ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.162 by root, Mon May 26 03:54:40 2008 UTC vs.
Revision 1.351 by root, Mon Jan 10 14:24:26 2011 UTC

…		…
2		2
3	libev - a high performance full-featured event loop written in C	3	libev - a high performance full-featured event loop written in C
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	#include <ev.h>	7	#include <ev.h>
8		8
9	=head2 EXAMPLE PROGRAM	9	=head2 EXAMPLE PROGRAM
10		10
11	// a single header file is required	11	// a single header file is required
12	#include <ev.h>	12	#include <ev.h>
13		13
		14	#include <stdio.h> // for puts
		15
14	// every watcher type has its own typedef'd struct	16	// every watcher type has its own typedef'd struct
15	// with the name ev_<type>	17	// with the name ev_TYPE
16	ev_io stdin_watcher;	18	ev_io stdin_watcher;
17	ev_timer timeout_watcher;	19	ev_timer timeout_watcher;
18		20
19	// all watcher callbacks have a similar signature	21	// all watcher callbacks have a similar signature
20	// this callback is called when data is readable on stdin	22	// this callback is called when data is readable on stdin
21	static void	23	static void
22	stdin_cb (EV_P_ struct ev_io *w, int revents)	24	stdin_cb (EV_P_ ev_io *w, int revents)
23	{	25	{
24	puts ("stdin ready");	26	puts ("stdin ready");
25	// for one-shot events, one must manually stop the watcher	27	// for one-shot events, one must manually stop the watcher
26	// with its corresponding stop function.	28	// with its corresponding stop function.
27	ev_io_stop (EV_A_ w);	29	ev_io_stop (EV_A_ w);
28		30
29	// this causes all nested ev_loop's to stop iterating	31	// this causes all nested ev_run's to stop iterating
30	ev_unloop (EV_A_ EVUNLOOP_ALL);	32	ev_break (EV_A_ EVBREAK_ALL);
31	}	33	}
32		34
33	// another callback, this time for a time-out	35	// another callback, this time for a time-out
34	static void	36	static void
35	timeout_cb (EV_P_ struct ev_timer *w, int revents)	37	timeout_cb (EV_P_ ev_timer *w, int revents)
36	{	38	{
37	puts ("timeout");	39	puts ("timeout");
38	// this causes the innermost ev_loop to stop iterating	40	// this causes the innermost ev_run to stop iterating
39	ev_unloop (EV_A_ EVUNLOOP_ONE);	41	ev_break (EV_A_ EVBREAK_ONE);
40	}	42	}
41		43
42	int	44	int
43	main (void)	45	main (void)
44	{	46	{
45	// use the default event loop unless you have special needs	47	// use the default event loop unless you have special needs
46	struct ev_loop *loop = ev_default_loop (0);	48	struct ev_loop *loop = EV_DEFAULT;
47		49
48	// initialise an io watcher, then start it	50	// initialise an io watcher, then start it
49	// this one will watch for stdin to become readable	51	// this one will watch for stdin to become readable
50	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	52	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
51	ev_io_start (loop, &stdin_watcher);	53	ev_io_start (loop, &stdin_watcher);
52		54
53	// initialise a timer watcher, then start it	55	// initialise a timer watcher, then start it
54	// simple non-repeating 5.5 second timeout	56	// simple non-repeating 5.5 second timeout
55	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);	57	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
56	ev_timer_start (loop, &timeout_watcher);	58	ev_timer_start (loop, &timeout_watcher);
57		59
58	// now wait for events to arrive	60	// now wait for events to arrive
59	ev_loop (loop, 0);	61	ev_run (loop, 0);
60		62
61	// unloop was called, so exit	63	// unloop was called, so exit
62	return 0;	64	return 0;
63	}	65	}
64		66
65	=head1 DESCRIPTION	67	=head1 ABOUT THIS DOCUMENT
		68
		69	This document documents the libev software package.
66		70
67	The newest version of this document is also available as an html-formatted	71	The newest version of this document is also available as an html-formatted
68	web page you might find easier to navigate when reading it for the first	72	web page you might find easier to navigate when reading it for the first
69	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.	73	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		74
		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.
		79
		80	Familiarity with event based programming techniques in general is assumed
		81	throughout this document.
		82
		83	=head1 WHAT TO READ WHEN IN A HURRY
		84
		85	This manual tries to be very detailed, but unfortunately, this also makes
		86	it very long. If you just want to know the basics of libev, I suggest
		87	reading L<ANATOMY OF A WATCHER>, then the L<EXAMPLE PROGRAM> above and
		88	look up the missing functions in L<GLOBAL FUNCTIONS> and the C<ev_io> and
		89	C<ev_timer> sections in L<WATCHER TYPES>.
		90
		91	=head1 ABOUT LIBEV
70		92
71	Libev is an event loop: you register interest in certain events (such as a	93	Libev is an event loop: you register interest in certain events (such as a
72	file descriptor being readable or a timeout occurring), and it will manage	94	file descriptor being readable or a timeout occurring), and it will manage
73	these event sources and provide your program with events.	95	these event sources and provide your program with events.
74		96
…		…
84	=head2 FEATURES	106	=head2 FEATURES
85		107
86	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	108	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
87	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	109	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
88	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	110	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
89	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers	111	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
90	with customised rescheduling (C<ev_periodic>), synchronous signals	112	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
91	(C<ev_signal>), process status change events (C<ev_child>), and event	113	timers (C<ev_timer>), absolute timers with customised rescheduling
92	watchers dealing with the event loop mechanism itself (C<ev_idle>,	114	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
93	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as	115	change events (C<ev_child>), and event watchers dealing with the event
94	file watchers (C<ev_stat>) and even limited support for fork events	116	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
95	(C<ev_fork>).	117	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		118	limited support for fork events (C<ev_fork>).
96		119
97	It also is quite fast (see this	120	It also is quite fast (see this
98	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent	121	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
99	for example).	122	for example).
100		123
…		…
108	name C<loop> (which is always of type C<struct ev_loop *>) will not have	131	name C<loop> (which is always of type C<struct ev_loop *>) will not have
109	this argument.	132	this argument.
110		133
111	=head2 TIME REPRESENTATION	134	=head2 TIME REPRESENTATION
112		135
113	Libev represents time as a single floating point number, representing the	136	Libev represents time as a single floating point number, representing
114	(fractional) number of seconds since the (POSIX) epoch (somewhere near	137	the (fractional) number of seconds since the (POSIX) epoch (in practice
115	the beginning of 1970, details are complicated, don't ask). This type is	138	somewhere near the beginning of 1970, details are complicated, don't
116	called C<ev_tstamp>, which is what you should use too. It usually aliases	139	ask). This type is called C<ev_tstamp>, which is what you should use
117	to the C<double> type in C, and when you need to do any calculations on	140	too. It usually aliases to the C<double> type in C. When you need to do
118	it, you should treat it as some floating point value. Unlike the name	141	any calculations on it, you should treat it as some floating point value.
		142
119	component C<stamp> might indicate, it is also used for time differences	143	Unlike the name component C<stamp> might indicate, it is also used for
120	throughout libev.	144	time differences (e.g. delays) throughout libev.
121		145
122	=head1 ERROR HANDLING	146	=head1 ERROR HANDLING
123		147
124	Libev knows three classes of errors: operating system errors, usage errors	148	Libev knows three classes of errors: operating system errors, usage errors
125	and internal errors (bugs).	149	and internal errors (bugs).
…		…
149		173
150	=item ev_tstamp ev_time ()	174	=item ev_tstamp ev_time ()
151		175
152	Returns the current time as libev would use it. Please note that the	176	Returns the current time as libev would use it. Please note that the
153	C<ev_now> function is usually faster and also often returns the timestamp	177	C<ev_now> function is usually faster and also often returns the timestamp
154	you actually want to know.	178	you actually want to know. Also interesting is the combination of
		179	C<ev_update_now> and C<ev_now>.
155		180
156	=item ev_sleep (ev_tstamp interval)	181	=item ev_sleep (ev_tstamp interval)
157		182
158	Sleep for the given interval: The current thread will be blocked until	183	Sleep for the given interval: The current thread will be blocked until
159	either it is interrupted or the given time interval has passed. Basically	184	either it is interrupted or the given time interval has passed. Basically
…		…
176	as this indicates an incompatible change. Minor versions are usually	201	as this indicates an incompatible change. Minor versions are usually
177	compatible to older versions, so a larger minor version alone is usually	202	compatible to older versions, so a larger minor version alone is usually
178	not a problem.	203	not a problem.
179		204
180	Example: Make sure we haven't accidentally been linked against the wrong	205	Example: Make sure we haven't accidentally been linked against the wrong
181	version.	206	version (note, however, that this will not detect other ABI mismatches,
		207	such as LFS or reentrancy).
182		208
183	assert (("libev version mismatch",	209	assert (("libev version mismatch",
184	ev_version_major () == EV_VERSION_MAJOR	210	ev_version_major () == EV_VERSION_MAJOR
185	&& ev_version_minor () >= EV_VERSION_MINOR));	211	&& ev_version_minor () >= EV_VERSION_MINOR));
186		212
187	=item unsigned int ev_supported_backends ()	213	=item unsigned int ev_supported_backends ()
188		214
189	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	215	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
190	value) compiled into this binary of libev (independent of their	216	value) compiled into this binary of libev (independent of their
…		…
192	a description of the set values.	218	a description of the set values.
193		219
194	Example: make sure we have the epoll method, because yeah this is cool and	220	Example: make sure we have the epoll method, because yeah this is cool and
195	a must have and can we have a torrent of it please!!!11	221	a must have and can we have a torrent of it please!!!11
196		222
197	assert (("sorry, no epoll, no sex",	223	assert (("sorry, no epoll, no sex",
198	ev_supported_backends () & EVBACKEND_EPOLL));	224	ev_supported_backends () & EVBACKEND_EPOLL));
199		225
200	=item unsigned int ev_recommended_backends ()	226	=item unsigned int ev_recommended_backends ()
201		227
202	Return the set of all backends compiled into this binary of libev and also	228	Return the set of all backends compiled into this binary of libev and
203	recommended for this platform. This set is often smaller than the one	229	also recommended for this platform, meaning it will work for most file
		230	descriptor types. This set is often smaller than the one returned by
204	returned by C<ev_supported_backends>, as for example kqueue is broken on	231	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
205	most BSDs and will not be auto-detected unless you explicitly request it	232	and will not be auto-detected unless you explicitly request it (assuming
206	(assuming you know what you are doing). This is the set of backends that	233	you know what you are doing). This is the set of backends that libev will
207	libev will probe for if you specify no backends explicitly.	234	probe for if you specify no backends explicitly.
208		235
209	=item unsigned int ev_embeddable_backends ()	236	=item unsigned int ev_embeddable_backends ()
210		237
211	Returns the set of backends that are embeddable in other event loops. This	238	Returns the set of backends that are embeddable in other event loops. This
212	is the theoretical, all-platform, value. To find which backends	239	value is platform-specific but can include backends not available on the
213	might be supported on the current system, you would need to look at	240	current system. To find which embeddable backends might be supported on
214	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	241	the current system, you would need to look at C<ev_embeddable_backends ()
215	recommended ones.	242	& ev_supported_backends ()>, likewise for recommended ones.
216		243
217	See the description of C<ev_embed> watchers for more info.	244	See the description of C<ev_embed> watchers for more info.
218		245
219	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	246	=item ev_set_allocator (void *(*cb)(void *ptr, long size))
220		247
…		…
250	}	277	}
251		278
252	...	279	...
253	ev_set_allocator (persistent_realloc);	280	ev_set_allocator (persistent_realloc);
254		281
255	=item ev_set_syserr_cb (void (*cb)(const char *msg));	282	=item ev_set_syserr_cb (void (*cb)(const char *msg))
256		283
257	Set the callback function to call on a retryable system call error (such	284	Set the callback function to call on a retryable system call error (such
258	as failed select, poll, epoll_wait). The message is a printable string	285	as failed select, poll, epoll_wait). The message is a printable string
259	indicating the system call or subsystem causing the problem. If this	286	indicating the system call or subsystem causing the problem. If this
260	callback is set, then libev will expect it to remedy the situation, no	287	callback is set, then libev will expect it to remedy the situation, no
…		…
272	}	299	}
273		300
274	...	301	...
275	ev_set_syserr_cb (fatal_error);	302	ev_set_syserr_cb (fatal_error);
276		303
		304	=item ev_feed_signal (int signum)
		305
		306	This function can be used to "simulate" a signal receive. It is completely
		307	safe to call this function at any time, from any context, including signal
		308	handlers or random threads.
		309
		310	Its main use is to customise signal handling in your process, especially
		311	in the presence of threads. For example, you could block signals
		312	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		313	creating any loops), and in one thread, use C<sigwait> or any other
		314	mechanism to wait for signals, then "deliver" them to libev by calling
		315	C<ev_feed_signal>.
		316
277	=back	317	=back
278		318
279	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	319	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
280		320
281	An event loop is described by a C<struct ev_loop *>. The library knows two	321	An event loop is described by a C<struct ev_loop *> (the C<struct> is
282	types of such loops, the I<default> loop, which supports signals and child	322	I<not> optional in this case unless libev 3 compatibility is disabled, as
283	events, and dynamically created loops which do not.	323	libev 3 had an C<ev_loop> function colliding with the struct name).
		324
		325	The library knows two types of such loops, the I<default> loop, which
		326	supports child process events, and dynamically created event loops which
		327	do not.
284		328
285	=over 4	329	=over 4
286		330
287	=item struct ev_loop *ev_default_loop (unsigned int flags)	331	=item struct ev_loop *ev_default_loop (unsigned int flags)
288		332
289	This will initialise the default event loop if it hasn't been initialised	333	This returns the "default" event loop object, which is what you should
290	yet and return it. If the default loop could not be initialised, returns	334	normally use when you just need "the event loop". Event loop objects and
291	false. If it already was initialised it simply returns it (and ignores the	335	the C<flags> parameter are described in more detail in the entry for
292	flags. If that is troubling you, check C<ev_backend ()> afterwards).	336	C<ev_loop_new>.
		337
		338	If the default loop is already initialised then this function simply
		339	returns it (and ignores the flags. If that is troubling you, check
		340	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		341	flags, which should almost always be C<0>, unless the caller is also the
		342	one calling C<ev_run> or otherwise qualifies as "the main program".
293		343
294	If you don't know what event loop to use, use the one returned from this	344	If you don't know what event loop to use, use the one returned from this
295	function.	345	function (or via the C<EV_DEFAULT> macro).
296		346
297	Note that this function is I<not> thread-safe, so if you want to use it	347	Note that this function is I<not> thread-safe, so if you want to use it
298	from multiple threads, you have to lock (note also that this is unlikely,	348	from multiple threads, you have to employ some kind of mutex (note also
299	as loops cannot bes hared easily between threads anyway).	349	that this case is unlikely, as loops cannot be shared easily between
		350	threads anyway).
300		351
301	The default loop is the only loop that can handle C<ev_signal> and	352	The default loop is the only loop that can handle C<ev_child> watchers,
302	C<ev_child> watchers, and to do this, it always registers a handler	353	and to do this, it always registers a handler for C<SIGCHLD>. If this is
303	for C<SIGCHLD>. If this is a problem for your application you can either	354	a problem for your application you can either create a dynamic loop with
304	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you	355	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
305	can simply overwrite the C<SIGCHLD> signal handler I<after> calling	356	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
306	C<ev_default_init>.	357
		358	Example: This is the most typical usage.
		359
		360	if (!ev_default_loop (0))
		361	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		362
		363	Example: Restrict libev to the select and poll backends, and do not allow
		364	environment settings to be taken into account:
		365
		366	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		367
		368	=item struct ev_loop *ev_loop_new (unsigned int flags)
		369
		370	This will create and initialise a new event loop object. If the loop
		371	could not be initialised, returns false.
		372
		373	This function is thread-safe, and one common way to use libev with
		374	threads is indeed to create one loop per thread, and using the default
		375	loop in the "main" or "initial" thread.
307		376
308	The flags argument can be used to specify special behaviour or specific	377	The flags argument can be used to specify special behaviour or specific
309	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	378	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
310		379
311	The following flags are supported:	380	The following flags are supported:
…		…
326	useful to try out specific backends to test their performance, or to work	395	useful to try out specific backends to test their performance, or to work
327	around bugs.	396	around bugs.
328		397
329	=item C<EVFLAG_FORKCHECK>	398	=item C<EVFLAG_FORKCHECK>
330		399
331	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after	400	Instead of calling C<ev_loop_fork> manually after a fork, you can also
332	a fork, you can also make libev check for a fork in each iteration by	401	make libev check for a fork in each iteration by enabling this flag.
333	enabling this flag.
334		402
335	This works by calling C<getpid ()> on every iteration of the loop,	403	This works by calling C<getpid ()> on every iteration of the loop,
336	and thus this might slow down your event loop if you do a lot of loop	404	and thus this might slow down your event loop if you do a lot of loop
337	iterations and little real work, but is usually not noticeable (on my	405	iterations and little real work, but is usually not noticeable (on my
338	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence	406	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
…		…
344	flag.	412	flag.
345		413
346	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>	414	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
347	environment variable.	415	environment variable.
348		416
		417	=item C<EVFLAG_NOINOTIFY>
		418
		419	When this flag is specified, then libev will not attempt to use the
		420	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
		421	testing, this flag can be useful to conserve inotify file descriptors, as
		422	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		423
		424	=item C<EVFLAG_SIGNALFD>
		425
		426	When this flag is specified, then libev will attempt to use the
		427	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
		428	delivers signals synchronously, which makes it both faster and might make
		429	it possible to get the queued signal data. It can also simplify signal
		430	handling with threads, as long as you properly block signals in your
		431	threads that are not interested in handling them.
		432
		433	Signalfd will not be used by default as this changes your signal mask, and
		434	there are a lot of shoddy libraries and programs (glib's threadpool for
		435	example) that can't properly initialise their signal masks.
		436
		437	=item C<EVFLAG_NOSIGMASK>
		438
		439	When this flag is specified, then libev will avoid to modify the signal
		440	mask. Specifically, this means you ahve to make sure signals are unblocked
		441	when you want to receive them.
		442
		443	This behaviour is useful when you want to do your own signal handling, or
		444	want to handle signals only in specific threads and want to avoid libev
		445	unblocking the signals.
		446
		447	This flag's behaviour will become the default in future versions of libev.
		448
349	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	449	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
350		450
351	This is your standard select(2) backend. Not I<completely> standard, as	451	This is your standard select(2) backend. Not I<completely> standard, as
352	libev tries to roll its own fd_set with no limits on the number of fds,	452	libev tries to roll its own fd_set with no limits on the number of fds,
353	but if that fails, expect a fairly low limit on the number of fds when	453	but if that fails, expect a fairly low limit on the number of fds when
…		…
359	writing a server, you should C<accept ()> in a loop to accept as many	459	writing a server, you should C<accept ()> in a loop to accept as many
360	connections as possible during one iteration. You might also want to have	460	connections as possible during one iteration. You might also want to have
361	a look at C<ev_set_io_collect_interval ()> to increase the amount of	461	a look at C<ev_set_io_collect_interval ()> to increase the amount of
362	readiness notifications you get per iteration.	462	readiness notifications you get per iteration.
363		463
		464	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		465	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		466	C<exceptfds> set on that platform).
		467
364	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	468	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
365		469
366	And this is your standard poll(2) backend. It's more complicated	470	And this is your standard poll(2) backend. It's more complicated
367	than select, but handles sparse fds better and has no artificial	471	than select, but handles sparse fds better and has no artificial
368	limit on the number of fds you can use (except it will slow down	472	limit on the number of fds you can use (except it will slow down
369	considerably with a lot of inactive fds). It scales similarly to select,	473	considerably with a lot of inactive fds). It scales similarly to select,
370	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for	474	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
371	performance tips.	475	performance tips.
372		476
		477	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		478	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		479
373	=item C<EVBACKEND_EPOLL> (value 4, Linux)	480	=item C<EVBACKEND_EPOLL> (value 4, Linux)
		481
		482	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		483	kernels).
374		484
375	For few fds, this backend is a bit little slower than poll and select,	485	For few fds, this backend is a bit little slower than poll and select,
376	but it scales phenomenally better. While poll and select usually scale	486	but it scales phenomenally better. While poll and select usually scale
377	like O(total_fds) where n is the total number of fds (or the highest fd),	487	like O(total_fds) where n is the total number of fds (or the highest fd),
378	epoll scales either O(1) or O(active_fds). The epoll design has a number	488	epoll scales either O(1) or O(active_fds).
379	of shortcomings, such as silently dropping events in some hard-to-detect	489
380	cases and requiring a system call per fd change, no fork support and bad	490	The epoll mechanism deserves honorable mention as the most misdesigned
381	support for dup.	491	of the more advanced event mechanisms: mere annoyances include silently
		492	dropping file descriptors, requiring a system call per change per file
		493	descriptor (and unnecessary guessing of parameters), problems with dup,
		494	returning before the timeout value, resulting in additional iterations
		495	(and only giving 5ms accuracy while select on the same platform gives
		496	0.1ms) and so on. The biggest issue is fork races, however - if a program
		497	forks then I<both> parent and child process have to recreate the epoll
		498	set, which can take considerable time (one syscall per file descriptor)
		499	and is of course hard to detect.
		500
		501	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		502	of course I<doesn't>, and epoll just loves to report events for totally
		503	I<different> file descriptors (even already closed ones, so one cannot
		504	even remove them from the set) than registered in the set (especially
		505	on SMP systems). Libev tries to counter these spurious notifications by
		506	employing an additional generation counter and comparing that against the
		507	events to filter out spurious ones, recreating the set when required. Last
		508	not least, it also refuses to work with some file descriptors which work
		509	perfectly fine with C<select> (files, many character devices...).
		510
		511	Epoll is truly the train wreck analog among event poll mechanisms.
382		512
383	While stopping, setting and starting an I/O watcher in the same iteration	513	While stopping, setting and starting an I/O watcher in the same iteration
384	will result in some caching, there is still a system call per such incident	514	will result in some caching, there is still a system call per such
385	(because the fd could point to a different file description now), so its	515	incident (because the same I<file descriptor> could point to a different
386	best to avoid that. Also, C<dup ()>'ed file descriptors might not work	516	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
387	very well if you register events for both fds.	517	file descriptors might not work very well if you register events for both
388		518	file descriptors.
389	Please note that epoll sometimes generates spurious notifications, so you
390	need to use non-blocking I/O or other means to avoid blocking when no data
391	(or space) is available.
392		519
393	Best performance from this backend is achieved by not unregistering all	520	Best performance from this backend is achieved by not unregistering all
394	watchers for a file descriptor until it has been closed, if possible, i.e.	521	watchers for a file descriptor until it has been closed, if possible,
395	keep at least one watcher active per fd at all times.	522	i.e. keep at least one watcher active per fd at all times. Stopping and
		523	starting a watcher (without re-setting it) also usually doesn't cause
		524	extra overhead. A fork can both result in spurious notifications as well
		525	as in libev having to destroy and recreate the epoll object, which can
		526	take considerable time and thus should be avoided.
		527
		528	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		529	faster than epoll for maybe up to a hundred file descriptors, depending on
		530	the usage. So sad.
396		531
397	While nominally embeddable in other event loops, this feature is broken in	532	While nominally embeddable in other event loops, this feature is broken in
398	all kernel versions tested so far.	533	all kernel versions tested so far.
		534
		535	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		536	C<EVBACKEND_POLL>.
399		537
400	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	538	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
401		539
402	Kqueue deserves special mention, as at the time of this writing, it	540	Kqueue deserves special mention, as at the time of this writing, it
403	was broken on all BSDs except NetBSD (usually it doesn't work reliably	541	was broken on all BSDs except NetBSD (usually it doesn't work reliably
404	with anything but sockets and pipes, except on Darwin, where of course	542	with anything but sockets and pipes, except on Darwin, where of course
405	it's completely useless). For this reason it's not being "auto-detected"	543	it's completely useless). Unlike epoll, however, whose brokenness
		544	is by design, these kqueue bugs can (and eventually will) be fixed
		545	without API changes to existing programs. For this reason it's not being
406	unless you explicitly specify it explicitly in the flags (i.e. using	546	"auto-detected" unless you explicitly specify it in the flags (i.e. using
407	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	547	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
408	system like NetBSD.	548	system like NetBSD.
409		549
410	You still can embed kqueue into a normal poll or select backend and use it	550	You still can embed kqueue into a normal poll or select backend and use it
411	only for sockets (after having made sure that sockets work with kqueue on	551	only for sockets (after having made sure that sockets work with kqueue on
…		…
413		553
414	It scales in the same way as the epoll backend, but the interface to the	554	It scales in the same way as the epoll backend, but the interface to the
415	kernel is more efficient (which says nothing about its actual speed, of	555	kernel is more efficient (which says nothing about its actual speed, of
416	course). While stopping, setting and starting an I/O watcher does never	556	course). While stopping, setting and starting an I/O watcher does never
417	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to	557	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
418	two event changes per incident, support for C<fork ()> is very bad and it	558	two event changes per incident. Support for C<fork ()> is very bad (but
419	drops fds silently in similarly hard-to-detect cases.	559	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		560	cases
420		561
421	This backend usually performs well under most conditions.	562	This backend usually performs well under most conditions.
422		563
423	While nominally embeddable in other event loops, this doesn't work	564	While nominally embeddable in other event loops, this doesn't work
424	everywhere, so you might need to test for this. And since it is broken	565	everywhere, so you might need to test for this. And since it is broken
425	almost everywhere, you should only use it when you have a lot of sockets	566	almost everywhere, you should only use it when you have a lot of sockets
426	(for which it usually works), by embedding it into another event loop	567	(for which it usually works), by embedding it into another event loop
427	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for	568	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
428	sockets.	569	also broken on OS X)) and, did I mention it, using it only for sockets.
		570
		571	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		572	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		573	C<NOTE_EOF>.
429		574
430	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	575	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
431		576
432	This is not implemented yet (and might never be, unless you send me an	577	This is not implemented yet (and might never be, unless you send me an
433	implementation). According to reports, C</dev/poll> only supports sockets	578	implementation). According to reports, C</dev/poll> only supports sockets
…		…
437	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	582	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
438		583
439	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	584	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
440	it's really slow, but it still scales very well (O(active_fds)).	585	it's really slow, but it still scales very well (O(active_fds)).
441		586
442	Please note that Solaris event ports can deliver a lot of spurious
443	notifications, so you need to use non-blocking I/O or other means to avoid
444	blocking when no data (or space) is available.
445
446	While this backend scales well, it requires one system call per active	587	While this backend scales well, it requires one system call per active
447	file descriptor per loop iteration. For small and medium numbers of file	588	file descriptor per loop iteration. For small and medium numbers of file
448	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend	589	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
449	might perform better.	590	might perform better.
450		591
451	On the positive side, ignoring the spurious readiness notifications, this	592	On the positive side, this backend actually performed fully to
452	backend actually performed to specification in all tests and is fully	593	specification in all tests and is fully embeddable, which is a rare feat
453	embeddable, which is a rare feat among the OS-specific backends.	594	among the OS-specific backends (I vastly prefer correctness over speed
		595	hacks).
		596
		597	On the negative side, the interface is I<bizarre>, with the event polling
		598	function sometimes returning events to the caller even though an error
		599	occured, but with no indication whether it has done so or not (yes, it's
		600	even documented that way) - deadly for edge-triggered interfaces, but
		601	fortunately libev seems to be able to work around it.
		602
		603	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		604	C<EVBACKEND_POLL>.
454		605
455	=item C<EVBACKEND_ALL>	606	=item C<EVBACKEND_ALL>
456		607
457	Try all backends (even potentially broken ones that wouldn't be tried	608	Try all backends (even potentially broken ones that wouldn't be tried
458	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	609	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
459	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	610	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
460		611
461	It is definitely not recommended to use this flag.	612	It is definitely not recommended to use this flag, use whatever
		613	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		614	at all.
		615
		616	=item C<EVBACKEND_MASK>
		617
		618	Not a backend at all, but a mask to select all backend bits from a
		619	C<flags> value, in case you want to mask out any backends from a flags
		620	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
462		621
463	=back	622	=back
464		623
465	If one or more of these are or'ed into the flags value, then only these	624	If one or more of the backend flags are or'ed into the flags value,
466	backends will be tried (in the reverse order as listed here). If none are	625	then only these backends will be tried (in the reverse order as listed
467	specified, all backends in C<ev_recommended_backends ()> will be tried.	626	here). If none are specified, all backends in C<ev_recommended_backends
468		627	()> will be tried.
469	The most typical usage is like this:
470
471	if (!ev_default_loop (0))
472	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
473
474	Restrict libev to the select and poll backends, and do not allow
475	environment settings to be taken into account:
476
477	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
478
479	Use whatever libev has to offer, but make sure that kqueue is used if
480	available (warning, breaks stuff, best use only with your own private
481	event loop and only if you know the OS supports your types of fds):
482
483	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
484
485	=item struct ev_loop *ev_loop_new (unsigned int flags)
486
487	Similar to C<ev_default_loop>, but always creates a new event loop that is
488	always distinct from the default loop. Unlike the default loop, it cannot
489	handle signal and child watchers, and attempts to do so will be greeted by
490	undefined behaviour (or a failed assertion if assertions are enabled).
491
492	Note that this function I<is> thread-safe, and the recommended way to use
493	libev with threads is indeed to create one loop per thread, and using the
494	default loop in the "main" or "initial" thread.
495		628
496	Example: Try to create a event loop that uses epoll and nothing else.	629	Example: Try to create a event loop that uses epoll and nothing else.
497		630
498	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	631	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
499	if (!epoller)	632	if (!epoller)
500	fatal ("no epoll found here, maybe it hides under your chair");	633	fatal ("no epoll found here, maybe it hides under your chair");
501		634
		635	Example: Use whatever libev has to offer, but make sure that kqueue is
		636	used if available.
		637
		638	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
		639
502	=item ev_default_destroy ()	640	=item ev_loop_destroy (loop)
503		641
504	Destroys the default loop again (frees all memory and kernel state	642	Destroys an event loop object (frees all memory and kernel state
505	etc.). None of the active event watchers will be stopped in the normal	643	etc.). None of the active event watchers will be stopped in the normal
506	sense, so e.g. C<ev_is_active> might still return true. It is your	644	sense, so e.g. C<ev_is_active> might still return true. It is your
507	responsibility to either stop all watchers cleanly yourself I<before>	645	responsibility to either stop all watchers cleanly yourself I<before>
508	calling this function, or cope with the fact afterwards (which is usually	646	calling this function, or cope with the fact afterwards (which is usually
509	the easiest thing, you can just ignore the watchers and/or C<free ()> them	647	the easiest thing, you can just ignore the watchers and/or C<free ()> them
510	for example).	648	for example).
511		649
512	Note that certain global state, such as signal state, will not be freed by	650	Note that certain global state, such as signal state (and installed signal
513	this function, and related watchers (such as signal and child watchers)	651	handlers), will not be freed by this function, and related watchers (such
514	would need to be stopped manually.	652	as signal and child watchers) would need to be stopped manually.
515		653
516	In general it is not advisable to call this function except in the	654	This function is normally used on loop objects allocated by
517	rare occasion where you really need to free e.g. the signal handling	655	C<ev_loop_new>, but it can also be used on the default loop returned by
		656	C<ev_default_loop>, in which case it is not thread-safe.
		657
		658	Note that it is not advisable to call this function on the default loop
		659	except in the rare occasion where you really need to free its resources.
518	pipe fds. If you need dynamically allocated loops it is better to use	660	If you need dynamically allocated loops it is better to use C<ev_loop_new>
519	C<ev_loop_new> and C<ev_loop_destroy>).	661	and C<ev_loop_destroy>.
520		662
521	=item ev_loop_destroy (loop)	663	=item ev_loop_fork (loop)
522		664
523	Like C<ev_default_destroy>, but destroys an event loop created by an
524	earlier call to C<ev_loop_new>.
525
526	=item ev_default_fork ()
527
528	This function sets a flag that causes subsequent C<ev_loop> iterations	665	This function sets a flag that causes subsequent C<ev_run> iterations to
529	to reinitialise the kernel state for backends that have one. Despite the	666	reinitialise the kernel state for backends that have one. Despite the
530	name, you can call it anytime, but it makes most sense after forking, in	667	name, you can call it anytime, but it makes most sense after forking, in
531	the child process (or both child and parent, but that again makes little	668	the child process. You I<must> call it (or use C<EVFLAG_FORKCHECK>) in the
532	sense). You I<must> call it in the child before using any of the libev	669	child before resuming or calling C<ev_run>.
533	functions, and it will only take effect at the next C<ev_loop> iteration.	670
		671	Again, you I<have> to call it on I<any> loop that you want to re-use after
		672	a fork, I<even if you do not plan to use the loop in the parent>. This is
		673	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		674	during fork.
534		675
535	On the other hand, you only need to call this function in the child	676	On the other hand, you only need to call this function in the child
536	process if and only if you want to use the event library in the child. If	677	process if and only if you want to use the event loop in the child. If
537	you just fork+exec, you don't have to call it at all.	678	you just fork+exec or create a new loop in the child, you don't have to
		679	call it at all (in fact, C<epoll> is so badly broken that it makes a
		680	difference, but libev will usually detect this case on its own and do a
		681	costly reset of the backend).
538		682
539	The function itself is quite fast and it's usually not a problem to call	683	The function itself is quite fast and it's usually not a problem to call
540	it just in case after a fork. To make this easy, the function will fit in	684	it just in case after a fork.
541	quite nicely into a call to C<pthread_atfork>:
542		685
		686	Example: Automate calling C<ev_loop_fork> on the default loop when
		687	using pthreads.
		688
		689	static void
		690	post_fork_child (void)
		691	{
		692	ev_loop_fork (EV_DEFAULT);
		693	}
		694
		695	...
543	pthread_atfork (0, 0, ev_default_fork);	696	pthread_atfork (0, 0, post_fork_child);
544
545	=item ev_loop_fork (loop)
546
547	Like C<ev_default_fork>, but acts on an event loop created by
548	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
549	after fork, and how you do this is entirely your own problem.
550		697
551	=item int ev_is_default_loop (loop)	698	=item int ev_is_default_loop (loop)
552		699
553	Returns true when the given loop actually is the default loop, false otherwise.	700	Returns true when the given loop is, in fact, the default loop, and false
		701	otherwise.
554		702
555	=item unsigned int ev_loop_count (loop)	703	=item unsigned int ev_iteration (loop)
556		704
557	Returns the count of loop iterations for the loop, which is identical to	705	Returns the current iteration count for the event loop, which is identical
558	the number of times libev did poll for new events. It starts at C<0> and	706	to the number of times libev did poll for new events. It starts at C<0>
559	happily wraps around with enough iterations.	707	and happily wraps around with enough iterations.
560		708
561	This value can sometimes be useful as a generation counter of sorts (it	709	This value can sometimes be useful as a generation counter of sorts (it
562	"ticks" the number of loop iterations), as it roughly corresponds with	710	"ticks" the number of loop iterations), as it roughly corresponds with
563	C<ev_prepare> and C<ev_check> calls.	711	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		712	prepare and check phases.
		713
		714	=item unsigned int ev_depth (loop)
		715
		716	Returns the number of times C<ev_run> was entered minus the number of
		717	times C<ev_run> was exited normally, in other words, the recursion depth.
		718
		719	Outside C<ev_run>, this number is zero. In a callback, this number is
		720	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		721	in which case it is higher.
		722
		723	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
		724	throwing an exception etc.), doesn't count as "exit" - consider this
		725	as a hint to avoid such ungentleman-like behaviour unless it's really
		726	convenient, in which case it is fully supported.
564		727
565	=item unsigned int ev_backend (loop)	728	=item unsigned int ev_backend (loop)
566		729
567	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	730	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
568	use.	731	use.
…		…
573	received events and started processing them. This timestamp does not	736	received events and started processing them. This timestamp does not
574	change as long as callbacks are being processed, and this is also the base	737	change as long as callbacks are being processed, and this is also the base
575	time used for relative timers. You can treat it as the timestamp of the	738	time used for relative timers. You can treat it as the timestamp of the
576	event occurring (or more correctly, libev finding out about it).	739	event occurring (or more correctly, libev finding out about it).
577		740
		741	=item ev_now_update (loop)
		742
		743	Establishes the current time by querying the kernel, updating the time
		744	returned by C<ev_now ()> in the progress. This is a costly operation and
		745	is usually done automatically within C<ev_run ()>.
		746
		747	This function is rarely useful, but when some event callback runs for a
		748	very long time without entering the event loop, updating libev's idea of
		749	the current time is a good idea.
		750
		751	See also L<The special problem of time updates> in the C<ev_timer> section.
		752
		753	=item ev_suspend (loop)
		754
		755	=item ev_resume (loop)
		756
		757	These two functions suspend and resume an event loop, for use when the
		758	loop is not used for a while and timeouts should not be processed.
		759
		760	A typical use case would be an interactive program such as a game: When
		761	the user presses C<^Z> to suspend the game and resumes it an hour later it
		762	would be best to handle timeouts as if no time had actually passed while
		763	the program was suspended. This can be achieved by calling C<ev_suspend>
		764	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		765	C<ev_resume> directly afterwards to resume timer processing.
		766
		767	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		768	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		769	will be rescheduled (that is, they will lose any events that would have
		770	occurred while suspended).
		771
		772	After calling C<ev_suspend> you B<must not> call I<any> function on the
		773	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		774	without a previous call to C<ev_suspend>.
		775
		776	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		777	event loop time (see C<ev_now_update>).
		778
578	=item ev_loop (loop, int flags)	779	=item ev_run (loop, int flags)
579		780
580	Finally, this is it, the event handler. This function usually is called	781	Finally, this is it, the event handler. This function usually is called
581	after you initialised all your watchers and you want to start handling	782	after you have initialised all your watchers and you want to start
582	events.	783	handling events. It will ask the operating system for any new events, call
		784	the watcher callbacks, an then repeat the whole process indefinitely: This
		785	is why event loops are called I<loops>.
583		786
584	If the flags argument is specified as C<0>, it will not return until	787	If the flags argument is specified as C<0>, it will keep handling events
585	either no event watchers are active anymore or C<ev_unloop> was called.	788	until either no event watchers are active anymore or C<ev_break> was
		789	called.
586		790
587	Please note that an explicit C<ev_unloop> is usually better than	791	Please note that an explicit C<ev_break> is usually better than
588	relying on all watchers to be stopped when deciding when a program has	792	relying on all watchers to be stopped when deciding when a program has
589	finished (especially in interactive programs), but having a program that	793	finished (especially in interactive programs), but having a program
590	automatically loops as long as it has to and no longer by virtue of	794	that automatically loops as long as it has to and no longer by virtue
591	relying on its watchers stopping correctly is a thing of beauty.	795	of relying on its watchers stopping correctly, that is truly a thing of
		796	beauty.
592		797
		798	This function is also I<mostly> exception-safe - you can break out of
		799	a C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		800	exception and so on. This does not decrement the C<ev_depth> value, nor
		801	will it clear any outstanding C<EVBREAK_ONE> breaks.
		802
593	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	803	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
594	those events and any outstanding ones, but will not block your process in	804	those events and any already outstanding ones, but will not wait and
595	case there are no events and will return after one iteration of the loop.	805	block your process in case there are no events and will return after one
		806	iteration of the loop. This is sometimes useful to poll and handle new
		807	events while doing lengthy calculations, to keep the program responsive.
596		808
597	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	809	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
598	necessary) and will handle those and any outstanding ones. It will block	810	necessary) and will handle those and any already outstanding ones. It
599	your process until at least one new event arrives, and will return after	811	will block your process until at least one new event arrives (which could
600	one iteration of the loop. This is useful if you are waiting for some	812	be an event internal to libev itself, so there is no guarantee that a
601	external event in conjunction with something not expressible using other	813	user-registered callback will be called), and will return after one
		814	iteration of the loop.
		815
		816	This is useful if you are waiting for some external event in conjunction
		817	with something not expressible using other libev watchers (i.e. "roll your
602	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	818	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
603	usually a better approach for this kind of thing.	819	usually a better approach for this kind of thing.
604		820
605	Here are the gory details of what C<ev_loop> does:	821	Here are the gory details of what C<ev_run> does:
606		822
		823	- Increment loop depth.
		824	- Reset the ev_break status.
607	- Before the first iteration, call any pending watchers.	825	- Before the first iteration, call any pending watchers.
		826	LOOP:
608	* If EVFLAG_FORKCHECK was used, check for a fork.	827	- If EVFLAG_FORKCHECK was used, check for a fork.
609	- If a fork was detected, queue and call all fork watchers.	828	- If a fork was detected (by any means), queue and call all fork watchers.
610	- Queue and call all prepare watchers.	829	- Queue and call all prepare watchers.
		830	- If ev_break was called, goto FINISH.
611	- If we have been forked, recreate the kernel state.	831	- If we have been forked, detach and recreate the kernel state
		832	as to not disturb the other process.
612	- Update the kernel state with all outstanding changes.	833	- Update the kernel state with all outstanding changes.
613	- Update the "event loop time".	834	- Update the "event loop time" (ev_now ()).
614	- Calculate for how long to sleep or block, if at all	835	- Calculate for how long to sleep or block, if at all
615	(active idle watchers, EVLOOP_NONBLOCK or not having	836	(active idle watchers, EVRUN_NOWAIT or not having
616	any active watchers at all will result in not sleeping).	837	any active watchers at all will result in not sleeping).
617	- Sleep if the I/O and timer collect interval say so.	838	- Sleep if the I/O and timer collect interval say so.
		839	- Increment loop iteration counter.
618	- Block the process, waiting for any events.	840	- Block the process, waiting for any events.
619	- Queue all outstanding I/O (fd) events.	841	- Queue all outstanding I/O (fd) events.
620	- Update the "event loop time" and do time jump handling.	842	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
621	- Queue all outstanding timers.	843	- Queue all expired timers.
622	- Queue all outstanding periodics.	844	- Queue all expired periodics.
623	- If no events are pending now, queue all idle watchers.	845	- Queue all idle watchers with priority higher than that of pending events.
624	- Queue all check watchers.	846	- Queue all check watchers.
625	- Call all queued watchers in reverse order (i.e. check watchers first).	847	- Call all queued watchers in reverse order (i.e. check watchers first).
626	Signals and child watchers are implemented as I/O watchers, and will	848	Signals and child watchers are implemented as I/O watchers, and will
627	be handled here by queueing them when their watcher gets executed.	849	be handled here by queueing them when their watcher gets executed.
628	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	850	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
629	were used, or there are no active watchers, return, otherwise	851	were used, or there are no active watchers, goto FINISH, otherwise
630	continue with step *.	852	continue with step LOOP.
		853	FINISH:
		854	- Reset the ev_break status iff it was EVBREAK_ONE.
		855	- Decrement the loop depth.
		856	- Return.
631		857
632	Example: Queue some jobs and then loop until no events are outstanding	858	Example: Queue some jobs and then loop until no events are outstanding
633	anymore.	859	anymore.
634		860
635	... queue jobs here, make sure they register event watchers as long	861	... queue jobs here, make sure they register event watchers as long
636	... as they still have work to do (even an idle watcher will do..)	862	... as they still have work to do (even an idle watcher will do..)
637	ev_loop (my_loop, 0);	863	ev_run (my_loop, 0);
638	... jobs done. yeah!	864	... jobs done or somebody called unloop. yeah!
639		865
640	=item ev_unloop (loop, how)	866	=item ev_break (loop, how)
641		867
642	Can be used to make a call to C<ev_loop> return early (but only after it	868	Can be used to make a call to C<ev_run> return early (but only after it
643	has processed all outstanding events). The C<how> argument must be either	869	has processed all outstanding events). The C<how> argument must be either
644	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	870	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
645	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	871	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
646		872
647	This "unloop state" will be cleared when entering C<ev_loop> again.	873	This "break state" will be cleared on the next call to C<ev_run>.
		874
		875	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		876	which case it will have no effect.
648		877
649	=item ev_ref (loop)	878	=item ev_ref (loop)
650		879
651	=item ev_unref (loop)	880	=item ev_unref (loop)
652		881
653	Ref/unref can be used to add or remove a reference count on the event	882	Ref/unref can be used to add or remove a reference count on the event
654	loop: Every watcher keeps one reference, and as long as the reference	883	loop: Every watcher keeps one reference, and as long as the reference
655	count is nonzero, C<ev_loop> will not return on its own. If you have	884	count is nonzero, C<ev_run> will not return on its own.
656	a watcher you never unregister that should not keep C<ev_loop> from	885
657	returning, ev_unref() after starting, and ev_ref() before stopping it. For	886	This is useful when you have a watcher that you never intend to
		887	unregister, but that nevertheless should not keep C<ev_run> from
		888	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		889	before stopping it.
		890
658	example, libev itself uses this for its internal signal pipe: It is not	891	As an example, libev itself uses this for its internal signal pipe: It
659	visible to the libev user and should not keep C<ev_loop> from exiting if	892	is not visible to the libev user and should not keep C<ev_run> from
660	no event watchers registered by it are active. It is also an excellent	893	exiting if no event watchers registered by it are active. It is also an
661	way to do this for generic recurring timers or from within third-party	894	excellent way to do this for generic recurring timers or from within
662	libraries. Just remember to I<unref after start> and I<ref before stop>	895	third-party libraries. Just remember to I<unref after start> and I<ref
663	(but only if the watcher wasn't active before, or was active before,	896	before stop> (but only if the watcher wasn't active before, or was active
664	respectively).	897	before, respectively. Note also that libev might stop watchers itself
		898	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		899	in the callback).
665		900
666	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	901	Example: Create a signal watcher, but keep it from keeping C<ev_run>
667	running when nothing else is active.	902	running when nothing else is active.
668		903
669	struct ev_signal exitsig;	904	ev_signal exitsig;
670	ev_signal_init (&exitsig, sig_cb, SIGINT);	905	ev_signal_init (&exitsig, sig_cb, SIGINT);
671	ev_signal_start (loop, &exitsig);	906	ev_signal_start (loop, &exitsig);
672	evf_unref (loop);	907	ev_unref (loop);
673		908
674	Example: For some weird reason, unregister the above signal handler again.	909	Example: For some weird reason, unregister the above signal handler again.
675		910
676	ev_ref (loop);	911	ev_ref (loop);
677	ev_signal_stop (loop, &exitsig);	912	ev_signal_stop (loop, &exitsig);
678		913
679	=item ev_set_io_collect_interval (loop, ev_tstamp interval)	914	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
680		915
681	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)	916	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
682		917
683	These advanced functions influence the time that libev will spend waiting	918	These advanced functions influence the time that libev will spend waiting
684	for events. Both are by default C<0>, meaning that libev will try to	919	for events. Both time intervals are by default C<0>, meaning that libev
685	invoke timer/periodic callbacks and I/O callbacks with minimum latency.	920	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		921	latency.
686		922
687	Setting these to a higher value (the C<interval> I<must> be >= C<0>)	923	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
688	allows libev to delay invocation of I/O and timer/periodic callbacks to	924	allows libev to delay invocation of I/O and timer/periodic callbacks
689	increase efficiency of loop iterations.	925	to increase efficiency of loop iterations (or to increase power-saving
		926	opportunities).
690		927
691	The background is that sometimes your program runs just fast enough to	928	The idea is that sometimes your program runs just fast enough to handle
692	handle one (or very few) event(s) per loop iteration. While this makes	929	one (or very few) event(s) per loop iteration. While this makes the
693	the program responsive, it also wastes a lot of CPU time to poll for new	930	program responsive, it also wastes a lot of CPU time to poll for new
694	events, especially with backends like C<select ()> which have a high	931	events, especially with backends like C<select ()> which have a high
695	overhead for the actual polling but can deliver many events at once.	932	overhead for the actual polling but can deliver many events at once.
696		933
697	By setting a higher I<io collect interval> you allow libev to spend more	934	By setting a higher I<io collect interval> you allow libev to spend more
698	time collecting I/O events, so you can handle more events per iteration,	935	time collecting I/O events, so you can handle more events per iteration,
699	at the cost of increasing latency. Timeouts (both C<ev_periodic> and	936	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
700	C<ev_timer>) will be not affected. Setting this to a non-null value will	937	C<ev_timer>) will be not affected. Setting this to a non-null value will
701	introduce an additional C<ev_sleep ()> call into most loop iterations.	938	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		939	sleep time ensures that libev will not poll for I/O events more often then
		940	once per this interval, on average.
702		941
703	Likewise, by setting a higher I<timeout collect interval> you allow libev	942	Likewise, by setting a higher I<timeout collect interval> you allow libev
704	to spend more time collecting timeouts, at the expense of increased	943	to spend more time collecting timeouts, at the expense of increased
705	latency (the watcher callback will be called later). C<ev_io> watchers	944	latency/jitter/inexactness (the watcher callback will be called
706	will not be affected. Setting this to a non-null value will not introduce	945	later). C<ev_io> watchers will not be affected. Setting this to a non-null
707	any overhead in libev.	946	value will not introduce any overhead in libev.
708		947
709	Many (busy) programs can usually benefit by setting the I/O collect	948	Many (busy) programs can usually benefit by setting the I/O collect
710	interval to a value near C<0.1> or so, which is often enough for	949	interval to a value near C<0.1> or so, which is often enough for
711	interactive servers (of course not for games), likewise for timeouts. It	950	interactive servers (of course not for games), likewise for timeouts. It
712	usually doesn't make much sense to set it to a lower value than C<0.01>,	951	usually doesn't make much sense to set it to a lower value than C<0.01>,
713	as this approaches the timing granularity of most systems.	952	as this approaches the timing granularity of most systems. Note that if
		953	you do transactions with the outside world and you can't increase the
		954	parallelity, then this setting will limit your transaction rate (if you
		955	need to poll once per transaction and the I/O collect interval is 0.01,
		956	then you can't do more than 100 transactions per second).
714		957
		958	Setting the I<timeout collect interval> can improve the opportunity for
		959	saving power, as the program will "bundle" timer callback invocations that
		960	are "near" in time together, by delaying some, thus reducing the number of
		961	times the process sleeps and wakes up again. Another useful technique to
		962	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		963	they fire on, say, one-second boundaries only.
		964
		965	Example: we only need 0.1s timeout granularity, and we wish not to poll
		966	more often than 100 times per second:
		967
		968	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		969	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		970
		971	=item ev_invoke_pending (loop)
		972
		973	This call will simply invoke all pending watchers while resetting their
		974	pending state. Normally, C<ev_run> does this automatically when required,
		975	but when overriding the invoke callback this call comes handy. This
		976	function can be invoked from a watcher - this can be useful for example
		977	when you want to do some lengthy calculation and want to pass further
		978	event handling to another thread (you still have to make sure only one
		979	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		980
		981	=item int ev_pending_count (loop)
		982
		983	Returns the number of pending watchers - zero indicates that no watchers
		984	are pending.
		985
		986	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		987
		988	This overrides the invoke pending functionality of the loop: Instead of
		989	invoking all pending watchers when there are any, C<ev_run> will call
		990	this callback instead. This is useful, for example, when you want to
		991	invoke the actual watchers inside another context (another thread etc.).
		992
		993	If you want to reset the callback, use C<ev_invoke_pending> as new
		994	callback.
		995
		996	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		997
		998	Sometimes you want to share the same loop between multiple threads. This
		999	can be done relatively simply by putting mutex_lock/unlock calls around
		1000	each call to a libev function.
		1001
		1002	However, C<ev_run> can run an indefinite time, so it is not feasible
		1003	to wait for it to return. One way around this is to wake up the event
		1004	loop via C<ev_break> and C<av_async_send>, another way is to set these
		1005	I<release> and I<acquire> callbacks on the loop.
		1006
		1007	When set, then C<release> will be called just before the thread is
		1008	suspended waiting for new events, and C<acquire> is called just
		1009	afterwards.
		1010
		1011	Ideally, C<release> will just call your mutex_unlock function, and
		1012	C<acquire> will just call the mutex_lock function again.
		1013
		1014	While event loop modifications are allowed between invocations of
		1015	C<release> and C<acquire> (that's their only purpose after all), no
		1016	modifications done will affect the event loop, i.e. adding watchers will
		1017	have no effect on the set of file descriptors being watched, or the time
		1018	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1019	to take note of any changes you made.
		1020
		1021	In theory, threads executing C<ev_run> will be async-cancel safe between
		1022	invocations of C<release> and C<acquire>.
		1023
		1024	See also the locking example in the C<THREADS> section later in this
		1025	document.
		1026
		1027	=item ev_set_userdata (loop, void *data)
		1028
		1029	=item void *ev_userdata (loop)
		1030
		1031	Set and retrieve a single C<void *> associated with a loop. When
		1032	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1033	C<0>.
		1034
		1035	These two functions can be used to associate arbitrary data with a loop,
		1036	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1037	C<acquire> callbacks described above, but of course can be (ab-)used for
		1038	any other purpose as well.
		1039
715	=item ev_loop_verify (loop)	1040	=item ev_verify (loop)
716		1041
717	This function only does something when C<EV_VERIFY> support has been	1042	This function only does something when C<EV_VERIFY> support has been
718	compiled in. It tries to go through all internal structures and checks	1043	compiled in, which is the default for non-minimal builds. It tries to go
719	them for validity. If anything is found to be inconsistent, it will print	1044	through all internal structures and checks them for validity. If anything
720	an error message to standard error and call C<abort ()>.	1045	is found to be inconsistent, it will print an error message to standard
		1046	error and call C<abort ()>.
721		1047
722	This can be used to catch bugs inside libev itself: under normal	1048	This can be used to catch bugs inside libev itself: under normal
723	circumstances, this function will never abort as of course libev keeps its	1049	circumstances, this function will never abort as of course libev keeps its
724	data structures consistent.	1050	data structures consistent.
725		1051
726	=back	1052	=back
727		1053
728		1054
729	=head1 ANATOMY OF A WATCHER	1055	=head1 ANATOMY OF A WATCHER
730		1056
		1057	In the following description, uppercase C<TYPE> in names stands for the
		1058	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1059	watchers and C<ev_io_start> for I/O watchers.
		1060
731	A watcher is a structure that you create and register to record your	1061	A watcher is an opaque structure that you allocate and register to record
732	interest in some event. For instance, if you want to wait for STDIN to	1062	your interest in some event. To make a concrete example, imagine you want
733	become readable, you would create an C<ev_io> watcher for that:	1063	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1064	for that:
734		1065
735	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1066	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
736	{	1067	{
737	ev_io_stop (w);	1068	ev_io_stop (w);
738	ev_unloop (loop, EVUNLOOP_ALL);	1069	ev_break (loop, EVBREAK_ALL);
739	}	1070	}
740		1071
741	struct ev_loop *loop = ev_default_loop (0);	1072	struct ev_loop *loop = ev_default_loop (0);
		1073
742	struct ev_io stdin_watcher;	1074	ev_io stdin_watcher;
		1075
743	ev_init (&stdin_watcher, my_cb);	1076	ev_init (&stdin_watcher, my_cb);
744	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1077	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
745	ev_io_start (loop, &stdin_watcher);	1078	ev_io_start (loop, &stdin_watcher);
		1079
746	ev_loop (loop, 0);	1080	ev_run (loop, 0);
747		1081
748	As you can see, you are responsible for allocating the memory for your	1082	As you can see, you are responsible for allocating the memory for your
749	watcher structures (and it is usually a bad idea to do this on the stack,	1083	watcher structures (and it is I<usually> a bad idea to do this on the
750	although this can sometimes be quite valid).	1084	stack).
751		1085
		1086	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1087	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1088
752	Each watcher structure must be initialised by a call to C<ev_init	1089	Each watcher structure must be initialised by a call to C<ev_init (watcher
753	(watcher *, callback)>, which expects a callback to be provided. This	1090	*, callback)>, which expects a callback to be provided. This callback is
754	callback gets invoked each time the event occurs (or, in the case of I/O	1091	invoked each time the event occurs (or, in the case of I/O watchers, each
755	watchers, each time the event loop detects that the file descriptor given	1092	time the event loop detects that the file descriptor given is readable
756	is readable and/or writable).	1093	and/or writable).
757		1094
758	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1095	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
759	with arguments specific to this watcher type. There is also a macro	1096	macro to configure it, with arguments specific to the watcher type. There
760	to combine initialisation and setting in one call: C<< ev_<type>_init	1097	is also a macro to combine initialisation and setting in one call: C<<
761	(watcher *, callback, ...) >>.	1098	ev_TYPE_init (watcher *, callback, ...) >>.
762		1099
763	To make the watcher actually watch out for events, you have to start it	1100	To make the watcher actually watch out for events, you have to start it
764	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1101	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
765	*) >>), and you can stop watching for events at any time by calling the	1102	*) >>), and you can stop watching for events at any time by calling the
766	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1103	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
767		1104
768	As long as your watcher is active (has been started but not stopped) you	1105	As long as your watcher is active (has been started but not stopped) you
769	must not touch the values stored in it. Most specifically you must never	1106	must not touch the values stored in it. Most specifically you must never
770	reinitialise it or call its C<set> macro.	1107	reinitialise it or call its C<ev_TYPE_set> macro.
771		1108
772	Each and every callback receives the event loop pointer as first, the	1109	Each and every callback receives the event loop pointer as first, the
773	registered watcher structure as second, and a bitset of received events as	1110	registered watcher structure as second, and a bitset of received events as
774	third argument.	1111	third argument.
775		1112
…		…
784	=item C<EV_WRITE>	1121	=item C<EV_WRITE>
785		1122
786	The file descriptor in the C<ev_io> watcher has become readable and/or	1123	The file descriptor in the C<ev_io> watcher has become readable and/or
787	writable.	1124	writable.
788		1125
789	=item C<EV_TIMEOUT>	1126	=item C<EV_TIMER>
790		1127
791	The C<ev_timer> watcher has timed out.	1128	The C<ev_timer> watcher has timed out.
792		1129
793	=item C<EV_PERIODIC>	1130	=item C<EV_PERIODIC>
794		1131
…		…
812		1149
813	=item C<EV_PREPARE>	1150	=item C<EV_PREPARE>
814		1151
815	=item C<EV_CHECK>	1152	=item C<EV_CHECK>
816		1153
817	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1154	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts
818	to gather new events, and all C<ev_check> watchers are invoked just after	1155	to gather new events, and all C<ev_check> watchers are invoked just after
819	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1156	C<ev_run> has gathered them, but before it invokes any callbacks for any
820	received events. Callbacks of both watcher types can start and stop as	1157	received events. Callbacks of both watcher types can start and stop as
821	many watchers as they want, and all of them will be taken into account	1158	many watchers as they want, and all of them will be taken into account
822	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1159	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
823	C<ev_loop> from blocking).	1160	C<ev_run> from blocking).
824		1161
825	=item C<EV_EMBED>	1162	=item C<EV_EMBED>
826		1163
827	The embedded event loop specified in the C<ev_embed> watcher needs attention.	1164	The embedded event loop specified in the C<ev_embed> watcher needs attention.
828		1165
829	=item C<EV_FORK>	1166	=item C<EV_FORK>
830		1167
831	The event loop has been resumed in the child process after fork (see	1168	The event loop has been resumed in the child process after fork (see
832	C<ev_fork>).	1169	C<ev_fork>).
833		1170
		1171	=item C<EV_CLEANUP>
		1172
		1173	The event loop is about to be destroyed (see C<ev_cleanup>).
		1174
834	=item C<EV_ASYNC>	1175	=item C<EV_ASYNC>
835		1176
836	The given async watcher has been asynchronously notified (see C<ev_async>).	1177	The given async watcher has been asynchronously notified (see C<ev_async>).
		1178
		1179	=item C<EV_CUSTOM>
		1180
		1181	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1182	by libev users to signal watchers (e.g. via C<ev_feed_event>).
837		1183
838	=item C<EV_ERROR>	1184	=item C<EV_ERROR>
839		1185
840	An unspecified error has occurred, the watcher has been stopped. This might	1186	An unspecified error has occurred, the watcher has been stopped. This might
841	happen because the watcher could not be properly started because libev	1187	happen because the watcher could not be properly started because libev
842	ran out of memory, a file descriptor was found to be closed or any other	1188	ran out of memory, a file descriptor was found to be closed or any other
		1189	problem. Libev considers these application bugs.
		1190
843	problem. You best act on it by reporting the problem and somehow coping	1191	You best act on it by reporting the problem and somehow coping with the
844	with the watcher being stopped.	1192	watcher being stopped. Note that well-written programs should not receive
		1193	an error ever, so when your watcher receives it, this usually indicates a
		1194	bug in your program.
845		1195
846	Libev will usually signal a few "dummy" events together with an error,	1196	Libev will usually signal a few "dummy" events together with an error, for
847	for example it might indicate that a fd is readable or writable, and if	1197	example it might indicate that a fd is readable or writable, and if your
848	your callbacks is well-written it can just attempt the operation and cope	1198	callbacks is well-written it can just attempt the operation and cope with
849	with the error from read() or write(). This will not work in multi-threaded	1199	the error from read() or write(). This will not work in multi-threaded
850	programs, though, so beware.	1200	programs, though, as the fd could already be closed and reused for another
		1201	thing, so beware.
851		1202
852	=back	1203	=back
853		1204
854	=head2 GENERIC WATCHER FUNCTIONS	1205	=head2 GENERIC WATCHER FUNCTIONS
855
856	In the following description, C<TYPE> stands for the watcher type,
857	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
858		1206
859	=over 4	1207	=over 4
860		1208
861	=item C<ev_init> (ev_TYPE *watcher, callback)	1209	=item C<ev_init> (ev_TYPE *watcher, callback)
862		1210
…		…
868	which rolls both calls into one.	1216	which rolls both calls into one.
869		1217
870	You can reinitialise a watcher at any time as long as it has been stopped	1218	You can reinitialise a watcher at any time as long as it has been stopped
871	(or never started) and there are no pending events outstanding.	1219	(or never started) and there are no pending events outstanding.
872		1220
873	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	1221	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
874	int revents)>.	1222	int revents)>.
875		1223
		1224	Example: Initialise an C<ev_io> watcher in two steps.
		1225
		1226	ev_io w;
		1227	ev_init (&w, my_cb);
		1228	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1229
876	=item C<ev_TYPE_set> (ev_TYPE *, [args])	1230	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
877		1231
878	This macro initialises the type-specific parts of a watcher. You need to	1232	This macro initialises the type-specific parts of a watcher. You need to
879	call C<ev_init> at least once before you call this macro, but you can	1233	call C<ev_init> at least once before you call this macro, but you can
880	call C<ev_TYPE_set> any number of times. You must not, however, call this	1234	call C<ev_TYPE_set> any number of times. You must not, however, call this
881	macro on a watcher that is active (it can be pending, however, which is a	1235	macro on a watcher that is active (it can be pending, however, which is a
882	difference to the C<ev_init> macro).	1236	difference to the C<ev_init> macro).
883		1237
884	Although some watcher types do not have type-specific arguments	1238	Although some watcher types do not have type-specific arguments
885	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	1239	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
886		1240
		1241	See C<ev_init>, above, for an example.
		1242
887	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	1243	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
888		1244
889	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	1245	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
890	calls into a single call. This is the most convenient method to initialise	1246	calls into a single call. This is the most convenient method to initialise
891	a watcher. The same limitations apply, of course.	1247	a watcher. The same limitations apply, of course.
892		1248
		1249	Example: Initialise and set an C<ev_io> watcher in one step.
		1250
		1251	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1252
893	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	1253	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
894		1254
895	Starts (activates) the given watcher. Only active watchers will receive	1255	Starts (activates) the given watcher. Only active watchers will receive
896	events. If the watcher is already active nothing will happen.	1256	events. If the watcher is already active nothing will happen.
897		1257
		1258	Example: Start the C<ev_io> watcher that is being abused as example in this
		1259	whole section.
		1260
		1261	ev_io_start (EV_DEFAULT_UC, &w);
		1262
898	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	1263	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
899		1264
900	Stops the given watcher again (if active) and clears the pending	1265	Stops the given watcher if active, and clears the pending status (whether
		1266	the watcher was active or not).
		1267
901	status. It is possible that stopped watchers are pending (for example,	1268	It is possible that stopped watchers are pending - for example,
902	non-repeating timers are being stopped when they become pending), but	1269	non-repeating timers are being stopped when they become pending - but
903	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1270	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
904	you want to free or reuse the memory used by the watcher it is therefore a	1271	pending. If you want to free or reuse the memory used by the watcher it is
905	good idea to always call its C<ev_TYPE_stop> function.	1272	therefore a good idea to always call its C<ev_TYPE_stop> function.
906		1273
907	=item bool ev_is_active (ev_TYPE *watcher)	1274	=item bool ev_is_active (ev_TYPE *watcher)
908		1275
909	Returns a true value iff the watcher is active (i.e. it has been started	1276	Returns a true value iff the watcher is active (i.e. it has been started
910	and not yet been stopped). As long as a watcher is active you must not modify	1277	and not yet been stopped). As long as a watcher is active you must not modify
…		…
926	=item ev_cb_set (ev_TYPE *watcher, callback)	1293	=item ev_cb_set (ev_TYPE *watcher, callback)
927		1294
928	Change the callback. You can change the callback at virtually any time	1295	Change the callback. You can change the callback at virtually any time
929	(modulo threads).	1296	(modulo threads).
930		1297
931	=item ev_set_priority (ev_TYPE *watcher, priority)	1298	=item ev_set_priority (ev_TYPE *watcher, int priority)
932		1299
933	=item int ev_priority (ev_TYPE *watcher)	1300	=item int ev_priority (ev_TYPE *watcher)
934		1301
935	Set and query the priority of the watcher. The priority is a small	1302	Set and query the priority of the watcher. The priority is a small
936	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>	1303	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
937	(default: C<-2>). Pending watchers with higher priority will be invoked	1304	(default: C<-2>). Pending watchers with higher priority will be invoked
938	before watchers with lower priority, but priority will not keep watchers	1305	before watchers with lower priority, but priority will not keep watchers
939	from being executed (except for C<ev_idle> watchers).	1306	from being executed (except for C<ev_idle> watchers).
940		1307
941	This means that priorities are I<only> used for ordering callback
942	invocation after new events have been received. This is useful, for
943	example, to reduce latency after idling, or more often, to bind two
944	watchers on the same event and make sure one is called first.
945
946	If you need to suppress invocation when higher priority events are pending	1308	If you need to suppress invocation when higher priority events are pending
947	you need to look at C<ev_idle> watchers, which provide this functionality.	1309	you need to look at C<ev_idle> watchers, which provide this functionality.
948		1310
949	You I<must not> change the priority of a watcher as long as it is active or	1311	You I<must not> change the priority of a watcher as long as it is active or
950	pending.	1312	pending.
951		1313
		1314	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1315	fine, as long as you do not mind that the priority value you query might
		1316	or might not have been clamped to the valid range.
		1317
952	The default priority used by watchers when no priority has been set is	1318	The default priority used by watchers when no priority has been set is
953	always C<0>, which is supposed to not be too high and not be too low :).	1319	always C<0>, which is supposed to not be too high and not be too low :).
954		1320
955	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is	1321	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
956	fine, as long as you do not mind that the priority value you query might	1322	priorities.
957	or might not have been adjusted to be within valid range.
958		1323
959	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1324	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
960		1325
961	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1326	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
962	C<loop> nor C<revents> need to be valid as long as the watcher callback	1327	C<loop> nor C<revents> need to be valid as long as the watcher callback
963	can deal with that fact.	1328	can deal with that fact, as both are simply passed through to the
		1329	callback.
964		1330
965	=item int ev_clear_pending (loop, ev_TYPE *watcher)	1331	=item int ev_clear_pending (loop, ev_TYPE *watcher)
966		1332
967	If the watcher is pending, this function returns clears its pending status	1333	If the watcher is pending, this function clears its pending status and
968	and returns its C<revents> bitset (as if its callback was invoked). If the	1334	returns its C<revents> bitset (as if its callback was invoked). If the
969	watcher isn't pending it does nothing and returns C<0>.	1335	watcher isn't pending it does nothing and returns C<0>.
970		1336
		1337	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1338	callback to be invoked, which can be accomplished with this function.
		1339
		1340	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1341
		1342	Feeds the given event set into the event loop, as if the specified event
		1343	had happened for the specified watcher (which must be a pointer to an
		1344	initialised but not necessarily started event watcher). Obviously you must
		1345	not free the watcher as long as it has pending events.
		1346
		1347	Stopping the watcher, letting libev invoke it, or calling
		1348	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1349	not started in the first place.
		1350
		1351	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1352	functions that do not need a watcher.
		1353
971	=back	1354	=back
972		1355
973
974	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1356	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
975		1357
976	Each watcher has, by default, a member C<void *data> that you can change	1358	Each watcher has, by default, a member C<void *data> that you can change
977	and read at any time, libev will completely ignore it. This can be used	1359	and read at any time: libev will completely ignore it. This can be used
978	to associate arbitrary data with your watcher. If you need more data and	1360	to associate arbitrary data with your watcher. If you need more data and
979	don't want to allocate memory and store a pointer to it in that data	1361	don't want to allocate memory and store a pointer to it in that data
980	member, you can also "subclass" the watcher type and provide your own	1362	member, you can also "subclass" the watcher type and provide your own
981	data:	1363	data:
982		1364
983	struct my_io	1365	struct my_io
984	{	1366	{
985	struct ev_io io;	1367	ev_io io;
986	int otherfd;	1368	int otherfd;
987	void *somedata;	1369	void *somedata;
988	struct whatever *mostinteresting;	1370	struct whatever *mostinteresting;
989	}	1371	};
		1372
		1373	...
		1374	struct my_io w;
		1375	ev_io_init (&w.io, my_cb, fd, EV_READ);
990		1376
991	And since your callback will be called with a pointer to the watcher, you	1377	And since your callback will be called with a pointer to the watcher, you
992	can cast it back to your own type:	1378	can cast it back to your own type:
993		1379
994	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1380	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
995	{	1381	{
996	struct my_io *w = (struct my_io *)w_;	1382	struct my_io *w = (struct my_io *)w_;
997	...	1383	...
998	}	1384	}
999		1385
1000	More interesting and less C-conformant ways of casting your callback type	1386	More interesting and less C-conformant ways of casting your callback type
1001	instead have been omitted.	1387	instead have been omitted.
1002		1388
1003	Another common scenario is having some data structure with multiple	1389	Another common scenario is to use some data structure with multiple
1004	watchers:	1390	embedded watchers:
1005		1391
1006	struct my_biggy	1392	struct my_biggy
1007	{	1393	{
1008	int some_data;	1394	int some_data;
1009	ev_timer t1;	1395	ev_timer t1;
1010	ev_timer t2;	1396	ev_timer t2;
1011	}	1397	}
1012		1398
1013	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1399	In this case getting the pointer to C<my_biggy> is a bit more
1014	you need to use C<offsetof>:	1400	complicated: Either you store the address of your C<my_biggy> struct
		1401	in the C<data> member of the watcher (for woozies), or you need to use
		1402	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1403	programmers):
1015		1404
1016	#include <stddef.h>	1405	#include <stddef.h>
1017		1406
1018	static void	1407	static void
1019	t1_cb (EV_P_ struct ev_timer *w, int revents)	1408	t1_cb (EV_P_ ev_timer *w, int revents)
1020	{	1409	{
1021	struct my_biggy big = (struct my_biggy *	1410	struct my_biggy big = (struct my_biggy *)
1022	(((char *)w) - offsetof (struct my_biggy, t1));	1411	(((char *)w) - offsetof (struct my_biggy, t1));
1023	}	1412	}
1024		1413
1025	static void	1414	static void
1026	t2_cb (EV_P_ struct ev_timer *w, int revents)	1415	t2_cb (EV_P_ ev_timer *w, int revents)
1027	{	1416	{
1028	struct my_biggy big = (struct my_biggy *	1417	struct my_biggy big = (struct my_biggy *)
1029	(((char *)w) - offsetof (struct my_biggy, t2));	1418	(((char *)w) - offsetof (struct my_biggy, t2));
1030	}	1419	}
		1420
		1421	=head2 WATCHER STATES
		1422
		1423	There are various watcher states mentioned throughout this manual -
		1424	active, pending and so on. In this section these states and the rules to
		1425	transition between them will be described in more detail - and while these
		1426	rules might look complicated, they usually do "the right thing".
		1427
		1428	=over 4
		1429
		1430	=item initialiased
		1431
		1432	Before a watcher can be registered with the event looop it has to be
		1433	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1434	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1435
		1436	In this state it is simply some block of memory that is suitable for use
		1437	in an event loop. It can be moved around, freed, reused etc. at will.
		1438
		1439	=item started/running/active
		1440
		1441	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1442	property of the event loop, and is actively waiting for events. While in
		1443	this state it cannot be accessed (except in a few documented ways), moved,
		1444	freed or anything else - the only legal thing is to keep a pointer to it,
		1445	and call libev functions on it that are documented to work on active watchers.
		1446
		1447	=item pending
		1448
		1449	If a watcher is active and libev determines that an event it is interested
		1450	in has occurred (such as a timer expiring), it will become pending. It will
		1451	stay in this pending state until either it is stopped or its callback is
		1452	about to be invoked, so it is not normally pending inside the watcher
		1453	callback.
		1454
		1455	The watcher might or might not be active while it is pending (for example,
		1456	an expired non-repeating timer can be pending but no longer active). If it
		1457	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1458	but it is still property of the event loop at this time, so cannot be
		1459	moved, freed or reused. And if it is active the rules described in the
		1460	previous item still apply.
		1461
		1462	It is also possible to feed an event on a watcher that is not active (e.g.
		1463	via C<ev_feed_event>), in which case it becomes pending without being
		1464	active.
		1465
		1466	=item stopped
		1467
		1468	A watcher can be stopped implicitly by libev (in which case it might still
		1469	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1470	latter will clear any pending state the watcher might be in, regardless
		1471	of whether it was active or not, so stopping a watcher explicitly before
		1472	freeing it is often a good idea.
		1473
		1474	While stopped (and not pending) the watcher is essentially in the
		1475	initialised state, that is it can be reused, moved, modified in any way
		1476	you wish.
		1477
		1478	=back
		1479
		1480	=head2 WATCHER PRIORITY MODELS
		1481
		1482	Many event loops support I<watcher priorities>, which are usually small
		1483	integers that influence the ordering of event callback invocation
		1484	between watchers in some way, all else being equal.
		1485
		1486	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1487	description for the more technical details such as the actual priority
		1488	range.
		1489
		1490	There are two common ways how these these priorities are being interpreted
		1491	by event loops:
		1492
		1493	In the more common lock-out model, higher priorities "lock out" invocation
		1494	of lower priority watchers, which means as long as higher priority
		1495	watchers receive events, lower priority watchers are not being invoked.
		1496
		1497	The less common only-for-ordering model uses priorities solely to order
		1498	callback invocation within a single event loop iteration: Higher priority
		1499	watchers are invoked before lower priority ones, but they all get invoked
		1500	before polling for new events.
		1501
		1502	Libev uses the second (only-for-ordering) model for all its watchers
		1503	except for idle watchers (which use the lock-out model).
		1504
		1505	The rationale behind this is that implementing the lock-out model for
		1506	watchers is not well supported by most kernel interfaces, and most event
		1507	libraries will just poll for the same events again and again as long as
		1508	their callbacks have not been executed, which is very inefficient in the
		1509	common case of one high-priority watcher locking out a mass of lower
		1510	priority ones.
		1511
		1512	Static (ordering) priorities are most useful when you have two or more
		1513	watchers handling the same resource: a typical usage example is having an
		1514	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1515	timeouts. Under load, data might be received while the program handles
		1516	other jobs, but since timers normally get invoked first, the timeout
		1517	handler will be executed before checking for data. In that case, giving
		1518	the timer a lower priority than the I/O watcher ensures that I/O will be
		1519	handled first even under adverse conditions (which is usually, but not
		1520	always, what you want).
		1521
		1522	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1523	will only be executed when no same or higher priority watchers have
		1524	received events, they can be used to implement the "lock-out" model when
		1525	required.
		1526
		1527	For example, to emulate how many other event libraries handle priorities,
		1528	you can associate an C<ev_idle> watcher to each such watcher, and in
		1529	the normal watcher callback, you just start the idle watcher. The real
		1530	processing is done in the idle watcher callback. This causes libev to
		1531	continuously poll and process kernel event data for the watcher, but when
		1532	the lock-out case is known to be rare (which in turn is rare :), this is
		1533	workable.
		1534
		1535	Usually, however, the lock-out model implemented that way will perform
		1536	miserably under the type of load it was designed to handle. In that case,
		1537	it might be preferable to stop the real watcher before starting the
		1538	idle watcher, so the kernel will not have to process the event in case
		1539	the actual processing will be delayed for considerable time.
		1540
		1541	Here is an example of an I/O watcher that should run at a strictly lower
		1542	priority than the default, and which should only process data when no
		1543	other events are pending:
		1544
		1545	ev_idle idle; // actual processing watcher
		1546	ev_io io; // actual event watcher
		1547
		1548	static void
		1549	io_cb (EV_P_ ev_io *w, int revents)
		1550	{
		1551	// stop the I/O watcher, we received the event, but
		1552	// are not yet ready to handle it.
		1553	ev_io_stop (EV_A_ w);
		1554
		1555	// start the idle watcher to handle the actual event.
		1556	// it will not be executed as long as other watchers
		1557	// with the default priority are receiving events.
		1558	ev_idle_start (EV_A_ &idle);
		1559	}
		1560
		1561	static void
		1562	idle_cb (EV_P_ ev_idle *w, int revents)
		1563	{
		1564	// actual processing
		1565	read (STDIN_FILENO, ...);
		1566
		1567	// have to start the I/O watcher again, as
		1568	// we have handled the event
		1569	ev_io_start (EV_P_ &io);
		1570	}
		1571
		1572	// initialisation
		1573	ev_idle_init (&idle, idle_cb);
		1574	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1575	ev_io_start (EV_DEFAULT_ &io);
		1576
		1577	In the "real" world, it might also be beneficial to start a timer, so that
		1578	low-priority connections can not be locked out forever under load. This
		1579	enables your program to keep a lower latency for important connections
		1580	during short periods of high load, while not completely locking out less
		1581	important ones.
1031		1582
1032		1583
1033	=head1 WATCHER TYPES	1584	=head1 WATCHER TYPES
1034		1585
1035	This section describes each watcher in detail, but will not repeat	1586	This section describes each watcher in detail, but will not repeat
…		…
1059	In general you can register as many read and/or write event watchers per	1610	In general you can register as many read and/or write event watchers per
1060	fd as you want (as long as you don't confuse yourself). Setting all file	1611	fd as you want (as long as you don't confuse yourself). Setting all file
1061	descriptors to non-blocking mode is also usually a good idea (but not	1612	descriptors to non-blocking mode is also usually a good idea (but not
1062	required if you know what you are doing).	1613	required if you know what you are doing).
1063		1614
1064	If you must do this, then force the use of a known-to-be-good backend	1615	If you cannot use non-blocking mode, then force the use of a
1065	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	1616	known-to-be-good backend (at the time of this writing, this includes only
1066	C<EVBACKEND_POLL>).	1617	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
		1618	descriptors for which non-blocking operation makes no sense (such as
		1619	files) - libev doesn't guarantee any specific behaviour in that case.
1067		1620
1068	Another thing you have to watch out for is that it is quite easy to	1621	Another thing you have to watch out for is that it is quite easy to
1069	receive "spurious" readiness notifications, that is your callback might	1622	receive "spurious" readiness notifications, that is your callback might
1070	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1623	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1071	because there is no data. Not only are some backends known to create a	1624	because there is no data. Not only are some backends known to create a
1072	lot of those (for example Solaris ports), it is very easy to get into	1625	lot of those (for example Solaris ports), it is very easy to get into
1073	this situation even with a relatively standard program structure. Thus	1626	this situation even with a relatively standard program structure. Thus
1074	it is best to always use non-blocking I/O: An extra C<read>(2) returning	1627	it is best to always use non-blocking I/O: An extra C<read>(2) returning
1075	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1628	C<EAGAIN> is far preferable to a program hanging until some data arrives.
1076		1629
1077	If you cannot run the fd in non-blocking mode (for example you should not	1630	If you cannot run the fd in non-blocking mode (for example you should
1078	play around with an Xlib connection), then you have to separately re-test	1631	not play around with an Xlib connection), then you have to separately
1079	whether a file descriptor is really ready with a known-to-be good interface	1632	re-test whether a file descriptor is really ready with a known-to-be good
1080	such as poll (fortunately in our Xlib example, Xlib already does this on	1633	interface such as poll (fortunately in our Xlib example, Xlib already
1081	its own, so its quite safe to use).	1634	does this on its own, so its quite safe to use). Some people additionally
		1635	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1636	indefinitely.
		1637
		1638	But really, best use non-blocking mode.
1082		1639
1083	=head3 The special problem of disappearing file descriptors	1640	=head3 The special problem of disappearing file descriptors
1084		1641
1085	Some backends (e.g. kqueue, epoll) need to be told about closing a file	1642	Some backends (e.g. kqueue, epoll) need to be told about closing a file
1086	descriptor (either by calling C<close> explicitly or by any other means,	1643	descriptor (either due to calling C<close> explicitly or any other means,
1087	such as C<dup>). The reason is that you register interest in some file	1644	such as C<dup2>). The reason is that you register interest in some file
1088	descriptor, but when it goes away, the operating system will silently drop	1645	descriptor, but when it goes away, the operating system will silently drop
1089	this interest. If another file descriptor with the same number then is	1646	this interest. If another file descriptor with the same number then is
1090	registered with libev, there is no efficient way to see that this is, in	1647	registered with libev, there is no efficient way to see that this is, in
1091	fact, a different file descriptor.	1648	fact, a different file descriptor.
1092		1649
…		…
1123	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or	1680	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
1124	C<EVBACKEND_POLL>.	1681	C<EVBACKEND_POLL>.
1125		1682
1126	=head3 The special problem of SIGPIPE	1683	=head3 The special problem of SIGPIPE
1127		1684
1128	While not really specific to libev, it is easy to forget about SIGPIPE:	1685	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1129	when reading from a pipe whose other end has been closed, your program	1686	when writing to a pipe whose other end has been closed, your program gets
1130	gets send a SIGPIPE, which, by default, aborts your program. For most	1687	sent a SIGPIPE, which, by default, aborts your program. For most programs
1131	programs this is sensible behaviour, for daemons, this is usually	1688	this is sensible behaviour, for daemons, this is usually undesirable.
1132	undesirable.
1133		1689
1134	So when you encounter spurious, unexplained daemon exits, make sure you	1690	So when you encounter spurious, unexplained daemon exits, make sure you
1135	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon	1691	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
1136	somewhere, as that would have given you a big clue).	1692	somewhere, as that would have given you a big clue).
1137		1693
		1694	=head3 The special problem of accept()ing when you can't
		1695
		1696	Many implementations of the POSIX C<accept> function (for example,
		1697	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1698	connection from the pending queue in all error cases.
		1699
		1700	For example, larger servers often run out of file descriptors (because
		1701	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1702	rejecting the connection, leading to libev signalling readiness on
		1703	the next iteration again (the connection still exists after all), and
		1704	typically causing the program to loop at 100% CPU usage.
		1705
		1706	Unfortunately, the set of errors that cause this issue differs between
		1707	operating systems, there is usually little the app can do to remedy the
		1708	situation, and no known thread-safe method of removing the connection to
		1709	cope with overload is known (to me).
		1710
		1711	One of the easiest ways to handle this situation is to just ignore it
		1712	- when the program encounters an overload, it will just loop until the
		1713	situation is over. While this is a form of busy waiting, no OS offers an
		1714	event-based way to handle this situation, so it's the best one can do.
		1715
		1716	A better way to handle the situation is to log any errors other than
		1717	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1718	messages, and continue as usual, which at least gives the user an idea of
		1719	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1720	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1721	usage.
		1722
		1723	If your program is single-threaded, then you could also keep a dummy file
		1724	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1725	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1726	close that fd, and create a new dummy fd. This will gracefully refuse
		1727	clients under typical overload conditions.
		1728
		1729	The last way to handle it is to simply log the error and C<exit>, as
		1730	is often done with C<malloc> failures, but this results in an easy
		1731	opportunity for a DoS attack.
1138		1732
1139	=head3 Watcher-Specific Functions	1733	=head3 Watcher-Specific Functions
1140		1734
1141	=over 4	1735	=over 4
1142		1736
1143	=item ev_io_init (ev_io *, callback, int fd, int events)	1737	=item ev_io_init (ev_io *, callback, int fd, int events)
1144		1738
1145	=item ev_io_set (ev_io *, int fd, int events)	1739	=item ev_io_set (ev_io *, int fd, int events)
1146		1740
1147	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1741	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
1148	receive events for and events is either C<EV_READ>, C<EV_WRITE> or	1742	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
1149	C<EV_READ | EV_WRITE> to receive the given events.	1743	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
1150		1744
1151	=item int fd [read-only]	1745	=item int fd [read-only]
1152		1746
1153	The file descriptor being watched.	1747	The file descriptor being watched.
1154		1748
…		…
1162		1756
1163	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1757	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1164	readable, but only once. Since it is likely line-buffered, you could	1758	readable, but only once. Since it is likely line-buffered, you could
1165	attempt to read a whole line in the callback.	1759	attempt to read a whole line in the callback.
1166		1760
1167	static void	1761	static void
1168	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1762	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
1169	{	1763	{
1170	ev_io_stop (loop, w);	1764	ev_io_stop (loop, w);
1171	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1765	.. read from stdin here (or from w->fd) and handle any I/O errors
1172	}	1766	}
1173		1767
1174	...	1768	...
1175	struct ev_loop *loop = ev_default_init (0);	1769	struct ev_loop *loop = ev_default_init (0);
1176	struct ev_io stdin_readable;	1770	ev_io stdin_readable;
1177	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1771	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1178	ev_io_start (loop, &stdin_readable);	1772	ev_io_start (loop, &stdin_readable);
1179	ev_loop (loop, 0);	1773	ev_run (loop, 0);
1180		1774
1181		1775
1182	=head2 C<ev_timer> - relative and optionally repeating timeouts	1776	=head2 C<ev_timer> - relative and optionally repeating timeouts
1183		1777
1184	Timer watchers are simple relative timers that generate an event after a	1778	Timer watchers are simple relative timers that generate an event after a
1185	given time, and optionally repeating in regular intervals after that.	1779	given time, and optionally repeating in regular intervals after that.
1186		1780
1187	The timers are based on real time, that is, if you register an event that	1781	The timers are based on real time, that is, if you register an event that
1188	times out after an hour and you reset your system clock to January last	1782	times out after an hour and you reset your system clock to January last
1189	year, it will still time out after (roughly) and hour. "Roughly" because	1783	year, it will still time out after (roughly) one hour. "Roughly" because
1190	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1784	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1191	monotonic clock option helps a lot here).	1785	monotonic clock option helps a lot here).
		1786
		1787	The callback is guaranteed to be invoked only I<after> its timeout has
		1788	passed (not I<at>, so on systems with very low-resolution clocks this
		1789	might introduce a small delay). If multiple timers become ready during the
		1790	same loop iteration then the ones with earlier time-out values are invoked
		1791	before ones of the same priority with later time-out values (but this is
		1792	no longer true when a callback calls C<ev_run> recursively).
		1793
		1794	=head3 Be smart about timeouts
		1795
		1796	Many real-world problems involve some kind of timeout, usually for error
		1797	recovery. A typical example is an HTTP request - if the other side hangs,
		1798	you want to raise some error after a while.
		1799
		1800	What follows are some ways to handle this problem, from obvious and
		1801	inefficient to smart and efficient.
		1802
		1803	In the following, a 60 second activity timeout is assumed - a timeout that
		1804	gets reset to 60 seconds each time there is activity (e.g. each time some
		1805	data or other life sign was received).
		1806
		1807	=over 4
		1808
		1809	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1810
		1811	This is the most obvious, but not the most simple way: In the beginning,
		1812	start the watcher:
		1813
		1814	ev_timer_init (timer, callback, 60., 0.);
		1815	ev_timer_start (loop, timer);
		1816
		1817	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1818	and start it again:
		1819
		1820	ev_timer_stop (loop, timer);
		1821	ev_timer_set (timer, 60., 0.);
		1822	ev_timer_start (loop, timer);
		1823
		1824	This is relatively simple to implement, but means that each time there is
		1825	some activity, libev will first have to remove the timer from its internal
		1826	data structure and then add it again. Libev tries to be fast, but it's
		1827	still not a constant-time operation.
		1828
		1829	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1830
		1831	This is the easiest way, and involves using C<ev_timer_again> instead of
		1832	C<ev_timer_start>.
		1833
		1834	To implement this, configure an C<ev_timer> with a C<repeat> value
		1835	of C<60> and then call C<ev_timer_again> at start and each time you
		1836	successfully read or write some data. If you go into an idle state where
		1837	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1838	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1839
		1840	That means you can ignore both the C<ev_timer_start> function and the
		1841	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1842	member and C<ev_timer_again>.
		1843
		1844	At start:
		1845
		1846	ev_init (timer, callback);
		1847	timer->repeat = 60.;
		1848	ev_timer_again (loop, timer);
		1849
		1850	Each time there is some activity:
		1851
		1852	ev_timer_again (loop, timer);
		1853
		1854	It is even possible to change the time-out on the fly, regardless of
		1855	whether the watcher is active or not:
		1856
		1857	timer->repeat = 30.;
		1858	ev_timer_again (loop, timer);
		1859
		1860	This is slightly more efficient then stopping/starting the timer each time
		1861	you want to modify its timeout value, as libev does not have to completely
		1862	remove and re-insert the timer from/into its internal data structure.
		1863
		1864	It is, however, even simpler than the "obvious" way to do it.
		1865
		1866	=item 3. Let the timer time out, but then re-arm it as required.
		1867
		1868	This method is more tricky, but usually most efficient: Most timeouts are
		1869	relatively long compared to the intervals between other activity - in
		1870	our example, within 60 seconds, there are usually many I/O events with
		1871	associated activity resets.
		1872
		1873	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1874	but remember the time of last activity, and check for a real timeout only
		1875	within the callback:
		1876
		1877	ev_tstamp last_activity; // time of last activity
		1878
		1879	static void
		1880	callback (EV_P_ ev_timer *w, int revents)
		1881	{
		1882	ev_tstamp now = ev_now (EV_A);
		1883	ev_tstamp timeout = last_activity + 60.;
		1884
		1885	// if last_activity + 60. is older than now, we did time out
		1886	if (timeout < now)
		1887	{
		1888	// timeout occurred, take action
		1889	}
		1890	else
		1891	{
		1892	// callback was invoked, but there was some activity, re-arm
		1893	// the watcher to fire in last_activity + 60, which is
		1894	// guaranteed to be in the future, so "again" is positive:
		1895	w->repeat = timeout - now;
		1896	ev_timer_again (EV_A_ w);
		1897	}
		1898	}
		1899
		1900	To summarise the callback: first calculate the real timeout (defined
		1901	as "60 seconds after the last activity"), then check if that time has
		1902	been reached, which means something I<did>, in fact, time out. Otherwise
		1903	the callback was invoked too early (C<timeout> is in the future), so
		1904	re-schedule the timer to fire at that future time, to see if maybe we have
		1905	a timeout then.
		1906
		1907	Note how C<ev_timer_again> is used, taking advantage of the
		1908	C<ev_timer_again> optimisation when the timer is already running.
		1909
		1910	This scheme causes more callback invocations (about one every 60 seconds
		1911	minus half the average time between activity), but virtually no calls to
		1912	libev to change the timeout.
		1913
		1914	To start the timer, simply initialise the watcher and set C<last_activity>
		1915	to the current time (meaning we just have some activity :), then call the
		1916	callback, which will "do the right thing" and start the timer:
		1917
		1918	ev_init (timer, callback);
		1919	last_activity = ev_now (loop);
		1920	callback (loop, timer, EV_TIMER);
		1921
		1922	And when there is some activity, simply store the current time in
		1923	C<last_activity>, no libev calls at all:
		1924
		1925	last_activity = ev_now (loop);
		1926
		1927	This technique is slightly more complex, but in most cases where the
		1928	time-out is unlikely to be triggered, much more efficient.
		1929
		1930	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1931	callback :) - just change the timeout and invoke the callback, which will
		1932	fix things for you.
		1933
		1934	=item 4. Wee, just use a double-linked list for your timeouts.
		1935
		1936	If there is not one request, but many thousands (millions...), all
		1937	employing some kind of timeout with the same timeout value, then one can
		1938	do even better:
		1939
		1940	When starting the timeout, calculate the timeout value and put the timeout
		1941	at the I<end> of the list.
		1942
		1943	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1944	the list is expected to fire (for example, using the technique #3).
		1945
		1946	When there is some activity, remove the timer from the list, recalculate
		1947	the timeout, append it to the end of the list again, and make sure to
		1948	update the C<ev_timer> if it was taken from the beginning of the list.
		1949
		1950	This way, one can manage an unlimited number of timeouts in O(1) time for
		1951	starting, stopping and updating the timers, at the expense of a major
		1952	complication, and having to use a constant timeout. The constant timeout
		1953	ensures that the list stays sorted.
		1954
		1955	=back
		1956
		1957	So which method the best?
		1958
		1959	Method #2 is a simple no-brain-required solution that is adequate in most
		1960	situations. Method #3 requires a bit more thinking, but handles many cases
		1961	better, and isn't very complicated either. In most case, choosing either
		1962	one is fine, with #3 being better in typical situations.
		1963
		1964	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1965	rather complicated, but extremely efficient, something that really pays
		1966	off after the first million or so of active timers, i.e. it's usually
		1967	overkill :)
		1968
		1969	=head3 The special problem of time updates
		1970
		1971	Establishing the current time is a costly operation (it usually takes at
		1972	least two system calls): EV therefore updates its idea of the current
		1973	time only before and after C<ev_run> collects new events, which causes a
		1974	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1975	lots of events in one iteration.
1192		1976
1193	The relative timeouts are calculated relative to the C<ev_now ()>	1977	The relative timeouts are calculated relative to the C<ev_now ()>
1194	time. This is usually the right thing as this timestamp refers to the time	1978	time. This is usually the right thing as this timestamp refers to the time
1195	of the event triggering whatever timeout you are modifying/starting. If	1979	of the event triggering whatever timeout you are modifying/starting. If
1196	you suspect event processing to be delayed and you I<need> to base the timeout	1980	you suspect event processing to be delayed and you I<need> to base the
1197	on the current time, use something like this to adjust for this:	1981	timeout on the current time, use something like this to adjust for this:
1198		1982
1199	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1983	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1200		1984
1201	The callback is guaranteed to be invoked only after its timeout has passed,	1985	If the event loop is suspended for a long time, you can also force an
1202	but if multiple timers become ready during the same loop iteration then	1986	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1203	order of execution is undefined.	1987	()>.
		1988
		1989	=head3 The special problems of suspended animation
		1990
		1991	When you leave the server world it is quite customary to hit machines that
		1992	can suspend/hibernate - what happens to the clocks during such a suspend?
		1993
		1994	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		1995	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		1996	to run until the system is suspended, but they will not advance while the
		1997	system is suspended. That means, on resume, it will be as if the program
		1998	was frozen for a few seconds, but the suspend time will not be counted
		1999	towards C<ev_timer> when a monotonic clock source is used. The real time
		2000	clock advanced as expected, but if it is used as sole clocksource, then a
		2001	long suspend would be detected as a time jump by libev, and timers would
		2002	be adjusted accordingly.
		2003
		2004	I would not be surprised to see different behaviour in different between
		2005	operating systems, OS versions or even different hardware.
		2006
		2007	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2008	time jump in the monotonic clocks and the realtime clock. If the program
		2009	is suspended for a very long time, and monotonic clock sources are in use,
		2010	then you can expect C<ev_timer>s to expire as the full suspension time
		2011	will be counted towards the timers. When no monotonic clock source is in
		2012	use, then libev will again assume a timejump and adjust accordingly.
		2013
		2014	It might be beneficial for this latter case to call C<ev_suspend>
		2015	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2016	deterministic behaviour in this case (you can do nothing against
		2017	C<SIGSTOP>).
1204		2018
1205	=head3 Watcher-Specific Functions and Data Members	2019	=head3 Watcher-Specific Functions and Data Members
1206		2020
1207	=over 4	2021	=over 4
1208		2022
…		…
1232	If the timer is started but non-repeating, stop it (as if it timed out).	2046	If the timer is started but non-repeating, stop it (as if it timed out).
1233		2047
1234	If the timer is repeating, either start it if necessary (with the	2048	If the timer is repeating, either start it if necessary (with the
1235	C<repeat> value), or reset the running timer to the C<repeat> value.	2049	C<repeat> value), or reset the running timer to the C<repeat> value.
1236		2050
1237	This sounds a bit complicated, but here is a useful and typical	2051	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
1238	example: Imagine you have a TCP connection and you want a so-called idle	2052	usage example.
1239	timeout, that is, you want to be called when there have been, say, 60
1240	seconds of inactivity on the socket. The easiest way to do this is to
1241	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
1242	C<ev_timer_again> each time you successfully read or write some data. If
1243	you go into an idle state where you do not expect data to travel on the
1244	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
1245	automatically restart it if need be.
1246		2053
1247	That means you can ignore the C<after> value and C<ev_timer_start>	2054	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
1248	altogether and only ever use the C<repeat> value and C<ev_timer_again>:
1249		2055
1250	ev_timer_init (timer, callback, 0., 5.);	2056	Returns the remaining time until a timer fires. If the timer is active,
1251	ev_timer_again (loop, timer);	2057	then this time is relative to the current event loop time, otherwise it's
1252	...	2058	the timeout value currently configured.
1253	timer->again = 17.;
1254	ev_timer_again (loop, timer);
1255	...
1256	timer->again = 10.;
1257	ev_timer_again (loop, timer);
1258		2059
1259	This is more slightly efficient then stopping/starting the timer each time	2060	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
1260	you want to modify its timeout value.	2061	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2062	will return C<4>. When the timer expires and is restarted, it will return
		2063	roughly C<7> (likely slightly less as callback invocation takes some time,
		2064	too), and so on.
1261		2065
1262	=item ev_tstamp repeat [read-write]	2066	=item ev_tstamp repeat [read-write]
1263		2067
1264	The current C<repeat> value. Will be used each time the watcher times out	2068	The current C<repeat> value. Will be used each time the watcher times out
1265	or C<ev_timer_again> is called and determines the next timeout (if any),	2069	or C<ev_timer_again> is called, and determines the next timeout (if any),
1266	which is also when any modifications are taken into account.	2070	which is also when any modifications are taken into account.
1267		2071
1268	=back	2072	=back
1269		2073
1270	=head3 Examples	2074	=head3 Examples
1271		2075
1272	Example: Create a timer that fires after 60 seconds.	2076	Example: Create a timer that fires after 60 seconds.
1273		2077
1274	static void	2078	static void
1275	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2079	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1276	{	2080	{
1277	.. one minute over, w is actually stopped right here	2081	.. one minute over, w is actually stopped right here
1278	}	2082	}
1279		2083
1280	struct ev_timer mytimer;	2084	ev_timer mytimer;
1281	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	2085	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1282	ev_timer_start (loop, &mytimer);	2086	ev_timer_start (loop, &mytimer);
1283		2087
1284	Example: Create a timeout timer that times out after 10 seconds of	2088	Example: Create a timeout timer that times out after 10 seconds of
1285	inactivity.	2089	inactivity.
1286		2090
1287	static void	2091	static void
1288	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2092	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1289	{	2093	{
1290	.. ten seconds without any activity	2094	.. ten seconds without any activity
1291	}	2095	}
1292		2096
1293	struct ev_timer mytimer;	2097	ev_timer mytimer;
1294	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	2098	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1295	ev_timer_again (&mytimer); /* start timer */	2099	ev_timer_again (&mytimer); /* start timer */
1296	ev_loop (loop, 0);	2100	ev_run (loop, 0);
1297		2101
1298	// and in some piece of code that gets executed on any "activity":	2102	// and in some piece of code that gets executed on any "activity":
1299	// reset the timeout to start ticking again at 10 seconds	2103	// reset the timeout to start ticking again at 10 seconds
1300	ev_timer_again (&mytimer);	2104	ev_timer_again (&mytimer);
1301		2105
1302		2106
1303	=head2 C<ev_periodic> - to cron or not to cron?	2107	=head2 C<ev_periodic> - to cron or not to cron?
1304		2108
1305	Periodic watchers are also timers of a kind, but they are very versatile	2109	Periodic watchers are also timers of a kind, but they are very versatile
1306	(and unfortunately a bit complex).	2110	(and unfortunately a bit complex).
1307		2111
1308	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2112	Unlike C<ev_timer>, periodic watchers are not based on real time (or
1309	but on wall clock time (absolute time). You can tell a periodic watcher	2113	relative time, the physical time that passes) but on wall clock time
1310	to trigger after some specific point in time. For example, if you tell a	2114	(absolute time, the thing you can read on your calender or clock). The
1311	periodic watcher to trigger in 10 seconds (by specifying e.g. C<ev_now ()	2115	difference is that wall clock time can run faster or slower than real
1312	+ 10.>, that is, an absolute time not a delay) and then reset your system	2116	time, and time jumps are not uncommon (e.g. when you adjust your
1313	clock to January of the previous year, then it will take more than year	2117	wrist-watch).
1314	to trigger the event (unlike an C<ev_timer>, which would still trigger
1315	roughly 10 seconds later as it uses a relative timeout).
1316		2118
		2119	You can tell a periodic watcher to trigger after some specific point
		2120	in time: for example, if you tell a periodic watcher to trigger "in 10
		2121	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2122	not a delay) and then reset your system clock to January of the previous
		2123	year, then it will take a year or more to trigger the event (unlike an
		2124	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2125	it, as it uses a relative timeout).
		2126
1317	C<ev_periodic>s can also be used to implement vastly more complex timers,	2127	C<ev_periodic> watchers can also be used to implement vastly more complex
1318	such as triggering an event on each "midnight, local time", or other	2128	timers, such as triggering an event on each "midnight, local time", or
1319	complicated, rules.	2129	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2130	those cannot react to time jumps.
1320		2131
1321	As with timers, the callback is guaranteed to be invoked only when the	2132	As with timers, the callback is guaranteed to be invoked only when the
1322	time (C<at>) has passed, but if multiple periodic timers become ready	2133	point in time where it is supposed to trigger has passed. If multiple
1323	during the same loop iteration then order of execution is undefined.	2134	timers become ready during the same loop iteration then the ones with
		2135	earlier time-out values are invoked before ones with later time-out values
		2136	(but this is no longer true when a callback calls C<ev_run> recursively).
1324		2137
1325	=head3 Watcher-Specific Functions and Data Members	2138	=head3 Watcher-Specific Functions and Data Members
1326		2139
1327	=over 4	2140	=over 4
1328		2141
1329	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2142	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1330		2143
1331	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2144	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1332		2145
1333	Lots of arguments, lets sort it out... There are basically three modes of	2146	Lots of arguments, let's sort it out... There are basically three modes of
1334	operation, and we will explain them from simplest to complex:	2147	operation, and we will explain them from simplest to most complex:
1335		2148
1336	=over 4	2149	=over 4
1337		2150
1338	=item * absolute timer (at = time, interval = reschedule_cb = 0)	2151	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
1339		2152
1340	In this configuration the watcher triggers an event after the wall clock	2153	In this configuration the watcher triggers an event after the wall clock
1341	time C<at> has passed and doesn't repeat. It will not adjust when a time	2154	time C<offset> has passed. It will not repeat and will not adjust when a
1342	jump occurs, that is, if it is to be run at January 1st 2011 then it will	2155	time jump occurs, that is, if it is to be run at January 1st 2011 then it
1343	run when the system time reaches or surpasses this time.	2156	will be stopped and invoked when the system clock reaches or surpasses
		2157	this point in time.
1344		2158
1345	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	2159	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
1346		2160
1347	In this mode the watcher will always be scheduled to time out at the next	2161	In this mode the watcher will always be scheduled to time out at the next
1348	C<at + N * interval> time (for some integer N, which can also be negative)	2162	C<offset + N * interval> time (for some integer N, which can also be
1349	and then repeat, regardless of any time jumps.	2163	negative) and then repeat, regardless of any time jumps. The C<offset>
		2164	argument is merely an offset into the C<interval> periods.
1350		2165
1351	This can be used to create timers that do not drift with respect to system	2166	This can be used to create timers that do not drift with respect to the
1352	time, for example, here is a C<ev_periodic> that triggers each hour, on	2167	system clock, for example, here is an C<ev_periodic> that triggers each
1353	the hour:	2168	hour, on the hour (with respect to UTC):
1354		2169
1355	ev_periodic_set (&periodic, 0., 3600., 0);	2170	ev_periodic_set (&periodic, 0., 3600., 0);
1356		2171
1357	This doesn't mean there will always be 3600 seconds in between triggers,	2172	This doesn't mean there will always be 3600 seconds in between triggers,
1358	but only that the callback will be called when the system time shows a	2173	but only that the callback will be called when the system time shows a
1359	full hour (UTC), or more correctly, when the system time is evenly divisible	2174	full hour (UTC), or more correctly, when the system time is evenly divisible
1360	by 3600.	2175	by 3600.
1361		2176
1362	Another way to think about it (for the mathematically inclined) is that	2177	Another way to think about it (for the mathematically inclined) is that
1363	C<ev_periodic> will try to run the callback in this mode at the next possible	2178	C<ev_periodic> will try to run the callback in this mode at the next possible
1364	time where C<time = at (mod interval)>, regardless of any time jumps.	2179	time where C<time = offset (mod interval)>, regardless of any time jumps.
1365		2180
1366	For numerical stability it is preferable that the C<at> value is near	2181	For numerical stability it is preferable that the C<offset> value is near
1367	C<ev_now ()> (the current time), but there is no range requirement for	2182	C<ev_now ()> (the current time), but there is no range requirement for
1368	this value, and in fact is often specified as zero.	2183	this value, and in fact is often specified as zero.
1369		2184
1370	Note also that there is an upper limit to how often a timer can fire (CPU	2185	Note also that there is an upper limit to how often a timer can fire (CPU
1371	speed for example), so if C<interval> is very small then timing stability	2186	speed for example), so if C<interval> is very small then timing stability
1372	will of course deteriorate. Libev itself tries to be exact to be about one	2187	will of course deteriorate. Libev itself tries to be exact to be about one
1373	millisecond (if the OS supports it and the machine is fast enough).	2188	millisecond (if the OS supports it and the machine is fast enough).
1374		2189
1375	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)	2190	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
1376		2191
1377	In this mode the values for C<interval> and C<at> are both being	2192	In this mode the values for C<interval> and C<offset> are both being
1378	ignored. Instead, each time the periodic watcher gets scheduled, the	2193	ignored. Instead, each time the periodic watcher gets scheduled, the
1379	reschedule callback will be called with the watcher as first, and the	2194	reschedule callback will be called with the watcher as first, and the
1380	current time as second argument.	2195	current time as second argument.
1381		2196
1382	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2197	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
1383	ever, or make ANY event loop modifications whatsoever>.	2198	or make ANY other event loop modifications whatsoever, unless explicitly
		2199	allowed by documentation here>.
1384		2200
1385	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop	2201	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
1386	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the	2202	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
1387	only event loop modification you are allowed to do).	2203	only event loop modification you are allowed to do).
1388		2204
1389	The callback prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic	2205	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1390	*w, ev_tstamp now)>, e.g.:	2206	*w, ev_tstamp now)>, e.g.:
1391		2207
		2208	static ev_tstamp
1392	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2209	my_rescheduler (ev_periodic *w, ev_tstamp now)
1393	{	2210	{
1394	return now + 60.;	2211	return now + 60.;
1395	}	2212	}
1396		2213
1397	It must return the next time to trigger, based on the passed time value	2214	It must return the next time to trigger, based on the passed time value
…		…
1417	a different time than the last time it was called (e.g. in a crond like	2234	a different time than the last time it was called (e.g. in a crond like
1418	program when the crontabs have changed).	2235	program when the crontabs have changed).
1419		2236
1420	=item ev_tstamp ev_periodic_at (ev_periodic *)	2237	=item ev_tstamp ev_periodic_at (ev_periodic *)
1421		2238
1422	When active, returns the absolute time that the watcher is supposed to	2239	When active, returns the absolute time that the watcher is supposed
1423	trigger next.	2240	to trigger next. This is not the same as the C<offset> argument to
		2241	C<ev_periodic_set>, but indeed works even in interval and manual
		2242	rescheduling modes.
1424		2243
1425	=item ev_tstamp offset [read-write]	2244	=item ev_tstamp offset [read-write]
1426		2245
1427	When repeating, this contains the offset value, otherwise this is the	2246	When repeating, this contains the offset value, otherwise this is the
1428	absolute point in time (the C<at> value passed to C<ev_periodic_set>).	2247	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2248	although libev might modify this value for better numerical stability).
1429		2249
1430	Can be modified any time, but changes only take effect when the periodic	2250	Can be modified any time, but changes only take effect when the periodic
1431	timer fires or C<ev_periodic_again> is being called.	2251	timer fires or C<ev_periodic_again> is being called.
1432		2252
1433	=item ev_tstamp interval [read-write]	2253	=item ev_tstamp interval [read-write]
1434		2254
1435	The current interval value. Can be modified any time, but changes only	2255	The current interval value. Can be modified any time, but changes only
1436	take effect when the periodic timer fires or C<ev_periodic_again> is being	2256	take effect when the periodic timer fires or C<ev_periodic_again> is being
1437	called.	2257	called.
1438		2258
1439	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	2259	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1440		2260
1441	The current reschedule callback, or C<0>, if this functionality is	2261	The current reschedule callback, or C<0>, if this functionality is
1442	switched off. Can be changed any time, but changes only take effect when	2262	switched off. Can be changed any time, but changes only take effect when
1443	the periodic timer fires or C<ev_periodic_again> is being called.	2263	the periodic timer fires or C<ev_periodic_again> is being called.
1444		2264
1445	=back	2265	=back
1446		2266
1447	=head3 Examples	2267	=head3 Examples
1448		2268
1449	Example: Call a callback every hour, or, more precisely, whenever the	2269	Example: Call a callback every hour, or, more precisely, whenever the
1450	system clock is divisible by 3600. The callback invocation times have	2270	system time is divisible by 3600. The callback invocation times have
1451	potentially a lot of jitter, but good long-term stability.	2271	potentially a lot of jitter, but good long-term stability.
1452		2272
1453	static void	2273	static void
1454	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	2274	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
1455	{	2275	{
1456	... its now a full hour (UTC, or TAI or whatever your clock follows)	2276	... its now a full hour (UTC, or TAI or whatever your clock follows)
1457	}	2277	}
1458		2278
1459	struct ev_periodic hourly_tick;	2279	ev_periodic hourly_tick;
1460	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	2280	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1461	ev_periodic_start (loop, &hourly_tick);	2281	ev_periodic_start (loop, &hourly_tick);
1462		2282
1463	Example: The same as above, but use a reschedule callback to do it:	2283	Example: The same as above, but use a reschedule callback to do it:
1464		2284
1465	#include <math.h>	2285	#include <math.h>
1466		2286
1467	static ev_tstamp	2287	static ev_tstamp
1468	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	2288	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1469	{	2289	{
1470	return fmod (now, 3600.) + 3600.;	2290	return now + (3600. - fmod (now, 3600.));
1471	}	2291	}
1472		2292
1473	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	2293	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1474		2294
1475	Example: Call a callback every hour, starting now:	2295	Example: Call a callback every hour, starting now:
1476		2296
1477	struct ev_periodic hourly_tick;	2297	ev_periodic hourly_tick;
1478	ev_periodic_init (&hourly_tick, clock_cb,	2298	ev_periodic_init (&hourly_tick, clock_cb,
1479	fmod (ev_now (loop), 3600.), 3600., 0);	2299	fmod (ev_now (loop), 3600.), 3600., 0);
1480	ev_periodic_start (loop, &hourly_tick);	2300	ev_periodic_start (loop, &hourly_tick);
1481		2301
1482		2302
1483	=head2 C<ev_signal> - signal me when a signal gets signalled!	2303	=head2 C<ev_signal> - signal me when a signal gets signalled!
1484		2304
1485	Signal watchers will trigger an event when the process receives a specific	2305	Signal watchers will trigger an event when the process receives a specific
1486	signal one or more times. Even though signals are very asynchronous, libev	2306	signal one or more times. Even though signals are very asynchronous, libev
1487	will try it's best to deliver signals synchronously, i.e. as part of the	2307	will try its best to deliver signals synchronously, i.e. as part of the
1488	normal event processing, like any other event.	2308	normal event processing, like any other event.
1489		2309
		2310	If you want signals to be delivered truly asynchronously, just use
		2311	C<sigaction> as you would do without libev and forget about sharing
		2312	the signal. You can even use C<ev_async> from a signal handler to
		2313	synchronously wake up an event loop.
		2314
1490	You can configure as many watchers as you like per signal. Only when the	2315	You can configure as many watchers as you like for the same signal, but
		2316	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2317	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2318	C<SIGINT> in both the default loop and another loop at the same time. At
		2319	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2320
1491	first watcher gets started will libev actually register a signal watcher	2321	When the first watcher gets started will libev actually register something
1492	with the kernel (thus it coexists with your own signal handlers as long	2322	with the kernel (thus it coexists with your own signal handlers as long as
1493	as you don't register any with libev). Similarly, when the last signal	2323	you don't register any with libev for the same signal).
1494	watcher for a signal is stopped libev will reset the signal handler to
1495	SIG_DFL (regardless of what it was set to before).
1496		2324
1497	If possible and supported, libev will install its handlers with	2325	If possible and supported, libev will install its handlers with
1498	C<SA_RESTART> behaviour enabled, so system calls should not be unduly	2326	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
1499	interrupted. If you have a problem with system calls getting interrupted by	2327	not be unduly interrupted. If you have a problem with system calls getting
1500	signals you can block all signals in an C<ev_check> watcher and unblock	2328	interrupted by signals you can block all signals in an C<ev_check> watcher
1501	them in an C<ev_prepare> watcher.	2329	and unblock them in an C<ev_prepare> watcher.
		2330
		2331	=head3 The special problem of inheritance over fork/execve/pthread_create
		2332
		2333	Both the signal mask (C<sigprocmask>) and the signal disposition
		2334	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2335	stopping it again), that is, libev might or might not block the signal,
		2336	and might or might not set or restore the installed signal handler.
		2337
		2338	While this does not matter for the signal disposition (libev never
		2339	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2340	C<execve>), this matters for the signal mask: many programs do not expect
		2341	certain signals to be blocked.
		2342
		2343	This means that before calling C<exec> (from the child) you should reset
		2344	the signal mask to whatever "default" you expect (all clear is a good
		2345	choice usually).
		2346
		2347	The simplest way to ensure that the signal mask is reset in the child is
		2348	to install a fork handler with C<pthread_atfork> that resets it. That will
		2349	catch fork calls done by libraries (such as the libc) as well.
		2350
		2351	In current versions of libev, the signal will not be blocked indefinitely
		2352	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2353	the window of opportunity for problems, it will not go away, as libev
		2354	I<has> to modify the signal mask, at least temporarily.
		2355
		2356	So I can't stress this enough: I<If you do not reset your signal mask when
		2357	you expect it to be empty, you have a race condition in your code>. This
		2358	is not a libev-specific thing, this is true for most event libraries.
		2359
		2360	=head3 The special problem of threads signal handling
		2361
		2362	POSIX threads has problematic signal handling semantics, specifically,
		2363	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2364	threads in a process block signals, which is hard to achieve.
		2365
		2366	When you want to use sigwait (or mix libev signal handling with your own
		2367	for the same signals), you can tackle this problem by globally blocking
		2368	all signals before creating any threads (or creating them with a fully set
		2369	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2370	loops. Then designate one thread as "signal receiver thread" which handles
		2371	these signals. You can pass on any signals that libev might be interested
		2372	in by calling C<ev_feed_signal>.
1502		2373
1503	=head3 Watcher-Specific Functions and Data Members	2374	=head3 Watcher-Specific Functions and Data Members
1504		2375
1505	=over 4	2376	=over 4
1506		2377
…		…
1517		2388
1518	=back	2389	=back
1519		2390
1520	=head3 Examples	2391	=head3 Examples
1521		2392
1522	Example: Try to exit cleanly on SIGINT and SIGTERM.	2393	Example: Try to exit cleanly on SIGINT.
1523		2394
1524	static void	2395	static void
1525	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	2396	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
1526	{	2397	{
1527	ev_unloop (loop, EVUNLOOP_ALL);	2398	ev_break (loop, EVBREAK_ALL);
1528	}	2399	}
1529		2400
1530	struct ev_signal signal_watcher;	2401	ev_signal signal_watcher;
1531	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	2402	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1532	ev_signal_start (loop, &sigint_cb);	2403	ev_signal_start (loop, &signal_watcher);
1533		2404
1534		2405
1535	=head2 C<ev_child> - watch out for process status changes	2406	=head2 C<ev_child> - watch out for process status changes
1536		2407
1537	Child watchers trigger when your process receives a SIGCHLD in response to	2408	Child watchers trigger when your process receives a SIGCHLD in response to
1538	some child status changes (most typically when a child of yours dies). It	2409	some child status changes (most typically when a child of yours dies or
1539	is permissible to install a child watcher I<after> the child has been	2410	exits). It is permissible to install a child watcher I<after> the child
1540	forked (which implies it might have already exited), as long as the event	2411	has been forked (which implies it might have already exited), as long
1541	loop isn't entered (or is continued from a watcher).	2412	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2413	forking and then immediately registering a watcher for the child is fine,
		2414	but forking and registering a watcher a few event loop iterations later or
		2415	in the next callback invocation is not.
1542		2416
1543	Only the default event loop is capable of handling signals, and therefore	2417	Only the default event loop is capable of handling signals, and therefore
1544	you can only register child watchers in the default event loop.	2418	you can only register child watchers in the default event loop.
1545		2419
		2420	Due to some design glitches inside libev, child watchers will always be
		2421	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2422	libev)
		2423
1546	=head3 Process Interaction	2424	=head3 Process Interaction
1547		2425
1548	Libev grabs C<SIGCHLD> as soon as the default event loop is	2426	Libev grabs C<SIGCHLD> as soon as the default event loop is
1549	initialised. This is necessary to guarantee proper behaviour even if	2427	initialised. This is necessary to guarantee proper behaviour even if the
1550	the first child watcher is started after the child exits. The occurrence	2428	first child watcher is started after the child exits. The occurrence
1551	of C<SIGCHLD> is recorded asynchronously, but child reaping is done	2429	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
1552	synchronously as part of the event loop processing. Libev always reaps all	2430	synchronously as part of the event loop processing. Libev always reaps all
1553	children, even ones not watched.	2431	children, even ones not watched.
1554		2432
1555	=head3 Overriding the Built-In Processing	2433	=head3 Overriding the Built-In Processing
…		…
1559	handler, you can override it easily by installing your own handler for	2437	handler, you can override it easily by installing your own handler for
1560	C<SIGCHLD> after initialising the default loop, and making sure the	2438	C<SIGCHLD> after initialising the default loop, and making sure the
1561	default loop never gets destroyed. You are encouraged, however, to use an	2439	default loop never gets destroyed. You are encouraged, however, to use an
1562	event-based approach to child reaping and thus use libev's support for	2440	event-based approach to child reaping and thus use libev's support for
1563	that, so other libev users can use C<ev_child> watchers freely.	2441	that, so other libev users can use C<ev_child> watchers freely.
		2442
		2443	=head3 Stopping the Child Watcher
		2444
		2445	Currently, the child watcher never gets stopped, even when the
		2446	child terminates, so normally one needs to stop the watcher in the
		2447	callback. Future versions of libev might stop the watcher automatically
		2448	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2449	problem).
1564		2450
1565	=head3 Watcher-Specific Functions and Data Members	2451	=head3 Watcher-Specific Functions and Data Members
1566		2452
1567	=over 4	2453	=over 4
1568		2454
…		…
1597	=head3 Examples	2483	=head3 Examples
1598		2484
1599	Example: C<fork()> a new process and install a child handler to wait for	2485	Example: C<fork()> a new process and install a child handler to wait for
1600	its completion.	2486	its completion.
1601		2487
1602	ev_child cw;	2488	ev_child cw;
1603		2489
1604	static void	2490	static void
1605	child_cb (EV_P_ struct ev_child *w, int revents)	2491	child_cb (EV_P_ ev_child *w, int revents)
1606	{	2492	{
1607	ev_child_stop (EV_A_ w);	2493	ev_child_stop (EV_A_ w);
1608	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);	2494	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1609	}	2495	}
1610		2496
1611	pid_t pid = fork ();	2497	pid_t pid = fork ();
1612		2498
1613	if (pid < 0)	2499	if (pid < 0)
1614	// error	2500	// error
1615	else if (pid == 0)	2501	else if (pid == 0)
1616	{	2502	{
1617	// the forked child executes here	2503	// the forked child executes here
1618	exit (1);	2504	exit (1);
1619	}	2505	}
1620	else	2506	else
1621	{	2507	{
1622	ev_child_init (&cw, child_cb, pid, 0);	2508	ev_child_init (&cw, child_cb, pid, 0);
1623	ev_child_start (EV_DEFAULT_ &cw);	2509	ev_child_start (EV_DEFAULT_ &cw);
1624	}	2510	}
1625		2511
1626		2512
1627	=head2 C<ev_stat> - did the file attributes just change?	2513	=head2 C<ev_stat> - did the file attributes just change?
1628		2514
1629	This watches a file system path for attribute changes. That is, it calls	2515	This watches a file system path for attribute changes. That is, it calls
1630	C<stat> regularly (or when the OS says it changed) and sees if it changed	2516	C<stat> on that path in regular intervals (or when the OS says it changed)
1631	compared to the last time, invoking the callback if it did.	2517	and sees if it changed compared to the last time, invoking the callback if
		2518	it did.
1632		2519
1633	The path does not need to exist: changing from "path exists" to "path does	2520	The path does not need to exist: changing from "path exists" to "path does
1634	not exist" is a status change like any other. The condition "path does	2521	not exist" is a status change like any other. The condition "path does not
1635	not exist" is signified by the C<st_nlink> field being zero (which is	2522	exist" (or more correctly "path cannot be stat'ed") is signified by the
1636	otherwise always forced to be at least one) and all the other fields of	2523	C<st_nlink> field being zero (which is otherwise always forced to be at
1637	the stat buffer having unspecified contents.	2524	least one) and all the other fields of the stat buffer having unspecified
		2525	contents.
1638		2526
1639	The path I<should> be absolute and I<must not> end in a slash. If it is	2527	The path I<must not> end in a slash or contain special components such as
		2528	C<.> or C<..>. The path I<should> be absolute: If it is relative and
1640	relative and your working directory changes, the behaviour is undefined.	2529	your working directory changes, then the behaviour is undefined.
1641		2530
1642	Since there is no standard to do this, the portable implementation simply	2531	Since there is no portable change notification interface available, the
1643	calls C<stat (2)> regularly on the path to see if it changed somehow. You	2532	portable implementation simply calls C<stat(2)> regularly on the path
1644	can specify a recommended polling interval for this case. If you specify	2533	to see if it changed somehow. You can specify a recommended polling
1645	a polling interval of C<0> (highly recommended!) then a I<suitable,	2534	interval for this case. If you specify a polling interval of C<0> (highly
1646	unspecified default> value will be used (which you can expect to be around	2535	recommended!) then a I<suitable, unspecified default> value will be used
1647	five seconds, although this might change dynamically). Libev will also	2536	(which you can expect to be around five seconds, although this might
1648	impose a minimum interval which is currently around C<0.1>, but thats	2537	change dynamically). Libev will also impose a minimum interval which is
1649	usually overkill.	2538	currently around C<0.1>, but that's usually overkill.
1650		2539
1651	This watcher type is not meant for massive numbers of stat watchers,	2540	This watcher type is not meant for massive numbers of stat watchers,
1652	as even with OS-supported change notifications, this can be	2541	as even with OS-supported change notifications, this can be
1653	resource-intensive.	2542	resource-intensive.
1654		2543
1655	At the time of this writing, only the Linux inotify interface is	2544	At the time of this writing, the only OS-specific interface implemented
1656	implemented (implementing kqueue support is left as an exercise for the	2545	is the Linux inotify interface (implementing kqueue support is left as an
1657	reader, note, however, that the author sees no way of implementing ev_stat	2546	exercise for the reader. Note, however, that the author sees no way of
1658	semantics with kqueue). Inotify will be used to give hints only and should	2547	implementing C<ev_stat> semantics with kqueue, except as a hint).
1659	not change the semantics of C<ev_stat> watchers, which means that libev
1660	sometimes needs to fall back to regular polling again even with inotify,
1661	but changes are usually detected immediately, and if the file exists there
1662	will be no polling.
1663		2548
1664	=head3 ABI Issues (Largefile Support)	2549	=head3 ABI Issues (Largefile Support)
1665		2550
1666	Libev by default (unless the user overrides this) uses the default	2551	Libev by default (unless the user overrides this) uses the default
1667	compilation environment, which means that on systems with optionally	2552	compilation environment, which means that on systems with large file
1668	disabled large file support, you get the 32 bit version of the stat	2553	support disabled by default, you get the 32 bit version of the stat
1669	structure. When using the library from programs that change the ABI to	2554	structure. When using the library from programs that change the ABI to
1670	use 64 bit file offsets the programs will fail. In that case you have to	2555	use 64 bit file offsets the programs will fail. In that case you have to
1671	compile libev with the same flags to get binary compatibility. This is	2556	compile libev with the same flags to get binary compatibility. This is
1672	obviously the case with any flags that change the ABI, but the problem is	2557	obviously the case with any flags that change the ABI, but the problem is
1673	most noticeably with ev_stat and large file support.	2558	most noticeably displayed with ev_stat and large file support.
1674		2559
1675	=head3 Inotify	2560	The solution for this is to lobby your distribution maker to make large
		2561	file interfaces available by default (as e.g. FreeBSD does) and not
		2562	optional. Libev cannot simply switch on large file support because it has
		2563	to exchange stat structures with application programs compiled using the
		2564	default compilation environment.
1676		2565
		2566	=head3 Inotify and Kqueue
		2567
1677	When C<inotify (7)> support has been compiled into libev (generally only	2568	When C<inotify (7)> support has been compiled into libev and present at
1678	available on Linux) and present at runtime, it will be used to speed up	2569	runtime, it will be used to speed up change detection where possible. The
1679	change detection where possible. The inotify descriptor will be created lazily	2570	inotify descriptor will be created lazily when the first C<ev_stat>
1680	when the first C<ev_stat> watcher is being started.	2571	watcher is being started.
1681		2572
1682	Inotify presence does not change the semantics of C<ev_stat> watchers	2573	Inotify presence does not change the semantics of C<ev_stat> watchers
1683	except that changes might be detected earlier, and in some cases, to avoid	2574	except that changes might be detected earlier, and in some cases, to avoid
1684	making regular C<stat> calls. Even in the presence of inotify support	2575	making regular C<stat> calls. Even in the presence of inotify support
1685	there are many cases where libev has to resort to regular C<stat> polling.	2576	there are many cases where libev has to resort to regular C<stat> polling,
		2577	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2578	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2579	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2580	xfs are fully working) libev usually gets away without polling.
1686		2581
1687	(There is no support for kqueue, as apparently it cannot be used to	2582	There is no support for kqueue, as apparently it cannot be used to
1688	implement this functionality, due to the requirement of having a file	2583	implement this functionality, due to the requirement of having a file
1689	descriptor open on the object at all times).	2584	descriptor open on the object at all times, and detecting renames, unlinks
		2585	etc. is difficult.
		2586
		2587	=head3 C<stat ()> is a synchronous operation
		2588
		2589	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2590	the process. The exception are C<ev_stat> watchers - those call C<stat
		2591	()>, which is a synchronous operation.
		2592
		2593	For local paths, this usually doesn't matter: unless the system is very
		2594	busy or the intervals between stat's are large, a stat call will be fast,
		2595	as the path data is usually in memory already (except when starting the
		2596	watcher).
		2597
		2598	For networked file systems, calling C<stat ()> can block an indefinite
		2599	time due to network issues, and even under good conditions, a stat call
		2600	often takes multiple milliseconds.
		2601
		2602	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2603	paths, although this is fully supported by libev.
1690		2604
1691	=head3 The special problem of stat time resolution	2605	=head3 The special problem of stat time resolution
1692		2606
1693	The C<stat ()> system call only supports full-second resolution portably, and	2607	The C<stat ()> system call only supports full-second resolution portably,
1694	even on systems where the resolution is higher, many file systems still	2608	and even on systems where the resolution is higher, most file systems
1695	only support whole seconds.	2609	still only support whole seconds.
1696		2610
1697	That means that, if the time is the only thing that changes, you can	2611	That means that, if the time is the only thing that changes, you can
1698	easily miss updates: on the first update, C<ev_stat> detects a change and	2612	easily miss updates: on the first update, C<ev_stat> detects a change and
1699	calls your callback, which does something. When there is another update	2613	calls your callback, which does something. When there is another update
1700	within the same second, C<ev_stat> will be unable to detect it as the stat	2614	within the same second, C<ev_stat> will be unable to detect unless the
1701	data does not change.	2615	stat data does change in other ways (e.g. file size).
1702		2616
1703	The solution to this is to delay acting on a change for slightly more	2617	The solution to this is to delay acting on a change for slightly more
1704	than a second (or till slightly after the next full second boundary), using	2618	than a second (or till slightly after the next full second boundary), using
1705	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);	2619	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
1706	ev_timer_again (loop, w)>).	2620	ev_timer_again (loop, w)>).
…		…
1726	C<path>. The C<interval> is a hint on how quickly a change is expected to	2640	C<path>. The C<interval> is a hint on how quickly a change is expected to
1727	be detected and should normally be specified as C<0> to let libev choose	2641	be detected and should normally be specified as C<0> to let libev choose
1728	a suitable value. The memory pointed to by C<path> must point to the same	2642	a suitable value. The memory pointed to by C<path> must point to the same
1729	path for as long as the watcher is active.	2643	path for as long as the watcher is active.
1730		2644
1731	The callback will receive C<EV_STAT> when a change was detected, relative	2645	The callback will receive an C<EV_STAT> event when a change was detected,
1732	to the attributes at the time the watcher was started (or the last change	2646	relative to the attributes at the time the watcher was started (or the
1733	was detected).	2647	last change was detected).
1734		2648
1735	=item ev_stat_stat (loop, ev_stat *)	2649	=item ev_stat_stat (loop, ev_stat *)
1736		2650
1737	Updates the stat buffer immediately with new values. If you change the	2651	Updates the stat buffer immediately with new values. If you change the
1738	watched path in your callback, you could call this function to avoid	2652	watched path in your callback, you could call this function to avoid
…		…
1767		2681
1768	=head3 Examples	2682	=head3 Examples
1769		2683
1770	Example: Watch C</etc/passwd> for attribute changes.	2684	Example: Watch C</etc/passwd> for attribute changes.
1771		2685
1772	static void	2686	static void
1773	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2687	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1774	{	2688	{
1775	/* /etc/passwd changed in some way */	2689	/* /etc/passwd changed in some way */
1776	if (w->attr.st_nlink)	2690	if (w->attr.st_nlink)
1777	{	2691	{
1778	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2692	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1779	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2693	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1780	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2694	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1781	}	2695	}
1782	else	2696	else
1783	/* you shalt not abuse printf for puts */	2697	/* you shalt not abuse printf for puts */
1784	puts ("wow, /etc/passwd is not there, expect problems. "	2698	puts ("wow, /etc/passwd is not there, expect problems. "
1785	"if this is windows, they already arrived\n");	2699	"if this is windows, they already arrived\n");
1786	}	2700	}
1787		2701
1788	...	2702	...
1789	ev_stat passwd;	2703	ev_stat passwd;
1790		2704
1791	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);	2705	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1792	ev_stat_start (loop, &passwd);	2706	ev_stat_start (loop, &passwd);
1793		2707
1794	Example: Like above, but additionally use a one-second delay so we do not	2708	Example: Like above, but additionally use a one-second delay so we do not
1795	miss updates (however, frequent updates will delay processing, too, so	2709	miss updates (however, frequent updates will delay processing, too, so
1796	one might do the work both on C<ev_stat> callback invocation I<and> on	2710	one might do the work both on C<ev_stat> callback invocation I<and> on
1797	C<ev_timer> callback invocation).	2711	C<ev_timer> callback invocation).
1798		2712
1799	static ev_stat passwd;	2713	static ev_stat passwd;
1800	static ev_timer timer;	2714	static ev_timer timer;
1801		2715
1802	static void	2716	static void
1803	timer_cb (EV_P_ ev_timer *w, int revents)	2717	timer_cb (EV_P_ ev_timer *w, int revents)
1804	{	2718	{
1805	ev_timer_stop (EV_A_ w);	2719	ev_timer_stop (EV_A_ w);
1806		2720
1807	/* now it's one second after the most recent passwd change */	2721	/* now it's one second after the most recent passwd change */
1808	}	2722	}
1809		2723
1810	static void	2724	static void
1811	stat_cb (EV_P_ ev_stat *w, int revents)	2725	stat_cb (EV_P_ ev_stat *w, int revents)
1812	{	2726	{
1813	/* reset the one-second timer */	2727	/* reset the one-second timer */
1814	ev_timer_again (EV_A_ &timer);	2728	ev_timer_again (EV_A_ &timer);
1815	}	2729	}
1816		2730
1817	...	2731	...
1818	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);	2732	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1819	ev_stat_start (loop, &passwd);	2733	ev_stat_start (loop, &passwd);
1820	ev_timer_init (&timer, timer_cb, 0., 1.02);	2734	ev_timer_init (&timer, timer_cb, 0., 1.02);
1821		2735
1822		2736
1823	=head2 C<ev_idle> - when you've got nothing better to do...	2737	=head2 C<ev_idle> - when you've got nothing better to do...
1824		2738
1825	Idle watchers trigger events when no other events of the same or higher	2739	Idle watchers trigger events when no other events of the same or higher
1826	priority are pending (prepare, check and other idle watchers do not	2740	priority are pending (prepare, check and other idle watchers do not count
1827	count).	2741	as receiving "events").
1828		2742
1829	That is, as long as your process is busy handling sockets or timeouts	2743	That is, as long as your process is busy handling sockets or timeouts
1830	(or even signals, imagine) of the same or higher priority it will not be	2744	(or even signals, imagine) of the same or higher priority it will not be
1831	triggered. But when your process is idle (or only lower-priority watchers	2745	triggered. But when your process is idle (or only lower-priority watchers
1832	are pending), the idle watchers are being called once per event loop	2746	are pending), the idle watchers are being called once per event loop
…		…
1843		2757
1844	=head3 Watcher-Specific Functions and Data Members	2758	=head3 Watcher-Specific Functions and Data Members
1845		2759
1846	=over 4	2760	=over 4
1847		2761
1848	=item ev_idle_init (ev_signal *, callback)	2762	=item ev_idle_init (ev_idle *, callback)
1849		2763
1850	Initialises and configures the idle watcher - it has no parameters of any	2764	Initialises and configures the idle watcher - it has no parameters of any
1851	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2765	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1852	believe me.	2766	believe me.
1853		2767
…		…
1856	=head3 Examples	2770	=head3 Examples
1857		2771
1858	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	2772	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1859	callback, free it. Also, use no error checking, as usual.	2773	callback, free it. Also, use no error checking, as usual.
1860		2774
1861	static void	2775	static void
1862	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2776	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1863	{	2777	{
1864	free (w);	2778	free (w);
1865	// now do something you wanted to do when the program has	2779	// now do something you wanted to do when the program has
1866	// no longer anything immediate to do.	2780	// no longer anything immediate to do.
1867	}	2781	}
1868		2782
1869	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2783	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1870	ev_idle_init (idle_watcher, idle_cb);	2784	ev_idle_init (idle_watcher, idle_cb);
1871	ev_idle_start (loop, idle_cb);	2785	ev_idle_start (loop, idle_watcher);
1872		2786
1873		2787
1874	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	2788	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1875		2789
1876	Prepare and check watchers are usually (but not always) used in tandem:	2790	Prepare and check watchers are usually (but not always) used in pairs:
1877	prepare watchers get invoked before the process blocks and check watchers	2791	prepare watchers get invoked before the process blocks and check watchers
1878	afterwards.	2792	afterwards.
1879		2793
1880	You I<must not> call C<ev_loop> or similar functions that enter	2794	You I<must not> call C<ev_run> or similar functions that enter
1881	the current event loop from either C<ev_prepare> or C<ev_check>	2795	the current event loop from either C<ev_prepare> or C<ev_check>
1882	watchers. Other loops than the current one are fine, however. The	2796	watchers. Other loops than the current one are fine, however. The
1883	rationale behind this is that you do not need to check for recursion in	2797	rationale behind this is that you do not need to check for recursion in
1884	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	2798	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
1885	C<ev_check> so if you have one watcher of each kind they will always be	2799	C<ev_check> so if you have one watcher of each kind they will always be
1886	called in pairs bracketing the blocking call.	2800	called in pairs bracketing the blocking call.
1887		2801
1888	Their main purpose is to integrate other event mechanisms into libev and	2802	Their main purpose is to integrate other event mechanisms into libev and
1889	their use is somewhat advanced. This could be used, for example, to track	2803	their use is somewhat advanced. They could be used, for example, to track
1890	variable changes, implement your own watchers, integrate net-snmp or a	2804	variable changes, implement your own watchers, integrate net-snmp or a
1891	coroutine library and lots more. They are also occasionally useful if	2805	coroutine library and lots more. They are also occasionally useful if
1892	you cache some data and want to flush it before blocking (for example,	2806	you cache some data and want to flush it before blocking (for example,
1893	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	2807	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1894	watcher).	2808	watcher).
1895		2809
1896	This is done by examining in each prepare call which file descriptors need	2810	This is done by examining in each prepare call which file descriptors
1897	to be watched by the other library, registering C<ev_io> watchers for	2811	need to be watched by the other library, registering C<ev_io> watchers
1898	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2812	for them and starting an C<ev_timer> watcher for any timeouts (many
1899	provide just this functionality). Then, in the check watcher you check for	2813	libraries provide exactly this functionality). Then, in the check watcher,
1900	any events that occurred (by checking the pending status of all watchers	2814	you check for any events that occurred (by checking the pending status
1901	and stopping them) and call back into the library. The I/O and timer	2815	of all watchers and stopping them) and call back into the library. The
1902	callbacks will never actually be called (but must be valid nevertheless,	2816	I/O and timer callbacks will never actually be called (but must be valid
1903	because you never know, you know?).	2817	nevertheless, because you never know, you know?).
1904		2818
1905	As another example, the Perl Coro module uses these hooks to integrate	2819	As another example, the Perl Coro module uses these hooks to integrate
1906	coroutines into libev programs, by yielding to other active coroutines	2820	coroutines into libev programs, by yielding to other active coroutines
1907	during each prepare and only letting the process block if no coroutines	2821	during each prepare and only letting the process block if no coroutines
1908	are ready to run (it's actually more complicated: it only runs coroutines	2822	are ready to run (it's actually more complicated: it only runs coroutines
…		…
1911	loop from blocking if lower-priority coroutines are active, thus mapping	2825	loop from blocking if lower-priority coroutines are active, thus mapping
1912	low-priority coroutines to idle/background tasks).	2826	low-priority coroutines to idle/background tasks).
1913		2827
1914	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	2828	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1915	priority, to ensure that they are being run before any other watchers	2829	priority, to ensure that they are being run before any other watchers
		2830	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2831
1916	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	2832	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
1917	too) should not activate ("feed") events into libev. While libev fully	2833	activate ("feed") events into libev. While libev fully supports this, they
1918	supports this, they might get executed before other C<ev_check> watchers	2834	might get executed before other C<ev_check> watchers did their job. As
1919	did their job. As C<ev_check> watchers are often used to embed other	2835	C<ev_check> watchers are often used to embed other (non-libev) event
1920	(non-libev) event loops those other event loops might be in an unusable	2836	loops those other event loops might be in an unusable state until their
1921	state until their C<ev_check> watcher ran (always remind yourself to	2837	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
1922	coexist peacefully with others).	2838	others).
1923		2839
1924	=head3 Watcher-Specific Functions and Data Members	2840	=head3 Watcher-Specific Functions and Data Members
1925		2841
1926	=over 4	2842	=over 4
1927		2843
…		…
1929		2845
1930	=item ev_check_init (ev_check *, callback)	2846	=item ev_check_init (ev_check *, callback)
1931		2847
1932	Initialises and configures the prepare or check watcher - they have no	2848	Initialises and configures the prepare or check watcher - they have no
1933	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2849	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1934	macros, but using them is utterly, utterly and completely pointless.	2850	macros, but using them is utterly, utterly, utterly and completely
		2851	pointless.
1935		2852
1936	=back	2853	=back
1937		2854
1938	=head3 Examples	2855	=head3 Examples
1939		2856
…		…
1948	and in a check watcher, destroy them and call into libadns. What follows	2865	and in a check watcher, destroy them and call into libadns. What follows
1949	is pseudo-code only of course. This requires you to either use a low	2866	is pseudo-code only of course. This requires you to either use a low
1950	priority for the check watcher or use C<ev_clear_pending> explicitly, as	2867	priority for the check watcher or use C<ev_clear_pending> explicitly, as
1951	the callbacks for the IO/timeout watchers might not have been called yet.	2868	the callbacks for the IO/timeout watchers might not have been called yet.
1952		2869
1953	static ev_io iow [nfd];	2870	static ev_io iow [nfd];
1954	static ev_timer tw;	2871	static ev_timer tw;
1955		2872
1956	static void	2873	static void
1957	io_cb (ev_loop *loop, ev_io *w, int revents)	2874	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1958	{	2875	{
1959	}	2876	}
1960		2877
1961	// create io watchers for each fd and a timer before blocking	2878	// create io watchers for each fd and a timer before blocking
1962	static void	2879	static void
1963	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	2880	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1964	{	2881	{
1965	int timeout = 3600000;	2882	int timeout = 3600000;
1966	struct pollfd fds [nfd];	2883	struct pollfd fds [nfd];
1967	// actual code will need to loop here and realloc etc.	2884	// actual code will need to loop here and realloc etc.
1968	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	2885	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1969		2886
1970	/* the callback is illegal, but won't be called as we stop during check */	2887	/* the callback is illegal, but won't be called as we stop during check */
1971	ev_timer_init (&tw, 0, timeout * 1e-3);	2888	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
1972	ev_timer_start (loop, &tw);	2889	ev_timer_start (loop, &tw);
1973		2890
1974	// create one ev_io per pollfd	2891	// create one ev_io per pollfd
1975	for (int i = 0; i < nfd; ++i)	2892	for (int i = 0; i < nfd; ++i)
1976	{	2893	{
1977	ev_io_init (iow + i, io_cb, fds [i].fd,	2894	ev_io_init (iow + i, io_cb, fds [i].fd,
1978	((fds [i].events & POLLIN ? EV_READ : 0)	2895	((fds [i].events & POLLIN ? EV_READ : 0)
1979	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	2896	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1980		2897
1981	fds [i].revents = 0;	2898	fds [i].revents = 0;
1982	ev_io_start (loop, iow + i);	2899	ev_io_start (loop, iow + i);
1983	}	2900	}
1984	}	2901	}
1985		2902
1986	// stop all watchers after blocking	2903	// stop all watchers after blocking
1987	static void	2904	static void
1988	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	2905	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1989	{	2906	{
1990	ev_timer_stop (loop, &tw);	2907	ev_timer_stop (loop, &tw);
1991		2908
1992	for (int i = 0; i < nfd; ++i)	2909	for (int i = 0; i < nfd; ++i)
1993	{	2910	{
1994	// set the relevant poll flags	2911	// set the relevant poll flags
1995	// could also call adns_processreadable etc. here	2912	// could also call adns_processreadable etc. here
1996	struct pollfd *fd = fds + i;	2913	struct pollfd *fd = fds + i;
1997	int revents = ev_clear_pending (iow + i);	2914	int revents = ev_clear_pending (iow + i);
1998	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;	2915	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1999	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;	2916	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
2000		2917
2001	// now stop the watcher	2918	// now stop the watcher
2002	ev_io_stop (loop, iow + i);	2919	ev_io_stop (loop, iow + i);
2003	}	2920	}
2004		2921
2005	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	2922	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
2006	}	2923	}
2007		2924
2008	Method 2: This would be just like method 1, but you run C<adns_afterpoll>	2925	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
2009	in the prepare watcher and would dispose of the check watcher.	2926	in the prepare watcher and would dispose of the check watcher.
2010		2927
2011	Method 3: If the module to be embedded supports explicit event	2928	Method 3: If the module to be embedded supports explicit event
2012	notification (libadns does), you can also make use of the actual watcher	2929	notification (libadns does), you can also make use of the actual watcher
2013	callbacks, and only destroy/create the watchers in the prepare watcher.	2930	callbacks, and only destroy/create the watchers in the prepare watcher.
2014		2931
2015	static void	2932	static void
2016	timer_cb (EV_P_ ev_timer *w, int revents)	2933	timer_cb (EV_P_ ev_timer *w, int revents)
2017	{	2934	{
2018	adns_state ads = (adns_state)w->data;	2935	adns_state ads = (adns_state)w->data;
2019	update_now (EV_A);	2936	update_now (EV_A);
2020		2937
2021	adns_processtimeouts (ads, &tv_now);	2938	adns_processtimeouts (ads, &tv_now);
2022	}	2939	}
2023		2940
2024	static void	2941	static void
2025	io_cb (EV_P_ ev_io *w, int revents)	2942	io_cb (EV_P_ ev_io *w, int revents)
2026	{	2943	{
2027	adns_state ads = (adns_state)w->data;	2944	adns_state ads = (adns_state)w->data;
2028	update_now (EV_A);	2945	update_now (EV_A);
2029		2946
2030	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);	2947	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
2031	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);	2948	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
2032	}	2949	}
2033		2950
2034	// do not ever call adns_afterpoll	2951	// do not ever call adns_afterpoll
2035		2952
2036	Method 4: Do not use a prepare or check watcher because the module you	2953	Method 4: Do not use a prepare or check watcher because the module you
2037	want to embed is too inflexible to support it. Instead, you can override	2954	want to embed is not flexible enough to support it. Instead, you can
2038	their poll function. The drawback with this solution is that the main	2955	override their poll function. The drawback with this solution is that the
2039	loop is now no longer controllable by EV. The C<Glib::EV> module does	2956	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
2040	this.	2957	this approach, effectively embedding EV as a client into the horrible
		2958	libglib event loop.
2041		2959
2042	static gint	2960	static gint
2043	event_poll_func (GPollFD *fds, guint nfds, gint timeout)	2961	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
2044	{	2962	{
2045	int got_events = 0;	2963	int got_events = 0;
2046		2964
2047	for (n = 0; n < nfds; ++n)	2965	for (n = 0; n < nfds; ++n)
2048	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events	2966	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
2049		2967
2050	if (timeout >= 0)	2968	if (timeout >= 0)
2051	// create/start timer	2969	// create/start timer
2052		2970
2053	// poll	2971	// poll
2054	ev_loop (EV_A_ 0);	2972	ev_run (EV_A_ 0);
2055		2973
2056	// stop timer again	2974	// stop timer again
2057	if (timeout >= 0)	2975	if (timeout >= 0)
2058	ev_timer_stop (EV_A_ &to);	2976	ev_timer_stop (EV_A_ &to);
2059		2977
2060	// stop io watchers again - their callbacks should have set	2978	// stop io watchers again - their callbacks should have set
2061	for (n = 0; n < nfds; ++n)	2979	for (n = 0; n < nfds; ++n)
2062	ev_io_stop (EV_A_ iow [n]);	2980	ev_io_stop (EV_A_ iow [n]);
2063		2981
2064	return got_events;	2982	return got_events;
2065	}	2983	}
2066		2984
2067		2985
2068	=head2 C<ev_embed> - when one backend isn't enough...	2986	=head2 C<ev_embed> - when one backend isn't enough...
2069		2987
2070	This is a rather advanced watcher type that lets you embed one event loop	2988	This is a rather advanced watcher type that lets you embed one event loop
…		…
2076	prioritise I/O.	2994	prioritise I/O.
2077		2995
2078	As an example for a bug workaround, the kqueue backend might only support	2996	As an example for a bug workaround, the kqueue backend might only support
2079	sockets on some platform, so it is unusable as generic backend, but you	2997	sockets on some platform, so it is unusable as generic backend, but you
2080	still want to make use of it because you have many sockets and it scales	2998	still want to make use of it because you have many sockets and it scales
2081	so nicely. In this case, you would create a kqueue-based loop and embed it	2999	so nicely. In this case, you would create a kqueue-based loop and embed
2082	into your default loop (which might use e.g. poll). Overall operation will	3000	it into your default loop (which might use e.g. poll). Overall operation
2083	be a bit slower because first libev has to poll and then call kevent, but	3001	will be a bit slower because first libev has to call C<poll> and then
2084	at least you can use both at what they are best.	3002	C<kevent>, but at least you can use both mechanisms for what they are
		3003	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
2085		3004
2086	As for prioritising I/O: rarely you have the case where some fds have	3005	As for prioritising I/O: under rare circumstances you have the case where
2087	to be watched and handled very quickly (with low latency), and even	3006	some fds have to be watched and handled very quickly (with low latency),
2088	priorities and idle watchers might have too much overhead. In this case	3007	and even priorities and idle watchers might have too much overhead. In
2089	you would put all the high priority stuff in one loop and all the rest in	3008	this case you would put all the high priority stuff in one loop and all
2090	a second one, and embed the second one in the first.	3009	the rest in a second one, and embed the second one in the first.
2091		3010
2092	As long as the watcher is active, the callback will be invoked every time	3011	As long as the watcher is active, the callback will be invoked every
2093	there might be events pending in the embedded loop. The callback must then	3012	time there might be events pending in the embedded loop. The callback
2094	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	3013	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
2095	their callbacks (you could also start an idle watcher to give the embedded	3014	sweep and invoke their callbacks (the callback doesn't need to invoke the
2096	loop strictly lower priority for example). You can also set the callback	3015	C<ev_embed_sweep> function directly, it could also start an idle watcher
2097	to C<0>, in which case the embed watcher will automatically execute the	3016	to give the embedded loop strictly lower priority for example).
2098	embedded loop sweep.
2099		3017
2100	As long as the watcher is started it will automatically handle events. The	3018	You can also set the callback to C<0>, in which case the embed watcher
2101	callback will be invoked whenever some events have been handled. You can	3019	will automatically execute the embedded loop sweep whenever necessary.
2102	set the callback to C<0> to avoid having to specify one if you are not
2103	interested in that.
2104		3020
2105	Also, there have not currently been made special provisions for forking:	3021	Fork detection will be handled transparently while the C<ev_embed> watcher
2106	when you fork, you not only have to call C<ev_loop_fork> on both loops,	3022	is active, i.e., the embedded loop will automatically be forked when the
2107	but you will also have to stop and restart any C<ev_embed> watchers	3023	embedding loop forks. In other cases, the user is responsible for calling
2108	yourself.	3024	C<ev_loop_fork> on the embedded loop.
2109		3025
2110	Unfortunately, not all backends are embeddable, only the ones returned by	3026	Unfortunately, not all backends are embeddable: only the ones returned by
2111	C<ev_embeddable_backends> are, which, unfortunately, does not include any	3027	C<ev_embeddable_backends> are, which, unfortunately, does not include any
2112	portable one.	3028	portable one.
2113		3029
2114	So when you want to use this feature you will always have to be prepared	3030	So when you want to use this feature you will always have to be prepared
2115	that you cannot get an embeddable loop. The recommended way to get around	3031	that you cannot get an embeddable loop. The recommended way to get around
2116	this is to have a separate variables for your embeddable loop, try to	3032	this is to have a separate variables for your embeddable loop, try to
2117	create it, and if that fails, use the normal loop for everything.	3033	create it, and if that fails, use the normal loop for everything.
		3034
		3035	=head3 C<ev_embed> and fork
		3036
		3037	While the C<ev_embed> watcher is running, forks in the embedding loop will
		3038	automatically be applied to the embedded loop as well, so no special
		3039	fork handling is required in that case. When the watcher is not running,
		3040	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		3041	as applicable.
2118		3042
2119	=head3 Watcher-Specific Functions and Data Members	3043	=head3 Watcher-Specific Functions and Data Members
2120		3044
2121	=over 4	3045	=over 4
2122		3046
…		…
2131	if you do not want that, you need to temporarily stop the embed watcher).	3055	if you do not want that, you need to temporarily stop the embed watcher).
2132		3056
2133	=item ev_embed_sweep (loop, ev_embed *)	3057	=item ev_embed_sweep (loop, ev_embed *)
2134		3058
2135	Make a single, non-blocking sweep over the embedded loop. This works	3059	Make a single, non-blocking sweep over the embedded loop. This works
2136	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	3060	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
2137	appropriate way for embedded loops.	3061	appropriate way for embedded loops.
2138		3062
2139	=item struct ev_loop *other [read-only]	3063	=item struct ev_loop *other [read-only]
2140		3064
2141	The embedded event loop.	3065	The embedded event loop.
…		…
2148	event loop. If that is not possible, use the default loop. The default	3072	event loop. If that is not possible, use the default loop. The default
2149	loop is stored in C<loop_hi>, while the embeddable loop is stored in	3073	loop is stored in C<loop_hi>, while the embeddable loop is stored in
2150	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be	3074	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
2151	used).	3075	used).
2152		3076
2153	struct ev_loop *loop_hi = ev_default_init (0);	3077	struct ev_loop *loop_hi = ev_default_init (0);
2154	struct ev_loop *loop_lo = 0;	3078	struct ev_loop *loop_lo = 0;
2155	struct ev_embed embed;	3079	ev_embed embed;
2156		3080
2157	// see if there is a chance of getting one that works	3081	// see if there is a chance of getting one that works
2158	// (remember that a flags value of 0 means autodetection)	3082	// (remember that a flags value of 0 means autodetection)
2159	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()	3083	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2160	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())	3084	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2161	: 0;	3085	: 0;
2162		3086
2163	// if we got one, then embed it, otherwise default to loop_hi	3087	// if we got one, then embed it, otherwise default to loop_hi
2164	if (loop_lo)	3088	if (loop_lo)
2165	{	3089	{
2166	ev_embed_init (&embed, 0, loop_lo);	3090	ev_embed_init (&embed, 0, loop_lo);
2167	ev_embed_start (loop_hi, &embed);	3091	ev_embed_start (loop_hi, &embed);
2168	}	3092	}
2169	else	3093	else
2170	loop_lo = loop_hi;	3094	loop_lo = loop_hi;
2171		3095
2172	Example: Check if kqueue is available but not recommended and create	3096	Example: Check if kqueue is available but not recommended and create
2173	a kqueue backend for use with sockets (which usually work with any	3097	a kqueue backend for use with sockets (which usually work with any
2174	kqueue implementation). Store the kqueue/socket-only event loop in	3098	kqueue implementation). Store the kqueue/socket-only event loop in
2175	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).	3099	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
2176		3100
2177	struct ev_loop *loop = ev_default_init (0);	3101	struct ev_loop *loop = ev_default_init (0);
2178	struct ev_loop *loop_socket = 0;	3102	struct ev_loop *loop_socket = 0;
2179	struct ev_embed embed;	3103	ev_embed embed;
2180		3104
2181	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)	3105	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2182	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))	3106	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2183	{	3107	{
2184	ev_embed_init (&embed, 0, loop_socket);	3108	ev_embed_init (&embed, 0, loop_socket);
2185	ev_embed_start (loop, &embed);	3109	ev_embed_start (loop, &embed);
2186	}	3110	}
2187		3111
2188	if (!loop_socket)	3112	if (!loop_socket)
2189	loop_socket = loop;	3113	loop_socket = loop;
2190		3114
2191	// now use loop_socket for all sockets, and loop for everything else	3115	// now use loop_socket for all sockets, and loop for everything else
2192		3116
2193		3117
2194	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	3118	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
2195		3119
2196	Fork watchers are called when a C<fork ()> was detected (usually because	3120	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
2199	event loop blocks next and before C<ev_check> watchers are being called,	3123	event loop blocks next and before C<ev_check> watchers are being called,
2200	and only in the child after the fork. If whoever good citizen calling	3124	and only in the child after the fork. If whoever good citizen calling
2201	C<ev_default_fork> cheats and calls it in the wrong process, the fork	3125	C<ev_default_fork> cheats and calls it in the wrong process, the fork
2202	handlers will be invoked, too, of course.	3126	handlers will be invoked, too, of course.
2203		3127
		3128	=head3 The special problem of life after fork - how is it possible?
		3129
		3130	Most uses of C<fork()> consist of forking, then some simple calls to set
		3131	up/change the process environment, followed by a call to C<exec()>. This
		3132	sequence should be handled by libev without any problems.
		3133
		3134	This changes when the application actually wants to do event handling
		3135	in the child, or both parent in child, in effect "continuing" after the
		3136	fork.
		3137
		3138	The default mode of operation (for libev, with application help to detect
		3139	forks) is to duplicate all the state in the child, as would be expected
		3140	when I<either> the parent I<or> the child process continues.
		3141
		3142	When both processes want to continue using libev, then this is usually the
		3143	wrong result. In that case, usually one process (typically the parent) is
		3144	supposed to continue with all watchers in place as before, while the other
		3145	process typically wants to start fresh, i.e. without any active watchers.
		3146
		3147	The cleanest and most efficient way to achieve that with libev is to
		3148	simply create a new event loop, which of course will be "empty", and
		3149	use that for new watchers. This has the advantage of not touching more
		3150	memory than necessary, and thus avoiding the copy-on-write, and the
		3151	disadvantage of having to use multiple event loops (which do not support
		3152	signal watchers).
		3153
		3154	When this is not possible, or you want to use the default loop for
		3155	other reasons, then in the process that wants to start "fresh", call
		3156	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3157	Destroying the default loop will "orphan" (not stop) all registered
		3158	watchers, so you have to be careful not to execute code that modifies
		3159	those watchers. Note also that in that case, you have to re-register any
		3160	signal watchers.
		3161
2204	=head3 Watcher-Specific Functions and Data Members	3162	=head3 Watcher-Specific Functions and Data Members
2205		3163
2206	=over 4	3164	=over 4
2207		3165
2208	=item ev_fork_init (ev_signal *, callback)	3166	=item ev_fork_init (ev_fork *, callback)
2209		3167
2210	Initialises and configures the fork watcher - it has no parameters of any	3168	Initialises and configures the fork watcher - it has no parameters of any
2211	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,	3169	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
2212	believe me.	3170	really.
2213		3171
2214	=back	3172	=back
2215		3173
2216		3174
		3175	=head2 C<ev_cleanup> - even the best things end
		3176
		3177	Cleanup watchers are called just before the event loop is being destroyed
		3178	by a call to C<ev_loop_destroy>.
		3179
		3180	While there is no guarantee that the event loop gets destroyed, cleanup
		3181	watchers provide a convenient method to install cleanup hooks for your
		3182	program, worker threads and so on - you just to make sure to destroy the
		3183	loop when you want them to be invoked.
		3184
		3185	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3186	all other watchers, they do not keep a reference to the event loop (which
		3187	makes a lot of sense if you think about it). Like all other watchers, you
		3188	can call libev functions in the callback, except C<ev_cleanup_start>.
		3189
		3190	=head3 Watcher-Specific Functions and Data Members
		3191
		3192	=over 4
		3193
		3194	=item ev_cleanup_init (ev_cleanup *, callback)
		3195
		3196	Initialises and configures the cleanup watcher - it has no parameters of
		3197	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3198	pointless, I assure you.
		3199
		3200	=back
		3201
		3202	Example: Register an atexit handler to destroy the default loop, so any
		3203	cleanup functions are called.
		3204
		3205	static void
		3206	program_exits (void)
		3207	{
		3208	ev_loop_destroy (EV_DEFAULT_UC);
		3209	}
		3210
		3211	...
		3212	atexit (program_exits);
		3213
		3214
2217	=head2 C<ev_async> - how to wake up another event loop	3215	=head2 C<ev_async> - how to wake up an event loop
2218		3216
2219	In general, you cannot use an C<ev_loop> from multiple threads or other	3217	In general, you cannot use an C<ev_run> from multiple threads or other
2220	asynchronous sources such as signal handlers (as opposed to multiple event	3218	asynchronous sources such as signal handlers (as opposed to multiple event
2221	loops - those are of course safe to use in different threads).	3219	loops - those are of course safe to use in different threads).
2222		3220
2223	Sometimes, however, you need to wake up another event loop you do not	3221	Sometimes, however, you need to wake up an event loop you do not control,
2224	control, for example because it belongs to another thread. This is what	3222	for example because it belongs to another thread. This is what C<ev_async>
2225	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you	3223	watchers do: as long as the C<ev_async> watcher is active, you can signal
2226	can signal it by calling C<ev_async_send>, which is thread- and signal	3224	it by calling C<ev_async_send>, which is thread- and signal safe.
2227	safe.
2228		3225
2229	This functionality is very similar to C<ev_signal> watchers, as signals,	3226	This functionality is very similar to C<ev_signal> watchers, as signals,
2230	too, are asynchronous in nature, and signals, too, will be compressed	3227	too, are asynchronous in nature, and signals, too, will be compressed
2231	(i.e. the number of callback invocations may be less than the number of	3228	(i.e. the number of callback invocations may be less than the number of
2232	C<ev_async_sent> calls).	3229	C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
		3230	of "global async watchers" by using a watcher on an otherwise unused
		3231	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3232	even without knowing which loop owns the signal.
2233		3233
2234	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not	3234	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
2235	just the default loop.	3235	just the default loop.
2236		3236
2237	=head3 Queueing	3237	=head3 Queueing
2238		3238
2239	C<ev_async> does not support queueing of data in any way. The reason	3239	C<ev_async> does not support queueing of data in any way. The reason
2240	is that the author does not know of a simple (or any) algorithm for a	3240	is that the author does not know of a simple (or any) algorithm for a
2241	multiple-writer-single-reader queue that works in all cases and doesn't	3241	multiple-writer-single-reader queue that works in all cases and doesn't
2242	need elaborate support such as pthreads.	3242	need elaborate support such as pthreads or unportable memory access
		3243	semantics.
2243		3244
2244	That means that if you want to queue data, you have to provide your own	3245	That means that if you want to queue data, you have to provide your own
2245	queue. But at least I can tell you would implement locking around your	3246	queue. But at least I can tell you how to implement locking around your
2246	queue:	3247	queue:
2247		3248
2248	=over 4	3249	=over 4
2249		3250
2250	=item queueing from a signal handler context	3251	=item queueing from a signal handler context
2251		3252
2252	To implement race-free queueing, you simply add to the queue in the signal	3253	To implement race-free queueing, you simply add to the queue in the signal
2253	handler but you block the signal handler in the watcher callback. Here is an example that does that for	3254	handler but you block the signal handler in the watcher callback. Here is
2254	some fictitious SIGUSR1 handler:	3255	an example that does that for some fictitious SIGUSR1 handler:
2255		3256
2256	static ev_async mysig;	3257	static ev_async mysig;
2257		3258
2258	static void	3259	static void
2259	sigusr1_handler (void)	3260	sigusr1_handler (void)
…		…
2325	=over 4	3326	=over 4
2326		3327
2327	=item ev_async_init (ev_async *, callback)	3328	=item ev_async_init (ev_async *, callback)
2328		3329
2329	Initialises and configures the async watcher - it has no parameters of any	3330	Initialises and configures the async watcher - it has no parameters of any
2330	kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,	3331	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
2331	believe me.	3332	trust me.
2332		3333
2333	=item ev_async_send (loop, ev_async *)	3334	=item ev_async_send (loop, ev_async *)
2334		3335
2335	Sends/signals/activates the given C<ev_async> watcher, that is, feeds	3336	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
2336	an C<EV_ASYNC> event on the watcher into the event loop. Unlike	3337	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
2337	C<ev_feed_event>, this call is safe to do in other threads, signal or	3338	C<ev_feed_event>, this call is safe to do from other threads, signal or
2338	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding	3339	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
2339	section below on what exactly this means).	3340	section below on what exactly this means).
2340		3341
		3342	Note that, as with other watchers in libev, multiple events might get
		3343	compressed into a single callback invocation (another way to look at this
		3344	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3345	reset when the event loop detects that).
		3346
2341	This call incurs the overhead of a system call only once per loop iteration,	3347	This call incurs the overhead of a system call only once per event loop
2342	so while the overhead might be noticeable, it doesn't apply to repeated	3348	iteration, so while the overhead might be noticeable, it doesn't apply to
2343	calls to C<ev_async_send>.	3349	repeated calls to C<ev_async_send> for the same event loop.
2344		3350
2345	=item bool = ev_async_pending (ev_async *)	3351	=item bool = ev_async_pending (ev_async *)
2346		3352
2347	Returns a non-zero value when C<ev_async_send> has been called on the	3353	Returns a non-zero value when C<ev_async_send> has been called on the
2348	watcher but the event has not yet been processed (or even noted) by the	3354	watcher but the event has not yet been processed (or even noted) by the
…		…
2351	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When	3357	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
2352	the loop iterates next and checks for the watcher to have become active,	3358	the loop iterates next and checks for the watcher to have become active,
2353	it will reset the flag again. C<ev_async_pending> can be used to very	3359	it will reset the flag again. C<ev_async_pending> can be used to very
2354	quickly check whether invoking the loop might be a good idea.	3360	quickly check whether invoking the loop might be a good idea.
2355		3361
2356	Not that this does I<not> check whether the watcher itself is pending, only	3362	Not that this does I<not> check whether the watcher itself is pending,
2357	whether it has been requested to make this watcher pending.	3363	only whether it has been requested to make this watcher pending: there
		3364	is a time window between the event loop checking and resetting the async
		3365	notification, and the callback being invoked.
2358		3366
2359	=back	3367	=back
2360		3368
2361		3369
2362	=head1 OTHER FUNCTIONS	3370	=head1 OTHER FUNCTIONS
…		…
2366	=over 4	3374	=over 4
2367		3375
2368	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3376	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
2369		3377
2370	This function combines a simple timer and an I/O watcher, calls your	3378	This function combines a simple timer and an I/O watcher, calls your
2371	callback on whichever event happens first and automatically stop both	3379	callback on whichever event happens first and automatically stops both
2372	watchers. This is useful if you want to wait for a single event on an fd	3380	watchers. This is useful if you want to wait for a single event on an fd
2373	or timeout without having to allocate/configure/start/stop/free one or	3381	or timeout without having to allocate/configure/start/stop/free one or
2374	more watchers yourself.	3382	more watchers yourself.
2375		3383
2376	If C<fd> is less than 0, then no I/O watcher will be started and events	3384	If C<fd> is less than 0, then no I/O watcher will be started and the
2377	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3385	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
2378	C<events> set will be created and started.	3386	the given C<fd> and C<events> set will be created and started.
2379		3387
2380	If C<timeout> is less than 0, then no timeout watcher will be	3388	If C<timeout> is less than 0, then no timeout watcher will be
2381	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3389	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
2382	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3390	repeat = 0) will be started. C<0> is a valid timeout.
2383	dubious value.
2384		3391
2385	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3392	The callback has the type C<void (*cb)(int revents, void *arg)> and is
2386	passed an C<revents> set like normal event callbacks (a combination of	3393	passed an C<revents> set like normal event callbacks (a combination of
2387	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3394	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
2388	value passed to C<ev_once>:	3395	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3396	a timeout and an io event at the same time - you probably should give io
		3397	events precedence.
2389		3398
		3399	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3400
2390	static void stdin_ready (int revents, void *arg)	3401	static void stdin_ready (int revents, void *arg)
2391	{	3402	{
2392	if (revents & EV_TIMEOUT)
2393	/* doh, nothing entered */;
2394	else if (revents & EV_READ)	3403	if (revents & EV_READ)
2395	/* stdin might have data for us, joy! */;	3404	/* stdin might have data for us, joy! */;
		3405	else if (revents & EV_TIMER)
		3406	/* doh, nothing entered */;
2396	}	3407	}
2397		3408
2398	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3409	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
2399		3410
2400	=item ev_feed_event (ev_loop *, watcher *, int revents)
2401
2402	Feeds the given event set into the event loop, as if the specified event
2403	had happened for the specified watcher (which must be a pointer to an
2404	initialised but not necessarily started event watcher).
2405
2406	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	3411	=item ev_feed_fd_event (loop, int fd, int revents)
2407		3412
2408	Feed an event on the given fd, as if a file descriptor backend detected	3413	Feed an event on the given fd, as if a file descriptor backend detected
2409	the given events it.	3414	the given events it.
2410		3415
2411	=item ev_feed_signal_event (ev_loop *loop, int signum)	3416	=item ev_feed_signal_event (loop, int signum)
2412		3417
2413	Feed an event as if the given signal occurred (C<loop> must be the default	3418	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
2414	loop!).	3419	which is async-safe.
		3420
		3421	=back
		3422
		3423
		3424	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3425
		3426	This section explains some common idioms that are not immediately
		3427	obvious. Note that examples are sprinkled over the whole manual, and this
		3428	section only contains stuff that wouldn't fit anywhere else.
		3429
		3430	=over 4
		3431
		3432	=item Model/nested event loop invocations and exit conditions.
		3433
		3434	Often (especially in GUI toolkits) there are places where you have
		3435	I<modal> interaction, which is most easily implemented by recursively
		3436	invoking C<ev_run>.
		3437
		3438	This brings the problem of exiting - a callback might want to finish the
		3439	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3440	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3441	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3442	other combination: In these cases, C<ev_break> will not work alone.
		3443
		3444	The solution is to maintain "break this loop" variable for each C<ev_run>
		3445	invocation, and use a loop around C<ev_run> until the condition is
		3446	triggered, using C<EVRUN_ONCE>:
		3447
		3448	// main loop
		3449	int exit_main_loop = 0;
		3450
		3451	while (!exit_main_loop)
		3452	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3453
		3454	// in a model watcher
		3455	int exit_nested_loop = 0;
		3456
		3457	while (!exit_nested_loop)
		3458	ev_run (EV_A_ EVRUN_ONCE);
		3459
		3460	To exit from any of these loops, just set the corresponding exit variable:
		3461
		3462	// exit modal loop
		3463	exit_nested_loop = 1;
		3464
		3465	// exit main program, after modal loop is finished
		3466	exit_main_loop = 1;
		3467
		3468	// exit both
		3469	exit_main_loop = exit_nested_loop = 1;
2415		3470
2416	=back	3471	=back
2417		3472
2418		3473
2419	=head1 LIBEVENT EMULATION	3474	=head1 LIBEVENT EMULATION
2420		3475
2421	Libev offers a compatibility emulation layer for libevent. It cannot	3476	Libev offers a compatibility emulation layer for libevent. It cannot
2422	emulate the internals of libevent, so here are some usage hints:	3477	emulate the internals of libevent, so here are some usage hints:
2423		3478
2424	=over 4	3479	=over 4
		3480
		3481	=item * Only the libevent-1.4.1-beta API is being emulated.
		3482
		3483	This was the newest libevent version available when libev was implemented,
		3484	and is still mostly unchanged in 2010.
2425		3485
2426	=item * Use it by including <event.h>, as usual.	3486	=item * Use it by including <event.h>, as usual.
2427		3487
2428	=item * The following members are fully supported: ev_base, ev_callback,	3488	=item * The following members are fully supported: ev_base, ev_callback,
2429	ev_arg, ev_fd, ev_res, ev_events.	3489	ev_arg, ev_fd, ev_res, ev_events.
…		…
2435	=item * Priorities are not currently supported. Initialising priorities	3495	=item * Priorities are not currently supported. Initialising priorities
2436	will fail and all watchers will have the same priority, even though there	3496	will fail and all watchers will have the same priority, even though there
2437	is an ev_pri field.	3497	is an ev_pri field.
2438		3498
2439	=item * In libevent, the last base created gets the signals, in libev, the	3499	=item * In libevent, the last base created gets the signals, in libev, the
2440	first base created (== the default loop) gets the signals.	3500	base that registered the signal gets the signals.
2441		3501
2442	=item * Other members are not supported.	3502	=item * Other members are not supported.
2443		3503
2444	=item * The libev emulation is I<not> ABI compatible to libevent, you need	3504	=item * The libev emulation is I<not> ABI compatible to libevent, you need
2445	to use the libev header file and library.	3505	to use the libev header file and library.
…		…
2452	you to use some convenience methods to start/stop watchers and also change	3512	you to use some convenience methods to start/stop watchers and also change
2453	the callback model to a model using method callbacks on objects.	3513	the callback model to a model using method callbacks on objects.
2454		3514
2455	To use it,	3515	To use it,
2456		3516
2457	#include <ev++.h>	3517	#include <ev++.h>
2458		3518
2459	This automatically includes F<ev.h> and puts all of its definitions (many	3519	This automatically includes F<ev.h> and puts all of its definitions (many
2460	of them macros) into the global namespace. All C++ specific things are	3520	of them macros) into the global namespace. All C++ specific things are
2461	put into the C<ev> namespace. It should support all the same embedding	3521	put into the C<ev> namespace. It should support all the same embedding
2462	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.	3522	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
…		…
2464	Care has been taken to keep the overhead low. The only data member the C++	3524	Care has been taken to keep the overhead low. The only data member the C++
2465	classes add (compared to plain C-style watchers) is the event loop pointer	3525	classes add (compared to plain C-style watchers) is the event loop pointer
2466	that the watcher is associated with (or no additional members at all if	3526	that the watcher is associated with (or no additional members at all if
2467	you disable C<EV_MULTIPLICITY> when embedding libev).	3527	you disable C<EV_MULTIPLICITY> when embedding libev).
2468		3528
2469	Currently, functions, and static and non-static member functions can be	3529	Currently, functions, static and non-static member functions and classes
2470	used as callbacks. Other types should be easy to add as long as they only	3530	with C<operator ()> can be used as callbacks. Other types should be easy
2471	need one additional pointer for context. If you need support for other	3531	to add as long as they only need one additional pointer for context. If
2472	types of functors please contact the author (preferably after implementing	3532	you need support for other types of functors please contact the author
2473	it).	3533	(preferably after implementing it).
2474		3534
2475	Here is a list of things available in the C<ev> namespace:	3535	Here is a list of things available in the C<ev> namespace:
2476		3536
2477	=over 4	3537	=over 4
2478		3538
…		…
2496		3556
2497	=over 4	3557	=over 4
2498		3558
2499	=item ev::TYPE::TYPE ()	3559	=item ev::TYPE::TYPE ()
2500		3560
2501	=item ev::TYPE::TYPE (struct ev_loop *)	3561	=item ev::TYPE::TYPE (loop)
2502		3562
2503	=item ev::TYPE::~TYPE	3563	=item ev::TYPE::~TYPE
2504		3564
2505	The constructor (optionally) takes an event loop to associate the watcher	3565	The constructor (optionally) takes an event loop to associate the watcher
2506	with. If it is omitted, it will use C<EV_DEFAULT>.	3566	with. If it is omitted, it will use C<EV_DEFAULT>.
…		…
2529	your compiler is good :), then the method will be fully inlined into the	3589	your compiler is good :), then the method will be fully inlined into the
2530	thunking function, making it as fast as a direct C callback.	3590	thunking function, making it as fast as a direct C callback.
2531		3591
2532	Example: simple class declaration and watcher initialisation	3592	Example: simple class declaration and watcher initialisation
2533		3593
2534	struct myclass	3594	struct myclass
2535	{	3595	{
2536	void io_cb (ev::io &w, int revents) { }	3596	void io_cb (ev::io &w, int revents) { }
2537	}	3597	}
2538		3598
2539	myclass obj;	3599	myclass obj;
2540	ev::io iow;	3600	ev::io iow;
2541	iow.set <myclass, &myclass::io_cb> (&obj);	3601	iow.set <myclass, &myclass::io_cb> (&obj);
		3602
		3603	=item w->set (object *)
		3604
		3605	This is a variation of a method callback - leaving out the method to call
		3606	will default the method to C<operator ()>, which makes it possible to use
		3607	functor objects without having to manually specify the C<operator ()> all
		3608	the time. Incidentally, you can then also leave out the template argument
		3609	list.
		3610
		3611	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3612	int revents)>.
		3613
		3614	See the method-C<set> above for more details.
		3615
		3616	Example: use a functor object as callback.
		3617
		3618	struct myfunctor
		3619	{
		3620	void operator() (ev::io &w, int revents)
		3621	{
		3622	...
		3623	}
		3624	}
		3625
		3626	myfunctor f;
		3627
		3628	ev::io w;
		3629	w.set (&f);
2542		3630
2543	=item w->set<function> (void *data = 0)	3631	=item w->set<function> (void *data = 0)
2544		3632
2545	Also sets a callback, but uses a static method or plain function as	3633	Also sets a callback, but uses a static method or plain function as
2546	callback. The optional C<data> argument will be stored in the watcher's	3634	callback. The optional C<data> argument will be stored in the watcher's
…		…
2548		3636
2549	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.	3637	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
2550		3638
2551	See the method-C<set> above for more details.	3639	See the method-C<set> above for more details.
2552		3640
2553	Example:	3641	Example: Use a plain function as callback.
2554		3642
2555	static void io_cb (ev::io &w, int revents) { }	3643	static void io_cb (ev::io &w, int revents) { }
2556	iow.set <io_cb> ();	3644	iow.set <io_cb> ();
2557		3645
2558	=item w->set (struct ev_loop *)	3646	=item w->set (loop)
2559		3647
2560	Associates a different C<struct ev_loop> with this watcher. You can only	3648	Associates a different C<struct ev_loop> with this watcher. You can only
2561	do this when the watcher is inactive (and not pending either).	3649	do this when the watcher is inactive (and not pending either).
2562		3650
2563	=item w->set ([arguments])	3651	=item w->set ([arguments])
2564		3652
2565	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be	3653	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this
2566	called at least once. Unlike the C counterpart, an active watcher gets	3654	method or a suitable start method must be called at least once. Unlike the
2567	automatically stopped and restarted when reconfiguring it with this	3655	C counterpart, an active watcher gets automatically stopped and restarted
2568	method.	3656	when reconfiguring it with this method.
2569		3657
2570	=item w->start ()	3658	=item w->start ()
2571		3659
2572	Starts the watcher. Note that there is no C<loop> argument, as the	3660	Starts the watcher. Note that there is no C<loop> argument, as the
2573	constructor already stores the event loop.	3661	constructor already stores the event loop.
2574		3662
		3663	=item w->start ([arguments])
		3664
		3665	Instead of calling C<set> and C<start> methods separately, it is often
		3666	convenient to wrap them in one call. Uses the same type of arguments as
		3667	the configure C<set> method of the watcher.
		3668
2575	=item w->stop ()	3669	=item w->stop ()
2576		3670
2577	Stops the watcher if it is active. Again, no C<loop> argument.	3671	Stops the watcher if it is active. Again, no C<loop> argument.
2578		3672
2579	=item w->again () (C<ev::timer>, C<ev::periodic> only)	3673	=item w->again () (C<ev::timer>, C<ev::periodic> only)
…		…
2591		3685
2592	=back	3686	=back
2593		3687
2594	=back	3688	=back
2595		3689
2596	Example: Define a class with an IO and idle watcher, start one of them in	3690	Example: Define a class with two I/O and idle watchers, start the I/O
2597	the constructor.	3691	watchers in the constructor.
2598		3692
2599	class myclass	3693	class myclass
2600	{	3694	{
2601	ev::io io; void io_cb (ev::io &w, int revents);	3695	ev::io io ; void io_cb (ev::io &w, int revents);
		3696	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);
2602	ev:idle idle void idle_cb (ev::idle &w, int revents);	3697	ev::idle idle; void idle_cb (ev::idle &w, int revents);
2603		3698
2604	myclass (int fd)	3699	myclass (int fd)
2605	{	3700	{
2606	io .set <myclass, &myclass::io_cb > (this);	3701	io .set <myclass, &myclass::io_cb > (this);
		3702	io2 .set <myclass, &myclass::io2_cb > (this);
2607	idle.set <myclass, &myclass::idle_cb> (this);	3703	idle.set <myclass, &myclass::idle_cb> (this);
2608		3704
2609	io.start (fd, ev::READ);	3705	io.set (fd, ev::WRITE); // configure the watcher
		3706	io.start (); // start it whenever convenient
		3707
		3708	io2.start (fd, ev::READ); // set + start in one call
2610	}	3709	}
2611	};	3710	};
2612		3711
2613		3712
2614	=head1 OTHER LANGUAGE BINDINGS	3713	=head1 OTHER LANGUAGE BINDINGS
2615		3714
2616	Libev does not offer other language bindings itself, but bindings for a	3715	Libev does not offer other language bindings itself, but bindings for a
…		…
2623	=item Perl	3722	=item Perl
2624		3723
2625	The EV module implements the full libev API and is actually used to test	3724	The EV module implements the full libev API and is actually used to test
2626	libev. EV is developed together with libev. Apart from the EV core module,	3725	libev. EV is developed together with libev. Apart from the EV core module,
2627	there are additional modules that implement libev-compatible interfaces	3726	there are additional modules that implement libev-compatible interfaces
2628	to C<libadns> (C<EV::ADNS>), C<Net::SNMP> (C<Net::SNMP::EV>) and the	3727	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
2629	C<libglib> event core (C<Glib::EV> and C<EV::Glib>).	3728	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3729	and C<EV::Glib>).
2630		3730
2631	It can be found and installed via CPAN, its homepage is found at	3731	It can be found and installed via CPAN, its homepage is at
2632	L<http://software.schmorp.de/pkg/EV>.	3732	L<http://software.schmorp.de/pkg/EV>.
		3733
		3734	=item Python
		3735
		3736	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3737	seems to be quite complete and well-documented.
2633		3738
2634	=item Ruby	3739	=item Ruby
2635		3740
2636	Tony Arcieri has written a ruby extension that offers access to a subset	3741	Tony Arcieri has written a ruby extension that offers access to a subset
2637	of the libev API and adds file handle abstractions, asynchronous DNS and	3742	of the libev API and adds file handle abstractions, asynchronous DNS and
2638	more on top of it. It can be found via gem servers. Its homepage is at	3743	more on top of it. It can be found via gem servers. Its homepage is at
2639	L<http://rev.rubyforge.org/>.	3744	L<http://rev.rubyforge.org/>.
2640		3745
		3746	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3747	makes rev work even on mingw.
		3748
		3749	=item Haskell
		3750
		3751	A haskell binding to libev is available at
		3752	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3753
2641	=item D	3754	=item D
2642		3755
2643	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to	3756	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
2644	be found at L<http://git.llucax.com.ar/?p=software/ev.d.git;a=summary>.	3757	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3758
		3759	=item Ocaml
		3760
		3761	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3762	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3763
		3764	=item Lua
		3765
		3766	Brian Maher has written a partial interface to libev for lua (at the
		3767	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		3768	L<http://github.com/brimworks/lua-ev>.
2645		3769
2646	=back	3770	=back
2647		3771
2648		3772
2649	=head1 MACRO MAGIC	3773	=head1 MACRO MAGIC
…		…
2661		3785
2662	This provides the loop I<argument> for functions, if one is required ("ev	3786	This provides the loop I<argument> for functions, if one is required ("ev
2663	loop argument"). The C<EV_A> form is used when this is the sole argument,	3787	loop argument"). The C<EV_A> form is used when this is the sole argument,
2664	C<EV_A_> is used when other arguments are following. Example:	3788	C<EV_A_> is used when other arguments are following. Example:
2665		3789
2666	ev_unref (EV_A);	3790	ev_unref (EV_A);
2667	ev_timer_add (EV_A_ watcher);	3791	ev_timer_add (EV_A_ watcher);
2668	ev_loop (EV_A_ 0);	3792	ev_run (EV_A_ 0);
2669		3793
2670	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	3794	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
2671	which is often provided by the following macro.	3795	which is often provided by the following macro.
2672		3796
2673	=item C<EV_P>, C<EV_P_>	3797	=item C<EV_P>, C<EV_P_>
2674		3798
2675	This provides the loop I<parameter> for functions, if one is required ("ev	3799	This provides the loop I<parameter> for functions, if one is required ("ev
2676	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	3800	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
2677	C<EV_P_> is used when other parameters are following. Example:	3801	C<EV_P_> is used when other parameters are following. Example:
2678		3802
2679	// this is how ev_unref is being declared	3803	// this is how ev_unref is being declared
2680	static void ev_unref (EV_P);	3804	static void ev_unref (EV_P);
2681		3805
2682	// this is how you can declare your typical callback	3806	// this is how you can declare your typical callback
2683	static void cb (EV_P_ ev_timer *w, int revents)	3807	static void cb (EV_P_ ev_timer *w, int revents)
2684		3808
2685	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	3809	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
2686	suitable for use with C<EV_A>.	3810	suitable for use with C<EV_A>.
2687		3811
2688	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	3812	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
…		…
2704		3828
2705	Example: Declare and initialise a check watcher, utilising the above	3829	Example: Declare and initialise a check watcher, utilising the above
2706	macros so it will work regardless of whether multiple loops are supported	3830	macros so it will work regardless of whether multiple loops are supported
2707	or not.	3831	or not.
2708		3832
2709	static void	3833	static void
2710	check_cb (EV_P_ ev_timer *w, int revents)	3834	check_cb (EV_P_ ev_timer *w, int revents)
2711	{	3835	{
2712	ev_check_stop (EV_A_ w);	3836	ev_check_stop (EV_A_ w);
2713	}	3837	}
2714		3838
2715	ev_check check;	3839	ev_check check;
2716	ev_check_init (&check, check_cb);	3840	ev_check_init (&check, check_cb);
2717	ev_check_start (EV_DEFAULT_ &check);	3841	ev_check_start (EV_DEFAULT_ &check);
2718	ev_loop (EV_DEFAULT_ 0);	3842	ev_run (EV_DEFAULT_ 0);
2719		3843
2720	=head1 EMBEDDING	3844	=head1 EMBEDDING
2721		3845
2722	Libev can (and often is) directly embedded into host	3846	Libev can (and often is) directly embedded into host
2723	applications. Examples of applications that embed it include the Deliantra	3847	applications. Examples of applications that embed it include the Deliantra
…		…
2737	=head3 CORE EVENT LOOP	3861	=head3 CORE EVENT LOOP
2738		3862
2739	To include only the libev core (all the C<ev_*> functions), with manual	3863	To include only the libev core (all the C<ev_*> functions), with manual
2740	configuration (no autoconf):	3864	configuration (no autoconf):
2741		3865
2742	#define EV_STANDALONE 1	3866	#define EV_STANDALONE 1
2743	#include "ev.c"	3867	#include "ev.c"
2744		3868
2745	This will automatically include F<ev.h>, too, and should be done in a	3869	This will automatically include F<ev.h>, too, and should be done in a
2746	single C source file only to provide the function implementations. To use	3870	single C source file only to provide the function implementations. To use
2747	it, do the same for F<ev.h> in all files wishing to use this API (best	3871	it, do the same for F<ev.h> in all files wishing to use this API (best
2748	done by writing a wrapper around F<ev.h> that you can include instead and	3872	done by writing a wrapper around F<ev.h> that you can include instead and
2749	where you can put other configuration options):	3873	where you can put other configuration options):
2750		3874
2751	#define EV_STANDALONE 1	3875	#define EV_STANDALONE 1
2752	#include "ev.h"	3876	#include "ev.h"
2753		3877
2754	Both header files and implementation files can be compiled with a C++	3878	Both header files and implementation files can be compiled with a C++
2755	compiler (at least, thats a stated goal, and breakage will be treated	3879	compiler (at least, that's a stated goal, and breakage will be treated
2756	as a bug).	3880	as a bug).
2757		3881
2758	You need the following files in your source tree, or in a directory	3882	You need the following files in your source tree, or in a directory
2759	in your include path (e.g. in libev/ when using -Ilibev):	3883	in your include path (e.g. in libev/ when using -Ilibev):
2760		3884
2761	ev.h	3885	ev.h
2762	ev.c	3886	ev.c
2763	ev_vars.h	3887	ev_vars.h
2764	ev_wrap.h	3888	ev_wrap.h
2765		3889
2766	ev_win32.c required on win32 platforms only	3890	ev_win32.c required on win32 platforms only
2767		3891
2768	ev_select.c only when select backend is enabled (which is enabled by default)	3892	ev_select.c only when select backend is enabled (which is enabled by default)
2769	ev_poll.c only when poll backend is enabled (disabled by default)	3893	ev_poll.c only when poll backend is enabled (disabled by default)
2770	ev_epoll.c only when the epoll backend is enabled (disabled by default)	3894	ev_epoll.c only when the epoll backend is enabled (disabled by default)
2771	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	3895	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2772	ev_port.c only when the solaris port backend is enabled (disabled by default)	3896	ev_port.c only when the solaris port backend is enabled (disabled by default)
2773		3897
2774	F<ev.c> includes the backend files directly when enabled, so you only need	3898	F<ev.c> includes the backend files directly when enabled, so you only need
2775	to compile this single file.	3899	to compile this single file.
2776		3900
2777	=head3 LIBEVENT COMPATIBILITY API	3901	=head3 LIBEVENT COMPATIBILITY API
2778		3902
2779	To include the libevent compatibility API, also include:	3903	To include the libevent compatibility API, also include:
2780		3904
2781	#include "event.c"	3905	#include "event.c"
2782		3906
2783	in the file including F<ev.c>, and:	3907	in the file including F<ev.c>, and:
2784		3908
2785	#include "event.h"	3909	#include "event.h"
2786		3910
2787	in the files that want to use the libevent API. This also includes F<ev.h>.	3911	in the files that want to use the libevent API. This also includes F<ev.h>.
2788		3912
2789	You need the following additional files for this:	3913	You need the following additional files for this:
2790		3914
2791	event.h	3915	event.h
2792	event.c	3916	event.c
2793		3917
2794	=head3 AUTOCONF SUPPORT	3918	=head3 AUTOCONF SUPPORT
2795		3919
2796	Instead of using C<EV_STANDALONE=1> and providing your configuration in	3920	Instead of using C<EV_STANDALONE=1> and providing your configuration in
2797	whatever way you want, you can also C<m4_include([libev.m4])> in your	3921	whatever way you want, you can also C<m4_include([libev.m4])> in your
2798	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	3922	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
2799	include F<config.h> and configure itself accordingly.	3923	include F<config.h> and configure itself accordingly.
2800		3924
2801	For this of course you need the m4 file:	3925	For this of course you need the m4 file:
2802		3926
2803	libev.m4	3927	libev.m4
2804		3928
2805	=head2 PREPROCESSOR SYMBOLS/MACROS	3929	=head2 PREPROCESSOR SYMBOLS/MACROS
2806		3930
2807	Libev can be configured via a variety of preprocessor symbols you have to	3931	Libev can be configured via a variety of preprocessor symbols you have to
2808	define before including any of its files. The default in the absence of	3932	define before including (or compiling) any of its files. The default in
2809	autoconf is noted for every option.	3933	the absence of autoconf is documented for every option.
		3934
		3935	Symbols marked with "(h)" do not change the ABI, and can have different
		3936	values when compiling libev vs. including F<ev.h>, so it is permissible
		3937	to redefine them before including F<ev.h> without breaking compatibility
		3938	to a compiled library. All other symbols change the ABI, which means all
		3939	users of libev and the libev code itself must be compiled with compatible
		3940	settings.
2810		3941
2811	=over 4	3942	=over 4
2812		3943
		3944	=item EV_COMPAT3 (h)
		3945
		3946	Backwards compatibility is a major concern for libev. This is why this
		3947	release of libev comes with wrappers for the functions and symbols that
		3948	have been renamed between libev version 3 and 4.
		3949
		3950	You can disable these wrappers (to test compatibility with future
		3951	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		3952	sources. This has the additional advantage that you can drop the C<struct>
		3953	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		3954	typedef in that case.
		3955
		3956	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		3957	and in some even more future version the compatibility code will be
		3958	removed completely.
		3959
2813	=item EV_STANDALONE	3960	=item EV_STANDALONE (h)
2814		3961
2815	Must always be C<1> if you do not use autoconf configuration, which	3962	Must always be C<1> if you do not use autoconf configuration, which
2816	keeps libev from including F<config.h>, and it also defines dummy	3963	keeps libev from including F<config.h>, and it also defines dummy
2817	implementations for some libevent functions (such as logging, which is not	3964	implementations for some libevent functions (such as logging, which is not
2818	supported). It will also not define any of the structs usually found in	3965	supported). It will also not define any of the structs usually found in
2819	F<event.h> that are not directly supported by the libev core alone.	3966	F<event.h> that are not directly supported by the libev core alone.
2820		3967
		3968	In standalone mode, libev will still try to automatically deduce the
		3969	configuration, but has to be more conservative.
		3970
2821	=item EV_USE_MONOTONIC	3971	=item EV_USE_MONOTONIC
2822		3972
2823	If defined to be C<1>, libev will try to detect the availability of the	3973	If defined to be C<1>, libev will try to detect the availability of the
2824	monotonic clock option at both compile time and runtime. Otherwise no use	3974	monotonic clock option at both compile time and runtime. Otherwise no
2825	of the monotonic clock option will be attempted. If you enable this, you	3975	use of the monotonic clock option will be attempted. If you enable this,
2826	usually have to link against librt or something similar. Enabling it when	3976	you usually have to link against librt or something similar. Enabling it
2827	the functionality isn't available is safe, though, although you have	3977	when the functionality isn't available is safe, though, although you have
2828	to make sure you link against any libraries where the C<clock_gettime>	3978	to make sure you link against any libraries where the C<clock_gettime>
2829	function is hiding in (often F<-lrt>).	3979	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
2830		3980
2831	=item EV_USE_REALTIME	3981	=item EV_USE_REALTIME
2832		3982
2833	If defined to be C<1>, libev will try to detect the availability of the	3983	If defined to be C<1>, libev will try to detect the availability of the
2834	real-time clock option at compile time (and assume its availability at	3984	real-time clock option at compile time (and assume its availability
2835	runtime if successful). Otherwise no use of the real-time clock option will	3985	at runtime if successful). Otherwise no use of the real-time clock
2836	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	3986	option will be attempted. This effectively replaces C<gettimeofday>
2837	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the	3987	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
2838	note about libraries in the description of C<EV_USE_MONOTONIC>, though.	3988	correctness. See the note about libraries in the description of
		3989	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3990	C<EV_USE_CLOCK_SYSCALL>.
		3991
		3992	=item EV_USE_CLOCK_SYSCALL
		3993
		3994	If defined to be C<1>, libev will try to use a direct syscall instead
		3995	of calling the system-provided C<clock_gettime> function. This option
		3996	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3997	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3998	programs needlessly. Using a direct syscall is slightly slower (in
		3999	theory), because no optimised vdso implementation can be used, but avoids
		4000	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4001	higher, as it simplifies linking (no need for C<-lrt>).
2839		4002
2840	=item EV_USE_NANOSLEEP	4003	=item EV_USE_NANOSLEEP
2841		4004
2842	If defined to be C<1>, libev will assume that C<nanosleep ()> is available	4005	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
2843	and will use it for delays. Otherwise it will use C<select ()>.	4006	and will use it for delays. Otherwise it will use C<select ()>.
…		…
2859		4022
2860	=item EV_SELECT_USE_FD_SET	4023	=item EV_SELECT_USE_FD_SET
2861		4024
2862	If defined to C<1>, then the select backend will use the system C<fd_set>	4025	If defined to C<1>, then the select backend will use the system C<fd_set>
2863	structure. This is useful if libev doesn't compile due to a missing	4026	structure. This is useful if libev doesn't compile due to a missing
2864	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout on	4027	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
2865	exotic systems. This usually limits the range of file descriptors to some	4028	on exotic systems. This usually limits the range of file descriptors to
2866	low limit such as 1024 or might have other limitations (winsocket only	4029	some low limit such as 1024 or might have other limitations (winsocket
2867	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	4030	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
2868	influence the size of the C<fd_set> used.	4031	configures the maximum size of the C<fd_set>.
2869		4032
2870	=item EV_SELECT_IS_WINSOCKET	4033	=item EV_SELECT_IS_WINSOCKET
2871		4034
2872	When defined to C<1>, the select backend will assume that	4035	When defined to C<1>, the select backend will assume that
2873	select/socket/connect etc. don't understand file descriptors but	4036	select/socket/connect etc. don't understand file descriptors but
…		…
2875	be used is the winsock select). This means that it will call	4038	be used is the winsock select). This means that it will call
2876	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	4039	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2877	it is assumed that all these functions actually work on fds, even	4040	it is assumed that all these functions actually work on fds, even
2878	on win32. Should not be defined on non-win32 platforms.	4041	on win32. Should not be defined on non-win32 platforms.
2879		4042
2880	=item EV_FD_TO_WIN32_HANDLE	4043	=item EV_FD_TO_WIN32_HANDLE(fd)
2881		4044
2882	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map	4045	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
2883	file descriptors to socket handles. When not defining this symbol (the	4046	file descriptors to socket handles. When not defining this symbol (the
2884	default), then libev will call C<_get_osfhandle>, which is usually	4047	default), then libev will call C<_get_osfhandle>, which is usually
2885	correct. In some cases, programs use their own file descriptor management,	4048	correct. In some cases, programs use their own file descriptor management,
2886	in which case they can provide this function to map fds to socket handles.	4049	in which case they can provide this function to map fds to socket handles.
		4050
		4051	=item EV_WIN32_HANDLE_TO_FD(handle)
		4052
		4053	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4054	using the standard C<_open_osfhandle> function. For programs implementing
		4055	their own fd to handle mapping, overwriting this function makes it easier
		4056	to do so. This can be done by defining this macro to an appropriate value.
		4057
		4058	=item EV_WIN32_CLOSE_FD(fd)
		4059
		4060	If programs implement their own fd to handle mapping on win32, then this
		4061	macro can be used to override the C<close> function, useful to unregister
		4062	file descriptors again. Note that the replacement function has to close
		4063	the underlying OS handle.
2887		4064
2888	=item EV_USE_POLL	4065	=item EV_USE_POLL
2889		4066
2890	If defined to be C<1>, libev will compile in support for the C<poll>(2)	4067	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2891	backend. Otherwise it will be enabled on non-win32 platforms. It	4068	backend. Otherwise it will be enabled on non-win32 platforms. It
…		…
2938	as well as for signal and thread safety in C<ev_async> watchers.	4115	as well as for signal and thread safety in C<ev_async> watchers.
2939		4116
2940	In the absence of this define, libev will use C<sig_atomic_t volatile>	4117	In the absence of this define, libev will use C<sig_atomic_t volatile>
2941	(from F<signal.h>), which is usually good enough on most platforms.	4118	(from F<signal.h>), which is usually good enough on most platforms.
2942		4119
2943	=item EV_H	4120	=item EV_H (h)
2944		4121
2945	The name of the F<ev.h> header file used to include it. The default if	4122	The name of the F<ev.h> header file used to include it. The default if
2946	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be	4123	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2947	used to virtually rename the F<ev.h> header file in case of conflicts.	4124	used to virtually rename the F<ev.h> header file in case of conflicts.
2948		4125
2949	=item EV_CONFIG_H	4126	=item EV_CONFIG_H (h)
2950		4127
2951	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	4128	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2952	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	4129	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2953	C<EV_H>, above.	4130	C<EV_H>, above.
2954		4131
2955	=item EV_EVENT_H	4132	=item EV_EVENT_H (h)
2956		4133
2957	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	4134	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2958	of how the F<event.h> header can be found, the default is C<"event.h">.	4135	of how the F<event.h> header can be found, the default is C<"event.h">.
2959		4136
2960	=item EV_PROTOTYPES	4137	=item EV_PROTOTYPES (h)
2961		4138
2962	If defined to be C<0>, then F<ev.h> will not define any function	4139	If defined to be C<0>, then F<ev.h> will not define any function
2963	prototypes, but still define all the structs and other symbols. This is	4140	prototypes, but still define all the structs and other symbols. This is
2964	occasionally useful if you want to provide your own wrapper functions	4141	occasionally useful if you want to provide your own wrapper functions
2965	around libev functions.	4142	around libev functions.
…		…
2984	When doing priority-based operations, libev usually has to linearly search	4161	When doing priority-based operations, libev usually has to linearly search
2985	all the priorities, so having many of them (hundreds) uses a lot of space	4162	all the priorities, so having many of them (hundreds) uses a lot of space
2986	and time, so using the defaults of five priorities (-2 .. +2) is usually	4163	and time, so using the defaults of five priorities (-2 .. +2) is usually
2987	fine.	4164	fine.
2988		4165
2989	If your embedding application does not need any priorities, defining these both to	4166	If your embedding application does not need any priorities, defining these
2990	C<0> will save some memory and CPU.	4167	both to C<0> will save some memory and CPU.
2991		4168
2992	=item EV_PERIODIC_ENABLE	4169	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4170	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4171	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
2993		4172
2994	If undefined or defined to be C<1>, then periodic timers are supported. If	4173	If undefined or defined to be C<1> (and the platform supports it), then
2995	defined to be C<0>, then they are not. Disabling them saves a few kB of	4174	the respective watcher type is supported. If defined to be C<0>, then it
2996	code.	4175	is not. Disabling watcher types mainly saves code size.
2997		4176
2998	=item EV_IDLE_ENABLE	4177	=item EV_FEATURES
2999
3000	If undefined or defined to be C<1>, then idle watchers are supported. If
3001	defined to be C<0>, then they are not. Disabling them saves a few kB of
3002	code.
3003
3004	=item EV_EMBED_ENABLE
3005
3006	If undefined or defined to be C<1>, then embed watchers are supported. If
3007	defined to be C<0>, then they are not.
3008
3009	=item EV_STAT_ENABLE
3010
3011	If undefined or defined to be C<1>, then stat watchers are supported. If
3012	defined to be C<0>, then they are not.
3013
3014	=item EV_FORK_ENABLE
3015
3016	If undefined or defined to be C<1>, then fork watchers are supported. If
3017	defined to be C<0>, then they are not.
3018
3019	=item EV_ASYNC_ENABLE
3020
3021	If undefined or defined to be C<1>, then async watchers are supported. If
3022	defined to be C<0>, then they are not.
3023
3024	=item EV_MINIMAL
3025		4178
3026	If you need to shave off some kilobytes of code at the expense of some	4179	If you need to shave off some kilobytes of code at the expense of some
3027	speed, define this symbol to C<1>. Currently this is used to override some	4180	speed (but with the full API), you can define this symbol to request
3028	inlining decisions, saves roughly 30% code size on amd64. It also selects a	4181	certain subsets of functionality. The default is to enable all features
3029	much smaller 2-heap for timer management over the default 4-heap.	4182	that can be enabled on the platform.
		4183
		4184	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4185	with some broad features you want) and then selectively re-enable
		4186	additional parts you want, for example if you want everything minimal,
		4187	but multiple event loop support, async and child watchers and the poll
		4188	backend, use this:
		4189
		4190	#define EV_FEATURES 0
		4191	#define EV_MULTIPLICITY 1
		4192	#define EV_USE_POLL 1
		4193	#define EV_CHILD_ENABLE 1
		4194	#define EV_ASYNC_ENABLE 1
		4195
		4196	The actual value is a bitset, it can be a combination of the following
		4197	values:
		4198
		4199	=over 4
		4200
		4201	=item C<1> - faster/larger code
		4202
		4203	Use larger code to speed up some operations.
		4204
		4205	Currently this is used to override some inlining decisions (enlarging the
		4206	code size by roughly 30% on amd64).
		4207
		4208	When optimising for size, use of compiler flags such as C<-Os> with
		4209	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4210	assertions.
		4211
		4212	=item C<2> - faster/larger data structures
		4213
		4214	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4215	hash table sizes and so on. This will usually further increase code size
		4216	and can additionally have an effect on the size of data structures at
		4217	runtime.
		4218
		4219	=item C<4> - full API configuration
		4220
		4221	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4222	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4223
		4224	=item C<8> - full API
		4225
		4226	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4227	details on which parts of the API are still available without this
		4228	feature, and do not complain if this subset changes over time.
		4229
		4230	=item C<16> - enable all optional watcher types
		4231
		4232	Enables all optional watcher types. If you want to selectively enable
		4233	only some watcher types other than I/O and timers (e.g. prepare,
		4234	embed, async, child...) you can enable them manually by defining
		4235	C<EV_watchertype_ENABLE> to C<1> instead.
		4236
		4237	=item C<32> - enable all backends
		4238
		4239	This enables all backends - without this feature, you need to enable at
		4240	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4241
		4242	=item C<64> - enable OS-specific "helper" APIs
		4243
		4244	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4245	default.
		4246
		4247	=back
		4248
		4249	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4250	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4251	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4252	watchers, timers and monotonic clock support.
		4253
		4254	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4255	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4256	your program might be left out as well - a binary starting a timer and an
		4257	I/O watcher then might come out at only 5Kb.
		4258
		4259	=item EV_AVOID_STDIO
		4260
		4261	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4262	functions (printf, scanf, perror etc.). This will increase the code size
		4263	somewhat, but if your program doesn't otherwise depend on stdio and your
		4264	libc allows it, this avoids linking in the stdio library which is quite
		4265	big.
		4266
		4267	Note that error messages might become less precise when this option is
		4268	enabled.
		4269
		4270	=item EV_NSIG
		4271
		4272	The highest supported signal number, +1 (or, the number of
		4273	signals): Normally, libev tries to deduce the maximum number of signals
		4274	automatically, but sometimes this fails, in which case it can be
		4275	specified. Also, using a lower number than detected (C<32> should be
		4276	good for about any system in existence) can save some memory, as libev
		4277	statically allocates some 12-24 bytes per signal number.
3030		4278
3031	=item EV_PID_HASHSIZE	4279	=item EV_PID_HASHSIZE
3032		4280
3033	C<ev_child> watchers use a small hash table to distribute workload by	4281	C<ev_child> watchers use a small hash table to distribute workload by
3034	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	4282	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
3035	than enough. If you need to manage thousands of children you might want to	4283	usually more than enough. If you need to manage thousands of children you
3036	increase this value (I<must> be a power of two).	4284	might want to increase this value (I<must> be a power of two).
3037		4285
3038	=item EV_INOTIFY_HASHSIZE	4286	=item EV_INOTIFY_HASHSIZE
3039		4287
3040	C<ev_stat> watchers use a small hash table to distribute workload by	4288	C<ev_stat> watchers use a small hash table to distribute workload by
3041	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	4289	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
3042	usually more than enough. If you need to manage thousands of C<ev_stat>	4290	disabled), usually more than enough. If you need to manage thousands of
3043	watchers you might want to increase this value (I<must> be a power of	4291	C<ev_stat> watchers you might want to increase this value (I<must> be a
3044	two).	4292	power of two).
3045		4293
3046	=item EV_USE_4HEAP	4294	=item EV_USE_4HEAP
3047		4295
3048	Heaps are not very cache-efficient. To improve the cache-efficiency of the	4296	Heaps are not very cache-efficient. To improve the cache-efficiency of the
3049	timer and periodics heap, libev uses a 4-heap when this symbol is defined	4297	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
3050	to C<1>. The 4-heap uses more complicated (longer) code but has	4298	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
3051	noticeably faster performance with many (thousands) of watchers.	4299	faster performance with many (thousands) of watchers.
3052		4300
3053	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>	4301	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
3054	(disabled).	4302	will be C<0>.
3055		4303
3056	=item EV_HEAP_CACHE_AT	4304	=item EV_HEAP_CACHE_AT
3057		4305
3058	Heaps are not very cache-efficient. To improve the cache-efficiency of the	4306	Heaps are not very cache-efficient. To improve the cache-efficiency of the
3059	timer and periodics heap, libev can cache the timestamp (I<at>) within	4307	timer and periodics heaps, libev can cache the timestamp (I<at>) within
3060	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),	4308	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
3061	which uses 8-12 bytes more per watcher and a few hundred bytes more code,	4309	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
3062	but avoids random read accesses on heap changes. This improves performance	4310	but avoids random read accesses on heap changes. This improves performance
3063	noticeably with with many (hundreds) of watchers.	4311	noticeably with many (hundreds) of watchers.
3064		4312
3065	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>	4313	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
3066	(disabled).	4314	will be C<0>.
3067		4315
3068	=item EV_VERIFY	4316	=item EV_VERIFY
3069		4317
3070	Controls how much internal verification (see C<ev_loop_verify ()>) will	4318	Controls how much internal verification (see C<ev_verify ()>) will
3071	be done: If set to C<0>, no internal verification code will be compiled	4319	be done: If set to C<0>, no internal verification code will be compiled
3072	in. If set to C<1>, then verification code will be compiled in, but not	4320	in. If set to C<1>, then verification code will be compiled in, but not
3073	called. If set to C<2>, then the internal verification code will be	4321	called. If set to C<2>, then the internal verification code will be
3074	called once per loop, which can slow down libev. If set to C<3>, then the	4322	called once per loop, which can slow down libev. If set to C<3>, then the
3075	verification code will be called very frequently, which will slow down	4323	verification code will be called very frequently, which will slow down
3076	libev considerably.	4324	libev considerably.
3077		4325
3078	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be	4326	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
3079	C<0.>	4327	will be C<0>.
3080		4328
3081	=item EV_COMMON	4329	=item EV_COMMON
3082		4330
3083	By default, all watchers have a C<void *data> member. By redefining	4331	By default, all watchers have a C<void *data> member. By redefining
3084	this macro to a something else you can include more and other types of	4332	this macro to something else you can include more and other types of
3085	members. You have to define it each time you include one of the files,	4333	members. You have to define it each time you include one of the files,
3086	though, and it must be identical each time.	4334	though, and it must be identical each time.
3087		4335
3088	For example, the perl EV module uses something like this:	4336	For example, the perl EV module uses something like this:
3089		4337
3090	#define EV_COMMON \	4338	#define EV_COMMON \
3091	SV *self; /* contains this struct */ \	4339	SV *self; /* contains this struct */ \
3092	SV *cb_sv, *fh /* note no trailing ";" */	4340	SV *cb_sv, *fh /* note no trailing ";" */
3093		4341
3094	=item EV_CB_DECLARE (type)	4342	=item EV_CB_DECLARE (type)
3095		4343
3096	=item EV_CB_INVOKE (watcher, revents)	4344	=item EV_CB_INVOKE (watcher, revents)
3097		4345
…		…
3102	definition and a statement, respectively. See the F<ev.h> header file for	4350	definition and a statement, respectively. See the F<ev.h> header file for
3103	their default definitions. One possible use for overriding these is to	4351	their default definitions. One possible use for overriding these is to
3104	avoid the C<struct ev_loop *> as first argument in all cases, or to use	4352	avoid the C<struct ev_loop *> as first argument in all cases, or to use
3105	method calls instead of plain function calls in C++.	4353	method calls instead of plain function calls in C++.
3106		4354
		4355	=back
		4356
3107	=head2 EXPORTED API SYMBOLS	4357	=head2 EXPORTED API SYMBOLS
3108		4358
3109	If you need to re-export the API (e.g. via a DLL) and you need a list of	4359	If you need to re-export the API (e.g. via a DLL) and you need a list of
3110	exported symbols, you can use the provided F<Symbol.*> files which list	4360	exported symbols, you can use the provided F<Symbol.*> files which list
3111	all public symbols, one per line:	4361	all public symbols, one per line:
3112		4362
3113	Symbols.ev for libev proper	4363	Symbols.ev for libev proper
3114	Symbols.event for the libevent emulation	4364	Symbols.event for the libevent emulation
3115		4365
3116	This can also be used to rename all public symbols to avoid clashes with	4366	This can also be used to rename all public symbols to avoid clashes with
3117	multiple versions of libev linked together (which is obviously bad in	4367	multiple versions of libev linked together (which is obviously bad in
3118	itself, but sometimes it is inconvenient to avoid this).	4368	itself, but sometimes it is inconvenient to avoid this).
3119		4369
…		…
3140	file.	4390	file.
3141		4391
3142	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	4392	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
3143	that everybody includes and which overrides some configure choices:	4393	that everybody includes and which overrides some configure choices:
3144		4394
3145	#define EV_MINIMAL 1	4395	#define EV_FEATURES 8
3146	#define EV_USE_POLL 0	4396	#define EV_USE_SELECT 1
3147	#define EV_MULTIPLICITY 0
3148	#define EV_PERIODIC_ENABLE 0	4397	#define EV_PREPARE_ENABLE 1
		4398	#define EV_IDLE_ENABLE 1
3149	#define EV_STAT_ENABLE 0	4399	#define EV_SIGNAL_ENABLE 1
3150	#define EV_FORK_ENABLE 0	4400	#define EV_CHILD_ENABLE 1
		4401	#define EV_USE_STDEXCEPT 0
3151	#define EV_CONFIG_H <config.h>	4402	#define EV_CONFIG_H <config.h>
3152	#define EV_MINPRI 0
3153	#define EV_MAXPRI 0
3154		4403
3155	#include "ev++.h"	4404	#include "ev++.h"
3156		4405
3157	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	4406	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
3158		4407
3159	#include "ev_cpp.h"	4408	#include "ev_cpp.h"
3160	#include "ev.c"	4409	#include "ev.c"
3161		4410
		4411	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
3162		4412
3163	=head1 THREADS AND COROUTINES	4413	=head2 THREADS AND COROUTINES
3164		4414
3165	=head2 THREADS	4415	=head3 THREADS
3166		4416
3167	Libev itself is completely thread-safe, but it uses no locking. This	4417	All libev functions are reentrant and thread-safe unless explicitly
		4418	documented otherwise, but libev implements no locking itself. This means
3168	means that you can use as many loops as you want in parallel, as long as	4419	that you can use as many loops as you want in parallel, as long as there
3169	only one thread ever calls into one libev function with the same loop	4420	are no concurrent calls into any libev function with the same loop
3170	parameter.	4421	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4422	of course): libev guarantees that different event loops share no data
		4423	structures that need any locking.
3171		4424
3172	Or put differently: calls with different loop parameters can be done in	4425	Or to put it differently: calls with different loop parameters can be done
3173	parallel from multiple threads, calls with the same loop parameter must be	4426	concurrently from multiple threads, calls with the same loop parameter
3174	done serially (but can be done from different threads, as long as only one	4427	must be done serially (but can be done from different threads, as long as
3175	thread ever is inside a call at any point in time, e.g. by using a mutex	4428	only one thread ever is inside a call at any point in time, e.g. by using
3176	per loop).	4429	a mutex per loop).
3177		4430
3178	If you want to know which design is best for your problem, then I cannot	4431	Specifically to support threads (and signal handlers), libev implements
		4432	so-called C<ev_async> watchers, which allow some limited form of
		4433	concurrency on the same event loop, namely waking it up "from the
		4434	outside".
		4435
		4436	If you want to know which design (one loop, locking, or multiple loops
		4437	without or something else still) is best for your problem, then I cannot
3179	help you but by giving some generic advice:	4438	help you, but here is some generic advice:
3180		4439
3181	=over 4	4440	=over 4
3182		4441
3183	=item * most applications have a main thread: use the default libev loop	4442	=item * most applications have a main thread: use the default libev loop
3184	in that thread, or create a separate thread running only the default loop.	4443	in that thread, or create a separate thread running only the default loop.
…		…
3196		4455
3197	Choosing a model is hard - look around, learn, know that usually you can do	4456	Choosing a model is hard - look around, learn, know that usually you can do
3198	better than you currently do :-)	4457	better than you currently do :-)
3199		4458
3200	=item * often you need to talk to some other thread which blocks in the	4459	=item * often you need to talk to some other thread which blocks in the
		4460	event loop.
		4461
3201	event loop - C<ev_async> watchers can be used to wake them up from other	4462	C<ev_async> watchers can be used to wake them up from other threads safely
3202	threads safely (or from signal contexts...).	4463	(or from signal contexts...).
		4464
		4465	An example use would be to communicate signals or other events that only
		4466	work in the default loop by registering the signal watcher with the
		4467	default loop and triggering an C<ev_async> watcher from the default loop
		4468	watcher callback into the event loop interested in the signal.
3203		4469
3204	=back	4470	=back
3205		4471
		4472	=head4 THREAD LOCKING EXAMPLE
		4473
		4474	Here is a fictitious example of how to run an event loop in a different
		4475	thread than where callbacks are being invoked and watchers are
		4476	created/added/removed.
		4477
		4478	For a real-world example, see the C<EV::Loop::Async> perl module,
		4479	which uses exactly this technique (which is suited for many high-level
		4480	languages).
		4481
		4482	The example uses a pthread mutex to protect the loop data, a condition
		4483	variable to wait for callback invocations, an async watcher to notify the
		4484	event loop thread and an unspecified mechanism to wake up the main thread.
		4485
		4486	First, you need to associate some data with the event loop:
		4487
		4488	typedef struct {
		4489	mutex_t lock; /* global loop lock */
		4490	ev_async async_w;
		4491	thread_t tid;
		4492	cond_t invoke_cv;
		4493	} userdata;
		4494
		4495	void prepare_loop (EV_P)
		4496	{
		4497	// for simplicity, we use a static userdata struct.
		4498	static userdata u;
		4499
		4500	ev_async_init (&u->async_w, async_cb);
		4501	ev_async_start (EV_A_ &u->async_w);
		4502
		4503	pthread_mutex_init (&u->lock, 0);
		4504	pthread_cond_init (&u->invoke_cv, 0);
		4505
		4506	// now associate this with the loop
		4507	ev_set_userdata (EV_A_ u);
		4508	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		4509	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		4510
		4511	// then create the thread running ev_loop
		4512	pthread_create (&u->tid, 0, l_run, EV_A);
		4513	}
		4514
		4515	The callback for the C<ev_async> watcher does nothing: the watcher is used
		4516	solely to wake up the event loop so it takes notice of any new watchers
		4517	that might have been added:
		4518
		4519	static void
		4520	async_cb (EV_P_ ev_async *w, int revents)
		4521	{
		4522	// just used for the side effects
		4523	}
		4524
		4525	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		4526	protecting the loop data, respectively.
		4527
		4528	static void
		4529	l_release (EV_P)
		4530	{
		4531	userdata *u = ev_userdata (EV_A);
		4532	pthread_mutex_unlock (&u->lock);
		4533	}
		4534
		4535	static void
		4536	l_acquire (EV_P)
		4537	{
		4538	userdata *u = ev_userdata (EV_A);
		4539	pthread_mutex_lock (&u->lock);
		4540	}
		4541
		4542	The event loop thread first acquires the mutex, and then jumps straight
		4543	into C<ev_run>:
		4544
		4545	void *
		4546	l_run (void *thr_arg)
		4547	{
		4548	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		4549
		4550	l_acquire (EV_A);
		4551	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		4552	ev_run (EV_A_ 0);
		4553	l_release (EV_A);
		4554
		4555	return 0;
		4556	}
		4557
		4558	Instead of invoking all pending watchers, the C<l_invoke> callback will
		4559	signal the main thread via some unspecified mechanism (signals? pipe
		4560	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		4561	have been called (in a while loop because a) spurious wakeups are possible
		4562	and b) skipping inter-thread-communication when there are no pending
		4563	watchers is very beneficial):
		4564
		4565	static void
		4566	l_invoke (EV_P)
		4567	{
		4568	userdata *u = ev_userdata (EV_A);
		4569
		4570	while (ev_pending_count (EV_A))
		4571	{
		4572	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		4573	pthread_cond_wait (&u->invoke_cv, &u->lock);
		4574	}
		4575	}
		4576
		4577	Now, whenever the main thread gets told to invoke pending watchers, it
		4578	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		4579	thread to continue:
		4580
		4581	static void
		4582	real_invoke_pending (EV_P)
		4583	{
		4584	userdata *u = ev_userdata (EV_A);
		4585
		4586	pthread_mutex_lock (&u->lock);
		4587	ev_invoke_pending (EV_A);
		4588	pthread_cond_signal (&u->invoke_cv);
		4589	pthread_mutex_unlock (&u->lock);
		4590	}
		4591
		4592	Whenever you want to start/stop a watcher or do other modifications to an
		4593	event loop, you will now have to lock:
		4594
		4595	ev_timer timeout_watcher;
		4596	userdata *u = ev_userdata (EV_A);
		4597
		4598	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		4599
		4600	pthread_mutex_lock (&u->lock);
		4601	ev_timer_start (EV_A_ &timeout_watcher);
		4602	ev_async_send (EV_A_ &u->async_w);
		4603	pthread_mutex_unlock (&u->lock);
		4604
		4605	Note that sending the C<ev_async> watcher is required because otherwise
		4606	an event loop currently blocking in the kernel will have no knowledge
		4607	about the newly added timer. By waking up the loop it will pick up any new
		4608	watchers in the next event loop iteration.
		4609
3206	=head2 COROUTINES	4610	=head3 COROUTINES
3207		4611
3208	Libev is much more accommodating to coroutines ("cooperative threads"):	4612	Libev is very accommodating to coroutines ("cooperative threads"):
3209	libev fully supports nesting calls to it's functions from different	4613	libev fully supports nesting calls to its functions from different
3210	coroutines (e.g. you can call C<ev_loop> on the same loop from two	4614	coroutines (e.g. you can call C<ev_run> on the same loop from two
3211	different coroutines and switch freely between both coroutines running the	4615	different coroutines, and switch freely between both coroutines running
3212	loop, as long as you don't confuse yourself). The only exception is that	4616	the loop, as long as you don't confuse yourself). The only exception is
3213	you must not do this from C<ev_periodic> reschedule callbacks.	4617	that you must not do this from C<ev_periodic> reschedule callbacks.
3214		4618
3215	Care has been invested into making sure that libev does not keep local	4619	Care has been taken to ensure that libev does not keep local state inside
3216	state inside C<ev_loop>, and other calls do not usually allow coroutine	4620	C<ev_run>, and other calls do not usually allow for coroutine switches as
3217	switches.	4621	they do not call any callbacks.
3218		4622
		4623	=head2 COMPILER WARNINGS
3219		4624
3220	=head1 COMPLEXITIES	4625	Depending on your compiler and compiler settings, you might get no or a
		4626	lot of warnings when compiling libev code. Some people are apparently
		4627	scared by this.
3221		4628
3222	In this section the complexities of (many of) the algorithms used inside	4629	However, these are unavoidable for many reasons. For one, each compiler
3223	libev will be explained. For complexity discussions about backends see the	4630	has different warnings, and each user has different tastes regarding
3224	documentation for C<ev_default_init>.	4631	warning options. "Warn-free" code therefore cannot be a goal except when
		4632	targeting a specific compiler and compiler-version.
3225		4633
3226	All of the following are about amortised time: If an array needs to be	4634	Another reason is that some compiler warnings require elaborate
3227	extended, libev needs to realloc and move the whole array, but this	4635	workarounds, or other changes to the code that make it less clear and less
3228	happens asymptotically never with higher number of elements, so O(1) might	4636	maintainable.
3229	mean it might do a lengthy realloc operation in rare cases, but on average
3230	it is much faster and asymptotically approaches constant time.
3231		4637
3232	=over 4	4638	And of course, some compiler warnings are just plain stupid, or simply
		4639	wrong (because they don't actually warn about the condition their message
		4640	seems to warn about). For example, certain older gcc versions had some
		4641	warnings that resulted in an extreme number of false positives. These have
		4642	been fixed, but some people still insist on making code warn-free with
		4643	such buggy versions.
3233		4644
3234	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	4645	While libev is written to generate as few warnings as possible,
		4646	"warn-free" code is not a goal, and it is recommended not to build libev
		4647	with any compiler warnings enabled unless you are prepared to cope with
		4648	them (e.g. by ignoring them). Remember that warnings are just that:
		4649	warnings, not errors, or proof of bugs.
3235		4650
3236	This means that, when you have a watcher that triggers in one hour and
3237	there are 100 watchers that would trigger before that then inserting will
3238	have to skip roughly seven (C<ld 100>) of these watchers.
3239		4651
3240	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)	4652	=head2 VALGRIND
3241		4653
3242	That means that changing a timer costs less than removing/adding them	4654	Valgrind has a special section here because it is a popular tool that is
3243	as only the relative motion in the event queue has to be paid for.	4655	highly useful. Unfortunately, valgrind reports are very hard to interpret.
3244		4656
3245	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)	4657	If you think you found a bug (memory leak, uninitialised data access etc.)
		4658	in libev, then check twice: If valgrind reports something like:
3246		4659
3247	These just add the watcher into an array or at the head of a list.	4660	==2274== definitely lost: 0 bytes in 0 blocks.
		4661	==2274== possibly lost: 0 bytes in 0 blocks.
		4662	==2274== still reachable: 256 bytes in 1 blocks.
3248		4663
3249	=item Stopping check/prepare/idle/fork/async watchers: O(1)	4664	Then there is no memory leak, just as memory accounted to global variables
		4665	is not a memleak - the memory is still being referenced, and didn't leak.
3250		4666
3251	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	4667	Similarly, under some circumstances, valgrind might report kernel bugs
		4668	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4669	although an acceptable workaround has been found here), or it might be
		4670	confused.
3252		4671
3253	These watchers are stored in lists then need to be walked to find the	4672	Keep in mind that valgrind is a very good tool, but only a tool. Don't
3254	correct watcher to remove. The lists are usually short (you don't usually	4673	make it into some kind of religion.
3255	have many watchers waiting for the same fd or signal).
3256		4674
3257	=item Finding the next timer in each loop iteration: O(1)	4675	If you are unsure about something, feel free to contact the mailing list
		4676	with the full valgrind report and an explanation on why you think this
		4677	is a bug in libev (best check the archives, too :). However, don't be
		4678	annoyed when you get a brisk "this is no bug" answer and take the chance
		4679	of learning how to interpret valgrind properly.
3258		4680
3259	By virtue of using a binary or 4-heap, the next timer is always found at a	4681	If you need, for some reason, empty reports from valgrind for your project
3260	fixed position in the storage array.	4682	I suggest using suppression lists.
3261		4683
3262	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
3263		4684
3264	A change means an I/O watcher gets started or stopped, which requires	4685	=head1 PORTABILITY NOTES
3265	libev to recalculate its status (and possibly tell the kernel, depending
3266	on backend and whether C<ev_io_set> was used).
3267		4686
3268	=item Activating one watcher (putting it into the pending state): O(1)	4687	=head2 GNU/LINUX 32 BIT LIMITATIONS
3269		4688
3270	=item Priority handling: O(number_of_priorities)	4689	GNU/Linux is the only common platform that supports 64 bit file/large file
		4690	interfaces but I<disables> them by default.
3271		4691
3272	Priorities are implemented by allocating some space for each	4692	That means that libev compiled in the default environment doesn't support
3273	priority. When doing priority-based operations, libev usually has to	4693	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
3274	linearly search all the priorities, but starting/stopping and activating
3275	watchers becomes O(1) w.r.t. priority handling.
3276		4694
3277	=item Sending an ev_async: O(1)	4695	Unfortunately, many programs try to work around this GNU/Linux issue
		4696	by enabling the large file API, which makes them incompatible with the
		4697	standard libev compiled for their system.
3278		4698
3279	=item Processing ev_async_send: O(number_of_async_watchers)	4699	Likewise, libev cannot enable the large file API itself as this would
		4700	suddenly make it incompatible to the default compile time environment,
		4701	i.e. all programs not using special compile switches.
3280		4702
3281	=item Processing signals: O(max_signal_number)	4703	=head2 OS/X AND DARWIN BUGS
3282		4704
3283	Sending involves a system call I<iff> there were no other C<ev_async_send>	4705	The whole thing is a bug if you ask me - basically any system interface
3284	calls in the current loop iteration. Checking for async and signal events	4706	you touch is broken, whether it is locales, poll, kqueue or even the
3285	involves iterating over all running async watchers or all signal numbers.	4707	OpenGL drivers.
3286		4708
3287	=back	4709	=head3 C<kqueue> is buggy
3288		4710
		4711	The kqueue syscall is broken in all known versions - most versions support
		4712	only sockets, many support pipes.
3289		4713
3290	=head1 Win32 platform limitations and workarounds	4714	Libev tries to work around this by not using C<kqueue> by default on this
		4715	rotten platform, but of course you can still ask for it when creating a
		4716	loop - embedding a socket-only kqueue loop into a select-based one is
		4717	probably going to work well.
		4718
		4719	=head3 C<poll> is buggy
		4720
		4721	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		4722	implementation by something calling C<kqueue> internally around the 10.5.6
		4723	release, so now C<kqueue> I<and> C<poll> are broken.
		4724
		4725	Libev tries to work around this by not using C<poll> by default on
		4726	this rotten platform, but of course you can still ask for it when creating
		4727	a loop.
		4728
		4729	=head3 C<select> is buggy
		4730
		4731	All that's left is C<select>, and of course Apple found a way to fuck this
		4732	one up as well: On OS/X, C<select> actively limits the number of file
		4733	descriptors you can pass in to 1024 - your program suddenly crashes when
		4734	you use more.
		4735
		4736	There is an undocumented "workaround" for this - defining
		4737	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		4738	work on OS/X.
		4739
		4740	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		4741
		4742	=head3 C<errno> reentrancy
		4743
		4744	The default compile environment on Solaris is unfortunately so
		4745	thread-unsafe that you can't even use components/libraries compiled
		4746	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		4747	defined by default. A valid, if stupid, implementation choice.
		4748
		4749	If you want to use libev in threaded environments you have to make sure
		4750	it's compiled with C<_REENTRANT> defined.
		4751
		4752	=head3 Event port backend
		4753
		4754	The scalable event interface for Solaris is called "event
		4755	ports". Unfortunately, this mechanism is very buggy in all major
		4756	releases. If you run into high CPU usage, your program freezes or you get
		4757	a large number of spurious wakeups, make sure you have all the relevant
		4758	and latest kernel patches applied. No, I don't know which ones, but there
		4759	are multiple ones to apply, and afterwards, event ports actually work
		4760	great.
		4761
		4762	If you can't get it to work, you can try running the program by setting
		4763	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		4764	C<select> backends.
		4765
		4766	=head2 AIX POLL BUG
		4767
		4768	AIX unfortunately has a broken C<poll.h> header. Libev works around
		4769	this by trying to avoid the poll backend altogether (i.e. it's not even
		4770	compiled in), which normally isn't a big problem as C<select> works fine
		4771	with large bitsets on AIX, and AIX is dead anyway.
		4772
		4773	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4774
		4775	=head3 General issues
3291		4776
3292	Win32 doesn't support any of the standards (e.g. POSIX) that libev	4777	Win32 doesn't support any of the standards (e.g. POSIX) that libev
3293	requires, and its I/O model is fundamentally incompatible with the POSIX	4778	requires, and its I/O model is fundamentally incompatible with the POSIX
3294	model. Libev still offers limited functionality on this platform in	4779	model. Libev still offers limited functionality on this platform in
3295	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	4780	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
3296	descriptors. This only applies when using Win32 natively, not when using	4781	descriptors. This only applies when using Win32 natively, not when using
3297	e.g. cygwin.	4782	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		4783	as every compielr comes with a slightly differently broken/incompatible
		4784	environment.
3298		4785
3299	Lifting these limitations would basically require the full	4786	Lifting these limitations would basically require the full
3300	re-implementation of the I/O system. If you are into these kinds of	4787	re-implementation of the I/O system. If you are into this kind of thing,
3301	things, then note that glib does exactly that for you in a very portable	4788	then note that glib does exactly that for you in a very portable way (note
3302	way (note also that glib is the slowest event library known to man).	4789	also that glib is the slowest event library known to man).
3303		4790
3304	There is no supported compilation method available on windows except	4791	There is no supported compilation method available on windows except
3305	embedding it into other applications.	4792	embedding it into other applications.
		4793
		4794	Sensible signal handling is officially unsupported by Microsoft - libev
		4795	tries its best, but under most conditions, signals will simply not work.
3306		4796
3307	Not a libev limitation but worth mentioning: windows apparently doesn't	4797	Not a libev limitation but worth mentioning: windows apparently doesn't
3308	accept large writes: instead of resulting in a partial write, windows will	4798	accept large writes: instead of resulting in a partial write, windows will
3309	either accept everything or return C<ENOBUFS> if the buffer is too large,	4799	either accept everything or return C<ENOBUFS> if the buffer is too large,
3310	so make sure you only write small amounts into your sockets (less than a	4800	so make sure you only write small amounts into your sockets (less than a
3311	megabyte seems safe, but thsi apparently depends on the amount of memory	4801	megabyte seems safe, but this apparently depends on the amount of memory
3312	available).	4802	available).
3313		4803
3314	Due to the many, low, and arbitrary limits on the win32 platform and	4804	Due to the many, low, and arbitrary limits on the win32 platform and
3315	the abysmal performance of winsockets, using a large number of sockets	4805	the abysmal performance of winsockets, using a large number of sockets
3316	is not recommended (and not reasonable). If your program needs to use	4806	is not recommended (and not reasonable). If your program needs to use
3317	more than a hundred or so sockets, then likely it needs to use a totally	4807	more than a hundred or so sockets, then likely it needs to use a totally
3318	different implementation for windows, as libev offers the POSIX readiness	4808	different implementation for windows, as libev offers the POSIX readiness
3319	notification model, which cannot be implemented efficiently on windows	4809	notification model, which cannot be implemented efficiently on windows
3320	(Microsoft monopoly games).	4810	(due to Microsoft monopoly games).
3321		4811
3322	=over 4	4812	A typical way to use libev under windows is to embed it (see the embedding
		4813	section for details) and use the following F<evwrap.h> header file instead
		4814	of F<ev.h>:
3323		4815
		4816	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4817	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4818
		4819	#include "ev.h"
		4820
		4821	And compile the following F<evwrap.c> file into your project (make sure
		4822	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4823
		4824	#include "evwrap.h"
		4825	#include "ev.c"
		4826
3324	=item The winsocket select function	4827	=head3 The winsocket C<select> function
3325		4828
3326	The winsocket C<select> function doesn't follow POSIX in that it	4829	The winsocket C<select> function doesn't follow POSIX in that it
3327	requires socket I<handles> and not socket I<file descriptors> (it is	4830	requires socket I<handles> and not socket I<file descriptors> (it is
3328	also extremely buggy). This makes select very inefficient, and also	4831	also extremely buggy). This makes select very inefficient, and also
3329	requires a mapping from file descriptors to socket handles. See the	4832	requires a mapping from file descriptors to socket handles (the Microsoft
		4833	C runtime provides the function C<_open_osfhandle> for this). See the
3330	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and	4834	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
3331	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.	4835	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
3332		4836
3333	The configuration for a "naked" win32 using the Microsoft runtime	4837	The configuration for a "naked" win32 using the Microsoft runtime
3334	libraries and raw winsocket select is:	4838	libraries and raw winsocket select is:
3335		4839
3336	#define EV_USE_SELECT 1	4840	#define EV_USE_SELECT 1
3337	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */	4841	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
3338		4842
3339	Note that winsockets handling of fd sets is O(n), so you can easily get a	4843	Note that winsockets handling of fd sets is O(n), so you can easily get a
3340	complexity in the O(n²) range when using win32.	4844	complexity in the O(n²) range when using win32.
3341		4845
3342	=item Limited number of file descriptors	4846	=head3 Limited number of file descriptors
3343		4847
3344	Windows has numerous arbitrary (and low) limits on things.	4848	Windows has numerous arbitrary (and low) limits on things.
3345		4849
3346	Early versions of winsocket's select only supported waiting for a maximum	4850	Early versions of winsocket's select only supported waiting for a maximum
3347	of C<64> handles (probably owning to the fact that all windows kernels	4851	of C<64> handles (probably owning to the fact that all windows kernels
3348	can only wait for C<64> things at the same time internally; Microsoft	4852	can only wait for C<64> things at the same time internally; Microsoft
3349	recommends spawning a chain of threads and wait for 63 handles and the	4853	recommends spawning a chain of threads and wait for 63 handles and the
3350	previous thread in each. Great).	4854	previous thread in each. Sounds great!).
3351		4855
3352	Newer versions support more handles, but you need to define C<FD_SETSIZE>	4856	Newer versions support more handles, but you need to define C<FD_SETSIZE>
3353	to some high number (e.g. C<2048>) before compiling the winsocket select	4857	to some high number (e.g. C<2048>) before compiling the winsocket select
3354	call (which might be in libev or elsewhere, for example, perl does its own	4858	call (which might be in libev or elsewhere, for example, perl and many
3355	select emulation on windows).	4859	other interpreters do their own select emulation on windows).
3356		4860
3357	Another limit is the number of file descriptors in the Microsoft runtime	4861	Another limit is the number of file descriptors in the Microsoft runtime
3358	libraries, which by default is C<64> (there must be a hidden I<64> fetish	4862	libraries, which by default is C<64> (there must be a hidden I<64>
3359	or something like this inside Microsoft). You can increase this by calling	4863	fetish or something like this inside Microsoft). You can increase this
3360	C<_setmaxstdio>, which can increase this limit to C<2048> (another	4864	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
3361	arbitrary limit), but is broken in many versions of the Microsoft runtime	4865	(another arbitrary limit), but is broken in many versions of the Microsoft
3362	libraries.
3363
3364	This might get you to about C<512> or C<2048> sockets (depending on	4866	runtime libraries. This might get you to about C<512> or C<2048> sockets
3365	windows version and/or the phase of the moon). To get more, you need to	4867	(depending on windows version and/or the phase of the moon). To get more,
3366	wrap all I/O functions and provide your own fd management, but the cost of	4868	you need to wrap all I/O functions and provide your own fd management, but
3367	calling select (O(n²)) will likely make this unworkable.	4869	the cost of calling select (O(n²)) will likely make this unworkable.
3368		4870
3369	=back
3370
3371
3372	=head1 PORTABILITY REQUIREMENTS	4871	=head2 PORTABILITY REQUIREMENTS
3373		4872
3374	In addition to a working ISO-C implementation, libev relies on a few	4873	In addition to a working ISO-C implementation and of course the
3375	additional extensions:	4874	backend-specific APIs, libev relies on a few additional extensions:
3376		4875
3377	=over 4	4876	=over 4
3378		4877
		4878	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4879	calling conventions regardless of C<ev_watcher_type *>.
		4880
		4881	Libev assumes not only that all watcher pointers have the same internal
		4882	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4883	assumes that the same (machine) code can be used to call any watcher
		4884	callback: The watcher callbacks have different type signatures, but libev
		4885	calls them using an C<ev_watcher *> internally.
		4886
		4887	=item pointer accesses must be thread-atomic
		4888
		4889	Accessing a pointer value must be atomic, it must both be readable and
		4890	writable in one piece - this is the case on all current architectures.
		4891
3379	=item C<sig_atomic_t volatile> must be thread-atomic as well	4892	=item C<sig_atomic_t volatile> must be thread-atomic as well
3380		4893
3381	The type C<sig_atomic_t volatile> (or whatever is defined as	4894	The type C<sig_atomic_t volatile> (or whatever is defined as
3382	C<EV_ATOMIC_T>) must be atomic w.r.t. accesses from different	4895	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
3383	threads. This is not part of the specification for C<sig_atomic_t>, but is	4896	threads. This is not part of the specification for C<sig_atomic_t>, but is
3384	believed to be sufficiently portable.	4897	believed to be sufficiently portable.
3385		4898
3386	=item C<sigprocmask> must work in a threaded environment	4899	=item C<sigprocmask> must work in a threaded environment
3387		4900
…		…
3396	except the initial one, and run the default loop in the initial thread as	4909	except the initial one, and run the default loop in the initial thread as
3397	well.	4910	well.
3398		4911
3399	=item C<long> must be large enough for common memory allocation sizes	4912	=item C<long> must be large enough for common memory allocation sizes
3400		4913
3401	To improve portability and simplify using libev, libev uses C<long>	4914	To improve portability and simplify its API, libev uses C<long> internally
3402	internally instead of C<size_t> when allocating its data structures. On	4915	instead of C<size_t> when allocating its data structures. On non-POSIX
3403	non-POSIX systems (Microsoft...) this might be unexpectedly low, but	4916	systems (Microsoft...) this might be unexpectedly low, but is still at
3404	is still at least 31 bits everywhere, which is enough for hundreds of	4917	least 31 bits everywhere, which is enough for hundreds of millions of
3405	millions of watchers.	4918	watchers.
3406		4919
3407	=item C<double> must hold a time value in seconds with enough accuracy	4920	=item C<double> must hold a time value in seconds with enough accuracy
3408		4921
3409	The type C<double> is used to represent timestamps. It is required to	4922	The type C<double> is used to represent timestamps. It is required to
3410	have at least 51 bits of mantissa (and 9 bits of exponent), which is good	4923	have at least 51 bits of mantissa (and 9 bits of exponent), which is
3411	enough for at least into the year 4000. This requirement is fulfilled by	4924	good enough for at least into the year 4000 with millisecond accuracy
		4925	(the design goal for libev). This requirement is overfulfilled by
3412	implementations implementing IEEE 754 (basically all existing ones).	4926	implementations using IEEE 754, which is basically all existing ones. With
		4927	IEEE 754 doubles, you get microsecond accuracy until at least 2200.
3413		4928
3414	=back	4929	=back
3415		4930
3416	If you know of other additional requirements drop me a note.	4931	If you know of other additional requirements drop me a note.
3417		4932
3418		4933
3419	=head1 COMPILER WARNINGS	4934	=head1 ALGORITHMIC COMPLEXITIES
3420		4935
3421	Depending on your compiler and compiler settings, you might get no or a	4936	In this section the complexities of (many of) the algorithms used inside
3422	lot of warnings when compiling libev code. Some people are apparently	4937	libev will be documented. For complexity discussions about backends see
3423	scared by this.	4938	the documentation for C<ev_default_init>.
3424		4939
3425	However, these are unavoidable for many reasons. For one, each compiler	4940	All of the following are about amortised time: If an array needs to be
3426	has different warnings, and each user has different tastes regarding	4941	extended, libev needs to realloc and move the whole array, but this
3427	warning options. "Warn-free" code therefore cannot be a goal except when	4942	happens asymptotically rarer with higher number of elements, so O(1) might
3428	targeting a specific compiler and compiler-version.	4943	mean that libev does a lengthy realloc operation in rare cases, but on
		4944	average it is much faster and asymptotically approaches constant time.
3429		4945
3430	Another reason is that some compiler warnings require elaborate	4946	=over 4
3431	workarounds, or other changes to the code that make it less clear and less
3432	maintainable.
3433		4947
3434	And of course, some compiler warnings are just plain stupid, or simply	4948	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
3435	wrong (because they don't actually warn about the condition their message
3436	seems to warn about).
3437		4949
3438	While libev is written to generate as few warnings as possible,	4950	This means that, when you have a watcher that triggers in one hour and
3439	"warn-free" code is not a goal, and it is recommended not to build libev	4951	there are 100 watchers that would trigger before that, then inserting will
3440	with any compiler warnings enabled unless you are prepared to cope with	4952	have to skip roughly seven (C<ld 100>) of these watchers.
3441	them (e.g. by ignoring them). Remember that warnings are just that:
3442	warnings, not errors, or proof of bugs.
3443		4953
		4954	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
3444		4955
3445	=head1 VALGRIND	4956	That means that changing a timer costs less than removing/adding them,
		4957	as only the relative motion in the event queue has to be paid for.
3446		4958
3447	Valgrind has a special section here because it is a popular tool that is	4959	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
3448	highly useful, but valgrind reports are very hard to interpret.
3449		4960
3450	If you think you found a bug (memory leak, uninitialised data access etc.)	4961	These just add the watcher into an array or at the head of a list.
3451	in libev, then check twice: If valgrind reports something like:
3452		4962
3453	==2274== definitely lost: 0 bytes in 0 blocks.	4963	=item Stopping check/prepare/idle/fork/async watchers: O(1)
3454	==2274== possibly lost: 0 bytes in 0 blocks.
3455	==2274== still reachable: 256 bytes in 1 blocks.
3456		4964
3457	Then there is no memory leak. Similarly, under some circumstances,	4965	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
3458	valgrind might report kernel bugs as if it were a bug in libev, or it
3459	might be confused (it is a very good tool, but only a tool).
3460		4966
3461	If you are unsure about something, feel free to contact the mailing list	4967	These watchers are stored in lists, so they need to be walked to find the
3462	with the full valgrind report and an explanation on why you think this is	4968	correct watcher to remove. The lists are usually short (you don't usually
3463	a bug in libev. However, don't be annoyed when you get a brisk "this is	4969	have many watchers waiting for the same fd or signal: one is typical, two
3464	no bug" answer and take the chance of learning how to interpret valgrind	4970	is rare).
3465	properly.
3466		4971
3467	If you need, for some reason, empty reports from valgrind for your project	4972	=item Finding the next timer in each loop iteration: O(1)
3468	I suggest using suppression lists.
3469		4973
		4974	By virtue of using a binary or 4-heap, the next timer is always found at a
		4975	fixed position in the storage array.
		4976
		4977	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		4978
		4979	A change means an I/O watcher gets started or stopped, which requires
		4980	libev to recalculate its status (and possibly tell the kernel, depending
		4981	on backend and whether C<ev_io_set> was used).
		4982
		4983	=item Activating one watcher (putting it into the pending state): O(1)
		4984
		4985	=item Priority handling: O(number_of_priorities)
		4986
		4987	Priorities are implemented by allocating some space for each
		4988	priority. When doing priority-based operations, libev usually has to
		4989	linearly search all the priorities, but starting/stopping and activating
		4990	watchers becomes O(1) with respect to priority handling.
		4991
		4992	=item Sending an ev_async: O(1)
		4993
		4994	=item Processing ev_async_send: O(number_of_async_watchers)
		4995
		4996	=item Processing signals: O(max_signal_number)
		4997
		4998	Sending involves a system call I<iff> there were no other C<ev_async_send>
		4999	calls in the current loop iteration. Checking for async and signal events
		5000	involves iterating over all running async watchers or all signal numbers.
		5001
		5002	=back
		5003
		5004
		5005	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5006
		5007	The major version 4 introduced some incompatible changes to the API.
		5008
		5009	At the moment, the C<ev.h> header file provides compatibility definitions
		5010	for all changes, so most programs should still compile. The compatibility
		5011	layer might be removed in later versions of libev, so better update to the
		5012	new API early than late.
		5013
		5014	=over 4
		5015
		5016	=item C<EV_COMPAT3> backwards compatibility mechanism
		5017
		5018	The backward compatibility mechanism can be controlled by
		5019	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
		5020	section.
		5021
		5022	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5023
		5024	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5025
		5026	ev_loop_destroy (EV_DEFAULT_UC);
		5027	ev_loop_fork (EV_DEFAULT);
		5028
		5029	=item function/symbol renames
		5030
		5031	A number of functions and symbols have been renamed:
		5032
		5033	ev_loop => ev_run
		5034	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5035	EVLOOP_ONESHOT => EVRUN_ONCE
		5036
		5037	ev_unloop => ev_break
		5038	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5039	EVUNLOOP_ONE => EVBREAK_ONE
		5040	EVUNLOOP_ALL => EVBREAK_ALL
		5041
		5042	EV_TIMEOUT => EV_TIMER
		5043
		5044	ev_loop_count => ev_iteration
		5045	ev_loop_depth => ev_depth
		5046	ev_loop_verify => ev_verify
		5047
		5048	Most functions working on C<struct ev_loop> objects don't have an
		5049	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5050	associated constants have been renamed to not collide with the C<struct
		5051	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5052	as all other watcher types. Note that C<ev_loop_fork> is still called
		5053	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5054	typedef.
		5055
		5056	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5057
		5058	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5059	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5060	and work, but the library code will of course be larger.
		5061
		5062	=back
		5063
		5064
		5065	=head1 GLOSSARY
		5066
		5067	=over 4
		5068
		5069	=item active
		5070
		5071	A watcher is active as long as it has been started and not yet stopped.
		5072	See L<WATCHER STATES> for details.
		5073
		5074	=item application
		5075
		5076	In this document, an application is whatever is using libev.
		5077
		5078	=item backend
		5079
		5080	The part of the code dealing with the operating system interfaces.
		5081
		5082	=item callback
		5083
		5084	The address of a function that is called when some event has been
		5085	detected. Callbacks are being passed the event loop, the watcher that
		5086	received the event, and the actual event bitset.
		5087
		5088	=item callback/watcher invocation
		5089
		5090	The act of calling the callback associated with a watcher.
		5091
		5092	=item event
		5093
		5094	A change of state of some external event, such as data now being available
		5095	for reading on a file descriptor, time having passed or simply not having
		5096	any other events happening anymore.
		5097
		5098	In libev, events are represented as single bits (such as C<EV_READ> or
		5099	C<EV_TIMER>).
		5100
		5101	=item event library
		5102
		5103	A software package implementing an event model and loop.
		5104
		5105	=item event loop
		5106
		5107	An entity that handles and processes external events and converts them
		5108	into callback invocations.
		5109
		5110	=item event model
		5111
		5112	The model used to describe how an event loop handles and processes
		5113	watchers and events.
		5114
		5115	=item pending
		5116
		5117	A watcher is pending as soon as the corresponding event has been
		5118	detected. See L<WATCHER STATES> for details.
		5119
		5120	=item real time
		5121
		5122	The physical time that is observed. It is apparently strictly monotonic :)
		5123
		5124	=item wall-clock time
		5125
		5126	The time and date as shown on clocks. Unlike real time, it can actually
		5127	be wrong and jump forwards and backwards, e.g. when the you adjust your
		5128	clock.
		5129
		5130	=item watcher
		5131
		5132	A data structure that describes interest in certain events. Watchers need
		5133	to be started (attached to an event loop) before they can receive events.
		5134
		5135	=back
3470		5136
3471	=head1 AUTHOR	5137	=head1 AUTHOR
3472		5138
3473	Marc Lehmann <libev@schmorp.de>.	5139	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5140	Magnusson and Emanuele Giaquinta.
3474		5141

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines