ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.30 by root, Fri Nov 23 04:36:03 2007 UTC vs.
Revision 1.461 by root, Wed Jan 22 12:15:52 2020 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	=head1 DESCRIPTION	11	=head2 EXAMPLE PROGRAM
		12
		13	// a single header file is required
		14	#include <ev.h>
		15
		16	#include <stdio.h> // for puts
		17
		18	// every watcher type has its own typedef'd struct
		19	// with the name ev_TYPE
		20	ev_io stdin_watcher;
		21	ev_timer timeout_watcher;
		22
		23	// all watcher callbacks have a similar signature
		24	// this callback is called when data is readable on stdin
		25	static void
		26	stdin_cb (EV_P_ ev_io *w, int revents)
		27	{
		28	puts ("stdin ready");
		29	// for one-shot events, one must manually stop the watcher
		30	// with its corresponding stop function.
		31	ev_io_stop (EV_A_ w);
		32
		33	// this causes all nested ev_run's to stop iterating
		34	ev_break (EV_A_ EVBREAK_ALL);
		35	}
		36
		37	// another callback, this time for a time-out
		38	static void
		39	timeout_cb (EV_P_ ev_timer *w, int revents)
		40	{
		41	puts ("timeout");
		42	// this causes the innermost ev_run to stop iterating
		43	ev_break (EV_A_ EVBREAK_ONE);
		44	}
		45
		46	int
		47	main (void)
		48	{
		49	// use the default event loop unless you have special needs
		50	struct ev_loop *loop = EV_DEFAULT;
		51
		52	// initialise an io watcher, then start it
		53	// this one will watch for stdin to become readable
		54	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
		55	ev_io_start (loop, &stdin_watcher);
		56
		57	// initialise a timer watcher, then start it
		58	// simple non-repeating 5.5 second timeout
		59	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		60	ev_timer_start (loop, &timeout_watcher);
		61
		62	// now wait for events to arrive
		63	ev_run (loop, 0);
		64
		65	// break was called, so exit
		66	return 0;
		67	}
		68
		69	=head1 ABOUT THIS DOCUMENT
		70
		71	This document documents the libev software package.
		72
		73	The newest version of this document is also available as an html-formatted
		74	web page you might find easier to navigate when reading it for the first
		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
10		94
11	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
12	file descriptor being readable or a timeout occuring), and it will manage	96	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	97	these event sources and provide your program with events.
14		98
15	To do this, it must take more or less complete control over your process	99	To do this, it must take more or less complete control over your process
16	(or thread) by executing the I<event loop> handler, and will then	100	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	101	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	103	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	104	watchers>, which are relatively small C structures you initialise with the
21	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
22	watcher.	106	watcher.
23		107
24	=head1 FEATURES	108	=head2 FEATURES
25		109
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	110	Libev supports C<select>, C<poll>, the Linux-specific aio and C<epoll>
27	kqueue mechanisms for file descriptor events, relative timers, absolute	111	interfaces, the BSD-specific C<kqueue> and the Solaris-specific event port
28	timers with customised rescheduling, signal events, process status change	112	mechanisms for file descriptor events (C<ev_io>), the Linux C<inotify>
		113	interface (for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
		114	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
		115	timers (C<ev_timer>), absolute timers with customised rescheduling
		116	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
29	events (related to SIGCHLD), and event watchers dealing with the event	117	change events (C<ev_child>), and event watchers dealing with the event
30	loop mechanism itself (idle, prepare and check watchers). It also is quite	118	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
		119	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		120	limited support for fork events (C<ev_fork>).
		121
		122	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	123	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	124	for example).
33		125
34	=head1 CONVENTIONS	126	=head2 CONVENTIONS
35		127
36	Libev is very configurable. In this manual the default configuration	128	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	129	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	130	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	131	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	132	for multiple event loops, then all functions taking an initial argument of
41	argument of name C<loop> (which is always of type C<struct ev_loop *>)	133	name C<loop> (which is always of type C<struct ev_loop *>) will not have
42	will not have this argument.	134	this argument.
43		135
44	=head1 TIME REPRESENTATION	136	=head2 TIME REPRESENTATION
45		137
46	Libev represents time as a single floating point number, representing the	138	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	139	the (fractional) number of seconds since the (POSIX) epoch (in practice
48	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
49	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
50	to the double type in C.	142	too. It usually aliases to the C<double> type in C. When you need to do
		143	any calculations on it, you should treat it as some floating point value.
		144
		145	Unlike the name component C<stamp> might indicate, it is also used for
		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
51		172
52	=head1 GLOBAL FUNCTIONS	173	=head1 GLOBAL FUNCTIONS
53		174
54	These functions can be called anytime, even before initialising the	175	These functions can be called anytime, even before initialising the
55	library in any way.	176	library in any way.
…		…
58		179
59	=item ev_tstamp ev_time ()	180	=item ev_tstamp ev_time ()
60		181
61	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
62	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
63	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>.
		186
		187	=item ev_sleep (ev_tstamp interval)
		188
		189	Sleep for the given interval: The current thread will be blocked
		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
		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 >>).
64		198
65	=item int ev_version_major ()	199	=item int ev_version_major ()
66		200
67	=item int ev_version_minor ()	201	=item int ev_version_minor ()
68		202
69	You can find out the major and minor version numbers of the library	203	You can find out the major and minor ABI version numbers of the library
70	you linked against by calling the functions C<ev_version_major> and	204	you linked against by calling the functions C<ev_version_major> and
71	C<ev_version_minor>. If you want, you can compare against the global	205	C<ev_version_minor>. If you want, you can compare against the global
72	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	206	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
73	version of the library your program was compiled against.	207	version of the library your program was compiled against.
74		208
		209	These version numbers refer to the ABI version of the library, not the
		210	release version.
		211
75	Usually, it's a good idea to terminate if the major versions mismatch,	212	Usually, it's a good idea to terminate if the major versions mismatch,
76	as this indicates an incompatible change. Minor versions are usually	213	as this indicates an incompatible change. Minor versions are usually
77	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
78	not a problem.	215	not a problem.
79		216
		217	Example: Make sure we haven't accidentally been linked against the wrong
		218	version (note, however, that this will not detect other ABI mismatches,
		219	such as LFS or reentrancy).
		220
		221	assert (("libev version mismatch",
		222	ev_version_major () == EV_VERSION_MAJOR
		223	&& ev_version_minor () >= EV_VERSION_MINOR));
		224
		225	=item unsigned int ev_supported_backends ()
		226
		227	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		228	value) compiled into this binary of libev (independent of their
		229	availability on the system you are running on). See C<ev_default_loop> for
		230	a description of the set values.
		231
		232	Example: make sure we have the epoll method, because yeah this is cool and
		233	a must have and can we have a torrent of it please!!!11
		234
		235	assert (("sorry, no epoll, no sex",
		236	ev_supported_backends () & EVBACKEND_EPOLL));
		237
		238	=item unsigned int ev_recommended_backends ()
		239
		240	Return the set of all backends compiled into this binary of libev and
		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
		243	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
		244	and will not be auto-detected unless you explicitly request it (assuming
		245	you know what you are doing). This is the set of backends that libev will
		246	probe for if you specify no backends explicitly.
		247
		248	=item unsigned int ev_embeddable_backends ()
		249
		250	Returns the set of backends that are embeddable in other event loops. This
		251	value is platform-specific but can include backends not available on the
		252	current system. To find which embeddable backends might be supported on
		253	the current system, you would need to look at C<ev_embeddable_backends ()
		254	& ev_supported_backends ()>, likewise for recommended ones.
		255
		256	See the description of C<ev_embed> watchers for more info.
		257
80	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	258	=item ev_set_allocator (void *(*cb)(void *ptr, long size) throw ())
81		259
82	Sets the allocation function to use (the prototype is similar to the	260	Sets the allocation function to use (the prototype is similar - the
83	realloc C function, the semantics are identical). It is used to allocate	261	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
84	and free memory (no surprises here). If it returns zero when memory	262	used to allocate and free memory (no surprises here). If it returns zero
85	needs to be allocated, the library might abort or take some potentially	263	when memory needs to be allocated (C<size != 0>), the library might abort
86	destructive action. The default is your system realloc function.	264	or take some potentially destructive action.
		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.
87		269
88	You could override this function in high-availability programs to, say,	270	You could override this function in high-availability programs to, say,
89	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,
90	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.
91		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
		288	Example: Replace the libev allocator with one that waits a bit and then
		289	retries.
		290
		291	static void *
		292	persistent_realloc (void *ptr, size_t size)
		293	{
		294	if (!size)
		295	{
		296	free (ptr);
		297	return 0;
		298	}
		299
		300	for (;;)
		301	{
		302	void *newptr = realloc (ptr, size);
		303
		304	if (newptr)
		305	return newptr;
		306
		307	sleep (60);
		308	}
		309	}
		310
		311	...
		312	ev_set_allocator (persistent_realloc);
		313
92	=item ev_set_syserr_cb (void (*cb)(const char *msg));	314	=item ev_set_syserr_cb (void (*cb)(const char *msg) throw ())
93		315
94	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
95	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
96	indicating the system call or subsystem causing the problem. If this	318	indicating the system call or subsystem causing the problem. If this
97	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
98	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
99	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
100	(such as abort).	322	(such as abort).
101		323
		324	Example: This is basically the same thing that libev does internally, too.
		325
		326	static void
		327	fatal_error (const char *msg)
		328	{
		329	perror (msg);
		330	abort ();
		331	}
		332
		333	...
		334	ev_set_syserr_cb (fatal_error);
		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
102	=back	349	=back
103		350
104	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	351	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
105		352
106	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
107	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
108	events, and dynamically created loops which do not.	355	libev 3 had an C<ev_loop> function colliding with the struct name).
109		356
110	If you use threads, a common model is to run the default event loop	357	The library knows two types of such loops, the I<default> loop, which
111	in your main thread (or in a separate thread) and for each thread you	358	supports child process events, and dynamically created event loops which
112	create, you also create another event loop. Libev itself does no locking	359	do not.
113	whatsoever, so if you mix calls to the same event loop in different
114	threads, make sure you lock (this is usually a bad idea, though, even if
115	done correctly, because it's hideous and inefficient).
116		360
117	=over 4	361	=over 4
118		362
119	=item struct ev_loop *ev_default_loop (unsigned int flags)	363	=item struct ev_loop *ev_default_loop (unsigned int flags)
120		364
121	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
122	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
123	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
124	flags).	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".
125		375
126	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
127	function.	377	function (or via the C<EV_DEFAULT> macro).
		378
		379	Note that this function is I<not> thread-safe, so if you want to use it
		380	from multiple threads, you have to employ some kind of mutex (note also
		381	that this case is unlikely, as loops cannot be shared easily between
		382	threads anyway).
		383
		384	The default loop is the only loop that can handle C<ev_child> watchers,
		385	and to do this, it always registers a handler for C<SIGCHLD>. If this is
		386	a problem for your application you can either create a dynamic loop with
		387	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
		388	C<SIGCHLD> signal handler I<after> calling 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.
128		408
129	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
130	backends to use, and is usually specified as 0 (or EVFLAG_AUTO).	410	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
131		411
132	It supports the following flags:	412	The following flags are supported:
133		413
134	=over 4	414	=over 4
135		415
136	=item C<EVFLAG_AUTO>	416	=item C<EVFLAG_AUTO>
137		417
138	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
139	thing, believe me).	419	thing, believe me).
140		420
141	=item C<EVFLAG_NOENV>	421	=item C<EVFLAG_NOENV>
142		422
143	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
144	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
145	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	425	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
146	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
147	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
148	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).
149		431
		432	=item C<EVFLAG_FORKCHECK>
		433
		434	Instead of calling C<ev_loop_fork> manually after a fork, you can also
		435	make libev check for a fork in each iteration by enabling this flag.
		436
		437	This works by calling C<getpid ()> on every iteration of the loop,
		438	and thus this might slow down your event loop if you do a lot of loop
		439	iterations and little real work, but is usually not noticeable (on my
		440	GNU/Linux system for example, C<getpid> is actually a simple 5-insn
		441	sequence without a system call and thus I<very> fast, but my GNU/Linux
		442	system also has C<pthread_atfork> which is even faster). (Update: glibc
		443	versions 2.25 apparently removed the C<getpid> optimisation again).
		444
		445	The big advantage of this flag is that you can forget about fork (and
		446	forget about forgetting to tell libev about forking, although you still
		447	have to ignore C<SIGPIPE>) when you use this flag.
		448
		449	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		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.
		495
150	=item C<EVMETHOD_SELECT> (value 1, portable select backend)	496	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
151		497
152	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
153	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,
154	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
155	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	501	using this backend. It doesn't scale too well (O(highest_fd)), but its
156	the fastest backend for a low number of fds.	502	usually the fastest backend for a low number of (low-numbered :) fds.
157		503
		504	To get good performance out of this backend you need a high amount of
		505	parallelism (most of the file descriptors should be busy). If you are
		506	writing a server, you should C<accept ()> in a loop to accept as many
		507	connections as possible during one iteration. You might also want to have
		508	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		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).
		514
158	=item C<EVMETHOD_POLL> (value 2, poll backend, available everywhere except on windows)	515	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
159		516
160	And this is your standard poll(2) backend. It's more complicated than	517	And this is your standard poll(2) backend. It's more complicated
161	select, but handles sparse fds better and has no artificial limit on the	518	than select, but handles sparse fds better and has no artificial
162	number of fds you can use (except it will slow down considerably with a	519	limit on the number of fds you can use (except it will slow down
163	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	520	considerably with a lot of inactive fds). It scales similarly to select,
		521	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		522	performance tips.
164		523
		524	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		525	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		526
165	=item C<EVMETHOD_EPOLL> (value 4, Linux)	527	=item C<EVBACKEND_EPOLL> (value 4, Linux)
166		528
		529	Use the Linux-specific epoll(7) interface (for both pre- and post-2.6.9
		530	kernels).
		531
167	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
168	but it scales phenomenally better. While poll and select usually scale like	533	it scales phenomenally better. While poll and select usually scale like
169	O(total_fds) where n is the total number of fds (or the highest fd), epoll scales	534	O(total_fds) where total_fds is the total number of fds (or the highest
170	either O(1) or O(active_fds).	535	fd), epoll scales either O(1) or O(active_fds).
171		536
		537	The epoll mechanism deserves honorable mention as the most misdesigned
		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...
		564
172	While stopping and starting an I/O watcher in the same iteration will	565	While stopping, setting and starting an I/O watcher in the same iteration
173	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
174	(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
175	best to avoid that. Also, dup()ed file descriptors might not work very	568	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
176	well if you register events for both fds.	569	file descriptors might not work very well if you register events for both
		570	file descriptors.
177		571
		572	Best performance from this backend is achieved by not unregistering all
		573	watchers for a file descriptor until it has been closed, if possible,
		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.
		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
		584	While nominally embeddable in other event loops, this feature is broken in
		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>.
		633
178	=item C<EVMETHOD_KQUEUE> (value 8, most BSD clones)	634	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
179		635
180	Kqueue deserves special mention, as at the time of this writing, it	636	Kqueue deserves special mention, as at the time this backend was
181	was broken on all BSDs except NetBSD (usually it doesn't work with	637	implemented, it was broken on all BSDs except NetBSD (usually it doesn't
182	anything but sockets and pipes, except on Darwin, where of course its	638	work reliably with anything but sockets and pipes, except on Darwin,
183	completely useless). For this reason its not being "autodetected" unless	639	where of course it's completely useless). Unlike epoll, however, whose
184	you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO).	640	brokenness is by design, these kqueue bugs can be (and mostly have been)
		641	fixed without API changes to existing programs. For this reason it's not
		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.
		645
		646	You still can embed kqueue into a normal poll or select backend and use it
		647	only for sockets (after having made sure that sockets work with kqueue on
		648	the target platform). See C<ev_embed> watchers for more info.
185		649
186	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
187	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
188	course). While starting and stopping an I/O watcher does not cause an	652	course). While stopping, setting and starting an I/O watcher does never
189	extra syscall as with epoll, it still adds up to four event changes per	653	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
190	incident, so its best to avoid that.	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
		656	drops fds silently in similarly hard-to-detect cases.
191		657
		658	This backend usually performs well under most conditions.
		659
		660	While nominally embeddable in other event loops, this doesn't work
		661	everywhere, so you might need to test for this. And since it is broken
		662	almost everywhere, you should only use it when you have a lot of sockets
		663	(for which it usually works), by embedding it into another event loop
		664	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		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>.
		670
192	=item C<EVMETHOD_DEVPOLL> (value 16, Solaris 8)	671	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
193		672
194	This is not implemented yet (and might never be).	673	This is not implemented yet (and might never be, unless you send me an
		674	implementation). According to reports, C</dev/poll> only supports sockets
		675	and is not embeddable, which would limit the usefulness of this backend
		676	immensely.
195		677
196	=item C<EVMETHOD_PORT> (value 32, Solaris 10)	678	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
197		679
198	This uses the Solaris 10 port mechanism. As with everything on Solaris,	680	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
199	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)).
200		682
		683	While this backend scales well, it requires one system call per active
		684	file descriptor per loop iteration. For small and medium numbers of file
		685	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		686	might perform better.
		687
		688	On the positive side, this backend actually performed fully to
		689	specification in all tests and is fully embeddable, which is a rare feat
		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>.
		705
201	=item C<EVMETHOD_ALL>	706	=item C<EVBACKEND_ALL>
202		707
203	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
204	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
205	C<EVMETHOD_ALL & ~EVMETHOD_KQUEUE>.	710	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		711
		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).
206		721
207	=back	722	=back
208		723
209	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,
210	backends will be tried (in the reverse order as given here). If none are	725	then only these backends will be tried (in the reverse order as listed
211	specified, most compiled-in backend will be tried, usually in reverse	726	here). If none are specified, all backends in C<ev_recommended_backends
212	order of their flag values :)	727	()> will be tried.
213		728
214	=item struct ev_loop *ev_loop_new (unsigned int flags)	729	Example: Try to create a event loop that uses epoll and nothing else.
215		730
216	Similar to C<ev_default_loop>, but always creates a new event loop that is	731	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
217	always distinct from the default loop. Unlike the default loop, it cannot	732	if (!epoller)
218	handle signal and child watchers, and attempts to do so will be greeted by	733	fatal ("no epoll found here, maybe it hides under your chair");
219	undefined behaviour (or a failed assertion if assertions are enabled).
220		734
221	=item ev_default_destroy ()	735	Example: Use whatever libev has to offer, but make sure that kqueue is
		736	used if available.
222		737
223	Destroys the default loop again (frees all memory and kernel state	738	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
224	etc.). This stops all registered event watchers (by not touching them in	739
225	any way whatsoever, although you cannot rely on this :).	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);
226		745
227	=item ev_loop_destroy (loop)	746	=item ev_loop_destroy (loop)
228		747
229	Like C<ev_default_destroy>, but destroys an event loop created by an	748	Destroys an event loop object (frees all memory and kernel state
230	earlier call to C<ev_loop_new>.	749	etc.). None of the active event watchers will be stopped in the normal
		750	sense, so e.g. C<ev_is_active> might still return true. It is your
		751	responsibility to either stop all watchers cleanly yourself I<before>
		752	calling this function, or cope with the fact afterwards (which is usually
		753	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		754	for example).
231		755
232	=item ev_default_fork ()	756	Note that certain global state, such as signal state (and installed signal
		757	handlers), will not be freed by this function, and related watchers (such
		758	as signal and child watchers) would need to be stopped manually.
233		759
		760	This function is normally used on loop objects allocated by
		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.
		766	If you need dynamically allocated loops it is better to use C<ev_loop_new>
		767	and C<ev_loop_destroy>.
		768
		769	=item ev_loop_fork (loop)
		770
		771	This function sets a flag that causes subsequent C<ev_run> iterations
234	This function reinitialises the kernel state for backends that have	772	to reinitialise the kernel state for backends that have one. Despite
235	one. Despite the name, you can call it anytime, but it makes most sense	773	the name, you can call it anytime you are allowed to start or stop
236	after forking, in either the parent or child process (or both, but that	774	watchers (except inside an C<ev_prepare> callback), but it makes most
237	again makes little sense).	775	sense after forking, in the child process. You I<must> call it (or use
		776	C<EVFLAG_FORKCHECK>) in the child before resuming or calling C<ev_run>.
238		777
239	You I<must> call this function in the child process after forking if and	778	In addition, if you want to reuse a loop (via this function or
240	only if you want to use the event library in both processes. If you just	779	C<EVFLAG_FORKCHECK>), you I<also> have to ignore C<SIGPIPE>.
241	fork+exec, you don't have to call it.	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.
		785
		786	On the other hand, you only need to call this function in the child
		787	process if and only if you want to use the event loop in the child. If
		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).
242		792
243	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
244	it just in case after a fork. To make this easy, the function will fit in	794	it just in case after a fork.
245	quite nicely into a call to C<pthread_atfork>:
246		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	...
247	pthread_atfork (0, 0, ev_default_fork);	806	pthread_atfork (0, 0, post_fork_child);
248		807
249	=item ev_loop_fork (loop)	808	=item int ev_is_default_loop (loop)
250		809
251	Like C<ev_default_fork>, but acts on an event loop created by	810	Returns true when the given loop is, in fact, the default loop, and false
252	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	811	otherwise.
253	after fork, and how you do this is entirely your own problem.
254		812
		813	=item unsigned int ev_iteration (loop)
		814
		815	Returns the current iteration count for the event loop, which is identical
		816	to the number of times libev did poll for new events. It starts at C<0>
		817	and happily wraps around with enough iterations.
		818
		819	This value can sometimes be useful as a generation counter of sorts (it
		820	"ticks" the number of loop iterations), as it roughly corresponds with
		821	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		822	prepare and check phases.
		823
255	=item unsigned int ev_method (loop)	824	=item unsigned int ev_depth (loop)
256		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.
		837
		838	=item unsigned int ev_backend (loop)
		839
257	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	840	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
258	use.	841	use.
259		842
260	=item ev_tstamp ev_now (loop)	843	=item ev_tstamp ev_now (loop)
261		844
262	Returns the current "event loop time", which is the time the event loop	845	Returns the current "event loop time", which is the time the event loop
263	got events and started processing them. This timestamp does not change	846	received events and started processing them. This timestamp does not
264	as long as callbacks are being processed, and this is also the base time	847	change as long as callbacks are being processed, and this is also the base
265	used for relative timers. You can treat it as the timestamp of the event	848	time used for relative timers. You can treat it as the timestamp of the
266	occuring (or more correctly, the mainloop finding out about it).	849	event occurring (or more correctly, libev finding out about it).
267		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
268	=item ev_loop (loop, int flags)	889	=item bool ev_run (loop, int flags)
269		890
270	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
271	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
272	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>.
273		896
274	If the flags argument is specified as 0, it will not return until either	897	If the flags argument is specified as C<0>, it will keep handling events
275	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.
276		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
		905	Please note that an explicit C<ev_break> is usually better than
		906	relying on all watchers to be stopped when deciding when a program has
		907	finished (especially in interactive programs), but having a program
		908	that automatically loops as long as it has to and no longer by virtue
		909	of relying on its watchers stopping correctly, that is truly a thing of
		910	beauty.
		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
277	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
278	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
279	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.
280		922
281	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
282	neccessary) and will handle those and any outstanding ones. It will block	924	necessary) and will handle those and any already outstanding ones. It
283	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
		926	be an event internal to libev itself, so there is no guarantee that a
		927	user-registered callback will be called), and will return after one
284	one iteration of the loop.	928	iteration of the loop.
285		929
286	This flags value could be used to implement alternative looping	930	This is useful if you are waiting for some external event in conjunction
287	constructs, but the C<prepare> and C<check> watchers provide a better and	931	with something not expressible using other libev watchers (i.e. "roll your
288	more generic mechanism.	932	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		933	usually a better approach for this kind of thing.
289		934
290	Here are the gory details of what 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):
291		938
292	1. If there are no active watchers (reference count is zero), return.	939	- Increment loop depth.
		940	- Reset the ev_break status.
		941	- Before the first iteration, call any pending watchers.
		942	LOOP:
		943	- If EVFLAG_FORKCHECK was used, check for a fork.
		944	- If a fork was detected (by any means), queue and call all fork watchers.
293	2. Queue and immediately call all prepare watchers.	945	- Queue and call all prepare watchers.
		946	- If ev_break was called, goto FINISH.
294	3. 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.
295	4. Update the kernel state with all outstanding changes.	949	- Update the kernel state with all outstanding changes.
296	5. Update the "event loop time".	950	- Update the "event loop time" (ev_now ()).
297	6. Calculate for how long to block.	951	- Calculate for how long to sleep or block, if at all
		952	(active idle watchers, EVRUN_NOWAIT or not having
		953	any active watchers at all will result in not sleeping).
		954	- Sleep if the I/O and timer collect interval say so.
		955	- Increment loop iteration counter.
298	7. Block the process, waiting for events.	956	- Block the process, waiting for any events.
		957	- Queue all outstanding I/O (fd) events.
299	8. Update the "event loop time" and do time jump handling.	958	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
300	9. Queue all outstanding timers.	959	- Queue all expired timers.
301	10. Queue all outstanding periodics.	960	- Queue all expired periodics.
302	11. If no events are pending now, queue all idle watchers.	961	- Queue all idle watchers with priority higher than that of pending events.
303	12. Queue all check watchers.	962	- Queue all check watchers.
304	13. 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).
305	14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	964	Signals and child watchers are implemented as I/O watchers, and will
306	was used, return, otherwise continue with step #1.	965	be handled here by queueing them when their watcher gets executed.
		966	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
		967	were used, or there are no active watchers, goto FINISH, otherwise
		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.
307		973
		974	Example: Queue some jobs and then loop until no events are outstanding
		975	anymore.
		976
		977	... queue jobs here, make sure they register event watchers as long
		978	... as they still have work to do (even an idle watcher will do..)
		979	ev_run (my_loop, 0);
		980	... jobs done or somebody called break. yeah!
		981
308	=item ev_unloop (loop, how)	982	=item ev_break (loop, how)
309		983
310	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
311	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
312	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
313	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.
		988
		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.
314		993
315	=item ev_ref (loop)	994	=item ev_ref (loop)
316		995
317	=item ev_unref (loop)	996	=item ev_unref (loop)
318		997
319	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
320	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
321	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.
322	a watcher you never unregister that should not keep C<ev_loop> from	1001
323	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
324	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
325	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
326	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
327	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
328	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
		1012	before stop> (but only if the watcher wasn't active before, or was active
		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).
		1016
		1017	Example: Create a signal watcher, but keep it from keeping C<ev_run>
		1018	running when nothing else is active.
		1019
		1020	ev_signal exitsig;
		1021	ev_signal_init (&exitsig, sig_cb, SIGINT);
		1022	ev_signal_start (loop, &exitsig);
		1023	ev_unref (loop);
		1024
		1025	Example: For some weird reason, unregister the above signal handler again.
		1026
		1027	ev_ref (loop);
		1028	ev_signal_stop (loop, &exitsig);
		1029
		1030	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		1031
		1032	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		1033
		1034	These advanced functions influence the time that libev will spend waiting
		1035	for events. Both time intervals are by default C<0>, meaning that libev
		1036	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		1037	latency.
		1038
		1039	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		1040	allows libev to delay invocation of I/O and timer/periodic callbacks
		1041	to increase efficiency of loop iterations (or to increase power-saving
		1042	opportunities).
		1043
		1044	The idea is that sometimes your program runs just fast enough to handle
		1045	one (or very few) event(s) per loop iteration. While this makes the
		1046	program responsive, it also wastes a lot of CPU time to poll for new
		1047	events, especially with backends like C<select ()> which have a high
		1048	overhead for the actual polling but can deliver many events at once.
		1049
		1050	By setting a higher I<io collect interval> you allow libev to spend more
		1051	time collecting I/O events, so you can handle more events per iteration,
		1052	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		1053	C<ev_timer>) will not be affected. Setting this to a non-null value will
		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).
		1058
		1059	Likewise, by setting a higher I<timeout collect interval> you allow libev
		1060	to spend more time collecting timeouts, at the expense of increased
		1061	latency/jitter/inexactness (the watcher callback will be called
		1062	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		1063	value will not introduce any overhead in libev.
		1064
		1065	Many (busy) programs can usually benefit by setting the I/O collect
		1066	interval to a value near C<0.1> or so, which is often enough for
		1067	interactive servers (of course not for games), likewise for timeouts. It
		1068	usually doesn't make much sense to set it to a lower value than C<0.01>,
		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.
329		1168
330	=back	1169	=back
331		1170
		1171
332	=head1 ANATOMY OF A WATCHER	1172	=head1 ANATOMY OF A WATCHER
333		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
334	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
335	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
336	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:
337		1182
338	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)
339	{	1184	{
340	ev_io_stop (w);	1185	ev_io_stop (w);
341	ev_unloop (loop, EVUNLOOP_ALL);	1186	ev_break (loop, EVBREAK_ALL);
342	}	1187	}
343		1188
344	struct ev_loop *loop = ev_default_loop (0);	1189	struct ev_loop *loop = ev_default_loop (0);
		1190
345	struct ev_io stdin_watcher;	1191	ev_io stdin_watcher;
		1192
346	ev_init (&stdin_watcher, my_cb);	1193	ev_init (&stdin_watcher, my_cb);
347	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1194	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
348	ev_io_start (loop, &stdin_watcher);	1195	ev_io_start (loop, &stdin_watcher);
		1196
349	ev_loop (loop, 0);	1197	ev_run (loop, 0);
350		1198
351	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
352	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
353	although this can sometimes be quite valid).	1201	stack).
354		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
355	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
356	(watcher *, callback)>, which expects a callback to be provided. This	1207	*, callback)>, which expects a callback to be provided. This callback is
357	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
358	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
359	is readable and/or writable).	1210	and/or writable).
360		1211
361	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 *, ...) >>
362	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
363	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<<
364	(watcher *, callback, ...) >>.	1215	ev_TYPE_init (watcher *, callback, ...) >>.
365		1216
366	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
367	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
368	*) >>), 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
369	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1220	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
370		1221
371	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
372	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
373	reinitialise it or call its set method.	1224	otherwise. Most specifically you must never reinitialise it or call its
374		1225	C<ev_TYPE_set> macro.
375	You can check whether an event is active by calling the C<ev_is_active
376	(watcher *)> macro. To see whether an event is outstanding (but the
377	callback for it has not been called yet) you can use the C<ev_is_pending
378	(watcher *)> macro.
379		1226
380	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
381	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
382	third argument.	1229	third argument.
383		1230
…		…
392	=item C<EV_WRITE>	1239	=item C<EV_WRITE>
393		1240
394	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
395	writable.	1242	writable.
396		1243
397	=item C<EV_TIMEOUT>	1244	=item C<EV_TIMER>
398		1245
399	The C<ev_timer> watcher has timed out.	1246	The C<ev_timer> watcher has timed out.
400		1247
401	=item C<EV_PERIODIC>	1248	=item C<EV_PERIODIC>
402		1249
…		…
408		1255
409	=item C<EV_CHILD>	1256	=item C<EV_CHILD>
410		1257
411	The pid specified in the C<ev_child> watcher has received a status change.	1258	The pid specified in the C<ev_child> watcher has received a status change.
412		1259
		1260	=item C<EV_STAT>
		1261
		1262	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		1263
413	=item C<EV_IDLE>	1264	=item C<EV_IDLE>
414		1265
415	The C<ev_idle> watcher has determined that you have nothing better to do.	1266	The C<ev_idle> watcher has determined that you have nothing better to do.
416		1267
417	=item C<EV_PREPARE>	1268	=item C<EV_PREPARE>
418		1269
419	=item C<EV_CHECK>	1270	=item C<EV_CHECK>
420		1271
421	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
422	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)
423	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
424	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
425	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
426	(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
427	C<ev_loop> from blocking).	1283	blocking).
		1284
		1285	=item C<EV_EMBED>
		1286
		1287	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1288
		1289	=item C<EV_FORK>
		1290
		1291	The event loop has been resumed in the child process after fork (see
		1292	C<ev_fork>).
		1293
		1294	=item C<EV_CLEANUP>
		1295
		1296	The event loop is about to be destroyed (see C<ev_cleanup>).
		1297
		1298	=item C<EV_ASYNC>
		1299
		1300	The given async watcher has been asynchronously notified (see C<ev_async>).
		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>).
428		1306
429	=item C<EV_ERROR>	1307	=item C<EV_ERROR>
430		1308
431	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
432	happen because the watcher could not be properly started because libev	1310	happen because the watcher could not be properly started because libev
433	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
434	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
435	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.
436		1318
437	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
438	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
439	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
440	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
441	programs, though, so beware.	1323	programs, though, as the fd could already be closed and reused for another
		1324	thing, so beware.
442		1325
443	=back	1326	=back
444		1327
445	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1328	=head2 GENERIC WATCHER FUNCTIONS
446		1329
447	Each watcher has, by default, a member C<void *data> that you can change	1330	=over 4
448	and read at any time, libev will completely ignore it. This can be used
449	to associate arbitrary data with your watcher. If you need more data and
450	don't want to allocate memory and store a pointer to it in that data
451	member, you can also "subclass" the watcher type and provide your own
452	data:
453		1331
454	struct my_io	1332	=item C<ev_init> (ev_TYPE *watcher, callback)
		1333
		1334	This macro initialises the generic portion of a watcher. The contents
		1335	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1336	the generic parts of the watcher are initialised, you I<need> to call
		1337	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1338	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1339	which rolls both calls into one.
		1340
		1341	You can reinitialise a watcher at any time as long as it has been stopped
		1342	(or never started) and there are no pending events outstanding.
		1343
		1344	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1345	int revents)>.
		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
		1353	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1354
		1355	This macro initialises the type-specific parts of a watcher. You need to
		1356	call C<ev_init> at least once before you call this macro, but you can
		1357	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1358	macro on a watcher that is active (it can be pending, however, which is a
		1359	difference to the C<ev_init> macro).
		1360
		1361	Although some watcher types do not have type-specific arguments
		1362	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1363
		1364	See C<ev_init>, above, for an example.
		1365
		1366	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1367
		1368	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1369	calls into a single call. This is the most convenient method to initialise
		1370	a watcher. The same limitations apply, of course.
		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
		1376	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1377
		1378	Starts (activates) the given watcher. Only active watchers will receive
		1379	events. If the watcher is already active nothing will happen.
		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
		1386	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1387
		1388	Stops the given watcher if active, and clears the pending status (whether
		1389	the watcher was active or not).
		1390
		1391	It is possible that stopped watchers are pending - for example,
		1392	non-repeating timers are being stopped when they become pending - but
		1393	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1394	pending. If you want to free or reuse the memory used by the watcher it is
		1395	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1396
		1397	=item bool ev_is_active (ev_TYPE *watcher)
		1398
		1399	Returns a true value iff the watcher is active (i.e. it has been started
		1400	and not yet been stopped). As long as a watcher is active you must not modify
		1401	it.
		1402
		1403	=item bool ev_is_pending (ev_TYPE *watcher)
		1404
		1405	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1406	events but its callback has not yet been invoked). As long as a watcher
		1407	is pending (but not active) you must not call an init function on it (but
		1408	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1409	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1410	it).
		1411
		1412	=item callback ev_cb (ev_TYPE *watcher)
		1413
		1414	Returns the callback currently set on the watcher.
		1415
		1416	=item ev_set_cb (ev_TYPE *watcher, callback)
		1417
		1418	Change the callback. You can change the callback at virtually any time
		1419	(modulo threads).
		1420
		1421	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1422
		1423	=item int ev_priority (ev_TYPE *watcher)
		1424
		1425	Set and query the priority of the watcher. The priority is a small
		1426	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1427	(default: C<-2>). Pending watchers with higher priority will be invoked
		1428	before watchers with lower priority, but priority will not keep watchers
		1429	from being executed (except for C<ev_idle> watchers).
		1430
		1431	If you need to suppress invocation when higher priority events are pending
		1432	you need to look at C<ev_idle> watchers, which provide this functionality.
		1433
		1434	You I<must not> change the priority of a watcher as long as it is active or
		1435	pending.
		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
		1441	The default priority used by watchers when no priority has been set is
		1442	always C<0>, which is supposed to not be too high and not be too low :).
		1443
		1444	See L</WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1445	priorities.
		1446
		1447	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1448
		1449	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1450	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1451	can deal with that fact, as both are simply passed through to the
		1452	callback.
		1453
		1454	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1455
		1456	If the watcher is pending, this function clears its pending status and
		1457	returns its C<revents> bitset (as if its callback was invoked). If the
		1458	watcher isn't pending it does nothing and returns C<0>.
		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
		1477	=back
		1478
		1479	See also the L</ASSOCIATING CUSTOM DATA WITH A WATCHER> and L</BUILDING YOUR
		1480	OWN COMPOSITE WATCHERS> idioms.
		1481
		1482	=head2 WATCHER STATES
		1483
		1484	There are various watcher states mentioned throughout this manual -
		1485	active, pending and so on. In this section these states and the rules to
		1486	transition between them will be described in more detail - and while these
		1487	rules might look complicated, they usually do "the right thing".
		1488
		1489	=over 4
		1490
		1491	=item initialised
		1492
		1493	Before a watcher can be registered with the event loop it has to be
		1494	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1495	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1496
		1497	In this state it is simply some block of memory that is suitable for
		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.
		1501
		1502	=item started/running/active
		1503
		1504	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1505	property of the event loop, and is actively waiting for events. While in
		1506	this state it cannot be accessed (except in a few documented ways), moved,
		1507	freed or anything else - the only legal thing is to keep a pointer to it,
		1508	and call libev functions on it that are documented to work on active watchers.
		1509
		1510	=item pending
		1511
		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.
		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
		1612	static void
		1613	io_cb (EV_P_ ev_io *w, int revents)
455	{	1614	{
456	struct ev_io io;	1615	// stop the I/O watcher, we received the event, but
457	int otherfd;	1616	// are not yet ready to handle it.
458	void *somedata;	1617	ev_io_stop (EV_A_ w);
459	struct whatever *mostinteresting;	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);
460	}	1623	}
461		1624
462	And since your callback will be called with a pointer to the watcher, you	1625	static void
463	can cast it back to your own type:	1626	idle_cb (EV_P_ ev_idle *w, int revents)
464
465	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
466	{	1627	{
467	struct my_io *w = (struct my_io *)w_;	1628	// actual processing
468	...	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);
469	}	1634	}
470		1635
471	More interesting and less C-conformant ways of catsing your callback type	1636	// initialisation
472	have been omitted....	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.
473		1646
474		1647
475	=head1 WATCHER TYPES	1648	=head1 WATCHER TYPES
476		1649
477	This section describes each watcher in detail, but will not repeat	1650	This section describes each watcher in detail, but will not repeat
478	information given in the last section.	1651	information given in the last section. Any initialisation/set macros,
		1652	functions and members specific to the watcher type are explained.
479		1653
		1654	Most members are additionally marked with either I<[read-only]>, meaning
		1655	that, while the watcher is active, you can look at the member and expect
		1656	some sensible content, but you must not modify it (you can modify it while
		1657	the watcher is stopped to your hearts content), or I<[read-write]>, which
		1658	means you can expect it to have some sensible content while the watcher
		1659	is active, but you can also modify it. Modifying it may not do something
		1660	sensible or take immediate effect (or do anything at all), but libev will
		1661	not crash or malfunction in any way.
		1662
		1663	In any case, the documentation for each member will explain what the
		1664	effects are, and if there are any additional access restrictions.
		1665
480	=head2 C<ev_io> - is this file descriptor readable or writable	1666	=head2 C<ev_io> - is this file descriptor readable or writable?
481		1667
482	I/O watchers check whether a file descriptor is readable or writable	1668	I/O watchers check whether a file descriptor is readable or writable
483	in each iteration of the event loop (This behaviour is called	1669	in each iteration of the event loop, or, more precisely, when reading
484	level-triggering because you keep receiving events as long as the	1670	would not block the process and writing would at least be able to write
485	condition persists. Remember you can stop the watcher if you don't want to	1671	some data. This behaviour is called level-triggering because you keep
486	act on the event and neither want to receive future events).	1672	receiving events as long as the condition persists. Remember you can stop
		1673	the watcher if you don't want to act on the event and neither want to
		1674	receive future events.
487		1675
488	In general you can register as many read and/or write event watchers per	1676	In general you can register as many read and/or write event watchers per
489	fd as you want (as long as you don't confuse yourself). Setting all file	1677	fd as you want (as long as you don't confuse yourself). Setting all file
490	descriptors to non-blocking mode is also usually a good idea (but not	1678	descriptors to non-blocking mode is also usually a good idea (but not
491	required if you know what you are doing).	1679	required if you know what you are doing).
492		1680
493	You have to be careful with dup'ed file descriptors, though. Some backends	1681	Another thing you have to watch out for is that it is quite easy to
494	(the linux epoll backend is a notable example) cannot handle dup'ed file	1682	receive "spurious" readiness notifications, that is, your callback might
495	descriptors correctly if you register interest in two or more fds pointing	1683	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
496	to the same underlying file/socket etc. description (that is, they share	1684	because there is no data. It is very easy to get into this situation even
497	the same underlying "file open").	1685	with a relatively standard program structure. Thus it is best to always
		1686	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
		1687	preferable to a program hanging until some data arrives.
498		1688
499	If you must do this, then force the use of a known-to-be-good backend	1689	If you cannot run the fd in non-blocking mode (for example you should
500	(at the time of this writing, this includes only EVMETHOD_SELECT and	1690	not play around with an Xlib connection), then you have to separately
501	EVMETHOD_POLL).	1691	re-test whether a file descriptor is really ready with a known-to-be good
		1692	interface such as poll (fortunately in the case of Xlib, it already does
		1693	this on its own, so its quite safe to use). Some people additionally
		1694	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1695	indefinitely.
		1696
		1697	But really, best use non-blocking mode.
		1698
		1699	=head3 The special problem of disappearing file descriptors
		1700
		1701	Some backends (e.g. kqueue, epoll, linuxaio) need to be told about closing
		1702	a file descriptor (either due to calling C<close> explicitly or any other
		1703	means, such as C<dup2>). The reason is that you register interest in some
		1704	file descriptor, but when it goes away, the operating system will silently
		1705	drop this interest. If another file descriptor with the same number then
		1706	is registered with libev, there is no efficient way to see that this is,
		1707	in fact, a different file descriptor.
		1708
		1709	To avoid having to explicitly tell libev about such cases, libev follows
		1710	the following policy: Each time C<ev_io_set> is being called, libev
		1711	will assume that this is potentially a new file descriptor, otherwise
		1712	it is assumed that the file descriptor stays the same. That means that
		1713	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1714	descriptor even if the file descriptor number itself did not change.
		1715
		1716	This is how one would do it normally anyway, the important point is that
		1717	the libev application should not optimise around libev but should leave
		1718	optimisations to libev.
		1719
		1720	=head3 The special problem of dup'ed file descriptors
		1721
		1722	Some backends (e.g. epoll), cannot register events for file descriptors,
		1723	but only events for the underlying file descriptions. That means when you
		1724	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1725	events for them, only one file descriptor might actually receive events.
		1726
		1727	There is no workaround possible except not registering events
		1728	for potentially C<dup ()>'ed file descriptors, or to resort to
		1729	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1730
		1731	=head3 The special problem of files
		1732
		1733	Many people try to use C<select> (or libev) on file descriptors
		1734	representing files, and expect it to become ready when their program
		1735	doesn't block on disk accesses (which can take a long time on their own).
		1736
		1737	However, this cannot ever work in the "expected" way - you get a readiness
		1738	notification as soon as the kernel knows whether and how much data is
		1739	there, and in the case of open files, that's always the case, so you
		1740	always get a readiness notification instantly, and your read (or possibly
		1741	write) will still block on the disk I/O.
		1742
		1743	Another way to view it is that in the case of sockets, pipes, character
		1744	devices and so on, there is another party (the sender) that delivers data
		1745	on its own, but in the case of files, there is no such thing: the disk
		1746	will not send data on its own, simply because it doesn't know what you
		1747	wish to read - you would first have to request some data.
		1748
		1749	Since files are typically not-so-well supported by advanced notification
		1750	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1751	to files, even though you should not use it. The reason for this is
		1752	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1753	usually a tty, often a pipe, but also sometimes files or special devices
		1754	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1755	F</dev/urandom>), and even though the file might better be served with
		1756	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1757	it "just works" instead of freezing.
		1758
		1759	So avoid file descriptors pointing to files when you know it (e.g. use
		1760	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1761	when you rarely read from a file instead of from a socket, and want to
		1762	reuse the same code path.
		1763
		1764	=head3 The special problem of fork
		1765
		1766	Some backends (epoll, kqueue, linuxaio, iouring) do not support C<fork ()>
		1767	at all or exhibit useless behaviour. Libev fully supports fork, but needs
		1768	to be told about it in the child if you want to continue to use it in the
		1769	child.
		1770
		1771	To support fork in your child processes, you have to call C<ev_loop_fork
		1772	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
		1773	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1774
		1775	=head3 The special problem of SIGPIPE
		1776
		1777	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1778	when writing to a pipe whose other end has been closed, your program gets
		1779	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1780	this is sensible behaviour, for daemons, this is usually undesirable.
		1781
		1782	So when you encounter spurious, unexplained daemon exits, make sure you
		1783	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1784	somewhere, as that would have given you a big clue).
		1785
		1786	=head3 The special problem of accept()ing when you can't
		1787
		1788	Many implementations of the POSIX C<accept> function (for example,
		1789	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1790	connection from the pending queue in all error cases.
		1791
		1792	For example, larger servers often run out of file descriptors (because
		1793	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1794	rejecting the connection, leading to libev signalling readiness on
		1795	the next iteration again (the connection still exists after all), and
		1796	typically causing the program to loop at 100% CPU usage.
		1797
		1798	Unfortunately, the set of errors that cause this issue differs between
		1799	operating systems, there is usually little the app can do to remedy the
		1800	situation, and no known thread-safe method of removing the connection to
		1801	cope with overload is known (to me).
		1802
		1803	One of the easiest ways to handle this situation is to just ignore it
		1804	- when the program encounters an overload, it will just loop until the
		1805	situation is over. While this is a form of busy waiting, no OS offers an
		1806	event-based way to handle this situation, so it's the best one can do.
		1807
		1808	A better way to handle the situation is to log any errors other than
		1809	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1810	messages, and continue as usual, which at least gives the user an idea of
		1811	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1812	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1813	usage.
		1814
		1815	If your program is single-threaded, then you could also keep a dummy file
		1816	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1817	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1818	close that fd, and create a new dummy fd. This will gracefully refuse
		1819	clients under typical overload conditions.
		1820
		1821	The last way to handle it is to simply log the error and C<exit>, as
		1822	is often done with C<malloc> failures, but this results in an easy
		1823	opportunity for a DoS attack.
		1824
		1825	=head3 Watcher-Specific Functions
502		1826
503	=over 4	1827	=over 4
504		1828
505	=item ev_io_init (ev_io *, callback, int fd, int events)	1829	=item ev_io_init (ev_io *, callback, int fd, int events)
506		1830
507	=item ev_io_set (ev_io *, int fd, int events)	1831	=item ev_io_set (ev_io *, int fd, int events)
508		1832
509	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1833	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
510	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1834	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
511	EV_WRITE> to receive the given events.	1835	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1836
		1837	=item ev_io_modify (ev_io *, int events)
		1838
		1839	Similar to C<ev_io_set>, but only changes the event mask. Using this might
		1840	be faster with some backends, as libev can assume that the C<fd> still
		1841	refers to the same underlying file description, something it cannot do
		1842	when using C<ev_io_set>.
		1843
		1844	=item int fd [no-modify]
		1845
		1846	The file descriptor being watched. While it can be read at any time, you
		1847	must not modify this member even when the watcher is stopped - always use
		1848	C<ev_io_set> for that.
		1849
		1850	=item int events [no-modify]
		1851
		1852	The set of events the fd is being watched for, among other flags. Remember
		1853	that this is a bit set - to test for C<EV_READ>, use C<< w->events &
		1854	EV_READ >>, and similarly for C<EV_WRITE>.
		1855
		1856	As with C<fd>, you must not modify this member even when the watcher is
		1857	stopped, always use C<ev_io_set> or C<ev_io_modify> for that.
512		1858
513	=back	1859	=back
514		1860
		1861	=head3 Examples
		1862
		1863	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1864	readable, but only once. Since it is likely line-buffered, you could
		1865	attempt to read a whole line in the callback.
		1866
		1867	static void
		1868	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1869	{
		1870	ev_io_stop (loop, w);
		1871	.. read from stdin here (or from w->fd) and handle any I/O errors
		1872	}
		1873
		1874	...
		1875	struct ev_loop *loop = ev_default_init (0);
		1876	ev_io stdin_readable;
		1877	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1878	ev_io_start (loop, &stdin_readable);
		1879	ev_run (loop, 0);
		1880
		1881
515	=head2 C<ev_timer> - relative and optionally recurring timeouts	1882	=head2 C<ev_timer> - relative and optionally repeating timeouts
516		1883
517	Timer watchers are simple relative timers that generate an event after a	1884	Timer watchers are simple relative timers that generate an event after a
518	given time, and optionally repeating in regular intervals after that.	1885	given time, and optionally repeating in regular intervals after that.
519		1886
520	The timers are based on real time, that is, if you register an event that	1887	The timers are based on real time, that is, if you register an event that
521	times out after an hour and you reset your system clock to last years	1888	times out after an hour and you reset your system clock to January last
522	time, it will still time out after (roughly) and hour. "Roughly" because	1889	year, it will still time out after (roughly) one hour. "Roughly" because
523	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1890	detecting time jumps is hard, and some inaccuracies are unavoidable (the
524	monotonic clock option helps a lot here).	1891	monotonic clock option helps a lot here).
		1892
		1893	The callback is guaranteed to be invoked only I<after> its timeout has
		1894	passed (not I<at>, so on systems with very low-resolution clocks this
		1895	might introduce a small delay, see "the special problem of being too
		1896	early", below). If multiple timers become ready during the same loop
		1897	iteration then the ones with earlier time-out values are invoked before
		1898	ones of the same priority with later time-out values (but this is no
		1899	longer true when a callback calls C<ev_run> recursively).
		1900
		1901	=head3 Be smart about timeouts
		1902
		1903	Many real-world problems involve some kind of timeout, usually for error
		1904	recovery. A typical example is an HTTP request - if the other side hangs,
		1905	you want to raise some error after a while.
		1906
		1907	What follows are some ways to handle this problem, from obvious and
		1908	inefficient to smart and efficient.
		1909
		1910	In the following, a 60 second activity timeout is assumed - a timeout that
		1911	gets reset to 60 seconds each time there is activity (e.g. each time some
		1912	data or other life sign was received).
		1913
		1914	=over 4
		1915
		1916	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1917
		1918	This is the most obvious, but not the most simple way: In the beginning,
		1919	start the watcher:
		1920
		1921	ev_timer_init (timer, callback, 60., 0.);
		1922	ev_timer_start (loop, timer);
		1923
		1924	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1925	and start it again:
		1926
		1927	ev_timer_stop (loop, timer);
		1928	ev_timer_set (timer, 60., 0.);
		1929	ev_timer_start (loop, timer);
		1930
		1931	This is relatively simple to implement, but means that each time there is
		1932	some activity, libev will first have to remove the timer from its internal
		1933	data structure and then add it again. Libev tries to be fast, but it's
		1934	still not a constant-time operation.
		1935
		1936	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1937
		1938	This is the easiest way, and involves using C<ev_timer_again> instead of
		1939	C<ev_timer_start>.
		1940
		1941	To implement this, configure an C<ev_timer> with a C<repeat> value
		1942	of C<60> and then call C<ev_timer_again> at start and each time you
		1943	successfully read or write some data. If you go into an idle state where
		1944	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1945	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1946
		1947	That means you can ignore both the C<ev_timer_start> function and the
		1948	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1949	member and C<ev_timer_again>.
		1950
		1951	At start:
		1952
		1953	ev_init (timer, callback);
		1954	timer->repeat = 60.;
		1955	ev_timer_again (loop, timer);
		1956
		1957	Each time there is some activity:
		1958
		1959	ev_timer_again (loop, timer);
		1960
		1961	It is even possible to change the time-out on the fly, regardless of
		1962	whether the watcher is active or not:
		1963
		1964	timer->repeat = 30.;
		1965	ev_timer_again (loop, timer);
		1966
		1967	This is slightly more efficient then stopping/starting the timer each time
		1968	you want to modify its timeout value, as libev does not have to completely
		1969	remove and re-insert the timer from/into its internal data structure.
		1970
		1971	It is, however, even simpler than the "obvious" way to do it.
		1972
		1973	=item 3. Let the timer time out, but then re-arm it as required.
		1974
		1975	This method is more tricky, but usually most efficient: Most timeouts are
		1976	relatively long compared to the intervals between other activity - in
		1977	our example, within 60 seconds, there are usually many I/O events with
		1978	associated activity resets.
		1979
		1980	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1981	but remember the time of last activity, and check for a real timeout only
		1982	within the callback:
		1983
		1984	ev_tstamp timeout = 60.;
		1985	ev_tstamp last_activity; // time of last activity
		1986	ev_timer timer;
		1987
		1988	static void
		1989	callback (EV_P_ ev_timer *w, int revents)
		1990	{
		1991	// calculate when the timeout would happen
		1992	ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
		1993
		1994	// if negative, it means we the timeout already occurred
		1995	if (after < 0.)
		1996	{
		1997	// timeout occurred, take action
		1998	}
		1999	else
		2000	{
		2001	// callback was invoked, but there was some recent
		2002	// activity. simply restart the timer to time out
		2003	// after "after" seconds, which is the earliest time
		2004	// the timeout can occur.
		2005	ev_timer_set (w, after, 0.);
		2006	ev_timer_start (EV_A_ w);
		2007	}
		2008	}
		2009
		2010	To summarise the callback: first calculate in how many seconds the
		2011	timeout will occur (by calculating the absolute time when it would occur,
		2012	C<last_activity + timeout>, and subtracting the current time, C<ev_now
		2013	(EV_A)> from that).
		2014
		2015	If this value is negative, then we are already past the timeout, i.e. we
		2016	timed out, and need to do whatever is needed in this case.
		2017
		2018	Otherwise, we now the earliest time at which the timeout would trigger,
		2019	and simply start the timer with this timeout value.
		2020
		2021	In other words, each time the callback is invoked it will check whether
		2022	the timeout occurred. If not, it will simply reschedule itself to check
		2023	again at the earliest time it could time out. Rinse. Repeat.
		2024
		2025	This scheme causes more callback invocations (about one every 60 seconds
		2026	minus half the average time between activity), but virtually no calls to
		2027	libev to change the timeout.
		2028
		2029	To start the machinery, simply initialise the watcher and set
		2030	C<last_activity> to the current time (meaning there was some activity just
		2031	now), then call the callback, which will "do the right thing" and start
		2032	the timer:
		2033
		2034	last_activity = ev_now (EV_A);
		2035	ev_init (&timer, callback);
		2036	callback (EV_A_ &timer, 0);
		2037
		2038	When there is some activity, simply store the current time in
		2039	C<last_activity>, no libev calls at all:
		2040
		2041	if (activity detected)
		2042	last_activity = ev_now (EV_A);
		2043
		2044	When your timeout value changes, then the timeout can be changed by simply
		2045	providing a new value, stopping the timer and calling the callback, which
		2046	will again do the right thing (for example, time out immediately :).
		2047
		2048	timeout = new_value;
		2049	ev_timer_stop (EV_A_ &timer);
		2050	callback (EV_A_ &timer, 0);
		2051
		2052	This technique is slightly more complex, but in most cases where the
		2053	time-out is unlikely to be triggered, much more efficient.
		2054
		2055	=item 4. Wee, just use a double-linked list for your timeouts.
		2056
		2057	If there is not one request, but many thousands (millions...), all
		2058	employing some kind of timeout with the same timeout value, then one can
		2059	do even better:
		2060
		2061	When starting the timeout, calculate the timeout value and put the timeout
		2062	at the I<end> of the list.
		2063
		2064	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		2065	the list is expected to fire (for example, using the technique #3).
		2066
		2067	When there is some activity, remove the timer from the list, recalculate
		2068	the timeout, append it to the end of the list again, and make sure to
		2069	update the C<ev_timer> if it was taken from the beginning of the list.
		2070
		2071	This way, one can manage an unlimited number of timeouts in O(1) time for
		2072	starting, stopping and updating the timers, at the expense of a major
		2073	complication, and having to use a constant timeout. The constant timeout
		2074	ensures that the list stays sorted.
		2075
		2076	=back
		2077
		2078	So which method the best?
		2079
		2080	Method #2 is a simple no-brain-required solution that is adequate in most
		2081	situations. Method #3 requires a bit more thinking, but handles many cases
		2082	better, and isn't very complicated either. In most case, choosing either
		2083	one is fine, with #3 being better in typical situations.
		2084
		2085	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		2086	rather complicated, but extremely efficient, something that really pays
		2087	off after the first million or so of active timers, i.e. it's usually
		2088	overkill :)
		2089
		2090	=head3 The special problem of being too early
		2091
		2092	If you ask a timer to call your callback after three seconds, then
		2093	you expect it to be invoked after three seconds - but of course, this
		2094	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		2095	guaranteed to any precision by libev - imagine somebody suspending the
		2096	process with a STOP signal for a few hours for example.
		2097
		2098	So, libev tries to invoke your callback as soon as possible I<after> the
		2099	delay has occurred, but cannot guarantee this.
		2100
		2101	A less obvious failure mode is calling your callback too early: many event
		2102	loops compare timestamps with a "elapsed delay >= requested delay", but
		2103	this can cause your callback to be invoked much earlier than you would
		2104	expect.
		2105
		2106	To see why, imagine a system with a clock that only offers full second
		2107	resolution (think windows if you can't come up with a broken enough OS
		2108	yourself). If you schedule a one-second timer at the time 500.9, then the
		2109	event loop will schedule your timeout to elapse at a system time of 500
		2110	(500.9 truncated to the resolution) + 1, or 501.
		2111
		2112	If an event library looks at the timeout 0.1s later, it will see "501 >=
		2113	501" and invoke the callback 0.1s after it was started, even though a
		2114	one-second delay was requested - this is being "too early", despite best
		2115	intentions.
		2116
		2117	This is the reason why libev will never invoke the callback if the elapsed
		2118	delay equals the requested delay, but only when the elapsed delay is
		2119	larger than the requested delay. In the example above, libev would only invoke
		2120	the callback at system time 502, or 1.1s after the timer was started.
		2121
		2122	So, while libev cannot guarantee that your callback will be invoked
		2123	exactly when requested, it I<can> and I<does> guarantee that the requested
		2124	delay has actually elapsed, or in other words, it always errs on the "too
		2125	late" side of things.
		2126
		2127	=head3 The special problem of time updates
		2128
		2129	Establishing the current time is a costly operation (it usually takes
		2130	at least one system call): EV therefore updates its idea of the current
		2131	time only before and after C<ev_run> collects new events, which causes a
		2132	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		2133	lots of events in one iteration.
525		2134
526	The relative timeouts are calculated relative to the C<ev_now ()>	2135	The relative timeouts are calculated relative to the C<ev_now ()>
527	time. This is usually the right thing as this timestamp refers to the time	2136	time. This is usually the right thing as this timestamp refers to the time
528	of the event triggering whatever timeout you are modifying/starting. If	2137	of the event triggering whatever timeout you are modifying/starting. If
529	you suspect event processing to be delayed and you I<need> to base the timeout	2138	you suspect event processing to be delayed and you I<need> to base the
530	on the current time, use something like this to adjust for this:	2139	timeout on the current time, use something like the following to adjust
		2140	for it:
531		2141
532	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2142	ev_timer_set (&timer, after + (ev_time () - ev_now ()), 0.);
533		2143
534	The callback is guarenteed to be invoked only when its timeout has passed,	2144	If the event loop is suspended for a long time, you can also force an
535	but if multiple timers become ready during the same loop iteration then	2145	update of the time returned by C<ev_now ()> by calling C<ev_now_update
536	order of execution is undefined.	2146	()>, although that will push the event time of all outstanding events
		2147	further into the future.
		2148
		2149	=head3 The special problem of unsynchronised clocks
		2150
		2151	Modern systems have a variety of clocks - libev itself uses the normal
		2152	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2153	jumps).
		2154
		2155	Neither of these clocks is synchronised with each other or any other clock
		2156	on the system, so C<ev_time ()> might return a considerably different time
		2157	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2158	a call to C<gettimeofday> might return a second count that is one higher
		2159	than a directly following call to C<time>.
		2160
		2161	The moral of this is to only compare libev-related timestamps with
		2162	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2163	a second or so.
		2164
		2165	One more problem arises due to this lack of synchronisation: if libev uses
		2166	the system monotonic clock and you compare timestamps from C<ev_time>
		2167	or C<ev_now> from when you started your timer and when your callback is
		2168	invoked, you will find that sometimes the callback is a bit "early".
		2169
		2170	This is because C<ev_timer>s work in real time, not wall clock time, so
		2171	libev makes sure your callback is not invoked before the delay happened,
		2172	I<measured according to the real time>, not the system clock.
		2173
		2174	If your timeouts are based on a physical timescale (e.g. "time out this
		2175	connection after 100 seconds") then this shouldn't bother you as it is
		2176	exactly the right behaviour.
		2177
		2178	If you want to compare wall clock/system timestamps to your timers, then
		2179	you need to use C<ev_periodic>s, as these are based on the wall clock
		2180	time, where your comparisons will always generate correct results.
		2181
		2182	=head3 The special problems of suspended animation
		2183
		2184	When you leave the server world it is quite customary to hit machines that
		2185	can suspend/hibernate - what happens to the clocks during such a suspend?
		2186
		2187	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		2188	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		2189	to run until the system is suspended, but they will not advance while the
		2190	system is suspended. That means, on resume, it will be as if the program
		2191	was frozen for a few seconds, but the suspend time will not be counted
		2192	towards C<ev_timer> when a monotonic clock source is used. The real time
		2193	clock advanced as expected, but if it is used as sole clocksource, then a
		2194	long suspend would be detected as a time jump by libev, and timers would
		2195	be adjusted accordingly.
		2196
		2197	I would not be surprised to see different behaviour in different between
		2198	operating systems, OS versions or even different hardware.
		2199
		2200	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2201	time jump in the monotonic clocks and the realtime clock. If the program
		2202	is suspended for a very long time, and monotonic clock sources are in use,
		2203	then you can expect C<ev_timer>s to expire as the full suspension time
		2204	will be counted towards the timers. When no monotonic clock source is in
		2205	use, then libev will again assume a timejump and adjust accordingly.
		2206
		2207	It might be beneficial for this latter case to call C<ev_suspend>
		2208	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2209	deterministic behaviour in this case (you can do nothing against
		2210	C<SIGSTOP>).
		2211
		2212	=head3 Watcher-Specific Functions and Data Members
537		2213
538	=over 4	2214	=over 4
539		2215
540	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2216	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
541		2217
542	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2218	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
543		2219
544	Configure the timer to trigger after C<after> seconds. If C<repeat> is	2220	Configure the timer to trigger after C<after> seconds (fractional and
545	C<0.>, then it will automatically be stopped. If it is positive, then the	2221	negative values are supported). If C<repeat> is C<0.>, then it will
		2222	automatically be stopped once the timeout is reached. If it is positive,
546	timer will automatically be configured to trigger again C<repeat> seconds	2223	then the timer will automatically be configured to trigger again C<repeat>
547	later, again, and again, until stopped manually.	2224	seconds later, again, and again, until stopped manually.
548		2225
549	The timer itself will do a best-effort at avoiding drift, that is, if you	2226	The timer itself will do a best-effort at avoiding drift, that is, if
550	configure a timer to trigger every 10 seconds, then it will trigger at	2227	you configure a timer to trigger every 10 seconds, then it will normally
551	exactly 10 second intervals. If, however, your program cannot keep up with	2228	trigger at exactly 10 second intervals. If, however, your program cannot
552	the timer (because it takes longer than those 10 seconds to do stuff) the	2229	keep up with the timer (because it takes longer than those 10 seconds to
553	timer will not fire more than once per event loop iteration.	2230	do stuff) the timer will not fire more than once per event loop iteration.
554		2231
555	=item ev_timer_again (loop)	2232	=item ev_timer_again (loop, ev_timer *)
556		2233
557	This will act as if the timer timed out and restart it again if it is	2234	This will act as if the timer timed out, and restarts it again if it is
558	repeating. The exact semantics are:	2235	repeating. It basically works like calling C<ev_timer_stop>, updating the
		2236	timeout to the C<repeat> value and calling C<ev_timer_start>.
559		2237
		2238	The exact semantics are as in the following rules, all of which will be
		2239	applied to the watcher:
		2240
		2241	=over 4
		2242
		2243	=item If the timer is pending, the pending status is always cleared.
		2244
560	If the timer is started but nonrepeating, stop it.	2245	=item If the timer is started but non-repeating, stop it (as if it timed
		2246	out, without invoking it).
561		2247
562	If the timer is repeating, either start it if necessary (with the repeat	2248	=item If the timer is repeating, make the C<repeat> value the new timeout
563	value), or reset the running timer to the repeat value.	2249	and start the timer, if necessary.
564
565	This sounds a bit complicated, but here is a useful and typical
566	example: Imagine you have a tcp connection and you want a so-called idle
567	timeout, that is, you want to be called when there have been, say, 60
568	seconds of inactivity on the socket. The easiest way to do this is to
569	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each
570	time you successfully read or write some data. If you go into an idle
571	state where you do not expect data to travel on the socket, you can stop
572	the timer, and again will automatically restart it if need be.
573		2250
574	=back	2251	=back
575		2252
		2253	This sounds a bit complicated, see L</Be smart about timeouts>, above, for a
		2254	usage example.
		2255
		2256	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
		2257
		2258	Returns the remaining time until a timer fires. If the timer is active,
		2259	then this time is relative to the current event loop time, otherwise it's
		2260	the timeout value currently configured.
		2261
		2262	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2263	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2264	will return C<4>. When the timer expires and is restarted, it will return
		2265	roughly C<7> (likely slightly less as callback invocation takes some time,
		2266	too), and so on.
		2267
		2268	=item ev_tstamp repeat [read-write]
		2269
		2270	The current C<repeat> value. Will be used each time the watcher times out
		2271	or C<ev_timer_again> is called, and determines the next timeout (if any),
		2272	which is also when any modifications are taken into account.
		2273
		2274	=back
		2275
		2276	=head3 Examples
		2277
		2278	Example: Create a timer that fires after 60 seconds.
		2279
		2280	static void
		2281	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2282	{
		2283	.. one minute over, w is actually stopped right here
		2284	}
		2285
		2286	ev_timer mytimer;
		2287	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		2288	ev_timer_start (loop, &mytimer);
		2289
		2290	Example: Create a timeout timer that times out after 10 seconds of
		2291	inactivity.
		2292
		2293	static void
		2294	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2295	{
		2296	.. ten seconds without any activity
		2297	}
		2298
		2299	ev_timer mytimer;
		2300	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		2301	ev_timer_again (&mytimer); /* start timer */
		2302	ev_run (loop, 0);
		2303
		2304	// and in some piece of code that gets executed on any "activity":
		2305	// reset the timeout to start ticking again at 10 seconds
		2306	ev_timer_again (&mytimer);
		2307
		2308
576	=head2 C<ev_periodic> - to cron or not to cron	2309	=head2 C<ev_periodic> - to cron or not to cron?
577		2310
578	Periodic watchers are also timers of a kind, but they are very versatile	2311	Periodic watchers are also timers of a kind, but they are very versatile
579	(and unfortunately a bit complex).	2312	(and unfortunately a bit complex).
580		2313
581	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2314	Unlike C<ev_timer>, periodic watchers are not based on real time (or
582	but on wallclock time (absolute time). You can tell a periodic watcher	2315	relative time, the physical time that passes) but on wall clock time
583	to trigger "at" some specific point in time. For example, if you tell a	2316	(absolute time, the thing you can read on your calendar or clock). The
584	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	2317	difference is that wall clock time can run faster or slower than real
585	+ 10.>) and then reset your system clock to the last year, then it will	2318	time, and time jumps are not uncommon (e.g. when you adjust your
586	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2319	wrist-watch).
587	roughly 10 seconds later and of course not if you reset your system time
588	again).
589		2320
590	They can also be used to implement vastly more complex timers, such as	2321	You can tell a periodic watcher to trigger after some specific point
		2322	in time: for example, if you tell a periodic watcher to trigger "in 10
		2323	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2324	not a delay) and then reset your system clock to January of the previous
		2325	year, then it will take a year or more to trigger the event (unlike an
		2326	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2327	it, as it uses a relative timeout).
		2328
		2329	C<ev_periodic> watchers can also be used to implement vastly more complex
591	triggering an event on eahc midnight, local time.	2330	timers, such as triggering an event on each "midnight, local time", or
		2331	other complicated rules. This cannot easily be done with C<ev_timer>
		2332	watchers, as those cannot react to time jumps.
592		2333
593	As with timers, the callback is guarenteed to be invoked only when the	2334	As with timers, the callback is guaranteed to be invoked only when the
594	time (C<at>) has been passed, but if multiple periodic timers become ready	2335	point in time where it is supposed to trigger has passed. If multiple
595	during the same loop iteration then order of execution is undefined.	2336	timers become ready during the same loop iteration then the ones with
		2337	earlier time-out values are invoked before ones with later time-out values
		2338	(but this is no longer true when a callback calls C<ev_run> recursively).
		2339
		2340	=head3 Watcher-Specific Functions and Data Members
596		2341
597	=over 4	2342	=over 4
598		2343
599	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2344	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
600		2345
601	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2346	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
602		2347
603	Lots of arguments, lets sort it out... There are basically three modes of	2348	Lots of arguments, let's sort it out... There are basically three modes of
604	operation, and we will explain them from simplest to complex:	2349	operation, and we will explain them from simplest to most complex:
605		2350
606	=over 4	2351	=over 4
607		2352
608	=item * absolute timer (interval = reschedule_cb = 0)	2353	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
609		2354
610	In this configuration the watcher triggers an event at the wallclock time	2355	In this configuration the watcher triggers an event after the wall clock
611	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2356	time C<offset> has passed. It will not repeat and will not adjust when a
612	that is, if it is to be run at January 1st 2011 then it will run when the	2357	time jump occurs, that is, if it is to be run at January 1st 2011 then it
613	system time reaches or surpasses this time.	2358	will be stopped and invoked when the system clock reaches or surpasses
		2359	this point in time.
614		2360
615	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	2361	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
616		2362
617	In this mode the watcher will always be scheduled to time out at the next	2363	In this mode the watcher will always be scheduled to time out at the next
618	C<at + N * interval> time (for some integer N) and then repeat, regardless	2364	C<offset + N * interval> time (for some integer N, which can also be
619	of any time jumps.	2365	negative) and then repeat, regardless of any time jumps. The C<offset>
		2366	argument is merely an offset into the C<interval> periods.
620		2367
621	This can be used to create timers that do not drift with respect to system	2368	This can be used to create timers that do not drift with respect to the
622	time:	2369	system clock, for example, here is an C<ev_periodic> that triggers each
		2370	hour, on the hour (with respect to UTC):
623		2371
624	ev_periodic_set (&periodic, 0., 3600., 0);	2372	ev_periodic_set (&periodic, 0., 3600., 0);
625		2373
626	This doesn't mean there will always be 3600 seconds in between triggers,	2374	This doesn't mean there will always be 3600 seconds in between triggers,
627	but only that the the callback will be called when the system time shows a	2375	but only that the callback will be called when the system time shows a
628	full hour (UTC), or more correctly, when the system time is evenly divisible	2376	full hour (UTC), or more correctly, when the system time is evenly divisible
629	by 3600.	2377	by 3600.
630		2378
631	Another way to think about it (for the mathematically inclined) is that	2379	Another way to think about it (for the mathematically inclined) is that
632	C<ev_periodic> will try to run the callback in this mode at the next possible	2380	C<ev_periodic> will try to run the callback in this mode at the next possible
633	time where C<time = at (mod interval)>, regardless of any time jumps.	2381	time where C<time = offset (mod interval)>, regardless of any time jumps.
634		2382
		2383	The C<interval> I<MUST> be positive, and for numerical stability, the
		2384	interval value should be higher than C<1/8192> (which is around 100
		2385	microseconds) and C<offset> should be higher than C<0> and should have
		2386	at most a similar magnitude as the current time (say, within a factor of
		2387	ten). Typical values for offset are, in fact, C<0> or something between
		2388	C<0> and C<interval>, which is also the recommended range.
		2389
		2390	Note also that there is an upper limit to how often a timer can fire (CPU
		2391	speed for example), so if C<interval> is very small then timing stability
		2392	will of course deteriorate. Libev itself tries to be exact to be about one
		2393	millisecond (if the OS supports it and the machine is fast enough).
		2394
635	=item * manual reschedule mode (reschedule_cb = callback)	2395	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
636		2396
637	In this mode the values for C<interval> and C<at> are both being	2397	In this mode the values for C<interval> and C<offset> are both being
638	ignored. Instead, each time the periodic watcher gets scheduled, the	2398	ignored. Instead, each time the periodic watcher gets scheduled, the
639	reschedule callback will be called with the watcher as first, and the	2399	reschedule callback will be called with the watcher as first, and the
640	current time as second argument.	2400	current time as second argument.
641		2401
642	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2402	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
643	ever, or make any event loop modifications>. If you need to stop it,	2403	or make ANY other event loop modifications whatsoever, unless explicitly
644	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2404	allowed by documentation here>.
645	starting a prepare watcher).
646		2405
		2406	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2407	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2408	only event loop modification you are allowed to do).
		2409
647	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2410	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
648	ev_tstamp now)>, e.g.:	2411	*w, ev_tstamp now)>, e.g.:
649		2412
		2413	static ev_tstamp
650	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2414	my_rescheduler (ev_periodic *w, ev_tstamp now)
651	{	2415	{
652	return now + 60.;	2416	return now + 60.;
653	}	2417	}
654		2418
655	It must return the next time to trigger, based on the passed time value	2419	It must return the next time to trigger, based on the passed time value
656	(that is, the lowest time value larger than to the second argument). It	2420	(that is, the lowest time value larger than to the second argument). It
657	will usually be called just before the callback will be triggered, but	2421	will usually be called just before the callback will be triggered, but
658	might be called at other times, too.	2422	might be called at other times, too.
659		2423
660	NOTE: I<< This callback must always return a time that is later than the	2424	NOTE: I<< This callback must always return a time that is higher than or
661	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2425	equal to the passed C<now> value >>.
662		2426
663	This can be used to create very complex timers, such as a timer that	2427	This can be used to create very complex timers, such as a timer that
664	triggers on each midnight, local time. To do this, you would calculate the	2428	triggers on "next midnight, local time". To do this, you would calculate
665	next midnight after C<now> and return the timestamp value for this. How	2429	the next midnight after C<now> and return the timestamp value for
666	you do this is, again, up to you (but it is not trivial, which is the main	2430	this. Here is a (completely untested, no error checking) example on how to
667	reason I omitted it as an example).	2431	do this:
		2432
		2433	#include <time.h>
		2434
		2435	static ev_tstamp
		2436	my_rescheduler (ev_periodic *w, ev_tstamp now)
		2437	{
		2438	time_t tnow = (time_t)now;
		2439	struct tm tm;
		2440	localtime_r (&tnow, &tm);
		2441
		2442	tm.tm_sec = tm.tm_min = tm.tm_hour = 0; // midnight current day
		2443	++tm.tm_mday; // midnight next day
		2444
		2445	return mktime (&tm);
		2446	}
		2447
		2448	Note: this code might run into trouble on days that have more then two
		2449	midnights (beginning and end).
668		2450
669	=back	2451	=back
670		2452
671	=item ev_periodic_again (loop, ev_periodic *)	2453	=item ev_periodic_again (loop, ev_periodic *)
672		2454
673	Simply stops and restarts the periodic watcher again. This is only useful	2455	Simply stops and restarts the periodic watcher again. This is only useful
674	when you changed some parameters or the reschedule callback would return	2456	when you changed some parameters or the reschedule callback would return
675	a different time than the last time it was called (e.g. in a crond like	2457	a different time than the last time it was called (e.g. in a crond like
676	program when the crontabs have changed).	2458	program when the crontabs have changed).
677		2459
		2460	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2461
		2462	When active, returns the absolute time that the watcher is supposed
		2463	to trigger next. This is not the same as the C<offset> argument to
		2464	C<ev_periodic_set>, but indeed works even in interval and manual
		2465	rescheduling modes.
		2466
		2467	=item ev_tstamp offset [read-write]
		2468
		2469	When repeating, this contains the offset value, otherwise this is the
		2470	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2471	although libev might modify this value for better numerical stability).
		2472
		2473	Can be modified any time, but changes only take effect when the periodic
		2474	timer fires or C<ev_periodic_again> is being called.
		2475
		2476	=item ev_tstamp interval [read-write]
		2477
		2478	The current interval value. Can be modified any time, but changes only
		2479	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2480	called.
		2481
		2482	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2483
		2484	The current reschedule callback, or C<0>, if this functionality is
		2485	switched off. Can be changed any time, but changes only take effect when
		2486	the periodic timer fires or C<ev_periodic_again> is being called.
		2487
678	=back	2488	=back
679		2489
		2490	=head3 Examples
		2491
		2492	Example: Call a callback every hour, or, more precisely, whenever the
		2493	system time is divisible by 3600. The callback invocation times have
		2494	potentially a lot of jitter, but good long-term stability.
		2495
		2496	static void
		2497	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
		2498	{
		2499	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2500	}
		2501
		2502	ev_periodic hourly_tick;
		2503	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2504	ev_periodic_start (loop, &hourly_tick);
		2505
		2506	Example: The same as above, but use a reschedule callback to do it:
		2507
		2508	#include <math.h>
		2509
		2510	static ev_tstamp
		2511	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2512	{
		2513	return now + (3600. - fmod (now, 3600.));
		2514	}
		2515
		2516	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2517
		2518	Example: Call a callback every hour, starting now:
		2519
		2520	ev_periodic hourly_tick;
		2521	ev_periodic_init (&hourly_tick, clock_cb,
		2522	fmod (ev_now (loop), 3600.), 3600., 0);
		2523	ev_periodic_start (loop, &hourly_tick);
		2524
		2525
680	=head2 C<ev_signal> - signal me when a signal gets signalled	2526	=head2 C<ev_signal> - signal me when a signal gets signalled!
681		2527
682	Signal watchers will trigger an event when the process receives a specific	2528	Signal watchers will trigger an event when the process receives a specific
683	signal one or more times. Even though signals are very asynchronous, libev	2529	signal one or more times. Even though signals are very asynchronous, libev
684	will try it's best to deliver signals synchronously, i.e. as part of the	2530	will try its best to deliver signals synchronously, i.e. as part of the
685	normal event processing, like any other event.	2531	normal event processing, like any other event.
686		2532
		2533	If you want signals to be delivered truly asynchronously, just use
		2534	C<sigaction> as you would do without libev and forget about sharing
		2535	the signal. You can even use C<ev_async> from a signal handler to
		2536	synchronously wake up an event loop.
		2537
687	You can configure as many watchers as you like per signal. Only when the	2538	You can configure as many watchers as you like for the same signal, but
688	first watcher gets started will libev actually register a signal watcher	2539	only within the same loop, i.e. you can watch for C<SIGINT> in your
689	with the kernel (thus it coexists with your own signal handlers as long	2540	default loop and for C<SIGIO> in another loop, but you cannot watch for
690	as you don't register any with libev). Similarly, when the last signal	2541	C<SIGINT> in both the default loop and another loop at the same time. At
691	watcher for a signal is stopped libev will reset the signal handler to	2542	the moment, C<SIGCHLD> is permanently tied to the default loop.
692	SIG_DFL (regardless of what it was set to before).	2543
		2544	Only after the first watcher for a signal is started will libev actually
		2545	register something with the kernel. It thus coexists with your own signal
		2546	handlers as long as you don't register any with libev for the same signal.
		2547
		2548	If possible and supported, libev will install its handlers with
		2549	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2550	not be unduly interrupted. If you have a problem with system calls getting
		2551	interrupted by signals you can block all signals in an C<ev_check> watcher
		2552	and unblock them in an C<ev_prepare> watcher.
		2553
		2554	=head3 The special problem of inheritance over fork/execve/pthread_create
		2555
		2556	Both the signal mask (C<sigprocmask>) and the signal disposition
		2557	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2558	stopping it again), that is, libev might or might not block the signal,
		2559	and might or might not set or restore the installed signal handler (but
		2560	see C<EVFLAG_NOSIGMASK>).
		2561
		2562	While this does not matter for the signal disposition (libev never
		2563	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2564	C<execve>), this matters for the signal mask: many programs do not expect
		2565	certain signals to be blocked.
		2566
		2567	This means that before calling C<exec> (from the child) you should reset
		2568	the signal mask to whatever "default" you expect (all clear is a good
		2569	choice usually).
		2570
		2571	The simplest way to ensure that the signal mask is reset in the child is
		2572	to install a fork handler with C<pthread_atfork> that resets it. That will
		2573	catch fork calls done by libraries (such as the libc) as well.
		2574
		2575	In current versions of libev, the signal will not be blocked indefinitely
		2576	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2577	the window of opportunity for problems, it will not go away, as libev
		2578	I<has> to modify the signal mask, at least temporarily.
		2579
		2580	So I can't stress this enough: I<If you do not reset your signal mask when
		2581	you expect it to be empty, you have a race condition in your code>. This
		2582	is not a libev-specific thing, this is true for most event libraries.
		2583
		2584	=head3 The special problem of threads signal handling
		2585
		2586	POSIX threads has problematic signal handling semantics, specifically,
		2587	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2588	threads in a process block signals, which is hard to achieve.
		2589
		2590	When you want to use sigwait (or mix libev signal handling with your own
		2591	for the same signals), you can tackle this problem by globally blocking
		2592	all signals before creating any threads (or creating them with a fully set
		2593	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2594	loops. Then designate one thread as "signal receiver thread" which handles
		2595	these signals. You can pass on any signals that libev might be interested
		2596	in by calling C<ev_feed_signal>.
		2597
		2598	=head3 Watcher-Specific Functions and Data Members
693		2599
694	=over 4	2600	=over 4
695		2601
696	=item ev_signal_init (ev_signal *, callback, int signum)	2602	=item ev_signal_init (ev_signal *, callback, int signum)
697		2603
698	=item ev_signal_set (ev_signal *, int signum)	2604	=item ev_signal_set (ev_signal *, int signum)
699		2605
700	Configures the watcher to trigger on the given signal number (usually one	2606	Configures the watcher to trigger on the given signal number (usually one
701	of the C<SIGxxx> constants).	2607	of the C<SIGxxx> constants).
702		2608
		2609	=item int signum [read-only]
		2610
		2611	The signal the watcher watches out for.
		2612
703	=back	2613	=back
704		2614
		2615	=head3 Examples
		2616
		2617	Example: Try to exit cleanly on SIGINT.
		2618
		2619	static void
		2620	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2621	{
		2622	ev_break (loop, EVBREAK_ALL);
		2623	}
		2624
		2625	ev_signal signal_watcher;
		2626	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2627	ev_signal_start (loop, &signal_watcher);
		2628
		2629
705	=head2 C<ev_child> - wait for pid status changes	2630	=head2 C<ev_child> - watch out for process status changes
706		2631
707	Child watchers trigger when your process receives a SIGCHLD in response to	2632	Child watchers trigger when your process receives a SIGCHLD in response to
708	some child status changes (most typically when a child of yours dies).	2633	some child status changes (most typically when a child of yours dies or
		2634	exits). It is permissible to install a child watcher I<after> the child
		2635	has been forked (which implies it might have already exited), as long
		2636	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2637	forking and then immediately registering a watcher for the child is fine,
		2638	but forking and registering a watcher a few event loop iterations later or
		2639	in the next callback invocation is not.
		2640
		2641	Only the default event loop is capable of handling signals, and therefore
		2642	you can only register child watchers in the default event loop.
		2643
		2644	Due to some design glitches inside libev, child watchers will always be
		2645	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2646	libev)
		2647
		2648	=head3 Process Interaction
		2649
		2650	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2651	initialised. This is necessary to guarantee proper behaviour even if the
		2652	first child watcher is started after the child exits. The occurrence
		2653	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2654	synchronously as part of the event loop processing. Libev always reaps all
		2655	children, even ones not watched.
		2656
		2657	=head3 Overriding the Built-In Processing
		2658
		2659	Libev offers no special support for overriding the built-in child
		2660	processing, but if your application collides with libev's default child
		2661	handler, you can override it easily by installing your own handler for
		2662	C<SIGCHLD> after initialising the default loop, and making sure the
		2663	default loop never gets destroyed. You are encouraged, however, to use an
		2664	event-based approach to child reaping and thus use libev's support for
		2665	that, so other libev users can use C<ev_child> watchers freely.
		2666
		2667	=head3 Stopping the Child Watcher
		2668
		2669	Currently, the child watcher never gets stopped, even when the
		2670	child terminates, so normally one needs to stop the watcher in the
		2671	callback. Future versions of libev might stop the watcher automatically
		2672	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2673	problem).
		2674
		2675	=head3 Watcher-Specific Functions and Data Members
709		2676
710	=over 4	2677	=over 4
711		2678
712	=item ev_child_init (ev_child *, callback, int pid)	2679	=item ev_child_init (ev_child *, callback, int pid, int trace)
713		2680
714	=item ev_child_set (ev_child *, int pid)	2681	=item ev_child_set (ev_child *, int pid, int trace)
715		2682
716	Configures the watcher to wait for status changes of process C<pid> (or	2683	Configures the watcher to wait for status changes of process C<pid> (or
717	I<any> process if C<pid> is specified as C<0>). The callback can look	2684	I<any> process if C<pid> is specified as C<0>). The callback can look
718	at the C<rstatus> member of the C<ev_child> watcher structure to see	2685	at the C<rstatus> member of the C<ev_child> watcher structure to see
719	the status word (use the macros from C<sys/wait.h> and see your systems	2686	the status word (use the macros from C<sys/wait.h> and see your systems
720	C<waitpid> documentation). The C<rpid> member contains the pid of the	2687	C<waitpid> documentation). The C<rpid> member contains the pid of the
721	process causing the status change.	2688	process causing the status change. C<trace> must be either C<0> (only
		2689	activate the watcher when the process terminates) or C<1> (additionally
		2690	activate the watcher when the process is stopped or continued).
		2691
		2692	=item int pid [read-only]
		2693
		2694	The process id this watcher watches out for, or C<0>, meaning any process id.
		2695
		2696	=item int rpid [read-write]
		2697
		2698	The process id that detected a status change.
		2699
		2700	=item int rstatus [read-write]
		2701
		2702	The process exit/trace status caused by C<rpid> (see your systems
		2703	C<waitpid> and C<sys/wait.h> documentation for details).
722		2704
723	=back	2705	=back
724		2706
		2707	=head3 Examples
		2708
		2709	Example: C<fork()> a new process and install a child handler to wait for
		2710	its completion.
		2711
		2712	ev_child cw;
		2713
		2714	static void
		2715	child_cb (EV_P_ ev_child *w, int revents)
		2716	{
		2717	ev_child_stop (EV_A_ w);
		2718	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2719	}
		2720
		2721	pid_t pid = fork ();
		2722
		2723	if (pid < 0)
		2724	// error
		2725	else if (pid == 0)
		2726	{
		2727	// the forked child executes here
		2728	exit (1);
		2729	}
		2730	else
		2731	{
		2732	ev_child_init (&cw, child_cb, pid, 0);
		2733	ev_child_start (EV_DEFAULT_ &cw);
		2734	}
		2735
		2736
		2737	=head2 C<ev_stat> - did the file attributes just change?
		2738
		2739	This watches a file system path for attribute changes. That is, it calls
		2740	C<stat> on that path in regular intervals (or when the OS says it changed)
		2741	and sees if it changed compared to the last time, invoking the callback
		2742	if it did. Starting the watcher C<stat>'s the file, so only changes that
		2743	happen after the watcher has been started will be reported.
		2744
		2745	The path does not need to exist: changing from "path exists" to "path does
		2746	not exist" is a status change like any other. The condition "path does not
		2747	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2748	C<st_nlink> field being zero (which is otherwise always forced to be at
		2749	least one) and all the other fields of the stat buffer having unspecified
		2750	contents.
		2751
		2752	The path I<must not> end in a slash or contain special components such as
		2753	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2754	your working directory changes, then the behaviour is undefined.
		2755
		2756	Since there is no portable change notification interface available, the
		2757	portable implementation simply calls C<stat(2)> regularly on the path
		2758	to see if it changed somehow. You can specify a recommended polling
		2759	interval for this case. If you specify a polling interval of C<0> (highly
		2760	recommended!) then a I<suitable, unspecified default> value will be used
		2761	(which you can expect to be around five seconds, although this might
		2762	change dynamically). Libev will also impose a minimum interval which is
		2763	currently around C<0.1>, but that's usually overkill.
		2764
		2765	This watcher type is not meant for massive numbers of stat watchers,
		2766	as even with OS-supported change notifications, this can be
		2767	resource-intensive.
		2768
		2769	At the time of this writing, the only OS-specific interface implemented
		2770	is the Linux inotify interface (implementing kqueue support is left as an
		2771	exercise for the reader. Note, however, that the author sees no way of
		2772	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2773
		2774	=head3 ABI Issues (Largefile Support)
		2775
		2776	Libev by default (unless the user overrides this) uses the default
		2777	compilation environment, which means that on systems with large file
		2778	support disabled by default, you get the 32 bit version of the stat
		2779	structure. When using the library from programs that change the ABI to
		2780	use 64 bit file offsets the programs will fail. In that case you have to
		2781	compile libev with the same flags to get binary compatibility. This is
		2782	obviously the case with any flags that change the ABI, but the problem is
		2783	most noticeably displayed with ev_stat and large file support.
		2784
		2785	The solution for this is to lobby your distribution maker to make large
		2786	file interfaces available by default (as e.g. FreeBSD does) and not
		2787	optional. Libev cannot simply switch on large file support because it has
		2788	to exchange stat structures with application programs compiled using the
		2789	default compilation environment.
		2790
		2791	=head3 Inotify and Kqueue
		2792
		2793	When C<inotify (7)> support has been compiled into libev and present at
		2794	runtime, it will be used to speed up change detection where possible. The
		2795	inotify descriptor will be created lazily when the first C<ev_stat>
		2796	watcher is being started.
		2797
		2798	Inotify presence does not change the semantics of C<ev_stat> watchers
		2799	except that changes might be detected earlier, and in some cases, to avoid
		2800	making regular C<stat> calls. Even in the presence of inotify support
		2801	there are many cases where libev has to resort to regular C<stat> polling,
		2802	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2803	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2804	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2805	xfs are fully working) libev usually gets away without polling.
		2806
		2807	There is no support for kqueue, as apparently it cannot be used to
		2808	implement this functionality, due to the requirement of having a file
		2809	descriptor open on the object at all times, and detecting renames, unlinks
		2810	etc. is difficult.
		2811
		2812	=head3 C<stat ()> is a synchronous operation
		2813
		2814	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2815	the process. The exception are C<ev_stat> watchers - those call C<stat
		2816	()>, which is a synchronous operation.
		2817
		2818	For local paths, this usually doesn't matter: unless the system is very
		2819	busy or the intervals between stat's are large, a stat call will be fast,
		2820	as the path data is usually in memory already (except when starting the
		2821	watcher).
		2822
		2823	For networked file systems, calling C<stat ()> can block an indefinite
		2824	time due to network issues, and even under good conditions, a stat call
		2825	often takes multiple milliseconds.
		2826
		2827	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2828	paths, although this is fully supported by libev.
		2829
		2830	=head3 The special problem of stat time resolution
		2831
		2832	The C<stat ()> system call only supports full-second resolution portably,
		2833	and even on systems where the resolution is higher, most file systems
		2834	still only support whole seconds.
		2835
		2836	That means that, if the time is the only thing that changes, you can
		2837	easily miss updates: on the first update, C<ev_stat> detects a change and
		2838	calls your callback, which does something. When there is another update
		2839	within the same second, C<ev_stat> will be unable to detect unless the
		2840	stat data does change in other ways (e.g. file size).
		2841
		2842	The solution to this is to delay acting on a change for slightly more
		2843	than a second (or till slightly after the next full second boundary), using
		2844	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2845	ev_timer_again (loop, w)>).
		2846
		2847	The C<.02> offset is added to work around small timing inconsistencies
		2848	of some operating systems (where the second counter of the current time
		2849	might be be delayed. One such system is the Linux kernel, where a call to
		2850	C<gettimeofday> might return a timestamp with a full second later than
		2851	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2852	update file times then there will be a small window where the kernel uses
		2853	the previous second to update file times but libev might already execute
		2854	the timer callback).
		2855
		2856	=head3 Watcher-Specific Functions and Data Members
		2857
		2858	=over 4
		2859
		2860	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2861
		2862	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2863
		2864	Configures the watcher to wait for status changes of the given
		2865	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2866	be detected and should normally be specified as C<0> to let libev choose
		2867	a suitable value. The memory pointed to by C<path> must point to the same
		2868	path for as long as the watcher is active.
		2869
		2870	The callback will receive an C<EV_STAT> event when a change was detected,
		2871	relative to the attributes at the time the watcher was started (or the
		2872	last change was detected).
		2873
		2874	=item ev_stat_stat (loop, ev_stat *)
		2875
		2876	Updates the stat buffer immediately with new values. If you change the
		2877	watched path in your callback, you could call this function to avoid
		2878	detecting this change (while introducing a race condition if you are not
		2879	the only one changing the path). Can also be useful simply to find out the
		2880	new values.
		2881
		2882	=item ev_statdata attr [read-only]
		2883
		2884	The most-recently detected attributes of the file. Although the type is
		2885	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2886	suitable for your system, but you can only rely on the POSIX-standardised
		2887	members to be present. If the C<st_nlink> member is C<0>, then there was
		2888	some error while C<stat>ing the file.
		2889
		2890	=item ev_statdata prev [read-only]
		2891
		2892	The previous attributes of the file. The callback gets invoked whenever
		2893	C<prev> != C<attr>, or, more precisely, one or more of these members
		2894	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2895	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2896
		2897	=item ev_tstamp interval [read-only]
		2898
		2899	The specified interval.
		2900
		2901	=item const char *path [read-only]
		2902
		2903	The file system path that is being watched.
		2904
		2905	=back
		2906
		2907	=head3 Examples
		2908
		2909	Example: Watch C</etc/passwd> for attribute changes.
		2910
		2911	static void
		2912	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2913	{
		2914	/* /etc/passwd changed in some way */
		2915	if (w->attr.st_nlink)
		2916	{
		2917	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2918	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2919	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2920	}
		2921	else
		2922	/* you shalt not abuse printf for puts */
		2923	puts ("wow, /etc/passwd is not there, expect problems. "
		2924	"if this is windows, they already arrived\n");
		2925	}
		2926
		2927	...
		2928	ev_stat passwd;
		2929
		2930	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2931	ev_stat_start (loop, &passwd);
		2932
		2933	Example: Like above, but additionally use a one-second delay so we do not
		2934	miss updates (however, frequent updates will delay processing, too, so
		2935	one might do the work both on C<ev_stat> callback invocation I<and> on
		2936	C<ev_timer> callback invocation).
		2937
		2938	static ev_stat passwd;
		2939	static ev_timer timer;
		2940
		2941	static void
		2942	timer_cb (EV_P_ ev_timer *w, int revents)
		2943	{
		2944	ev_timer_stop (EV_A_ w);
		2945
		2946	/* now it's one second after the most recent passwd change */
		2947	}
		2948
		2949	static void
		2950	stat_cb (EV_P_ ev_stat *w, int revents)
		2951	{
		2952	/* reset the one-second timer */
		2953	ev_timer_again (EV_A_ &timer);
		2954	}
		2955
		2956	...
		2957	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2958	ev_stat_start (loop, &passwd);
		2959	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2960
		2961
725	=head2 C<ev_idle> - when you've got nothing better to do	2962	=head2 C<ev_idle> - when you've got nothing better to do...
726		2963
727	Idle watchers trigger events when there are no other events are pending	2964	Idle watchers trigger events when no other events of the same or higher
728	(prepare, check and other idle watchers do not count). That is, as long	2965	priority are pending (prepare, check and other idle watchers do not count
729	as your process is busy handling sockets or timeouts (or even signals,	2966	as receiving "events").
730	imagine) it will not be triggered. But when your process is idle all idle	2967
731	watchers are being called again and again, once per event loop iteration -	2968	That is, as long as your process is busy handling sockets or timeouts
		2969	(or even signals, imagine) of the same or higher priority it will not be
		2970	triggered. But when your process is idle (or only lower-priority watchers
		2971	are pending), the idle watchers are being called once per event loop
732	until stopped, that is, or your process receives more events and becomes	2972	iteration - until stopped, that is, or your process receives more events
733	busy.	2973	and becomes busy again with higher priority stuff.
734		2974
735	The most noteworthy effect is that as long as any idle watchers are	2975	The most noteworthy effect is that as long as any idle watchers are
736	active, the process will not block when waiting for new events.	2976	active, the process will not block when waiting for new events.
737		2977
738	Apart from keeping your process non-blocking (which is a useful	2978	Apart from keeping your process non-blocking (which is a useful
739	effect on its own sometimes), idle watchers are a good place to do	2979	effect on its own sometimes), idle watchers are a good place to do
740	"pseudo-background processing", or delay processing stuff to after the	2980	"pseudo-background processing", or delay processing stuff to after the
741	event loop has handled all outstanding events.	2981	event loop has handled all outstanding events.
742		2982
		2983	=head3 Abusing an C<ev_idle> watcher for its side-effect
		2984
		2985	As long as there is at least one active idle watcher, libev will never
		2986	sleep unnecessarily. Or in other words, it will loop as fast as possible.
		2987	For this to work, the idle watcher doesn't need to be invoked at all - the
		2988	lowest priority will do.
		2989
		2990	This mode of operation can be useful together with an C<ev_check> watcher,
		2991	to do something on each event loop iteration - for example to balance load
		2992	between different connections.
		2993
		2994	See L</Abusing an ev_check watcher for its side-effect> for a longer
		2995	example.
		2996
		2997	=head3 Watcher-Specific Functions and Data Members
		2998
743	=over 4	2999	=over 4
744		3000
745	=item ev_idle_init (ev_signal *, callback)	3001	=item ev_idle_init (ev_idle *, callback)
746		3002
747	Initialises and configures the idle watcher - it has no parameters of any	3003	Initialises and configures the idle watcher - it has no parameters of any
748	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	3004	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
749	believe me.	3005	believe me.
750		3006
751	=back	3007	=back
752		3008
		3009	=head3 Examples
		3010
		3011	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		3012	callback, free it. Also, use no error checking, as usual.
		3013
		3014	static void
		3015	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		3016	{
		3017	// stop the watcher
		3018	ev_idle_stop (loop, w);
		3019
		3020	// now we can free it
		3021	free (w);
		3022
		3023	// now do something you wanted to do when the program has
		3024	// no longer anything immediate to do.
		3025	}
		3026
		3027	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		3028	ev_idle_init (idle_watcher, idle_cb);
		3029	ev_idle_start (loop, idle_watcher);
		3030
		3031
753	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	3032	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
754		3033
755	Prepare and check watchers are usually (but not always) used in tandem:	3034	Prepare and check watchers are often (but not always) used in pairs:
756	prepare watchers get invoked before the process blocks and check watchers	3035	prepare watchers get invoked before the process blocks and check watchers
757	afterwards.	3036	afterwards.
758		3037
		3038	You I<must not> call C<ev_run> (or similar functions that enter the
		3039	current event loop) or C<ev_loop_fork> from either C<ev_prepare> or
		3040	C<ev_check> watchers. Other loops than the current one are fine,
		3041	however. The rationale behind this is that you do not need to check
		3042	for recursion in those watchers, i.e. the sequence will always be
		3043	C<ev_prepare>, blocking, C<ev_check> so if you have one watcher of each
		3044	kind they will always be called in pairs bracketing the blocking call.
		3045
759	Their main purpose is to integrate other event mechanisms into libev. This	3046	Their main purpose is to integrate other event mechanisms into libev and
760	could be used, for example, to track variable changes, implement your own	3047	their use is somewhat advanced. They could be used, for example, to track
761	watchers, integrate net-snmp or a coroutine library and lots more.	3048	variable changes, implement your own watchers, integrate net-snmp or a
		3049	coroutine library and lots more. They are also occasionally useful if
		3050	you cache some data and want to flush it before blocking (for example,
		3051	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		3052	watcher).
762		3053
763	This is done by examining in each prepare call which file descriptors need	3054	This is done by examining in each prepare call which file descriptors
764	to be watched by the other library, registering C<ev_io> watchers for	3055	need to be watched by the other library, registering C<ev_io> watchers
765	them and starting an C<ev_timer> watcher for any timeouts (many libraries	3056	for them and starting an C<ev_timer> watcher for any timeouts (many
766	provide just this functionality). Then, in the check watcher you check for	3057	libraries provide exactly this functionality). Then, in the check watcher,
767	any events that occured (by checking the pending status of all watchers	3058	you check for any events that occurred (by checking the pending status
768	and stopping them) and call back into the library. The I/O and timer	3059	of all watchers and stopping them) and call back into the library. The
769	callbacks will never actually be called (but must be valid nevertheless,	3060	I/O and timer callbacks will never actually be called (but must be valid
770	because you never know, you know?).	3061	nevertheless, because you never know, you know?).
771		3062
772	As another example, the Perl Coro module uses these hooks to integrate	3063	As another example, the Perl Coro module uses these hooks to integrate
773	coroutines into libev programs, by yielding to other active coroutines	3064	coroutines into libev programs, by yielding to other active coroutines
774	during each prepare and only letting the process block if no coroutines	3065	during each prepare and only letting the process block if no coroutines
775	are ready to run (it's actually more complicated: it only runs coroutines	3066	are ready to run (it's actually more complicated: it only runs coroutines
776	with priority higher than or equal to the event loop and one coroutine	3067	with priority higher than or equal to the event loop and one coroutine
777	of lower priority, but only once, using idle watchers to keep the event	3068	of lower priority, but only once, using idle watchers to keep the event
778	loop from blocking if lower-priority coroutines are active, thus mapping	3069	loop from blocking if lower-priority coroutines are active, thus mapping
779	low-priority coroutines to idle/background tasks).	3070	low-priority coroutines to idle/background tasks).
780		3071
		3072	When used for this purpose, it is recommended to give C<ev_check> watchers
		3073	highest (C<EV_MAXPRI>) priority, to ensure that they are being run before
		3074	any other watchers after the poll (this doesn't matter for C<ev_prepare>
		3075	watchers).
		3076
		3077	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		3078	activate ("feed") events into libev. While libev fully supports this, they
		3079	might get executed before other C<ev_check> watchers did their job. As
		3080	C<ev_check> watchers are often used to embed other (non-libev) event
		3081	loops those other event loops might be in an unusable state until their
		3082	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		3083	others).
		3084
		3085	=head3 Abusing an C<ev_check> watcher for its side-effect
		3086
		3087	C<ev_check> (and less often also C<ev_prepare>) watchers can also be
		3088	useful because they are called once per event loop iteration. For
		3089	example, if you want to handle a large number of connections fairly, you
		3090	normally only do a bit of work for each active connection, and if there
		3091	is more work to do, you wait for the next event loop iteration, so other
		3092	connections have a chance of making progress.
		3093
		3094	Using an C<ev_check> watcher is almost enough: it will be called on the
		3095	next event loop iteration. However, that isn't as soon as possible -
		3096	without external events, your C<ev_check> watcher will not be invoked.
		3097
		3098	This is where C<ev_idle> watchers come in handy - all you need is a
		3099	single global idle watcher that is active as long as you have one active
		3100	C<ev_check> watcher. The C<ev_idle> watcher makes sure the event loop
		3101	will not sleep, and the C<ev_check> watcher makes sure a callback gets
		3102	invoked. Neither watcher alone can do that.
		3103
		3104	=head3 Watcher-Specific Functions and Data Members
		3105
781	=over 4	3106	=over 4
782		3107
783	=item ev_prepare_init (ev_prepare *, callback)	3108	=item ev_prepare_init (ev_prepare *, callback)
784		3109
785	=item ev_check_init (ev_check *, callback)	3110	=item ev_check_init (ev_check *, callback)
786		3111
787	Initialises and configures the prepare or check watcher - they have no	3112	Initialises and configures the prepare or check watcher - they have no
788	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	3113	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
789	macros, but using them is utterly, utterly and completely pointless.	3114	macros, but using them is utterly, utterly, utterly and completely
		3115	pointless.
790		3116
791	=back	3117	=back
792		3118
		3119	=head3 Examples
		3120
		3121	There are a number of principal ways to embed other event loops or modules
		3122	into libev. Here are some ideas on how to include libadns into libev
		3123	(there is a Perl module named C<EV::ADNS> that does this, which you could
		3124	use as a working example. Another Perl module named C<EV::Glib> embeds a
		3125	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		3126	Glib event loop).
		3127
		3128	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		3129	and in a check watcher, destroy them and call into libadns. What follows
		3130	is pseudo-code only of course. This requires you to either use a low
		3131	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		3132	the callbacks for the IO/timeout watchers might not have been called yet.
		3133
		3134	static ev_io iow [nfd];
		3135	static ev_timer tw;
		3136
		3137	static void
		3138	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		3139	{
		3140	}
		3141
		3142	// create io watchers for each fd and a timer before blocking
		3143	static void
		3144	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		3145	{
		3146	int timeout = 3600000;
		3147	struct pollfd fds [nfd];
		3148	// actual code will need to loop here and realloc etc.
		3149	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		3150
		3151	/* the callback is illegal, but won't be called as we stop during check */
		3152	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		3153	ev_timer_start (loop, &tw);
		3154
		3155	// create one ev_io per pollfd
		3156	for (int i = 0; i < nfd; ++i)
		3157	{
		3158	ev_io_init (iow + i, io_cb, fds [i].fd,
		3159	((fds [i].events & POLLIN ? EV_READ : 0)
		3160	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		3161
		3162	fds [i].revents = 0;
		3163	ev_io_start (loop, iow + i);
		3164	}
		3165	}
		3166
		3167	// stop all watchers after blocking
		3168	static void
		3169	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		3170	{
		3171	ev_timer_stop (loop, &tw);
		3172
		3173	for (int i = 0; i < nfd; ++i)
		3174	{
		3175	// set the relevant poll flags
		3176	// could also call adns_processreadable etc. here
		3177	struct pollfd *fd = fds + i;
		3178	int revents = ev_clear_pending (iow + i);
		3179	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		3180	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		3181
		3182	// now stop the watcher
		3183	ev_io_stop (loop, iow + i);
		3184	}
		3185
		3186	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		3187	}
		3188
		3189	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		3190	in the prepare watcher and would dispose of the check watcher.
		3191
		3192	Method 3: If the module to be embedded supports explicit event
		3193	notification (libadns does), you can also make use of the actual watcher
		3194	callbacks, and only destroy/create the watchers in the prepare watcher.
		3195
		3196	static void
		3197	timer_cb (EV_P_ ev_timer *w, int revents)
		3198	{
		3199	adns_state ads = (adns_state)w->data;
		3200	update_now (EV_A);
		3201
		3202	adns_processtimeouts (ads, &tv_now);
		3203	}
		3204
		3205	static void
		3206	io_cb (EV_P_ ev_io *w, int revents)
		3207	{
		3208	adns_state ads = (adns_state)w->data;
		3209	update_now (EV_A);
		3210
		3211	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		3212	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		3213	}
		3214
		3215	// do not ever call adns_afterpoll
		3216
		3217	Method 4: Do not use a prepare or check watcher because the module you
		3218	want to embed is not flexible enough to support it. Instead, you can
		3219	override their poll function. The drawback with this solution is that the
		3220	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		3221	this approach, effectively embedding EV as a client into the horrible
		3222	libglib event loop.
		3223
		3224	static gint
		3225	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		3226	{
		3227	int got_events = 0;
		3228
		3229	for (n = 0; n < nfds; ++n)
		3230	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		3231
		3232	if (timeout >= 0)
		3233	// create/start timer
		3234
		3235	// poll
		3236	ev_run (EV_A_ 0);
		3237
		3238	// stop timer again
		3239	if (timeout >= 0)
		3240	ev_timer_stop (EV_A_ &to);
		3241
		3242	// stop io watchers again - their callbacks should have set
		3243	for (n = 0; n < nfds; ++n)
		3244	ev_io_stop (EV_A_ iow [n]);
		3245
		3246	return got_events;
		3247	}
		3248
		3249
		3250	=head2 C<ev_embed> - when one backend isn't enough...
		3251
		3252	This is a rather advanced watcher type that lets you embed one event loop
		3253	into another (currently only C<ev_io> events are supported in the embedded
		3254	loop, other types of watchers might be handled in a delayed or incorrect
		3255	fashion and must not be used).
		3256
		3257	There are primarily two reasons you would want that: work around bugs and
		3258	prioritise I/O.
		3259
		3260	As an example for a bug workaround, the kqueue backend might only support
		3261	sockets on some platform, so it is unusable as generic backend, but you
		3262	still want to make use of it because you have many sockets and it scales
		3263	so nicely. In this case, you would create a kqueue-based loop and embed
		3264	it into your default loop (which might use e.g. poll). Overall operation
		3265	will be a bit slower because first libev has to call C<poll> and then
		3266	C<kevent>, but at least you can use both mechanisms for what they are
		3267	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		3268
		3269	As for prioritising I/O: under rare circumstances you have the case where
		3270	some fds have to be watched and handled very quickly (with low latency),
		3271	and even priorities and idle watchers might have too much overhead. In
		3272	this case you would put all the high priority stuff in one loop and all
		3273	the rest in a second one, and embed the second one in the first.
		3274
		3275	As long as the watcher is active, the callback will be invoked every
		3276	time there might be events pending in the embedded loop. The callback
		3277	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		3278	sweep and invoke their callbacks (the callback doesn't need to invoke the
		3279	C<ev_embed_sweep> function directly, it could also start an idle watcher
		3280	to give the embedded loop strictly lower priority for example).
		3281
		3282	You can also set the callback to C<0>, in which case the embed watcher
		3283	will automatically execute the embedded loop sweep whenever necessary.
		3284
		3285	Fork detection will be handled transparently while the C<ev_embed> watcher
		3286	is active, i.e., the embedded loop will automatically be forked when the
		3287	embedding loop forks. In other cases, the user is responsible for calling
		3288	C<ev_loop_fork> on the embedded loop.
		3289
		3290	Unfortunately, not all backends are embeddable: only the ones returned by
		3291	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		3292	portable one.
		3293
		3294	So when you want to use this feature you will always have to be prepared
		3295	that you cannot get an embeddable loop. The recommended way to get around
		3296	this is to have a separate variables for your embeddable loop, try to
		3297	create it, and if that fails, use the normal loop for everything.
		3298
		3299	=head3 C<ev_embed> and fork
		3300
		3301	While the C<ev_embed> watcher is running, forks in the embedding loop will
		3302	automatically be applied to the embedded loop as well, so no special
		3303	fork handling is required in that case. When the watcher is not running,
		3304	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		3305	as applicable.
		3306
		3307	=head3 Watcher-Specific Functions and Data Members
		3308
		3309	=over 4
		3310
		3311	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		3312
		3313	=item ev_embed_set (ev_embed *, struct ev_loop *embedded_loop)
		3314
		3315	Configures the watcher to embed the given loop, which must be
		3316	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		3317	invoked automatically, otherwise it is the responsibility of the callback
		3318	to invoke it (it will continue to be called until the sweep has been done,
		3319	if you do not want that, you need to temporarily stop the embed watcher).
		3320
		3321	=item ev_embed_sweep (loop, ev_embed *)
		3322
		3323	Make a single, non-blocking sweep over the embedded loop. This works
		3324	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
		3325	appropriate way for embedded loops.
		3326
		3327	=item struct ev_loop *other [read-only]
		3328
		3329	The embedded event loop.
		3330
		3331	=back
		3332
		3333	=head3 Examples
		3334
		3335	Example: Try to get an embeddable event loop and embed it into the default
		3336	event loop. If that is not possible, use the default loop. The default
		3337	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		3338	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		3339	used).
		3340
		3341	struct ev_loop *loop_hi = ev_default_init (0);
		3342	struct ev_loop *loop_lo = 0;
		3343	ev_embed embed;
		3344
		3345	// see if there is a chance of getting one that works
		3346	// (remember that a flags value of 0 means autodetection)
		3347	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3348	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3349	: 0;
		3350
		3351	// if we got one, then embed it, otherwise default to loop_hi
		3352	if (loop_lo)
		3353	{
		3354	ev_embed_init (&embed, 0, loop_lo);
		3355	ev_embed_start (loop_hi, &embed);
		3356	}
		3357	else
		3358	loop_lo = loop_hi;
		3359
		3360	Example: Check if kqueue is available but not recommended and create
		3361	a kqueue backend for use with sockets (which usually work with any
		3362	kqueue implementation). Store the kqueue/socket-only event loop in
		3363	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3364
		3365	struct ev_loop *loop = ev_default_init (0);
		3366	struct ev_loop *loop_socket = 0;
		3367	ev_embed embed;
		3368
		3369	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3370	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3371	{
		3372	ev_embed_init (&embed, 0, loop_socket);
		3373	ev_embed_start (loop, &embed);
		3374	}
		3375
		3376	if (!loop_socket)
		3377	loop_socket = loop;
		3378
		3379	// now use loop_socket for all sockets, and loop for everything else
		3380
		3381
		3382	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		3383
		3384	Fork watchers are called when a C<fork ()> was detected (usually because
		3385	whoever is a good citizen cared to tell libev about it by calling
		3386	C<ev_loop_fork>). The invocation is done before the event loop blocks next
		3387	and before C<ev_check> watchers are being called, and only in the child
		3388	after the fork. If whoever good citizen calling C<ev_default_fork> cheats
		3389	and calls it in the wrong process, the fork handlers will be invoked, too,
		3390	of course.
		3391
		3392	=head3 The special problem of life after fork - how is it possible?
		3393
		3394	Most uses of C<fork ()> consist of forking, then some simple calls to set
		3395	up/change the process environment, followed by a call to C<exec()>. This
		3396	sequence should be handled by libev without any problems.
		3397
		3398	This changes when the application actually wants to do event handling
		3399	in the child, or both parent in child, in effect "continuing" after the
		3400	fork.
		3401
		3402	The default mode of operation (for libev, with application help to detect
		3403	forks) is to duplicate all the state in the child, as would be expected
		3404	when I<either> the parent I<or> the child process continues.
		3405
		3406	When both processes want to continue using libev, then this is usually the
		3407	wrong result. In that case, usually one process (typically the parent) is
		3408	supposed to continue with all watchers in place as before, while the other
		3409	process typically wants to start fresh, i.e. without any active watchers.
		3410
		3411	The cleanest and most efficient way to achieve that with libev is to
		3412	simply create a new event loop, which of course will be "empty", and
		3413	use that for new watchers. This has the advantage of not touching more
		3414	memory than necessary, and thus avoiding the copy-on-write, and the
		3415	disadvantage of having to use multiple event loops (which do not support
		3416	signal watchers).
		3417
		3418	When this is not possible, or you want to use the default loop for
		3419	other reasons, then in the process that wants to start "fresh", call
		3420	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3421	Destroying the default loop will "orphan" (not stop) all registered
		3422	watchers, so you have to be careful not to execute code that modifies
		3423	those watchers. Note also that in that case, you have to re-register any
		3424	signal watchers.
		3425
		3426	=head3 Watcher-Specific Functions and Data Members
		3427
		3428	=over 4
		3429
		3430	=item ev_fork_init (ev_fork *, callback)
		3431
		3432	Initialises and configures the fork watcher - it has no parameters of any
		3433	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		3434	really.
		3435
		3436	=back
		3437
		3438
		3439	=head2 C<ev_cleanup> - even the best things end
		3440
		3441	Cleanup watchers are called just before the event loop is being destroyed
		3442	by a call to C<ev_loop_destroy>.
		3443
		3444	While there is no guarantee that the event loop gets destroyed, cleanup
		3445	watchers provide a convenient method to install cleanup hooks for your
		3446	program, worker threads and so on - you just to make sure to destroy the
		3447	loop when you want them to be invoked.
		3448
		3449	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3450	all other watchers, they do not keep a reference to the event loop (which
		3451	makes a lot of sense if you think about it). Like all other watchers, you
		3452	can call libev functions in the callback, except C<ev_cleanup_start>.
		3453
		3454	=head3 Watcher-Specific Functions and Data Members
		3455
		3456	=over 4
		3457
		3458	=item ev_cleanup_init (ev_cleanup *, callback)
		3459
		3460	Initialises and configures the cleanup watcher - it has no parameters of
		3461	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3462	pointless, I assure you.
		3463
		3464	=back
		3465
		3466	Example: Register an atexit handler to destroy the default loop, so any
		3467	cleanup functions are called.
		3468
		3469	static void
		3470	program_exits (void)
		3471	{
		3472	ev_loop_destroy (EV_DEFAULT_UC);
		3473	}
		3474
		3475	...
		3476	atexit (program_exits);
		3477
		3478
		3479	=head2 C<ev_async> - how to wake up an event loop
		3480
		3481	In general, you cannot use an C<ev_loop> from multiple threads or other
		3482	asynchronous sources such as signal handlers (as opposed to multiple event
		3483	loops - those are of course safe to use in different threads).
		3484
		3485	Sometimes, however, you need to wake up an event loop you do not control,
		3486	for example because it belongs to another thread. This is what C<ev_async>
		3487	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3488	it by calling C<ev_async_send>, which is thread- and signal safe.
		3489
		3490	This functionality is very similar to C<ev_signal> watchers, as signals,
		3491	too, are asynchronous in nature, and signals, too, will be compressed
		3492	(i.e. the number of callback invocations may be less than the number of
		3493	C<ev_async_send> calls). In fact, you could use signal watchers as a kind
		3494	of "global async watchers" by using a watcher on an otherwise unused
		3495	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3496	even without knowing which loop owns the signal.
		3497
		3498	=head3 Queueing
		3499
		3500	C<ev_async> does not support queueing of data in any way. The reason
		3501	is that the author does not know of a simple (or any) algorithm for a
		3502	multiple-writer-single-reader queue that works in all cases and doesn't
		3503	need elaborate support such as pthreads or unportable memory access
		3504	semantics.
		3505
		3506	That means that if you want to queue data, you have to provide your own
		3507	queue. But at least I can tell you how to implement locking around your
		3508	queue:
		3509
		3510	=over 4
		3511
		3512	=item queueing from a signal handler context
		3513
		3514	To implement race-free queueing, you simply add to the queue in the signal
		3515	handler but you block the signal handler in the watcher callback. Here is
		3516	an example that does that for some fictitious SIGUSR1 handler:
		3517
		3518	static ev_async mysig;
		3519
		3520	static void
		3521	sigusr1_handler (void)
		3522	{
		3523	sometype data;
		3524
		3525	// no locking etc.
		3526	queue_put (data);
		3527	ev_async_send (EV_DEFAULT_ &mysig);
		3528	}
		3529
		3530	static void
		3531	mysig_cb (EV_P_ ev_async *w, int revents)
		3532	{
		3533	sometype data;
		3534	sigset_t block, prev;
		3535
		3536	sigemptyset (&block);
		3537	sigaddset (&block, SIGUSR1);
		3538	sigprocmask (SIG_BLOCK, &block, &prev);
		3539
		3540	while (queue_get (&data))
		3541	process (data);
		3542
		3543	if (sigismember (&prev, SIGUSR1)
		3544	sigprocmask (SIG_UNBLOCK, &block, 0);
		3545	}
		3546
		3547	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3548	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3549	either...).
		3550
		3551	=item queueing from a thread context
		3552
		3553	The strategy for threads is different, as you cannot (easily) block
		3554	threads but you can easily preempt them, so to queue safely you need to
		3555	employ a traditional mutex lock, such as in this pthread example:
		3556
		3557	static ev_async mysig;
		3558	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3559
		3560	static void
		3561	otherthread (void)
		3562	{
		3563	// only need to lock the actual queueing operation
		3564	pthread_mutex_lock (&mymutex);
		3565	queue_put (data);
		3566	pthread_mutex_unlock (&mymutex);
		3567
		3568	ev_async_send (EV_DEFAULT_ &mysig);
		3569	}
		3570
		3571	static void
		3572	mysig_cb (EV_P_ ev_async *w, int revents)
		3573	{
		3574	pthread_mutex_lock (&mymutex);
		3575
		3576	while (queue_get (&data))
		3577	process (data);
		3578
		3579	pthread_mutex_unlock (&mymutex);
		3580	}
		3581
		3582	=back
		3583
		3584
		3585	=head3 Watcher-Specific Functions and Data Members
		3586
		3587	=over 4
		3588
		3589	=item ev_async_init (ev_async *, callback)
		3590
		3591	Initialises and configures the async watcher - it has no parameters of any
		3592	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3593	trust me.
		3594
		3595	=item ev_async_send (loop, ev_async *)
		3596
		3597	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3598	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3599	returns.
		3600
		3601	Unlike C<ev_feed_event>, this call is safe to do from other threads,
		3602	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
		3603	embedding section below on what exactly this means).
		3604
		3605	Note that, as with other watchers in libev, multiple events might get
		3606	compressed into a single callback invocation (another way to look at
		3607	this is that C<ev_async> watchers are level-triggered: they are set on
		3608	C<ev_async_send>, reset when the event loop detects that).
		3609
		3610	This call incurs the overhead of at most one extra system call per event
		3611	loop iteration, if the event loop is blocked, and no syscall at all if
		3612	the event loop (or your program) is processing events. That means that
		3613	repeated calls are basically free (there is no need to avoid calls for
		3614	performance reasons) and that the overhead becomes smaller (typically
		3615	zero) under load.
		3616
		3617	=item bool = ev_async_pending (ev_async *)
		3618
		3619	Returns a non-zero value when C<ev_async_send> has been called on the
		3620	watcher but the event has not yet been processed (or even noted) by the
		3621	event loop.
		3622
		3623	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3624	the loop iterates next and checks for the watcher to have become active,
		3625	it will reset the flag again. C<ev_async_pending> can be used to very
		3626	quickly check whether invoking the loop might be a good idea.
		3627
		3628	Not that this does I<not> check whether the watcher itself is pending,
		3629	only whether it has been requested to make this watcher pending: there
		3630	is a time window between the event loop checking and resetting the async
		3631	notification, and the callback being invoked.
		3632
		3633	=back
		3634
		3635
793	=head1 OTHER FUNCTIONS	3636	=head1 OTHER FUNCTIONS
794		3637
795	There are some other functions of possible interest. Described. Here. Now.	3638	There are some other functions of possible interest. Described. Here. Now.
796		3639
797	=over 4	3640	=over 4
798		3641
799	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3642	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback, arg)
800		3643
801	This function combines a simple timer and an I/O watcher, calls your	3644	This function combines a simple timer and an I/O watcher, calls your
802	callback on whichever event happens first and automatically stop both	3645	callback on whichever event happens first and automatically stops both
803	watchers. This is useful if you want to wait for a single event on an fd	3646	watchers. This is useful if you want to wait for a single event on an fd
804	or timeout without having to allocate/configure/start/stop/free one or	3647	or timeout without having to allocate/configure/start/stop/free one or
805	more watchers yourself.	3648	more watchers yourself.
806		3649
807	If C<fd> is less than 0, then no I/O watcher will be started and events	3650	If C<fd> is less than 0, then no I/O watcher will be started and the
808	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3651	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
809	C<events> set will be craeted and started.	3652	the given C<fd> and C<events> set will be created and started.
810		3653
811	If C<timeout> is less than 0, then no timeout watcher will be	3654	If C<timeout> is less than 0, then no timeout watcher will be
812	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3655	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
813	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3656	repeat = 0) will be started. C<0> is a valid timeout.
814	dubious value.
815		3657
816	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3658	The callback has the type C<void (*cb)(int revents, void *arg)> and is
817	passed an C<revents> set like normal event callbacks (a combination of	3659	passed an C<revents> set like normal event callbacks (a combination of
818	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3660	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
819	value passed to C<ev_once>:	3661	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3662	a timeout and an io event at the same time - you probably should give io
		3663	events precedence.
820		3664
		3665	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3666
821	static void stdin_ready (int revents, void *arg)	3667	static void stdin_ready (int revents, void *arg)
		3668	{
		3669	if (revents & EV_READ)
		3670	/* stdin might have data for us, joy! */;
		3671	else if (revents & EV_TIMER)
		3672	/* doh, nothing entered */;
		3673	}
		3674
		3675	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
		3676
		3677	=item ev_feed_fd_event (loop, int fd, int revents)
		3678
		3679	Feed an event on the given fd, as if a file descriptor backend detected
		3680	the given events.
		3681
		3682	=item ev_feed_signal_event (loop, int signum)
		3683
		3684	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
		3685	which is async-safe.
		3686
		3687	=back
		3688
		3689
		3690	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3691
		3692	This section explains some common idioms that are not immediately
		3693	obvious. Note that examples are sprinkled over the whole manual, and this
		3694	section only contains stuff that wouldn't fit anywhere else.
		3695
		3696	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3697
		3698	Each watcher has, by default, a C<void *data> member that you can read
		3699	or modify at any time: libev will completely ignore it. This can be used
		3700	to associate arbitrary data with your watcher. If you need more data and
		3701	don't want to allocate memory separately and store a pointer to it in that
		3702	data member, you can also "subclass" the watcher type and provide your own
		3703	data:
		3704
		3705	struct my_io
		3706	{
		3707	ev_io io;
		3708	int otherfd;
		3709	void *somedata;
		3710	struct whatever *mostinteresting;
		3711	};
		3712
		3713	...
		3714	struct my_io w;
		3715	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3716
		3717	And since your callback will be called with a pointer to the watcher, you
		3718	can cast it back to your own type:
		3719
		3720	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3721	{
		3722	struct my_io *w = (struct my_io *)w_;
		3723	...
		3724	}
		3725
		3726	More interesting and less C-conformant ways of casting your callback
		3727	function type instead have been omitted.
		3728
		3729	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3730
		3731	Another common scenario is to use some data structure with multiple
		3732	embedded watchers, in effect creating your own watcher that combines
		3733	multiple libev event sources into one "super-watcher":
		3734
		3735	struct my_biggy
		3736	{
		3737	int some_data;
		3738	ev_timer t1;
		3739	ev_timer t2;
		3740	}
		3741
		3742	In this case getting the pointer to C<my_biggy> is a bit more
		3743	complicated: Either you store the address of your C<my_biggy> struct in
		3744	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3745	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3746	real programmers):
		3747
		3748	#include <stddef.h>
		3749
		3750	static void
		3751	t1_cb (EV_P_ ev_timer *w, int revents)
		3752	{
		3753	struct my_biggy big = (struct my_biggy *)
		3754	(((char *)w) - offsetof (struct my_biggy, t1));
		3755	}
		3756
		3757	static void
		3758	t2_cb (EV_P_ ev_timer *w, int revents)
		3759	{
		3760	struct my_biggy big = (struct my_biggy *)
		3761	(((char *)w) - offsetof (struct my_biggy, t2));
		3762	}
		3763
		3764	=head2 AVOIDING FINISHING BEFORE RETURNING
		3765
		3766	Often you have structures like this in event-based programs:
		3767
		3768	callback ()
822	{	3769	{
823	if (revents & EV_TIMEOUT)	3770	free (request);
824	/* doh, nothing entered */;
825	else if (revents & EV_READ)
826	/* stdin might have data for us, joy! */;
827	}	3771	}
828		3772
829	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3773	request = start_new_request (..., callback);
830		3774
831	=item ev_feed_event (loop, watcher, int events)	3775	The intent is to start some "lengthy" operation. The C<request> could be
		3776	used to cancel the operation, or do other things with it.
832		3777
833	Feeds the given event set into the event loop, as if the specified event	3778	It's not uncommon to have code paths in C<start_new_request> that
834	had happened for the specified watcher (which must be a pointer to an	3779	immediately invoke the callback, for example, to report errors. Or you add
835	initialised but not necessarily started event watcher).	3780	some caching layer that finds that it can skip the lengthy aspects of the
		3781	operation and simply invoke the callback with the result.
836		3782
837	=item ev_feed_fd_event (loop, int fd, int revents)	3783	The problem here is that this will happen I<before> C<start_new_request>
		3784	has returned, so C<request> is not set.
838		3785
839	Feed an event on the given fd, as if a file descriptor backend detected	3786	Even if you pass the request by some safer means to the callback, you
840	the given events it.	3787	might want to do something to the request after starting it, such as
		3788	canceling it, which probably isn't working so well when the callback has
		3789	already been invoked.
841		3790
842	=item ev_feed_signal_event (loop, int signum)	3791	A common way around all these issues is to make sure that
		3792	C<start_new_request> I<always> returns before the callback is invoked. If
		3793	C<start_new_request> immediately knows the result, it can artificially
		3794	delay invoking the callback by using a C<prepare> or C<idle> watcher for
		3795	example, or more sneakily, by reusing an existing (stopped) watcher and
		3796	pushing it into the pending queue:
843		3797
844	Feed an event as if the given signal occured (loop must be the default loop!).	3798	ev_set_cb (watcher, callback);
		3799	ev_feed_event (EV_A_ watcher, 0);
845		3800
846	=back	3801	This way, C<start_new_request> can safely return before the callback is
		3802	invoked, while not delaying callback invocation too much.
		3803
		3804	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3805
		3806	Often (especially in GUI toolkits) there are places where you have
		3807	I<modal> interaction, which is most easily implemented by recursively
		3808	invoking C<ev_run>.
		3809
		3810	This brings the problem of exiting - a callback might want to finish the
		3811	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3812	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3813	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3814	other combination: In these cases, a simple C<ev_break> will not work.
		3815
		3816	The solution is to maintain "break this loop" variable for each C<ev_run>
		3817	invocation, and use a loop around C<ev_run> until the condition is
		3818	triggered, using C<EVRUN_ONCE>:
		3819
		3820	// main loop
		3821	int exit_main_loop = 0;
		3822
		3823	while (!exit_main_loop)
		3824	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3825
		3826	// in a modal watcher
		3827	int exit_nested_loop = 0;
		3828
		3829	while (!exit_nested_loop)
		3830	ev_run (EV_A_ EVRUN_ONCE);
		3831
		3832	To exit from any of these loops, just set the corresponding exit variable:
		3833
		3834	// exit modal loop
		3835	exit_nested_loop = 1;
		3836
		3837	// exit main program, after modal loop is finished
		3838	exit_main_loop = 1;
		3839
		3840	// exit both
		3841	exit_main_loop = exit_nested_loop = 1;
		3842
		3843	=head2 THREAD LOCKING EXAMPLE
		3844
		3845	Here is a fictitious example of how to run an event loop in a different
		3846	thread from where callbacks are being invoked and watchers are
		3847	created/added/removed.
		3848
		3849	For a real-world example, see the C<EV::Loop::Async> perl module,
		3850	which uses exactly this technique (which is suited for many high-level
		3851	languages).
		3852
		3853	The example uses a pthread mutex to protect the loop data, a condition
		3854	variable to wait for callback invocations, an async watcher to notify the
		3855	event loop thread and an unspecified mechanism to wake up the main thread.
		3856
		3857	First, you need to associate some data with the event loop:
		3858
		3859	typedef struct {
		3860	mutex_t lock; /* global loop lock */
		3861	ev_async async_w;
		3862	thread_t tid;
		3863	cond_t invoke_cv;
		3864	} userdata;
		3865
		3866	void prepare_loop (EV_P)
		3867	{
		3868	// for simplicity, we use a static userdata struct.
		3869	static userdata u;
		3870
		3871	ev_async_init (&u->async_w, async_cb);
		3872	ev_async_start (EV_A_ &u->async_w);
		3873
		3874	pthread_mutex_init (&u->lock, 0);
		3875	pthread_cond_init (&u->invoke_cv, 0);
		3876
		3877	// now associate this with the loop
		3878	ev_set_userdata (EV_A_ u);
		3879	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3880	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3881
		3882	// then create the thread running ev_run
		3883	pthread_create (&u->tid, 0, l_run, EV_A);
		3884	}
		3885
		3886	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3887	solely to wake up the event loop so it takes notice of any new watchers
		3888	that might have been added:
		3889
		3890	static void
		3891	async_cb (EV_P_ ev_async *w, int revents)
		3892	{
		3893	// just used for the side effects
		3894	}
		3895
		3896	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3897	protecting the loop data, respectively.
		3898
		3899	static void
		3900	l_release (EV_P)
		3901	{
		3902	userdata *u = ev_userdata (EV_A);
		3903	pthread_mutex_unlock (&u->lock);
		3904	}
		3905
		3906	static void
		3907	l_acquire (EV_P)
		3908	{
		3909	userdata *u = ev_userdata (EV_A);
		3910	pthread_mutex_lock (&u->lock);
		3911	}
		3912
		3913	The event loop thread first acquires the mutex, and then jumps straight
		3914	into C<ev_run>:
		3915
		3916	void *
		3917	l_run (void *thr_arg)
		3918	{
		3919	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3920
		3921	l_acquire (EV_A);
		3922	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3923	ev_run (EV_A_ 0);
		3924	l_release (EV_A);
		3925
		3926	return 0;
		3927	}
		3928
		3929	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3930	signal the main thread via some unspecified mechanism (signals? pipe
		3931	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3932	have been called (in a while loop because a) spurious wakeups are possible
		3933	and b) skipping inter-thread-communication when there are no pending
		3934	watchers is very beneficial):
		3935
		3936	static void
		3937	l_invoke (EV_P)
		3938	{
		3939	userdata *u = ev_userdata (EV_A);
		3940
		3941	while (ev_pending_count (EV_A))
		3942	{
		3943	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3944	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3945	}
		3946	}
		3947
		3948	Now, whenever the main thread gets told to invoke pending watchers, it
		3949	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3950	thread to continue:
		3951
		3952	static void
		3953	real_invoke_pending (EV_P)
		3954	{
		3955	userdata *u = ev_userdata (EV_A);
		3956
		3957	pthread_mutex_lock (&u->lock);
		3958	ev_invoke_pending (EV_A);
		3959	pthread_cond_signal (&u->invoke_cv);
		3960	pthread_mutex_unlock (&u->lock);
		3961	}
		3962
		3963	Whenever you want to start/stop a watcher or do other modifications to an
		3964	event loop, you will now have to lock:
		3965
		3966	ev_timer timeout_watcher;
		3967	userdata *u = ev_userdata (EV_A);
		3968
		3969	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3970
		3971	pthread_mutex_lock (&u->lock);
		3972	ev_timer_start (EV_A_ &timeout_watcher);
		3973	ev_async_send (EV_A_ &u->async_w);
		3974	pthread_mutex_unlock (&u->lock);
		3975
		3976	Note that sending the C<ev_async> watcher is required because otherwise
		3977	an event loop currently blocking in the kernel will have no knowledge
		3978	about the newly added timer. By waking up the loop it will pick up any new
		3979	watchers in the next event loop iteration.
		3980
		3981	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3982
		3983	While the overhead of a callback that e.g. schedules a thread is small, it
		3984	is still an overhead. If you embed libev, and your main usage is with some
		3985	kind of threads or coroutines, you might want to customise libev so that
		3986	doesn't need callbacks anymore.
		3987
		3988	Imagine you have coroutines that you can switch to using a function
		3989	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3990	and that due to some magic, the currently active coroutine is stored in a
		3991	global called C<current_coro>. Then you can build your own "wait for libev
		3992	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3993	the differing C<;> conventions):
		3994
		3995	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3996	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3997
		3998	That means instead of having a C callback function, you store the
		3999	coroutine to switch to in each watcher, and instead of having libev call
		4000	your callback, you instead have it switch to that coroutine.
		4001
		4002	A coroutine might now wait for an event with a function called
		4003	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		4004	matter when, or whether the watcher is active or not when this function is
		4005	called):
		4006
		4007	void
		4008	wait_for_event (ev_watcher *w)
		4009	{
		4010	ev_set_cb (w, current_coro);
		4011	switch_to (libev_coro);
		4012	}
		4013
		4014	That basically suspends the coroutine inside C<wait_for_event> and
		4015	continues the libev coroutine, which, when appropriate, switches back to
		4016	this or any other coroutine.
		4017
		4018	You can do similar tricks if you have, say, threads with an event queue -
		4019	instead of storing a coroutine, you store the queue object and instead of
		4020	switching to a coroutine, you push the watcher onto the queue and notify
		4021	any waiters.
		4022
		4023	To embed libev, see L</EMBEDDING>, but in short, it's easiest to create two
		4024	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		4025
		4026	// my_ev.h
		4027	#define EV_CB_DECLARE(type) struct my_coro *cb;
		4028	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		4029	#include "../libev/ev.h"
		4030
		4031	// my_ev.c
		4032	#define EV_H "my_ev.h"
		4033	#include "../libev/ev.c"
		4034
		4035	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		4036	F<my_ev.c> into your project. When properly specifying include paths, you
		4037	can even use F<ev.h> as header file name directly.
		4038
847		4039
848	=head1 LIBEVENT EMULATION	4040	=head1 LIBEVENT EMULATION
849		4041
850	Libev offers a compatibility emulation layer for libevent. It cannot	4042	Libev offers a compatibility emulation layer for libevent. It cannot
851	emulate the internals of libevent, so here are some usage hints:	4043	emulate the internals of libevent, so here are some usage hints:
852		4044
853	=over 4	4045	=over 4
		4046
		4047	=item * Only the libevent-1.4.1-beta API is being emulated.
		4048
		4049	This was the newest libevent version available when libev was implemented,
		4050	and is still mostly unchanged in 2010.
854		4051
855	=item * Use it by including <event.h>, as usual.	4052	=item * Use it by including <event.h>, as usual.
856		4053
857	=item * The following members are fully supported: ev_base, ev_callback,	4054	=item * The following members are fully supported: ev_base, ev_callback,
858	ev_arg, ev_fd, ev_res, ev_events.	4055	ev_arg, ev_fd, ev_res, ev_events.
…		…
863		4060
864	=item * Priorities are not currently supported. Initialising priorities	4061	=item * Priorities are not currently supported. Initialising priorities
865	will fail and all watchers will have the same priority, even though there	4062	will fail and all watchers will have the same priority, even though there
866	is an ev_pri field.	4063	is an ev_pri field.
867		4064
		4065	=item * In libevent, the last base created gets the signals, in libev, the
		4066	base that registered the signal gets the signals.
		4067
868	=item * Other members are not supported.	4068	=item * Other members are not supported.
869		4069
870	=item * The libev emulation is I<not> ABI compatible to libevent, you need	4070	=item * The libev emulation is I<not> ABI compatible to libevent, you need
871	to use the libev header file and library.	4071	to use the libev header file and library.
872		4072
873	=back	4073	=back
874		4074
875	=head1 C++ SUPPORT	4075	=head1 C++ SUPPORT
876		4076
877	TBD.	4077	=head2 C API
		4078
		4079	The normal C API should work fine when used from C++: both ev.h and the
		4080	libev sources can be compiled as C++. Therefore, code that uses the C API
		4081	will work fine.
		4082
		4083	Proper exception specifications might have to be added to callbacks passed
		4084	to libev: exceptions may be thrown only from watcher callbacks, all other
		4085	callbacks (allocator, syserr, loop acquire/release and periodic reschedule
		4086	callbacks) must not throw exceptions, and might need a C<noexcept>
		4087	specification. If you have code that needs to be compiled as both C and
		4088	C++ you can use the C<EV_NOEXCEPT> macro for this:
		4089
		4090	static void
		4091	fatal_error (const char *msg) EV_NOEXCEPT
		4092	{
		4093	perror (msg);
		4094	abort ();
		4095	}
		4096
		4097	...
		4098	ev_set_syserr_cb (fatal_error);
		4099
		4100	The only API functions that can currently throw exceptions are C<ev_run>,
		4101	C<ev_invoke>, C<ev_invoke_pending> and C<ev_loop_destroy> (the latter
		4102	because it runs cleanup watchers).
		4103
		4104	Throwing exceptions in watcher callbacks is only supported if libev itself
		4105	is compiled with a C++ compiler or your C and C++ environments allow
		4106	throwing exceptions through C libraries (most do).
		4107
		4108	=head2 C++ API
		4109
		4110	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		4111	you to use some convenience methods to start/stop watchers and also change
		4112	the callback model to a model using method callbacks on objects.
		4113
		4114	To use it,
		4115
		4116	#include <ev++.h>
		4117
		4118	This automatically includes F<ev.h> and puts all of its definitions (many
		4119	of them macros) into the global namespace. All C++ specific things are
		4120	put into the C<ev> namespace. It should support all the same embedding
		4121	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		4122
		4123	Care has been taken to keep the overhead low. The only data member the C++
		4124	classes add (compared to plain C-style watchers) is the event loop pointer
		4125	that the watcher is associated with (or no additional members at all if
		4126	you disable C<EV_MULTIPLICITY> when embedding libev).
		4127
		4128	Currently, functions, static and non-static member functions and classes
		4129	with C<operator ()> can be used as callbacks. Other types should be easy
		4130	to add as long as they only need one additional pointer for context. If
		4131	you need support for other types of functors please contact the author
		4132	(preferably after implementing it).
		4133
		4134	For all this to work, your C++ compiler either has to use the same calling
		4135	conventions as your C compiler (for static member functions), or you have
		4136	to embed libev and compile libev itself as C++.
		4137
		4138	Here is a list of things available in the C<ev> namespace:
		4139
		4140	=over 4
		4141
		4142	=item C<ev::READ>, C<ev::WRITE> etc.
		4143
		4144	These are just enum values with the same values as the C<EV_READ> etc.
		4145	macros from F<ev.h>.
		4146
		4147	=item C<ev::tstamp>, C<ev::now>
		4148
		4149	Aliases to the same types/functions as with the C<ev_> prefix.
		4150
		4151	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		4152
		4153	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		4154	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		4155	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		4156	defined by many implementations.
		4157
		4158	All of those classes have these methods:
		4159
		4160	=over 4
		4161
		4162	=item ev::TYPE::TYPE ()
		4163
		4164	=item ev::TYPE::TYPE (loop)
		4165
		4166	=item ev::TYPE::~TYPE
		4167
		4168	The constructor (optionally) takes an event loop to associate the watcher
		4169	with. If it is omitted, it will use C<EV_DEFAULT>.
		4170
		4171	The constructor calls C<ev_init> for you, which means you have to call the
		4172	C<set> method before starting it.
		4173
		4174	It will not set a callback, however: You have to call the templated C<set>
		4175	method to set a callback before you can start the watcher.
		4176
		4177	(The reason why you have to use a method is a limitation in C++ which does
		4178	not allow explicit template arguments for constructors).
		4179
		4180	The destructor automatically stops the watcher if it is active.
		4181
		4182	=item w->set<class, &class::method> (object *)
		4183
		4184	This method sets the callback method to call. The method has to have a
		4185	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		4186	first argument and the C<revents> as second. The object must be given as
		4187	parameter and is stored in the C<data> member of the watcher.
		4188
		4189	This method synthesizes efficient thunking code to call your method from
		4190	the C callback that libev requires. If your compiler can inline your
		4191	callback (i.e. it is visible to it at the place of the C<set> call and
		4192	your compiler is good :), then the method will be fully inlined into the
		4193	thunking function, making it as fast as a direct C callback.
		4194
		4195	Example: simple class declaration and watcher initialisation
		4196
		4197	struct myclass
		4198	{
		4199	void io_cb (ev::io &w, int revents) { }
		4200	}
		4201
		4202	myclass obj;
		4203	ev::io iow;
		4204	iow.set <myclass, &myclass::io_cb> (&obj);
		4205
		4206	=item w->set (object *)
		4207
		4208	This is a variation of a method callback - leaving out the method to call
		4209	will default the method to C<operator ()>, which makes it possible to use
		4210	functor objects without having to manually specify the C<operator ()> all
		4211	the time. Incidentally, you can then also leave out the template argument
		4212	list.
		4213
		4214	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		4215	int revents)>.
		4216
		4217	See the method-C<set> above for more details.
		4218
		4219	Example: use a functor object as callback.
		4220
		4221	struct myfunctor
		4222	{
		4223	void operator() (ev::io &w, int revents)
		4224	{
		4225	...
		4226	}
		4227	}
		4228
		4229	myfunctor f;
		4230
		4231	ev::io w;
		4232	w.set (&f);
		4233
		4234	=item w->set<function> (void *data = 0)
		4235
		4236	Also sets a callback, but uses a static method or plain function as
		4237	callback. The optional C<data> argument will be stored in the watcher's
		4238	C<data> member and is free for you to use.
		4239
		4240	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		4241
		4242	See the method-C<set> above for more details.
		4243
		4244	Example: Use a plain function as callback.
		4245
		4246	static void io_cb (ev::io &w, int revents) { }
		4247	iow.set <io_cb> ();
		4248
		4249	=item w->set (loop)
		4250
		4251	Associates a different C<struct ev_loop> with this watcher. You can only
		4252	do this when the watcher is inactive (and not pending either).
		4253
		4254	=item w->set ([arguments])
		4255
		4256	Basically the same as C<ev_TYPE_set> (except for C<ev::embed> watchers>),
		4257	with the same arguments. Either this method or a suitable start method
		4258	must be called at least once. Unlike the C counterpart, an active watcher
		4259	gets automatically stopped and restarted when reconfiguring it with this
		4260	method.
		4261
		4262	For C<ev::embed> watchers this method is called C<set_embed>, to avoid
		4263	clashing with the C<set (loop)> method.
		4264
		4265	For C<ev::io> watchers there is an additional C<set> method that acepts a
		4266	new event mask only, and internally calls C<ev_io_modfify>.
		4267
		4268	=item w->start ()
		4269
		4270	Starts the watcher. Note that there is no C<loop> argument, as the
		4271	constructor already stores the event loop.
		4272
		4273	=item w->start ([arguments])
		4274
		4275	Instead of calling C<set> and C<start> methods separately, it is often
		4276	convenient to wrap them in one call. Uses the same type of arguments as
		4277	the configure C<set> method of the watcher.
		4278
		4279	=item w->stop ()
		4280
		4281	Stops the watcher if it is active. Again, no C<loop> argument.
		4282
		4283	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		4284
		4285	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		4286	C<ev_TYPE_again> function.
		4287
		4288	=item w->sweep () (C<ev::embed> only)
		4289
		4290	Invokes C<ev_embed_sweep>.
		4291
		4292	=item w->update () (C<ev::stat> only)
		4293
		4294	Invokes C<ev_stat_stat>.
		4295
		4296	=back
		4297
		4298	=back
		4299
		4300	Example: Define a class with two I/O and idle watchers, start the I/O
		4301	watchers in the constructor.
		4302
		4303	class myclass
		4304	{
		4305	ev::io io ; void io_cb (ev::io &w, int revents);
		4306	ev::io io2 ; void io2_cb (ev::io &w, int revents);
		4307	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		4308
		4309	myclass (int fd)
		4310	{
		4311	io .set <myclass, &myclass::io_cb > (this);
		4312	io2 .set <myclass, &myclass::io2_cb > (this);
		4313	idle.set <myclass, &myclass::idle_cb> (this);
		4314
		4315	io.set (fd, ev::WRITE); // configure the watcher
		4316	io.start (); // start it whenever convenient
		4317
		4318	io2.start (fd, ev::READ); // set + start in one call
		4319	}
		4320	};
		4321
		4322
		4323	=head1 OTHER LANGUAGE BINDINGS
		4324
		4325	Libev does not offer other language bindings itself, but bindings for a
		4326	number of languages exist in the form of third-party packages. If you know
		4327	any interesting language binding in addition to the ones listed here, drop
		4328	me a note.
		4329
		4330	=over 4
		4331
		4332	=item Perl
		4333
		4334	The EV module implements the full libev API and is actually used to test
		4335	libev. EV is developed together with libev. Apart from the EV core module,
		4336	there are additional modules that implement libev-compatible interfaces
		4337	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		4338	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		4339	and C<EV::Glib>).
		4340
		4341	It can be found and installed via CPAN, its homepage is at
		4342	L<http://software.schmorp.de/pkg/EV>.
		4343
		4344	=item Python
		4345
		4346	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		4347	seems to be quite complete and well-documented.
		4348
		4349	=item Ruby
		4350
		4351	Tony Arcieri has written a ruby extension that offers access to a subset
		4352	of the libev API and adds file handle abstractions, asynchronous DNS and
		4353	more on top of it. It can be found via gem servers. Its homepage is at
		4354	L<http://rev.rubyforge.org/>.
		4355
		4356	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		4357	makes rev work even on mingw.
		4358
		4359	=item Haskell
		4360
		4361	A haskell binding to libev is available at
		4362	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		4363
		4364	=item D
		4365
		4366	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		4367	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
		4368
		4369	=item Ocaml
		4370
		4371	Erkki Seppala has written Ocaml bindings for libev, to be found at
		4372	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		4373
		4374	=item Lua
		4375
		4376	Brian Maher has written a partial interface to libev for lua (at the
		4377	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		4378	L<http://github.com/brimworks/lua-ev>.
		4379
		4380	=item Javascript
		4381
		4382	Node.js (L<http://nodejs.org>) uses libev as the underlying event library.
		4383
		4384	=item Others
		4385
		4386	There are others, and I stopped counting.
		4387
		4388	=back
		4389
		4390
		4391	=head1 MACRO MAGIC
		4392
		4393	Libev can be compiled with a variety of options, the most fundamental
		4394	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		4395	functions and callbacks have an initial C<struct ev_loop *> argument.
		4396
		4397	To make it easier to write programs that cope with either variant, the
		4398	following macros are defined:
		4399
		4400	=over 4
		4401
		4402	=item C<EV_A>, C<EV_A_>
		4403
		4404	This provides the loop I<argument> for functions, if one is required ("ev
		4405	loop argument"). The C<EV_A> form is used when this is the sole argument,
		4406	C<EV_A_> is used when other arguments are following. Example:
		4407
		4408	ev_unref (EV_A);
		4409	ev_timer_add (EV_A_ watcher);
		4410	ev_run (EV_A_ 0);
		4411
		4412	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		4413	which is often provided by the following macro.
		4414
		4415	=item C<EV_P>, C<EV_P_>
		4416
		4417	This provides the loop I<parameter> for functions, if one is required ("ev
		4418	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		4419	C<EV_P_> is used when other parameters are following. Example:
		4420
		4421	// this is how ev_unref is being declared
		4422	static void ev_unref (EV_P);
		4423
		4424	// this is how you can declare your typical callback
		4425	static void cb (EV_P_ ev_timer *w, int revents)
		4426
		4427	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		4428	suitable for use with C<EV_A>.
		4429
		4430	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		4431
		4432	Similar to the other two macros, this gives you the value of the default
		4433	loop, if multiple loops are supported ("ev loop default"). The default loop
		4434	will be initialised if it isn't already initialised.
		4435
		4436	For non-multiplicity builds, these macros do nothing, so you always have
		4437	to initialise the loop somewhere.
		4438
		4439	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		4440
		4441	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		4442	default loop has been initialised (C<UC> == unchecked). Their behaviour
		4443	is undefined when the default loop has not been initialised by a previous
		4444	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		4445
		4446	It is often prudent to use C<EV_DEFAULT> when initialising the first
		4447	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		4448
		4449	=back
		4450
		4451	Example: Declare and initialise a check watcher, utilising the above
		4452	macros so it will work regardless of whether multiple loops are supported
		4453	or not.
		4454
		4455	static void
		4456	check_cb (EV_P_ ev_timer *w, int revents)
		4457	{
		4458	ev_check_stop (EV_A_ w);
		4459	}
		4460
		4461	ev_check check;
		4462	ev_check_init (&check, check_cb);
		4463	ev_check_start (EV_DEFAULT_ &check);
		4464	ev_run (EV_DEFAULT_ 0);
		4465
		4466	=head1 EMBEDDING
		4467
		4468	Libev can (and often is) directly embedded into host
		4469	applications. Examples of applications that embed it include the Deliantra
		4470	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		4471	and rxvt-unicode.
		4472
		4473	The goal is to enable you to just copy the necessary files into your
		4474	source directory without having to change even a single line in them, so
		4475	you can easily upgrade by simply copying (or having a checked-out copy of
		4476	libev somewhere in your source tree).
		4477
		4478	=head2 FILESETS
		4479
		4480	Depending on what features you need you need to include one or more sets of files
		4481	in your application.
		4482
		4483	=head3 CORE EVENT LOOP
		4484
		4485	To include only the libev core (all the C<ev_*> functions), with manual
		4486	configuration (no autoconf):
		4487
		4488	#define EV_STANDALONE 1
		4489	#include "ev.c"
		4490
		4491	This will automatically include F<ev.h>, too, and should be done in a
		4492	single C source file only to provide the function implementations. To use
		4493	it, do the same for F<ev.h> in all files wishing to use this API (best
		4494	done by writing a wrapper around F<ev.h> that you can include instead and
		4495	where you can put other configuration options):
		4496
		4497	#define EV_STANDALONE 1
		4498	#include "ev.h"
		4499
		4500	Both header files and implementation files can be compiled with a C++
		4501	compiler (at least, that's a stated goal, and breakage will be treated
		4502	as a bug).
		4503
		4504	You need the following files in your source tree, or in a directory
		4505	in your include path (e.g. in libev/ when using -Ilibev):
		4506
		4507	ev.h
		4508	ev.c
		4509	ev_vars.h
		4510	ev_wrap.h
		4511
		4512	ev_win32.c required on win32 platforms only
		4513
		4514	ev_select.c only when select backend is enabled
		4515	ev_poll.c only when poll backend is enabled
		4516	ev_epoll.c only when the epoll backend is enabled
		4517	ev_linuxaio.c only when the linux aio backend is enabled
		4518	ev_iouring.c only when the linux io_uring backend is enabled
		4519	ev_kqueue.c only when the kqueue backend is enabled
		4520	ev_port.c only when the solaris port backend is enabled
		4521
		4522	F<ev.c> includes the backend files directly when enabled, so you only need
		4523	to compile this single file.
		4524
		4525	=head3 LIBEVENT COMPATIBILITY API
		4526
		4527	To include the libevent compatibility API, also include:
		4528
		4529	#include "event.c"
		4530
		4531	in the file including F<ev.c>, and:
		4532
		4533	#include "event.h"
		4534
		4535	in the files that want to use the libevent API. This also includes F<ev.h>.
		4536
		4537	You need the following additional files for this:
		4538
		4539	event.h
		4540	event.c
		4541
		4542	=head3 AUTOCONF SUPPORT
		4543
		4544	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		4545	whatever way you want, you can also C<m4_include([libev.m4])> in your
		4546	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		4547	include F<config.h> and configure itself accordingly.
		4548
		4549	For this of course you need the m4 file:
		4550
		4551	libev.m4
		4552
		4553	=head2 PREPROCESSOR SYMBOLS/MACROS
		4554
		4555	Libev can be configured via a variety of preprocessor symbols you have to
		4556	define before including (or compiling) any of its files. The default in
		4557	the absence of autoconf is documented for every option.
		4558
		4559	Symbols marked with "(h)" do not change the ABI, and can have different
		4560	values when compiling libev vs. including F<ev.h>, so it is permissible
		4561	to redefine them before including F<ev.h> without breaking compatibility
		4562	to a compiled library. All other symbols change the ABI, which means all
		4563	users of libev and the libev code itself must be compiled with compatible
		4564	settings.
		4565
		4566	=over 4
		4567
		4568	=item EV_COMPAT3 (h)
		4569
		4570	Backwards compatibility is a major concern for libev. This is why this
		4571	release of libev comes with wrappers for the functions and symbols that
		4572	have been renamed between libev version 3 and 4.
		4573
		4574	You can disable these wrappers (to test compatibility with future
		4575	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4576	sources. This has the additional advantage that you can drop the C<struct>
		4577	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4578	typedef in that case.
		4579
		4580	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4581	and in some even more future version the compatibility code will be
		4582	removed completely.
		4583
		4584	=item EV_STANDALONE (h)
		4585
		4586	Must always be C<1> if you do not use autoconf configuration, which
		4587	keeps libev from including F<config.h>, and it also defines dummy
		4588	implementations for some libevent functions (such as logging, which is not
		4589	supported). It will also not define any of the structs usually found in
		4590	F<event.h> that are not directly supported by the libev core alone.
		4591
		4592	In standalone mode, libev will still try to automatically deduce the
		4593	configuration, but has to be more conservative.
		4594
		4595	=item EV_USE_FLOOR
		4596
		4597	If defined to be C<1>, libev will use the C<floor ()> function for its
		4598	periodic reschedule calculations, otherwise libev will fall back on a
		4599	portable (slower) implementation. If you enable this, you usually have to
		4600	link against libm or something equivalent. Enabling this when the C<floor>
		4601	function is not available will fail, so the safe default is to not enable
		4602	this.
		4603
		4604	=item EV_USE_MONOTONIC
		4605
		4606	If defined to be C<1>, libev will try to detect the availability of the
		4607	monotonic clock option at both compile time and runtime. Otherwise no
		4608	use of the monotonic clock option will be attempted. If you enable this,
		4609	you usually have to link against librt or something similar. Enabling it
		4610	when the functionality isn't available is safe, though, although you have
		4611	to make sure you link against any libraries where the C<clock_gettime>
		4612	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		4613
		4614	=item EV_USE_REALTIME
		4615
		4616	If defined to be C<1>, libev will try to detect the availability of the
		4617	real-time clock option at compile time (and assume its availability
		4618	at runtime if successful). Otherwise no use of the real-time clock
		4619	option will be attempted. This effectively replaces C<gettimeofday>
		4620	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		4621	correctness. See the note about libraries in the description of
		4622	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4623	C<EV_USE_CLOCK_SYSCALL>.
		4624
		4625	=item EV_USE_CLOCK_SYSCALL
		4626
		4627	If defined to be C<1>, libev will try to use a direct syscall instead
		4628	of calling the system-provided C<clock_gettime> function. This option
		4629	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4630	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4631	programs needlessly. Using a direct syscall is slightly slower (in
		4632	theory), because no optimised vdso implementation can be used, but avoids
		4633	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4634	higher, as it simplifies linking (no need for C<-lrt>).
		4635
		4636	=item EV_USE_NANOSLEEP
		4637
		4638	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		4639	and will use it for delays. Otherwise it will use C<select ()>.
		4640
		4641	=item EV_USE_EVENTFD
		4642
		4643	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4644	available and will probe for kernel support at runtime. This will improve
		4645	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4646	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4647	2.7 or newer, otherwise disabled.
		4648
		4649	=item EV_USE_SIGNALFD
		4650
		4651	If defined to be C<1>, then libev will assume that C<signalfd ()> is
		4652	available and will probe for kernel support at runtime. This enables
		4653	the use of EVFLAG_SIGNALFD for faster and simpler signal handling. If
		4654	undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4655	2.7 or newer, otherwise disabled.
		4656
		4657	=item EV_USE_TIMERFD
		4658
		4659	If defined to be C<1>, then libev will assume that C<timerfd ()> is
		4660	available and will probe for kernel support at runtime. This allows
		4661	libev to detect time jumps accurately. If undefined, it will be enabled
		4662	if the headers indicate GNU/Linux + Glibc 2.8 or newer and define
		4663	C<TFD_TIMER_CANCEL_ON_SET>, otherwise disabled.
		4664
		4665	=item EV_USE_EVENTFD
		4666
		4667	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4668	available and will probe for kernel support at runtime. This will improve
		4669	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4670	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4671	2.7 or newer, otherwise disabled.
		4672
		4673	=item EV_USE_SELECT
		4674
		4675	If undefined or defined to be C<1>, libev will compile in support for the
		4676	C<select>(2) backend. No attempt at auto-detection will be done: if no
		4677	other method takes over, select will be it. Otherwise the select backend
		4678	will not be compiled in.
		4679
		4680	=item EV_SELECT_USE_FD_SET
		4681
		4682	If defined to C<1>, then the select backend will use the system C<fd_set>
		4683	structure. This is useful if libev doesn't compile due to a missing
		4684	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		4685	on exotic systems. This usually limits the range of file descriptors to
		4686	some low limit such as 1024 or might have other limitations (winsocket
		4687	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		4688	configures the maximum size of the C<fd_set>.
		4689
		4690	=item EV_SELECT_IS_WINSOCKET
		4691
		4692	When defined to C<1>, the select backend will assume that
		4693	select/socket/connect etc. don't understand file descriptors but
		4694	wants osf handles on win32 (this is the case when the select to
		4695	be used is the winsock select). This means that it will call
		4696	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		4697	it is assumed that all these functions actually work on fds, even
		4698	on win32. Should not be defined on non-win32 platforms.
		4699
		4700	=item EV_FD_TO_WIN32_HANDLE(fd)
		4701
		4702	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		4703	file descriptors to socket handles. When not defining this symbol (the
		4704	default), then libev will call C<_get_osfhandle>, which is usually
		4705	correct. In some cases, programs use their own file descriptor management,
		4706	in which case they can provide this function to map fds to socket handles.
		4707
		4708	=item EV_WIN32_HANDLE_TO_FD(handle)
		4709
		4710	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4711	using the standard C<_open_osfhandle> function. For programs implementing
		4712	their own fd to handle mapping, overwriting this function makes it easier
		4713	to do so. This can be done by defining this macro to an appropriate value.
		4714
		4715	=item EV_WIN32_CLOSE_FD(fd)
		4716
		4717	If programs implement their own fd to handle mapping on win32, then this
		4718	macro can be used to override the C<close> function, useful to unregister
		4719	file descriptors again. Note that the replacement function has to close
		4720	the underlying OS handle.
		4721
		4722	=item EV_USE_WSASOCKET
		4723
		4724	If defined to be C<1>, libev will use C<WSASocket> to create its internal
		4725	communication socket, which works better in some environments. Otherwise,
		4726	the normal C<socket> function will be used, which works better in other
		4727	environments.
		4728
		4729	=item EV_USE_POLL
		4730
		4731	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		4732	backend. Otherwise it will be enabled on non-win32 platforms. It
		4733	takes precedence over select.
		4734
		4735	=item EV_USE_EPOLL
		4736
		4737	If defined to be C<1>, libev will compile in support for the Linux
		4738	C<epoll>(7) backend. Its availability will be detected at runtime,
		4739	otherwise another method will be used as fallback. This is the preferred
		4740	backend for GNU/Linux systems. If undefined, it will be enabled if the
		4741	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4742
		4743	=item EV_USE_LINUXAIO
		4744
		4745	If defined to be C<1>, libev will compile in support for the Linux aio
		4746	backend (C<EV_USE_EPOLL> must also be enabled). If undefined, it will be
		4747	enabled on linux, otherwise disabled.
		4748
		4749	=item EV_USE_IOURING
		4750
		4751	If defined to be C<1>, libev will compile in support for the Linux
		4752	io_uring backend (C<EV_USE_EPOLL> must also be enabled). Due to it's
		4753	current limitations it has to be requested explicitly. If undefined, it
		4754	will be enabled on linux, otherwise disabled.
		4755
		4756	=item EV_USE_KQUEUE
		4757
		4758	If defined to be C<1>, libev will compile in support for the BSD style
		4759	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		4760	otherwise another method will be used as fallback. This is the preferred
		4761	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		4762	supports some types of fds correctly (the only platform we found that
		4763	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		4764	not be used unless explicitly requested. The best way to use it is to find
		4765	out whether kqueue supports your type of fd properly and use an embedded
		4766	kqueue loop.
		4767
		4768	=item EV_USE_PORT
		4769
		4770	If defined to be C<1>, libev will compile in support for the Solaris
		4771	10 port style backend. Its availability will be detected at runtime,
		4772	otherwise another method will be used as fallback. This is the preferred
		4773	backend for Solaris 10 systems.
		4774
		4775	=item EV_USE_DEVPOLL
		4776
		4777	Reserved for future expansion, works like the USE symbols above.
		4778
		4779	=item EV_USE_INOTIFY
		4780
		4781	If defined to be C<1>, libev will compile in support for the Linux inotify
		4782	interface to speed up C<ev_stat> watchers. Its actual availability will
		4783	be detected at runtime. If undefined, it will be enabled if the headers
		4784	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4785
		4786	=item EV_NO_SMP
		4787
		4788	If defined to be C<1>, libev will assume that memory is always coherent
		4789	between threads, that is, threads can be used, but threads never run on
		4790	different cpus (or different cpu cores). This reduces dependencies
		4791	and makes libev faster.
		4792
		4793	=item EV_NO_THREADS
		4794
		4795	If defined to be C<1>, libev will assume that it will never be called from
		4796	different threads (that includes signal handlers), which is a stronger
		4797	assumption than C<EV_NO_SMP>, above. This reduces dependencies and makes
		4798	libev faster.
		4799
		4800	=item EV_ATOMIC_T
		4801
		4802	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		4803	access is atomic with respect to other threads or signal contexts. No
		4804	such type is easily found in the C language, so you can provide your own
		4805	type that you know is safe for your purposes. It is used both for signal
		4806	handler "locking" as well as for signal and thread safety in C<ev_async>
		4807	watchers.
		4808
		4809	In the absence of this define, libev will use C<sig_atomic_t volatile>
		4810	(from F<signal.h>), which is usually good enough on most platforms.
		4811
		4812	=item EV_H (h)
		4813
		4814	The name of the F<ev.h> header file used to include it. The default if
		4815	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		4816	used to virtually rename the F<ev.h> header file in case of conflicts.
		4817
		4818	=item EV_CONFIG_H (h)
		4819
		4820	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		4821	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		4822	C<EV_H>, above.
		4823
		4824	=item EV_EVENT_H (h)
		4825
		4826	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		4827	of how the F<event.h> header can be found, the default is C<"event.h">.
		4828
		4829	=item EV_PROTOTYPES (h)
		4830
		4831	If defined to be C<0>, then F<ev.h> will not define any function
		4832	prototypes, but still define all the structs and other symbols. This is
		4833	occasionally useful if you want to provide your own wrapper functions
		4834	around libev functions.
		4835
		4836	=item EV_MULTIPLICITY
		4837
		4838	If undefined or defined to C<1>, then all event-loop-specific functions
		4839	will have the C<struct ev_loop *> as first argument, and you can create
		4840	additional independent event loops. Otherwise there will be no support
		4841	for multiple event loops and there is no first event loop pointer
		4842	argument. Instead, all functions act on the single default loop.
		4843
		4844	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4845	default loop when multiplicity is switched off - you always have to
		4846	initialise the loop manually in this case.
		4847
		4848	=item EV_MINPRI
		4849
		4850	=item EV_MAXPRI
		4851
		4852	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		4853	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		4854	provide for more priorities by overriding those symbols (usually defined
		4855	to be C<-2> and C<2>, respectively).
		4856
		4857	When doing priority-based operations, libev usually has to linearly search
		4858	all the priorities, so having many of them (hundreds) uses a lot of space
		4859	and time, so using the defaults of five priorities (-2 .. +2) is usually
		4860	fine.
		4861
		4862	If your embedding application does not need any priorities, defining these
		4863	both to C<0> will save some memory and CPU.
		4864
		4865	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4866	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4867	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
		4868
		4869	If undefined or defined to be C<1> (and the platform supports it), then
		4870	the respective watcher type is supported. If defined to be C<0>, then it
		4871	is not. Disabling watcher types mainly saves code size.
		4872
		4873	=item EV_FEATURES
		4874
		4875	If you need to shave off some kilobytes of code at the expense of some
		4876	speed (but with the full API), you can define this symbol to request
		4877	certain subsets of functionality. The default is to enable all features
		4878	that can be enabled on the platform.
		4879
		4880	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4881	with some broad features you want) and then selectively re-enable
		4882	additional parts you want, for example if you want everything minimal,
		4883	but multiple event loop support, async and child watchers and the poll
		4884	backend, use this:
		4885
		4886	#define EV_FEATURES 0
		4887	#define EV_MULTIPLICITY 1
		4888	#define EV_USE_POLL 1
		4889	#define EV_CHILD_ENABLE 1
		4890	#define EV_ASYNC_ENABLE 1
		4891
		4892	The actual value is a bitset, it can be a combination of the following
		4893	values (by default, all of these are enabled):
		4894
		4895	=over 4
		4896
		4897	=item C<1> - faster/larger code
		4898
		4899	Use larger code to speed up some operations.
		4900
		4901	Currently this is used to override some inlining decisions (enlarging the
		4902	code size by roughly 30% on amd64).
		4903
		4904	When optimising for size, use of compiler flags such as C<-Os> with
		4905	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4906	assertions.
		4907
		4908	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4909	(e.g. gcc with C<-Os>).
		4910
		4911	=item C<2> - faster/larger data structures
		4912
		4913	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4914	hash table sizes and so on. This will usually further increase code size
		4915	and can additionally have an effect on the size of data structures at
		4916	runtime.
		4917
		4918	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4919	(e.g. gcc with C<-Os>).
		4920
		4921	=item C<4> - full API configuration
		4922
		4923	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4924	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4925
		4926	=item C<8> - full API
		4927
		4928	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4929	details on which parts of the API are still available without this
		4930	feature, and do not complain if this subset changes over time.
		4931
		4932	=item C<16> - enable all optional watcher types
		4933
		4934	Enables all optional watcher types. If you want to selectively enable
		4935	only some watcher types other than I/O and timers (e.g. prepare,
		4936	embed, async, child...) you can enable them manually by defining
		4937	C<EV_watchertype_ENABLE> to C<1> instead.
		4938
		4939	=item C<32> - enable all backends
		4940
		4941	This enables all backends - without this feature, you need to enable at
		4942	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4943
		4944	=item C<64> - enable OS-specific "helper" APIs
		4945
		4946	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4947	default.
		4948
		4949	=back
		4950
		4951	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4952	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4953	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4954	watchers, timers and monotonic clock support.
		4955
		4956	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4957	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4958	your program might be left out as well - a binary starting a timer and an
		4959	I/O watcher then might come out at only 5Kb.
		4960
		4961	=item EV_API_STATIC
		4962
		4963	If this symbol is defined (by default it is not), then all identifiers
		4964	will have static linkage. This means that libev will not export any
		4965	identifiers, and you cannot link against libev anymore. This can be useful
		4966	when you embed libev, only want to use libev functions in a single file,
		4967	and do not want its identifiers to be visible.
		4968
		4969	To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
		4970	wants to use libev.
		4971
		4972	This option only works when libev is compiled with a C compiler, as C++
		4973	doesn't support the required declaration syntax.
		4974
		4975	=item EV_AVOID_STDIO
		4976
		4977	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4978	functions (printf, scanf, perror etc.). This will increase the code size
		4979	somewhat, but if your program doesn't otherwise depend on stdio and your
		4980	libc allows it, this avoids linking in the stdio library which is quite
		4981	big.
		4982
		4983	Note that error messages might become less precise when this option is
		4984	enabled.
		4985
		4986	=item EV_NSIG
		4987
		4988	The highest supported signal number, +1 (or, the number of
		4989	signals): Normally, libev tries to deduce the maximum number of signals
		4990	automatically, but sometimes this fails, in which case it can be
		4991	specified. Also, using a lower number than detected (C<32> should be
		4992	good for about any system in existence) can save some memory, as libev
		4993	statically allocates some 12-24 bytes per signal number.
		4994
		4995	=item EV_PID_HASHSIZE
		4996
		4997	C<ev_child> watchers use a small hash table to distribute workload by
		4998	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
		4999	usually more than enough. If you need to manage thousands of children you
		5000	might want to increase this value (I<must> be a power of two).
		5001
		5002	=item EV_INOTIFY_HASHSIZE
		5003
		5004	C<ev_stat> watchers use a small hash table to distribute workload by
		5005	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
		5006	disabled), usually more than enough. If you need to manage thousands of
		5007	C<ev_stat> watchers you might want to increase this value (I<must> be a
		5008	power of two).
		5009
		5010	=item EV_USE_4HEAP
		5011
		5012	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		5013	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		5014	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		5015	faster performance with many (thousands) of watchers.
		5016
		5017	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		5018	will be C<0>.
		5019
		5020	=item EV_HEAP_CACHE_AT
		5021
		5022	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		5023	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		5024	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		5025	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		5026	but avoids random read accesses on heap changes. This improves performance
		5027	noticeably with many (hundreds) of watchers.
		5028
		5029	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		5030	will be C<0>.
		5031
		5032	=item EV_VERIFY
		5033
		5034	Controls how much internal verification (see C<ev_verify ()>) will
		5035	be done: If set to C<0>, no internal verification code will be compiled
		5036	in. If set to C<1>, then verification code will be compiled in, but not
		5037	called. If set to C<2>, then the internal verification code will be
		5038	called once per loop, which can slow down libev. If set to C<3>, then the
		5039	verification code will be called very frequently, which will slow down
		5040	libev considerably.
		5041
		5042	Verification errors are reported via C's C<assert> mechanism, so if you
		5043	disable that (e.g. by defining C<NDEBUG>) then no errors will be reported.
		5044
		5045	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		5046	will be C<0>.
		5047
		5048	=item EV_COMMON
		5049
		5050	By default, all watchers have a C<void *data> member. By redefining
		5051	this macro to something else you can include more and other types of
		5052	members. You have to define it each time you include one of the files,
		5053	though, and it must be identical each time.
		5054
		5055	For example, the perl EV module uses something like this:
		5056
		5057	#define EV_COMMON \
		5058	SV *self; /* contains this struct */ \
		5059	SV *cb_sv, *fh /* note no trailing ";" */
		5060
		5061	=item EV_CB_DECLARE (type)
		5062
		5063	=item EV_CB_INVOKE (watcher, revents)
		5064
		5065	=item ev_set_cb (ev, cb)
		5066
		5067	Can be used to change the callback member declaration in each watcher,
		5068	and the way callbacks are invoked and set. Must expand to a struct member
		5069	definition and a statement, respectively. See the F<ev.h> header file for
		5070	their default definitions. One possible use for overriding these is to
		5071	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		5072	method calls instead of plain function calls in C++.
		5073
		5074	=back
		5075
		5076	=head2 EXPORTED API SYMBOLS
		5077
		5078	If you need to re-export the API (e.g. via a DLL) and you need a list of
		5079	exported symbols, you can use the provided F<Symbol.*> files which list
		5080	all public symbols, one per line:
		5081
		5082	Symbols.ev for libev proper
		5083	Symbols.event for the libevent emulation
		5084
		5085	This can also be used to rename all public symbols to avoid clashes with
		5086	multiple versions of libev linked together (which is obviously bad in
		5087	itself, but sometimes it is inconvenient to avoid this).
		5088
		5089	A sed command like this will create wrapper C<#define>'s that you need to
		5090	include before including F<ev.h>:
		5091
		5092	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		5093
		5094	This would create a file F<wrap.h> which essentially looks like this:
		5095
		5096	#define ev_backend myprefix_ev_backend
		5097	#define ev_check_start myprefix_ev_check_start
		5098	#define ev_check_stop myprefix_ev_check_stop
		5099	...
		5100
		5101	=head2 EXAMPLES
		5102
		5103	For a real-world example of a program the includes libev
		5104	verbatim, you can have a look at the EV perl module
		5105	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		5106	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		5107	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		5108	will be compiled. It is pretty complex because it provides its own header
		5109	file.
		5110
		5111	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		5112	that everybody includes and which overrides some configure choices:
		5113
		5114	#define EV_FEATURES 8
		5115	#define EV_USE_SELECT 1
		5116	#define EV_PREPARE_ENABLE 1
		5117	#define EV_IDLE_ENABLE 1
		5118	#define EV_SIGNAL_ENABLE 1
		5119	#define EV_CHILD_ENABLE 1
		5120	#define EV_USE_STDEXCEPT 0
		5121	#define EV_CONFIG_H <config.h>
		5122
		5123	#include "ev++.h"
		5124
		5125	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		5126
		5127	#include "ev_cpp.h"
		5128	#include "ev.c"
		5129
		5130	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
		5131
		5132	=head2 THREADS AND COROUTINES
		5133
		5134	=head3 THREADS
		5135
		5136	All libev functions are reentrant and thread-safe unless explicitly
		5137	documented otherwise, but libev implements no locking itself. This means
		5138	that you can use as many loops as you want in parallel, as long as there
		5139	are no concurrent calls into any libev function with the same loop
		5140	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		5141	of course): libev guarantees that different event loops share no data
		5142	structures that need any locking.
		5143
		5144	Or to put it differently: calls with different loop parameters can be done
		5145	concurrently from multiple threads, calls with the same loop parameter
		5146	must be done serially (but can be done from different threads, as long as
		5147	only one thread ever is inside a call at any point in time, e.g. by using
		5148	a mutex per loop).
		5149
		5150	Specifically to support threads (and signal handlers), libev implements
		5151	so-called C<ev_async> watchers, which allow some limited form of
		5152	concurrency on the same event loop, namely waking it up "from the
		5153	outside".
		5154
		5155	If you want to know which design (one loop, locking, or multiple loops
		5156	without or something else still) is best for your problem, then I cannot
		5157	help you, but here is some generic advice:
		5158
		5159	=over 4
		5160
		5161	=item * most applications have a main thread: use the default libev loop
		5162	in that thread, or create a separate thread running only the default loop.
		5163
		5164	This helps integrating other libraries or software modules that use libev
		5165	themselves and don't care/know about threading.
		5166
		5167	=item * one loop per thread is usually a good model.
		5168
		5169	Doing this is almost never wrong, sometimes a better-performance model
		5170	exists, but it is always a good start.
		5171
		5172	=item * other models exist, such as the leader/follower pattern, where one
		5173	loop is handed through multiple threads in a kind of round-robin fashion.
		5174
		5175	Choosing a model is hard - look around, learn, know that usually you can do
		5176	better than you currently do :-)
		5177
		5178	=item * often you need to talk to some other thread which blocks in the
		5179	event loop.
		5180
		5181	C<ev_async> watchers can be used to wake them up from other threads safely
		5182	(or from signal contexts...).
		5183
		5184	An example use would be to communicate signals or other events that only
		5185	work in the default loop by registering the signal watcher with the
		5186	default loop and triggering an C<ev_async> watcher from the default loop
		5187	watcher callback into the event loop interested in the signal.
		5188
		5189	=back
		5190
		5191	See also L</THREAD LOCKING EXAMPLE>.
		5192
		5193	=head3 COROUTINES
		5194
		5195	Libev is very accommodating to coroutines ("cooperative threads"):
		5196	libev fully supports nesting calls to its functions from different
		5197	coroutines (e.g. you can call C<ev_run> on the same loop from two
		5198	different coroutines, and switch freely between both coroutines running
		5199	the loop, as long as you don't confuse yourself). The only exception is
		5200	that you must not do this from C<ev_periodic> reschedule callbacks.
		5201
		5202	Care has been taken to ensure that libev does not keep local state inside
		5203	C<ev_run>, and other calls do not usually allow for coroutine switches as
		5204	they do not call any callbacks.
		5205
		5206	=head2 COMPILER WARNINGS
		5207
		5208	Depending on your compiler and compiler settings, you might get no or a
		5209	lot of warnings when compiling libev code. Some people are apparently
		5210	scared by this.
		5211
		5212	However, these are unavoidable for many reasons. For one, each compiler
		5213	has different warnings, and each user has different tastes regarding
		5214	warning options. "Warn-free" code therefore cannot be a goal except when
		5215	targeting a specific compiler and compiler-version.
		5216
		5217	Another reason is that some compiler warnings require elaborate
		5218	workarounds, or other changes to the code that make it less clear and less
		5219	maintainable.
		5220
		5221	And of course, some compiler warnings are just plain stupid, or simply
		5222	wrong (because they don't actually warn about the condition their message
		5223	seems to warn about). For example, certain older gcc versions had some
		5224	warnings that resulted in an extreme number of false positives. These have
		5225	been fixed, but some people still insist on making code warn-free with
		5226	such buggy versions.
		5227
		5228	While libev is written to generate as few warnings as possible,
		5229	"warn-free" code is not a goal, and it is recommended not to build libev
		5230	with any compiler warnings enabled unless you are prepared to cope with
		5231	them (e.g. by ignoring them). Remember that warnings are just that:
		5232	warnings, not errors, or proof of bugs.
		5233
		5234
		5235	=head2 VALGRIND
		5236
		5237	Valgrind has a special section here because it is a popular tool that is
		5238	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		5239
		5240	If you think you found a bug (memory leak, uninitialised data access etc.)
		5241	in libev, then check twice: If valgrind reports something like:
		5242
		5243	==2274== definitely lost: 0 bytes in 0 blocks.
		5244	==2274== possibly lost: 0 bytes in 0 blocks.
		5245	==2274== still reachable: 256 bytes in 1 blocks.
		5246
		5247	Then there is no memory leak, just as memory accounted to global variables
		5248	is not a memleak - the memory is still being referenced, and didn't leak.
		5249
		5250	Similarly, under some circumstances, valgrind might report kernel bugs
		5251	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		5252	although an acceptable workaround has been found here), or it might be
		5253	confused.
		5254
		5255	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		5256	make it into some kind of religion.
		5257
		5258	If you are unsure about something, feel free to contact the mailing list
		5259	with the full valgrind report and an explanation on why you think this
		5260	is a bug in libev (best check the archives, too :). However, don't be
		5261	annoyed when you get a brisk "this is no bug" answer and take the chance
		5262	of learning how to interpret valgrind properly.
		5263
		5264	If you need, for some reason, empty reports from valgrind for your project
		5265	I suggest using suppression lists.
		5266
		5267
		5268	=head1 PORTABILITY NOTES
		5269
		5270	=head2 GNU/LINUX 32 BIT LIMITATIONS
		5271
		5272	GNU/Linux is the only common platform that supports 64 bit file/large file
		5273	interfaces but I<disables> them by default.
		5274
		5275	That means that libev compiled in the default environment doesn't support
		5276	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		5277
		5278	Unfortunately, many programs try to work around this GNU/Linux issue
		5279	by enabling the large file API, which makes them incompatible with the
		5280	standard libev compiled for their system.
		5281
		5282	Likewise, libev cannot enable the large file API itself as this would
		5283	suddenly make it incompatible to the default compile time environment,
		5284	i.e. all programs not using special compile switches.
		5285
		5286	=head2 OS/X AND DARWIN BUGS
		5287
		5288	The whole thing is a bug if you ask me - basically any system interface
		5289	you touch is broken, whether it is locales, poll, kqueue or even the
		5290	OpenGL drivers.
		5291
		5292	=head3 C<kqueue> is buggy
		5293
		5294	The kqueue syscall is broken in all known versions - most versions support
		5295	only sockets, many support pipes.
		5296
		5297	Libev tries to work around this by not using C<kqueue> by default on this
		5298	rotten platform, but of course you can still ask for it when creating a
		5299	loop - embedding a socket-only kqueue loop into a select-based one is
		5300	probably going to work well.
		5301
		5302	=head3 C<poll> is buggy
		5303
		5304	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		5305	implementation by something calling C<kqueue> internally around the 10.5.6
		5306	release, so now C<kqueue> I<and> C<poll> are broken.
		5307
		5308	Libev tries to work around this by not using C<poll> by default on
		5309	this rotten platform, but of course you can still ask for it when creating
		5310	a loop.
		5311
		5312	=head3 C<select> is buggy
		5313
		5314	All that's left is C<select>, and of course Apple found a way to fuck this
		5315	one up as well: On OS/X, C<select> actively limits the number of file
		5316	descriptors you can pass in to 1024 - your program suddenly crashes when
		5317	you use more.
		5318
		5319	There is an undocumented "workaround" for this - defining
		5320	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		5321	work on OS/X.
		5322
		5323	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		5324
		5325	=head3 C<errno> reentrancy
		5326
		5327	The default compile environment on Solaris is unfortunately so
		5328	thread-unsafe that you can't even use components/libraries compiled
		5329	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		5330	defined by default. A valid, if stupid, implementation choice.
		5331
		5332	If you want to use libev in threaded environments you have to make sure
		5333	it's compiled with C<_REENTRANT> defined.
		5334
		5335	=head3 Event port backend
		5336
		5337	The scalable event interface for Solaris is called "event
		5338	ports". Unfortunately, this mechanism is very buggy in all major
		5339	releases. If you run into high CPU usage, your program freezes or you get
		5340	a large number of spurious wakeups, make sure you have all the relevant
		5341	and latest kernel patches applied. No, I don't know which ones, but there
		5342	are multiple ones to apply, and afterwards, event ports actually work
		5343	great.
		5344
		5345	If you can't get it to work, you can try running the program by setting
		5346	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		5347	C<select> backends.
		5348
		5349	=head2 AIX POLL BUG
		5350
		5351	AIX unfortunately has a broken C<poll.h> header. Libev works around
		5352	this by trying to avoid the poll backend altogether (i.e. it's not even
		5353	compiled in), which normally isn't a big problem as C<select> works fine
		5354	with large bitsets on AIX, and AIX is dead anyway.
		5355
		5356	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		5357
		5358	=head3 General issues
		5359
		5360	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		5361	requires, and its I/O model is fundamentally incompatible with the POSIX
		5362	model. Libev still offers limited functionality on this platform in
		5363	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		5364	descriptors. This only applies when using Win32 natively, not when using
		5365	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		5366	as every compiler comes with a slightly differently broken/incompatible
		5367	environment.
		5368
		5369	Lifting these limitations would basically require the full
		5370	re-implementation of the I/O system. If you are into this kind of thing,
		5371	then note that glib does exactly that for you in a very portable way (note
		5372	also that glib is the slowest event library known to man).
		5373
		5374	There is no supported compilation method available on windows except
		5375	embedding it into other applications.
		5376
		5377	Sensible signal handling is officially unsupported by Microsoft - libev
		5378	tries its best, but under most conditions, signals will simply not work.
		5379
		5380	Not a libev limitation but worth mentioning: windows apparently doesn't
		5381	accept large writes: instead of resulting in a partial write, windows will
		5382	either accept everything or return C<ENOBUFS> if the buffer is too large,
		5383	so make sure you only write small amounts into your sockets (less than a
		5384	megabyte seems safe, but this apparently depends on the amount of memory
		5385	available).
		5386
		5387	Due to the many, low, and arbitrary limits on the win32 platform and
		5388	the abysmal performance of winsockets, using a large number of sockets
		5389	is not recommended (and not reasonable). If your program needs to use
		5390	more than a hundred or so sockets, then likely it needs to use a totally
		5391	different implementation for windows, as libev offers the POSIX readiness
		5392	notification model, which cannot be implemented efficiently on windows
		5393	(due to Microsoft monopoly games).
		5394
		5395	A typical way to use libev under windows is to embed it (see the embedding
		5396	section for details) and use the following F<evwrap.h> header file instead
		5397	of F<ev.h>:
		5398
		5399	#define EV_STANDALONE /* keeps ev from requiring config.h */
		5400	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		5401
		5402	#include "ev.h"
		5403
		5404	And compile the following F<evwrap.c> file into your project (make sure
		5405	you do I<not> compile the F<ev.c> or any other embedded source files!):
		5406
		5407	#include "evwrap.h"
		5408	#include "ev.c"
		5409
		5410	=head3 The winsocket C<select> function
		5411
		5412	The winsocket C<select> function doesn't follow POSIX in that it
		5413	requires socket I<handles> and not socket I<file descriptors> (it is
		5414	also extremely buggy). This makes select very inefficient, and also
		5415	requires a mapping from file descriptors to socket handles (the Microsoft
		5416	C runtime provides the function C<_open_osfhandle> for this). See the
		5417	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		5418	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		5419
		5420	The configuration for a "naked" win32 using the Microsoft runtime
		5421	libraries and raw winsocket select is:
		5422
		5423	#define EV_USE_SELECT 1
		5424	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		5425
		5426	Note that winsockets handling of fd sets is O(n), so you can easily get a
		5427	complexity in the O(n²) range when using win32.
		5428
		5429	=head3 Limited number of file descriptors
		5430
		5431	Windows has numerous arbitrary (and low) limits on things.
		5432
		5433	Early versions of winsocket's select only supported waiting for a maximum
		5434	of C<64> handles (probably owning to the fact that all windows kernels
		5435	can only wait for C<64> things at the same time internally; Microsoft
		5436	recommends spawning a chain of threads and wait for 63 handles and the
		5437	previous thread in each. Sounds great!).
		5438
		5439	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		5440	to some high number (e.g. C<2048>) before compiling the winsocket select
		5441	call (which might be in libev or elsewhere, for example, perl and many
		5442	other interpreters do their own select emulation on windows).
		5443
		5444	Another limit is the number of file descriptors in the Microsoft runtime
		5445	libraries, which by default is C<64> (there must be a hidden I<64>
		5446	fetish or something like this inside Microsoft). You can increase this
		5447	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		5448	(another arbitrary limit), but is broken in many versions of the Microsoft
		5449	runtime libraries. This might get you to about C<512> or C<2048> sockets
		5450	(depending on windows version and/or the phase of the moon). To get more,
		5451	you need to wrap all I/O functions and provide your own fd management, but
		5452	the cost of calling select (O(n²)) will likely make this unworkable.
		5453
		5454	=head2 PORTABILITY REQUIREMENTS
		5455
		5456	In addition to a working ISO-C implementation and of course the
		5457	backend-specific APIs, libev relies on a few additional extensions:
		5458
		5459	=over 4
		5460
		5461	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		5462	calling conventions regardless of C<ev_watcher_type *>.
		5463
		5464	Libev assumes not only that all watcher pointers have the same internal
		5465	structure (guaranteed by POSIX but not by ISO C for example), but it also
		5466	assumes that the same (machine) code can be used to call any watcher
		5467	callback: The watcher callbacks have different type signatures, but libev
		5468	calls them using an C<ev_watcher *> internally.
		5469
		5470	=item null pointers and integer zero are represented by 0 bytes
		5471
		5472	Libev uses C<memset> to initialise structs and arrays to C<0> bytes, and
		5473	relies on this setting pointers and integers to null.
		5474
		5475	=item pointer accesses must be thread-atomic
		5476
		5477	Accessing a pointer value must be atomic, it must both be readable and
		5478	writable in one piece - this is the case on all current architectures.
		5479
		5480	=item C<sig_atomic_t volatile> must be thread-atomic as well
		5481
		5482	The type C<sig_atomic_t volatile> (or whatever is defined as
		5483	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		5484	threads. This is not part of the specification for C<sig_atomic_t>, but is
		5485	believed to be sufficiently portable.
		5486
		5487	=item C<sigprocmask> must work in a threaded environment
		5488
		5489	Libev uses C<sigprocmask> to temporarily block signals. This is not
		5490	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		5491	pthread implementations will either allow C<sigprocmask> in the "main
		5492	thread" or will block signals process-wide, both behaviours would
		5493	be compatible with libev. Interaction between C<sigprocmask> and
		5494	C<pthread_sigmask> could complicate things, however.
		5495
		5496	The most portable way to handle signals is to block signals in all threads
		5497	except the initial one, and run the signal handling loop in the initial
		5498	thread as well.
		5499
		5500	=item C<long> must be large enough for common memory allocation sizes
		5501
		5502	To improve portability and simplify its API, libev uses C<long> internally
		5503	instead of C<size_t> when allocating its data structures. On non-POSIX
		5504	systems (Microsoft...) this might be unexpectedly low, but is still at
		5505	least 31 bits everywhere, which is enough for hundreds of millions of
		5506	watchers.
		5507
		5508	=item C<double> must hold a time value in seconds with enough accuracy
		5509
		5510	The type C<double> is used to represent timestamps. It is required to
		5511	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		5512	good enough for at least into the year 4000 with millisecond accuracy
		5513	(the design goal for libev). This requirement is overfulfilled by
		5514	implementations using IEEE 754, which is basically all existing ones.
		5515
		5516	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5517	year 2255 (and millisecond accuracy till the year 287396 - by then, libev
		5518	is either obsolete or somebody patched it to use C<long double> or
		5519	something like that, just kidding).
		5520
		5521	=back
		5522
		5523	If you know of other additional requirements drop me a note.
		5524
		5525
		5526	=head1 ALGORITHMIC COMPLEXITIES
		5527
		5528	In this section the complexities of (many of) the algorithms used inside
		5529	libev will be documented. For complexity discussions about backends see
		5530	the documentation for C<ev_default_init>.
		5531
		5532	All of the following are about amortised time: If an array needs to be
		5533	extended, libev needs to realloc and move the whole array, but this
		5534	happens asymptotically rarer with higher number of elements, so O(1) might
		5535	mean that libev does a lengthy realloc operation in rare cases, but on
		5536	average it is much faster and asymptotically approaches constant time.
		5537
		5538	=over 4
		5539
		5540	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		5541
		5542	This means that, when you have a watcher that triggers in one hour and
		5543	there are 100 watchers that would trigger before that, then inserting will
		5544	have to skip roughly seven (C<ld 100>) of these watchers.
		5545
		5546	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		5547
		5548	That means that changing a timer costs less than removing/adding them,
		5549	as only the relative motion in the event queue has to be paid for.
		5550
		5551	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		5552
		5553	These just add the watcher into an array or at the head of a list.
		5554
		5555	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		5556
		5557	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		5558
		5559	These watchers are stored in lists, so they need to be walked to find the
		5560	correct watcher to remove. The lists are usually short (you don't usually
		5561	have many watchers waiting for the same fd or signal: one is typical, two
		5562	is rare).
		5563
		5564	=item Finding the next timer in each loop iteration: O(1)
		5565
		5566	By virtue of using a binary or 4-heap, the next timer is always found at a
		5567	fixed position in the storage array.
		5568
		5569	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		5570
		5571	A change means an I/O watcher gets started or stopped, which requires
		5572	libev to recalculate its status (and possibly tell the kernel, depending
		5573	on backend and whether C<ev_io_set> was used).
		5574
		5575	=item Activating one watcher (putting it into the pending state): O(1)
		5576
		5577	=item Priority handling: O(number_of_priorities)
		5578
		5579	Priorities are implemented by allocating some space for each
		5580	priority. When doing priority-based operations, libev usually has to
		5581	linearly search all the priorities, but starting/stopping and activating
		5582	watchers becomes O(1) with respect to priority handling.
		5583
		5584	=item Sending an ev_async: O(1)
		5585
		5586	=item Processing ev_async_send: O(number_of_async_watchers)
		5587
		5588	=item Processing signals: O(max_signal_number)
		5589
		5590	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5591	calls in the current loop iteration and the loop is currently
		5592	blocked. Checking for async and signal events involves iterating over all
		5593	running async watchers or all signal numbers.
		5594
		5595	=back
		5596
		5597
		5598	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5599
		5600	The major version 4 introduced some incompatible changes to the API.
		5601
		5602	At the moment, the C<ev.h> header file provides compatibility definitions
		5603	for all changes, so most programs should still compile. The compatibility
		5604	layer might be removed in later versions of libev, so better update to the
		5605	new API early than late.
		5606
		5607	=over 4
		5608
		5609	=item C<EV_COMPAT3> backwards compatibility mechanism
		5610
		5611	The backward compatibility mechanism can be controlled by
		5612	C<EV_COMPAT3>. See L</"PREPROCESSOR SYMBOLS/MACROS"> in the L</EMBEDDING>
		5613	section.
		5614
		5615	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5616
		5617	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5618
		5619	ev_loop_destroy (EV_DEFAULT_UC);
		5620	ev_loop_fork (EV_DEFAULT);
		5621
		5622	=item function/symbol renames
		5623
		5624	A number of functions and symbols have been renamed:
		5625
		5626	ev_loop => ev_run
		5627	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5628	EVLOOP_ONESHOT => EVRUN_ONCE
		5629
		5630	ev_unloop => ev_break
		5631	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5632	EVUNLOOP_ONE => EVBREAK_ONE
		5633	EVUNLOOP_ALL => EVBREAK_ALL
		5634
		5635	EV_TIMEOUT => EV_TIMER
		5636
		5637	ev_loop_count => ev_iteration
		5638	ev_loop_depth => ev_depth
		5639	ev_loop_verify => ev_verify
		5640
		5641	Most functions working on C<struct ev_loop> objects don't have an
		5642	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5643	associated constants have been renamed to not collide with the C<struct
		5644	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5645	as all other watcher types. Note that C<ev_loop_fork> is still called
		5646	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5647	typedef.
		5648
		5649	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5650
		5651	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5652	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5653	and work, but the library code will of course be larger.
		5654
		5655	=back
		5656
		5657
		5658	=head1 GLOSSARY
		5659
		5660	=over 4
		5661
		5662	=item active
		5663
		5664	A watcher is active as long as it has been started and not yet stopped.
		5665	See L</WATCHER STATES> for details.
		5666
		5667	=item application
		5668
		5669	In this document, an application is whatever is using libev.
		5670
		5671	=item backend
		5672
		5673	The part of the code dealing with the operating system interfaces.
		5674
		5675	=item callback
		5676
		5677	The address of a function that is called when some event has been
		5678	detected. Callbacks are being passed the event loop, the watcher that
		5679	received the event, and the actual event bitset.
		5680
		5681	=item callback/watcher invocation
		5682
		5683	The act of calling the callback associated with a watcher.
		5684
		5685	=item event
		5686
		5687	A change of state of some external event, such as data now being available
		5688	for reading on a file descriptor, time having passed or simply not having
		5689	any other events happening anymore.
		5690
		5691	In libev, events are represented as single bits (such as C<EV_READ> or
		5692	C<EV_TIMER>).
		5693
		5694	=item event library
		5695
		5696	A software package implementing an event model and loop.
		5697
		5698	=item event loop
		5699
		5700	An entity that handles and processes external events and converts them
		5701	into callback invocations.
		5702
		5703	=item event model
		5704
		5705	The model used to describe how an event loop handles and processes
		5706	watchers and events.
		5707
		5708	=item pending
		5709
		5710	A watcher is pending as soon as the corresponding event has been
		5711	detected. See L</WATCHER STATES> for details.
		5712
		5713	=item real time
		5714
		5715	The physical time that is observed. It is apparently strictly monotonic :)
		5716
		5717	=item wall-clock time
		5718
		5719	The time and date as shown on clocks. Unlike real time, it can actually
		5720	be wrong and jump forwards and backwards, e.g. when you adjust your
		5721	clock.
		5722
		5723	=item watcher
		5724
		5725	A data structure that describes interest in certain events. Watchers need
		5726	to be started (attached to an event loop) before they can receive events.
		5727
		5728	=back
878		5729
879	=head1 AUTHOR	5730	=head1 AUTHOR
880		5731
881	Marc Lehmann <libev@schmorp.de>.	5732	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5733	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
882		5734

Diff Legend

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