3601 in the files that want to use the libevent API. This also includes F<ev.h>.
3603 You need the following additional files for this:
3605 event.h
3606 event.c
3610 Instead of using C<EV_STANDALONE=1> and providing your configuration in
3611 whatever way you want, you can also C<m4_include([libev.m4])> in your
3612 F<> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
3613 include F<config.h> and configure itself accordingly.
3615 For this of course you need the m4 file:
3617 libev.m4
3621 Libev can be configured via a variety of preprocessor symbols you have to
3622 define before including any of its files. The default in the absence of
3623 autoconf is documented for every option.
3625 =over 4
3627 =item EV_STANDALONE
3629 Must always be C<1> if you do not use autoconf configuration, which
3630 keeps libev from including F<config.h>, and it also defines dummy
3631 implementations for some libevent functions (such as logging, which is not
3632 supported). It will also not define any of the structs usually found in
3633 F<event.h> that are not directly supported by the libev core alone.
3635 In standalone mode, libev will still try to automatically deduce the
3636 configuration, but has to be more conservative.
3640 If defined to be C<1>, libev will try to detect the availability of the
3641 monotonic clock option at both compile time and runtime. Otherwise no
3642 use of the monotonic clock option will be attempted. If you enable this,
3643 you usually have to link against librt or something similar. Enabling it
3644 when the functionality isn't available is safe, though, although you have
3645 to make sure you link against any libraries where the C<clock_gettime>
3646 function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
3648 =item EV_USE_REALTIME
3650 If defined to be C<1>, libev will try to detect the availability of the
3651 real-time clock option at compile time (and assume its availability
3652 at runtime if successful). Otherwise no use of the real-time clock
3653 option will be attempted. This effectively replaces C<gettimeofday>
3654 by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
3655 correctness. See the note about libraries in the description of
3656 C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
3661 If defined to be C<1>, libev will try to use a direct syscall instead
3662 of calling the system-provided C<clock_gettime> function. This option
3663 exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
3664 unconditionally pulls in C<libpthread>, slowing down single-threaded
3665 programs needlessly. Using a direct syscall is slightly slower (in
3666 theory), because no optimised vdso implementation can be used, but avoids
3667 the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
3668 higher, as it simplifies linking (no need for C<-lrt>).
3672 If defined to be C<1>, libev will assume that C<nanosleep ()> is available
3673 and will use it for delays. Otherwise it will use C<select ()>.
3675 =item EV_USE_EVENTFD
3677 If defined to be C<1>, then libev will assume that C<eventfd ()> is
3678 available and will probe for kernel support at runtime. This will improve
3679 C<ev_signal> and C<ev_async> performance and reduce resource consumption.
3680 If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
3681 2.7 or newer, otherwise disabled.
3683 =item EV_USE_SELECT
3685 If undefined or defined to be C<1>, libev will compile in support for the
3686 C<select>(2) backend. No attempt at auto-detection will be done: if no
3687 other method takes over, select will be it. Otherwise the select backend
3688 will not be compiled in.
3692 If defined to C<1>, then the select backend will use the system C<fd_set>
3693 structure. This is useful if libev doesn't compile due to a missing
3694 C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
3695 on exotic systems. This usually limits the range of file descriptors to
3696 some low limit such as 1024 or might have other limitations (winsocket
3697 only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
3698 configures the maximum size of the C<fd_set>.
3702 When defined to C<1>, the select backend will assume that
3703 select/socket/connect etc. don't understand file descriptors but
3704 wants osf handles on win32 (this is the case when the select to
3705 be used is the winsock select). This means that it will call
3706 C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
3707 it is assumed that all these functions actually work on fds, even
3708 on win32. Should not be defined on non-win32 platforms.
3710 =item EV_FD_TO_WIN32_HANDLE(fd)
3712 If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
3713 file descriptors to socket handles. When not defining this symbol (the
3714 default), then libev will call C<_get_osfhandle>, which is usually
3715 correct. In some cases, programs use their own file descriptor management,
3716 in which case they can provide this function to map fds to socket handles.
3718 =item EV_WIN32_HANDLE_TO_FD(handle)
3720 If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
3721 using the standard C<_open_osfhandle> function. For programs implementing
3722 their own fd to handle mapping, overwriting this function makes it easier
3723 to do so. This can be done by defining this macro to an appropriate value.
3725 =item EV_WIN32_CLOSE_FD(fd)
3727 If programs implement their own fd to handle mapping on win32, then this
3728 macro can be used to override the C<close> function, useful to unregister
3729 file descriptors again. Note that the replacement function has to close
3730 the underlying OS handle.
3732 =item EV_USE_POLL
3734 If defined to be C<1>, libev will compile in support for the C<poll>(2)
3735 backend. Otherwise it will be enabled on non-win32 platforms. It
3736 takes precedence over select.
3738 =item EV_USE_EPOLL
3740 If defined to be C<1>, libev will compile in support for the Linux
3741 C<epoll>(7) backend. Its availability will be detected at runtime,
3742 otherwise another method will be used as fallback. This is the preferred
3743 backend for GNU/Linux systems. If undefined, it will be enabled if the
3744 headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
3746 =item EV_USE_KQUEUE
3748 If defined to be C<1>, libev will compile in support for the BSD style
3749 C<kqueue>(2) backend. Its actual availability will be detected at runtime,
3750 otherwise another method will be used as fallback. This is the preferred
3751 backend for BSD and BSD-like systems, although on most BSDs kqueue only
3752 supports some types of fds correctly (the only platform we found that
3753 supports ptys for example was NetBSD), so kqueue might be compiled in, but
3754 not be used unless explicitly requested. The best way to use it is to find
3755 out whether kqueue supports your type of fd properly and use an embedded
3756 kqueue loop.
3758 =item EV_USE_PORT
3760 If defined to be C<1>, libev will compile in support for the Solaris
3761 10 port style backend. Its availability will be detected at runtime,
3762 otherwise another method will be used as fallback. This is the preferred
3763 backend for Solaris 10 systems.
3765 =item EV_USE_DEVPOLL
3767 Reserved for future expansion, works like the USE symbols above.
3769 =item EV_USE_INOTIFY
3771 If defined to be C<1>, libev will compile in support for the Linux inotify
3772 interface to speed up C<ev_stat> watchers. Its actual availability will
3773 be detected at runtime. If undefined, it will be enabled if the headers
3774 indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
3776 =item EV_ATOMIC_T
3778 Libev requires an integer type (suitable for storing C<0> or C<1>) whose
3779 access is atomic with respect to other threads or signal contexts. No such
3780 type is easily found in the C language, so you can provide your own type
3781 that you know is safe for your purposes. It is used both for signal handler "locking"
3782 as well as for signal and thread safety in C<ev_async> watchers.
3784 In the absence of this define, libev will use C<sig_atomic_t volatile>
3785 (from F<signal.h>), which is usually good enough on most platforms.
3787 =item EV_H
3789 The name of the F<ev.h> header file used to include it. The default if
3790 undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
3791 used to virtually rename the F<ev.h> header file in case of conflicts.
3793 =item EV_CONFIG_H
3795 If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
3796 F<ev.c>'s idea of where to find the F<config.h> file, similarly to
3797 C<EV_H>, above.
3799 =item EV_EVENT_H
3801 Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
3802 of how the F<event.h> header can be found, the default is C<"event.h">.
3804 =item EV_PROTOTYPES
3806 If defined to be C<0>, then F<ev.h> will not define any function
3807 prototypes, but still define all the structs and other symbols. This is
3808 occasionally useful if you want to provide your own wrapper functions
3809 around libev functions.
3813 If undefined or defined to C<1>, then all event-loop-specific functions
3814 will have the C<struct ev_loop *> as first argument, and you can create
3815 additional independent event loops. Otherwise there will be no support
3816 for multiple event loops and there is no first event loop pointer
3817 argument. Instead, all functions act on the single default loop.
3819 =item EV_MINPRI
3821 =item EV_MAXPRI
3823 The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
3824 C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
3825 provide for more priorities by overriding those symbols (usually defined
3826 to be C<-2> and C<2>, respectively).
3828 When doing priority-based operations, libev usually has to linearly search
3829 all the priorities, so having many of them (hundreds) uses a lot of space
3830 and time, so using the defaults of five priorities (-2 .. +2) is usually
3831 fine.
3833 If your embedding application does not need any priorities, defining these
3834 both to C<0> will save some memory and CPU.
3838 If undefined or defined to be C<1>, then periodic timers are supported. If
3839 defined to be C<0>, then they are not. Disabling them saves a few kB of
3840 code.
3842 =item EV_IDLE_ENABLE
3844 If undefined or defined to be C<1>, then idle watchers are supported. If
3845 defined to be C<0>, then they are not. Disabling them saves a few kB of
3846 code.
3848 =item EV_EMBED_ENABLE
3850 If undefined or defined to be C<1>, then embed watchers are supported. If
3851 defined to be C<0>, then they are not. Embed watchers rely on most other
3852 watcher types, which therefore must not be disabled.
3854 =item EV_STAT_ENABLE
3856 If undefined or defined to be C<1>, then stat watchers are supported. If
3857 defined to be C<0>, then they are not.
3859 =item EV_FORK_ENABLE
3861 If undefined or defined to be C<1>, then fork watchers are supported. If
3862 defined to be C<0>, then they are not.
3864 =item EV_ASYNC_ENABLE
3866 If undefined or defined to be C<1>, then async watchers are supported. If
3867 defined to be C<0>, then they are not.
3869 =item EV_MINIMAL
3871 If you need to shave off some kilobytes of code at the expense of some
3872 speed (but with the full API), define this symbol to C<1>. Currently this
3873 is used to override some inlining decisions, saves roughly 30% code size
3874 on amd64. It also selects a much smaller 2-heap for timer management over
3875 the default 4-heap.
3877 You can save even more by disabling watcher types you do not need
3878 and setting C<EV_MAXPRI> == C<EV_MINPRI>. Also, disabling C<assert>
3879 (C<-DNDEBUG>) will usually reduce code size a lot.
3881 Defining C<EV_MINIMAL> to C<2> will additionally reduce the core API to
3882 provide a bare-bones event library. See C<ev.h> for details on what parts
3883 of the API are still available, and do not complain if this subset changes
3884 over time.
3886 =item EV_NSIG
3888 The highest supported signal number, +1 (or, the number of
3889 signals): Normally, libev tries to deduce the maximum number of signals
3890 automatically, but sometimes this fails, in which case it can be
3891 specified. Also, using a lower number than detected (C<32> should be
3892 good for about any system in existance) can save some memory, as libev
3893 statically allocates some 12-24 bytes per signal number.
3895 =item EV_PID_HASHSIZE
3897 C<ev_child> watchers use a small hash table to distribute workload by
3898 pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
3899 than enough. If you need to manage thousands of children you might want to
3900 increase this value (I<must> be a power of two).
3904 C<ev_stat> watchers use a small hash table to distribute workload by
3905 inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
3906 usually more than enough. If you need to manage thousands of C<ev_stat>
3907 watchers you might want to increase this value (I<must> be a power of
3908 two).
3910 =item EV_USE_4HEAP
3912 Heaps are not very cache-efficient. To improve the cache-efficiency of the
3913 timer and periodics heaps, libev uses a 4-heap when this symbol is defined
3914 to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
3915 faster performance with many (thousands) of watchers.
3917 The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
3918 (disabled).
3920 =item EV_HEAP_CACHE_AT
3922 Heaps are not very cache-efficient. To improve the cache-efficiency of the
3923 timer and periodics heaps, libev can cache the timestamp (I<at>) within
3924 the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
3925 which uses 8-12 bytes more per watcher and a few hundred bytes more code,
3926 but avoids random read accesses on heap changes. This improves performance
3927 noticeably with many (hundreds) of watchers.
3929 The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
3930 (disabled).
3932 =item EV_VERIFY
3934 Controls how much internal verification (see C<ev_loop_verify ()>) will
3935 be done: If set to C<0>, no internal verification code will be compiled
3936 in. If set to C<1>, then verification code will be compiled in, but not
3937 called. If set to C<2>, then the internal verification code will be
3938 called once per loop, which can slow down libev. If set to C<3>, then the
3939 verification code will be called very frequently, which will slow down
3940 libev considerably.
3942 The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
3943 C<0>.
3945 =item EV_COMMON
3947 By default, all watchers have a C<void *data> member. By redefining
3948 this macro to a something else you can include more and other types of
3949 members. You have to define it each time you include one of the files,
3950 though, and it must be identical each time.
3952 For example, the perl EV module uses something like this:
3954 #define EV_COMMON \
3955 SV *self; /* contains this struct */ \
3956 SV *cb_sv, *fh /* note no trailing ";" */
3958 =item EV_CB_DECLARE (type)
3960 =item EV_CB_INVOKE (watcher, revents)
3962 =item ev_set_cb (ev, cb)
3964 Can be used to change the callback member declaration in each watcher,
3965 and the way callbacks are invoked and set. Must expand to a struct member
3966 definition and a statement, respectively. See the F<ev.h> header file for
3967 their default definitions. One possible use for overriding these is to
3968 avoid the C<struct ev_loop *> as first argument in all cases, or to use
3969 method calls instead of plain function calls in C++.
3971 =back
3975 If you need to re-export the API (e.g. via a DLL) and you need a list of
3976 exported symbols, you can use the provided F<Symbol.*> files which list
3977 all public symbols, one per line:
3979 Symbols.ev for libev proper
3980 Symbols.event for the libevent emulation
3982 This can also be used to rename all public symbols to avoid clashes with
3983 multiple versions of libev linked together (which is obviously bad in
3984 itself, but sometimes it is inconvenient to avoid this).
3986 A sed command like this will create wrapper C<#define>'s that you need to
3987 include before including F<ev.h>:
3989 <Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
3991 This would create a file F<wrap.h> which essentially looks like this:
3993 #define ev_backend myprefix_ev_backend
3994 #define ev_check_start myprefix_ev_check_start
3995 #define ev_check_stop myprefix_ev_check_stop
3996 ...
3998 =head2 EXAMPLES
4000 For a real-world example of a program the includes libev
4001 verbatim, you can have a look at the EV perl module
4002 (L<>). It has the libev files in
4003 the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
4004 interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
4005 will be compiled. It is pretty complex because it provides its own header
4006 file.
4008 The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
4009 that everybody includes and which overrides some configure choices:
4011 #define EV_MINIMAL 1
4012 #define EV_USE_POLL 0
4013 #define EV_MULTIPLICITY 0
4014 #define EV_PERIODIC_ENABLE 0
4015 #define EV_STAT_ENABLE 0
4016 #define EV_FORK_ENABLE 0
4017 #define EV_CONFIG_H <config.h>
4018 #define EV_MINPRI 0
4019 #define EV_MAXPRI 0
4021 #include "ev++.h"
4023 And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
4025 #include "ev_cpp.h"
4026 #include "ev.c"
4032 =head3 THREADS
4034 All libev functions are reentrant and thread-safe unless explicitly
4035 documented otherwise, but libev implements no locking itself. This means
4036 that you can use as many loops as you want in parallel, as long as there
4037 are no concurrent calls into any libev function with the same loop
4038 parameter (C<ev_default_*> calls have an implicit default loop parameter,
4039 of course): libev guarantees that different event loops share no data
4040 structures that need any locking.
4042 Or to put it differently: calls with different loop parameters can be done
4043 concurrently from multiple threads, calls with the same loop parameter
4044 must be done serially (but can be done from different threads, as long as
4045 only one thread ever is inside a call at any point in time, e.g. by using
4046 a mutex per loop).
4048 Specifically to support threads (and signal handlers), libev implements
4049 so-called C<ev_async> watchers, which allow some limited form of
4050 concurrency on the same event loop, namely waking it up "from the
4051 outside".
4053 If you want to know which design (one loop, locking, or multiple loops
4054 without or something else still) is best for your problem, then I cannot
4055 help you, but here is some generic advice:
4057 =over 4
4059 =item * most applications have a main thread: use the default libev loop
4060 in that thread, or create a separate thread running only the default loop.
4062 This helps integrating other libraries or software modules that use libev
4063 themselves and don't care/know about threading.
4065 =item * one loop per thread is usually a good model.
4067 Doing this is almost never wrong, sometimes a better-performance model
4068 exists, but it is always a good start.
4070 =item * other models exist, such as the leader/follower pattern, where one
4071 loop is handed through multiple threads in a kind of round-robin fashion.
4073 Choosing a model is hard - look around, learn, know that usually you can do
4074 better than you currently do :-)
4076 =item * often you need to talk to some other thread which blocks in the
4077 event loop.
4079 C<ev_async> watchers can be used to wake them up from other threads safely
4080 (or from signal contexts...).
4082 An example use would be to communicate signals or other events that only
4083 work in the default loop by registering the signal watcher with the
4084 default loop and triggering an C<ev_async> watcher from the default loop
4085 watcher callback into the event loop interested in the signal.
4087 =back
4091 Here is a fictitious example of how to run an event loop in a different
4092 thread than where callbacks are being invoked and watchers are
4093 created/added/removed.
4095 For a real-world example, see the C<EV::Loop::Async> perl module,
4096 which uses exactly this technique (which is suited for many high-level
4097 languages).
4099 The example uses a pthread mutex to protect the loop data, a condition
4100 variable to wait for callback invocations, an async watcher to notify the
4101 event loop thread and an unspecified mechanism to wake up the main thread.
4103 First, you need to associate some data with the event loop:
4105 typedef struct {
4106 mutex_t lock; /* global loop lock */
4107 ev_async async_w;
4108 thread_t tid;
4109 cond_t invoke_cv;
4110 } userdata;
4112 void prepare_loop (EV_P)
4113 {
4114 // for simplicity, we use a static userdata struct.
4115 static userdata u;
4117 ev_async_init (&u->async_w, async_cb);
4118 ev_async_start (EV_A_ &u->async_w);
4120 pthread_mutex_init (&u->lock, 0);
4121 pthread_cond_init (&u->invoke_cv, 0);
4123 // now associate this with the loop
4124 ev_set_userdata (EV_A_ u);
4125 ev_set_invoke_pending_cb (EV_A_ l_invoke);
4126 ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
4128 // then create the thread running ev_loop
4129 pthread_create (&u->tid, 0, l_run, EV_A);
4130 }
4132 The callback for the C<ev_async> watcher does nothing: the watcher is used
4133 solely to wake up the event loop so it takes notice of any new watchers
4134 that might have been added:
4136 static void
4137 async_cb (EV_P_ ev_async *w, int revents)
4138 {
4139 // just used for the side effects
4140 }
4142 The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
4143 protecting the loop data, respectively.
4145 static void
4146 l_release (EV_P)
4147 {
4148 userdata *u = ev_userdata (EV_A);
4149 pthread_mutex_unlock (&u->lock);
4150 }
4152 static void
4153 l_acquire (EV_P)
4154 {
4155 userdata *u = ev_userdata (EV_A);
4156 pthread_mutex_lock (&u->lock);
4157 }
4159 The event loop thread first acquires the mutex, and then jumps straight
4160 into C<ev_loop>:
4162 void *
4163 l_run (void *thr_arg)
4164 {
4165 struct ev_loop *loop = (struct ev_loop *)thr_arg;
4167 l_acquire (EV_A);
4168 pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
4169 ev_loop (EV_A_ 0);
4170 l_release (EV_A);
4172 return 0;
4173 }
4175 Instead of invoking all pending watchers, the C<l_invoke> callback will
4176 signal the main thread via some unspecified mechanism (signals? pipe
4177 writes? C<Async::Interrupt>?) and then waits until all pending watchers
4178 have been called (in a while loop because a) spurious wakeups are possible
4179 and b) skipping inter-thread-communication when there are no pending
4180 watchers is very beneficial):
4182 static void
4183 l_invoke (EV_P)
4184 {
4185 userdata *u = ev_userdata (EV_A);
4187 while (ev_pending_count (EV_A))
4188 {
4189 wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
4190 pthread_cond_wait (&u->invoke_cv, &u->lock);
4191 }
4192 }
4194 Now, whenever the main thread gets told to invoke pending watchers, it
4195 will grab the lock, call C<ev_invoke_pending> and then signal the loop
4196 thread to continue:
4198 static void
4199 real_invoke_pending (EV_P)
4200 {
4201 userdata *u = ev_userdata (EV_A);
4203 pthread_mutex_lock (&u->lock);
4204 ev_invoke_pending (EV_A);
4205 pthread_cond_signal (&u->invoke_cv);
4206 pthread_mutex_unlock (&u->lock);
4207 }
4209 Whenever you want to start/stop a watcher or do other modifications to an
4210 event loop, you will now have to lock:
4212 ev_timer timeout_watcher;
4213 userdata *u = ev_userdata (EV_A);
4215 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
4217 pthread_mutex_lock (&u->lock);
4218 ev_timer_start (EV_A_ &timeout_watcher);
4219 ev_async_send (EV_A_ &u->async_w);
4220 pthread_mutex_unlock (&u->lock);
4222 Note that sending the C<ev_async> watcher is required because otherwise
4223 an event loop currently blocking in the kernel will have no knowledge
4224 about the newly added timer. By waking up the loop it will pick up any new
4225 watchers in the next event loop iteration.
4227 =head3 COROUTINES
4229 Libev is very accommodating to coroutines ("cooperative threads"):
4230 libev fully supports nesting calls to its functions from different
4231 coroutines (e.g. you can call C<ev_loop> on the same loop from two
4232 different coroutines, and switch freely between both coroutines running
4233 the loop, as long as you don't confuse yourself). The only exception is
4234 that you must not do this from C<ev_periodic> reschedule callbacks.
4236 Care has been taken to ensure that libev does not keep local state inside
4237 C<ev_loop>, and other calls do not usually allow for coroutine switches as
4238 they do not call any callbacks.
4242 Depending on your compiler and compiler settings, you might get no or a
4243 lot of warnings when compiling libev code. Some people are apparently
4244 scared by this.
4246 However, these are unavoidable for many reasons. For one, each compiler
4247 has different warnings, and each user has different tastes regarding
4248 warning options. "Warn-free" code therefore cannot be a goal except when
4249 targeting a specific compiler and compiler-version.
4251 Another reason is that some compiler warnings require elaborate
4252 workarounds, or other changes to the code that make it less clear and less
4253 maintainable.
4255 And of course, some compiler warnings are just plain stupid, or simply
4256 wrong (because they don't actually warn about the condition their message
4257 seems to warn about). For example, certain older gcc versions had some
4258 warnings that resulted an extreme number of false positives. These have
4259 been fixed, but some people still insist on making code warn-free with
4260 such buggy versions.
4262 While libev is written to generate as few warnings as possible,
4263 "warn-free" code is not a goal, and it is recommended not to build libev
4264 with any compiler warnings enabled unless you are prepared to cope with
4265 them (e.g. by ignoring them). Remember that warnings are just that:
4266 warnings, not errors, or proof of bugs.
4269 =head2 VALGRIND
4271 Valgrind has a special section here because it is a popular tool that is
4272 highly useful. Unfortunately, valgrind reports are very hard to interpret.
4274 If you think you found a bug (memory leak, uninitialised data access etc.)
4275 in libev, then check twice: If valgrind reports something like:
4277 ==2274== definitely lost: 0 bytes in 0 blocks.
4278 ==2274== possibly lost: 0 bytes in 0 blocks.
4279 ==2274== still reachable: 256 bytes in 1 blocks.
4281 Then there is no memory leak, just as memory accounted to global variables
4282 is not a memleak - the memory is still being referenced, and didn't leak.
4284 Similarly, under some circumstances, valgrind might report kernel bugs
4285 as if it were a bug in libev (e.g. in realloc or in the poll backend,
4286 although an acceptable workaround has been found here), or it might be
4287 confused.
4289 Keep in mind that valgrind is a very good tool, but only a tool. Don't
4290 make it into some kind of religion.
4292 If you are unsure about something, feel free to contact the mailing list
4293 with the full valgrind report and an explanation on why you think this
4294 is a bug in libev (best check the archives, too :). However, don't be
4295 annoyed when you get a brisk "this is no bug" answer and take the chance
4296 of learning how to interpret valgrind properly.
4298 If you need, for some reason, empty reports from valgrind for your project
4299 I suggest using suppression lists.
4306 Win32 doesn't support any of the standards (e.g. POSIX) that libev
4307 requires, and its I/O model is fundamentally incompatible with the POSIX
4308 model. Libev still offers limited functionality on this platform in
4309 the form of the C<EVBACKEND_SELECT> backend, and only supports socket
4310 descriptors. This only applies when using Win32 natively, not when using
4311 e.g. cygwin.
4313 Lifting these limitations would basically require the full
4314 re-implementation of the I/O system. If you are into these kinds of
4315 things, then note that glib does exactly that for you in a very portable
4316 way (note also that glib is the slowest event library known to man).
4318 There is no supported compilation method available on windows except
4319 embedding it into other applications.
4321 Sensible signal handling is officially unsupported by Microsoft - libev
4322 tries its best, but under most conditions, signals will simply not work.
4324 Not a libev limitation but worth mentioning: windows apparently doesn't
4325 accept large writes: instead of resulting in a partial write, windows will
4326 either accept everything or return C<ENOBUFS> if the buffer is too large,
4327 so make sure you only write small amounts into your sockets (less than a
4328 megabyte seems safe, but this apparently depends on the amount of memory
4329 available).
4331 Due to the many, low, and arbitrary limits on the win32 platform and
4332 the abysmal performance of winsockets, using a large number of sockets
4333 is not recommended (and not reasonable). If your program needs to use
4334 more than a hundred or so sockets, then likely it needs to use a totally
4335 different implementation for windows, as libev offers the POSIX readiness
4336 notification model, which cannot be implemented efficiently on windows
4337 (due to Microsoft monopoly games).
4339 A typical way to use libev under windows is to embed it (see the embedding
4340 section for details) and use the following F<evwrap.h> header file instead
4341 of F<ev.h>:
4343 #define EV_STANDALONE /* keeps ev from requiring config.h */
4344 #define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
4346 #include "ev.h"
4348 And compile the following F<evwrap.c> file into your project (make sure
4349 you do I<not> compile the F<ev.c> or any other embedded source files!):
4351 #include "evwrap.h"
4352 #include "ev.c"
4354 =over 4
4356 =item The winsocket select function
4358 The winsocket C<select> function doesn't follow POSIX in that it
4359 requires socket I<handles> and not socket I<file descriptors> (it is
4360 also extremely buggy). This makes select very inefficient, and also
4361 requires a mapping from file descriptors to socket handles (the Microsoft
4362 C runtime provides the function C<_open_osfhandle> for this). See the
4363 discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
4364 C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
4366 The configuration for a "naked" win32 using the Microsoft runtime
4367 libraries and raw winsocket select is:
4369 #define EV_USE_SELECT 1
4370 #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
4372 Note that winsockets handling of fd sets is O(n), so you can easily get a
4373 complexity in the O(n²) range when using win32.
4375 =item Limited number of file descriptors
4377 Windows has numerous arbitrary (and low) limits on things.
4379 Early versions of winsocket's select only supported waiting for a maximum
4380 of C<64> handles (probably owning to the fact that all windows kernels
4381 can only wait for C<64> things at the same time internally; Microsoft
4382 recommends spawning a chain of threads and wait for 63 handles and the
4383 previous thread in each. Sounds great!).
4385 Newer versions support more handles, but you need to define C<FD_SETSIZE>
4386 to some high number (e.g. C<2048>) before compiling the winsocket select
4387 call (which might be in libev or elsewhere, for example, perl and many
4388 other interpreters do their own select emulation on windows).
4390 Another limit is the number of file descriptors in the Microsoft runtime
4391 libraries, which by default is C<64> (there must be a hidden I<64>
4392 fetish or something like this inside Microsoft). You can increase this
4393 by calling C<_setmaxstdio>, which can increase this limit to C<2048>
4394 (another arbitrary limit), but is broken in many versions of the Microsoft
4395 runtime libraries. This might get you to about C<512> or C<2048> sockets
4396 (depending on windows version and/or the phase of the moon). To get more,
4397 you need to wrap all I/O functions and provide your own fd management, but
4398 the cost of calling select (O(n²)) will likely make this unworkable.
4400 =back
4404 In addition to a working ISO-C implementation and of course the
4405 backend-specific APIs, libev relies on a few additional extensions:
4407 =over 4
4409 =item C<void (*)(ev_watcher_type *, int revents)> must have compatible
4410 calling conventions regardless of C<ev_watcher_type *>.
4412 Libev assumes not only that all watcher pointers have the same internal
4413 structure (guaranteed by POSIX but not by ISO C for example), but it also
4414 assumes that the same (machine) code can be used to call any watcher
4415 callback: The watcher callbacks have different type signatures, but libev
4416 calls them using an C<ev_watcher *> internally.
4418 =item C<sig_atomic_t volatile> must be thread-atomic as well
4420 The type C<sig_atomic_t volatile> (or whatever is defined as
4421 C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
4422 threads. This is not part of the specification for C<sig_atomic_t>, but is
4423 believed to be sufficiently portable.
4425 =item C<sigprocmask> must work in a threaded environment
4427 Libev uses C<sigprocmask> to temporarily block signals. This is not
4428 allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
4429 pthread implementations will either allow C<sigprocmask> in the "main
4430 thread" or will block signals process-wide, both behaviours would
4431 be compatible with libev. Interaction between C<sigprocmask> and
4432 C<pthread_sigmask> could complicate things, however.
4434 The most portable way to handle signals is to block signals in all threads
4435 except the initial one, and run the default loop in the initial thread as
4436 well.
4438 =item C<long> must be large enough for common memory allocation sizes
4440 To improve portability and simplify its API, libev uses C<long> internally
4441 instead of C<size_t> when allocating its data structures. On non-POSIX
4442 systems (Microsoft...) this might be unexpectedly low, but is still at
4443 least 31 bits everywhere, which is enough for hundreds of millions of
4444 watchers.
4446 =item C<double> must hold a time value in seconds with enough accuracy
4448 The type C<double> is used to represent timestamps. It is required to
4449 have at least 51 bits of mantissa (and 9 bits of exponent), which is good
4450 enough for at least into the year 4000. This requirement is fulfilled by
4451 implementations implementing IEEE 754, which is basically all existing
4452 ones. With IEEE 754 doubles, you get microsecond accuracy until at least
4453 2200.
4455 =back
4457 If you know of other additional requirements drop me a note.
4462 In this section the complexities of (many of) the algorithms used inside
4463 libev will be documented. For complexity discussions about backends see
4464 the documentation for C<ev_default_init>.
4466 All of the following are about amortised time: If an array needs to be
4467 extended, libev needs to realloc and move the whole array, but this
4468 happens asymptotically rarer with higher number of elements, so O(1) might
4469 mean that libev does a lengthy realloc operation in rare cases, but on
4470 average it is much faster and asymptotically approaches constant time.
4472 =over 4
4474 =item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
4476 This means that, when you have a watcher that triggers in one hour and
4477 there are 100 watchers that would trigger before that, then inserting will
4478 have to skip roughly seven (C<ld 100>) of these watchers.
4480 =item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
4482 That means that changing a timer costs less than removing/adding them,
4483 as only the relative motion in the event queue has to be paid for.
4485 =item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
4487 These just add the watcher into an array or at the head of a list.
4489 =item Stopping check/prepare/idle/fork/async watchers: O(1)
4491 =item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
4493 These watchers are stored in lists, so they need to be walked to find the
4494 correct watcher to remove. The lists are usually short (you don't usually
4495 have many watchers waiting for the same fd or signal: one is typical, two
4496 is rare).
4498 =item Finding the next timer in each loop iteration: O(1)
4500 By virtue of using a binary or 4-heap, the next timer is always found at a
4501 fixed position in the storage array.
4503 =item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
4505 A change means an I/O watcher gets started or stopped, which requires
4506 libev to recalculate its status (and possibly tell the kernel, depending
4507 on backend and whether C<ev_io_set> was used).
4509 =item Activating one watcher (putting it into the pending state): O(1)
4511 =item Priority handling: O(number_of_priorities)
4513 Priorities are implemented by allocating some space for each
4514 priority. When doing priority-based operations, libev usually has to
4515 linearly search all the priorities, but starting/stopping and activating
4516 watchers becomes O(1) with respect to priority handling.
4518 =item Sending an ev_async: O(1)
4520 =item Processing ev_async_send: O(number_of_async_watchers)
4522 =item Processing signals: O(max_signal_number)
4524 Sending involves a system call I<iff> there were no other C<ev_async_send>
4525 calls in the current loop iteration. Checking for async and signal events
4526 involves iterating over all running async watchers or all signal numbers.
4528 =back
4531 =head1 GLOSSARY
4533 =over 4
4535 =item active
4537 A watcher is active as long as it has been started (has been attached to
4538 an event loop) but not yet stopped (disassociated from the event loop).
4540 =item application
4542 In this document, an application is whatever is using libev.
4544 =item callback
4546 The address of a function that is called when some event has been
4547 detected. Callbacks are being passed the event loop, the watcher that
4548 received the event, and the actual event bitset.
4550 =item callback invocation
4552 The act of calling the callback associated with a watcher.
4554 =item event
4556 A change of state of some external event, such as data now being available
4557 for reading on a file descriptor, time having passed or simply not having
4558 any other events happening anymore.
4560 In libev, events are represented as single bits (such as C<EV_READ> or
4561 C<EV_TIMEOUT>).
4563 =item event library
4565 A software package implementing an event model and loop.
4567 =item event loop
4569 An entity that handles and processes external events and converts them
4570 into callback invocations.
4572 =item event model
4574 The model used to describe how an event loop handles and processes
4575 watchers and events.
4577 =item pending
4579 A watcher is pending as soon as the corresponding event has been detected,
4580 and stops being pending as soon as the watcher will be invoked or its
4581 pending status is explicitly cleared by the application.
4583 A watcher can be pending, but not active. Stopping a watcher also clears
4584 its pending status.
4586 =item real time
4588 The physical time that is observed. It is apparently strictly monotonic :)
4590 =item wall-clock time
4592 The time and date as shown on clocks. Unlike real time, it can actually
4593 be wrong and jump forwards and backwards, e.g. when the you adjust your
4594 clock.
4596 =item watcher
4598 A data structure that describes interest in certain events. Watchers need
4599 to be started (attached to an event loop) before they can receive events.
4601 =item watcher invocation
4603 The act of calling the callback associated with a watcher.
4605 =back
4607 =head1 AUTHOR
4609 Marc Lehmann <>, with repeated corrections by Mikael Magnusson.