ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.144 by root, Mon Apr 7 12:33:29 2008 UTC vs.
Revision 1.468 by sf-exg, Sun May 14 19:02:31 2023 UTC

		1	=encoding utf-8
		2
1	=head1 NAME	3	=head1 NAME
2		4
3	libev - a high performance full-featured event loop written in C	5	libev - a high performance full-featured event loop written in C
4		6
5	=head1 SYNOPSIS	7	=head1 SYNOPSIS
6		8
7	#include <ev.h>	9	#include <ev.h>
8		10
9	=head2 EXAMPLE PROGRAM	11	=head2 EXAMPLE PROGRAM
10		12
11	// a single header file is required	13	// a single header file is required
12	#include <ev.h>	14	#include <ev.h>
13		15
		16	#include <stdio.h> // for puts
		17
14	// every watcher type has its own typedef'd struct	18	// every watcher type has its own typedef'd struct
15	// with the name ev_<type>	19	// with the name ev_TYPE
16	ev_io stdin_watcher;	20	ev_io stdin_watcher;
17	ev_timer timeout_watcher;	21	ev_timer timeout_watcher;
18		22
19	// all watcher callbacks have a similar signature	23	// all watcher callbacks have a similar signature
20	// this callback is called when data is readable on stdin	24	// this callback is called when data is readable on stdin
21	static void	25	static void
22	stdin_cb (EV_P_ struct ev_io *w, int revents)	26	stdin_cb (EV_P_ ev_io *w, int revents)
23	{	27	{
24	puts ("stdin ready");	28	puts ("stdin ready");
25	// for one-shot events, one must manually stop the watcher	29	// for one-shot events, one must manually stop the watcher
26	// with its corresponding stop function.	30	// with its corresponding stop function.
27	ev_io_stop (EV_A_ w);	31	ev_io_stop (EV_A_ w);
28		32
29	// this causes all nested ev_loop's to stop iterating	33	// this causes all nested ev_run's to stop iterating
30	ev_unloop (EV_A_ EVUNLOOP_ALL);	34	ev_break (EV_A_ EVBREAK_ALL);
31	}	35	}
32		36
33	// another callback, this time for a time-out	37	// another callback, this time for a time-out
34	static void	38	static void
35	timeout_cb (EV_P_ struct ev_timer *w, int revents)	39	timeout_cb (EV_P_ ev_timer *w, int revents)
36	{	40	{
37	puts ("timeout");	41	puts ("timeout");
38	// this causes the innermost ev_loop to stop iterating	42	// this causes the innermost ev_run to stop iterating
39	ev_unloop (EV_A_ EVUNLOOP_ONE);	43	ev_break (EV_A_ EVBREAK_ONE);
40	}	44	}
41		45
42	int	46	int
43	main (void)	47	main (void)
44	{	48	{
45	// use the default event loop unless you have special needs	49	// use the default event loop unless you have special needs
46	struct ev_loop *loop = ev_default_loop (0);	50	struct ev_loop *loop = EV_DEFAULT;
47		51
48	// initialise an io watcher, then start it	52	// initialise an io watcher, then start it
49	// this one will watch for stdin to become readable	53	// this one will watch for stdin to become readable
50	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	54	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
51	ev_io_start (loop, &stdin_watcher);	55	ev_io_start (loop, &stdin_watcher);
52		56
53	// initialise a timer watcher, then start it	57	// initialise a timer watcher, then start it
54	// simple non-repeating 5.5 second timeout	58	// simple non-repeating 5.5 second timeout
55	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);	59	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
56	ev_timer_start (loop, &timeout_watcher);	60	ev_timer_start (loop, &timeout_watcher);
57		61
58	// now wait for events to arrive	62	// now wait for events to arrive
59	ev_loop (loop, 0);	63	ev_run (loop, 0);
60		64
61	// unloop was called, so exit	65	// break was called, so exit
62	return 0;	66	return 0;
63	}	67	}
64		68
65	=head1 DESCRIPTION	69	=head1 ABOUT THIS DOCUMENT
		70
		71	This document documents the libev software package.
66		72
67	The newest version of this document is also available as an html-formatted	73	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	74	web page you might find easier to navigate when reading it for the first
69	time: L<http://cvs.schmorp.de/libev/ev.html>.	75	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		76
		77	While this document tries to be as complete as possible in documenting
		78	libev, its usage and the rationale behind its design, it is not a tutorial
		79	on event-based programming, nor will it introduce event-based programming
		80	with libev.
		81
		82	Familiarity with event based programming techniques in general is assumed
		83	throughout this document.
		84
		85	=head1 WHAT TO READ WHEN IN A HURRY
		86
		87	This manual tries to be very detailed, but unfortunately, this also makes
		88	it very long. If you just want to know the basics of libev, I suggest
		89	reading L</ANATOMY OF A WATCHER>, then the L</EXAMPLE PROGRAM> above and
		90	look up the missing functions in L</GLOBAL FUNCTIONS> and the C<ev_io> and
		91	C<ev_timer> sections in L</WATCHER TYPES>.
		92
		93	=head1 ABOUT LIBEV
70		94
71	Libev is an event loop: you register interest in certain events (such as a	95	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	96	file descriptor being readable or a timeout occurring), and it will manage
73	these event sources and provide your program with events.	97	these event sources and provide your program with events.
74		98
…		…
81	details of the event, and then hand it over to libev by I<starting> the	105	details of the event, and then hand it over to libev by I<starting> the
82	watcher.	106	watcher.
83		107
84	=head2 FEATURES	108	=head2 FEATURES
85		109
86	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	110	Libev supports C<select>, C<poll>, the Linux-specific aio and C<epoll>
87	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	111	interfaces, the BSD-specific C<kqueue> and the Solaris-specific event port
88	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	112	mechanisms for file descriptor events (C<ev_io>), the Linux C<inotify>
89	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers	113	interface (for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
90	with customised rescheduling (C<ev_periodic>), synchronous signals	114	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	115	timers (C<ev_timer>), absolute timers with customised rescheduling
92	watchers dealing with the event loop mechanism itself (C<ev_idle>,	116	(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	117	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	118	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
95	(C<ev_fork>).	119	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		120	limited support for fork events (C<ev_fork>).
96		121
97	It also is quite fast (see this	122	It also is quite fast (see this
98	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent	123	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
99	for example).	124	for example).
100		125
…		…
108	name C<loop> (which is always of type C<struct ev_loop *>) will not have	133	name C<loop> (which is always of type C<struct ev_loop *>) will not have
109	this argument.	134	this argument.
110		135
111	=head2 TIME REPRESENTATION	136	=head2 TIME REPRESENTATION
112		137
113	Libev represents time as a single floating point number, representing the	138	Libev represents time as a single floating point number, representing
114	(fractional) number of seconds since the (POSIX) epoch (somewhere near	139	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	140	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	141	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	142	too. It usually aliases to the C<double> type in C. When you need to do
118	it, you should treat it as some floatingpoint value. Unlike the name	143	any calculations on it, you should treat it as some floating point value.
		144
119	component C<stamp> might indicate, it is also used for time differences	145	Unlike the name component C<stamp> might indicate, it is also used for
120	throughout libev.	146	time differences (e.g. delays) throughout libev.
		147
		148	=head1 ERROR HANDLING
		149
		150	Libev knows three classes of errors: operating system errors, usage errors
		151	and internal errors (bugs).
		152
		153	When libev catches an operating system error it cannot handle (for example
		154	a system call indicating a condition libev cannot fix), it calls the callback
		155	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		156	abort. The default is to print a diagnostic message and to call C<abort
		157	()>.
		158
		159	When libev detects a usage error such as a negative timer interval, then
		160	it will print a diagnostic message and abort (via the C<assert> mechanism,
		161	so C<NDEBUG> will disable this checking): these are programming errors in
		162	the libev caller and need to be fixed there.
		163
		164	Via the C<EV_FREQUENT> macro you can compile in and/or enable extensive
		165	consistency checking code inside libev that can be used to check for
		166	internal inconsistencies, suually caused by application bugs.
		167
		168	Libev also has a few internal error-checking C<assert>ions. These do not
		169	trigger under normal circumstances, as they indicate either a bug in libev
		170	or worse.
		171
121		172
122	=head1 GLOBAL FUNCTIONS	173	=head1 GLOBAL FUNCTIONS
123		174
124	These functions can be called anytime, even before initialising the	175	These functions can be called anytime, even before initialising the
125	library in any way.	176	library in any way.
…		…
128		179
129	=item ev_tstamp ev_time ()	180	=item ev_tstamp ev_time ()
130		181
131	Returns the current time as libev would use it. Please note that the	182	Returns the current time as libev would use it. Please note that the
132	C<ev_now> function is usually faster and also often returns the timestamp	183	C<ev_now> function is usually faster and also often returns the timestamp
133	you actually want to know.	184	you actually want to know. Also interesting is the combination of
		185	C<ev_now_update> and C<ev_now>.
134		186
135	=item ev_sleep (ev_tstamp interval)	187	=item ev_sleep (ev_tstamp interval)
136		188
137	Sleep for the given interval: The current thread will be blocked until	189	Sleep for the given interval: The current thread will be blocked
138	either it is interrupted or the given time interval has passed. Basically	190	until either it is interrupted or the given time interval has
		191	passed (approximately - it might return a bit earlier even if not
		192	interrupted). Returns immediately if C<< interval <= 0 >>.
		193
139	this is a subsecond-resolution C<sleep ()>.	194	Basically this is a sub-second-resolution C<sleep ()>.
		195
		196	The range of the C<interval> is limited - libev only guarantees to work
		197	with sleep times of up to one day (C<< interval <= 86400 >>).
140		198
141	=item int ev_version_major ()	199	=item int ev_version_major ()
142		200
143	=item int ev_version_minor ()	201	=item int ev_version_minor ()
144		202
…		…
155	as this indicates an incompatible change. Minor versions are usually	213	as this indicates an incompatible change. Minor versions are usually
156	compatible to older versions, so a larger minor version alone is usually	214	compatible to older versions, so a larger minor version alone is usually
157	not a problem.	215	not a problem.
158		216
159	Example: Make sure we haven't accidentally been linked against the wrong	217	Example: Make sure we haven't accidentally been linked against the wrong
160	version.	218	version (note, however, that this will not detect other ABI mismatches,
		219	such as LFS or reentrancy).
161		220
162	assert (("libev version mismatch",	221	assert (("libev version mismatch",
163	ev_version_major () == EV_VERSION_MAJOR	222	ev_version_major () == EV_VERSION_MAJOR
164	&& ev_version_minor () >= EV_VERSION_MINOR));	223	&& ev_version_minor () >= EV_VERSION_MINOR));
165		224
166	=item unsigned int ev_supported_backends ()	225	=item unsigned int ev_supported_backends ()
167		226
168	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	227	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
169	value) compiled into this binary of libev (independent of their	228	value) compiled into this binary of libev (independent of their
…		…
171	a description of the set values.	230	a description of the set values.
172		231
173	Example: make sure we have the epoll method, because yeah this is cool and	232	Example: make sure we have the epoll method, because yeah this is cool and
174	a must have and can we have a torrent of it please!!!11	233	a must have and can we have a torrent of it please!!!11
175		234
176	assert (("sorry, no epoll, no sex",	235	assert (("sorry, no epoll, no sex",
177	ev_supported_backends () & EVBACKEND_EPOLL));	236	ev_supported_backends () & EVBACKEND_EPOLL));
178		237
179	=item unsigned int ev_recommended_backends ()	238	=item unsigned int ev_recommended_backends ()
180		239
181	Return the set of all backends compiled into this binary of libev and also	240	Return the set of all backends compiled into this binary of libev and
182	recommended for this platform. This set is often smaller than the one	241	also recommended for this platform, meaning it will work for most file
		242	descriptor types. This set is often smaller than the one returned by
183	returned by C<ev_supported_backends>, as for example kqueue is broken on	243	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
184	most BSDs and will not be autodetected unless you explicitly request it	244	and will not be auto-detected unless you explicitly request it (assuming
185	(assuming you know what you are doing). This is the set of backends that	245	you know what you are doing). This is the set of backends that libev will
186	libev will probe for if you specify no backends explicitly.	246	probe for if you specify no backends explicitly.
187		247
188	=item unsigned int ev_embeddable_backends ()	248	=item unsigned int ev_embeddable_backends ()
189		249
190	Returns the set of backends that are embeddable in other event loops. This	250	Returns the set of backends that are embeddable in other event loops. This
191	is the theoretical, all-platform, value. To find which backends	251	value is platform-specific but can include backends not available on the
192	might be supported on the current system, you would need to look at	252	current system. To find which embeddable backends might be supported on
193	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	253	the current system, you would need to look at C<ev_embeddable_backends ()
194	recommended ones.	254	& ev_supported_backends ()>, likewise for recommended ones.
195		255
196	See the description of C<ev_embed> watchers for more info.	256	See the description of C<ev_embed> watchers for more info.
197		257
198	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	258	=item ev_set_allocator (void *(*cb)(void *ptr, long size) throw ())
199		259
200	Sets the allocation function to use (the prototype is similar - the	260	Sets the allocation function to use (the prototype is similar - the
201	semantics is identical - to the realloc C function). It is used to	261	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
202	allocate and free memory (no surprises here). If it returns zero when	262	used to allocate and free memory (no surprises here). If it returns zero
203	memory needs to be allocated, the library might abort or take some	263	when memory needs to be allocated (C<size != 0>), the library might abort
204	potentially destructive action. The default is your system realloc	264	or take some potentially destructive action.
205	function.	265
		266	Since some systems (at least OpenBSD and Darwin) fail to implement
		267	correct C<realloc> semantics, libev will use a wrapper around the system
		268	C<realloc> and C<free> functions by default.
206		269
207	You could override this function in high-availability programs to, say,	270	You could override this function in high-availability programs to, say,
208	free some memory if it cannot allocate memory, to use a special allocator,	271	free some memory if it cannot allocate memory, to use a special allocator,
209	or even to sleep a while and retry until some memory is available.	272	or even to sleep a while and retry until some memory is available.
210		273
		274	Example: The following is the C<realloc> function that libev itself uses
		275	which should work with C<realloc> and C<free> functions of all kinds and
		276	is probably a good basis for your own implementation.
		277
		278	static void *
		279	ev_realloc_emul (void *ptr, long size) EV_NOEXCEPT
		280	{
		281	if (size)
		282	return realloc (ptr, size);
		283
		284	free (ptr);
		285	return 0;
		286	}
		287
211	Example: Replace the libev allocator with one that waits a bit and then	288	Example: Replace the libev allocator with one that waits a bit and then
212	retries).	289	retries.
213		290
214	static void *	291	static void *
215	persistent_realloc (void *ptr, size_t size)	292	persistent_realloc (void *ptr, size_t size)
216	{	293	{
		294	if (!size)
		295	{
		296	free (ptr);
		297	return 0;
		298	}
		299
217	for (;;)	300	for (;;)
218	{	301	{
219	void *newptr = realloc (ptr, size);	302	void *newptr = realloc (ptr, size);
220		303
221	if (newptr)	304	if (newptr)
…		…
226	}	309	}
227		310
228	...	311	...
229	ev_set_allocator (persistent_realloc);	312	ev_set_allocator (persistent_realloc);
230		313
231	=item ev_set_syserr_cb (void (*cb)(const char *msg));	314	=item ev_set_syserr_cb (void (*cb)(const char *msg) throw ())
232		315
233	Set the callback function to call on a retryable syscall error (such	316	Set the callback function to call on a retryable system call error (such
234	as failed select, poll, epoll_wait). The message is a printable string	317	as failed select, poll, epoll_wait). The message is a printable string
235	indicating the system call or subsystem causing the problem. If this	318	indicating the system call or subsystem causing the problem. If this
236	callback is set, then libev will expect it to remedy the sitution, no	319	callback is set, then libev will expect it to remedy the situation, no
237	matter what, when it returns. That is, libev will generally retry the	320	matter what, when it returns. That is, libev will generally retry the
238	requested operation, or, if the condition doesn't go away, do bad stuff	321	requested operation, or, if the condition doesn't go away, do bad stuff
239	(such as abort).	322	(such as abort).
240		323
241	Example: This is basically the same thing that libev does internally, too.	324	Example: This is basically the same thing that libev does internally, too.
…		…
248	}	331	}
249		332
250	...	333	...
251	ev_set_syserr_cb (fatal_error);	334	ev_set_syserr_cb (fatal_error);
252		335
		336	=item ev_feed_signal (int signum)
		337
		338	This function can be used to "simulate" a signal receive. It is completely
		339	safe to call this function at any time, from any context, including signal
		340	handlers or random threads.
		341
		342	Its main use is to customise signal handling in your process, especially
		343	in the presence of threads. For example, you could block signals
		344	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		345	creating any loops), and in one thread, use C<sigwait> or any other
		346	mechanism to wait for signals, then "deliver" them to libev by calling
		347	C<ev_feed_signal>.
		348
253	=back	349	=back
254		350
255	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	351	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
256		352
257	An event loop is described by a C<struct ev_loop *>. The library knows two	353	An event loop is described by a C<struct ev_loop *> (the C<struct> is
258	types of such loops, the I<default> loop, which supports signals and child	354	I<not> optional in this case unless libev 3 compatibility is disabled, as
259	events, and dynamically created loops which do not.	355	libev 3 had an C<ev_loop> function colliding with the struct name).
		356
		357	The library knows two types of such loops, the I<default> loop, which
		358	supports child process events, and dynamically created event loops which
		359	do not.
260		360
261	=over 4	361	=over 4
262		362
263	=item struct ev_loop *ev_default_loop (unsigned int flags)	363	=item struct ev_loop *ev_default_loop (unsigned int flags)
264		364
265	This will initialise the default event loop if it hasn't been initialised	365	This returns the "default" event loop object, which is what you should
266	yet and return it. If the default loop could not be initialised, returns	366	normally use when you just need "the event loop". Event loop objects and
267	false. If it already was initialised it simply returns it (and ignores the	367	the C<flags> parameter are described in more detail in the entry for
268	flags. If that is troubling you, check C<ev_backend ()> afterwards).	368	C<ev_loop_new>.
		369
		370	If the default loop is already initialised then this function simply
		371	returns it (and ignores the flags. If that is troubling you, check
		372	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		373	flags, which should almost always be C<0>, unless the caller is also the
		374	one calling C<ev_run> or otherwise qualifies as "the main program".
269		375
270	If you don't know what event loop to use, use the one returned from this	376	If you don't know what event loop to use, use the one returned from this
271	function.	377	function (or via the C<EV_DEFAULT> macro).
272		378
273	Note that this function is I<not> thread-safe, so if you want to use it	379	Note that this function is I<not> thread-safe, so if you want to use it
274	from multiple threads, you have to lock (note also that this is unlikely,	380	from multiple threads, you have to employ some kind of mutex (note also
275	as loops cannot bes hared easily between threads anyway).	381	that this case is unlikely, as loops cannot be shared easily between
		382	threads anyway).
276		383
277	The default loop is the only loop that can handle C<ev_signal> and	384	The default loop is the only loop that can handle C<ev_child> watchers,
278	C<ev_child> watchers, and to do this, it always registers a handler	385	and to do this, it always registers a handler for C<SIGCHLD>. If this is
279	for C<SIGCHLD>. If this is a problem for your app you can either	386	a problem for your application you can either create a dynamic loop with
280	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you	387	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
281	can simply overwrite the C<SIGCHLD> signal handler I<after> calling	388	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
282	C<ev_default_init>.	389
		390	Example: This is the most typical usage.
		391
		392	if (!ev_default_loop (0))
		393	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		394
		395	Example: Restrict libev to the select and poll backends, and do not allow
		396	environment settings to be taken into account:
		397
		398	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		399
		400	=item struct ev_loop *ev_loop_new (unsigned int flags)
		401
		402	This will create and initialise a new event loop object. If the loop
		403	could not be initialised, returns false.
		404
		405	This function is thread-safe, and one common way to use libev with
		406	threads is indeed to create one loop per thread, and using the default
		407	loop in the "main" or "initial" thread.
283		408
284	The flags argument can be used to specify special behaviour or specific	409	The flags argument can be used to specify special behaviour or specific
285	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	410	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
286		411
287	The following flags are supported:	412	The following flags are supported:
…		…
293	The default flags value. Use this if you have no clue (it's the right	418	The default flags value. Use this if you have no clue (it's the right
294	thing, believe me).	419	thing, believe me).
295		420
296	=item C<EVFLAG_NOENV>	421	=item C<EVFLAG_NOENV>
297		422
298	If this flag bit is ored into the flag value (or the program runs setuid	423	If this flag bit is or'ed into the flag value (or the program runs setuid
299	or setgid) then libev will I<not> look at the environment variable	424	or setgid) then libev will I<not> look at the environment variable
300	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	425	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
301	override the flags completely if it is found in the environment. This is	426	override the flags completely if it is found in the environment. This is
302	useful to try out specific backends to test their performance, or to work	427	useful to try out specific backends to test their performance, to work
303	around bugs.	428	around bugs, or to make libev threadsafe (accessing environment variables
		429	cannot be done in a threadsafe way, but usually it works if no other
		430	thread modifies them).
304		431
305	=item C<EVFLAG_FORKCHECK>	432	=item C<EVFLAG_FORKCHECK>
306		433
307	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after	434	Instead of calling C<ev_loop_fork> manually after a fork, you can also
308	a fork, you can also make libev check for a fork in each iteration by	435	make libev check for a fork in each iteration by enabling this flag.
309	enabling this flag.
310		436
311	This works by calling C<getpid ()> on every iteration of the loop,	437	This works by calling C<getpid ()> on every iteration of the loop,
312	and thus this might slow down your event loop if you do a lot of loop	438	and thus this might slow down your event loop if you do a lot of loop
313	iterations and little real work, but is usually not noticeable (on my	439	iterations and little real work, but is usually not noticeable (on my
314	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence	440	GNU/Linux system for example, C<getpid> is actually a simple 5-insn
315	without a syscall and thus I<very> fast, but my GNU/Linux system also has	441	sequence without a system call and thus I<very> fast, but my GNU/Linux
316	C<pthread_atfork> which is even faster).	442	system also has C<pthread_atfork> which is even faster). (Update: glibc
		443	versions 2.25 apparently removed the C<getpid> optimisation again).
317		444
318	The big advantage of this flag is that you can forget about fork (and	445	The big advantage of this flag is that you can forget about fork (and
319	forget about forgetting to tell libev about forking) when you use this	446	forget about forgetting to tell libev about forking, although you still
320	flag.	447	have to ignore C<SIGPIPE>) when you use this flag.
321		448
322	This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS>	449	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
323	environment variable.	450	environment variable.
		451
		452	=item C<EVFLAG_NOINOTIFY>
		453
		454	When this flag is specified, then libev will not attempt to use the
		455	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
		456	testing, this flag can be useful to conserve inotify file descriptors, as
		457	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		458
		459	=item C<EVFLAG_SIGNALFD>
		460
		461	When this flag is specified, then libev will attempt to use the
		462	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
		463	delivers signals synchronously, which makes it both faster and might make
		464	it possible to get the queued signal data. It can also simplify signal
		465	handling with threads, as long as you properly block signals in your
		466	threads that are not interested in handling them.
		467
		468	Signalfd will not be used by default as this changes your signal mask, and
		469	there are a lot of shoddy libraries and programs (glib's threadpool for
		470	example) that can't properly initialise their signal masks.
		471
		472	=item C<EVFLAG_NOSIGMASK>
		473
		474	When this flag is specified, then libev will avoid to modify the signal
		475	mask. Specifically, this means you have to make sure signals are unblocked
		476	when you want to receive them.
		477
		478	This behaviour is useful when you want to do your own signal handling, or
		479	want to handle signals only in specific threads and want to avoid libev
		480	unblocking the signals.
		481
		482	It's also required by POSIX in a threaded program, as libev calls
		483	C<sigprocmask>, whose behaviour is officially unspecified.
		484
		485	=item C<EVFLAG_NOTIMERFD>
		486
		487	When this flag is specified, the libev will avoid using a C<timerfd> to
		488	detect time jumps. It will still be able to detect time jumps, but takes
		489	longer and has a lower accuracy in doing so, but saves a file descriptor
		490	per loop.
		491
		492	The current implementation only tries to use a C<timerfd> when the first
		493	C<ev_periodic> watcher is started and falls back on other methods if it
		494	cannot be created, but this behaviour might change in the future.
324		495
325	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	496	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
326		497
327	This is your standard select(2) backend. Not I<completely> standard, as	498	This is your standard select(2) backend. Not I<completely> standard, as
328	libev tries to roll its own fd_set with no limits on the number of fds,	499	libev tries to roll its own fd_set with no limits on the number of fds,
329	but if that fails, expect a fairly low limit on the number of fds when	500	but if that fails, expect a fairly low limit on the number of fds when
330	using this backend. It doesn't scale too well (O(highest_fd)), but its	501	using this backend. It doesn't scale too well (O(highest_fd)), but its
331	usually the fastest backend for a low number of (low-numbered :) fds.	502	usually the fastest backend for a low number of (low-numbered :) fds.
332		503
333	To get good performance out of this backend you need a high amount of	504	To get good performance out of this backend you need a high amount of
334	parallelity (most of the file descriptors should be busy). If you are	505	parallelism (most of the file descriptors should be busy). If you are
335	writing a server, you should C<accept ()> in a loop to accept as many	506	writing a server, you should C<accept ()> in a loop to accept as many
336	connections as possible during one iteration. You might also want to have	507	connections as possible during one iteration. You might also want to have
337	a look at C<ev_set_io_collect_interval ()> to increase the amount of	508	a look at C<ev_set_io_collect_interval ()> to increase the amount of
338	readyness notifications you get per iteration.	509	readiness notifications you get per iteration.
		510
		511	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		512	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		513	C<exceptfds> set on that platform).
339		514
340	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	515	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
341		516
342	And this is your standard poll(2) backend. It's more complicated	517	And this is your standard poll(2) backend. It's more complicated
343	than select, but handles sparse fds better and has no artificial	518	than select, but handles sparse fds better and has no artificial
344	limit on the number of fds you can use (except it will slow down	519	limit on the number of fds you can use (except it will slow down
345	considerably with a lot of inactive fds). It scales similarly to select,	520	considerably with a lot of inactive fds). It scales similarly to select,
346	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for	521	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
347	performance tips.	522	performance tips.
348		523
		524	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		525	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		526
349	=item C<EVBACKEND_EPOLL> (value 4, Linux)	527	=item C<EVBACKEND_EPOLL> (value 4, Linux)
350		528
		529	Use the Linux-specific epoll(7) interface (for both pre- and post-2.6.9
		530	kernels).
		531
351	For few fds, this backend is a bit little slower than poll and select,	532	For few fds, this backend is a bit little slower than poll and select, but
352	but it scales phenomenally better. While poll and select usually scale	533	it scales phenomenally better. While poll and select usually scale like
353	like O(total_fds) where n is the total number of fds (or the highest fd),	534	O(total_fds) where total_fds is the total number of fds (or the highest
354	epoll scales either O(1) or O(active_fds). The epoll design has a number	535	fd), epoll scales either O(1) or O(active_fds).
355	of shortcomings, such as silently dropping events in some hard-to-detect	536
356	cases and requiring a syscall per fd change, no fork support and bad	537	The epoll mechanism deserves honorable mention as the most misdesigned
357	support for dup.	538	of the more advanced event mechanisms: mere annoyances include silently
		539	dropping file descriptors, requiring a system call per change per file
		540	descriptor (and unnecessary guessing of parameters), problems with dup,
		541	returning before the timeout value, resulting in additional iterations
		542	(and only giving 5ms accuracy while select on the same platform gives
		543	0.1ms) and so on. The biggest issue is fork races, however - if a program
		544	forks then I<both> parent and child process have to recreate the epoll
		545	set, which can take considerable time (one syscall per file descriptor)
		546	and is of course hard to detect.
		547
		548	Epoll is also notoriously buggy - embedding epoll fds I<should> work,
		549	but of course I<doesn't>, and epoll just loves to report events for
		550	totally I<different> file descriptors (even already closed ones, so
		551	one cannot even remove them from the set) than registered in the set
		552	(especially on SMP systems). Libev tries to counter these spurious
		553	notifications by employing an additional generation counter and comparing
		554	that against the events to filter out spurious ones, recreating the set
		555	when required. Epoll also erroneously rounds down timeouts, but gives you
		556	no way to know when and by how much, so sometimes you have to busy-wait
		557	because epoll returns immediately despite a nonzero timeout. And last
		558	not least, it also refuses to work with some file descriptors which work
		559	perfectly fine with C<select> (files, many character devices...).
		560
		561	Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
		562	cobbled together in a hurry, no thought to design or interaction with
		563	others. Oh, the pain, will it ever stop...
358		564
359	While stopping, setting and starting an I/O watcher in the same iteration	565	While stopping, setting and starting an I/O watcher in the same iteration
360	will result in some caching, there is still a syscall per such incident	566	will result in some caching, there is still a system call per such
361	(because the fd could point to a different file description now), so its	567	incident (because the same I<file descriptor> could point to a different
362	best to avoid that. Also, C<dup ()>'ed file descriptors might not work	568	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
363	very well if you register events for both fds.	569	file descriptors might not work very well if you register events for both
364		570	file descriptors.
365	Please note that epoll sometimes generates spurious notifications, so you
366	need to use non-blocking I/O or other means to avoid blocking when no data
367	(or space) is available.
368		571
369	Best performance from this backend is achieved by not unregistering all	572	Best performance from this backend is achieved by not unregistering all
370	watchers for a file descriptor until it has been closed, if possible, i.e.	573	watchers for a file descriptor until it has been closed, if possible,
371	keep at least one watcher active per fd at all times.	574	i.e. keep at least one watcher active per fd at all times. Stopping and
		575	starting a watcher (without re-setting it) also usually doesn't cause
		576	extra overhead. A fork can both result in spurious notifications as well
		577	as in libev having to destroy and recreate the epoll object, which can
		578	take considerable time and thus should be avoided.
372		579
		580	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		581	faster than epoll for maybe up to a hundred file descriptors, depending on
		582	the usage. So sad.
		583
373	While nominally embeddeble in other event loops, this feature is broken in	584	While nominally embeddable in other event loops, this feature is broken in
374	all kernel versions tested so far.	585	a lot of kernel revisions, but probably(!) works in current versions.
		586
		587	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		588	C<EVBACKEND_POLL>.
		589
		590	=item C<EVBACKEND_LINUXAIO> (value 64, Linux)
		591
		592	Use the Linux-specific Linux AIO (I<not> C<< aio(7) >> but C<<
		593	io_submit(2) >>) event interface available in post-4.18 kernels (but libev
		594	only tries to use it in 4.19+).
		595
		596	This is another Linux train wreck of an event interface.
		597
		598	If this backend works for you (as of this writing, it was very
		599	experimental), it is the best event interface available on Linux and might
		600	be well worth enabling it - if it isn't available in your kernel this will
		601	be detected and this backend will be skipped.
		602
		603	This backend can batch oneshot requests and supports a user-space ring
		604	buffer to receive events. It also doesn't suffer from most of the design
		605	problems of epoll (such as not being able to remove event sources from
		606	the epoll set), and generally sounds too good to be true. Because, this
		607	being the Linux kernel, of course it suffers from a whole new set of
		608	limitations, forcing you to fall back to epoll, inheriting all its design
		609	issues.
		610
		611	For one, it is not easily embeddable (but probably could be done using
		612	an event fd at some extra overhead). It also is subject to a system wide
		613	limit that can be configured in F</proc/sys/fs/aio-max-nr>. If no AIO
		614	requests are left, this backend will be skipped during initialisation, and
		615	will switch to epoll when the loop is active.
		616
		617	Most problematic in practice, however, is that not all file descriptors
		618	work with it. For example, in Linux 5.1, TCP sockets, pipes, event fds,
		619	files, F</dev/null> and many others are supported, but ttys do not work
		620	properly (a known bug that the kernel developers don't care about, see
		621	L<https://lore.kernel.org/patchwork/patch/1047453/>), so this is not
		622	(yet?) a generic event polling interface.
		623
		624	Overall, it seems the Linux developers just don't want it to have a
		625	generic event handling mechanism other than C<select> or C<poll>.
		626
		627	To work around all these problem, the current version of libev uses its
		628	epoll backend as a fallback for file descriptor types that do not work. Or
		629	falls back completely to epoll if the kernel acts up.
		630
		631	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		632	C<EVBACKEND_POLL>.
375		633
376	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	634	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
377		635
378	Kqueue deserves special mention, as at the time of this writing, it	636	Kqueue deserves special mention, as at the time this backend was
379	was broken on all BSDs except NetBSD (usually it doesn't work reliably	637	implemented, it was broken on all BSDs except NetBSD (usually it doesn't
380	with anything but sockets and pipes, except on Darwin, where of course	638	work reliably with anything but sockets and pipes, except on Darwin,
381	it's completely useless). For this reason it's not being "autodetected"	639	where of course it's completely useless). Unlike epoll, however, whose
382	unless you explicitly specify it explicitly in the flags (i.e. using	640	brokenness is by design, these kqueue bugs can be (and mostly have been)
383	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	641	fixed without API changes to existing programs. For this reason it's not
384	system like NetBSD.	642	being "auto-detected" on all platforms unless you explicitly specify it
		643	in the flags (i.e. using C<EVBACKEND_KQUEUE>) or libev was compiled on a
		644	known-to-be-good (-enough) system like NetBSD.
385		645
386	You still can embed kqueue into a normal poll or select backend and use it	646	You still can embed kqueue into a normal poll or select backend and use it
387	only for sockets (after having made sure that sockets work with kqueue on	647	only for sockets (after having made sure that sockets work with kqueue on
388	the target platform). See C<ev_embed> watchers for more info.	648	the target platform). See C<ev_embed> watchers for more info.
389		649
390	It scales in the same way as the epoll backend, but the interface to the	650	It scales in the same way as the epoll backend, but the interface to the
391	kernel is more efficient (which says nothing about its actual speed, of	651	kernel is more efficient (which says nothing about its actual speed, of
392	course). While stopping, setting and starting an I/O watcher does never	652	course). While stopping, setting and starting an I/O watcher does never
393	cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to	653	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
394	two event changes per incident, support for C<fork ()> is very bad and it	654	two event changes per incident. Support for C<fork ()> is very bad (you
		655	might have to leak fds on fork, but it's more sane than epoll) and it
395	drops fds silently in similarly hard-to-detect cases.	656	drops fds silently in similarly hard-to-detect cases.
396		657
397	This backend usually performs well under most conditions.	658	This backend usually performs well under most conditions.
398		659
399	While nominally embeddable in other event loops, this doesn't work	660	While nominally embeddable in other event loops, this doesn't work
400	everywhere, so you might need to test for this. And since it is broken	661	everywhere, so you might need to test for this. And since it is broken
401	almost everywhere, you should only use it when you have a lot of sockets	662	almost everywhere, you should only use it when you have a lot of sockets
402	(for which it usually works), by embedding it into another event loop	663	(for which it usually works), by embedding it into another event loop
403	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for	664	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
404	sockets.	665	also broken on OS X)) and, did I mention it, using it only for sockets.
		666
		667	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		668	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		669	C<NOTE_EOF>.
405		670
406	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	671	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
407		672
408	This is not implemented yet (and might never be, unless you send me an	673	This is not implemented yet (and might never be, unless you send me an
409	implementation). According to reports, C</dev/poll> only supports sockets	674	implementation). According to reports, C</dev/poll> only supports sockets
…		…
413	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	678	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
414		679
415	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	680	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
416	it's really slow, but it still scales very well (O(active_fds)).	681	it's really slow, but it still scales very well (O(active_fds)).
417		682
418	Please note that solaris event ports can deliver a lot of spurious
419	notifications, so you need to use non-blocking I/O or other means to avoid
420	blocking when no data (or space) is available.
421
422	While this backend scales well, it requires one system call per active	683	While this backend scales well, it requires one system call per active
423	file descriptor per loop iteration. For small and medium numbers of file	684	file descriptor per loop iteration. For small and medium numbers of file
424	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend	685	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
425	might perform better.	686	might perform better.
426		687
427	On the positive side, ignoring the spurious readyness notifications, this	688	On the positive side, this backend actually performed fully to
428	backend actually performed to specification in all tests and is fully	689	specification in all tests and is fully embeddable, which is a rare feat
429	embeddable, which is a rare feat among the OS-specific backends.	690	among the OS-specific backends (I vastly prefer correctness over speed
		691	hacks).
		692
		693	On the negative side, the interface is I<bizarre> - so bizarre that
		694	even sun itself gets it wrong in their code examples: The event polling
		695	function sometimes returns events to the caller even though an error
		696	occurred, but with no indication whether it has done so or not (yes, it's
		697	even documented that way) - deadly for edge-triggered interfaces where you
		698	absolutely have to know whether an event occurred or not because you have
		699	to re-arm the watcher.
		700
		701	Fortunately libev seems to be able to work around these idiocies.
		702
		703	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		704	C<EVBACKEND_POLL>.
430		705
431	=item C<EVBACKEND_ALL>	706	=item C<EVBACKEND_ALL>
432		707
433	Try all backends (even potentially broken ones that wouldn't be tried	708	Try all backends (even potentially broken ones that wouldn't be tried
434	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	709	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
435	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	710	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
436		711
437	It is definitely not recommended to use this flag.	712	It is definitely not recommended to use this flag, use whatever
		713	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		714	at all.
		715
		716	=item C<EVBACKEND_MASK>
		717
		718	Not a backend at all, but a mask to select all backend bits from a
		719	C<flags> value, in case you want to mask out any backends from a flags
		720	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
438		721
439	=back	722	=back
440		723
441	If one or more of these are ored into the flags value, then only these	724	If one or more of the backend flags are or'ed into the flags value,
442	backends will be tried (in the reverse order as listed here). If none are	725	then only these backends will be tried (in the reverse order as listed
443	specified, all backends in C<ev_recommended_backends ()> will be tried.	726	here). If none are specified, all backends in C<ev_recommended_backends
444		727	()> will be tried.
445	The most typical usage is like this:
446
447	if (!ev_default_loop (0))
448	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
449
450	Restrict libev to the select and poll backends, and do not allow
451	environment settings to be taken into account:
452
453	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
454
455	Use whatever libev has to offer, but make sure that kqueue is used if
456	available (warning, breaks stuff, best use only with your own private
457	event loop and only if you know the OS supports your types of fds):
458
459	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
460
461	=item struct ev_loop *ev_loop_new (unsigned int flags)
462
463	Similar to C<ev_default_loop>, but always creates a new event loop that is
464	always distinct from the default loop. Unlike the default loop, it cannot
465	handle signal and child watchers, and attempts to do so will be greeted by
466	undefined behaviour (or a failed assertion if assertions are enabled).
467
468	Note that this function I<is> thread-safe, and the recommended way to use
469	libev with threads is indeed to create one loop per thread, and using the
470	default loop in the "main" or "initial" thread.
471		728
472	Example: Try to create a event loop that uses epoll and nothing else.	729	Example: Try to create a event loop that uses epoll and nothing else.
473		730
474	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	731	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
475	if (!epoller)	732	if (!epoller)
476	fatal ("no epoll found here, maybe it hides under your chair");	733	fatal ("no epoll found here, maybe it hides under your chair");
477		734
		735	Example: Use whatever libev has to offer, but make sure that kqueue is
		736	used if available.
		737
		738	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
		739
		740	Example: Similarly, on linux, you mgiht want to take advantage of the
		741	linux aio backend if possible, but fall back to something else if that
		742	isn't available.
		743
		744	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_LINUXAIO);
		745
478	=item ev_default_destroy ()	746	=item ev_loop_destroy (loop)
479		747
480	Destroys the default loop again (frees all memory and kernel state	748	Destroys an event loop object (frees all memory and kernel state
481	etc.). None of the active event watchers will be stopped in the normal	749	etc.). None of the active event watchers will be stopped in the normal
482	sense, so e.g. C<ev_is_active> might still return true. It is your	750	sense, so e.g. C<ev_is_active> might still return true. It is your
483	responsibility to either stop all watchers cleanly yoursef I<before>	751	responsibility to either stop all watchers cleanly yourself I<before>
484	calling this function, or cope with the fact afterwards (which is usually	752	calling this function, or cope with the fact afterwards (which is usually
485	the easiest thing, you can just ignore the watchers and/or C<free ()> them	753	the easiest thing, you can just ignore the watchers and/or C<free ()> them
486	for example).	754	for example).
487		755
488	Note that certain global state, such as signal state, will not be freed by	756	Note that certain global state, such as signal state (and installed signal
489	this function, and related watchers (such as signal and child watchers)	757	handlers), will not be freed by this function, and related watchers (such
490	would need to be stopped manually.	758	as signal and child watchers) would need to be stopped manually.
491		759
492	In general it is not advisable to call this function except in the	760	This function is normally used on loop objects allocated by
493	rare occasion where you really need to free e.g. the signal handling	761	C<ev_loop_new>, but it can also be used on the default loop returned by
		762	C<ev_default_loop>, in which case it is not thread-safe.
		763
		764	Note that it is not advisable to call this function on the default loop
		765	except in the rare occasion where you really need to free its resources.
494	pipe fds. If you need dynamically allocated loops it is better to use	766	If you need dynamically allocated loops it is better to use C<ev_loop_new>
495	C<ev_loop_new> and C<ev_loop_destroy>).	767	and C<ev_loop_destroy>.
496		768
497	=item ev_loop_destroy (loop)	769	=item ev_loop_fork (loop)
498		770
499	Like C<ev_default_destroy>, but destroys an event loop created by an
500	earlier call to C<ev_loop_new>.
501
502	=item ev_default_fork ()
503
504	This function sets a flag that causes subsequent C<ev_loop> iterations	771	This function sets a flag that causes subsequent C<ev_run> iterations
505	to reinitialise the kernel state for backends that have one. Despite the	772	to reinitialise the kernel state for backends that have one. Despite
506	name, you can call it anytime, but it makes most sense after forking, in	773	the name, you can call it anytime you are allowed to start or stop
507	the child process (or both child and parent, but that again makes little	774	watchers (except inside an C<ev_prepare> callback), but it makes most
508	sense). You I<must> call it in the child before using any of the libev	775	sense after forking, in the child process. You I<must> call it (or use
509	functions, and it will only take effect at the next C<ev_loop> iteration.	776	C<EVFLAG_FORKCHECK>) in the child before resuming or calling C<ev_run>.
		777
		778	In addition, if you want to reuse a loop (via this function or
		779	C<EVFLAG_FORKCHECK>), you I<also> have to ignore C<SIGPIPE>.
		780
		781	Again, you I<have> to call it on I<any> loop that you want to re-use after
		782	a fork, I<even if you do not plan to use the loop in the parent>. This is
		783	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		784	during fork.
510		785
511	On the other hand, you only need to call this function in the child	786	On the other hand, you only need to call this function in the child
512	process if and only if you want to use the event library in the child. If	787	process if and only if you want to use the event loop in the child. If
513	you just fork+exec, you don't have to call it at all.	788	you just fork+exec or create a new loop in the child, you don't have to
		789	call it at all (in fact, C<epoll> is so badly broken that it makes a
		790	difference, but libev will usually detect this case on its own and do a
		791	costly reset of the backend).
514		792
515	The function itself is quite fast and it's usually not a problem to call	793	The function itself is quite fast and it's usually not a problem to call
516	it just in case after a fork. To make this easy, the function will fit in	794	it just in case after a fork.
517	quite nicely into a call to C<pthread_atfork>:
518		795
		796	Example: Automate calling C<ev_loop_fork> on the default loop when
		797	using pthreads.
		798
		799	static void
		800	post_fork_child (void)
		801	{
		802	ev_loop_fork (EV_DEFAULT);
		803	}
		804
		805	...
519	pthread_atfork (0, 0, ev_default_fork);	806	pthread_atfork (0, 0, post_fork_child);
520
521	=item ev_loop_fork (loop)
522
523	Like C<ev_default_fork>, but acts on an event loop created by
524	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
525	after fork, and how you do this is entirely your own problem.
526		807
527	=item int ev_is_default_loop (loop)	808	=item int ev_is_default_loop (loop)
528		809
529	Returns true when the given loop actually is the default loop, false otherwise.	810	Returns true when the given loop is, in fact, the default loop, and false
		811	otherwise.
530		812
531	=item unsigned int ev_loop_count (loop)	813	=item unsigned int ev_iteration (loop)
532		814
533	Returns the count of loop iterations for the loop, which is identical to	815	Returns the current iteration count for the event loop, which is identical
534	the number of times libev did poll for new events. It starts at C<0> and	816	to the number of times libev did poll for new events. It starts at C<0>
535	happily wraps around with enough iterations.	817	and happily wraps around with enough iterations.
536		818
537	This value can sometimes be useful as a generation counter of sorts (it	819	This value can sometimes be useful as a generation counter of sorts (it
538	"ticks" the number of loop iterations), as it roughly corresponds with	820	"ticks" the number of loop iterations), as it roughly corresponds with
539	C<ev_prepare> and C<ev_check> calls.	821	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		822	prepare and check phases.
		823
		824	=item unsigned int ev_depth (loop)
		825
		826	Returns the number of times C<ev_run> was entered minus the number of
		827	times C<ev_run> was exited normally, in other words, the recursion depth.
		828
		829	Outside C<ev_run>, this number is zero. In a callback, this number is
		830	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		831	in which case it is higher.
		832
		833	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
		834	throwing an exception etc.), doesn't count as "exit" - consider this
		835	as a hint to avoid such ungentleman-like behaviour unless it's really
		836	convenient, in which case it is fully supported.
540		837
541	=item unsigned int ev_backend (loop)	838	=item unsigned int ev_backend (loop)
542		839
543	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	840	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
544	use.	841	use.
…		…
549	received events and started processing them. This timestamp does not	846	received events and started processing them. This timestamp does not
550	change as long as callbacks are being processed, and this is also the base	847	change as long as callbacks are being processed, and this is also the base
551	time used for relative timers. You can treat it as the timestamp of the	848	time used for relative timers. You can treat it as the timestamp of the
552	event occurring (or more correctly, libev finding out about it).	849	event occurring (or more correctly, libev finding out about it).
553		850
		851	=item ev_now_update (loop)
		852
		853	Establishes the current time by querying the kernel, updating the time
		854	returned by C<ev_now ()> in the progress. This is a costly operation and
		855	is usually done automatically within C<ev_run ()>.
		856
		857	This function is rarely useful, but when some event callback runs for a
		858	very long time without entering the event loop, updating libev's idea of
		859	the current time is a good idea.
		860
		861	See also L</The special problem of time updates> in the C<ev_timer> section.
		862
		863	=item ev_suspend (loop)
		864
		865	=item ev_resume (loop)
		866
		867	These two functions suspend and resume an event loop, for use when the
		868	loop is not used for a while and timeouts should not be processed.
		869
		870	A typical use case would be an interactive program such as a game: When
		871	the user presses C<^Z> to suspend the game and resumes it an hour later it
		872	would be best to handle timeouts as if no time had actually passed while
		873	the program was suspended. This can be achieved by calling C<ev_suspend>
		874	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		875	C<ev_resume> directly afterwards to resume timer processing.
		876
		877	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		878	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		879	will be rescheduled (that is, they will lose any events that would have
		880	occurred while suspended).
		881
		882	After calling C<ev_suspend> you B<must not> call I<any> function on the
		883	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		884	without a previous call to C<ev_suspend>.
		885
		886	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		887	event loop time (see C<ev_now_update>).
		888
554	=item ev_loop (loop, int flags)	889	=item bool ev_run (loop, int flags)
555		890
556	Finally, this is it, the event handler. This function usually is called	891	Finally, this is it, the event handler. This function usually is called
557	after you initialised all your watchers and you want to start handling	892	after you have initialised all your watchers and you want to start
558	events.	893	handling events. It will ask the operating system for any new events, call
		894	the watcher callbacks, and then repeat the whole process indefinitely: This
		895	is why event loops are called I<loops>.
559		896
560	If the flags argument is specified as C<0>, it will not return until	897	If the flags argument is specified as C<0>, it will keep handling events
561	either no event watchers are active anymore or C<ev_unloop> was called.	898	until either no event watchers are active anymore or C<ev_break> was
		899	called.
562		900
		901	The return value is false if there are no more active watchers (which
		902	usually means "all jobs done" or "deadlock"), and true in all other cases
		903	(which usually means " you should call C<ev_run> again").
		904
563	Please note that an explicit C<ev_unloop> is usually better than	905	Please note that an explicit C<ev_break> is usually better than
564	relying on all watchers to be stopped when deciding when a program has	906	relying on all watchers to be stopped when deciding when a program has
565	finished (especially in interactive programs), but having a program that	907	finished (especially in interactive programs), but having a program
566	automatically loops as long as it has to and no longer by virtue of	908	that automatically loops as long as it has to and no longer by virtue
567	relying on its watchers stopping correctly is a thing of beauty.	909	of relying on its watchers stopping correctly, that is truly a thing of
		910	beauty.
568		911
		912	This function is I<mostly> exception-safe - you can break out of a
		913	C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		914	exception and so on. This does not decrement the C<ev_depth> value, nor
		915	will it clear any outstanding C<EVBREAK_ONE> breaks.
		916
569	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	917	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
570	those events and any outstanding ones, but will not block your process in	918	those events and any already outstanding ones, but will not wait and
571	case there are no events and will return after one iteration of the loop.	919	block your process in case there are no events and will return after one
		920	iteration of the loop. This is sometimes useful to poll and handle new
		921	events while doing lengthy calculations, to keep the program responsive.
572		922
573	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	923	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
574	neccessary) and will handle those and any outstanding ones. It will block	924	necessary) and will handle those and any already outstanding ones. It
575	your process until at least one new event arrives, and will return after	925	will block your process until at least one new event arrives (which could
576	one iteration of the loop. This is useful if you are waiting for some	926	be an event internal to libev itself, so there is no guarantee that a
577	external event in conjunction with something not expressible using other	927	user-registered callback will be called), and will return after one
		928	iteration of the loop.
		929
		930	This is useful if you are waiting for some external event in conjunction
		931	with something not expressible using other libev watchers (i.e. "roll your
578	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	932	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
579	usually a better approach for this kind of thing.	933	usually a better approach for this kind of thing.
580		934
581	Here are the gory details of what C<ev_loop> does:	935	Here are the gory details of what C<ev_run> does (this is for your
		936	understanding, not a guarantee that things will work exactly like this in
		937	future versions):
582		938
		939	- Increment loop depth.
		940	- Reset the ev_break status.
583	- Before the first iteration, call any pending watchers.	941	- Before the first iteration, call any pending watchers.
		942	LOOP:
584	* If EVFLAG_FORKCHECK was used, check for a fork.	943	- If EVFLAG_FORKCHECK was used, check for a fork.
585	- If a fork was detected, queue and call all fork watchers.	944	- If a fork was detected (by any means), queue and call all fork watchers.
586	- Queue and call all prepare watchers.	945	- Queue and call all prepare watchers.
		946	- If ev_break was called, goto FINISH.
587	- If we have been forked, recreate the kernel state.	947	- If we have been forked, detach and recreate the kernel state
		948	as to not disturb the other process.
588	- Update the kernel state with all outstanding changes.	949	- Update the kernel state with all outstanding changes.
589	- Update the "event loop time".	950	- Update the "event loop time" (ev_now ()).
590	- Calculate for how long to sleep or block, if at all	951	- Calculate for how long to sleep or block, if at all
591	(active idle watchers, EVLOOP_NONBLOCK or not having	952	(active idle watchers, EVRUN_NOWAIT or not having
592	any active watchers at all will result in not sleeping).	953	any active watchers at all will result in not sleeping).
593	- Sleep if the I/O and timer collect interval say so.	954	- Sleep if the I/O and timer collect interval say so.
		955	- Increment loop iteration counter.
594	- Block the process, waiting for any events.	956	- Block the process, waiting for any events.
595	- Queue all outstanding I/O (fd) events.	957	- Queue all outstanding I/O (fd) events.
596	- Update the "event loop time" and do time jump handling.	958	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
597	- Queue all outstanding timers.	959	- Queue all expired timers.
598	- Queue all outstanding periodics.	960	- Queue all expired periodics.
599	- If no events are pending now, queue all idle watchers.	961	- Queue all idle watchers with priority higher than that of pending events.
600	- Queue all check watchers.	962	- Queue all check watchers.
601	- Call all queued watchers in reverse order (i.e. check watchers first).	963	- Call all queued watchers in reverse order (i.e. check watchers first).
602	Signals and child watchers are implemented as I/O watchers, and will	964	Signals, async and child watchers are implemented as I/O watchers, and
603	be handled here by queueing them when their watcher gets executed.	965	will be handled here by queueing them when their watcher gets executed.
604	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	966	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
605	were used, or there are no active watchers, return, otherwise	967	were used, or there are no active watchers, goto FINISH, otherwise
606	continue with step *.	968	continue with step LOOP.
		969	FINISH:
		970	- Reset the ev_break status iff it was EVBREAK_ONE.
		971	- Decrement the loop depth.
		972	- Return.
607		973
608	Example: Queue some jobs and then loop until no events are outstanding	974	Example: Queue some jobs and then loop until no events are outstanding
609	anymore.	975	anymore.
610		976
611	... queue jobs here, make sure they register event watchers as long	977	... queue jobs here, make sure they register event watchers as long
612	... as they still have work to do (even an idle watcher will do..)	978	... as they still have work to do (even an idle watcher will do..)
613	ev_loop (my_loop, 0);	979	ev_run (my_loop, 0);
614	... jobs done. yeah!	980	... jobs done or somebody called break. yeah!
615		981
616	=item ev_unloop (loop, how)	982	=item ev_break (loop, how)
617		983
618	Can be used to make a call to C<ev_loop> return early (but only after it	984	Can be used to make a call to C<ev_run> return early (but only after it
619	has processed all outstanding events). The C<how> argument must be either	985	has processed all outstanding events). The C<how> argument must be either
620	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	986	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
621	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	987	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
622		988
623	This "unloop state" will be cleared when entering C<ev_loop> again.	989	This "break state" will be cleared on the next call to C<ev_run>.
		990
		991	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		992	which case it will have no effect.
624		993
625	=item ev_ref (loop)	994	=item ev_ref (loop)
626		995
627	=item ev_unref (loop)	996	=item ev_unref (loop)
628		997
629	Ref/unref can be used to add or remove a reference count on the event	998	Ref/unref can be used to add or remove a reference count on the event
630	loop: Every watcher keeps one reference, and as long as the reference	999	loop: Every watcher keeps one reference, and as long as the reference
631	count is nonzero, C<ev_loop> will not return on its own. If you have	1000	count is nonzero, C<ev_run> will not return on its own.
632	a watcher you never unregister that should not keep C<ev_loop> from	1001
633	returning, ev_unref() after starting, and ev_ref() before stopping it. For	1002	This is useful when you have a watcher that you never intend to
		1003	unregister, but that nevertheless should not keep C<ev_run> from
		1004	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		1005	before stopping it.
		1006
634	example, libev itself uses this for its internal signal pipe: It is not	1007	As an example, libev itself uses this for its internal signal pipe: It
635	visible to the libev user and should not keep C<ev_loop> from exiting if	1008	is not visible to the libev user and should not keep C<ev_run> from
636	no event watchers registered by it are active. It is also an excellent	1009	exiting if no event watchers registered by it are active. It is also an
637	way to do this for generic recurring timers or from within third-party	1010	excellent way to do this for generic recurring timers or from within
638	libraries. Just remember to I<unref after start> and I<ref before stop>	1011	third-party libraries. Just remember to I<unref after start> and I<ref
639	(but only if the watcher wasn't active before, or was active before,	1012	before stop> (but only if the watcher wasn't active before, or was active
640	respectively).	1013	before, respectively. Note also that libev might stop watchers itself
		1014	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		1015	in the callback).
641		1016
642	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	1017	Example: Create a signal watcher, but keep it from keeping C<ev_run>
643	running when nothing else is active.	1018	running when nothing else is active.
644		1019
645	struct ev_signal exitsig;	1020	ev_signal exitsig;
646	ev_signal_init (&exitsig, sig_cb, SIGINT);	1021	ev_signal_init (&exitsig, sig_cb, SIGINT);
647	ev_signal_start (loop, &exitsig);	1022	ev_signal_start (loop, &exitsig);
648	evf_unref (loop);	1023	ev_unref (loop);
649		1024
650	Example: For some weird reason, unregister the above signal handler again.	1025	Example: For some weird reason, unregister the above signal handler again.
651		1026
652	ev_ref (loop);	1027	ev_ref (loop);
653	ev_signal_stop (loop, &exitsig);	1028	ev_signal_stop (loop, &exitsig);
654		1029
655	=item ev_set_io_collect_interval (loop, ev_tstamp interval)	1030	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
656		1031
657	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)	1032	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
658		1033
659	These advanced functions influence the time that libev will spend waiting	1034	These advanced functions influence the time that libev will spend waiting
660	for events. Both are by default C<0>, meaning that libev will try to	1035	for events. Both time intervals are by default C<0>, meaning that libev
661	invoke timer/periodic callbacks and I/O callbacks with minimum latency.	1036	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		1037	latency.
662		1038
663	Setting these to a higher value (the C<interval> I<must> be >= C<0>)	1039	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
664	allows libev to delay invocation of I/O and timer/periodic callbacks to	1040	allows libev to delay invocation of I/O and timer/periodic callbacks
665	increase efficiency of loop iterations.	1041	to increase efficiency of loop iterations (or to increase power-saving
		1042	opportunities).
666		1043
667	The background is that sometimes your program runs just fast enough to	1044	The idea is that sometimes your program runs just fast enough to handle
668	handle one (or very few) event(s) per loop iteration. While this makes	1045	one (or very few) event(s) per loop iteration. While this makes the
669	the program responsive, it also wastes a lot of CPU time to poll for new	1046	program responsive, it also wastes a lot of CPU time to poll for new
670	events, especially with backends like C<select ()> which have a high	1047	events, especially with backends like C<select ()> which have a high
671	overhead for the actual polling but can deliver many events at once.	1048	overhead for the actual polling but can deliver many events at once.
672		1049
673	By setting a higher I<io collect interval> you allow libev to spend more	1050	By setting a higher I<io collect interval> you allow libev to spend more
674	time collecting I/O events, so you can handle more events per iteration,	1051	time collecting I/O events, so you can handle more events per iteration,
675	at the cost of increasing latency. Timeouts (both C<ev_periodic> and	1052	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
676	C<ev_timer>) will be not affected. Setting this to a non-null value will	1053	C<ev_timer>) will not be affected. Setting this to a non-null value will
677	introduce an additional C<ev_sleep ()> call into most loop iterations.	1054	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		1055	sleep time ensures that libev will not poll for I/O events more often then
		1056	once per this interval, on average (as long as the host time resolution is
		1057	good enough).
678		1058
679	Likewise, by setting a higher I<timeout collect interval> you allow libev	1059	Likewise, by setting a higher I<timeout collect interval> you allow libev
680	to spend more time collecting timeouts, at the expense of increased	1060	to spend more time collecting timeouts, at the expense of increased
681	latency (the watcher callback will be called later). C<ev_io> watchers	1061	latency/jitter/inexactness (the watcher callback will be called
682	will not be affected. Setting this to a non-null value will not introduce	1062	later). C<ev_io> watchers will not be affected. Setting this to a non-null
683	any overhead in libev.	1063	value will not introduce any overhead in libev.
684		1064
685	Many (busy) programs can usually benefit by setting the io collect	1065	Many (busy) programs can usually benefit by setting the I/O collect
686	interval to a value near C<0.1> or so, which is often enough for	1066	interval to a value near C<0.1> or so, which is often enough for
687	interactive servers (of course not for games), likewise for timeouts. It	1067	interactive servers (of course not for games), likewise for timeouts. It
688	usually doesn't make much sense to set it to a lower value than C<0.01>,	1068	usually doesn't make much sense to set it to a lower value than C<0.01>,
689	as this approsaches the timing granularity of most systems.	1069	as this approaches the timing granularity of most systems. Note that if
		1070	you do transactions with the outside world and you can't increase the
		1071	parallelity, then this setting will limit your transaction rate (if you
		1072	need to poll once per transaction and the I/O collect interval is 0.01,
		1073	then you can't do more than 100 transactions per second).
		1074
		1075	Setting the I<timeout collect interval> can improve the opportunity for
		1076	saving power, as the program will "bundle" timer callback invocations that
		1077	are "near" in time together, by delaying some, thus reducing the number of
		1078	times the process sleeps and wakes up again. Another useful technique to
		1079	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		1080	they fire on, say, one-second boundaries only.
		1081
		1082	Example: we only need 0.1s timeout granularity, and we wish not to poll
		1083	more often than 100 times per second:
		1084
		1085	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		1086	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		1087
		1088	=item ev_invoke_pending (loop)
		1089
		1090	This call will simply invoke all pending watchers while resetting their
		1091	pending state. Normally, C<ev_run> does this automatically when required,
		1092	but when overriding the invoke callback this call comes handy. This
		1093	function can be invoked from a watcher - this can be useful for example
		1094	when you want to do some lengthy calculation and want to pass further
		1095	event handling to another thread (you still have to make sure only one
		1096	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		1097
		1098	=item int ev_pending_count (loop)
		1099
		1100	Returns the number of pending watchers - zero indicates that no watchers
		1101	are pending.
		1102
		1103	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		1104
		1105	This overrides the invoke pending functionality of the loop: Instead of
		1106	invoking all pending watchers when there are any, C<ev_run> will call
		1107	this callback instead. This is useful, for example, when you want to
		1108	invoke the actual watchers inside another context (another thread etc.).
		1109
		1110	If you want to reset the callback, use C<ev_invoke_pending> as new
		1111	callback.
		1112
		1113	=item ev_set_loop_release_cb (loop, void (*release)(EV_P) throw (), void (*acquire)(EV_P) throw ())
		1114
		1115	Sometimes you want to share the same loop between multiple threads. This
		1116	can be done relatively simply by putting mutex_lock/unlock calls around
		1117	each call to a libev function.
		1118
		1119	However, C<ev_run> can run an indefinite time, so it is not feasible
		1120	to wait for it to return. One way around this is to wake up the event
		1121	loop via C<ev_break> and C<ev_async_send>, another way is to set these
		1122	I<release> and I<acquire> callbacks on the loop.
		1123
		1124	When set, then C<release> will be called just before the thread is
		1125	suspended waiting for new events, and C<acquire> is called just
		1126	afterwards.
		1127
		1128	Ideally, C<release> will just call your mutex_unlock function, and
		1129	C<acquire> will just call the mutex_lock function again.
		1130
		1131	While event loop modifications are allowed between invocations of
		1132	C<release> and C<acquire> (that's their only purpose after all), no
		1133	modifications done will affect the event loop, i.e. adding watchers will
		1134	have no effect on the set of file descriptors being watched, or the time
		1135	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1136	to take note of any changes you made.
		1137
		1138	In theory, threads executing C<ev_run> will be async-cancel safe between
		1139	invocations of C<release> and C<acquire>.
		1140
		1141	See also the locking example in the C<THREADS> section later in this
		1142	document.
		1143
		1144	=item ev_set_userdata (loop, void *data)
		1145
		1146	=item void *ev_userdata (loop)
		1147
		1148	Set and retrieve a single C<void *> associated with a loop. When
		1149	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1150	C<0>.
		1151
		1152	These two functions can be used to associate arbitrary data with a loop,
		1153	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1154	C<acquire> callbacks described above, but of course can be (ab-)used for
		1155	any other purpose as well.
		1156
		1157	=item ev_verify (loop)
		1158
		1159	This function only does something when C<EV_VERIFY> support has been
		1160	compiled in, which is the default for non-minimal builds. It tries to go
		1161	through all internal structures and checks them for validity. If anything
		1162	is found to be inconsistent, it will print an error message to standard
		1163	error and call C<abort ()>.
		1164
		1165	This can be used to catch bugs inside libev itself: under normal
		1166	circumstances, this function will never abort as of course libev keeps its
		1167	data structures consistent.
690		1168
691	=back	1169	=back
692		1170
693		1171
694	=head1 ANATOMY OF A WATCHER	1172	=head1 ANATOMY OF A WATCHER
695		1173
		1174	In the following description, uppercase C<TYPE> in names stands for the
		1175	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1176	watchers and C<ev_io_start> for I/O watchers.
		1177
696	A watcher is a structure that you create and register to record your	1178	A watcher is an opaque structure that you allocate and register to record
697	interest in some event. For instance, if you want to wait for STDIN to	1179	your interest in some event. To make a concrete example, imagine you want
698	become readable, you would create an C<ev_io> watcher for that:	1180	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1181	for that:
699		1182
700	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1183	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
701	{	1184	{
702	ev_io_stop (w);	1185	ev_io_stop (w);
703	ev_unloop (loop, EVUNLOOP_ALL);	1186	ev_break (loop, EVBREAK_ALL);
704	}	1187	}
705		1188
706	struct ev_loop *loop = ev_default_loop (0);	1189	struct ev_loop *loop = ev_default_loop (0);
		1190
707	struct ev_io stdin_watcher;	1191	ev_io stdin_watcher;
		1192
708	ev_init (&stdin_watcher, my_cb);	1193	ev_init (&stdin_watcher, my_cb);
709	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1194	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
710	ev_io_start (loop, &stdin_watcher);	1195	ev_io_start (loop, &stdin_watcher);
		1196
711	ev_loop (loop, 0);	1197	ev_run (loop, 0);
712		1198
713	As you can see, you are responsible for allocating the memory for your	1199	As you can see, you are responsible for allocating the memory for your
714	watcher structures (and it is usually a bad idea to do this on the stack,	1200	watcher structures (and it is I<usually> a bad idea to do this on the
715	although this can sometimes be quite valid).	1201	stack).
716		1202
		1203	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1204	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1205
717	Each watcher structure must be initialised by a call to C<ev_init	1206	Each watcher structure must be initialised by a call to C<ev_init (watcher
718	(watcher *, callback)>, which expects a callback to be provided. This	1207	*, callback)>, which expects a callback to be provided. This callback is
719	callback gets invoked each time the event occurs (or, in the case of io	1208	invoked each time the event occurs (or, in the case of I/O watchers, each
720	watchers, each time the event loop detects that the file descriptor given	1209	time the event loop detects that the file descriptor given is readable
721	is readable and/or writable).	1210	and/or writable).
722		1211
723	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1212	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
724	with arguments specific to this watcher type. There is also a macro	1213	macro to configure it, with arguments specific to the watcher type. There
725	to combine initialisation and setting in one call: C<< ev_<type>_init	1214	is also a macro to combine initialisation and setting in one call: C<<
726	(watcher *, callback, ...) >>.	1215	ev_TYPE_init (watcher *, callback, ...) >>.
727		1216
728	To make the watcher actually watch out for events, you have to start it	1217	To make the watcher actually watch out for events, you have to start it
729	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1218	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
730	*) >>), and you can stop watching for events at any time by calling the	1219	*) >>), and you can stop watching for events at any time by calling the
731	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1220	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
732		1221
733	As long as your watcher is active (has been started but not stopped) you	1222	As long as your watcher is active (has been started but not stopped) you
734	must not touch the values stored in it. Most specifically you must never	1223	must not touch the values stored in it except when explicitly documented
735	reinitialise it or call its C<set> macro.	1224	otherwise. Most specifically you must never reinitialise it or call its
		1225	C<ev_TYPE_set> macro.
736		1226
737	Each and every callback receives the event loop pointer as first, the	1227	Each and every callback receives the event loop pointer as first, the
738	registered watcher structure as second, and a bitset of received events as	1228	registered watcher structure as second, and a bitset of received events as
739	third argument.	1229	third argument.
740		1230
…		…
749	=item C<EV_WRITE>	1239	=item C<EV_WRITE>
750		1240
751	The file descriptor in the C<ev_io> watcher has become readable and/or	1241	The file descriptor in the C<ev_io> watcher has become readable and/or
752	writable.	1242	writable.
753		1243
754	=item C<EV_TIMEOUT>	1244	=item C<EV_TIMER>
755		1245
756	The C<ev_timer> watcher has timed out.	1246	The C<ev_timer> watcher has timed out.
757		1247
758	=item C<EV_PERIODIC>	1248	=item C<EV_PERIODIC>
759		1249
…		…
777		1267
778	=item C<EV_PREPARE>	1268	=item C<EV_PREPARE>
779		1269
780	=item C<EV_CHECK>	1270	=item C<EV_CHECK>
781		1271
782	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1272	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts to
783	to gather new events, and all C<ev_check> watchers are invoked just after	1273	gather new events, and all C<ev_check> watchers are queued (not invoked)
784	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1274	just after C<ev_run> has gathered them, but before it queues any callbacks
		1275	for any received events. That means C<ev_prepare> watchers are the last
		1276	watchers invoked before the event loop sleeps or polls for new events, and
		1277	C<ev_check> watchers will be invoked before any other watchers of the same
		1278	or lower priority within an event loop iteration.
		1279
785	received events. Callbacks of both watcher types can start and stop as	1280	Callbacks of both watcher types can start and stop as many watchers as
786	many watchers as they want, and all of them will be taken into account	1281	they want, and all of them will be taken into account (for example, a
787	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1282	C<ev_prepare> watcher might start an idle watcher to keep C<ev_run> from
788	C<ev_loop> from blocking).	1283	blocking).
789		1284
790	=item C<EV_EMBED>	1285	=item C<EV_EMBED>
791		1286
792	The embedded event loop specified in the C<ev_embed> watcher needs attention.	1287	The embedded event loop specified in the C<ev_embed> watcher needs attention.
793		1288
794	=item C<EV_FORK>	1289	=item C<EV_FORK>
795		1290
796	The event loop has been resumed in the child process after fork (see	1291	The event loop has been resumed in the child process after fork (see
797	C<ev_fork>).	1292	C<ev_fork>).
798		1293
		1294	=item C<EV_CLEANUP>
		1295
		1296	The event loop is about to be destroyed (see C<ev_cleanup>).
		1297
799	=item C<EV_ASYNC>	1298	=item C<EV_ASYNC>
800		1299
801	The given async watcher has been asynchronously notified (see C<ev_async>).	1300	The given async watcher has been asynchronously notified (see C<ev_async>).
802		1301
		1302	=item C<EV_CUSTOM>
		1303
		1304	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1305	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1306
803	=item C<EV_ERROR>	1307	=item C<EV_ERROR>
804		1308
805	An unspecified error has occured, the watcher has been stopped. This might	1309	An unspecified error has occurred, the watcher has been stopped. This might
806	happen because the watcher could not be properly started because libev	1310	happen because the watcher could not be properly started because libev
807	ran out of memory, a file descriptor was found to be closed or any other	1311	ran out of memory, a file descriptor was found to be closed or any other
		1312	problem. Libev considers these application bugs.
		1313
808	problem. You best act on it by reporting the problem and somehow coping	1314	You best act on it by reporting the problem and somehow coping with the
809	with the watcher being stopped.	1315	watcher being stopped. Note that well-written programs should not receive
		1316	an error ever, so when your watcher receives it, this usually indicates a
		1317	bug in your program.
810		1318
811	Libev will usually signal a few "dummy" events together with an error,	1319	Libev will usually signal a few "dummy" events together with an error, for
812	for example it might indicate that a fd is readable or writable, and if	1320	example it might indicate that a fd is readable or writable, and if your
813	your callbacks is well-written it can just attempt the operation and cope	1321	callbacks is well-written it can just attempt the operation and cope with
814	with the error from read() or write(). This will not work in multithreaded	1322	the error from read() or write(). This will not work in multi-threaded
815	programs, though, so beware.	1323	programs, though, as the fd could already be closed and reused for another
		1324	thing, so beware.
816		1325
817	=back	1326	=back
818		1327
819	=head2 GENERIC WATCHER FUNCTIONS	1328	=head2 GENERIC WATCHER FUNCTIONS
820
821	In the following description, C<TYPE> stands for the watcher type,
822	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
823		1329
824	=over 4	1330	=over 4
825		1331
826	=item C<ev_init> (ev_TYPE *watcher, callback)	1332	=item C<ev_init> (ev_TYPE *watcher, callback)
827		1333
…		…
833	which rolls both calls into one.	1339	which rolls both calls into one.
834		1340
835	You can reinitialise a watcher at any time as long as it has been stopped	1341	You can reinitialise a watcher at any time as long as it has been stopped
836	(or never started) and there are no pending events outstanding.	1342	(or never started) and there are no pending events outstanding.
837		1343
838	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	1344	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
839	int revents)>.	1345	int revents)>.
840		1346
		1347	Example: Initialise an C<ev_io> watcher in two steps.
		1348
		1349	ev_io w;
		1350	ev_init (&w, my_cb);
		1351	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1352
841	=item C<ev_TYPE_set> (ev_TYPE *, [args])	1353	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
842		1354
843	This macro initialises the type-specific parts of a watcher. You need to	1355	This macro initialises the type-specific parts of a watcher. You need to
844	call C<ev_init> at least once before you call this macro, but you can	1356	call C<ev_init> at least once before you call this macro, but you can
845	call C<ev_TYPE_set> any number of times. You must not, however, call this	1357	call C<ev_TYPE_set> any number of times. You must not, however, call this
846	macro on a watcher that is active (it can be pending, however, which is a	1358	macro on a watcher that is active (it can be pending, however, which is a
847	difference to the C<ev_init> macro).	1359	difference to the C<ev_init> macro).
848		1360
849	Although some watcher types do not have type-specific arguments	1361	Although some watcher types do not have type-specific arguments
850	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	1362	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
851		1363
		1364	See C<ev_init>, above, for an example.
		1365
852	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	1366	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
853		1367
854	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	1368	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
855	calls into a single call. This is the most convinient method to initialise	1369	calls into a single call. This is the most convenient method to initialise
856	a watcher. The same limitations apply, of course.	1370	a watcher. The same limitations apply, of course.
857		1371
		1372	Example: Initialise and set an C<ev_io> watcher in one step.
		1373
		1374	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1375
858	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	1376	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
859		1377
860	Starts (activates) the given watcher. Only active watchers will receive	1378	Starts (activates) the given watcher. Only active watchers will receive
861	events. If the watcher is already active nothing will happen.	1379	events. If the watcher is already active nothing will happen.
862		1380
		1381	Example: Start the C<ev_io> watcher that is being abused as example in this
		1382	whole section.
		1383
		1384	ev_io_start (EV_DEFAULT_UC, &w);
		1385
863	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	1386	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
864		1387
865	Stops the given watcher again (if active) and clears the pending	1388	Stops the given watcher if active, and clears the pending status (whether
		1389	the watcher was active or not).
		1390
866	status. It is possible that stopped watchers are pending (for example,	1391	It is possible that stopped watchers are pending - for example,
867	non-repeating timers are being stopped when they become pending), but	1392	non-repeating timers are being stopped when they become pending - but
868	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1393	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
869	you want to free or reuse the memory used by the watcher it is therefore a	1394	pending. If you want to free or reuse the memory used by the watcher it is
870	good idea to always call its C<ev_TYPE_stop> function.	1395	therefore a good idea to always call its C<ev_TYPE_stop> function.
871		1396
872	=item bool ev_is_active (ev_TYPE *watcher)	1397	=item bool ev_is_active (ev_TYPE *watcher)
873		1398
874	Returns a true value iff the watcher is active (i.e. it has been started	1399	Returns a true value iff the watcher is active (i.e. it has been started
875	and not yet been stopped). As long as a watcher is active you must not modify	1400	and not yet been stopped). As long as a watcher is active you must not modify
876	it.	1401	it unless documented otherwise.
877		1402
878	=item bool ev_is_pending (ev_TYPE *watcher)	1403	=item bool ev_is_pending (ev_TYPE *watcher)
879		1404
880	Returns a true value iff the watcher is pending, (i.e. it has outstanding	1405	Returns a true value iff the watcher is pending, (i.e. it has outstanding
881	events but its callback has not yet been invoked). As long as a watcher	1406	events but its callback has not yet been invoked). As long as a watcher
…		…
886		1411
887	=item callback ev_cb (ev_TYPE *watcher)	1412	=item callback ev_cb (ev_TYPE *watcher)
888		1413
889	Returns the callback currently set on the watcher.	1414	Returns the callback currently set on the watcher.
890		1415
891	=item ev_cb_set (ev_TYPE *watcher, callback)	1416	=item ev_set_cb (ev_TYPE *watcher, callback)
892		1417
893	Change the callback. You can change the callback at virtually any time	1418	Change the callback. You can change the callback at virtually any time
894	(modulo threads).	1419	(modulo threads).
895		1420
896	=item ev_set_priority (ev_TYPE *watcher, priority)	1421	=item ev_set_priority (ev_TYPE *watcher, int priority)
897		1422
898	=item int ev_priority (ev_TYPE *watcher)	1423	=item int ev_priority (ev_TYPE *watcher)
899		1424
900	Set and query the priority of the watcher. The priority is a small	1425	Set and query the priority of the watcher. The priority is a small
901	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>	1426	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
902	(default: C<-2>). Pending watchers with higher priority will be invoked	1427	(default: C<-2>). Pending watchers with higher priority will be invoked
903	before watchers with lower priority, but priority will not keep watchers	1428	before watchers with lower priority, but priority will not keep watchers
904	from being executed (except for C<ev_idle> watchers).	1429	from being executed (except for C<ev_idle> watchers).
905		1430
906	This means that priorities are I<only> used for ordering callback
907	invocation after new events have been received. This is useful, for
908	example, to reduce latency after idling, or more often, to bind two
909	watchers on the same event and make sure one is called first.
910
911	If you need to suppress invocation when higher priority events are pending	1431	If you need to suppress invocation when higher priority events are pending
912	you need to look at C<ev_idle> watchers, which provide this functionality.	1432	you need to look at C<ev_idle> watchers, which provide this functionality.
913		1433
914	You I<must not> change the priority of a watcher as long as it is active or	1434	You I<must not> change the priority of a watcher as long as it is active or
915	pending.	1435	pending.
916		1436
		1437	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1438	fine, as long as you do not mind that the priority value you query might
		1439	or might not have been clamped to the valid range.
		1440
917	The default priority used by watchers when no priority has been set is	1441	The default priority used by watchers when no priority has been set is
918	always C<0>, which is supposed to not be too high and not be too low :).	1442	always C<0>, which is supposed to not be too high and not be too low :).
919		1443
920	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is	1444	See L</WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
921	fine, as long as you do not mind that the priority value you query might	1445	priorities.
922	or might not have been adjusted to be within valid range.
923		1446
924	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1447	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
925		1448
926	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1449	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
927	C<loop> nor C<revents> need to be valid as long as the watcher callback	1450	C<loop> nor C<revents> need to be valid as long as the watcher callback
928	can deal with that fact.	1451	can deal with that fact, as both are simply passed through to the
		1452	callback.
929		1453
930	=item int ev_clear_pending (loop, ev_TYPE *watcher)	1454	=item int ev_clear_pending (loop, ev_TYPE *watcher)
931		1455
932	If the watcher is pending, this function returns clears its pending status	1456	If the watcher is pending, this function clears its pending status and
933	and returns its C<revents> bitset (as if its callback was invoked). If the	1457	returns its C<revents> bitset (as if its callback was invoked). If the
934	watcher isn't pending it does nothing and returns C<0>.	1458	watcher isn't pending it does nothing and returns C<0>.
935		1459
		1460	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1461	callback to be invoked, which can be accomplished with this function.
		1462
		1463	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1464
		1465	Feeds the given event set into the event loop, as if the specified event
		1466	had happened for the specified watcher (which must be a pointer to an
		1467	initialised but not necessarily started event watcher). Obviously you must
		1468	not free the watcher as long as it has pending events.
		1469
		1470	Stopping the watcher, letting libev invoke it, or calling
		1471	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1472	not started in the first place.
		1473
		1474	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1475	functions that do not need a watcher.
		1476
936	=back	1477	=back
937		1478
		1479	See also the L</ASSOCIATING CUSTOM DATA WITH A WATCHER> and L</BUILDING YOUR
		1480	OWN COMPOSITE WATCHERS> idioms.
938		1481
939	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1482	=head2 WATCHER STATES
940		1483
941	Each watcher has, by default, a member C<void *data> that you can change	1484	There are various watcher states mentioned throughout this manual -
942	and read at any time, libev will completely ignore it. This can be used	1485	active, pending and so on. In this section these states and the rules to
943	to associate arbitrary data with your watcher. If you need more data and	1486	transition between them will be described in more detail - and while these
944	don't want to allocate memory and store a pointer to it in that data	1487	rules might look complicated, they usually do "the right thing".
945	member, you can also "subclass" the watcher type and provide your own
946	data:
947		1488
948	struct my_io	1489	=over 4
949	{
950	struct ev_io io;
951	int otherfd;
952	void *somedata;
953	struct whatever *mostinteresting;
954	}
955		1490
956	And since your callback will be called with a pointer to the watcher, you	1491	=item initialised
957	can cast it back to your own type:
958		1492
959	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1493	Before a watcher can be registered with the event loop it has to be
960	{	1494	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
961	struct my_io *w = (struct my_io *)w_;	1495	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
962	...
963	}
964		1496
965	More interesting and less C-conformant ways of casting your callback type	1497	In this state it is simply some block of memory that is suitable for
966	instead have been omitted.	1498	use in an event loop. It can be moved around, freed, reused etc. at
		1499	will - as long as you either keep the memory contents intact, or call
		1500	C<ev_TYPE_init> again.
967		1501
968	Another common scenario is having some data structure with multiple	1502	=item started/running/active
969	watchers:
970		1503
971	struct my_biggy	1504	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
972	{	1505	property of the event loop, and is actively waiting for events. While in
973	int some_data;	1506	this state it cannot be accessed (except in a few documented ways), moved,
974	ev_timer t1;	1507	freed or anything else - the only legal thing is to keep a pointer to it,
975	ev_timer t2;	1508	and call libev functions on it that are documented to work on active watchers.
976	}
977		1509
978	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1510	=item pending
979	you need to use C<offsetof>:
980		1511
981	#include <stddef.h>	1512	If a watcher is active and libev determines that an event it is interested
		1513	in has occurred (such as a timer expiring), it will become pending. It will
		1514	stay in this pending state until either it is stopped or its callback is
		1515	about to be invoked, so it is not normally pending inside the watcher
		1516	callback.
982		1517
		1518	The watcher might or might not be active while it is pending (for example,
		1519	an expired non-repeating timer can be pending but no longer active). If it
		1520	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1521	but it is still property of the event loop at this time, so cannot be
		1522	moved, freed or reused. And if it is active the rules described in the
		1523	previous item still apply.
		1524
		1525	It is also possible to feed an event on a watcher that is not active (e.g.
		1526	via C<ev_feed_event>), in which case it becomes pending without being
		1527	active.
		1528
		1529	=item stopped
		1530
		1531	A watcher can be stopped implicitly by libev (in which case it might still
		1532	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1533	latter will clear any pending state the watcher might be in, regardless
		1534	of whether it was active or not, so stopping a watcher explicitly before
		1535	freeing it is often a good idea.
		1536
		1537	While stopped (and not pending) the watcher is essentially in the
		1538	initialised state, that is, it can be reused, moved, modified in any way
		1539	you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
		1540	it again).
		1541
		1542	=back
		1543
		1544	=head2 WATCHER PRIORITY MODELS
		1545
		1546	Many event loops support I<watcher priorities>, which are usually small
		1547	integers that influence the ordering of event callback invocation
		1548	between watchers in some way, all else being equal.
		1549
		1550	In libev, watcher priorities can be set using C<ev_set_priority>. See its
		1551	description for the more technical details such as the actual priority
		1552	range.
		1553
		1554	There are two common ways how these these priorities are being interpreted
		1555	by event loops:
		1556
		1557	In the more common lock-out model, higher priorities "lock out" invocation
		1558	of lower priority watchers, which means as long as higher priority
		1559	watchers receive events, lower priority watchers are not being invoked.
		1560
		1561	The less common only-for-ordering model uses priorities solely to order
		1562	callback invocation within a single event loop iteration: Higher priority
		1563	watchers are invoked before lower priority ones, but they all get invoked
		1564	before polling for new events.
		1565
		1566	Libev uses the second (only-for-ordering) model for all its watchers
		1567	except for idle watchers (which use the lock-out model).
		1568
		1569	The rationale behind this is that implementing the lock-out model for
		1570	watchers is not well supported by most kernel interfaces, and most event
		1571	libraries will just poll for the same events again and again as long as
		1572	their callbacks have not been executed, which is very inefficient in the
		1573	common case of one high-priority watcher locking out a mass of lower
		1574	priority ones.
		1575
		1576	Static (ordering) priorities are most useful when you have two or more
		1577	watchers handling the same resource: a typical usage example is having an
		1578	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1579	timeouts. Under load, data might be received while the program handles
		1580	other jobs, but since timers normally get invoked first, the timeout
		1581	handler will be executed before checking for data. In that case, giving
		1582	the timer a lower priority than the I/O watcher ensures that I/O will be
		1583	handled first even under adverse conditions (which is usually, but not
		1584	always, what you want).
		1585
		1586	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1587	will only be executed when no same or higher priority watchers have
		1588	received events, they can be used to implement the "lock-out" model when
		1589	required.
		1590
		1591	For example, to emulate how many other event libraries handle priorities,
		1592	you can associate an C<ev_idle> watcher to each such watcher, and in
		1593	the normal watcher callback, you just start the idle watcher. The real
		1594	processing is done in the idle watcher callback. This causes libev to
		1595	continuously poll and process kernel event data for the watcher, but when
		1596	the lock-out case is known to be rare (which in turn is rare :), this is
		1597	workable.
		1598
		1599	Usually, however, the lock-out model implemented that way will perform
		1600	miserably under the type of load it was designed to handle. In that case,
		1601	it might be preferable to stop the real watcher before starting the
		1602	idle watcher, so the kernel will not have to process the event in case
		1603	the actual processing will be delayed for considerable time.
		1604
		1605	Here is an example of an I/O watcher that should run at a strictly lower
		1606	priority than the default, and which should only process data when no
		1607	other events are pending:
		1608
		1609	ev_idle idle; // actual processing watcher
		1610	ev_io io; // actual event watcher
		1611
983	static void	1612	static void
984	t1_cb (EV_P_ struct ev_timer *w, int revents)	1613	io_cb (EV_P_ ev_io *w, int revents)
985	{	1614	{
986	struct my_biggy big = (struct my_biggy *	1615	// stop the I/O watcher, we received the event, but
987	(((char *)w) - offsetof (struct my_biggy, t1));	1616	// are not yet ready to handle it.
		1617	ev_io_stop (EV_A_ w);
		1618
		1619	// start the idle watcher to handle the actual event.
		1620	// it will not be executed as long as other watchers
		1621	// with the default priority are receiving events.
		1622	ev_idle_start (EV_A_ &idle);
988	}	1623	}
989		1624
990	static void	1625	static void
991	t2_cb (EV_P_ struct ev_timer *w, int revents)	1626	idle_cb (EV_P_ ev_idle *w, int revents)
992	{	1627	{
993	struct my_biggy big = (struct my_biggy *	1628	// actual processing
994	(((char *)w) - offsetof (struct my_biggy, t2));	1629	read (STDIN_FILENO, ...);
		1630
		1631	// have to start the I/O watcher again, as
		1632	// we have handled the event
		1633	ev_io_start (EV_P_ &io);
995	}	1634	}
		1635
		1636	// initialisation
		1637	ev_idle_init (&idle, idle_cb);
		1638	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1639	ev_io_start (EV_DEFAULT_ &io);
		1640
		1641	In the "real" world, it might also be beneficial to start a timer, so that
		1642	low-priority connections can not be locked out forever under load. This
		1643	enables your program to keep a lower latency for important connections
		1644	during short periods of high load, while not completely locking out less
		1645	important ones.
996		1646
997		1647
998	=head1 WATCHER TYPES	1648	=head1 WATCHER TYPES
999		1649
1000	This section describes each watcher in detail, but will not repeat	1650	This section describes each watcher in detail, but will not repeat
1001	information given in the last section. Any initialisation/set macros,	1651	information given in the last section. Any initialisation/set macros,
1002	functions and members specific to the watcher type are explained.	1652	functions and members specific to the watcher type are explained.
1003		1653
1004	Members are additionally marked with either I<[read-only]>, meaning that,	1654	Most members are additionally marked with either I<[read-only]>, meaning
1005	while the watcher is active, you can look at the member and expect some	1655	that, while the watcher is active, you can look at the member and expect
1006	sensible content, but you must not modify it (you can modify it while the	1656	some sensible content, but you must not modify it (you can modify it while
1007	watcher is stopped to your hearts content), or I<[read-write]>, which	1657	the watcher is stopped to your hearts content), or I<[read-write]>, which
1008	means you can expect it to have some sensible content while the watcher	1658	means you can expect it to have some sensible content while the watcher is
1009	is active, but you can also modify it. Modifying it may not do something	1659	active, but you can also modify it (within the same thread as the event
		1660	loop, i.e. without creating data races). Modifying it may not do something
1010	sensible or take immediate effect (or do anything at all), but libev will	1661	sensible or take immediate effect (or do anything at all), but libev will
1011	not crash or malfunction in any way.	1662	not crash or malfunction in any way.
1012		1663
		1664	In any case, the documentation for each member will explain what the
		1665	effects are, and if there are any additional access restrictions.
1013		1666
1014	=head2 C<ev_io> - is this file descriptor readable or writable?	1667	=head2 C<ev_io> - is this file descriptor readable or writable?
1015		1668
1016	I/O watchers check whether a file descriptor is readable or writable	1669	I/O watchers check whether a file descriptor is readable or writable
1017	in each iteration of the event loop, or, more precisely, when reading	1670	in each iteration of the event loop, or, more precisely, when reading
…		…
1024	In general you can register as many read and/or write event watchers per	1677	In general you can register as many read and/or write event watchers per
1025	fd as you want (as long as you don't confuse yourself). Setting all file	1678	fd as you want (as long as you don't confuse yourself). Setting all file
1026	descriptors to non-blocking mode is also usually a good idea (but not	1679	descriptors to non-blocking mode is also usually a good idea (but not
1027	required if you know what you are doing).	1680	required if you know what you are doing).
1028		1681
1029	If you must do this, then force the use of a known-to-be-good backend
1030	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
1031	C<EVBACKEND_POLL>).
1032
1033	Another thing you have to watch out for is that it is quite easy to	1682	Another thing you have to watch out for is that it is quite easy to
1034	receive "spurious" readyness notifications, that is your callback might	1683	receive "spurious" readiness notifications, that is, your callback might
1035	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1684	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1036	because there is no data. Not only are some backends known to create a	1685	because there is no data. It is very easy to get into this situation even
1037	lot of those (for example solaris ports), it is very easy to get into	1686	with a relatively standard program structure. Thus it is best to always
1038	this situation even with a relatively standard program structure. Thus	1687	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
1039	it is best to always use non-blocking I/O: An extra C<read>(2) returning
1040	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1688	preferable to a program hanging until some data arrives.
1041		1689
1042	If you cannot run the fd in non-blocking mode (for example you should not	1690	If you cannot run the fd in non-blocking mode (for example you should
1043	play around with an Xlib connection), then you have to seperately re-test	1691	not play around with an Xlib connection), then you have to separately
1044	whether a file descriptor is really ready with a known-to-be good interface	1692	re-test whether a file descriptor is really ready with a known-to-be good
1045	such as poll (fortunately in our Xlib example, Xlib already does this on	1693	interface such as poll (fortunately in the case of Xlib, it already does
1046	its own, so its quite safe to use).	1694	this on its own, so its quite safe to use). Some people additionally
		1695	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1696	indefinitely.
		1697
		1698	But really, best use non-blocking mode.
1047		1699
1048	=head3 The special problem of disappearing file descriptors	1700	=head3 The special problem of disappearing file descriptors
1049		1701
1050	Some backends (e.g. kqueue, epoll) need to be told about closing a file	1702	Some backends (e.g. kqueue, epoll, linuxaio) need to be told about closing
1051	descriptor (either by calling C<close> explicitly or by any other means,	1703	a file descriptor (either due to calling C<close> explicitly or any other
1052	such as C<dup>). The reason is that you register interest in some file	1704	means, such as C<dup2>). The reason is that you register interest in some
1053	descriptor, but when it goes away, the operating system will silently drop	1705	file descriptor, but when it goes away, the operating system will silently
1054	this interest. If another file descriptor with the same number then is	1706	drop this interest. If another file descriptor with the same number then
1055	registered with libev, there is no efficient way to see that this is, in	1707	is registered with libev, there is no efficient way to see that this is,
1056	fact, a different file descriptor.	1708	in fact, a different file descriptor.
1057		1709
1058	To avoid having to explicitly tell libev about such cases, libev follows	1710	To avoid having to explicitly tell libev about such cases, libev follows
1059	the following policy: Each time C<ev_io_set> is being called, libev	1711	the following policy: Each time C<ev_io_set> is being called, libev
1060	will assume that this is potentially a new file descriptor, otherwise	1712	will assume that this is potentially a new file descriptor, otherwise
1061	it is assumed that the file descriptor stays the same. That means that	1713	it is assumed that the file descriptor stays the same. That means that
…		…
1075		1727
1076	There is no workaround possible except not registering events	1728	There is no workaround possible except not registering events
1077	for potentially C<dup ()>'ed file descriptors, or to resort to	1729	for potentially C<dup ()>'ed file descriptors, or to resort to
1078	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.	1730	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1079		1731
		1732	=head3 The special problem of files
		1733
		1734	Many people try to use C<select> (or libev) on file descriptors
		1735	representing files, and expect it to become ready when their program
		1736	doesn't block on disk accesses (which can take a long time on their own).
		1737
		1738	However, this cannot ever work in the "expected" way - you get a readiness
		1739	notification as soon as the kernel knows whether and how much data is
		1740	there, and in the case of open files, that's always the case, so you
		1741	always get a readiness notification instantly, and your read (or possibly
		1742	write) will still block on the disk I/O.
		1743
		1744	Another way to view it is that in the case of sockets, pipes, character
		1745	devices and so on, there is another party (the sender) that delivers data
		1746	on its own, but in the case of files, there is no such thing: the disk
		1747	will not send data on its own, simply because it doesn't know what you
		1748	wish to read - you would first have to request some data.
		1749
		1750	Since files are typically not-so-well supported by advanced notification
		1751	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1752	to files, even though you should not use it. The reason for this is
		1753	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1754	usually a tty, often a pipe, but also sometimes files or special devices
		1755	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1756	F</dev/urandom>), and even though the file might better be served with
		1757	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1758	it "just works" instead of freezing.
		1759
		1760	So avoid file descriptors pointing to files when you know it (e.g. use
		1761	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1762	when you rarely read from a file instead of from a socket, and want to
		1763	reuse the same code path.
		1764
1080	=head3 The special problem of fork	1765	=head3 The special problem of fork
1081		1766
1082	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit	1767	Some backends (epoll, kqueue, linuxaio, iouring) do not support C<fork ()>
1083	useless behaviour. Libev fully supports fork, but needs to be told about	1768	at all or exhibit useless behaviour. Libev fully supports fork, but needs
1084	it in the child.	1769	to be told about it in the child if you want to continue to use it in the
		1770	child.
1085		1771
1086	To support fork in your programs, you either have to call	1772	To support fork in your child processes, you have to call C<ev_loop_fork
1087	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,	1773	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
1088	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or	1774	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1089	C<EVBACKEND_POLL>.
1090		1775
1091	=head3 The special problem of SIGPIPE	1776	=head3 The special problem of SIGPIPE
1092		1777
1093	While not really specific to libev, it is easy to forget about SIGPIPE:	1778	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1094	when reading from a pipe whose other end has been closed, your program	1779	when writing to a pipe whose other end has been closed, your program gets
1095	gets send a SIGPIPE, which, by default, aborts your program. For most	1780	sent a SIGPIPE, which, by default, aborts your program. For most programs
1096	programs this is sensible behaviour, for daemons, this is usually	1781	this is sensible behaviour, for daemons, this is usually undesirable.
1097	undesirable.
1098		1782
1099	So when you encounter spurious, unexplained daemon exits, make sure you	1783	So when you encounter spurious, unexplained daemon exits, make sure you
1100	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon	1784	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
1101	somewhere, as that would have given you a big clue).	1785	somewhere, as that would have given you a big clue).
1102		1786
		1787	=head3 The special problem of accept()ing when you can't
		1788
		1789	Many implementations of the POSIX C<accept> function (for example,
		1790	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1791	connection from the pending queue in all error cases.
		1792
		1793	For example, larger servers often run out of file descriptors (because
		1794	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1795	rejecting the connection, leading to libev signalling readiness on
		1796	the next iteration again (the connection still exists after all), and
		1797	typically causing the program to loop at 100% CPU usage.
		1798
		1799	Unfortunately, the set of errors that cause this issue differs between
		1800	operating systems, there is usually little the app can do to remedy the
		1801	situation, and no known thread-safe method of removing the connection to
		1802	cope with overload is known (to me).
		1803
		1804	One of the easiest ways to handle this situation is to just ignore it
		1805	- when the program encounters an overload, it will just loop until the
		1806	situation is over. While this is a form of busy waiting, no OS offers an
		1807	event-based way to handle this situation, so it's the best one can do.
		1808
		1809	A better way to handle the situation is to log any errors other than
		1810	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1811	messages, and continue as usual, which at least gives the user an idea of
		1812	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1813	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1814	usage.
		1815
		1816	If your program is single-threaded, then you could also keep a dummy file
		1817	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1818	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1819	close that fd, and create a new dummy fd. This will gracefully refuse
		1820	clients under typical overload conditions.
		1821
		1822	The last way to handle it is to simply log the error and C<exit>, as
		1823	is often done with C<malloc> failures, but this results in an easy
		1824	opportunity for a DoS attack.
1103		1825
1104	=head3 Watcher-Specific Functions	1826	=head3 Watcher-Specific Functions
1105		1827
1106	=over 4	1828	=over 4
1107		1829
1108	=item ev_io_init (ev_io *, callback, int fd, int events)	1830	=item ev_io_init (ev_io *, callback, int fd, int events)
1109		1831
1110	=item ev_io_set (ev_io *, int fd, int events)	1832	=item ev_io_set (ev_io *, int fd, int events)
1111		1833
1112	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1834	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
1113	rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or	1835	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE>, both
1114	C<EV_READ | EV_WRITE> to receive the given events.	1836	C<EV_READ | EV_WRITE> or C<0>, to express the desire to receive the given
		1837	events.
1115		1838
1116	=item int fd [read-only]	1839	Note that setting the C<events> to C<0> and starting the watcher is
		1840	supported, but not specially optimized - if your program sometimes happens
		1841	to generate this combination this is fine, but if it is easy to avoid
		1842	starting an io watcher watching for no events you should do so.
1117		1843
1118	The file descriptor being watched.	1844	=item ev_io_modify (ev_io *, int events)
1119		1845
		1846	Similar to C<ev_io_set>, but only changes the requested events. Using this
		1847	might be faster with some backends, as libev can assume that the C<fd>
		1848	still refers to the same underlying file description, something it cannot
		1849	do when using C<ev_io_set>.
		1850
		1851	=item int fd [no-modify]
		1852
		1853	The file descriptor being watched. While it can be read at any time, you
		1854	must not modify this member even when the watcher is stopped - always use
		1855	C<ev_io_set> for that.
		1856
1120	=item int events [read-only]	1857	=item int events [no-modify]
1121		1858
1122	The events being watched.	1859	The set of events the fd is being watched for, among other flags. Remember
		1860	that this is a bit set - to test for C<EV_READ>, use C<< w->events &
		1861	EV_READ >>, and similarly for C<EV_WRITE>.
		1862
		1863	As with C<fd>, you must not modify this member even when the watcher is
		1864	stopped, always use C<ev_io_set> or C<ev_io_modify> for that.
1123		1865
1124	=back	1866	=back
1125		1867
1126	=head3 Examples	1868	=head3 Examples
1127		1869
1128	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1870	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1129	readable, but only once. Since it is likely line-buffered, you could	1871	readable, but only once. Since it is likely line-buffered, you could
1130	attempt to read a whole line in the callback.	1872	attempt to read a whole line in the callback.
1131		1873
1132	static void	1874	static void
1133	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1875	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
1134	{	1876	{
1135	ev_io_stop (loop, w);	1877	ev_io_stop (loop, w);
1136	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1878	.. read from stdin here (or from w->fd) and handle any I/O errors
1137	}	1879	}
1138		1880
1139	...	1881	...
1140	struct ev_loop *loop = ev_default_init (0);	1882	struct ev_loop *loop = ev_default_init (0);
1141	struct ev_io stdin_readable;	1883	ev_io stdin_readable;
1142	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1884	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1143	ev_io_start (loop, &stdin_readable);	1885	ev_io_start (loop, &stdin_readable);
1144	ev_loop (loop, 0);	1886	ev_run (loop, 0);
1145		1887
1146		1888
1147	=head2 C<ev_timer> - relative and optionally repeating timeouts	1889	=head2 C<ev_timer> - relative and optionally repeating timeouts
1148		1890
1149	Timer watchers are simple relative timers that generate an event after a	1891	Timer watchers are simple relative timers that generate an event after a
1150	given time, and optionally repeating in regular intervals after that.	1892	given time, and optionally repeating in regular intervals after that.
1151		1893
1152	The timers are based on real time, that is, if you register an event that	1894	The timers are based on real time, that is, if you register an event that
1153	times out after an hour and you reset your system clock to last years	1895	times out after an hour and you reset your system clock to January last
1154	time, it will still time out after (roughly) and hour. "Roughly" because	1896	year, it will still time out after (roughly) one hour. "Roughly" because
1155	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1897	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1156	monotonic clock option helps a lot here).	1898	monotonic clock option helps a lot here).
		1899
		1900	The callback is guaranteed to be invoked only I<after> its timeout has
		1901	passed (not I<at>, so on systems with very low-resolution clocks this
		1902	might introduce a small delay, see "the special problem of being too
		1903	early", below). If multiple timers become ready during the same loop
		1904	iteration then the ones with earlier time-out values are invoked before
		1905	ones of the same priority with later time-out values (but this is no
		1906	longer true when a callback calls C<ev_run> recursively).
		1907
		1908	=head3 Be smart about timeouts
		1909
		1910	Many real-world problems involve some kind of timeout, usually for error
		1911	recovery. A typical example is an HTTP request - if the other side hangs,
		1912	you want to raise some error after a while.
		1913
		1914	What follows are some ways to handle this problem, from obvious and
		1915	inefficient to smart and efficient.
		1916
		1917	In the following, a 60 second activity timeout is assumed - a timeout that
		1918	gets reset to 60 seconds each time there is activity (e.g. each time some
		1919	data or other life sign was received).
		1920
		1921	=over 4
		1922
		1923	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1924
		1925	This is the most obvious, but not the most simple way: In the beginning,
		1926	start the watcher:
		1927
		1928	ev_timer_init (timer, callback, 60., 0.);
		1929	ev_timer_start (loop, timer);
		1930
		1931	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1932	and start it again:
		1933
		1934	ev_timer_stop (loop, timer);
		1935	ev_timer_set (timer, 60., 0.);
		1936	ev_timer_start (loop, timer);
		1937
		1938	This is relatively simple to implement, but means that each time there is
		1939	some activity, libev will first have to remove the timer from its internal
		1940	data structure and then add it again. Libev tries to be fast, but it's
		1941	still not a constant-time operation.
		1942
		1943	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1944
		1945	This is the easiest way, and involves using C<ev_timer_again> instead of
		1946	C<ev_timer_start>.
		1947
		1948	To implement this, configure an C<ev_timer> with a C<repeat> value
		1949	of C<60> and then call C<ev_timer_again> at start and each time you
		1950	successfully read or write some data. If you go into an idle state where
		1951	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1952	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1953
		1954	That means you can ignore both the C<ev_timer_start> function and the
		1955	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1956	member and C<ev_timer_again>.
		1957
		1958	At start:
		1959
		1960	ev_init (timer, callback);
		1961	timer->repeat = 60.;
		1962	ev_timer_again (loop, timer);
		1963
		1964	Each time there is some activity:
		1965
		1966	ev_timer_again (loop, timer);
		1967
		1968	It is even possible to change the time-out on the fly, regardless of
		1969	whether the watcher is active or not:
		1970
		1971	timer->repeat = 30.;
		1972	ev_timer_again (loop, timer);
		1973
		1974	This is slightly more efficient then stopping/starting the timer each time
		1975	you want to modify its timeout value, as libev does not have to completely
		1976	remove and re-insert the timer from/into its internal data structure.
		1977
		1978	It is, however, even simpler than the "obvious" way to do it.
		1979
		1980	=item 3. Let the timer time out, but then re-arm it as required.
		1981
		1982	This method is more tricky, but usually most efficient: Most timeouts are
		1983	relatively long compared to the intervals between other activity - in
		1984	our example, within 60 seconds, there are usually many I/O events with
		1985	associated activity resets.
		1986
		1987	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1988	but remember the time of last activity, and check for a real timeout only
		1989	within the callback:
		1990
		1991	ev_tstamp timeout = 60.;
		1992	ev_tstamp last_activity; // time of last activity
		1993	ev_timer timer;
		1994
		1995	static void
		1996	callback (EV_P_ ev_timer *w, int revents)
		1997	{
		1998	// calculate when the timeout would happen
		1999	ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
		2000
		2001	// if negative, it means we the timeout already occurred
		2002	if (after < 0.)
		2003	{
		2004	// timeout occurred, take action
		2005	}
		2006	else
		2007	{
		2008	// callback was invoked, but there was some recent
		2009	// activity. simply restart the timer to time out
		2010	// after "after" seconds, which is the earliest time
		2011	// the timeout can occur.
		2012	ev_timer_set (w, after, 0.);
		2013	ev_timer_start (EV_A_ w);
		2014	}
		2015	}
		2016
		2017	To summarise the callback: first calculate in how many seconds the
		2018	timeout will occur (by calculating the absolute time when it would occur,
		2019	C<last_activity + timeout>, and subtracting the current time, C<ev_now
		2020	(EV_A)> from that).
		2021
		2022	If this value is negative, then we are already past the timeout, i.e. we
		2023	timed out, and need to do whatever is needed in this case.
		2024
		2025	Otherwise, we now the earliest time at which the timeout would trigger,
		2026	and simply start the timer with this timeout value.
		2027
		2028	In other words, each time the callback is invoked it will check whether
		2029	the timeout occurred. If not, it will simply reschedule itself to check
		2030	again at the earliest time it could time out. Rinse. Repeat.
		2031
		2032	This scheme causes more callback invocations (about one every 60 seconds
		2033	minus half the average time between activity), but virtually no calls to
		2034	libev to change the timeout.
		2035
		2036	To start the machinery, simply initialise the watcher and set
		2037	C<last_activity> to the current time (meaning there was some activity just
		2038	now), then call the callback, which will "do the right thing" and start
		2039	the timer:
		2040
		2041	last_activity = ev_now (EV_A);
		2042	ev_init (&timer, callback);
		2043	callback (EV_A_ &timer, 0);
		2044
		2045	When there is some activity, simply store the current time in
		2046	C<last_activity>, no libev calls at all:
		2047
		2048	if (activity detected)
		2049	last_activity = ev_now (EV_A);
		2050
		2051	When your timeout value changes, then the timeout can be changed by simply
		2052	providing a new value, stopping the timer and calling the callback, which
		2053	will again do the right thing (for example, time out immediately :).
		2054
		2055	timeout = new_value;
		2056	ev_timer_stop (EV_A_ &timer);
		2057	callback (EV_A_ &timer, 0);
		2058
		2059	This technique is slightly more complex, but in most cases where the
		2060	time-out is unlikely to be triggered, much more efficient.
		2061
		2062	=item 4. Wee, just use a double-linked list for your timeouts.
		2063
		2064	If there is not one request, but many thousands (millions...), all
		2065	employing some kind of timeout with the same timeout value, then one can
		2066	do even better:
		2067
		2068	When starting the timeout, calculate the timeout value and put the timeout
		2069	at the I<end> of the list.
		2070
		2071	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		2072	the list is expected to fire (for example, using the technique #3).
		2073
		2074	When there is some activity, remove the timer from the list, recalculate
		2075	the timeout, append it to the end of the list again, and make sure to
		2076	update the C<ev_timer> if it was taken from the beginning of the list.
		2077
		2078	This way, one can manage an unlimited number of timeouts in O(1) time for
		2079	starting, stopping and updating the timers, at the expense of a major
		2080	complication, and having to use a constant timeout. The constant timeout
		2081	ensures that the list stays sorted.
		2082
		2083	=back
		2084
		2085	So which method the best?
		2086
		2087	Method #2 is a simple no-brain-required solution that is adequate in most
		2088	situations. Method #3 requires a bit more thinking, but handles many cases
		2089	better, and isn't very complicated either. In most case, choosing either
		2090	one is fine, with #3 being better in typical situations.
		2091
		2092	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		2093	rather complicated, but extremely efficient, something that really pays
		2094	off after the first million or so of active timers, i.e. it's usually
		2095	overkill :)
		2096
		2097	=head3 The special problem of being too early
		2098
		2099	If you ask a timer to call your callback after three seconds, then
		2100	you expect it to be invoked after three seconds - but of course, this
		2101	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		2102	guaranteed to any precision by libev - imagine somebody suspending the
		2103	process with a STOP signal for a few hours for example.
		2104
		2105	So, libev tries to invoke your callback as soon as possible I<after> the
		2106	delay has occurred, but cannot guarantee this.
		2107
		2108	A less obvious failure mode is calling your callback too early: many event
		2109	loops compare timestamps with a "elapsed delay >= requested delay", but
		2110	this can cause your callback to be invoked much earlier than you would
		2111	expect.
		2112
		2113	To see why, imagine a system with a clock that only offers full second
		2114	resolution (think windows if you can't come up with a broken enough OS
		2115	yourself). If you schedule a one-second timer at the time 500.9, then the
		2116	event loop will schedule your timeout to elapse at a system time of 500
		2117	(500.9 truncated to the resolution) + 1, or 501.
		2118
		2119	If an event library looks at the timeout 0.1s later, it will see "501 >=
		2120	501" and invoke the callback 0.1s after it was started, even though a
		2121	one-second delay was requested - this is being "too early", despite best
		2122	intentions.
		2123
		2124	This is the reason why libev will never invoke the callback if the elapsed
		2125	delay equals the requested delay, but only when the elapsed delay is
		2126	larger than the requested delay. In the example above, libev would only invoke
		2127	the callback at system time 502, or 1.1s after the timer was started.
		2128
		2129	So, while libev cannot guarantee that your callback will be invoked
		2130	exactly when requested, it I<can> and I<does> guarantee that the requested
		2131	delay has actually elapsed, or in other words, it always errs on the "too
		2132	late" side of things.
		2133
		2134	=head3 The special problem of time updates
		2135
		2136	Establishing the current time is a costly operation (it usually takes
		2137	at least one system call): EV therefore updates its idea of the current
		2138	time only before and after C<ev_run> collects new events, which causes a
		2139	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		2140	lots of events in one iteration.
1157		2141
1158	The relative timeouts are calculated relative to the C<ev_now ()>	2142	The relative timeouts are calculated relative to the C<ev_now ()>
1159	time. This is usually the right thing as this timestamp refers to the time	2143	time. This is usually the right thing as this timestamp refers to the time
1160	of the event triggering whatever timeout you are modifying/starting. If	2144	of the event triggering whatever timeout you are modifying/starting. If
1161	you suspect event processing to be delayed and you I<need> to base the timeout	2145	you suspect event processing to be delayed and you I<need> to base the
1162	on the current time, use something like this to adjust for this:	2146	timeout on the current time, use something like the following to adjust
		2147	for it:
1163		2148
1164	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2149	ev_timer_set (&timer, after + (ev_time () - ev_now ()), 0.);
1165		2150
1166	The callback is guarenteed to be invoked only when its timeout has passed,	2151	If the event loop is suspended for a long time, you can also force an
1167	but if multiple timers become ready during the same loop iteration then	2152	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1168	order of execution is undefined.	2153	()>, although that will push the event time of all outstanding events
		2154	further into the future.
		2155
		2156	=head3 The special problem of unsynchronised clocks
		2157
		2158	Modern systems have a variety of clocks - libev itself uses the normal
		2159	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2160	jumps).
		2161
		2162	Neither of these clocks is synchronised with each other or any other clock
		2163	on the system, so C<ev_time ()> might return a considerably different time
		2164	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2165	a call to C<gettimeofday> might return a second count that is one higher
		2166	than a directly following call to C<time>.
		2167
		2168	The moral of this is to only compare libev-related timestamps with
		2169	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2170	a second or so.
		2171
		2172	One more problem arises due to this lack of synchronisation: if libev uses
		2173	the system monotonic clock and you compare timestamps from C<ev_time>
		2174	or C<ev_now> from when you started your timer and when your callback is
		2175	invoked, you will find that sometimes the callback is a bit "early".
		2176
		2177	This is because C<ev_timer>s work in real time, not wall clock time, so
		2178	libev makes sure your callback is not invoked before the delay happened,
		2179	I<measured according to the real time>, not the system clock.
		2180
		2181	If your timeouts are based on a physical timescale (e.g. "time out this
		2182	connection after 100 seconds") then this shouldn't bother you as it is
		2183	exactly the right behaviour.
		2184
		2185	If you want to compare wall clock/system timestamps to your timers, then
		2186	you need to use C<ev_periodic>s, as these are based on the wall clock
		2187	time, where your comparisons will always generate correct results.
		2188
		2189	=head3 The special problems of suspended animation
		2190
		2191	When you leave the server world it is quite customary to hit machines that
		2192	can suspend/hibernate - what happens to the clocks during such a suspend?
		2193
		2194	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		2195	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		2196	to run until the system is suspended, but they will not advance while the
		2197	system is suspended. That means, on resume, it will be as if the program
		2198	was frozen for a few seconds, but the suspend time will not be counted
		2199	towards C<ev_timer> when a monotonic clock source is used. The real time
		2200	clock advanced as expected, but if it is used as sole clocksource, then a
		2201	long suspend would be detected as a time jump by libev, and timers would
		2202	be adjusted accordingly.
		2203
		2204	I would not be surprised to see different behaviour in different between
		2205	operating systems, OS versions or even different hardware.
		2206
		2207	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2208	time jump in the monotonic clocks and the realtime clock. If the program
		2209	is suspended for a very long time, and monotonic clock sources are in use,
		2210	then you can expect C<ev_timer>s to expire as the full suspension time
		2211	will be counted towards the timers. When no monotonic clock source is in
		2212	use, then libev will again assume a timejump and adjust accordingly.
		2213
		2214	It might be beneficial for this latter case to call C<ev_suspend>
		2215	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2216	deterministic behaviour in this case (you can do nothing against
		2217	C<SIGSTOP>).
1169		2218
1170	=head3 Watcher-Specific Functions and Data Members	2219	=head3 Watcher-Specific Functions and Data Members
1171		2220
1172	=over 4	2221	=over 4
1173		2222
1174	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2223	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
1175		2224
1176	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2225	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
1177		2226
1178	Configure the timer to trigger after C<after> seconds. If C<repeat> is	2227	Configure the timer to trigger after C<after> seconds (fractional and
1179	C<0.>, then it will automatically be stopped. If it is positive, then the	2228	negative values are supported). If C<repeat> is C<0.>, then it will
		2229	automatically be stopped once the timeout is reached. If it is positive,
1180	timer will automatically be configured to trigger again C<repeat> seconds	2230	then the timer will automatically be configured to trigger again C<repeat>
1181	later, again, and again, until stopped manually.	2231	seconds later, again, and again, until stopped manually.
1182		2232
1183	The timer itself will do a best-effort at avoiding drift, that is, if you	2233	The timer itself will do a best-effort at avoiding drift, that is, if
1184	configure a timer to trigger every 10 seconds, then it will trigger at	2234	you configure a timer to trigger every 10 seconds, then it will normally
1185	exactly 10 second intervals. If, however, your program cannot keep up with	2235	trigger at exactly 10 second intervals. If, however, your program cannot
1186	the timer (because it takes longer than those 10 seconds to do stuff) the	2236	keep up with the timer (because it takes longer than those 10 seconds to
1187	timer will not fire more than once per event loop iteration.	2237	do stuff) the timer will not fire more than once per event loop iteration.
1188		2238
1189	=item ev_timer_again (loop, ev_timer *)	2239	=item ev_timer_again (loop, ev_timer *)
1190		2240
1191	This will act as if the timer timed out and restart it again if it is	2241	This will act as if the timer timed out, and restarts it again if it is
1192	repeating. The exact semantics are:	2242	repeating. It basically works like calling C<ev_timer_stop>, updating the
		2243	timeout to the C<repeat> value and calling C<ev_timer_start>.
1193		2244
		2245	The exact semantics are as in the following rules, all of which will be
		2246	applied to the watcher:
		2247
		2248	=over 4
		2249
1194	If the timer is pending, its pending status is cleared.	2250	=item If the timer is pending, the pending status is always cleared.
1195		2251
1196	If the timer is started but nonrepeating, stop it (as if it timed out).	2252	=item If the timer is started but non-repeating, stop it (as if it timed
		2253	out, without invoking it).
1197		2254
1198	If the timer is repeating, either start it if necessary (with the	2255	=item If the timer is repeating, make the C<repeat> value the new timeout
1199	C<repeat> value), or reset the running timer to the C<repeat> value.	2256	and start the timer, if necessary.
1200		2257
1201	This sounds a bit complicated, but here is a useful and typical	2258	=back
1202	example: Imagine you have a tcp connection and you want a so-called idle
1203	timeout, that is, you want to be called when there have been, say, 60
1204	seconds of inactivity on the socket. The easiest way to do this is to
1205	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
1206	C<ev_timer_again> each time you successfully read or write some data. If
1207	you go into an idle state where you do not expect data to travel on the
1208	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
1209	automatically restart it if need be.
1210		2259
1211	That means you can ignore the C<after> value and C<ev_timer_start>	2260	This sounds a bit complicated, see L</Be smart about timeouts>, above, for a
1212	altogether and only ever use the C<repeat> value and C<ev_timer_again>:	2261	usage example.
1213		2262
1214	ev_timer_init (timer, callback, 0., 5.);	2263	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
1215	ev_timer_again (loop, timer);
1216	...
1217	timer->again = 17.;
1218	ev_timer_again (loop, timer);
1219	...
1220	timer->again = 10.;
1221	ev_timer_again (loop, timer);
1222		2264
1223	This is more slightly efficient then stopping/starting the timer each time	2265	Returns the remaining time until a timer fires. If the timer is active,
1224	you want to modify its timeout value.	2266	then this time is relative to the current event loop time, otherwise it's
		2267	the timeout value currently configured.
		2268
		2269	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2270	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2271	will return C<4>. When the timer expires and is restarted, it will return
		2272	roughly C<7> (likely slightly less as callback invocation takes some time,
		2273	too), and so on.
1225		2274
1226	=item ev_tstamp repeat [read-write]	2275	=item ev_tstamp repeat [read-write]
1227		2276
1228	The current C<repeat> value. Will be used each time the watcher times out	2277	The current C<repeat> value. Will be used each time the watcher times out
1229	or C<ev_timer_again> is called and determines the next timeout (if any),	2278	or C<ev_timer_again> is called, and determines the next timeout (if any),
1230	which is also when any modifications are taken into account.	2279	which is also when any modifications are taken into account.
1231		2280
1232	=back	2281	=back
1233		2282
1234	=head3 Examples	2283	=head3 Examples
1235		2284
1236	Example: Create a timer that fires after 60 seconds.	2285	Example: Create a timer that fires after 60 seconds.
1237		2286
1238	static void	2287	static void
1239	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2288	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1240	{	2289	{
1241	.. one minute over, w is actually stopped right here	2290	.. one minute over, w is actually stopped right here
1242	}	2291	}
1243		2292
1244	struct ev_timer mytimer;	2293	ev_timer mytimer;
1245	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	2294	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1246	ev_timer_start (loop, &mytimer);	2295	ev_timer_start (loop, &mytimer);
1247		2296
1248	Example: Create a timeout timer that times out after 10 seconds of	2297	Example: Create a timeout timer that times out after 10 seconds of
1249	inactivity.	2298	inactivity.
1250		2299
1251	static void	2300	static void
1252	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2301	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1253	{	2302	{
1254	.. ten seconds without any activity	2303	.. ten seconds without any activity
1255	}	2304	}
1256		2305
1257	struct ev_timer mytimer;	2306	ev_timer mytimer;
1258	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	2307	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1259	ev_timer_again (&mytimer); /* start timer */	2308	ev_timer_again (&mytimer); /* start timer */
1260	ev_loop (loop, 0);	2309	ev_run (loop, 0);
1261		2310
1262	// and in some piece of code that gets executed on any "activity":	2311	// and in some piece of code that gets executed on any "activity":
1263	// reset the timeout to start ticking again at 10 seconds	2312	// reset the timeout to start ticking again at 10 seconds
1264	ev_timer_again (&mytimer);	2313	ev_timer_again (&mytimer);
1265		2314
1266		2315
1267	=head2 C<ev_periodic> - to cron or not to cron?	2316	=head2 C<ev_periodic> - to cron or not to cron?
1268		2317
1269	Periodic watchers are also timers of a kind, but they are very versatile	2318	Periodic watchers are also timers of a kind, but they are very versatile
1270	(and unfortunately a bit complex).	2319	(and unfortunately a bit complex).
1271		2320
1272	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2321	Unlike C<ev_timer>, periodic watchers are not based on real time (or
1273	but on wallclock time (absolute time). You can tell a periodic watcher	2322	relative time, the physical time that passes) but on wall clock time
1274	to trigger "at" some specific point in time. For example, if you tell a	2323	(absolute time, the thing you can read on your calendar or clock). The
1275	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	2324	difference is that wall clock time can run faster or slower than real
1276	+ 10.>) and then reset your system clock to the last year, then it will	2325	time, and time jumps are not uncommon (e.g. when you adjust your
1277	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2326	wrist-watch).
1278	roughly 10 seconds later).
1279		2327
1280	They can also be used to implement vastly more complex timers, such as	2328	You can tell a periodic watcher to trigger after some specific point
1281	triggering an event on each midnight, local time or other, complicated,	2329	in time: for example, if you tell a periodic watcher to trigger "in 10
1282	rules.	2330	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2331	not a delay) and then reset your system clock to January of the previous
		2332	year, then it will take a year or more to trigger the event (unlike an
		2333	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2334	it, as it uses a relative timeout).
1283		2335
		2336	C<ev_periodic> watchers can also be used to implement vastly more complex
		2337	timers, such as triggering an event on each "midnight, local time", or
		2338	other complicated rules. This cannot easily be done with C<ev_timer>
		2339	watchers, as those cannot react to time jumps.
		2340
1284	As with timers, the callback is guarenteed to be invoked only when the	2341	As with timers, the callback is guaranteed to be invoked only when the
1285	time (C<at>) has been passed, but if multiple periodic timers become ready	2342	point in time where it is supposed to trigger has passed. If multiple
1286	during the same loop iteration then order of execution is undefined.	2343	timers become ready during the same loop iteration then the ones with
		2344	earlier time-out values are invoked before ones with later time-out values
		2345	(but this is no longer true when a callback calls C<ev_run> recursively).
1287		2346
1288	=head3 Watcher-Specific Functions and Data Members	2347	=head3 Watcher-Specific Functions and Data Members
1289		2348
1290	=over 4	2349	=over 4
1291		2350
1292	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2351	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1293		2352
1294	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2353	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1295		2354
1296	Lots of arguments, lets sort it out... There are basically three modes of	2355	Lots of arguments, let's sort it out... There are basically three modes of
1297	operation, and we will explain them from simplest to complex:	2356	operation, and we will explain them from simplest to most complex:
1298		2357
1299	=over 4	2358	=over 4
1300		2359
1301	=item * absolute timer (at = time, interval = reschedule_cb = 0)	2360	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
1302		2361
1303	In this configuration the watcher triggers an event at the wallclock time	2362	In this configuration the watcher triggers an event after the wall clock
1304	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2363	time C<offset> has passed. It will not repeat and will not adjust when a
1305	that is, if it is to be run at January 1st 2011 then it will run when the	2364	time jump occurs, that is, if it is to be run at January 1st 2011 then it
1306	system time reaches or surpasses this time.	2365	will be stopped and invoked when the system clock reaches or surpasses
		2366	this point in time.
1307		2367
1308	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	2368	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
1309		2369
1310	In this mode the watcher will always be scheduled to time out at the next	2370	In this mode the watcher will always be scheduled to time out at the next
1311	C<at + N * interval> time (for some integer N, which can also be negative)	2371	C<offset + N * interval> time (for some integer N, which can also be
1312	and then repeat, regardless of any time jumps.	2372	negative) and then repeat, regardless of any time jumps. The C<offset>
		2373	argument is merely an offset into the C<interval> periods.
1313		2374
1314	This can be used to create timers that do not drift with respect to system	2375	This can be used to create timers that do not drift with respect to the
1315	time:	2376	system clock, for example, here is an C<ev_periodic> that triggers each
		2377	hour, on the hour (with respect to UTC):
1316		2378
1317	ev_periodic_set (&periodic, 0., 3600., 0);	2379	ev_periodic_set (&periodic, 0., 3600., 0);
1318		2380
1319	This doesn't mean there will always be 3600 seconds in between triggers,	2381	This doesn't mean there will always be 3600 seconds in between triggers,
1320	but only that the the callback will be called when the system time shows a	2382	but only that the callback will be called when the system time shows a
1321	full hour (UTC), or more correctly, when the system time is evenly divisible	2383	full hour (UTC), or more correctly, when the system time is evenly divisible
1322	by 3600.	2384	by 3600.
1323		2385
1324	Another way to think about it (for the mathematically inclined) is that	2386	Another way to think about it (for the mathematically inclined) is that
1325	C<ev_periodic> will try to run the callback in this mode at the next possible	2387	C<ev_periodic> will try to run the callback in this mode at the next possible
1326	time where C<time = at (mod interval)>, regardless of any time jumps.	2388	time where C<time = offset (mod interval)>, regardless of any time jumps.
1327		2389
1328	For numerical stability it is preferable that the C<at> value is near	2390	The C<interval> I<MUST> be positive, and for numerical stability, the
1329	C<ev_now ()> (the current time), but there is no range requirement for	2391	interval value should be higher than C<1/8192> (which is around 100
1330	this value.	2392	microseconds) and C<offset> should be higher than C<0> and should have
		2393	at most a similar magnitude as the current time (say, within a factor of
		2394	ten). Typical values for offset are, in fact, C<0> or something between
		2395	C<0> and C<interval>, which is also the recommended range.
1331		2396
		2397	Note also that there is an upper limit to how often a timer can fire (CPU
		2398	speed for example), so if C<interval> is very small then timing stability
		2399	will of course deteriorate. Libev itself tries to be exact to be about one
		2400	millisecond (if the OS supports it and the machine is fast enough).
		2401
1332	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)	2402	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
1333		2403
1334	In this mode the values for C<interval> and C<at> are both being	2404	In this mode the values for C<interval> and C<offset> are both being
1335	ignored. Instead, each time the periodic watcher gets scheduled, the	2405	ignored. Instead, each time the periodic watcher gets scheduled, the
1336	reschedule callback will be called with the watcher as first, and the	2406	reschedule callback will be called with the watcher as first, and the
1337	current time as second argument.	2407	current time as second argument.
1338		2408
1339	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2409	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
1340	ever, or make any event loop modifications>. If you need to stop it,	2410	or make ANY other event loop modifications whatsoever, unless explicitly
1341	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2411	allowed by documentation here>.
1342	starting an C<ev_prepare> watcher, which is legal).
1343		2412
		2413	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2414	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2415	only event loop modification you are allowed to do).
		2416
1344	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2417	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1345	ev_tstamp now)>, e.g.:	2418	*w, ev_tstamp now)>, e.g.:
1346		2419
		2420	static ev_tstamp
1347	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2421	my_rescheduler (ev_periodic *w, ev_tstamp now)
1348	{	2422	{
1349	return now + 60.;	2423	return now + 60.;
1350	}	2424	}
1351		2425
1352	It must return the next time to trigger, based on the passed time value	2426	It must return the next time to trigger, based on the passed time value
1353	(that is, the lowest time value larger than to the second argument). It	2427	(that is, the lowest time value larger than to the second argument). It
1354	will usually be called just before the callback will be triggered, but	2428	will usually be called just before the callback will be triggered, but
1355	might be called at other times, too.	2429	might be called at other times, too.
1356		2430
1357	NOTE: I<< This callback must always return a time that is later than the	2431	NOTE: I<< This callback must always return a time that is higher than or
1358	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2432	equal to the passed C<now> value >>.
1359		2433
1360	This can be used to create very complex timers, such as a timer that	2434	This can be used to create very complex timers, such as a timer that
1361	triggers on each midnight, local time. To do this, you would calculate the	2435	triggers on "next midnight, local time". To do this, you would calculate
1362	next midnight after C<now> and return the timestamp value for this. How	2436	the next midnight after C<now> and return the timestamp value for
1363	you do this is, again, up to you (but it is not trivial, which is the main	2437	this. Here is a (completely untested, no error checking) example on how to
1364	reason I omitted it as an example).	2438	do this:
		2439
		2440	#include <time.h>
		2441
		2442	static ev_tstamp
		2443	my_rescheduler (ev_periodic *w, ev_tstamp now)
		2444	{
		2445	time_t tnow = (time_t)now;
		2446	struct tm tm;
		2447	localtime_r (&tnow, &tm);
		2448
		2449	tm.tm_sec = tm.tm_min = tm.tm_hour = 0; // midnight current day
		2450	++tm.tm_mday; // midnight next day
		2451
		2452	return mktime (&tm);
		2453	}
		2454
		2455	Note: this code might run into trouble on days that have more then two
		2456	midnights (beginning and end).
1365		2457
1366	=back	2458	=back
1367		2459
1368	=item ev_periodic_again (loop, ev_periodic *)	2460	=item ev_periodic_again (loop, ev_periodic *)
1369		2461
1370	Simply stops and restarts the periodic watcher again. This is only useful	2462	Simply stops and restarts the periodic watcher again. This is only useful
1371	when you changed some parameters or the reschedule callback would return	2463	when you changed some parameters or the reschedule callback would return
1372	a different time than the last time it was called (e.g. in a crond like	2464	a different time than the last time it was called (e.g. in a crond like
1373	program when the crontabs have changed).	2465	program when the crontabs have changed).
1374		2466
		2467	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2468
		2469	When active, returns the absolute time that the watcher is supposed
		2470	to trigger next. This is not the same as the C<offset> argument to
		2471	C<ev_periodic_set>, but indeed works even in interval and manual
		2472	rescheduling modes.
		2473
1375	=item ev_tstamp offset [read-write]	2474	=item ev_tstamp offset [read-write]
1376		2475
1377	When repeating, this contains the offset value, otherwise this is the	2476	When repeating, this contains the offset value, otherwise this is the
1378	absolute point in time (the C<at> value passed to C<ev_periodic_set>).	2477	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2478	although libev might modify this value for better numerical stability).
1379		2479
1380	Can be modified any time, but changes only take effect when the periodic	2480	Can be modified any time, but changes only take effect when the periodic
1381	timer fires or C<ev_periodic_again> is being called.	2481	timer fires or C<ev_periodic_again> is being called.
1382		2482
1383	=item ev_tstamp interval [read-write]	2483	=item ev_tstamp interval [read-write]
1384		2484
1385	The current interval value. Can be modified any time, but changes only	2485	The current interval value. Can be modified any time, but changes only
1386	take effect when the periodic timer fires or C<ev_periodic_again> is being	2486	take effect when the periodic timer fires or C<ev_periodic_again> is being
1387	called.	2487	called.
1388		2488
1389	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	2489	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1390		2490
1391	The current reschedule callback, or C<0>, if this functionality is	2491	The current reschedule callback, or C<0>, if this functionality is
1392	switched off. Can be changed any time, but changes only take effect when	2492	switched off. Can be changed any time, but changes only take effect when
1393	the periodic timer fires or C<ev_periodic_again> is being called.	2493	the periodic timer fires or C<ev_periodic_again> is being called.
1394		2494
1395	=item ev_tstamp at [read-only]
1396
1397	When active, contains the absolute time that the watcher is supposed to
1398	trigger next.
1399
1400	=back	2495	=back
1401		2496
1402	=head3 Examples	2497	=head3 Examples
1403		2498
1404	Example: Call a callback every hour, or, more precisely, whenever the	2499	Example: Call a callback every hour, or, more precisely, whenever the
1405	system clock is divisible by 3600. The callback invocation times have	2500	system time is divisible by 3600. The callback invocation times have
1406	potentially a lot of jittering, but good long-term stability.	2501	potentially a lot of jitter, but good long-term stability.
1407		2502
1408	static void	2503	static void
1409	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	2504	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
1410	{	2505	{
1411	... its now a full hour (UTC, or TAI or whatever your clock follows)	2506	... its now a full hour (UTC, or TAI or whatever your clock follows)
1412	}	2507	}
1413		2508
1414	struct ev_periodic hourly_tick;	2509	ev_periodic hourly_tick;
1415	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	2510	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1416	ev_periodic_start (loop, &hourly_tick);	2511	ev_periodic_start (loop, &hourly_tick);
1417		2512
1418	Example: The same as above, but use a reschedule callback to do it:	2513	Example: The same as above, but use a reschedule callback to do it:
1419		2514
1420	#include <math.h>	2515	#include <math.h>
1421		2516
1422	static ev_tstamp	2517	static ev_tstamp
1423	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	2518	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1424	{	2519	{
1425	return fmod (now, 3600.) + 3600.;	2520	return now + (3600. - fmod (now, 3600.));
1426	}	2521	}
1427		2522
1428	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	2523	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1429		2524
1430	Example: Call a callback every hour, starting now:	2525	Example: Call a callback every hour, starting now:
1431		2526
1432	struct ev_periodic hourly_tick;	2527	ev_periodic hourly_tick;
1433	ev_periodic_init (&hourly_tick, clock_cb,	2528	ev_periodic_init (&hourly_tick, clock_cb,
1434	fmod (ev_now (loop), 3600.), 3600., 0);	2529	fmod (ev_now (loop), 3600.), 3600., 0);
1435	ev_periodic_start (loop, &hourly_tick);	2530	ev_periodic_start (loop, &hourly_tick);
1436		2531
1437		2532
1438	=head2 C<ev_signal> - signal me when a signal gets signalled!	2533	=head2 C<ev_signal> - signal me when a signal gets signalled!
1439		2534
1440	Signal watchers will trigger an event when the process receives a specific	2535	Signal watchers will trigger an event when the process receives a specific
1441	signal one or more times. Even though signals are very asynchronous, libev	2536	signal one or more times. Even though signals are very asynchronous, libev
1442	will try it's best to deliver signals synchronously, i.e. as part of the	2537	will try its best to deliver signals synchronously, i.e. as part of the
1443	normal event processing, like any other event.	2538	normal event processing, like any other event.
1444		2539
		2540	If you want signals to be delivered truly asynchronously, just use
		2541	C<sigaction> as you would do without libev and forget about sharing
		2542	the signal. You can even use C<ev_async> from a signal handler to
		2543	synchronously wake up an event loop.
		2544
1445	You can configure as many watchers as you like per signal. Only when the	2545	You can configure as many watchers as you like for the same signal, but
1446	first watcher gets started will libev actually register a signal watcher	2546	only within the same loop, i.e. you can watch for C<SIGINT> in your
1447	with the kernel (thus it coexists with your own signal handlers as long	2547	default loop and for C<SIGIO> in another loop, but you cannot watch for
1448	as you don't register any with libev). Similarly, when the last signal	2548	C<SIGINT> in both the default loop and another loop at the same time. At
1449	watcher for a signal is stopped libev will reset the signal handler to	2549	the moment, C<SIGCHLD> is permanently tied to the default loop.
1450	SIG_DFL (regardless of what it was set to before).	2550
		2551	Only after the first watcher for a signal is started will libev actually
		2552	register something with the kernel. It thus coexists with your own signal
		2553	handlers as long as you don't register any with libev for the same signal.
1451		2554
1452	If possible and supported, libev will install its handlers with	2555	If possible and supported, libev will install its handlers with
1453	C<SA_RESTART> behaviour enabled, so syscalls should not be unduly	2556	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
1454	interrupted. If you have a problem with syscalls getting interrupted by	2557	not be unduly interrupted. If you have a problem with system calls getting
1455	signals you can block all signals in an C<ev_check> watcher and unblock	2558	interrupted by signals you can block all signals in an C<ev_check> watcher
1456	them in an C<ev_prepare> watcher.	2559	and unblock them in an C<ev_prepare> watcher.
		2560
		2561	=head3 The special problem of inheritance over fork/execve/pthread_create
		2562
		2563	Both the signal mask (C<sigprocmask>) and the signal disposition
		2564	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2565	stopping it again), that is, libev might or might not block the signal,
		2566	and might or might not set or restore the installed signal handler (but
		2567	see C<EVFLAG_NOSIGMASK>).
		2568
		2569	While this does not matter for the signal disposition (libev never
		2570	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2571	C<execve>), this matters for the signal mask: many programs do not expect
		2572	certain signals to be blocked.
		2573
		2574	This means that before calling C<exec> (from the child) you should reset
		2575	the signal mask to whatever "default" you expect (all clear is a good
		2576	choice usually).
		2577
		2578	The simplest way to ensure that the signal mask is reset in the child is
		2579	to install a fork handler with C<pthread_atfork> that resets it. That will
		2580	catch fork calls done by libraries (such as the libc) as well.
		2581
		2582	In current versions of libev, the signal will not be blocked indefinitely
		2583	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2584	the window of opportunity for problems, it will not go away, as libev
		2585	I<has> to modify the signal mask, at least temporarily.
		2586
		2587	So I can't stress this enough: I<If you do not reset your signal mask when
		2588	you expect it to be empty, you have a race condition in your code>. This
		2589	is not a libev-specific thing, this is true for most event libraries.
		2590
		2591	=head3 The special problem of threads signal handling
		2592
		2593	POSIX threads has problematic signal handling semantics, specifically,
		2594	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2595	threads in a process block signals, which is hard to achieve.
		2596
		2597	When you want to use sigwait (or mix libev signal handling with your own
		2598	for the same signals), you can tackle this problem by globally blocking
		2599	all signals before creating any threads (or creating them with a fully set
		2600	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2601	loops. Then designate one thread as "signal receiver thread" which handles
		2602	these signals. You can pass on any signals that libev might be interested
		2603	in by calling C<ev_feed_signal>.
1457		2604
1458	=head3 Watcher-Specific Functions and Data Members	2605	=head3 Watcher-Specific Functions and Data Members
1459		2606
1460	=over 4	2607	=over 4
1461		2608
…		…
1472		2619
1473	=back	2620	=back
1474		2621
1475	=head3 Examples	2622	=head3 Examples
1476		2623
1477	Example: Try to exit cleanly on SIGINT and SIGTERM.	2624	Example: Try to exit cleanly on SIGINT.
1478		2625
1479	static void	2626	static void
1480	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	2627	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
1481	{	2628	{
1482	ev_unloop (loop, EVUNLOOP_ALL);	2629	ev_break (loop, EVBREAK_ALL);
1483	}	2630	}
1484		2631
1485	struct ev_signal signal_watcher;	2632	ev_signal signal_watcher;
1486	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	2633	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1487	ev_signal_start (loop, &sigint_cb);	2634	ev_signal_start (loop, &signal_watcher);
1488		2635
1489		2636
1490	=head2 C<ev_child> - watch out for process status changes	2637	=head2 C<ev_child> - watch out for process status changes
1491		2638
1492	Child watchers trigger when your process receives a SIGCHLD in response to	2639	Child watchers trigger when your process receives a SIGCHLD in response to
1493	some child status changes (most typically when a child of yours dies). It	2640	some child status changes (most typically when a child of yours dies or
1494	is permissible to install a child watcher I<after> the child has been	2641	exits). It is permissible to install a child watcher I<after> the child
1495	forked (which implies it might have already exited), as long as the event	2642	has been forked (which implies it might have already exited), as long
1496	loop isn't entered (or is continued from a watcher).	2643	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2644	forking and then immediately registering a watcher for the child is fine,
		2645	but forking and registering a watcher a few event loop iterations later or
		2646	in the next callback invocation is not.
1497		2647
1498	Only the default event loop is capable of handling signals, and therefore	2648	Only the default event loop is capable of handling signals, and therefore
1499	you can only rgeister child watchers in the default event loop.	2649	you can only register child watchers in the default event loop.
		2650
		2651	Due to some design glitches inside libev, child watchers will always be
		2652	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2653	libev)
1500		2654
1501	=head3 Process Interaction	2655	=head3 Process Interaction
1502		2656
1503	Libev grabs C<SIGCHLD> as soon as the default event loop is	2657	Libev grabs C<SIGCHLD> as soon as the default event loop is
1504	initialised. This is necessary to guarantee proper behaviour even if	2658	initialised. This is necessary to guarantee proper behaviour even if the
1505	the first child watcher is started after the child exits. The occurance	2659	first child watcher is started after the child exits. The occurrence
1506	of C<SIGCHLD> is recorded asynchronously, but child reaping is done	2660	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
1507	synchronously as part of the event loop processing. Libev always reaps all	2661	synchronously as part of the event loop processing. Libev always reaps all
1508	children, even ones not watched.	2662	children, even ones not watched.
1509		2663
1510	=head3 Overriding the Built-In Processing	2664	=head3 Overriding the Built-In Processing
…		…
1514	handler, you can override it easily by installing your own handler for	2668	handler, you can override it easily by installing your own handler for
1515	C<SIGCHLD> after initialising the default loop, and making sure the	2669	C<SIGCHLD> after initialising the default loop, and making sure the
1516	default loop never gets destroyed. You are encouraged, however, to use an	2670	default loop never gets destroyed. You are encouraged, however, to use an
1517	event-based approach to child reaping and thus use libev's support for	2671	event-based approach to child reaping and thus use libev's support for
1518	that, so other libev users can use C<ev_child> watchers freely.	2672	that, so other libev users can use C<ev_child> watchers freely.
		2673
		2674	=head3 Stopping the Child Watcher
		2675
		2676	Currently, the child watcher never gets stopped, even when the
		2677	child terminates, so normally one needs to stop the watcher in the
		2678	callback. Future versions of libev might stop the watcher automatically
		2679	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2680	problem).
1519		2681
1520	=head3 Watcher-Specific Functions and Data Members	2682	=head3 Watcher-Specific Functions and Data Members
1521		2683
1522	=over 4	2684	=over 4
1523		2685
…		…
1552	=head3 Examples	2714	=head3 Examples
1553		2715
1554	Example: C<fork()> a new process and install a child handler to wait for	2716	Example: C<fork()> a new process and install a child handler to wait for
1555	its completion.	2717	its completion.
1556		2718
1557	ev_child cw;	2719	ev_child cw;
1558		2720
1559	static void	2721	static void
1560	child_cb (EV_P_ struct ev_child *w, int revents)	2722	child_cb (EV_P_ ev_child *w, int revents)
1561	{	2723	{
1562	ev_child_stop (EV_A_ w);	2724	ev_child_stop (EV_A_ w);
1563	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);	2725	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1564	}	2726	}
1565		2727
1566	pid_t pid = fork ();	2728	pid_t pid = fork ();
1567		2729
1568	if (pid < 0)	2730	if (pid < 0)
1569	// error	2731	// error
1570	else if (pid == 0)	2732	else if (pid == 0)
1571	{	2733	{
1572	// the forked child executes here	2734	// the forked child executes here
1573	exit (1);	2735	exit (1);
1574	}	2736	}
1575	else	2737	else
1576	{	2738	{
1577	ev_child_init (&cw, child_cb, pid, 0);	2739	ev_child_init (&cw, child_cb, pid, 0);
1578	ev_child_start (EV_DEFAULT_ &cw);	2740	ev_child_start (EV_DEFAULT_ &cw);
1579	}	2741	}
1580		2742
1581		2743
1582	=head2 C<ev_stat> - did the file attributes just change?	2744	=head2 C<ev_stat> - did the file attributes just change?
1583		2745
1584	This watches a filesystem path for attribute changes. That is, it calls	2746	This watches a file system path for attribute changes. That is, it calls
1585	C<stat> regularly (or when the OS says it changed) and sees if it changed	2747	C<stat> on that path in regular intervals (or when the OS says it changed)
1586	compared to the last time, invoking the callback if it did.	2748	and sees if it changed compared to the last time, invoking the callback
		2749	if it did. Starting the watcher C<stat>'s the file, so only changes that
		2750	happen after the watcher has been started will be reported.
1587		2751
1588	The path does not need to exist: changing from "path exists" to "path does	2752	The path does not need to exist: changing from "path exists" to "path does
1589	not exist" is a status change like any other. The condition "path does	2753	not exist" is a status change like any other. The condition "path does not
1590	not exist" is signified by the C<st_nlink> field being zero (which is	2754	exist" (or more correctly "path cannot be stat'ed") is signified by the
1591	otherwise always forced to be at least one) and all the other fields of	2755	C<st_nlink> field being zero (which is otherwise always forced to be at
1592	the stat buffer having unspecified contents.	2756	least one) and all the other fields of the stat buffer having unspecified
		2757	contents.
1593		2758
1594	The path I<should> be absolute and I<must not> end in a slash. If it is	2759	The path I<must not> end in a slash or contain special components such as
		2760	C<.> or C<..>. The path I<should> be absolute: If it is relative and
1595	relative and your working directory changes, the behaviour is undefined.	2761	your working directory changes, then the behaviour is undefined.
1596		2762
1597	Since there is no standard to do this, the portable implementation simply	2763	Since there is no portable change notification interface available, the
1598	calls C<stat (2)> regularly on the path to see if it changed somehow. You	2764	portable implementation simply calls C<stat(2)> regularly on the path
1599	can specify a recommended polling interval for this case. If you specify	2765	to see if it changed somehow. You can specify a recommended polling
1600	a polling interval of C<0> (highly recommended!) then a I<suitable,	2766	interval for this case. If you specify a polling interval of C<0> (highly
1601	unspecified default> value will be used (which you can expect to be around	2767	recommended!) then a I<suitable, unspecified default> value will be used
1602	five seconds, although this might change dynamically). Libev will also	2768	(which you can expect to be around five seconds, although this might
1603	impose a minimum interval which is currently around C<0.1>, but thats	2769	change dynamically). Libev will also impose a minimum interval which is
1604	usually overkill.	2770	currently around C<0.1>, but that's usually overkill.
1605		2771
1606	This watcher type is not meant for massive numbers of stat watchers,	2772	This watcher type is not meant for massive numbers of stat watchers,
1607	as even with OS-supported change notifications, this can be	2773	as even with OS-supported change notifications, this can be
1608	resource-intensive.	2774	resource-intensive.
1609		2775
1610	At the time of this writing, only the Linux inotify interface is	2776	At the time of this writing, the only OS-specific interface implemented
1611	implemented (implementing kqueue support is left as an exercise for the	2777	is the Linux inotify interface (implementing kqueue support is left as an
1612	reader). Inotify will be used to give hints only and should not change the	2778	exercise for the reader. Note, however, that the author sees no way of
1613	semantics of C<ev_stat> watchers, which means that libev sometimes needs	2779	implementing C<ev_stat> semantics with kqueue, except as a hint).
1614	to fall back to regular polling again even with inotify, but changes are
1615	usually detected immediately, and if the file exists there will be no
1616	polling.
1617		2780
1618	=head3 ABI Issues (Largefile Support)	2781	=head3 ABI Issues (Largefile Support)
1619		2782
1620	Libev by default (unless the user overrides this) uses the default	2783	Libev by default (unless the user overrides this) uses the default
1621	compilation environment, which means that on systems with optionally	2784	compilation environment, which means that on systems with large file
1622	disabled large file support, you get the 32 bit version of the stat	2785	support disabled by default, you get the 32 bit version of the stat
1623	structure. When using the library from programs that change the ABI to	2786	structure. When using the library from programs that change the ABI to
1624	use 64 bit file offsets the programs will fail. In that case you have to	2787	use 64 bit file offsets the programs will fail. In that case you have to
1625	compile libev with the same flags to get binary compatibility. This is	2788	compile libev with the same flags to get binary compatibility. This is
1626	obviously the case with any flags that change the ABI, but the problem is	2789	obviously the case with any flags that change the ABI, but the problem is
1627	most noticably with ev_stat and largefile support.	2790	most noticeably displayed with ev_stat and large file support.
1628		2791
1629	=head3 Inotify	2792	The solution for this is to lobby your distribution maker to make large
		2793	file interfaces available by default (as e.g. FreeBSD does) and not
		2794	optional. Libev cannot simply switch on large file support because it has
		2795	to exchange stat structures with application programs compiled using the
		2796	default compilation environment.
1630		2797
		2798	=head3 Inotify and Kqueue
		2799
1631	When C<inotify (7)> support has been compiled into libev (generally only	2800	When C<inotify (7)> support has been compiled into libev and present at
1632	available on Linux) and present at runtime, it will be used to speed up	2801	runtime, it will be used to speed up change detection where possible. The
1633	change detection where possible. The inotify descriptor will be created lazily	2802	inotify descriptor will be created lazily when the first C<ev_stat>
1634	when the first C<ev_stat> watcher is being started.	2803	watcher is being started.
1635		2804
1636	Inotify presense does not change the semantics of C<ev_stat> watchers	2805	Inotify presence does not change the semantics of C<ev_stat> watchers
1637	except that changes might be detected earlier, and in some cases, to avoid	2806	except that changes might be detected earlier, and in some cases, to avoid
1638	making regular C<stat> calls. Even in the presense of inotify support	2807	making regular C<stat> calls. Even in the presence of inotify support
1639	there are many cases where libev has to resort to regular C<stat> polling.	2808	there are many cases where libev has to resort to regular C<stat> polling,
		2809	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2810	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2811	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2812	xfs are fully working) libev usually gets away without polling.
1640		2813
1641	(There is no support for kqueue, as apparently it cannot be used to	2814	There is no support for kqueue, as apparently it cannot be used to
1642	implement this functionality, due to the requirement of having a file	2815	implement this functionality, due to the requirement of having a file
1643	descriptor open on the object at all times).	2816	descriptor open on the object at all times, and detecting renames, unlinks
		2817	etc. is difficult.
		2818
		2819	=head3 C<stat ()> is a synchronous operation
		2820
		2821	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2822	the process. The exception are C<ev_stat> watchers - those call C<stat
		2823	()>, which is a synchronous operation.
		2824
		2825	For local paths, this usually doesn't matter: unless the system is very
		2826	busy or the intervals between stat's are large, a stat call will be fast,
		2827	as the path data is usually in memory already (except when starting the
		2828	watcher).
		2829
		2830	For networked file systems, calling C<stat ()> can block an indefinite
		2831	time due to network issues, and even under good conditions, a stat call
		2832	often takes multiple milliseconds.
		2833
		2834	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2835	paths, although this is fully supported by libev.
1644		2836
1645	=head3 The special problem of stat time resolution	2837	=head3 The special problem of stat time resolution
1646		2838
1647	The C<stat ()> syscall only supports full-second resolution portably, and	2839	The C<stat ()> system call only supports full-second resolution portably,
1648	even on systems where the resolution is higher, many filesystems still	2840	and even on systems where the resolution is higher, most file systems
1649	only support whole seconds.	2841	still only support whole seconds.
1650		2842
1651	That means that, if the time is the only thing that changes, you might	2843	That means that, if the time is the only thing that changes, you can
1652	miss updates: on the first update, C<ev_stat> detects a change and calls	2844	easily miss updates: on the first update, C<ev_stat> detects a change and
1653	your callback, which does something. When there is another update within	2845	calls your callback, which does something. When there is another update
1654	the same second, C<ev_stat> will be unable to detect it.	2846	within the same second, C<ev_stat> will be unable to detect unless the
		2847	stat data does change in other ways (e.g. file size).
1655		2848
1656	The solution to this is to delay acting on a change for a second (or till	2849	The solution to this is to delay acting on a change for slightly more
1657	the next second boundary), using a roughly one-second delay C<ev_timer>	2850	than a second (or till slightly after the next full second boundary), using
1658	(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>	2851	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
1659	is added to work around small timing inconsistencies of some operating	2852	ev_timer_again (loop, w)>).
1660	systems.	2853
		2854	The C<.02> offset is added to work around small timing inconsistencies
		2855	of some operating systems (where the second counter of the current time
		2856	might be be delayed. One such system is the Linux kernel, where a call to
		2857	C<gettimeofday> might return a timestamp with a full second later than
		2858	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2859	update file times then there will be a small window where the kernel uses
		2860	the previous second to update file times but libev might already execute
		2861	the timer callback).
1661		2862
1662	=head3 Watcher-Specific Functions and Data Members	2863	=head3 Watcher-Specific Functions and Data Members
1663		2864
1664	=over 4	2865	=over 4
1665		2866
…		…
1671	C<path>. The C<interval> is a hint on how quickly a change is expected to	2872	C<path>. The C<interval> is a hint on how quickly a change is expected to
1672	be detected and should normally be specified as C<0> to let libev choose	2873	be detected and should normally be specified as C<0> to let libev choose
1673	a suitable value. The memory pointed to by C<path> must point to the same	2874	a suitable value. The memory pointed to by C<path> must point to the same
1674	path for as long as the watcher is active.	2875	path for as long as the watcher is active.
1675		2876
1676	The callback will be receive C<EV_STAT> when a change was detected,	2877	The callback will receive an C<EV_STAT> event when a change was detected,
1677	relative to the attributes at the time the watcher was started (or the	2878	relative to the attributes at the time the watcher was started (or the
1678	last change was detected).	2879	last change was detected).
1679		2880
1680	=item ev_stat_stat (loop, ev_stat *)	2881	=item ev_stat_stat (loop, ev_stat *)
1681		2882
1682	Updates the stat buffer immediately with new values. If you change the	2883	Updates the stat buffer immediately with new values. If you change the
1683	watched path in your callback, you could call this fucntion to avoid	2884	watched path in your callback, you could call this function to avoid
1684	detecting this change (while introducing a race condition). Can also be	2885	detecting this change (while introducing a race condition if you are not
1685	useful simply to find out the new values.	2886	the only one changing the path). Can also be useful simply to find out the
		2887	new values.
1686		2888
1687	=item ev_statdata attr [read-only]	2889	=item ev_statdata attr [read-only]
1688		2890
1689	The most-recently detected attributes of the file. Although the type is of	2891	The most-recently detected attributes of the file. Although the type is
1690	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types	2892	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1691	suitable for your system. If the C<st_nlink> member is C<0>, then there	2893	suitable for your system, but you can only rely on the POSIX-standardised
		2894	members to be present. If the C<st_nlink> member is C<0>, then there was
1692	was some error while C<stat>ing the file.	2895	some error while C<stat>ing the file.
1693		2896
1694	=item ev_statdata prev [read-only]	2897	=item ev_statdata prev [read-only]
1695		2898
1696	The previous attributes of the file. The callback gets invoked whenever	2899	The previous attributes of the file. The callback gets invoked whenever
1697	C<prev> != C<attr>.	2900	C<prev> != C<attr>, or, more precisely, one or more of these members
		2901	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2902	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
1698		2903
1699	=item ev_tstamp interval [read-only]	2904	=item ev_tstamp interval [read-only]
1700		2905
1701	The specified interval.	2906	The specified interval.
1702		2907
1703	=item const char *path [read-only]	2908	=item const char *path [read-only]
1704		2909
1705	The filesystem path that is being watched.	2910	The file system path that is being watched.
1706		2911
1707	=back	2912	=back
1708		2913
1709	=head3 Examples	2914	=head3 Examples
1710		2915
1711	Example: Watch C</etc/passwd> for attribute changes.	2916	Example: Watch C</etc/passwd> for attribute changes.
1712		2917
1713	static void	2918	static void
1714	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2919	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1715	{	2920	{
1716	/* /etc/passwd changed in some way */	2921	/* /etc/passwd changed in some way */
1717	if (w->attr.st_nlink)	2922	if (w->attr.st_nlink)
1718	{	2923	{
1719	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2924	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1720	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2925	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1721	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2926	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1722	}	2927	}
1723	else	2928	else
1724	/* you shalt not abuse printf for puts */	2929	/* you shalt not abuse printf for puts */
1725	puts ("wow, /etc/passwd is not there, expect problems. "	2930	puts ("wow, /etc/passwd is not there, expect problems. "
1726	"if this is windows, they already arrived\n");	2931	"if this is windows, they already arrived\n");
1727	}	2932	}
1728		2933
1729	...	2934	...
1730	ev_stat passwd;	2935	ev_stat passwd;
1731		2936
1732	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);	2937	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1733	ev_stat_start (loop, &passwd);	2938	ev_stat_start (loop, &passwd);
1734		2939
1735	Example: Like above, but additionally use a one-second delay so we do not	2940	Example: Like above, but additionally use a one-second delay so we do not
1736	miss updates (however, frequent updates will delay processing, too, so	2941	miss updates (however, frequent updates will delay processing, too, so
1737	one might do the work both on C<ev_stat> callback invocation I<and> on	2942	one might do the work both on C<ev_stat> callback invocation I<and> on
1738	C<ev_timer> callback invocation).	2943	C<ev_timer> callback invocation).
1739		2944
1740	static ev_stat passwd;	2945	static ev_stat passwd;
1741	static ev_timer timer;	2946	static ev_timer timer;
1742		2947
1743	static void	2948	static void
1744	timer_cb (EV_P_ ev_timer *w, int revents)	2949	timer_cb (EV_P_ ev_timer *w, int revents)
1745	{	2950	{
1746	ev_timer_stop (EV_A_ w);	2951	ev_timer_stop (EV_A_ w);
1747		2952
1748	/* now it's one second after the most recent passwd change */	2953	/* now it's one second after the most recent passwd change */
1749	}	2954	}
1750		2955
1751	static void	2956	static void
1752	stat_cb (EV_P_ ev_stat *w, int revents)	2957	stat_cb (EV_P_ ev_stat *w, int revents)
1753	{	2958	{
1754	/* reset the one-second timer */	2959	/* reset the one-second timer */
1755	ev_timer_again (EV_A_ &timer);	2960	ev_timer_again (EV_A_ &timer);
1756	}	2961	}
1757		2962
1758	...	2963	...
1759	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);	2964	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1760	ev_stat_start (loop, &passwd);	2965	ev_stat_start (loop, &passwd);
1761	ev_timer_init (&timer, timer_cb, 0., 1.01);	2966	ev_timer_init (&timer, timer_cb, 0., 1.02);
1762		2967
1763		2968
1764	=head2 C<ev_idle> - when you've got nothing better to do...	2969	=head2 C<ev_idle> - when you've got nothing better to do...
1765		2970
1766	Idle watchers trigger events when no other events of the same or higher	2971	Idle watchers trigger events when no other events of the same or higher
1767	priority are pending (prepare, check and other idle watchers do not	2972	priority are pending (prepare, check and other idle watchers do not count
1768	count).	2973	as receiving "events").
1769		2974
1770	That is, as long as your process is busy handling sockets or timeouts	2975	That is, as long as your process is busy handling sockets or timeouts
1771	(or even signals, imagine) of the same or higher priority it will not be	2976	(or even signals, imagine) of the same or higher priority it will not be
1772	triggered. But when your process is idle (or only lower-priority watchers	2977	triggered. But when your process is idle (or only lower-priority watchers
1773	are pending), the idle watchers are being called once per event loop	2978	are pending), the idle watchers are being called once per event loop
…		…
1780	Apart from keeping your process non-blocking (which is a useful	2985	Apart from keeping your process non-blocking (which is a useful
1781	effect on its own sometimes), idle watchers are a good place to do	2986	effect on its own sometimes), idle watchers are a good place to do
1782	"pseudo-background processing", or delay processing stuff to after the	2987	"pseudo-background processing", or delay processing stuff to after the
1783	event loop has handled all outstanding events.	2988	event loop has handled all outstanding events.
1784		2989
		2990	=head3 Abusing an C<ev_idle> watcher for its side-effect
		2991
		2992	As long as there is at least one active idle watcher, libev will never
		2993	sleep unnecessarily. Or in other words, it will loop as fast as possible.
		2994	For this to work, the idle watcher doesn't need to be invoked at all - the
		2995	lowest priority will do.
		2996
		2997	This mode of operation can be useful together with an C<ev_check> watcher,
		2998	to do something on each event loop iteration - for example to balance load
		2999	between different connections.
		3000
		3001	See L</Abusing an ev_check watcher for its side-effect> for a longer
		3002	example.
		3003
1785	=head3 Watcher-Specific Functions and Data Members	3004	=head3 Watcher-Specific Functions and Data Members
1786		3005
1787	=over 4	3006	=over 4
1788		3007
1789	=item ev_idle_init (ev_signal *, callback)	3008	=item ev_idle_init (ev_idle *, callback)
1790		3009
1791	Initialises and configures the idle watcher - it has no parameters of any	3010	Initialises and configures the idle watcher - it has no parameters of any
1792	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	3011	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1793	believe me.	3012	believe me.
1794		3013
…		…
1797	=head3 Examples	3016	=head3 Examples
1798		3017
1799	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	3018	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1800	callback, free it. Also, use no error checking, as usual.	3019	callback, free it. Also, use no error checking, as usual.
1801		3020
1802	static void	3021	static void
1803	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	3022	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1804	{	3023	{
		3024	// stop the watcher
		3025	ev_idle_stop (loop, w);
		3026
		3027	// now we can free it
1805	free (w);	3028	free (w);
		3029
1806	// now do something you wanted to do when the program has	3030	// now do something you wanted to do when the program has
1807	// no longer anything immediate to do.	3031	// no longer anything immediate to do.
1808	}	3032	}
1809		3033
1810	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	3034	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1811	ev_idle_init (idle_watcher, idle_cb);	3035	ev_idle_init (idle_watcher, idle_cb);
1812	ev_idle_start (loop, idle_cb);	3036	ev_idle_start (loop, idle_watcher);
1813		3037
1814		3038
1815	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	3039	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1816		3040
1817	Prepare and check watchers are usually (but not always) used in tandem:	3041	Prepare and check watchers are often (but not always) used in pairs:
1818	prepare watchers get invoked before the process blocks and check watchers	3042	prepare watchers get invoked before the process blocks and check watchers
1819	afterwards.	3043	afterwards.
1820		3044
1821	You I<must not> call C<ev_loop> or similar functions that enter	3045	You I<must not> call C<ev_run> (or similar functions that enter the
1822	the current event loop from either C<ev_prepare> or C<ev_check>	3046	current event loop) or C<ev_loop_fork> from either C<ev_prepare> or
1823	watchers. Other loops than the current one are fine, however. The	3047	C<ev_check> watchers. Other loops than the current one are fine,
1824	rationale behind this is that you do not need to check for recursion in	3048	however. The rationale behind this is that you do not need to check
1825	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	3049	for recursion in those watchers, i.e. the sequence will always be
1826	C<ev_check> so if you have one watcher of each kind they will always be	3050	C<ev_prepare>, blocking, C<ev_check> so if you have one watcher of each
1827	called in pairs bracketing the blocking call.	3051	kind they will always be called in pairs bracketing the blocking call.
1828		3052
1829	Their main purpose is to integrate other event mechanisms into libev and	3053	Their main purpose is to integrate other event mechanisms into libev and
1830	their use is somewhat advanced. This could be used, for example, to track	3054	their use is somewhat advanced. They could be used, for example, to track
1831	variable changes, implement your own watchers, integrate net-snmp or a	3055	variable changes, implement your own watchers, integrate net-snmp or a
1832	coroutine library and lots more. They are also occasionally useful if	3056	coroutine library and lots more. They are also occasionally useful if
1833	you cache some data and want to flush it before blocking (for example,	3057	you cache some data and want to flush it before blocking (for example,
1834	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	3058	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1835	watcher).	3059	watcher).
1836		3060
1837	This is done by examining in each prepare call which file descriptors need	3061	This is done by examining in each prepare call which file descriptors
1838	to be watched by the other library, registering C<ev_io> watchers for	3062	need to be watched by the other library, registering C<ev_io> watchers
1839	them and starting an C<ev_timer> watcher for any timeouts (many libraries	3063	for them and starting an C<ev_timer> watcher for any timeouts (many
1840	provide just this functionality). Then, in the check watcher you check for	3064	libraries provide exactly this functionality). Then, in the check watcher,
1841	any events that occured (by checking the pending status of all watchers	3065	you check for any events that occurred (by checking the pending status
1842	and stopping them) and call back into the library. The I/O and timer	3066	of all watchers and stopping them) and call back into the library. The
1843	callbacks will never actually be called (but must be valid nevertheless,	3067	I/O and timer callbacks will never actually be called (but must be valid
1844	because you never know, you know?).	3068	nevertheless, because you never know, you know?).
1845		3069
1846	As another example, the Perl Coro module uses these hooks to integrate	3070	As another example, the Perl Coro module uses these hooks to integrate
1847	coroutines into libev programs, by yielding to other active coroutines	3071	coroutines into libev programs, by yielding to other active coroutines
1848	during each prepare and only letting the process block if no coroutines	3072	during each prepare and only letting the process block if no coroutines
1849	are ready to run (it's actually more complicated: it only runs coroutines	3073	are ready to run (it's actually more complicated: it only runs coroutines
1850	with priority higher than or equal to the event loop and one coroutine	3074	with priority higher than or equal to the event loop and one coroutine
1851	of lower priority, but only once, using idle watchers to keep the event	3075	of lower priority, but only once, using idle watchers to keep the event
1852	loop from blocking if lower-priority coroutines are active, thus mapping	3076	loop from blocking if lower-priority coroutines are active, thus mapping
1853	low-priority coroutines to idle/background tasks).	3077	low-priority coroutines to idle/background tasks).
1854		3078
1855	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	3079	When used for this purpose, it is recommended to give C<ev_check> watchers
1856	priority, to ensure that they are being run before any other watchers	3080	highest (C<EV_MAXPRI>) priority, to ensure that they are being run before
		3081	any other watchers after the poll (this doesn't matter for C<ev_prepare>
		3082	watchers).
		3083
1857	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	3084	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
1858	too) should not activate ("feed") events into libev. While libev fully	3085	activate ("feed") events into libev. While libev fully supports this, they
1859	supports this, they will be called before other C<ev_check> watchers	3086	might get executed before other C<ev_check> watchers did their job. As
1860	did their job. As C<ev_check> watchers are often used to embed other	3087	C<ev_check> watchers are often used to embed other (non-libev) event
1861	(non-libev) event loops those other event loops might be in an unusable	3088	loops those other event loops might be in an unusable state until their
1862	state until their C<ev_check> watcher ran (always remind yourself to	3089	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
1863	coexist peacefully with others).	3090	others).
		3091
		3092	=head3 Abusing an C<ev_check> watcher for its side-effect
		3093
		3094	C<ev_check> (and less often also C<ev_prepare>) watchers can also be
		3095	useful because they are called once per event loop iteration. For
		3096	example, if you want to handle a large number of connections fairly, you
		3097	normally only do a bit of work for each active connection, and if there
		3098	is more work to do, you wait for the next event loop iteration, so other
		3099	connections have a chance of making progress.
		3100
		3101	Using an C<ev_check> watcher is almost enough: it will be called on the
		3102	next event loop iteration. However, that isn't as soon as possible -
		3103	without external events, your C<ev_check> watcher will not be invoked.
		3104
		3105	This is where C<ev_idle> watchers come in handy - all you need is a
		3106	single global idle watcher that is active as long as you have one active
		3107	C<ev_check> watcher. The C<ev_idle> watcher makes sure the event loop
		3108	will not sleep, and the C<ev_check> watcher makes sure a callback gets
		3109	invoked. Neither watcher alone can do that.
1864		3110
1865	=head3 Watcher-Specific Functions and Data Members	3111	=head3 Watcher-Specific Functions and Data Members
1866		3112
1867	=over 4	3113	=over 4
1868		3114
…		…
1870		3116
1871	=item ev_check_init (ev_check *, callback)	3117	=item ev_check_init (ev_check *, callback)
1872		3118
1873	Initialises and configures the prepare or check watcher - they have no	3119	Initialises and configures the prepare or check watcher - they have no
1874	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	3120	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1875	macros, but using them is utterly, utterly and completely pointless.	3121	macros, but using them is utterly, utterly, utterly and completely
		3122	pointless.
1876		3123
1877	=back	3124	=back
1878		3125
1879	=head3 Examples	3126	=head3 Examples
1880		3127
1881	There are a number of principal ways to embed other event loops or modules	3128	There are a number of principal ways to embed other event loops or modules
1882	into libev. Here are some ideas on how to include libadns into libev	3129	into libev. Here are some ideas on how to include libadns into libev
1883	(there is a Perl module named C<EV::ADNS> that does this, which you could	3130	(there is a Perl module named C<EV::ADNS> that does this, which you could
1884	use for an actually working example. Another Perl module named C<EV::Glib>	3131	use as a working example. Another Perl module named C<EV::Glib> embeds a
1885	embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV	3132	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
1886	into the Glib event loop).	3133	Glib event loop).
1887		3134
1888	Method 1: Add IO watchers and a timeout watcher in a prepare handler,	3135	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1889	and in a check watcher, destroy them and call into libadns. What follows	3136	and in a check watcher, destroy them and call into libadns. What follows
1890	is pseudo-code only of course. This requires you to either use a low	3137	is pseudo-code only of course. This requires you to either use a low
1891	priority for the check watcher or use C<ev_clear_pending> explicitly, as	3138	priority for the check watcher or use C<ev_clear_pending> explicitly, as
1892	the callbacks for the IO/timeout watchers might not have been called yet.	3139	the callbacks for the IO/timeout watchers might not have been called yet.
1893		3140
1894	static ev_io iow [nfd];	3141	static ev_io iow [nfd];
1895	static ev_timer tw;	3142	static ev_timer tw;
1896		3143
1897	static void	3144	static void
1898	io_cb (ev_loop *loop, ev_io *w, int revents)	3145	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1899	{	3146	{
1900	}	3147	}
1901		3148
1902	// create io watchers for each fd and a timer before blocking	3149	// create io watchers for each fd and a timer before blocking
1903	static void	3150	static void
1904	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	3151	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1905	{	3152	{
1906	int timeout = 3600000;	3153	int timeout = 3600000;
1907	struct pollfd fds [nfd];	3154	struct pollfd fds [nfd];
1908	// actual code will need to loop here and realloc etc.	3155	// actual code will need to loop here and realloc etc.
1909	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	3156	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1910		3157
1911	/* the callback is illegal, but won't be called as we stop during check */	3158	/* the callback is illegal, but won't be called as we stop during check */
1912	ev_timer_init (&tw, 0, timeout * 1e-3);	3159	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
1913	ev_timer_start (loop, &tw);	3160	ev_timer_start (loop, &tw);
1914		3161
1915	// create one ev_io per pollfd	3162	// create one ev_io per pollfd
1916	for (int i = 0; i < nfd; ++i)	3163	for (int i = 0; i < nfd; ++i)
1917	{	3164	{
1918	ev_io_init (iow + i, io_cb, fds [i].fd,	3165	ev_io_init (iow + i, io_cb, fds [i].fd,
1919	((fds [i].events & POLLIN ? EV_READ : 0)	3166	((fds [i].events & POLLIN ? EV_READ : 0)
1920	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	3167	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1921		3168
1922	fds [i].revents = 0;	3169	fds [i].revents = 0;
1923	ev_io_start (loop, iow + i);	3170	ev_io_start (loop, iow + i);
1924	}	3171	}
1925	}	3172	}
1926		3173
1927	// stop all watchers after blocking	3174	// stop all watchers after blocking
1928	static void	3175	static void
1929	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	3176	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1930	{	3177	{
1931	ev_timer_stop (loop, &tw);	3178	ev_timer_stop (loop, &tw);
1932		3179
1933	for (int i = 0; i < nfd; ++i)	3180	for (int i = 0; i < nfd; ++i)
1934	{	3181	{
1935	// set the relevant poll flags	3182	// set the relevant poll flags
1936	// could also call adns_processreadable etc. here	3183	// could also call adns_processreadable etc. here
1937	struct pollfd *fd = fds + i;	3184	struct pollfd *fd = fds + i;
1938	int revents = ev_clear_pending (iow + i);	3185	int revents = ev_clear_pending (iow + i);
1939	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;	3186	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1940	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;	3187	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1941		3188
1942	// now stop the watcher	3189	// now stop the watcher
1943	ev_io_stop (loop, iow + i);	3190	ev_io_stop (loop, iow + i);
1944	}	3191	}
1945		3192
1946	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	3193	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1947	}	3194	}
1948		3195
1949	Method 2: This would be just like method 1, but you run C<adns_afterpoll>	3196	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
1950	in the prepare watcher and would dispose of the check watcher.	3197	in the prepare watcher and would dispose of the check watcher.
1951		3198
1952	Method 3: If the module to be embedded supports explicit event	3199	Method 3: If the module to be embedded supports explicit event
1953	notification (adns does), you can also make use of the actual watcher	3200	notification (libadns does), you can also make use of the actual watcher
1954	callbacks, and only destroy/create the watchers in the prepare watcher.	3201	callbacks, and only destroy/create the watchers in the prepare watcher.
1955		3202
1956	static void	3203	static void
1957	timer_cb (EV_P_ ev_timer *w, int revents)	3204	timer_cb (EV_P_ ev_timer *w, int revents)
1958	{	3205	{
1959	adns_state ads = (adns_state)w->data;	3206	adns_state ads = (adns_state)w->data;
1960	update_now (EV_A);	3207	update_now (EV_A);
1961		3208
1962	adns_processtimeouts (ads, &tv_now);	3209	adns_processtimeouts (ads, &tv_now);
1963	}	3210	}
1964		3211
1965	static void	3212	static void
1966	io_cb (EV_P_ ev_io *w, int revents)	3213	io_cb (EV_P_ ev_io *w, int revents)
1967	{	3214	{
1968	adns_state ads = (adns_state)w->data;	3215	adns_state ads = (adns_state)w->data;
1969	update_now (EV_A);	3216	update_now (EV_A);
1970		3217
1971	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);	3218	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1972	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);	3219	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1973	}	3220	}
1974		3221
1975	// do not ever call adns_afterpoll	3222	// do not ever call adns_afterpoll
1976		3223
1977	Method 4: Do not use a prepare or check watcher because the module you	3224	Method 4: Do not use a prepare or check watcher because the module you
1978	want to embed is too inflexible to support it. Instead, youc na override	3225	want to embed is not flexible enough to support it. Instead, you can
1979	their poll function. The drawback with this solution is that the main	3226	override their poll function. The drawback with this solution is that the
1980	loop is now no longer controllable by EV. The C<Glib::EV> module does	3227	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
1981	this.	3228	this approach, effectively embedding EV as a client into the horrible
		3229	libglib event loop.
1982		3230
1983	static gint	3231	static gint
1984	event_poll_func (GPollFD *fds, guint nfds, gint timeout)	3232	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1985	{	3233	{
1986	int got_events = 0;	3234	int got_events = 0;
1987		3235
1988	for (n = 0; n < nfds; ++n)	3236	for (n = 0; n < nfds; ++n)
1989	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events	3237	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1990		3238
1991	if (timeout >= 0)	3239	if (timeout >= 0)
1992	// create/start timer	3240	// create/start timer
1993		3241
1994	// poll	3242	// poll
1995	ev_loop (EV_A_ 0);	3243	ev_run (EV_A_ 0);
1996		3244
1997	// stop timer again	3245	// stop timer again
1998	if (timeout >= 0)	3246	if (timeout >= 0)
1999	ev_timer_stop (EV_A_ &to);	3247	ev_timer_stop (EV_A_ &to);
2000		3248
2001	// stop io watchers again - their callbacks should have set	3249	// stop io watchers again - their callbacks should have set
2002	for (n = 0; n < nfds; ++n)	3250	for (n = 0; n < nfds; ++n)
2003	ev_io_stop (EV_A_ iow [n]);	3251	ev_io_stop (EV_A_ iow [n]);
2004		3252
2005	return got_events;	3253	return got_events;
2006	}	3254	}
2007		3255
2008		3256
2009	=head2 C<ev_embed> - when one backend isn't enough...	3257	=head2 C<ev_embed> - when one backend isn't enough...
2010		3258
2011	This is a rather advanced watcher type that lets you embed one event loop	3259	This is a rather advanced watcher type that lets you embed one event loop
…		…
2017	prioritise I/O.	3265	prioritise I/O.
2018		3266
2019	As an example for a bug workaround, the kqueue backend might only support	3267	As an example for a bug workaround, the kqueue backend might only support
2020	sockets on some platform, so it is unusable as generic backend, but you	3268	sockets on some platform, so it is unusable as generic backend, but you
2021	still want to make use of it because you have many sockets and it scales	3269	still want to make use of it because you have many sockets and it scales
2022	so nicely. In this case, you would create a kqueue-based loop and embed it	3270	so nicely. In this case, you would create a kqueue-based loop and embed
2023	into your default loop (which might use e.g. poll). Overall operation will	3271	it into your default loop (which might use e.g. poll). Overall operation
2024	be a bit slower because first libev has to poll and then call kevent, but	3272	will be a bit slower because first libev has to call C<poll> and then
2025	at least you can use both at what they are best.	3273	C<kevent>, but at least you can use both mechanisms for what they are
		3274	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
2026		3275
2027	As for prioritising I/O: rarely you have the case where some fds have	3276	As for prioritising I/O: under rare circumstances you have the case where
2028	to be watched and handled very quickly (with low latency), and even	3277	some fds have to be watched and handled very quickly (with low latency),
2029	priorities and idle watchers might have too much overhead. In this case	3278	and even priorities and idle watchers might have too much overhead. In
2030	you would put all the high priority stuff in one loop and all the rest in	3279	this case you would put all the high priority stuff in one loop and all
2031	a second one, and embed the second one in the first.	3280	the rest in a second one, and embed the second one in the first.
2032		3281
2033	As long as the watcher is active, the callback will be invoked every time	3282	As long as the watcher is active, the callback will be invoked every
2034	there might be events pending in the embedded loop. The callback must then	3283	time there might be events pending in the embedded loop. The callback
2035	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	3284	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
2036	their callbacks (you could also start an idle watcher to give the embedded	3285	sweep and invoke their callbacks (the callback doesn't need to invoke the
2037	loop strictly lower priority for example). You can also set the callback	3286	C<ev_embed_sweep> function directly, it could also start an idle watcher
2038	to C<0>, in which case the embed watcher will automatically execute the	3287	to give the embedded loop strictly lower priority for example).
2039	embedded loop sweep.
2040		3288
2041	As long as the watcher is started it will automatically handle events. The	3289	You can also set the callback to C<0>, in which case the embed watcher
2042	callback will be invoked whenever some events have been handled. You can	3290	will automatically execute the embedded loop sweep whenever necessary.
2043	set the callback to C<0> to avoid having to specify one if you are not
2044	interested in that.
2045		3291
2046	Also, there have not currently been made special provisions for forking:	3292	Fork detection will be handled transparently while the C<ev_embed> watcher
2047	when you fork, you not only have to call C<ev_loop_fork> on both loops,	3293	is active, i.e., the embedded loop will automatically be forked when the
2048	but you will also have to stop and restart any C<ev_embed> watchers	3294	embedding loop forks. In other cases, the user is responsible for calling
2049	yourself.	3295	C<ev_loop_fork> on the embedded loop.
2050		3296
2051	Unfortunately, not all backends are embeddable, only the ones returned by	3297	Unfortunately, not all backends are embeddable: only the ones returned by
2052	C<ev_embeddable_backends> are, which, unfortunately, does not include any	3298	C<ev_embeddable_backends> are, which, unfortunately, does not include any
2053	portable one.	3299	portable one.
2054		3300
2055	So when you want to use this feature you will always have to be prepared	3301	So when you want to use this feature you will always have to be prepared
2056	that you cannot get an embeddable loop. The recommended way to get around	3302	that you cannot get an embeddable loop. The recommended way to get around
2057	this is to have a separate variables for your embeddable loop, try to	3303	this is to have a separate variables for your embeddable loop, try to
2058	create it, and if that fails, use the normal loop for everything.	3304	create it, and if that fails, use the normal loop for everything.
2059		3305
		3306	=head3 C<ev_embed> and fork
		3307
		3308	While the C<ev_embed> watcher is running, forks in the embedding loop will
		3309	automatically be applied to the embedded loop as well, so no special
		3310	fork handling is required in that case. When the watcher is not running,
		3311	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		3312	as applicable.
		3313
2060	=head3 Watcher-Specific Functions and Data Members	3314	=head3 Watcher-Specific Functions and Data Members
2061		3315
2062	=over 4	3316	=over 4
2063		3317
2064	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	3318	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
2065		3319
2066	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)	3320	=item ev_embed_set (ev_embed *, struct ev_loop *embedded_loop)
2067		3321
2068	Configures the watcher to embed the given loop, which must be	3322	Configures the watcher to embed the given loop, which must be
2069	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	3323	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
2070	invoked automatically, otherwise it is the responsibility of the callback	3324	invoked automatically, otherwise it is the responsibility of the callback
2071	to invoke it (it will continue to be called until the sweep has been done,	3325	to invoke it (it will continue to be called until the sweep has been done,
2072	if you do not want thta, you need to temporarily stop the embed watcher).	3326	if you do not want that, you need to temporarily stop the embed watcher).
2073		3327
2074	=item ev_embed_sweep (loop, ev_embed *)	3328	=item ev_embed_sweep (loop, ev_embed *)
2075		3329
2076	Make a single, non-blocking sweep over the embedded loop. This works	3330	Make a single, non-blocking sweep over the embedded loop. This works
2077	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	3331	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
2078	apropriate way for embedded loops.	3332	appropriate way for embedded loops.
2079		3333
2080	=item struct ev_loop *other [read-only]	3334	=item struct ev_loop *other [read-only]
2081		3335
2082	The embedded event loop.	3336	The embedded event loop.
2083		3337
…		…
2085		3339
2086	=head3 Examples	3340	=head3 Examples
2087		3341
2088	Example: Try to get an embeddable event loop and embed it into the default	3342	Example: Try to get an embeddable event loop and embed it into the default
2089	event loop. If that is not possible, use the default loop. The default	3343	event loop. If that is not possible, use the default loop. The default
2090	loop is stored in C<loop_hi>, while the mebeddable loop is stored in	3344	loop is stored in C<loop_hi>, while the embeddable loop is stored in
2091	C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be	3345	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
2092	used).	3346	used).
2093		3347
2094	struct ev_loop *loop_hi = ev_default_init (0);	3348	struct ev_loop *loop_hi = ev_default_init (0);
2095	struct ev_loop *loop_lo = 0;	3349	struct ev_loop *loop_lo = 0;
2096	struct ev_embed embed;	3350	ev_embed embed;
2097		3351
2098	// see if there is a chance of getting one that works	3352	// see if there is a chance of getting one that works
2099	// (remember that a flags value of 0 means autodetection)	3353	// (remember that a flags value of 0 means autodetection)
2100	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()	3354	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2101	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())	3355	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2102	: 0;	3356	: 0;
2103		3357
2104	// if we got one, then embed it, otherwise default to loop_hi	3358	// if we got one, then embed it, otherwise default to loop_hi
2105	if (loop_lo)	3359	if (loop_lo)
2106	{	3360	{
2107	ev_embed_init (&embed, 0, loop_lo);	3361	ev_embed_init (&embed, 0, loop_lo);
2108	ev_embed_start (loop_hi, &embed);	3362	ev_embed_start (loop_hi, &embed);
2109	}	3363	}
2110	else	3364	else
2111	loop_lo = loop_hi;	3365	loop_lo = loop_hi;
2112		3366
2113	Example: Check if kqueue is available but not recommended and create	3367	Example: Check if kqueue is available but not recommended and create
2114	a kqueue backend for use with sockets (which usually work with any	3368	a kqueue backend for use with sockets (which usually work with any
2115	kqueue implementation). Store the kqueue/socket-only event loop in	3369	kqueue implementation). Store the kqueue/socket-only event loop in
2116	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).	3370	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
2117		3371
2118	struct ev_loop *loop = ev_default_init (0);	3372	struct ev_loop *loop = ev_default_init (0);
2119	struct ev_loop *loop_socket = 0;	3373	struct ev_loop *loop_socket = 0;
2120	struct ev_embed embed;	3374	ev_embed embed;
2121		3375
2122	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)	3376	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2123	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))	3377	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2124	{	3378	{
2125	ev_embed_init (&embed, 0, loop_socket);	3379	ev_embed_init (&embed, 0, loop_socket);
2126	ev_embed_start (loop, &embed);	3380	ev_embed_start (loop, &embed);
2127	}	3381	}
2128		3382
2129	if (!loop_socket)	3383	if (!loop_socket)
2130	loop_socket = loop;	3384	loop_socket = loop;
2131		3385
2132	// now use loop_socket for all sockets, and loop for everything else	3386	// now use loop_socket for all sockets, and loop for everything else
2133		3387
2134		3388
2135	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	3389	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
2136		3390
2137	Fork watchers are called when a C<fork ()> was detected (usually because	3391	Fork watchers are called when a C<fork ()> was detected (usually because
2138	whoever is a good citizen cared to tell libev about it by calling	3392	whoever is a good citizen cared to tell libev about it by calling
2139	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the	3393	C<ev_loop_fork>). The invocation is done before the event loop blocks next
2140	event loop blocks next and before C<ev_check> watchers are being called,	3394	and before C<ev_check> watchers are being called, and only in the child
2141	and only in the child after the fork. If whoever good citizen calling	3395	after the fork. If whoever good citizen calling C<ev_default_fork> cheats
2142	C<ev_default_fork> cheats and calls it in the wrong process, the fork	3396	and calls it in the wrong process, the fork handlers will be invoked, too,
2143	handlers will be invoked, too, of course.	3397	of course.
		3398
		3399	=head3 The special problem of life after fork - how is it possible?
		3400
		3401	Most uses of C<fork ()> consist of forking, then some simple calls to set
		3402	up/change the process environment, followed by a call to C<exec()>. This
		3403	sequence should be handled by libev without any problems.
		3404
		3405	This changes when the application actually wants to do event handling
		3406	in the child, or both parent in child, in effect "continuing" after the
		3407	fork.
		3408
		3409	The default mode of operation (for libev, with application help to detect
		3410	forks) is to duplicate all the state in the child, as would be expected
		3411	when I<either> the parent I<or> the child process continues.
		3412
		3413	When both processes want to continue using libev, then this is usually the
		3414	wrong result. In that case, usually one process (typically the parent) is
		3415	supposed to continue with all watchers in place as before, while the other
		3416	process typically wants to start fresh, i.e. without any active watchers.
		3417
		3418	The cleanest and most efficient way to achieve that with libev is to
		3419	simply create a new event loop, which of course will be "empty", and
		3420	use that for new watchers. This has the advantage of not touching more
		3421	memory than necessary, and thus avoiding the copy-on-write, and the
		3422	disadvantage of having to use multiple event loops (which do not support
		3423	signal watchers).
		3424
		3425	When this is not possible, or you want to use the default loop for
		3426	other reasons, then in the process that wants to start "fresh", call
		3427	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3428	Destroying the default loop will "orphan" (not stop) all registered
		3429	watchers, so you have to be careful not to execute code that modifies
		3430	those watchers. Note also that in that case, you have to re-register any
		3431	signal watchers.
2144		3432
2145	=head3 Watcher-Specific Functions and Data Members	3433	=head3 Watcher-Specific Functions and Data Members
2146		3434
2147	=over 4	3435	=over 4
2148		3436
2149	=item ev_fork_init (ev_signal *, callback)	3437	=item ev_fork_init (ev_fork *, callback)
2150		3438
2151	Initialises and configures the fork watcher - it has no parameters of any	3439	Initialises and configures the fork watcher - it has no parameters of any
2152	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,	3440	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
2153	believe me.	3441	really.
2154		3442
2155	=back	3443	=back
2156		3444
2157		3445
		3446	=head2 C<ev_cleanup> - even the best things end
		3447
		3448	Cleanup watchers are called just before the event loop is being destroyed
		3449	by a call to C<ev_loop_destroy>.
		3450
		3451	While there is no guarantee that the event loop gets destroyed, cleanup
		3452	watchers provide a convenient method to install cleanup hooks for your
		3453	program, worker threads and so on - you just to make sure to destroy the
		3454	loop when you want them to be invoked.
		3455
		3456	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3457	all other watchers, they do not keep a reference to the event loop (which
		3458	makes a lot of sense if you think about it). Like all other watchers, you
		3459	can call libev functions in the callback, except C<ev_cleanup_start>.
		3460
		3461	=head3 Watcher-Specific Functions and Data Members
		3462
		3463	=over 4
		3464
		3465	=item ev_cleanup_init (ev_cleanup *, callback)
		3466
		3467	Initialises and configures the cleanup watcher - it has no parameters of
		3468	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3469	pointless, I assure you.
		3470
		3471	=back
		3472
		3473	Example: Register an atexit handler to destroy the default loop, so any
		3474	cleanup functions are called.
		3475
		3476	static void
		3477	program_exits (void)
		3478	{
		3479	ev_loop_destroy (EV_DEFAULT_UC);
		3480	}
		3481
		3482	...
		3483	atexit (program_exits);
		3484
		3485
2158	=head2 C<ev_async> - how to wake up another event loop	3486	=head2 C<ev_async> - how to wake up an event loop
2159		3487
2160	In general, you cannot use an C<ev_loop> from multiple threads or other	3488	In general, you cannot use an C<ev_loop> from multiple threads or other
2161	asynchronous sources such as signal handlers (as opposed to multiple event	3489	asynchronous sources such as signal handlers (as opposed to multiple event
2162	loops - those are of course safe to use in different threads).	3490	loops - those are of course safe to use in different threads).
2163		3491
2164	Sometimes, however, you need to wake up another event loop you do not	3492	Sometimes, however, you need to wake up an event loop you do not control,
2165	control, for example because it belongs to another thread. This is what	3493	for example because it belongs to another thread. This is what C<ev_async>
2166	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you	3494	watchers do: as long as the C<ev_async> watcher is active, you can signal
2167	can signal it by calling C<ev_async_send>, which is thread- and signal	3495	it by calling C<ev_async_send>, which is thread- and signal safe.
2168	safe.
2169		3496
2170	This functionality is very similar to C<ev_signal> watchers, as signals,	3497	This functionality is very similar to C<ev_signal> watchers, as signals,
2171	too, are asynchronous in nature, and signals, too, will be compressed	3498	too, are asynchronous in nature, and signals, too, will be compressed
2172	(i.e. the number of callback invocations may be less than the number of	3499	(i.e. the number of callback invocations may be less than the number of
2173	C<ev_async_sent> calls).	3500	C<ev_async_send> calls). In fact, you could use signal watchers as a kind
2174		3501	of "global async watchers" by using a watcher on an otherwise unused
2175	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not	3502	signal, and C<ev_feed_signal> to signal this watcher from another thread,
2176	just the default loop.	3503	even without knowing which loop owns the signal.
2177		3504
2178	=head3 Queueing	3505	=head3 Queueing
2179		3506
2180	C<ev_async> does not support queueing of data in any way. The reason	3507	C<ev_async> does not support queueing of data in any way. The reason
2181	is that the author does not know of a simple (or any) algorithm for a	3508	is that the author does not know of a simple (or any) algorithm for a
2182	multiple-writer-single-reader queue that works in all cases and doesn't	3509	multiple-writer-single-reader queue that works in all cases and doesn't
2183	need elaborate support such as pthreads.	3510	need elaborate support such as pthreads or unportable memory access
		3511	semantics.
2184		3512
2185	That means that if you want to queue data, you have to provide your own	3513	That means that if you want to queue data, you have to provide your own
2186	queue. But at least I can tell you would implement locking around your	3514	queue. But at least I can tell you how to implement locking around your
2187	queue:	3515	queue:
2188		3516
2189	=over 4	3517	=over 4
2190		3518
2191	=item queueing from a signal handler context	3519	=item queueing from a signal handler context
2192		3520
2193	To implement race-free queueing, you simply add to the queue in the signal	3521	To implement race-free queueing, you simply add to the queue in the signal
2194	handler but you block the signal handler in the watcher callback. Here is an example that does that for	3522	handler but you block the signal handler in the watcher callback. Here is
2195	some fictitiuous SIGUSR1 handler:	3523	an example that does that for some fictitious SIGUSR1 handler:
2196		3524
2197	static ev_async mysig;	3525	static ev_async mysig;
2198		3526
2199	static void	3527	static void
2200	sigusr1_handler (void)	3528	sigusr1_handler (void)
…		…
2266	=over 4	3594	=over 4
2267		3595
2268	=item ev_async_init (ev_async *, callback)	3596	=item ev_async_init (ev_async *, callback)
2269		3597
2270	Initialises and configures the async watcher - it has no parameters of any	3598	Initialises and configures the async watcher - it has no parameters of any
2271	kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,	3599	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
2272	believe me.	3600	trust me.
2273		3601
2274	=item ev_async_send (loop, ev_async *)	3602	=item ev_async_send (loop, ev_async *)
2275		3603
2276	Sends/signals/activates the given C<ev_async> watcher, that is, feeds	3604	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
2277	an C<EV_ASYNC> event on the watcher into the event loop. Unlike	3605	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3606	returns.
		3607
2278	C<ev_feed_event>, this call is safe to do in other threads, signal or	3608	Unlike C<ev_feed_event>, this call is safe to do from other threads,
2279	similar contexts (see the dicusssion of C<EV_ATOMIC_T> in the embedding	3609	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
2280	section below on what exactly this means).	3610	embedding section below on what exactly this means).
2281		3611
2282	This call incurs the overhead of a syscall only once per loop iteration,	3612	Note that, as with other watchers in libev, multiple events might get
2283	so while the overhead might be noticable, it doesn't apply to repeated	3613	compressed into a single callback invocation (another way to look at
2284	calls to C<ev_async_send>.	3614	this is that C<ev_async> watchers are level-triggered: they are set on
		3615	C<ev_async_send>, reset when the event loop detects that).
		3616
		3617	This call incurs the overhead of at most one extra system call per event
		3618	loop iteration, if the event loop is blocked, and no syscall at all if
		3619	the event loop (or your program) is processing events. That means that
		3620	repeated calls are basically free (there is no need to avoid calls for
		3621	performance reasons) and that the overhead becomes smaller (typically
		3622	zero) under load.
2285		3623
2286	=item bool = ev_async_pending (ev_async *)	3624	=item bool = ev_async_pending (ev_async *)
2287		3625
2288	Returns a non-zero value when C<ev_async_send> has been called on the	3626	Returns a non-zero value when C<ev_async_send> has been called on the
2289	watcher but the event has not yet been processed (or even noted) by the	3627	watcher but the event has not yet been processed (or even noted) by the
2290	event loop.	3628	event loop.
2291		3629
2292	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When	3630	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
2293	the loop iterates next and checks for the watcher to have become active,	3631	the loop iterates next and checks for the watcher to have become active,
2294	it will reset the flag again. C<ev_async_pending> can be used to very	3632	it will reset the flag again. C<ev_async_pending> can be used to very
2295	quickly check wether invoking the loop might be a good idea.	3633	quickly check whether invoking the loop might be a good idea.
2296		3634
2297	Not that this does I<not> check wether the watcher itself is pending, only	3635	Not that this does I<not> check whether the watcher itself is pending,
2298	wether it has been requested to make this watcher pending.	3636	only whether it has been requested to make this watcher pending: there
		3637	is a time window between the event loop checking and resetting the async
		3638	notification, and the callback being invoked.
2299		3639
2300	=back	3640	=back
2301		3641
2302		3642
2303	=head1 OTHER FUNCTIONS	3643	=head1 OTHER FUNCTIONS
2304		3644
2305	There are some other functions of possible interest. Described. Here. Now.	3645	There are some other functions of possible interest. Described. Here. Now.
2306		3646
2307	=over 4	3647	=over 4
2308		3648
2309	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3649	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback, arg)
2310		3650
2311	This function combines a simple timer and an I/O watcher, calls your	3651	This function combines a simple timer and an I/O watcher, calls your
2312	callback on whichever event happens first and automatically stop both	3652	callback on whichever event happens first and automatically stops both
2313	watchers. This is useful if you want to wait for a single event on an fd	3653	watchers. This is useful if you want to wait for a single event on an fd
2314	or timeout without having to allocate/configure/start/stop/free one or	3654	or timeout without having to allocate/configure/start/stop/free one or
2315	more watchers yourself.	3655	more watchers yourself.
2316		3656
2317	If C<fd> is less than 0, then no I/O watcher will be started and events	3657	If C<fd> is less than 0, then no I/O watcher will be started and the
2318	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3658	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
2319	C<events> set will be craeted and started.	3659	the given C<fd> and C<events> set will be created and started.
2320		3660
2321	If C<timeout> is less than 0, then no timeout watcher will be	3661	If C<timeout> is less than 0, then no timeout watcher will be
2322	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3662	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
2323	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3663	repeat = 0) will be started. C<0> is a valid timeout.
2324	dubious value.
2325		3664
2326	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3665	The callback has the type C<void (*cb)(int revents, void *arg)> and is
2327	passed an C<revents> set like normal event callbacks (a combination of	3666	passed an C<revents> set like normal event callbacks (a combination of
2328	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3667	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
2329	value passed to C<ev_once>:	3668	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3669	a timeout and an io event at the same time - you probably should give io
		3670	events precedence.
2330		3671
		3672	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3673
2331	static void stdin_ready (int revents, void *arg)	3674	static void stdin_ready (int revents, void *arg)
		3675	{
		3676	if (revents & EV_READ)
		3677	/* stdin might have data for us, joy! */;
		3678	else if (revents & EV_TIMER)
		3679	/* doh, nothing entered */;
		3680	}
		3681
		3682	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
		3683
		3684	=item ev_feed_fd_event (loop, int fd, int revents)
		3685
		3686	Feed an event on the given fd, as if a file descriptor backend detected
		3687	the given events.
		3688
		3689	=item ev_feed_signal_event (loop, int signum)
		3690
		3691	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
		3692	which is async-safe.
		3693
		3694	=back
		3695
		3696
		3697	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3698
		3699	This section explains some common idioms that are not immediately
		3700	obvious. Note that examples are sprinkled over the whole manual, and this
		3701	section only contains stuff that wouldn't fit anywhere else.
		3702
		3703	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3704
		3705	Each watcher has, by default, a C<void *data> member that you can read
		3706	or modify at any time: libev will completely ignore it. This can be used
		3707	to associate arbitrary data with your watcher. If you need more data and
		3708	don't want to allocate memory separately and store a pointer to it in that
		3709	data member, you can also "subclass" the watcher type and provide your own
		3710	data:
		3711
		3712	struct my_io
		3713	{
		3714	ev_io io;
		3715	int otherfd;
		3716	void *somedata;
		3717	struct whatever *mostinteresting;
		3718	};
		3719
		3720	...
		3721	struct my_io w;
		3722	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3723
		3724	And since your callback will be called with a pointer to the watcher, you
		3725	can cast it back to your own type:
		3726
		3727	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3728	{
		3729	struct my_io *w = (struct my_io *)w_;
		3730	...
		3731	}
		3732
		3733	More interesting and less C-conformant ways of casting your callback
		3734	function type instead have been omitted.
		3735
		3736	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3737
		3738	Another common scenario is to use some data structure with multiple
		3739	embedded watchers, in effect creating your own watcher that combines
		3740	multiple libev event sources into one "super-watcher":
		3741
		3742	struct my_biggy
		3743	{
		3744	int some_data;
		3745	ev_timer t1;
		3746	ev_timer t2;
		3747	}
		3748
		3749	In this case getting the pointer to C<my_biggy> is a bit more
		3750	complicated: Either you store the address of your C<my_biggy> struct in
		3751	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3752	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3753	real programmers):
		3754
		3755	#include <stddef.h>
		3756
		3757	static void
		3758	t1_cb (EV_P_ ev_timer *w, int revents)
		3759	{
		3760	struct my_biggy big = (struct my_biggy *)
		3761	(((char *)w) - offsetof (struct my_biggy, t1));
		3762	}
		3763
		3764	static void
		3765	t2_cb (EV_P_ ev_timer *w, int revents)
		3766	{
		3767	struct my_biggy big = (struct my_biggy *)
		3768	(((char *)w) - offsetof (struct my_biggy, t2));
		3769	}
		3770
		3771	=head2 AVOIDING FINISHING BEFORE RETURNING
		3772
		3773	Often you have structures like this in event-based programs:
		3774
		3775	callback ()
2332	{	3776	{
2333	if (revents & EV_TIMEOUT)	3777	free (request);
2334	/* doh, nothing entered */;
2335	else if (revents & EV_READ)
2336	/* stdin might have data for us, joy! */;
2337	}	3778	}
2338		3779
2339	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3780	request = start_new_request (..., callback);
2340		3781
2341	=item ev_feed_event (ev_loop *, watcher *, int revents)	3782	The intent is to start some "lengthy" operation. The C<request> could be
		3783	used to cancel the operation, or do other things with it.
2342		3784
2343	Feeds the given event set into the event loop, as if the specified event	3785	It's not uncommon to have code paths in C<start_new_request> that
2344	had happened for the specified watcher (which must be a pointer to an	3786	immediately invoke the callback, for example, to report errors. Or you add
2345	initialised but not necessarily started event watcher).	3787	some caching layer that finds that it can skip the lengthy aspects of the
		3788	operation and simply invoke the callback with the result.
2346		3789
2347	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	3790	The problem here is that this will happen I<before> C<start_new_request>
		3791	has returned, so C<request> is not set.
2348		3792
2349	Feed an event on the given fd, as if a file descriptor backend detected	3793	Even if you pass the request by some safer means to the callback, you
2350	the given events it.	3794	might want to do something to the request after starting it, such as
		3795	canceling it, which probably isn't working so well when the callback has
		3796	already been invoked.
2351		3797
2352	=item ev_feed_signal_event (ev_loop *loop, int signum)	3798	A common way around all these issues is to make sure that
		3799	C<start_new_request> I<always> returns before the callback is invoked. If
		3800	C<start_new_request> immediately knows the result, it can artificially
		3801	delay invoking the callback by using a C<prepare> or C<idle> watcher for
		3802	example, or more sneakily, by reusing an existing (stopped) watcher and
		3803	pushing it into the pending queue:
2353		3804
2354	Feed an event as if the given signal occured (C<loop> must be the default	3805	ev_set_cb (watcher, callback);
2355	loop!).	3806	ev_feed_event (EV_A_ watcher, 0);
2356		3807
2357	=back	3808	This way, C<start_new_request> can safely return before the callback is
		3809	invoked, while not delaying callback invocation too much.
		3810
		3811	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3812
		3813	Often (especially in GUI toolkits) there are places where you have
		3814	I<modal> interaction, which is most easily implemented by recursively
		3815	invoking C<ev_run>.
		3816
		3817	This brings the problem of exiting - a callback might want to finish the
		3818	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3819	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3820	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3821	other combination: In these cases, a simple C<ev_break> will not work.
		3822
		3823	The solution is to maintain "break this loop" variable for each C<ev_run>
		3824	invocation, and use a loop around C<ev_run> until the condition is
		3825	triggered, using C<EVRUN_ONCE>:
		3826
		3827	// main loop
		3828	int exit_main_loop = 0;
		3829
		3830	while (!exit_main_loop)
		3831	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3832
		3833	// in a modal watcher
		3834	int exit_nested_loop = 0;
		3835
		3836	while (!exit_nested_loop)
		3837	ev_run (EV_A_ EVRUN_ONCE);
		3838
		3839	To exit from any of these loops, just set the corresponding exit variable:
		3840
		3841	// exit modal loop
		3842	exit_nested_loop = 1;
		3843
		3844	// exit main program, after modal loop is finished
		3845	exit_main_loop = 1;
		3846
		3847	// exit both
		3848	exit_main_loop = exit_nested_loop = 1;
		3849
		3850	=head2 THREAD LOCKING EXAMPLE
		3851
		3852	Here is a fictitious example of how to run an event loop in a different
		3853	thread from where callbacks are being invoked and watchers are
		3854	created/added/removed.
		3855
		3856	For a real-world example, see the C<EV::Loop::Async> perl module,
		3857	which uses exactly this technique (which is suited for many high-level
		3858	languages).
		3859
		3860	The example uses a pthread mutex to protect the loop data, a condition
		3861	variable to wait for callback invocations, an async watcher to notify the
		3862	event loop thread and an unspecified mechanism to wake up the main thread.
		3863
		3864	First, you need to associate some data with the event loop:
		3865
		3866	typedef struct {
		3867	pthread_mutex_t lock; /* global loop lock */
		3868	pthread_t tid;
		3869	pthread_cond_t invoke_cv;
		3870	ev_async async_w;
		3871	} userdata;
		3872
		3873	void prepare_loop (EV_P)
		3874	{
		3875	// for simplicity, we use a static userdata struct.
		3876	static userdata u;
		3877
		3878	ev_async_init (&u.async_w, async_cb);
		3879	ev_async_start (EV_A_ &u.async_w);
		3880
		3881	pthread_mutex_init (&u.lock, 0);
		3882	pthread_cond_init (&u.invoke_cv, 0);
		3883
		3884	// now associate this with the loop
		3885	ev_set_userdata (EV_A_ &u);
		3886	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3887	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3888
		3889	// then create the thread running ev_run
		3890	pthread_create (&u.tid, 0, l_run, EV_A);
		3891	}
		3892
		3893	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3894	solely to wake up the event loop so it takes notice of any new watchers
		3895	that might have been added:
		3896
		3897	static void
		3898	async_cb (EV_P_ ev_async *w, int revents)
		3899	{
		3900	// just used for the side effects
		3901	}
		3902
		3903	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3904	protecting the loop data, respectively.
		3905
		3906	static void
		3907	l_release (EV_P)
		3908	{
		3909	userdata *u = ev_userdata (EV_A);
		3910	pthread_mutex_unlock (&u->lock);
		3911	}
		3912
		3913	static void
		3914	l_acquire (EV_P)
		3915	{
		3916	userdata *u = ev_userdata (EV_A);
		3917	pthread_mutex_lock (&u->lock);
		3918	}
		3919
		3920	The event loop thread first acquires the mutex, and then jumps straight
		3921	into C<ev_run>:
		3922
		3923	void *
		3924	l_run (void *thr_arg)
		3925	{
		3926	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3927
		3928	l_acquire (EV_A);
		3929	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3930	ev_run (EV_A_ 0);
		3931	l_release (EV_A);
		3932
		3933	return 0;
		3934	}
		3935
		3936	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3937	signal the main thread via some unspecified mechanism (signals? pipe
		3938	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3939	have been called (in a while loop because a) spurious wakeups are possible
		3940	and b) skipping inter-thread-communication when there are no pending
		3941	watchers is very beneficial):
		3942
		3943	static void
		3944	l_invoke (EV_P)
		3945	{
		3946	userdata *u = ev_userdata (EV_A);
		3947
		3948	while (ev_pending_count (EV_A))
		3949	{
		3950	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3951	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3952	}
		3953	}
		3954
		3955	Now, whenever the main thread gets told to invoke pending watchers, it
		3956	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3957	thread to continue:
		3958
		3959	static void
		3960	real_invoke_pending (EV_P)
		3961	{
		3962	userdata *u = ev_userdata (EV_A);
		3963
		3964	pthread_mutex_lock (&u->lock);
		3965	ev_invoke_pending (EV_A);
		3966	pthread_cond_signal (&u->invoke_cv);
		3967	pthread_mutex_unlock (&u->lock);
		3968	}
		3969
		3970	Whenever you want to start/stop a watcher or do other modifications to an
		3971	event loop, you will now have to lock:
		3972
		3973	ev_timer timeout_watcher;
		3974	userdata *u = ev_userdata (EV_A);
		3975
		3976	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3977
		3978	pthread_mutex_lock (&u->lock);
		3979	ev_timer_start (EV_A_ &timeout_watcher);
		3980	ev_async_send (EV_A_ &u->async_w);
		3981	pthread_mutex_unlock (&u->lock);
		3982
		3983	Note that sending the C<ev_async> watcher is required because otherwise
		3984	an event loop currently blocking in the kernel will have no knowledge
		3985	about the newly added timer. By waking up the loop it will pick up any new
		3986	watchers in the next event loop iteration.
		3987
		3988	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3989
		3990	While the overhead of a callback that e.g. schedules a thread is small, it
		3991	is still an overhead. If you embed libev, and your main usage is with some
		3992	kind of threads or coroutines, you might want to customise libev so that
		3993	doesn't need callbacks anymore.
		3994
		3995	Imagine you have coroutines that you can switch to using a function
		3996	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3997	and that due to some magic, the currently active coroutine is stored in a
		3998	global called C<current_coro>. Then you can build your own "wait for libev
		3999	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		4000	the differing C<;> conventions):
		4001
		4002	#define EV_CB_DECLARE(type) struct my_coro *cb;
		4003	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		4004
		4005	That means instead of having a C callback function, you store the
		4006	coroutine to switch to in each watcher, and instead of having libev call
		4007	your callback, you instead have it switch to that coroutine.
		4008
		4009	A coroutine might now wait for an event with a function called
		4010	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		4011	matter when, or whether the watcher is active or not when this function is
		4012	called):
		4013
		4014	void
		4015	wait_for_event (ev_watcher *w)
		4016	{
		4017	ev_set_cb (w, current_coro);
		4018	switch_to (libev_coro);
		4019	}
		4020
		4021	That basically suspends the coroutine inside C<wait_for_event> and
		4022	continues the libev coroutine, which, when appropriate, switches back to
		4023	this or any other coroutine.
		4024
		4025	You can do similar tricks if you have, say, threads with an event queue -
		4026	instead of storing a coroutine, you store the queue object and instead of
		4027	switching to a coroutine, you push the watcher onto the queue and notify
		4028	any waiters.
		4029
		4030	To embed libev, see L</EMBEDDING>, but in short, it's easiest to create two
		4031	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		4032
		4033	// my_ev.h
		4034	#define EV_CB_DECLARE(type) struct my_coro *cb;
		4035	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		4036	#include "../libev/ev.h"
		4037
		4038	// my_ev.c
		4039	#define EV_H "my_ev.h"
		4040	#include "../libev/ev.c"
		4041
		4042	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		4043	F<my_ev.c> into your project. When properly specifying include paths, you
		4044	can even use F<ev.h> as header file name directly.
2358		4045
2359		4046
2360	=head1 LIBEVENT EMULATION	4047	=head1 LIBEVENT EMULATION
2361		4048
2362	Libev offers a compatibility emulation layer for libevent. It cannot	4049	Libev offers a compatibility emulation layer for libevent. It cannot
2363	emulate the internals of libevent, so here are some usage hints:	4050	emulate the internals of libevent, so here are some usage hints:
2364		4051
2365	=over 4	4052	=over 4
		4053
		4054	=item * Only the libevent-1.4.1-beta API is being emulated.
		4055
		4056	This was the newest libevent version available when libev was implemented,
		4057	and is still mostly unchanged in 2010.
2366		4058
2367	=item * Use it by including <event.h>, as usual.	4059	=item * Use it by including <event.h>, as usual.
2368		4060
2369	=item * The following members are fully supported: ev_base, ev_callback,	4061	=item * The following members are fully supported: ev_base, ev_callback,
2370	ev_arg, ev_fd, ev_res, ev_events.	4062	ev_arg, ev_fd, ev_res, ev_events.
…		…
2375		4067
2376	=item * Priorities are not currently supported. Initialising priorities	4068	=item * Priorities are not currently supported. Initialising priorities
2377	will fail and all watchers will have the same priority, even though there	4069	will fail and all watchers will have the same priority, even though there
2378	is an ev_pri field.	4070	is an ev_pri field.
2379		4071
		4072	=item * In libevent, the last base created gets the signals, in libev, the
		4073	base that registered the signal gets the signals.
		4074
2380	=item * Other members are not supported.	4075	=item * Other members are not supported.
2381		4076
2382	=item * The libev emulation is I<not> ABI compatible to libevent, you need	4077	=item * The libev emulation is I<not> ABI compatible to libevent, you need
2383	to use the libev header file and library.	4078	to use the libev header file and library.
2384		4079
2385	=back	4080	=back
2386		4081
2387	=head1 C++ SUPPORT	4082	=head1 C++ SUPPORT
2388		4083
		4084	=head2 C API
		4085
		4086	The normal C API should work fine when used from C++: both ev.h and the
		4087	libev sources can be compiled as C++. Therefore, code that uses the C API
		4088	will work fine.
		4089
		4090	Proper exception specifications might have to be added to callbacks passed
		4091	to libev: exceptions may be thrown only from watcher callbacks, all other
		4092	callbacks (allocator, syserr, loop acquire/release and periodic reschedule
		4093	callbacks) must not throw exceptions, and might need a C<noexcept>
		4094	specification. If you have code that needs to be compiled as both C and
		4095	C++ you can use the C<EV_NOEXCEPT> macro for this:
		4096
		4097	static void
		4098	fatal_error (const char *msg) EV_NOEXCEPT
		4099	{
		4100	perror (msg);
		4101	abort ();
		4102	}
		4103
		4104	...
		4105	ev_set_syserr_cb (fatal_error);
		4106
		4107	The only API functions that can currently throw exceptions are C<ev_run>,
		4108	C<ev_invoke>, C<ev_invoke_pending> and C<ev_loop_destroy> (the latter
		4109	because it runs cleanup watchers).
		4110
		4111	Throwing exceptions in watcher callbacks is only supported if libev itself
		4112	is compiled with a C++ compiler or your C and C++ environments allow
		4113	throwing exceptions through C libraries (most do).
		4114
		4115	=head2 C++ API
		4116
2389	Libev comes with some simplistic wrapper classes for C++ that mainly allow	4117	Libev comes with some simplistic wrapper classes for C++ that mainly allow
2390	you to use some convinience methods to start/stop watchers and also change	4118	you to use some convenience methods to start/stop watchers and also change
2391	the callback model to a model using method callbacks on objects.	4119	the callback model to a model using method callbacks on objects.
2392		4120
2393	To use it,	4121	To use it,
2394		4122
2395	#include <ev++.h>	4123	#include <ev++.h>
2396		4124
2397	This automatically includes F<ev.h> and puts all of its definitions (many	4125	This automatically includes F<ev.h> and puts all of its definitions (many
2398	of them macros) into the global namespace. All C++ specific things are	4126	of them macros) into the global namespace. All C++ specific things are
2399	put into the C<ev> namespace. It should support all the same embedding	4127	put into the C<ev> namespace. It should support all the same embedding
2400	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.	4128	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
…		…
2402	Care has been taken to keep the overhead low. The only data member the C++	4130	Care has been taken to keep the overhead low. The only data member the C++
2403	classes add (compared to plain C-style watchers) is the event loop pointer	4131	classes add (compared to plain C-style watchers) is the event loop pointer
2404	that the watcher is associated with (or no additional members at all if	4132	that the watcher is associated with (or no additional members at all if
2405	you disable C<EV_MULTIPLICITY> when embedding libev).	4133	you disable C<EV_MULTIPLICITY> when embedding libev).
2406		4134
2407	Currently, functions, and static and non-static member functions can be	4135	Currently, functions, static and non-static member functions and classes
2408	used as callbacks. Other types should be easy to add as long as they only	4136	with C<operator ()> can be used as callbacks. Other types should be easy
2409	need one additional pointer for context. If you need support for other	4137	to add as long as they only need one additional pointer for context. If
2410	types of functors please contact the author (preferably after implementing	4138	you need support for other types of functors please contact the author
2411	it).	4139	(preferably after implementing it).
		4140
		4141	For all this to work, your C++ compiler either has to use the same calling
		4142	conventions as your C compiler (for static member functions), or you have
		4143	to embed libev and compile libev itself as C++.
2412		4144
2413	Here is a list of things available in the C<ev> namespace:	4145	Here is a list of things available in the C<ev> namespace:
2414		4146
2415	=over 4	4147	=over 4
2416		4148
…		…
2426	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.	4158	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
2427		4159
2428	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of	4160	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
2429	the same name in the C<ev> namespace, with the exception of C<ev_signal>	4161	the same name in the C<ev> namespace, with the exception of C<ev_signal>
2430	which is called C<ev::sig> to avoid clashes with the C<signal> macro	4162	which is called C<ev::sig> to avoid clashes with the C<signal> macro
2431	defines by many implementations.	4163	defined by many implementations.
2432		4164
2433	All of those classes have these methods:	4165	All of those classes have these methods:
2434		4166
2435	=over 4	4167	=over 4
2436		4168
2437	=item ev::TYPE::TYPE ()	4169	=item ev::TYPE::TYPE ()
2438		4170
2439	=item ev::TYPE::TYPE (struct ev_loop *)	4171	=item ev::TYPE::TYPE (loop)
2440		4172
2441	=item ev::TYPE::~TYPE	4173	=item ev::TYPE::~TYPE
2442		4174
2443	The constructor (optionally) takes an event loop to associate the watcher	4175	The constructor (optionally) takes an event loop to associate the watcher
2444	with. If it is omitted, it will use C<EV_DEFAULT>.	4176	with. If it is omitted, it will use C<EV_DEFAULT>.
…		…
2467	your compiler is good :), then the method will be fully inlined into the	4199	your compiler is good :), then the method will be fully inlined into the
2468	thunking function, making it as fast as a direct C callback.	4200	thunking function, making it as fast as a direct C callback.
2469		4201
2470	Example: simple class declaration and watcher initialisation	4202	Example: simple class declaration and watcher initialisation
2471		4203
2472	struct myclass	4204	struct myclass
2473	{	4205	{
2474	void io_cb (ev::io &w, int revents) { }	4206	void io_cb (ev::io &w, int revents) { }
2475	}	4207	}
2476		4208
2477	myclass obj;	4209	myclass obj;
2478	ev::io iow;	4210	ev::io iow;
2479	iow.set <myclass, &myclass::io_cb> (&obj);	4211	iow.set <myclass, &myclass::io_cb> (&obj);
		4212
		4213	=item w->set (object *)
		4214
		4215	This is a variation of a method callback - leaving out the method to call
		4216	will default the method to C<operator ()>, which makes it possible to use
		4217	functor objects without having to manually specify the C<operator ()> all
		4218	the time. Incidentally, you can then also leave out the template argument
		4219	list.
		4220
		4221	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		4222	int revents)>.
		4223
		4224	See the method-C<set> above for more details.
		4225
		4226	Example: use a functor object as callback.
		4227
		4228	struct myfunctor
		4229	{
		4230	void operator() (ev::io &w, int revents)
		4231	{
		4232	...
		4233	}
		4234	}
		4235
		4236	myfunctor f;
		4237
		4238	ev::io w;
		4239	w.set (&f);
2480		4240
2481	=item w->set<function> (void *data = 0)	4241	=item w->set<function> (void *data = 0)
2482		4242
2483	Also sets a callback, but uses a static method or plain function as	4243	Also sets a callback, but uses a static method or plain function as
2484	callback. The optional C<data> argument will be stored in the watcher's	4244	callback. The optional C<data> argument will be stored in the watcher's
…		…
2486		4246
2487	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.	4247	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
2488		4248
2489	See the method-C<set> above for more details.	4249	See the method-C<set> above for more details.
2490		4250
2491	Example:	4251	Example: Use a plain function as callback.
2492		4252
2493	static void io_cb (ev::io &w, int revents) { }	4253	static void io_cb (ev::io &w, int revents) { }
2494	iow.set <io_cb> ();	4254	iow.set <io_cb> ();
2495		4255
2496	=item w->set (struct ev_loop *)	4256	=item w->set (loop)
2497		4257
2498	Associates a different C<struct ev_loop> with this watcher. You can only	4258	Associates a different C<struct ev_loop> with this watcher. You can only
2499	do this when the watcher is inactive (and not pending either).	4259	do this when the watcher is inactive (and not pending either).
2500		4260
2501	=item w->set ([args])	4261	=item w->set ([arguments])
2502		4262
2503	Basically the same as C<ev_TYPE_set>, with the same args. Must be	4263	Basically the same as C<ev_TYPE_set> (except for C<ev::embed> watchers>),
		4264	with the same arguments. Either this method or a suitable start method
2504	called at least once. Unlike the C counterpart, an active watcher gets	4265	must be called at least once. Unlike the C counterpart, an active watcher
2505	automatically stopped and restarted when reconfiguring it with this	4266	gets automatically stopped and restarted when reconfiguring it with this
2506	method.	4267	method.
		4268
		4269	For C<ev::embed> watchers this method is called C<set_embed>, to avoid
		4270	clashing with the C<set (loop)> method.
		4271
		4272	For C<ev::io> watchers there is an additional C<set> method that acepts a
		4273	new event mask only, and internally calls C<ev_io_modify>.
2507		4274
2508	=item w->start ()	4275	=item w->start ()
2509		4276
2510	Starts the watcher. Note that there is no C<loop> argument, as the	4277	Starts the watcher. Note that there is no C<loop> argument, as the
2511	constructor already stores the event loop.	4278	constructor already stores the event loop.
2512		4279
		4280	=item w->start ([arguments])
		4281
		4282	Instead of calling C<set> and C<start> methods separately, it is often
		4283	convenient to wrap them in one call. Uses the same type of arguments as
		4284	the configure C<set> method of the watcher.
		4285
2513	=item w->stop ()	4286	=item w->stop ()
2514		4287
2515	Stops the watcher if it is active. Again, no C<loop> argument.	4288	Stops the watcher if it is active. Again, no C<loop> argument.
2516		4289
2517	=item w->again () (C<ev::timer>, C<ev::periodic> only)	4290	=item w->again () (C<ev::timer>, C<ev::periodic> only)
…		…
2529		4302
2530	=back	4303	=back
2531		4304
2532	=back	4305	=back
2533		4306
2534	Example: Define a class with an IO and idle watcher, start one of them in	4307	Example: Define a class with two I/O and idle watchers, start the I/O
2535	the constructor.	4308	watchers in the constructor.
2536		4309
2537	class myclass	4310	class myclass
2538	{	4311	{
2539	ev::io io; void io_cb (ev::io &w, int revents);	4312	ev::io io ; void io_cb (ev::io &w, int revents);
		4313	ev::io io2 ; void io2_cb (ev::io &w, int revents);
2540	ev:idle idle void idle_cb (ev::idle &w, int revents);	4314	ev::idle idle; void idle_cb (ev::idle &w, int revents);
2541		4315
2542	myclass (int fd)	4316	myclass (int fd)
2543	{	4317	{
2544	io .set <myclass, &myclass::io_cb > (this);	4318	io .set <myclass, &myclass::io_cb > (this);
		4319	io2 .set <myclass, &myclass::io2_cb > (this);
2545	idle.set <myclass, &myclass::idle_cb> (this);	4320	idle.set <myclass, &myclass::idle_cb> (this);
2546		4321
2547	io.start (fd, ev::READ);	4322	io.set (fd, ev::WRITE); // configure the watcher
		4323	io.start (); // start it whenever convenient
		4324
		4325	io2.start (fd, ev::READ); // set + start in one call
2548	}	4326	}
2549	};	4327	};
2550		4328
2551		4329
2552	=head1 OTHER LANGUAGE BINDINGS	4330	=head1 OTHER LANGUAGE BINDINGS
2553		4331
2554	Libev does not offer other language bindings itself, but bindings for a	4332	Libev does not offer other language bindings itself, but bindings for a
2555	numbe rof languages exist in the form of third-party packages. If you know	4333	number of languages exist in the form of third-party packages. If you know
2556	any interesting language binding in addition to the ones listed here, drop	4334	any interesting language binding in addition to the ones listed here, drop
2557	me a note.	4335	me a note.
2558		4336
2559	=over 4	4337	=over 4
2560		4338
2561	=item Perl	4339	=item Perl
2562		4340
2563	The EV module implements the full libev API and is actually used to test	4341	The EV module implements the full libev API and is actually used to test
2564	libev. EV is developed together with libev. Apart from the EV core module,	4342	libev. EV is developed together with libev. Apart from the EV core module,
2565	there are additional modules that implement libev-compatible interfaces	4343	there are additional modules that implement libev-compatible interfaces
2566	to C<libadns> (C<EV::ADNS>), C<Net::SNMP> (C<Net::SNMP::EV>) and the	4344	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
2567	C<libglib> event core (C<Glib::EV> and C<EV::Glib>).	4345	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		4346	and C<EV::Glib>).
2568		4347
2569	It can be found and installed via CPAN, its homepage is found at	4348	It can be found and installed via CPAN, its homepage is at
2570	L<http://software.schmorp.de/pkg/EV>.	4349	L<http://software.schmorp.de/pkg/EV>.
2571		4350
		4351	=item Python
		4352
		4353	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		4354	seems to be quite complete and well-documented.
		4355
2572	=item Ruby	4356	=item Ruby
2573		4357
2574	Tony Arcieri has written a ruby extension that offers access to a subset	4358	Tony Arcieri has written a ruby extension that offers access to a subset
2575	of the libev API and adds filehandle abstractions, asynchronous DNS and	4359	of the libev API and adds file handle abstractions, asynchronous DNS and
2576	more on top of it. It can be found via gem servers. Its homepage is at	4360	more on top of it. It can be found via gem servers. Its homepage is at
2577	L<http://rev.rubyforge.org/>.	4361	L<http://rev.rubyforge.org/>.
2578		4362
		4363	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		4364	makes rev work even on mingw.
		4365
		4366	=item Haskell
		4367
		4368	A haskell binding to libev is available at
		4369	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		4370
2579	=item D	4371	=item D
2580		4372
2581	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to	4373	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
2582	be found at L<http://git.llucax.com.ar/?p=software/ev.d.git;a=summary>.	4374	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
		4375
		4376	=item Ocaml
		4377
		4378	Erkki Seppala has written Ocaml bindings for libev, to be found at
		4379	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		4380
		4381	=item Lua
		4382
		4383	Brian Maher has written a partial interface to libev for lua (at the
		4384	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		4385	L<http://github.com/brimworks/lua-ev>.
		4386
		4387	=item Javascript
		4388
		4389	Node.js (L<http://nodejs.org>) uses libev as the underlying event library.
		4390
		4391	=item Others
		4392
		4393	There are others, and I stopped counting.
2583		4394
2584	=back	4395	=back
2585		4396
2586		4397
2587	=head1 MACRO MAGIC	4398	=head1 MACRO MAGIC
2588		4399
2589	Libev can be compiled with a variety of options, the most fundamantal	4400	Libev can be compiled with a variety of options, the most fundamental
2590	of which is C<EV_MULTIPLICITY>. This option determines whether (most)	4401	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
2591	functions and callbacks have an initial C<struct ev_loop *> argument.	4402	functions and callbacks have an initial C<struct ev_loop *> argument.
2592		4403
2593	To make it easier to write programs that cope with either variant, the	4404	To make it easier to write programs that cope with either variant, the
2594	following macros are defined:	4405	following macros are defined:
…		…
2599		4410
2600	This provides the loop I<argument> for functions, if one is required ("ev	4411	This provides the loop I<argument> for functions, if one is required ("ev
2601	loop argument"). The C<EV_A> form is used when this is the sole argument,	4412	loop argument"). The C<EV_A> form is used when this is the sole argument,
2602	C<EV_A_> is used when other arguments are following. Example:	4413	C<EV_A_> is used when other arguments are following. Example:
2603		4414
2604	ev_unref (EV_A);	4415	ev_unref (EV_A);
2605	ev_timer_add (EV_A_ watcher);	4416	ev_timer_add (EV_A_ watcher);
2606	ev_loop (EV_A_ 0);	4417	ev_run (EV_A_ 0);
2607		4418
2608	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	4419	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
2609	which is often provided by the following macro.	4420	which is often provided by the following macro.
2610		4421
2611	=item C<EV_P>, C<EV_P_>	4422	=item C<EV_P>, C<EV_P_>
2612		4423
2613	This provides the loop I<parameter> for functions, if one is required ("ev	4424	This provides the loop I<parameter> for functions, if one is required ("ev
2614	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	4425	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
2615	C<EV_P_> is used when other parameters are following. Example:	4426	C<EV_P_> is used when other parameters are following. Example:
2616		4427
2617	// this is how ev_unref is being declared	4428	// this is how ev_unref is being declared
2618	static void ev_unref (EV_P);	4429	static void ev_unref (EV_P);
2619		4430
2620	// this is how you can declare your typical callback	4431	// this is how you can declare your typical callback
2621	static void cb (EV_P_ ev_timer *w, int revents)	4432	static void cb (EV_P_ ev_timer *w, int revents)
2622		4433
2623	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	4434	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
2624	suitable for use with C<EV_A>.	4435	suitable for use with C<EV_A>.
2625		4436
2626	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	4437	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
2627		4438
2628	Similar to the other two macros, this gives you the value of the default	4439	Similar to the other two macros, this gives you the value of the default
2629	loop, if multiple loops are supported ("ev loop default").	4440	loop, if multiple loops are supported ("ev loop default"). The default loop
		4441	will be initialised if it isn't already initialised.
		4442
		4443	For non-multiplicity builds, these macros do nothing, so you always have
		4444	to initialise the loop somewhere.
2630		4445
2631	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>	4446	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
2632		4447
2633	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the	4448	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
2634	default loop has been initialised (C<UC> == unchecked). Their behaviour	4449	default loop has been initialised (C<UC> == unchecked). Their behaviour
…		…
2642		4457
2643	Example: Declare and initialise a check watcher, utilising the above	4458	Example: Declare and initialise a check watcher, utilising the above
2644	macros so it will work regardless of whether multiple loops are supported	4459	macros so it will work regardless of whether multiple loops are supported
2645	or not.	4460	or not.
2646		4461
2647	static void	4462	static void
2648	check_cb (EV_P_ ev_timer *w, int revents)	4463	check_cb (EV_P_ ev_timer *w, int revents)
2649	{	4464	{
2650	ev_check_stop (EV_A_ w);	4465	ev_check_stop (EV_A_ w);
2651	}	4466	}
2652		4467
2653	ev_check check;	4468	ev_check check;
2654	ev_check_init (&check, check_cb);	4469	ev_check_init (&check, check_cb);
2655	ev_check_start (EV_DEFAULT_ &check);	4470	ev_check_start (EV_DEFAULT_ &check);
2656	ev_loop (EV_DEFAULT_ 0);	4471	ev_run (EV_DEFAULT_ 0);
2657		4472
2658	=head1 EMBEDDING	4473	=head1 EMBEDDING
2659		4474
2660	Libev can (and often is) directly embedded into host	4475	Libev can (and often is) directly embedded into host
2661	applications. Examples of applications that embed it include the Deliantra	4476	applications. Examples of applications that embed it include the Deliantra
…		…
2668	libev somewhere in your source tree).	4483	libev somewhere in your source tree).
2669		4484
2670	=head2 FILESETS	4485	=head2 FILESETS
2671		4486
2672	Depending on what features you need you need to include one or more sets of files	4487	Depending on what features you need you need to include one or more sets of files
2673	in your app.	4488	in your application.
2674		4489
2675	=head3 CORE EVENT LOOP	4490	=head3 CORE EVENT LOOP
2676		4491
2677	To include only the libev core (all the C<ev_*> functions), with manual	4492	To include only the libev core (all the C<ev_*> functions), with manual
2678	configuration (no autoconf):	4493	configuration (no autoconf):
2679		4494
2680	#define EV_STANDALONE 1	4495	#define EV_STANDALONE 1
2681	#include "ev.c"	4496	#include "ev.c"
2682		4497
2683	This will automatically include F<ev.h>, too, and should be done in a	4498	This will automatically include F<ev.h>, too, and should be done in a
2684	single C source file only to provide the function implementations. To use	4499	single C source file only to provide the function implementations. To use
2685	it, do the same for F<ev.h> in all files wishing to use this API (best	4500	it, do the same for F<ev.h> in all files wishing to use this API (best
2686	done by writing a wrapper around F<ev.h> that you can include instead and	4501	done by writing a wrapper around F<ev.h> that you can include instead and
2687	where you can put other configuration options):	4502	where you can put other configuration options):
2688		4503
2689	#define EV_STANDALONE 1	4504	#define EV_STANDALONE 1
2690	#include "ev.h"	4505	#include "ev.h"
2691		4506
2692	Both header files and implementation files can be compiled with a C++	4507	Both header files and implementation files can be compiled with a C++
2693	compiler (at least, thats a stated goal, and breakage will be treated	4508	compiler (at least, that's a stated goal, and breakage will be treated
2694	as a bug).	4509	as a bug).
2695		4510
2696	You need the following files in your source tree, or in a directory	4511	You need the following files in your source tree, or in a directory
2697	in your include path (e.g. in libev/ when using -Ilibev):	4512	in your include path (e.g. in libev/ when using -Ilibev):
2698		4513
2699	ev.h	4514	ev.h
2700	ev.c	4515	ev.c
2701	ev_vars.h	4516	ev_vars.h
2702	ev_wrap.h	4517	ev_wrap.h
2703		4518
2704	ev_win32.c required on win32 platforms only	4519	ev_win32.c required on win32 platforms only
2705		4520
2706	ev_select.c only when select backend is enabled (which is enabled by default)	4521	ev_select.c only when select backend is enabled
2707	ev_poll.c only when poll backend is enabled (disabled by default)	4522	ev_poll.c only when poll backend is enabled
2708	ev_epoll.c only when the epoll backend is enabled (disabled by default)	4523	ev_epoll.c only when the epoll backend is enabled
		4524	ev_linuxaio.c only when the linux aio backend is enabled
		4525	ev_iouring.c only when the linux io_uring backend is enabled
2709	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	4526	ev_kqueue.c only when the kqueue backend is enabled
2710	ev_port.c only when the solaris port backend is enabled (disabled by default)	4527	ev_port.c only when the solaris port backend is enabled
2711		4528
2712	F<ev.c> includes the backend files directly when enabled, so you only need	4529	F<ev.c> includes the backend files directly when enabled, so you only need
2713	to compile this single file.	4530	to compile this single file.
2714		4531
2715	=head3 LIBEVENT COMPATIBILITY API	4532	=head3 LIBEVENT COMPATIBILITY API
2716		4533
2717	To include the libevent compatibility API, also include:	4534	To include the libevent compatibility API, also include:
2718		4535
2719	#include "event.c"	4536	#include "event.c"
2720		4537
2721	in the file including F<ev.c>, and:	4538	in the file including F<ev.c>, and:
2722		4539
2723	#include "event.h"	4540	#include "event.h"
2724		4541
2725	in the files that want to use the libevent API. This also includes F<ev.h>.	4542	in the files that want to use the libevent API. This also includes F<ev.h>.
2726		4543
2727	You need the following additional files for this:	4544	You need the following additional files for this:
2728		4545
2729	event.h	4546	event.h
2730	event.c	4547	event.c
2731		4548
2732	=head3 AUTOCONF SUPPORT	4549	=head3 AUTOCONF SUPPORT
2733		4550
2734	Instead of using C<EV_STANDALONE=1> and providing your config in	4551	Instead of using C<EV_STANDALONE=1> and providing your configuration in
2735	whatever way you want, you can also C<m4_include([libev.m4])> in your	4552	whatever way you want, you can also C<m4_include([libev.m4])> in your
2736	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	4553	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
2737	include F<config.h> and configure itself accordingly.	4554	include F<config.h> and configure itself accordingly.
2738		4555
2739	For this of course you need the m4 file:	4556	For this of course you need the m4 file:
2740		4557
2741	libev.m4	4558	libev.m4
2742		4559
2743	=head2 PREPROCESSOR SYMBOLS/MACROS	4560	=head2 PREPROCESSOR SYMBOLS/MACROS
2744		4561
2745	Libev can be configured via a variety of preprocessor symbols you have to	4562	Libev can be configured via a variety of preprocessor symbols you have to
2746	define before including any of its files. The default in the absense of	4563	define before including (or compiling) any of its files. The default in
2747	autoconf is noted for every option.	4564	the absence of autoconf is documented for every option.
		4565
		4566	Symbols marked with "(h)" do not change the ABI, and can have different
		4567	values when compiling libev vs. including F<ev.h>, so it is permissible
		4568	to redefine them before including F<ev.h> without breaking compatibility
		4569	to a compiled library. All other symbols change the ABI, which means all
		4570	users of libev and the libev code itself must be compiled with compatible
		4571	settings.
2748		4572
2749	=over 4	4573	=over 4
2750		4574
		4575	=item EV_COMPAT3 (h)
		4576
		4577	Backwards compatibility is a major concern for libev. This is why this
		4578	release of libev comes with wrappers for the functions and symbols that
		4579	have been renamed between libev version 3 and 4.
		4580
		4581	You can disable these wrappers (to test compatibility with future
		4582	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4583	sources. This has the additional advantage that you can drop the C<struct>
		4584	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4585	typedef in that case.
		4586
		4587	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4588	and in some even more future version the compatibility code will be
		4589	removed completely.
		4590
2751	=item EV_STANDALONE	4591	=item EV_STANDALONE (h)
2752		4592
2753	Must always be C<1> if you do not use autoconf configuration, which	4593	Must always be C<1> if you do not use autoconf configuration, which
2754	keeps libev from including F<config.h>, and it also defines dummy	4594	keeps libev from including F<config.h>, and it also defines dummy
2755	implementations for some libevent functions (such as logging, which is not	4595	implementations for some libevent functions (such as logging, which is not
2756	supported). It will also not define any of the structs usually found in	4596	supported). It will also not define any of the structs usually found in
2757	F<event.h> that are not directly supported by the libev core alone.	4597	F<event.h> that are not directly supported by the libev core alone.
2758		4598
		4599	In standalone mode, libev will still try to automatically deduce the
		4600	configuration, but has to be more conservative.
		4601
		4602	=item EV_USE_FLOOR
		4603
		4604	If defined to be C<1>, libev will use the C<floor ()> function for its
		4605	periodic reschedule calculations, otherwise libev will fall back on a
		4606	portable (slower) implementation. If you enable this, you usually have to
		4607	link against libm or something equivalent. Enabling this when the C<floor>
		4608	function is not available will fail, so the safe default is to not enable
		4609	this.
		4610
2759	=item EV_USE_MONOTONIC	4611	=item EV_USE_MONOTONIC
2760		4612
2761	If defined to be C<1>, libev will try to detect the availability of the	4613	If defined to be C<1>, libev will try to detect the availability of the
2762	monotonic clock option at both compiletime and runtime. Otherwise no use	4614	monotonic clock option at both compile time and runtime. Otherwise no
2763	of the monotonic clock option will be attempted. If you enable this, you	4615	use of the monotonic clock option will be attempted. If you enable this,
2764	usually have to link against librt or something similar. Enabling it when	4616	you usually have to link against librt or something similar. Enabling it
2765	the functionality isn't available is safe, though, although you have	4617	when the functionality isn't available is safe, though, although you have
2766	to make sure you link against any libraries where the C<clock_gettime>	4618	to make sure you link against any libraries where the C<clock_gettime>
2767	function is hiding in (often F<-lrt>).	4619	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
2768		4620
2769	=item EV_USE_REALTIME	4621	=item EV_USE_REALTIME
2770		4622
2771	If defined to be C<1>, libev will try to detect the availability of the	4623	If defined to be C<1>, libev will try to detect the availability of the
2772	realtime clock option at compiletime (and assume its availability at	4624	real-time clock option at compile time (and assume its availability
2773	runtime if successful). Otherwise no use of the realtime clock option will	4625	at runtime if successful). Otherwise no use of the real-time clock
2774	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	4626	option will be attempted. This effectively replaces C<gettimeofday>
2775	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the	4627	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
2776	note about libraries in the description of C<EV_USE_MONOTONIC>, though.	4628	correctness. See the note about libraries in the description of
		4629	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4630	C<EV_USE_CLOCK_SYSCALL>.
		4631
		4632	=item EV_USE_CLOCK_SYSCALL
		4633
		4634	If defined to be C<1>, libev will try to use a direct syscall instead
		4635	of calling the system-provided C<clock_gettime> function. This option
		4636	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4637	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4638	programs needlessly. Using a direct syscall is slightly slower (in
		4639	theory), because no optimised vdso implementation can be used, but avoids
		4640	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4641	higher, as it simplifies linking (no need for C<-lrt>).
2777		4642
2778	=item EV_USE_NANOSLEEP	4643	=item EV_USE_NANOSLEEP
2779		4644
2780	If defined to be C<1>, libev will assume that C<nanosleep ()> is available	4645	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
2781	and will use it for delays. Otherwise it will use C<select ()>.	4646	and will use it for delays. Otherwise it will use C<select ()>.
…		…
2786	available and will probe for kernel support at runtime. This will improve	4651	available and will probe for kernel support at runtime. This will improve
2787	C<ev_signal> and C<ev_async> performance and reduce resource consumption.	4652	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
2788	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc	4653	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
2789	2.7 or newer, otherwise disabled.	4654	2.7 or newer, otherwise disabled.
2790		4655
		4656	=item EV_USE_SIGNALFD
		4657
		4658	If defined to be C<1>, then libev will assume that C<signalfd ()> is
		4659	available and will probe for kernel support at runtime. This enables
		4660	the use of EVFLAG_SIGNALFD for faster and simpler signal handling. If
		4661	undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4662	2.7 or newer, otherwise disabled.
		4663
		4664	=item EV_USE_TIMERFD
		4665
		4666	If defined to be C<1>, then libev will assume that C<timerfd ()> is
		4667	available and will probe for kernel support at runtime. This allows
		4668	libev to detect time jumps accurately. If undefined, it will be enabled
		4669	if the headers indicate GNU/Linux + Glibc 2.8 or newer and define
		4670	C<TFD_TIMER_CANCEL_ON_SET>, otherwise disabled.
		4671
		4672	=item EV_USE_EVENTFD
		4673
		4674	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4675	available and will probe for kernel support at runtime. This will improve
		4676	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4677	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4678	2.7 or newer, otherwise disabled.
		4679
2791	=item EV_USE_SELECT	4680	=item EV_USE_SELECT
2792		4681
2793	If undefined or defined to be C<1>, libev will compile in support for the	4682	If undefined or defined to be C<1>, libev will compile in support for the
2794	C<select>(2) backend. No attempt at autodetection will be done: if no	4683	C<select>(2) backend. No attempt at auto-detection will be done: if no
2795	other method takes over, select will be it. Otherwise the select backend	4684	other method takes over, select will be it. Otherwise the select backend
2796	will not be compiled in.	4685	will not be compiled in.
2797		4686
2798	=item EV_SELECT_USE_FD_SET	4687	=item EV_SELECT_USE_FD_SET
2799		4688
2800	If defined to C<1>, then the select backend will use the system C<fd_set>	4689	If defined to C<1>, then the select backend will use the system C<fd_set>
2801	structure. This is useful if libev doesn't compile due to a missing	4690	structure. This is useful if libev doesn't compile due to a missing
2802	C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on	4691	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
2803	exotic systems. This usually limits the range of file descriptors to some	4692	on exotic systems. This usually limits the range of file descriptors to
2804	low limit such as 1024 or might have other limitations (winsocket only	4693	some low limit such as 1024 or might have other limitations (winsocket
2805	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	4694	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
2806	influence the size of the C<fd_set> used.	4695	configures the maximum size of the C<fd_set>.
2807		4696
2808	=item EV_SELECT_IS_WINSOCKET	4697	=item EV_SELECT_IS_WINSOCKET
2809		4698
2810	When defined to C<1>, the select backend will assume that	4699	When defined to C<1>, the select backend will assume that
2811	select/socket/connect etc. don't understand file descriptors but	4700	select/socket/connect etc. don't understand file descriptors but
…		…
2813	be used is the winsock select). This means that it will call	4702	be used is the winsock select). This means that it will call
2814	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	4703	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2815	it is assumed that all these functions actually work on fds, even	4704	it is assumed that all these functions actually work on fds, even
2816	on win32. Should not be defined on non-win32 platforms.	4705	on win32. Should not be defined on non-win32 platforms.
2817		4706
2818	=item EV_FD_TO_WIN32_HANDLE	4707	=item EV_FD_TO_WIN32_HANDLE(fd)
2819		4708
2820	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map	4709	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
2821	file descriptors to socket handles. When not defining this symbol (the	4710	file descriptors to socket handles. When not defining this symbol (the
2822	default), then libev will call C<_get_osfhandle>, which is usually	4711	default), then libev will call C<_get_osfhandle>, which is usually
2823	correct. In some cases, programs use their own file descriptor management,	4712	correct. In some cases, programs use their own file descriptor management,
2824	in which case they can provide this function to map fds to socket handles.	4713	in which case they can provide this function to map fds to socket handles.
2825		4714
		4715	=item EV_WIN32_HANDLE_TO_FD(handle)
		4716
		4717	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4718	using the standard C<_open_osfhandle> function. For programs implementing
		4719	their own fd to handle mapping, overwriting this function makes it easier
		4720	to do so. This can be done by defining this macro to an appropriate value.
		4721
		4722	=item EV_WIN32_CLOSE_FD(fd)
		4723
		4724	If programs implement their own fd to handle mapping on win32, then this
		4725	macro can be used to override the C<close> function, useful to unregister
		4726	file descriptors again. Note that the replacement function has to close
		4727	the underlying OS handle.
		4728
		4729	=item EV_USE_WSASOCKET
		4730
		4731	If defined to be C<1>, libev will use C<WSASocket> to create its internal
		4732	communication socket, which works better in some environments. Otherwise,
		4733	the normal C<socket> function will be used, which works better in other
		4734	environments.
		4735
2826	=item EV_USE_POLL	4736	=item EV_USE_POLL
2827		4737
2828	If defined to be C<1>, libev will compile in support for the C<poll>(2)	4738	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2829	backend. Otherwise it will be enabled on non-win32 platforms. It	4739	backend. Otherwise it will be enabled on non-win32 platforms. It
2830	takes precedence over select.	4740	takes precedence over select.
…		…
2834	If defined to be C<1>, libev will compile in support for the Linux	4744	If defined to be C<1>, libev will compile in support for the Linux
2835	C<epoll>(7) backend. Its availability will be detected at runtime,	4745	C<epoll>(7) backend. Its availability will be detected at runtime,
2836	otherwise another method will be used as fallback. This is the preferred	4746	otherwise another method will be used as fallback. This is the preferred
2837	backend for GNU/Linux systems. If undefined, it will be enabled if the	4747	backend for GNU/Linux systems. If undefined, it will be enabled if the
2838	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.	4748	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4749
		4750	=item EV_USE_LINUXAIO
		4751
		4752	If defined to be C<1>, libev will compile in support for the Linux aio
		4753	backend (C<EV_USE_EPOLL> must also be enabled). If undefined, it will be
		4754	enabled on linux, otherwise disabled.
		4755
		4756	=item EV_USE_IOURING
		4757
		4758	If defined to be C<1>, libev will compile in support for the Linux
		4759	io_uring backend (C<EV_USE_EPOLL> must also be enabled). Due to it's
		4760	current limitations it has to be requested explicitly. If undefined, it
		4761	will be enabled on linux, otherwise disabled.
2839		4762
2840	=item EV_USE_KQUEUE	4763	=item EV_USE_KQUEUE
2841		4764
2842	If defined to be C<1>, libev will compile in support for the BSD style	4765	If defined to be C<1>, libev will compile in support for the BSD style
2843	C<kqueue>(2) backend. Its actual availability will be detected at runtime,	4766	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
…		…
2856	otherwise another method will be used as fallback. This is the preferred	4779	otherwise another method will be used as fallback. This is the preferred
2857	backend for Solaris 10 systems.	4780	backend for Solaris 10 systems.
2858		4781
2859	=item EV_USE_DEVPOLL	4782	=item EV_USE_DEVPOLL
2860		4783
2861	reserved for future expansion, works like the USE symbols above.	4784	Reserved for future expansion, works like the USE symbols above.
2862		4785
2863	=item EV_USE_INOTIFY	4786	=item EV_USE_INOTIFY
2864		4787
2865	If defined to be C<1>, libev will compile in support for the Linux inotify	4788	If defined to be C<1>, libev will compile in support for the Linux inotify
2866	interface to speed up C<ev_stat> watchers. Its actual availability will	4789	interface to speed up C<ev_stat> watchers. Its actual availability will
2867	be detected at runtime. If undefined, it will be enabled if the headers	4790	be detected at runtime. If undefined, it will be enabled if the headers
2868	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.	4791	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
2869		4792
		4793	=item EV_NO_SMP
		4794
		4795	If defined to be C<1>, libev will assume that memory is always coherent
		4796	between threads, that is, threads can be used, but threads never run on
		4797	different cpus (or different cpu cores). This reduces dependencies
		4798	and makes libev faster.
		4799
		4800	=item EV_NO_THREADS
		4801
		4802	If defined to be C<1>, libev will assume that it will never be called from
		4803	different threads (that includes signal handlers), which is a stronger
		4804	assumption than C<EV_NO_SMP>, above. This reduces dependencies and makes
		4805	libev faster.
		4806
2870	=item EV_ATOMIC_T	4807	=item EV_ATOMIC_T
2871		4808
2872	Libev requires an integer type (suitable for storing C<0> or C<1>) whose	4809	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
2873	access is atomic with respect to other threads or signal contexts. No such	4810	access is atomic with respect to other threads or signal contexts. No
2874	type is easily found in the C language, so you can provide your own type	4811	such type is easily found in the C language, so you can provide your own
2875	that you know is safe for your purposes. It is used both for signal handler "locking"	4812	type that you know is safe for your purposes. It is used both for signal
2876	as well as for signal and thread safety in C<ev_async> watchers.	4813	handler "locking" as well as for signal and thread safety in C<ev_async>
		4814	watchers.
2877		4815
2878	In the absense of this define, libev will use C<sig_atomic_t volatile>	4816	In the absence of this define, libev will use C<sig_atomic_t volatile>
2879	(from F<signal.h>), which is usually good enough on most platforms.	4817	(from F<signal.h>), which is usually good enough on most platforms.
2880		4818
2881	=item EV_H	4819	=item EV_H (h)
2882		4820
2883	The name of the F<ev.h> header file used to include it. The default if	4821	The name of the F<ev.h> header file used to include it. The default if
2884	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be	4822	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2885	used to virtually rename the F<ev.h> header file in case of conflicts.	4823	used to virtually rename the F<ev.h> header file in case of conflicts.
2886		4824
2887	=item EV_CONFIG_H	4825	=item EV_CONFIG_H (h)
2888		4826
2889	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	4827	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2890	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	4828	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2891	C<EV_H>, above.	4829	C<EV_H>, above.
2892		4830
2893	=item EV_EVENT_H	4831	=item EV_EVENT_H (h)
2894		4832
2895	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	4833	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2896	of how the F<event.h> header can be found, the default is C<"event.h">.	4834	of how the F<event.h> header can be found, the default is C<"event.h">.
2897		4835
2898	=item EV_PROTOTYPES	4836	=item EV_PROTOTYPES (h)
2899		4837
2900	If defined to be C<0>, then F<ev.h> will not define any function	4838	If defined to be C<0>, then F<ev.h> will not define any function
2901	prototypes, but still define all the structs and other symbols. This is	4839	prototypes, but still define all the structs and other symbols. This is
2902	occasionally useful if you want to provide your own wrapper functions	4840	occasionally useful if you want to provide your own wrapper functions
2903	around libev functions.	4841	around libev functions.
…		…
2908	will have the C<struct ev_loop *> as first argument, and you can create	4846	will have the C<struct ev_loop *> as first argument, and you can create
2909	additional independent event loops. Otherwise there will be no support	4847	additional independent event loops. Otherwise there will be no support
2910	for multiple event loops and there is no first event loop pointer	4848	for multiple event loops and there is no first event loop pointer
2911	argument. Instead, all functions act on the single default loop.	4849	argument. Instead, all functions act on the single default loop.
2912		4850
		4851	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4852	default loop when multiplicity is switched off - you always have to
		4853	initialise the loop manually in this case.
		4854
2913	=item EV_MINPRI	4855	=item EV_MINPRI
2914		4856
2915	=item EV_MAXPRI	4857	=item EV_MAXPRI
2916		4858
2917	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to	4859	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
…		…
2922	When doing priority-based operations, libev usually has to linearly search	4864	When doing priority-based operations, libev usually has to linearly search
2923	all the priorities, so having many of them (hundreds) uses a lot of space	4865	all the priorities, so having many of them (hundreds) uses a lot of space
2924	and time, so using the defaults of five priorities (-2 .. +2) is usually	4866	and time, so using the defaults of five priorities (-2 .. +2) is usually
2925	fine.	4867	fine.
2926		4868
2927	If your embedding app does not need any priorities, defining these both to	4869	If your embedding application does not need any priorities, defining these
2928	C<0> will save some memory and cpu.	4870	both to C<0> will save some memory and CPU.
2929		4871
2930	=item EV_PERIODIC_ENABLE	4872	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4873	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4874	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
2931		4875
2932	If undefined or defined to be C<1>, then periodic timers are supported. If	4876	If undefined or defined to be C<1> (and the platform supports it), then
2933	defined to be C<0>, then they are not. Disabling them saves a few kB of	4877	the respective watcher type is supported. If defined to be C<0>, then it
2934	code.	4878	is not. Disabling watcher types mainly saves code size.
2935		4879
2936	=item EV_IDLE_ENABLE	4880	=item EV_FEATURES
2937
2938	If undefined or defined to be C<1>, then idle watchers are supported. If
2939	defined to be C<0>, then they are not. Disabling them saves a few kB of
2940	code.
2941
2942	=item EV_EMBED_ENABLE
2943
2944	If undefined or defined to be C<1>, then embed watchers are supported. If
2945	defined to be C<0>, then they are not.
2946
2947	=item EV_STAT_ENABLE
2948
2949	If undefined or defined to be C<1>, then stat watchers are supported. If
2950	defined to be C<0>, then they are not.
2951
2952	=item EV_FORK_ENABLE
2953
2954	If undefined or defined to be C<1>, then fork watchers are supported. If
2955	defined to be C<0>, then they are not.
2956
2957	=item EV_ASYNC_ENABLE
2958
2959	If undefined or defined to be C<1>, then async watchers are supported. If
2960	defined to be C<0>, then they are not.
2961
2962	=item EV_MINIMAL
2963		4881
2964	If you need to shave off some kilobytes of code at the expense of some	4882	If you need to shave off some kilobytes of code at the expense of some
2965	speed, define this symbol to C<1>. Currently only used for gcc to override	4883	speed (but with the full API), you can define this symbol to request
2966	some inlining decisions, saves roughly 30% codesize of amd64.	4884	certain subsets of functionality. The default is to enable all features
		4885	that can be enabled on the platform.
		4886
		4887	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4888	with some broad features you want) and then selectively re-enable
		4889	additional parts you want, for example if you want everything minimal,
		4890	but multiple event loop support, async and child watchers and the poll
		4891	backend, use this:
		4892
		4893	#define EV_FEATURES 0
		4894	#define EV_MULTIPLICITY 1
		4895	#define EV_USE_POLL 1
		4896	#define EV_CHILD_ENABLE 1
		4897	#define EV_ASYNC_ENABLE 1
		4898
		4899	The actual value is a bitset, it can be a combination of the following
		4900	values (by default, all of these are enabled):
		4901
		4902	=over 4
		4903
		4904	=item C<1> - faster/larger code
		4905
		4906	Use larger code to speed up some operations.
		4907
		4908	Currently this is used to override some inlining decisions (enlarging the
		4909	code size by roughly 30% on amd64).
		4910
		4911	When optimising for size, use of compiler flags such as C<-Os> with
		4912	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4913	assertions.
		4914
		4915	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4916	(e.g. gcc with C<-Os>).
		4917
		4918	=item C<2> - faster/larger data structures
		4919
		4920	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4921	hash table sizes and so on. This will usually further increase code size
		4922	and can additionally have an effect on the size of data structures at
		4923	runtime.
		4924
		4925	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4926	(e.g. gcc with C<-Os>).
		4927
		4928	=item C<4> - full API configuration
		4929
		4930	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4931	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4932
		4933	=item C<8> - full API
		4934
		4935	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4936	details on which parts of the API are still available without this
		4937	feature, and do not complain if this subset changes over time.
		4938
		4939	=item C<16> - enable all optional watcher types
		4940
		4941	Enables all optional watcher types. If you want to selectively enable
		4942	only some watcher types other than I/O and timers (e.g. prepare,
		4943	embed, async, child...) you can enable them manually by defining
		4944	C<EV_watchertype_ENABLE> to C<1> instead.
		4945
		4946	=item C<32> - enable all backends
		4947
		4948	This enables all backends - without this feature, you need to enable at
		4949	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4950
		4951	=item C<64> - enable OS-specific "helper" APIs
		4952
		4953	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4954	default.
		4955
		4956	=back
		4957
		4958	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4959	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4960	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4961	watchers, timers and monotonic clock support.
		4962
		4963	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4964	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4965	your program might be left out as well - a binary starting a timer and an
		4966	I/O watcher then might come out at only 5Kb.
		4967
		4968	=item EV_API_STATIC
		4969
		4970	If this symbol is defined (by default it is not), then all identifiers
		4971	will have static linkage. This means that libev will not export any
		4972	identifiers, and you cannot link against libev anymore. This can be useful
		4973	when you embed libev, only want to use libev functions in a single file,
		4974	and do not want its identifiers to be visible.
		4975
		4976	To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
		4977	wants to use libev.
		4978
		4979	This option only works when libev is compiled with a C compiler, as C++
		4980	doesn't support the required declaration syntax.
		4981
		4982	=item EV_AVOID_STDIO
		4983
		4984	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4985	functions (printf, scanf, perror etc.). This will increase the code size
		4986	somewhat, but if your program doesn't otherwise depend on stdio and your
		4987	libc allows it, this avoids linking in the stdio library which is quite
		4988	big.
		4989
		4990	Note that error messages might become less precise when this option is
		4991	enabled.
		4992
		4993	=item EV_NSIG
		4994
		4995	The highest supported signal number, +1 (or, the number of
		4996	signals): Normally, libev tries to deduce the maximum number of signals
		4997	automatically, but sometimes this fails, in which case it can be
		4998	specified. Also, using a lower number than detected (C<32> should be
		4999	good for about any system in existence) can save some memory, as libev
		5000	statically allocates some 12-24 bytes per signal number.
2967		5001
2968	=item EV_PID_HASHSIZE	5002	=item EV_PID_HASHSIZE
2969		5003
2970	C<ev_child> watchers use a small hash table to distribute workload by	5004	C<ev_child> watchers use a small hash table to distribute workload by
2971	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	5005	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
2972	than enough. If you need to manage thousands of children you might want to	5006	usually more than enough. If you need to manage thousands of children you
2973	increase this value (I<must> be a power of two).	5007	might want to increase this value (I<must> be a power of two).
2974		5008
2975	=item EV_INOTIFY_HASHSIZE	5009	=item EV_INOTIFY_HASHSIZE
2976		5010
2977	C<ev_stat> watchers use a small hash table to distribute workload by	5011	C<ev_stat> watchers use a small hash table to distribute workload by
2978	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	5012	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
2979	usually more than enough. If you need to manage thousands of C<ev_stat>	5013	disabled), usually more than enough. If you need to manage thousands of
2980	watchers you might want to increase this value (I<must> be a power of	5014	C<ev_stat> watchers you might want to increase this value (I<must> be a
2981	two).	5015	power of two).
		5016
		5017	=item EV_USE_4HEAP
		5018
		5019	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		5020	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		5021	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		5022	faster performance with many (thousands) of watchers.
		5023
		5024	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		5025	will be C<0>.
		5026
		5027	=item EV_HEAP_CACHE_AT
		5028
		5029	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		5030	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		5031	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		5032	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		5033	but avoids random read accesses on heap changes. This improves performance
		5034	noticeably with many (hundreds) of watchers.
		5035
		5036	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		5037	will be C<0>.
		5038
		5039	=item EV_VERIFY
		5040
		5041	Controls how much internal verification (see C<ev_verify ()>) will
		5042	be done: If set to C<0>, no internal verification code will be compiled
		5043	in. If set to C<1>, then verification code will be compiled in, but not
		5044	called. If set to C<2>, then the internal verification code will be
		5045	called once per loop, which can slow down libev. If set to C<3>, then the
		5046	verification code will be called very frequently, which will slow down
		5047	libev considerably.
		5048
		5049	Verification errors are reported via C's C<assert> mechanism, so if you
		5050	disable that (e.g. by defining C<NDEBUG>) then no errors will be reported.
		5051
		5052	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		5053	will be C<0>.
2982		5054
2983	=item EV_COMMON	5055	=item EV_COMMON
2984		5056
2985	By default, all watchers have a C<void *data> member. By redefining	5057	By default, all watchers have a C<void *data> member. By redefining
2986	this macro to a something else you can include more and other types of	5058	this macro to something else you can include more and other types of
2987	members. You have to define it each time you include one of the files,	5059	members. You have to define it each time you include one of the files,
2988	though, and it must be identical each time.	5060	though, and it must be identical each time.
2989		5061
2990	For example, the perl EV module uses something like this:	5062	For example, the perl EV module uses something like this:
2991		5063
2992	#define EV_COMMON \	5064	#define EV_COMMON \
2993	SV *self; /* contains this struct */ \	5065	SV *self; /* contains this struct */ \
2994	SV *cb_sv, *fh /* note no trailing ";" */	5066	SV *cb_sv, *fh /* note no trailing ";" */
2995		5067
2996	=item EV_CB_DECLARE (type)	5068	=item EV_CB_DECLARE (type)
2997		5069
2998	=item EV_CB_INVOKE (watcher, revents)	5070	=item EV_CB_INVOKE (watcher, revents)
2999		5071
…		…
3004	definition and a statement, respectively. See the F<ev.h> header file for	5076	definition and a statement, respectively. See the F<ev.h> header file for
3005	their default definitions. One possible use for overriding these is to	5077	their default definitions. One possible use for overriding these is to
3006	avoid the C<struct ev_loop *> as first argument in all cases, or to use	5078	avoid the C<struct ev_loop *> as first argument in all cases, or to use
3007	method calls instead of plain function calls in C++.	5079	method calls instead of plain function calls in C++.
3008		5080
		5081	=back
		5082
3009	=head2 EXPORTED API SYMBOLS	5083	=head2 EXPORTED API SYMBOLS
3010		5084
3011	If you need to re-export the API (e.g. via a dll) and you need a list of	5085	If you need to re-export the API (e.g. via a DLL) and you need a list of
3012	exported symbols, you can use the provided F<Symbol.*> files which list	5086	exported symbols, you can use the provided F<Symbol.*> files which list
3013	all public symbols, one per line:	5087	all public symbols, one per line:
3014		5088
3015	Symbols.ev for libev proper	5089	Symbols.ev for libev proper
3016	Symbols.event for the libevent emulation	5090	Symbols.event for the libevent emulation
3017		5091
3018	This can also be used to rename all public symbols to avoid clashes with	5092	This can also be used to rename all public symbols to avoid clashes with
3019	multiple versions of libev linked together (which is obviously bad in	5093	multiple versions of libev linked together (which is obviously bad in
3020	itself, but sometimes it is inconvinient to avoid this).	5094	itself, but sometimes it is inconvenient to avoid this).
3021		5095
3022	A sed command like this will create wrapper C<#define>'s that you need to	5096	A sed command like this will create wrapper C<#define>'s that you need to
3023	include before including F<ev.h>:	5097	include before including F<ev.h>:
3024		5098
3025	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h	5099	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
…		…
3042	file.	5116	file.
3043		5117
3044	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	5118	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
3045	that everybody includes and which overrides some configure choices:	5119	that everybody includes and which overrides some configure choices:
3046		5120
3047	#define EV_MINIMAL 1	5121	#define EV_FEATURES 8
3048	#define EV_USE_POLL 0	5122	#define EV_USE_SELECT 1
3049	#define EV_MULTIPLICITY 0
3050	#define EV_PERIODIC_ENABLE 0	5123	#define EV_PREPARE_ENABLE 1
		5124	#define EV_IDLE_ENABLE 1
3051	#define EV_STAT_ENABLE 0	5125	#define EV_SIGNAL_ENABLE 1
3052	#define EV_FORK_ENABLE 0	5126	#define EV_CHILD_ENABLE 1
		5127	#define EV_USE_STDEXCEPT 0
3053	#define EV_CONFIG_H <config.h>	5128	#define EV_CONFIG_H <config.h>
3054	#define EV_MINPRI 0
3055	#define EV_MAXPRI 0
3056		5129
3057	#include "ev++.h"	5130	#include "ev++.h"
3058		5131
3059	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	5132	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
3060		5133
3061	#include "ev_cpp.h"	5134	#include "ev_cpp.h"
3062	#include "ev.c"	5135	#include "ev.c"
3063		5136
		5137	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
3064		5138
3065	=head1 THREADS AND COROUTINES	5139	=head2 THREADS AND COROUTINES
3066		5140
3067	=head2 THREADS	5141	=head3 THREADS
3068		5142
3069	Libev itself is completely threadsafe, but it uses no locking. This	5143	All libev functions are reentrant and thread-safe unless explicitly
		5144	documented otherwise, but libev implements no locking itself. This means
3070	means that you can use as many loops as you want in parallel, as long as	5145	that you can use as many loops as you want in parallel, as long as there
3071	only one thread ever calls into one libev function with the same loop	5146	are no concurrent calls into any libev function with the same loop
3072	parameter.	5147	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		5148	of course): libev guarantees that different event loops share no data
		5149	structures that need any locking.
3073		5150
3074	Or put differently: calls with different loop parameters can be done in	5151	Or to put it differently: calls with different loop parameters can be done
3075	parallel from multiple threads, calls with the same loop parameter must be	5152	concurrently from multiple threads, calls with the same loop parameter
3076	done serially (but can be done from different threads, as long as only one	5153	must be done serially (but can be done from different threads, as long as
3077	thread ever is inside a call at any point in time, e.g. by using a mutex	5154	only one thread ever is inside a call at any point in time, e.g. by using
3078	per loop).	5155	a mutex per loop).
3079		5156
3080	If you want to know which design is best for your problem, then I cannot	5157	Specifically to support threads (and signal handlers), libev implements
		5158	so-called C<ev_async> watchers, which allow some limited form of
		5159	concurrency on the same event loop, namely waking it up "from the
		5160	outside".
		5161
		5162	If you want to know which design (one loop, locking, or multiple loops
		5163	without or something else still) is best for your problem, then I cannot
3081	help you but by giving some generic advice:	5164	help you, but here is some generic advice:
3082		5165
3083	=over 4	5166	=over 4
3084		5167
3085	=item * most applications have a main thread: use the default libev loop	5168	=item * most applications have a main thread: use the default libev loop
3086	in that thread, or create a seperate thread running only the default loop.	5169	in that thread, or create a separate thread running only the default loop.
3087		5170
3088	This helps integrating other libraries or software modules that use libev	5171	This helps integrating other libraries or software modules that use libev
3089	themselves and don't care/know about threading.	5172	themselves and don't care/know about threading.
3090		5173
3091	=item * one loop per thread is usually a good model.	5174	=item * one loop per thread is usually a good model.
3092		5175
3093	Doing this is almost never wrong, sometimes a better-performance model	5176	Doing this is almost never wrong, sometimes a better-performance model
3094	exists, but it is always a good start.	5177	exists, but it is always a good start.
3095		5178
3096	=item * other models exist, such as the leader/follower pattern, where one	5179	=item * other models exist, such as the leader/follower pattern, where one
3097	loop is handed through multiple threads in a kind of round-robbin fashion.	5180	loop is handed through multiple threads in a kind of round-robin fashion.
3098		5181
3099	Chosing a model is hard - look around, learn, know that usually you cna do	5182	Choosing a model is hard - look around, learn, know that usually you can do
3100	better than you currently do :-)	5183	better than you currently do :-)
3101		5184
3102	=item * often you need to talk to some other thread which blocks in the	5185	=item * often you need to talk to some other thread which blocks in the
		5186	event loop.
		5187
3103	event loop - C<ev_async> watchers can be used to wake them up from other	5188	C<ev_async> watchers can be used to wake them up from other threads safely
3104	threads safely (or from signal contexts...).	5189	(or from signal contexts...).
		5190
		5191	An example use would be to communicate signals or other events that only
		5192	work in the default loop by registering the signal watcher with the
		5193	default loop and triggering an C<ev_async> watcher from the default loop
		5194	watcher callback into the event loop interested in the signal.
3105		5195
3106	=back	5196	=back
3107		5197
		5198	See also L</THREAD LOCKING EXAMPLE>.
		5199
3108	=head2 COROUTINES	5200	=head3 COROUTINES
3109		5201
3110	Libev is much more accomodating to coroutines ("cooperative threads"):	5202	Libev is very accommodating to coroutines ("cooperative threads"):
3111	libev fully supports nesting calls to it's functions from different	5203	libev fully supports nesting calls to its functions from different
3112	coroutines (e.g. you can call C<ev_loop> on the same loop from two	5204	coroutines (e.g. you can call C<ev_run> on the same loop from two
3113	different coroutines and switch freely between both coroutines running the	5205	different coroutines, and switch freely between both coroutines running
3114	loop, as long as you don't confuse yourself). The only exception is that	5206	the loop, as long as you don't confuse yourself). The only exception is
3115	you must not do this from C<ev_periodic> reschedule callbacks.	5207	that you must not do this from C<ev_periodic> reschedule callbacks.
3116		5208
3117	Care has been invested into making sure that libev does not keep local	5209	Care has been taken to ensure that libev does not keep local state inside
3118	state inside C<ev_loop>, and other calls do not usually allow coroutine	5210	C<ev_run>, and other calls do not usually allow for coroutine switches as
3119	switches.	5211	they do not call any callbacks.
3120		5212
		5213	=head2 COMPILER WARNINGS
3121		5214
3122	=head1 COMPLEXITIES	5215	Depending on your compiler and compiler settings, you might get no or a
		5216	lot of warnings when compiling libev code. Some people are apparently
		5217	scared by this.
3123		5218
3124	In this section the complexities of (many of) the algorithms used inside	5219	However, these are unavoidable for many reasons. For one, each compiler
3125	libev will be explained. For complexity discussions about backends see the	5220	has different warnings, and each user has different tastes regarding
3126	documentation for C<ev_default_init>.	5221	warning options. "Warn-free" code therefore cannot be a goal except when
		5222	targeting a specific compiler and compiler-version.
3127		5223
3128	All of the following are about amortised time: If an array needs to be	5224	Another reason is that some compiler warnings require elaborate
3129	extended, libev needs to realloc and move the whole array, but this	5225	workarounds, or other changes to the code that make it less clear and less
3130	happens asymptotically never with higher number of elements, so O(1) might	5226	maintainable.
3131	mean it might do a lengthy realloc operation in rare cases, but on average
3132	it is much faster and asymptotically approaches constant time.
3133		5227
3134	=over 4	5228	And of course, some compiler warnings are just plain stupid, or simply
		5229	wrong (because they don't actually warn about the condition their message
		5230	seems to warn about). For example, certain older gcc versions had some
		5231	warnings that resulted in an extreme number of false positives. These have
		5232	been fixed, but some people still insist on making code warn-free with
		5233	such buggy versions.
3135		5234
3136	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	5235	While libev is written to generate as few warnings as possible,
		5236	"warn-free" code is not a goal, and it is recommended not to build libev
		5237	with any compiler warnings enabled unless you are prepared to cope with
		5238	them (e.g. by ignoring them). Remember that warnings are just that:
		5239	warnings, not errors, or proof of bugs.
3137		5240
3138	This means that, when you have a watcher that triggers in one hour and
3139	there are 100 watchers that would trigger before that then inserting will
3140	have to skip roughly seven (C<ld 100>) of these watchers.
3141		5241
3142	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)	5242	=head2 VALGRIND
3143		5243
3144	That means that changing a timer costs less than removing/adding them	5244	Valgrind has a special section here because it is a popular tool that is
3145	as only the relative motion in the event queue has to be paid for.	5245	highly useful. Unfortunately, valgrind reports are very hard to interpret.
3146		5246
3147	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)	5247	If you think you found a bug (memory leak, uninitialised data access etc.)
		5248	in libev, then check twice: If valgrind reports something like:
3148		5249
3149	These just add the watcher into an array or at the head of a list.	5250	==2274== definitely lost: 0 bytes in 0 blocks.
		5251	==2274== possibly lost: 0 bytes in 0 blocks.
		5252	==2274== still reachable: 256 bytes in 1 blocks.
3150		5253
3151	=item Stopping check/prepare/idle/fork/async watchers: O(1)	5254	Then there is no memory leak, just as memory accounted to global variables
		5255	is not a memleak - the memory is still being referenced, and didn't leak.
3152		5256
3153	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	5257	Similarly, under some circumstances, valgrind might report kernel bugs
		5258	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		5259	although an acceptable workaround has been found here), or it might be
		5260	confused.
3154		5261
3155	These watchers are stored in lists then need to be walked to find the	5262	Keep in mind that valgrind is a very good tool, but only a tool. Don't
3156	correct watcher to remove. The lists are usually short (you don't usually	5263	make it into some kind of religion.
3157	have many watchers waiting for the same fd or signal).
3158		5264
3159	=item Finding the next timer in each loop iteration: O(1)	5265	If you are unsure about something, feel free to contact the mailing list
		5266	with the full valgrind report and an explanation on why you think this
		5267	is a bug in libev (best check the archives, too :). However, don't be
		5268	annoyed when you get a brisk "this is no bug" answer and take the chance
		5269	of learning how to interpret valgrind properly.
3160		5270
3161	By virtue of using a binary heap, the next timer is always found at the	5271	If you need, for some reason, empty reports from valgrind for your project
3162	beginning of the storage array.	5272	I suggest using suppression lists.
3163		5273
3164	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
3165		5274
3166	A change means an I/O watcher gets started or stopped, which requires	5275	=head1 PORTABILITY NOTES
3167	libev to recalculate its status (and possibly tell the kernel, depending
3168	on backend and wether C<ev_io_set> was used).
3169		5276
3170	=item Activating one watcher (putting it into the pending state): O(1)	5277	=head2 GNU/LINUX 32 BIT LIMITATIONS
3171		5278
3172	=item Priority handling: O(number_of_priorities)	5279	GNU/Linux is the only common platform that supports 64 bit file/large file
		5280	interfaces but I<disables> them by default.
3173		5281
3174	Priorities are implemented by allocating some space for each	5282	That means that libev compiled in the default environment doesn't support
3175	priority. When doing priority-based operations, libev usually has to	5283	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
3176	linearly search all the priorities, but starting/stopping and activating
3177	watchers becomes O(1) w.r.t. priority handling.
3178		5284
3179	=item Sending an ev_async: O(1)	5285	Unfortunately, many programs try to work around this GNU/Linux issue
		5286	by enabling the large file API, which makes them incompatible with the
		5287	standard libev compiled for their system.
3180		5288
3181	=item Processing ev_async_send: O(number_of_async_watchers)	5289	Likewise, libev cannot enable the large file API itself as this would
		5290	suddenly make it incompatible to the default compile time environment,
		5291	i.e. all programs not using special compile switches.
3182		5292
3183	=item Processing signals: O(max_signal_number)	5293	=head2 OS/X AND DARWIN BUGS
3184		5294
3185	Sending involves a syscall I<iff> there were no other C<ev_async_send>	5295	The whole thing is a bug if you ask me - basically any system interface
3186	calls in the current loop iteration. Checking for async and signal events	5296	you touch is broken, whether it is locales, poll, kqueue or even the
3187	involves iterating over all running async watchers or all signal numbers.	5297	OpenGL drivers.
3188		5298
3189	=back	5299	=head3 C<kqueue> is buggy
3190		5300
		5301	The kqueue syscall is broken in all known versions - most versions support
		5302	only sockets, many support pipes.
3191		5303
3192	=head1 Win32 platform limitations and workarounds	5304	Libev tries to work around this by not using C<kqueue> by default on this
		5305	rotten platform, but of course you can still ask for it when creating a
		5306	loop - embedding a socket-only kqueue loop into a select-based one is
		5307	probably going to work well.
		5308
		5309	=head3 C<poll> is buggy
		5310
		5311	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		5312	implementation by something calling C<kqueue> internally around the 10.5.6
		5313	release, so now C<kqueue> I<and> C<poll> are broken.
		5314
		5315	Libev tries to work around this by not using C<poll> by default on
		5316	this rotten platform, but of course you can still ask for it when creating
		5317	a loop.
		5318
		5319	=head3 C<select> is buggy
		5320
		5321	All that's left is C<select>, and of course Apple found a way to fuck this
		5322	one up as well: On OS/X, C<select> actively limits the number of file
		5323	descriptors you can pass in to 1024 - your program suddenly crashes when
		5324	you use more.
		5325
		5326	There is an undocumented "workaround" for this - defining
		5327	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		5328	work on OS/X.
		5329
		5330	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		5331
		5332	=head3 C<errno> reentrancy
		5333
		5334	The default compile environment on Solaris is unfortunately so
		5335	thread-unsafe that you can't even use components/libraries compiled
		5336	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		5337	defined by default. A valid, if stupid, implementation choice.
		5338
		5339	If you want to use libev in threaded environments you have to make sure
		5340	it's compiled with C<_REENTRANT> defined.
		5341
		5342	=head3 Event port backend
		5343
		5344	The scalable event interface for Solaris is called "event
		5345	ports". Unfortunately, this mechanism is very buggy in all major
		5346	releases. If you run into high CPU usage, your program freezes or you get
		5347	a large number of spurious wakeups, make sure you have all the relevant
		5348	and latest kernel patches applied. No, I don't know which ones, but there
		5349	are multiple ones to apply, and afterwards, event ports actually work
		5350	great.
		5351
		5352	If you can't get it to work, you can try running the program by setting
		5353	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		5354	C<select> backends.
		5355
		5356	=head2 AIX POLL BUG
		5357
		5358	AIX unfortunately has a broken C<poll.h> header. Libev works around
		5359	this by trying to avoid the poll backend altogether (i.e. it's not even
		5360	compiled in), which normally isn't a big problem as C<select> works fine
		5361	with large bitsets on AIX, and AIX is dead anyway.
		5362
		5363	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		5364
		5365	=head3 General issues
3193		5366
3194	Win32 doesn't support any of the standards (e.g. POSIX) that libev	5367	Win32 doesn't support any of the standards (e.g. POSIX) that libev
3195	requires, and its I/O model is fundamentally incompatible with the POSIX	5368	requires, and its I/O model is fundamentally incompatible with the POSIX
3196	model. Libev still offers limited functionality on this platform in	5369	model. Libev still offers limited functionality on this platform in
3197	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	5370	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
3198	descriptors. This only applies when using Win32 natively, not when using	5371	descriptors. This only applies when using Win32 natively, not when using
3199	e.g. cygwin.	5372	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		5373	as every compiler comes with a slightly differently broken/incompatible
		5374	environment.
		5375
		5376	Lifting these limitations would basically require the full
		5377	re-implementation of the I/O system. If you are into this kind of thing,
		5378	then note that glib does exactly that for you in a very portable way (note
		5379	also that glib is the slowest event library known to man).
3200		5380
3201	There is no supported compilation method available on windows except	5381	There is no supported compilation method available on windows except
3202	embedding it into other applications.	5382	embedding it into other applications.
3203		5383
		5384	Sensible signal handling is officially unsupported by Microsoft - libev
		5385	tries its best, but under most conditions, signals will simply not work.
		5386
		5387	Not a libev limitation but worth mentioning: windows apparently doesn't
		5388	accept large writes: instead of resulting in a partial write, windows will
		5389	either accept everything or return C<ENOBUFS> if the buffer is too large,
		5390	so make sure you only write small amounts into your sockets (less than a
		5391	megabyte seems safe, but this apparently depends on the amount of memory
		5392	available).
		5393
3204	Due to the many, low, and arbitrary limits on the win32 platform and the	5394	Due to the many, low, and arbitrary limits on the win32 platform and
3205	abysmal performance of winsockets, using a large number of sockets is not	5395	the abysmal performance of winsockets, using a large number of sockets
3206	recommended (and not reasonable). If your program needs to use more than	5396	is not recommended (and not reasonable). If your program needs to use
3207	a hundred or so sockets, then likely it needs to use a totally different	5397	more than a hundred or so sockets, then likely it needs to use a totally
3208	implementation for windows, as libev offers the POSIX model, which cannot	5398	different implementation for windows, as libev offers the POSIX readiness
3209	be implemented efficiently on windows (microsoft monopoly games).	5399	notification model, which cannot be implemented efficiently on windows
		5400	(due to Microsoft monopoly games).
3210		5401
3211	=over 4	5402	A typical way to use libev under windows is to embed it (see the embedding
		5403	section for details) and use the following F<evwrap.h> header file instead
		5404	of F<ev.h>:
3212		5405
		5406	#define EV_STANDALONE /* keeps ev from requiring config.h */
		5407	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		5408
		5409	#include "ev.h"
		5410
		5411	And compile the following F<evwrap.c> file into your project (make sure
		5412	you do I<not> compile the F<ev.c> or any other embedded source files!):
		5413
		5414	#include "evwrap.h"
		5415	#include "ev.c"
		5416
3213	=item The winsocket select function	5417	=head3 The winsocket C<select> function
3214		5418
3215	The winsocket C<select> function doesn't follow POSIX in that it requires	5419	The winsocket C<select> function doesn't follow POSIX in that it
3216	socket I<handles> and not socket I<file descriptors>. This makes select	5420	requires socket I<handles> and not socket I<file descriptors> (it is
3217	very inefficient, and also requires a mapping from file descriptors	5421	also extremely buggy). This makes select very inefficient, and also
3218	to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,	5422	requires a mapping from file descriptors to socket handles (the Microsoft
3219	C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor	5423	C runtime provides the function C<_open_osfhandle> for this). See the
3220	symbols for more info.	5424	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		5425	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
3221		5426
3222	The configuration for a "naked" win32 using the microsoft runtime	5427	The configuration for a "naked" win32 using the Microsoft runtime
3223	libraries and raw winsocket select is:	5428	libraries and raw winsocket select is:
3224		5429
3225	#define EV_USE_SELECT 1	5430	#define EV_USE_SELECT 1
3226	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */	5431	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
3227		5432
3228	Note that winsockets handling of fd sets is O(n), so you can easily get a	5433	Note that winsockets handling of fd sets is O(n), so you can easily get a
3229	complexity in the O(n²) range when using win32.	5434	complexity in the O(n²) range when using win32.
3230		5435
3231	=item Limited number of file descriptors	5436	=head3 Limited number of file descriptors
3232		5437
3233	Windows has numerous arbitrary (and low) limits on things. Early versions	5438	Windows has numerous arbitrary (and low) limits on things.
3234	of winsocket's select only supported waiting for a max. of C<64> handles	5439
		5440	Early versions of winsocket's select only supported waiting for a maximum
3235	(probably owning to the fact that all windows kernels can only wait for	5441	of C<64> handles (probably owning to the fact that all windows kernels
3236	C<64> things at the same time internally; microsoft recommends spawning a	5442	can only wait for C<64> things at the same time internally; Microsoft
3237	chain of threads and wait for 63 handles and the previous thread in each).	5443	recommends spawning a chain of threads and wait for 63 handles and the
		5444	previous thread in each. Sounds great!).
3238		5445
3239	Newer versions support more handles, but you need to define C<FD_SETSIZE>	5446	Newer versions support more handles, but you need to define C<FD_SETSIZE>
3240	to some high number (e.g. C<2048>) before compiling the winsocket select	5447	to some high number (e.g. C<2048>) before compiling the winsocket select
3241	call (which might be in libev or elsewhere, for example, perl does its own	5448	call (which might be in libev or elsewhere, for example, perl and many
3242	select emulation on windows).	5449	other interpreters do their own select emulation on windows).
3243		5450
3244	Another limit is the number of file descriptors in the microsoft runtime	5451	Another limit is the number of file descriptors in the Microsoft runtime
3245	libraries, which by default is C<64> (there must be a hidden I<64> fetish	5452	libraries, which by default is C<64> (there must be a hidden I<64>
3246	or something like this inside microsoft). You can increase this by calling	5453	fetish or something like this inside Microsoft). You can increase this
3247	C<_setmaxstdio>, which can increase this limit to C<2048> (another	5454	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
3248	arbitrary limit), but is broken in many versions of the microsoft runtime	5455	(another arbitrary limit), but is broken in many versions of the Microsoft
3249	libraries.
3250
3251	This might get you to about C<512> or C<2048> sockets (depending on	5456	runtime libraries. This might get you to about C<512> or C<2048> sockets
3252	windows version and/or the phase of the moon). To get more, you need to	5457	(depending on windows version and/or the phase of the moon). To get more,
3253	wrap all I/O functions and provide your own fd management, but the cost of	5458	you need to wrap all I/O functions and provide your own fd management, but
3254	calling select (O(n²)) will likely make this unworkable.	5459	the cost of calling select (O(n²)) will likely make this unworkable.
		5460
		5461	=head2 PORTABILITY REQUIREMENTS
		5462
		5463	In addition to a working ISO-C implementation and of course the
		5464	backend-specific APIs, libev relies on a few additional extensions:
		5465
		5466	=over 4
		5467
		5468	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		5469	calling conventions regardless of C<ev_watcher_type *>.
		5470
		5471	Libev assumes not only that all watcher pointers have the same internal
		5472	structure (guaranteed by POSIX but not by ISO C for example), but it also
		5473	assumes that the same (machine) code can be used to call any watcher
		5474	callback: The watcher callbacks have different type signatures, but libev
		5475	calls them using an C<ev_watcher *> internally.
		5476
		5477	=item null pointers and integer zero are represented by 0 bytes
		5478
		5479	Libev uses C<memset> to initialise structs and arrays to C<0> bytes, and
		5480	relies on this setting pointers and integers to null.
		5481
		5482	=item pointer accesses must be thread-atomic
		5483
		5484	Accessing a pointer value must be atomic, it must both be readable and
		5485	writable in one piece - this is the case on all current architectures.
		5486
		5487	=item C<sig_atomic_t volatile> must be thread-atomic as well
		5488
		5489	The type C<sig_atomic_t volatile> (or whatever is defined as
		5490	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		5491	threads. This is not part of the specification for C<sig_atomic_t>, but is
		5492	believed to be sufficiently portable.
		5493
		5494	=item C<sigprocmask> must work in a threaded environment
		5495
		5496	Libev uses C<sigprocmask> to temporarily block signals. This is not
		5497	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		5498	pthread implementations will either allow C<sigprocmask> in the "main
		5499	thread" or will block signals process-wide, both behaviours would
		5500	be compatible with libev. Interaction between C<sigprocmask> and
		5501	C<pthread_sigmask> could complicate things, however.
		5502
		5503	The most portable way to handle signals is to block signals in all threads
		5504	except the initial one, and run the signal handling loop in the initial
		5505	thread as well.
		5506
		5507	=item C<long> must be large enough for common memory allocation sizes
		5508
		5509	To improve portability and simplify its API, libev uses C<long> internally
		5510	instead of C<size_t> when allocating its data structures. On non-POSIX
		5511	systems (Microsoft...) this might be unexpectedly low, but is still at
		5512	least 31 bits everywhere, which is enough for hundreds of millions of
		5513	watchers.
		5514
		5515	=item C<double> must hold a time value in seconds with enough accuracy
		5516
		5517	The type C<double> is used to represent timestamps. It is required to
		5518	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		5519	good enough for at least into the year 4000 with millisecond accuracy
		5520	(the design goal for libev). This requirement is overfulfilled by
		5521	implementations using IEEE 754, which is basically all existing ones.
		5522
		5523	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5524	year 2255 (and millisecond accuracy till the year 287396 - by then, libev
		5525	is either obsolete or somebody patched it to use C<long double> or
		5526	something like that, just kidding).
3255		5527
3256	=back	5528	=back
3257		5529
		5530	If you know of other additional requirements drop me a note.
		5531
		5532
		5533	=head1 ALGORITHMIC COMPLEXITIES
		5534
		5535	In this section the complexities of (many of) the algorithms used inside
		5536	libev will be documented. For complexity discussions about backends see
		5537	the documentation for C<ev_default_init>.
		5538
		5539	All of the following are about amortised time: If an array needs to be
		5540	extended, libev needs to realloc and move the whole array, but this
		5541	happens asymptotically rarer with higher number of elements, so O(1) might
		5542	mean that libev does a lengthy realloc operation in rare cases, but on
		5543	average it is much faster and asymptotically approaches constant time.
		5544
		5545	=over 4
		5546
		5547	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		5548
		5549	This means that, when you have a watcher that triggers in one hour and
		5550	there are 100 watchers that would trigger before that, then inserting will
		5551	have to skip roughly seven (C<ld 100>) of these watchers.
		5552
		5553	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		5554
		5555	That means that changing a timer costs less than removing/adding them,
		5556	as only the relative motion in the event queue has to be paid for.
		5557
		5558	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		5559
		5560	These just add the watcher into an array or at the head of a list.
		5561
		5562	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		5563
		5564	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		5565
		5566	These watchers are stored in lists, so they need to be walked to find the
		5567	correct watcher to remove. The lists are usually short (you don't usually
		5568	have many watchers waiting for the same fd or signal: one is typical, two
		5569	is rare).
		5570
		5571	=item Finding the next timer in each loop iteration: O(1)
		5572
		5573	By virtue of using a binary or 4-heap, the next timer is always found at a
		5574	fixed position in the storage array.
		5575
		5576	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		5577
		5578	A change means an I/O watcher gets started or stopped, which requires
		5579	libev to recalculate its status (and possibly tell the kernel, depending
		5580	on backend and whether C<ev_io_set> was used).
		5581
		5582	=item Activating one watcher (putting it into the pending state): O(1)
		5583
		5584	=item Priority handling: O(number_of_priorities)
		5585
		5586	Priorities are implemented by allocating some space for each
		5587	priority. When doing priority-based operations, libev usually has to
		5588	linearly search all the priorities, but starting/stopping and activating
		5589	watchers becomes O(1) with respect to priority handling.
		5590
		5591	=item Sending an ev_async: O(1)
		5592
		5593	=item Processing ev_async_send: O(number_of_async_watchers)
		5594
		5595	=item Processing signals: O(max_signal_number)
		5596
		5597	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5598	calls in the current loop iteration and the loop is currently
		5599	blocked. Checking for async and signal events involves iterating over all
		5600	running async watchers or all signal numbers.
		5601
		5602	=back
		5603
		5604
		5605	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5606
		5607	The major version 4 introduced some incompatible changes to the API.
		5608
		5609	At the moment, the C<ev.h> header file provides compatibility definitions
		5610	for all changes, so most programs should still compile. The compatibility
		5611	layer might be removed in later versions of libev, so better update to the
		5612	new API early than late.
		5613
		5614	=over 4
		5615
		5616	=item C<EV_COMPAT3> backwards compatibility mechanism
		5617
		5618	The backward compatibility mechanism can be controlled by
		5619	C<EV_COMPAT3>. See L</"PREPROCESSOR SYMBOLS/MACROS"> in the L</EMBEDDING>
		5620	section.
		5621
		5622	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5623
		5624	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5625
		5626	ev_loop_destroy (EV_DEFAULT_UC);
		5627	ev_loop_fork (EV_DEFAULT);
		5628
		5629	=item function/symbol renames
		5630
		5631	A number of functions and symbols have been renamed:
		5632
		5633	ev_loop => ev_run
		5634	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5635	EVLOOP_ONESHOT => EVRUN_ONCE
		5636
		5637	ev_unloop => ev_break
		5638	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5639	EVUNLOOP_ONE => EVBREAK_ONE
		5640	EVUNLOOP_ALL => EVBREAK_ALL
		5641
		5642	EV_TIMEOUT => EV_TIMER
		5643
		5644	ev_loop_count => ev_iteration
		5645	ev_loop_depth => ev_depth
		5646	ev_loop_verify => ev_verify
		5647
		5648	Most functions working on C<struct ev_loop> objects don't have an
		5649	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5650	associated constants have been renamed to not collide with the C<struct
		5651	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5652	as all other watcher types. Note that C<ev_loop_fork> is still called
		5653	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5654	typedef.
		5655
		5656	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5657
		5658	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5659	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5660	and work, but the library code will of course be larger.
		5661
		5662	=back
		5663
		5664
		5665	=head1 GLOSSARY
		5666
		5667	=over 4
		5668
		5669	=item active
		5670
		5671	A watcher is active as long as it has been started and not yet stopped.
		5672	See L</WATCHER STATES> for details.
		5673
		5674	=item application
		5675
		5676	In this document, an application is whatever is using libev.
		5677
		5678	=item backend
		5679
		5680	The part of the code dealing with the operating system interfaces.
		5681
		5682	=item callback
		5683
		5684	The address of a function that is called when some event has been
		5685	detected. Callbacks are being passed the event loop, the watcher that
		5686	received the event, and the actual event bitset.
		5687
		5688	=item callback/watcher invocation
		5689
		5690	The act of calling the callback associated with a watcher.
		5691
		5692	=item event
		5693
		5694	A change of state of some external event, such as data now being available
		5695	for reading on a file descriptor, time having passed or simply not having
		5696	any other events happening anymore.
		5697
		5698	In libev, events are represented as single bits (such as C<EV_READ> or
		5699	C<EV_TIMER>).
		5700
		5701	=item event library
		5702
		5703	A software package implementing an event model and loop.
		5704
		5705	=item event loop
		5706
		5707	An entity that handles and processes external events and converts them
		5708	into callback invocations.
		5709
		5710	=item event model
		5711
		5712	The model used to describe how an event loop handles and processes
		5713	watchers and events.
		5714
		5715	=item pending
		5716
		5717	A watcher is pending as soon as the corresponding event has been
		5718	detected. See L</WATCHER STATES> for details.
		5719
		5720	=item real time
		5721
		5722	The physical time that is observed. It is apparently strictly monotonic :)
		5723
		5724	=item wall-clock time
		5725
		5726	The time and date as shown on clocks. Unlike real time, it can actually
		5727	be wrong and jump forwards and backwards, e.g. when you adjust your
		5728	clock.
		5729
		5730	=item watcher
		5731
		5732	A data structure that describes interest in certain events. Watchers need
		5733	to be started (attached to an event loop) before they can receive events.
		5734
		5735	=back
3258		5736
3259	=head1 AUTHOR	5737	=head1 AUTHOR
3260		5738
3261	Marc Lehmann <libev@schmorp.de>.	5739	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5740	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
3262		5741

Diff Legend

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