ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.8 by root, Mon Nov 12 08:20:02 2007 UTC vs.
Revision 1.356 by root, Tue Jan 11 01:42:47 2011 UTC

…		…
2		2
3	libev - a high performance full-featured event loop written in C	3	libev - a high performance full-featured event loop written in C
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	#include <ev.h>	7	#include <ev.h>
8		8
9	=head1 DESCRIPTION	9	=head2 EXAMPLE PROGRAM
		10
		11	// a single header file is required
		12	#include <ev.h>
		13
		14	#include <stdio.h> // for puts
		15
		16	// every watcher type has its own typedef'd struct
		17	// with the name ev_TYPE
		18	ev_io stdin_watcher;
		19	ev_timer timeout_watcher;
		20
		21	// all watcher callbacks have a similar signature
		22	// this callback is called when data is readable on stdin
		23	static void
		24	stdin_cb (EV_P_ ev_io *w, int revents)
		25	{
		26	puts ("stdin ready");
		27	// for one-shot events, one must manually stop the watcher
		28	// with its corresponding stop function.
		29	ev_io_stop (EV_A_ w);
		30
		31	// this causes all nested ev_run's to stop iterating
		32	ev_break (EV_A_ EVBREAK_ALL);
		33	}
		34
		35	// another callback, this time for a time-out
		36	static void
		37	timeout_cb (EV_P_ ev_timer *w, int revents)
		38	{
		39	puts ("timeout");
		40	// this causes the innermost ev_run to stop iterating
		41	ev_break (EV_A_ EVBREAK_ONE);
		42	}
		43
		44	int
		45	main (void)
		46	{
		47	// use the default event loop unless you have special needs
		48	struct ev_loop *loop = EV_DEFAULT;
		49
		50	// initialise an io watcher, then start it
		51	// this one will watch for stdin to become readable
		52	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
		53	ev_io_start (loop, &stdin_watcher);
		54
		55	// initialise a timer watcher, then start it
		56	// simple non-repeating 5.5 second timeout
		57	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		58	ev_timer_start (loop, &timeout_watcher);
		59
		60	// now wait for events to arrive
		61	ev_run (loop, 0);
		62
		63	// unloop was called, so exit
		64	return 0;
		65	}
		66
		67	=head1 ABOUT THIS DOCUMENT
		68
		69	This document documents the libev software package.
		70
		71	The newest version of this document is also available as an html-formatted
		72	web page you might find easier to navigate when reading it for the first
		73	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		74
		75	While this document tries to be as complete as possible in documenting
		76	libev, its usage and the rationale behind its design, it is not a tutorial
		77	on event-based programming, nor will it introduce event-based programming
		78	with libev.
		79
		80	Familiarity with event based programming techniques in general is assumed
		81	throughout this document.
		82
		83	=head1 WHAT TO READ WHEN IN A HURRY
		84
		85	This manual tries to be very detailed, but unfortunately, this also makes
		86	it very long. If you just want to know the basics of libev, I suggest
		87	reading L<ANATOMY OF A WATCHER>, then the L<EXAMPLE PROGRAM> above and
		88	look up the missing functions in L<GLOBAL FUNCTIONS> and the C<ev_io> and
		89	C<ev_timer> sections in L<WATCHER TYPES>.
		90
		91	=head1 ABOUT LIBEV
10		92
11	Libev is an event loop: you register interest in certain events (such as a	93	Libev is an event loop: you register interest in certain events (such as a
12	file descriptor being readable or a timeout occuring), and it will manage	94	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	95	these event sources and provide your program with events.
14		96
15	To do this, it must take more or less complete control over your process	97	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	98	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	99	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	101	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	102	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	103	details of the event, and then hand it over to libev by I<starting> the
22	watcher.	104	watcher.
23		105
24	=head1 FEATURES	106	=head2 FEATURES
25		107
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	108	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27	kqueue mechanisms for file descriptor events, relative timers, absolute	109	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28	timers with customised rescheduling, signal events, process status change	110	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
		111	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
		112	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
		113	timers (C<ev_timer>), absolute timers with customised rescheduling
		114	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
29	events (related to SIGCHLD), and event watchers dealing with the event	115	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	116	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
		117	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		118	limited support for fork events (C<ev_fork>).
		119
		120	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	121	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	122	for example).
33		123
34	=head1 CONVENTIONS	124	=head2 CONVENTIONS
35		125
36	Libev is very configurable. In this manual the default configuration	126	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	127	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	128	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	129	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	130	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 *>)	131	name C<loop> (which is always of type C<struct ev_loop *>) will not have
42	will not have this argument.	132	this argument.
43		133
44	=head1 TIME AND OTHER GLOBAL FUNCTIONS	134	=head2 TIME REPRESENTATION
45		135
46	Libev represents time as a single floating point number, representing the	136	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	137	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	138	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	139	ask). This type is called C<ev_tstamp>, which is what you should use
50	to the double type in C.	140	too. It usually aliases to the C<double> type in C. When you need to do
		141	any calculations on it, you should treat it as some floating point value.
		142
		143	Unlike the name component C<stamp> might indicate, it is also used for
		144	time differences (e.g. delays) throughout libev.
		145
		146	=head1 ERROR HANDLING
		147
		148	Libev knows three classes of errors: operating system errors, usage errors
		149	and internal errors (bugs).
		150
		151	When libev catches an operating system error it cannot handle (for example
		152	a system call indicating a condition libev cannot fix), it calls the callback
		153	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		154	abort. The default is to print a diagnostic message and to call C<abort
		155	()>.
		156
		157	When libev detects a usage error such as a negative timer interval, then
		158	it will print a diagnostic message and abort (via the C<assert> mechanism,
		159	so C<NDEBUG> will disable this checking): these are programming errors in
		160	the libev caller and need to be fixed there.
		161
		162	Libev also has a few internal error-checking C<assert>ions, and also has
		163	extensive consistency checking code. These do not trigger under normal
		164	circumstances, as they indicate either a bug in libev or worse.
		165
		166
		167	=head1 GLOBAL FUNCTIONS
		168
		169	These functions can be called anytime, even before initialising the
		170	library in any way.
51		171
52	=over 4	172	=over 4
53		173
54	=item ev_tstamp ev_time ()	174	=item ev_tstamp ev_time ()
55		175
56	Returns the current time as libev would use it.	176	Returns the current time as libev would use it. Please note that the
		177	C<ev_now> function is usually faster and also often returns the timestamp
		178	you actually want to know. Also interesting is the combination of
		179	C<ev_update_now> and C<ev_now>.
		180
		181	=item ev_sleep (ev_tstamp interval)
		182
		183	Sleep for the given interval: The current thread will be blocked until
		184	either it is interrupted or the given time interval has passed. Basically
		185	this is a sub-second-resolution C<sleep ()>.
57		186
58	=item int ev_version_major ()	187	=item int ev_version_major ()
59		188
60	=item int ev_version_minor ()	189	=item int ev_version_minor ()
61		190
62	You can find out the major and minor version numbers of the library	191	You can find out the major and minor ABI version numbers of the library
63	you linked against by calling the functions C<ev_version_major> and	192	you linked against by calling the functions C<ev_version_major> and
64	C<ev_version_minor>. If you want, you can compare against the global	193	C<ev_version_minor>. If you want, you can compare against the global
65	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	194	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
66	version of the library your program was compiled against.	195	version of the library your program was compiled against.
67		196
		197	These version numbers refer to the ABI version of the library, not the
		198	release version.
		199
68	Usually, its a good idea to terminate if the major versions mismatch,	200	Usually, it's a good idea to terminate if the major versions mismatch,
69	as this indicates an incompatible change. Minor versions are usually	201	as this indicates an incompatible change. Minor versions are usually
70	compatible to older versions, so a larger minor version alone is usually	202	compatible to older versions, so a larger minor version alone is usually
71	not a problem.	203	not a problem.
72		204
		205	Example: Make sure we haven't accidentally been linked against the wrong
		206	version (note, however, that this will not detect other ABI mismatches,
		207	such as LFS or reentrancy).
		208
		209	assert (("libev version mismatch",
		210	ev_version_major () == EV_VERSION_MAJOR
		211	&& ev_version_minor () >= EV_VERSION_MINOR));
		212
		213	=item unsigned int ev_supported_backends ()
		214
		215	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		216	value) compiled into this binary of libev (independent of their
		217	availability on the system you are running on). See C<ev_default_loop> for
		218	a description of the set values.
		219
		220	Example: make sure we have the epoll method, because yeah this is cool and
		221	a must have and can we have a torrent of it please!!!11
		222
		223	assert (("sorry, no epoll, no sex",
		224	ev_supported_backends () & EVBACKEND_EPOLL));
		225
		226	=item unsigned int ev_recommended_backends ()
		227
		228	Return the set of all backends compiled into this binary of libev and
		229	also recommended for this platform, meaning it will work for most file
		230	descriptor types. This set is often smaller than the one returned by
		231	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
		232	and will not be auto-detected unless you explicitly request it (assuming
		233	you know what you are doing). This is the set of backends that libev will
		234	probe for if you specify no backends explicitly.
		235
		236	=item unsigned int ev_embeddable_backends ()
		237
		238	Returns the set of backends that are embeddable in other event loops. This
		239	value is platform-specific but can include backends not available on the
		240	current system. To find which embeddable backends might be supported on
		241	the current system, you would need to look at C<ev_embeddable_backends ()
		242	& ev_supported_backends ()>, likewise for recommended ones.
		243
		244	See the description of C<ev_embed> watchers for more info.
		245
73	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	246	=item ev_set_allocator (void *(*cb)(void *ptr, long size))
74		247
75	Sets the allocation function to use (the prototype is similar to the	248	Sets the allocation function to use (the prototype is similar - the
76	realloc C function, the semantics are identical). It is used to allocate	249	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
77	and free memory (no surprises here). If it returns zero when memory	250	used to allocate and free memory (no surprises here). If it returns zero
78	needs to be allocated, the library might abort or take some potentially	251	when memory needs to be allocated (C<size != 0>), the library might abort
79	destructive action. The default is your system realloc function.	252	or take some potentially destructive action.
		253
		254	Since some systems (at least OpenBSD and Darwin) fail to implement
		255	correct C<realloc> semantics, libev will use a wrapper around the system
		256	C<realloc> and C<free> functions by default.
80		257
81	You could override this function in high-availability programs to, say,	258	You could override this function in high-availability programs to, say,
82	free some memory if it cannot allocate memory, to use a special allocator,	259	free some memory if it cannot allocate memory, to use a special allocator,
83	or even to sleep a while and retry until some memory is available.	260	or even to sleep a while and retry until some memory is available.
84		261
		262	Example: Replace the libev allocator with one that waits a bit and then
		263	retries (example requires a standards-compliant C<realloc>).
		264
		265	static void *
		266	persistent_realloc (void *ptr, size_t size)
		267	{
		268	for (;;)
		269	{
		270	void *newptr = realloc (ptr, size);
		271
		272	if (newptr)
		273	return newptr;
		274
		275	sleep (60);
		276	}
		277	}
		278
		279	...
		280	ev_set_allocator (persistent_realloc);
		281
85	=item ev_set_syserr_cb (void (*cb)(const char *msg));	282	=item ev_set_syserr_cb (void (*cb)(const char *msg))
86		283
87	Set the callback function to call on a retryable syscall error (such	284	Set the callback function to call on a retryable system call error (such
88	as failed select, poll, epoll_wait). The message is a printable string	285	as failed select, poll, epoll_wait). The message is a printable string
89	indicating the system call or subsystem causing the problem. If this	286	indicating the system call or subsystem causing the problem. If this
90	callback is set, then libev will expect it to remedy the sitution, no	287	callback is set, then libev will expect it to remedy the situation, no
91	matter what, when it returns. That is, libev will generally retry the	288	matter what, when it returns. That is, libev will generally retry the
92	requested operation, or, if the condition doesn't go away, do bad stuff	289	requested operation, or, if the condition doesn't go away, do bad stuff
93	(such as abort).	290	(such as abort).
94		291
		292	Example: This is basically the same thing that libev does internally, too.
		293
		294	static void
		295	fatal_error (const char *msg)
		296	{
		297	perror (msg);
		298	abort ();
		299	}
		300
		301	...
		302	ev_set_syserr_cb (fatal_error);
		303
		304	=item ev_feed_signal (int signum)
		305
		306	This function can be used to "simulate" a signal receive. It is completely
		307	safe to call this function at any time, from any context, including signal
		308	handlers or random threads.
		309
		310	Its main use is to customise signal handling in your process, especially
		311	in the presence of threads. For example, you could block signals
		312	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		313	creating any loops), and in one thread, use C<sigwait> or any other
		314	mechanism to wait for signals, then "deliver" them to libev by calling
		315	C<ev_feed_signal>.
		316
95	=back	317	=back
96		318
97	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	319	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
98		320
99	An event loop is described by a C<struct ev_loop *>. The library knows two	321	An event loop is described by a C<struct ev_loop *> (the C<struct> is
100	types of such loops, the I<default> loop, which supports signals and child	322	I<not> optional in this case unless libev 3 compatibility is disabled, as
101	events, and dynamically created loops which do not.	323	libev 3 had an C<ev_loop> function colliding with the struct name).
102		324
103	If you use threads, a common model is to run the default event loop	325	The library knows two types of such loops, the I<default> loop, which
104	in your main thread (or in a separate thrad) and for each thread you	326	supports child process events, and dynamically created event loops which
105	create, you also create another event loop. Libev itself does no locking	327	do not.
106	whatsoever, so if you mix calls to the same event loop in different
107	threads, make sure you lock (this is usually a bad idea, though, even if
108	done correctly, because its hideous and inefficient).
109		328
110	=over 4	329	=over 4
111		330
112	=item struct ev_loop *ev_default_loop (unsigned int flags)	331	=item struct ev_loop *ev_default_loop (unsigned int flags)
113		332
114	This will initialise the default event loop if it hasn't been initialised	333	This returns the "default" event loop object, which is what you should
115	yet and return it. If the default loop could not be initialised, returns	334	normally use when you just need "the event loop". Event loop objects and
116	false. If it already was initialised it simply returns it (and ignores the	335	the C<flags> parameter are described in more detail in the entry for
117	flags).	336	C<ev_loop_new>.
		337
		338	If the default loop is already initialised then this function simply
		339	returns it (and ignores the flags. If that is troubling you, check
		340	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		341	flags, which should almost always be C<0>, unless the caller is also the
		342	one calling C<ev_run> or otherwise qualifies as "the main program".
118		343
119	If you don't know what event loop to use, use the one returned from this	344	If you don't know what event loop to use, use the one returned from this
120	function.	345	function (or via the C<EV_DEFAULT> macro).
		346
		347	Note that this function is I<not> thread-safe, so if you want to use it
		348	from multiple threads, you have to employ some kind of mutex (note also
		349	that this case is unlikely, as loops cannot be shared easily between
		350	threads anyway).
		351
		352	The default loop is the only loop that can handle C<ev_child> watchers,
		353	and to do this, it always registers a handler for C<SIGCHLD>. If this is
		354	a problem for your application you can either create a dynamic loop with
		355	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
		356	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
		357
		358	Example: This is the most typical usage.
		359
		360	if (!ev_default_loop (0))
		361	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		362
		363	Example: Restrict libev to the select and poll backends, and do not allow
		364	environment settings to be taken into account:
		365
		366	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		367
		368	=item struct ev_loop *ev_loop_new (unsigned int flags)
		369
		370	This will create and initialise a new event loop object. If the loop
		371	could not be initialised, returns false.
		372
		373	This function is thread-safe, and one common way to use libev with
		374	threads is indeed to create one loop per thread, and using the default
		375	loop in the "main" or "initial" thread.
121		376
122	The flags argument can be used to specify special behaviour or specific	377	The flags argument can be used to specify special behaviour or specific
123	backends to use, and is usually specified as 0 (or EVFLAG_AUTO).	378	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
124		379
125	It supports the following flags:	380	The following flags are supported:
126		381
127	=over 4	382	=over 4
128		383
129	=item EVFLAG_AUTO	384	=item C<EVFLAG_AUTO>
130		385
131	The default flags value. Use this if you have no clue (its the right	386	The default flags value. Use this if you have no clue (it's the right
132	thing, believe me).	387	thing, believe me).
133		388
134	=item EVFLAG_NOENV	389	=item C<EVFLAG_NOENV>
135		390
136	If this flag bit is ored into the flag value (or the program runs setuid	391	If this flag bit is or'ed into the flag value (or the program runs setuid
137	or setgid) then libev will I<not> look at the environment variable	392	or setgid) then libev will I<not> look at the environment variable
138	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	393	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
139	override the flags completely if it is found in the environment. This is	394	override the flags completely if it is found in the environment. This is
140	useful to try out specific backends to test their performance, or to work	395	useful to try out specific backends to test their performance, or to work
141	around bugs.	396	around bugs.
142		397
		398	=item C<EVFLAG_FORKCHECK>
		399
		400	Instead of calling C<ev_loop_fork> manually after a fork, you can also
		401	make libev check for a fork in each iteration by enabling this flag.
		402
		403	This works by calling C<getpid ()> on every iteration of the loop,
		404	and thus this might slow down your event loop if you do a lot of loop
		405	iterations and little real work, but is usually not noticeable (on my
		406	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		407	without a system call and thus I<very> fast, but my GNU/Linux system also has
		408	C<pthread_atfork> which is even faster).
		409
		410	The big advantage of this flag is that you can forget about fork (and
		411	forget about forgetting to tell libev about forking) when you use this
		412	flag.
		413
		414	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		415	environment variable.
		416
		417	=item C<EVFLAG_NOINOTIFY>
		418
		419	When this flag is specified, then libev will not attempt to use the
		420	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
		421	testing, this flag can be useful to conserve inotify file descriptors, as
		422	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		423
		424	=item C<EVFLAG_SIGNALFD>
		425
		426	When this flag is specified, then libev will attempt to use the
		427	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
		428	delivers signals synchronously, which makes it both faster and might make
		429	it possible to get the queued signal data. It can also simplify signal
		430	handling with threads, as long as you properly block signals in your
		431	threads that are not interested in handling them.
		432
		433	Signalfd will not be used by default as this changes your signal mask, and
		434	there are a lot of shoddy libraries and programs (glib's threadpool for
		435	example) that can't properly initialise their signal masks.
		436
		437	=item C<EVFLAG_NOSIGMASK>
		438
		439	When this flag is specified, then libev will avoid to modify the signal
		440	mask. Specifically, this means you ahve to make sure signals are unblocked
		441	when you want to receive them.
		442
		443	This behaviour is useful when you want to do your own signal handling, or
		444	want to handle signals only in specific threads and want to avoid libev
		445	unblocking the signals.
		446
		447	This flag's behaviour will become the default in future versions of libev.
		448
143	=item EVMETHOD_SELECT portable select backend	449	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
144		450
145	=item EVMETHOD_POLL poll backend (everywhere except windows)	451	This is your standard select(2) backend. Not I<completely> standard, as
		452	libev tries to roll its own fd_set with no limits on the number of fds,
		453	but if that fails, expect a fairly low limit on the number of fds when
		454	using this backend. It doesn't scale too well (O(highest_fd)), but its
		455	usually the fastest backend for a low number of (low-numbered :) fds.
146		456
147	=item EVMETHOD_EPOLL linux only	457	To get good performance out of this backend you need a high amount of
		458	parallelism (most of the file descriptors should be busy). If you are
		459	writing a server, you should C<accept ()> in a loop to accept as many
		460	connections as possible during one iteration. You might also want to have
		461	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		462	readiness notifications you get per iteration.
148		463
149	=item EVMETHOD_KQUEUE some bsds only	464	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		465	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		466	C<exceptfds> set on that platform).
150		467
151	=item EVMETHOD_DEVPOLL solaris 8 only	468	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
152		469
153	=item EVMETHOD_PORT solaris 10 only	470	And this is your standard poll(2) backend. It's more complicated
		471	than select, but handles sparse fds better and has no artificial
		472	limit on the number of fds you can use (except it will slow down
		473	considerably with a lot of inactive fds). It scales similarly to select,
		474	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		475	performance tips.
154		476
155	If one or more of these are ored into the flags value, then only these	477	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
156	backends will be tried (in the reverse order as given here). If one are	478	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
157	specified, any backend will do.	479
		480	=item C<EVBACKEND_EPOLL> (value 4, Linux)
		481
		482	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		483	kernels).
		484
		485	For few fds, this backend is a bit little slower than poll and select,
		486	but it scales phenomenally better. While poll and select usually scale
		487	like O(total_fds) where n is the total number of fds (or the highest fd),
		488	epoll scales either O(1) or O(active_fds).
		489
		490	The epoll mechanism deserves honorable mention as the most misdesigned
		491	of the more advanced event mechanisms: mere annoyances include silently
		492	dropping file descriptors, requiring a system call per change per file
		493	descriptor (and unnecessary guessing of parameters), problems with dup,
		494	returning before the timeout value, resulting in additional iterations
		495	(and only giving 5ms accuracy while select on the same platform gives
		496	0.1ms) and so on. The biggest issue is fork races, however - if a program
		497	forks then I<both> parent and child process have to recreate the epoll
		498	set, which can take considerable time (one syscall per file descriptor)
		499	and is of course hard to detect.
		500
		501	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		502	of course I<doesn't>, and epoll just loves to report events for totally
		503	I<different> file descriptors (even already closed ones, so one cannot
		504	even remove them from the set) than registered in the set (especially
		505	on SMP systems). Libev tries to counter these spurious notifications by
		506	employing an additional generation counter and comparing that against the
		507	events to filter out spurious ones, recreating the set when required. Last
		508	not least, it also refuses to work with some file descriptors which work
		509	perfectly fine with C<select> (files, many character devices...).
		510
		511	Epoll is truly the train wreck analog among event poll mechanisms,
		512	a frankenpoll, cobbled together in a hurry, no thought to design or
		513	interaction with others.
		514
		515	While stopping, setting and starting an I/O watcher in the same iteration
		516	will result in some caching, there is still a system call per such
		517	incident (because the same I<file descriptor> could point to a different
		518	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
		519	file descriptors might not work very well if you register events for both
		520	file descriptors.
		521
		522	Best performance from this backend is achieved by not unregistering all
		523	watchers for a file descriptor until it has been closed, if possible,
		524	i.e. keep at least one watcher active per fd at all times. Stopping and
		525	starting a watcher (without re-setting it) also usually doesn't cause
		526	extra overhead. A fork can both result in spurious notifications as well
		527	as in libev having to destroy and recreate the epoll object, which can
		528	take considerable time and thus should be avoided.
		529
		530	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		531	faster than epoll for maybe up to a hundred file descriptors, depending on
		532	the usage. So sad.
		533
		534	While nominally embeddable in other event loops, this feature is broken in
		535	all kernel versions tested so far.
		536
		537	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		538	C<EVBACKEND_POLL>.
		539
		540	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
		541
		542	Kqueue deserves special mention, as at the time of this writing, it
		543	was broken on all BSDs except NetBSD (usually it doesn't work reliably
		544	with anything but sockets and pipes, except on Darwin, where of course
		545	it's completely useless). Unlike epoll, however, whose brokenness
		546	is by design, these kqueue bugs can (and eventually will) be fixed
		547	without API changes to existing programs. For this reason it's not being
		548	"auto-detected" unless you explicitly specify it in the flags (i.e. using
		549	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		550	system like NetBSD.
		551
		552	You still can embed kqueue into a normal poll or select backend and use it
		553	only for sockets (after having made sure that sockets work with kqueue on
		554	the target platform). See C<ev_embed> watchers for more info.
		555
		556	It scales in the same way as the epoll backend, but the interface to the
		557	kernel is more efficient (which says nothing about its actual speed, of
		558	course). While stopping, setting and starting an I/O watcher does never
		559	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
		560	two event changes per incident. Support for C<fork ()> is very bad (but
		561	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		562	cases
		563
		564	This backend usually performs well under most conditions.
		565
		566	While nominally embeddable in other event loops, this doesn't work
		567	everywhere, so you might need to test for this. And since it is broken
		568	almost everywhere, you should only use it when you have a lot of sockets
		569	(for which it usually works), by embedding it into another event loop
		570	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		571	also broken on OS X)) and, did I mention it, using it only for sockets.
		572
		573	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		574	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		575	C<NOTE_EOF>.
		576
		577	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
		578
		579	This is not implemented yet (and might never be, unless you send me an
		580	implementation). According to reports, C</dev/poll> only supports sockets
		581	and is not embeddable, which would limit the usefulness of this backend
		582	immensely.
		583
		584	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
		585
		586	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
		587	it's really slow, but it still scales very well (O(active_fds)).
		588
		589	While this backend scales well, it requires one system call per active
		590	file descriptor per loop iteration. For small and medium numbers of file
		591	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		592	might perform better.
		593
		594	On the positive side, this backend actually performed fully to
		595	specification in all tests and is fully embeddable, which is a rare feat
		596	among the OS-specific backends (I vastly prefer correctness over speed
		597	hacks).
		598
		599	On the negative side, the interface is I<bizarre> - so bizarre that
		600	even sun itself gets it wrong in their code examples: The event polling
		601	function sometimes returning events to the caller even though an error
		602	occurred, but with no indication whether it has done so or not (yes, it's
		603	even documented that way) - deadly for edge-triggered interfaces where
		604	you absolutely have to know whether an event occurred or not because you
		605	have to re-arm the watcher.
		606
		607	Fortunately libev seems to be able to work around these idiocies.
		608
		609	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		610	C<EVBACKEND_POLL>.
		611
		612	=item C<EVBACKEND_ALL>
		613
		614	Try all backends (even potentially broken ones that wouldn't be tried
		615	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
		616	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		617
		618	It is definitely not recommended to use this flag, use whatever
		619	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		620	at all.
		621
		622	=item C<EVBACKEND_MASK>
		623
		624	Not a backend at all, but a mask to select all backend bits from a
		625	C<flags> value, in case you want to mask out any backends from a flags
		626	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
158		627
159	=back	628	=back
160		629
161	=item struct ev_loop *ev_loop_new (unsigned int flags)	630	If one or more of the backend flags are or'ed into the flags value,
		631	then only these backends will be tried (in the reverse order as listed
		632	here). If none are specified, all backends in C<ev_recommended_backends
		633	()> will be tried.
162		634
163	Similar to C<ev_default_loop>, but always creates a new event loop that is	635	Example: Try to create a event loop that uses epoll and nothing else.
164	always distinct from the default loop. Unlike the default loop, it cannot
165	handle signal and child watchers, and attempts to do so will be greeted by
166	undefined behaviour (or a failed assertion if assertions are enabled).
167		636
168	=item ev_default_destroy ()	637	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
		638	if (!epoller)
		639	fatal ("no epoll found here, maybe it hides under your chair");
169		640
170	Destroys the default loop again (frees all memory and kernel state	641	Example: Use whatever libev has to offer, but make sure that kqueue is
171	etc.). This stops all registered event watchers (by not touching them in	642	used if available.
172	any way whatsoever, although you cnanot rely on this :).	643
		644	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
173		645
174	=item ev_loop_destroy (loop)	646	=item ev_loop_destroy (loop)
175		647
176	Like C<ev_default_destroy>, but destroys an event loop created by an	648	Destroys an event loop object (frees all memory and kernel state
177	earlier call to C<ev_loop_new>.	649	etc.). None of the active event watchers will be stopped in the normal
		650	sense, so e.g. C<ev_is_active> might still return true. It is your
		651	responsibility to either stop all watchers cleanly yourself I<before>
		652	calling this function, or cope with the fact afterwards (which is usually
		653	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		654	for example).
178		655
179	=item ev_default_fork ()	656	Note that certain global state, such as signal state (and installed signal
		657	handlers), will not be freed by this function, and related watchers (such
		658	as signal and child watchers) would need to be stopped manually.
180		659
181	This function reinitialises the kernel state for backends that have	660	This function is normally used on loop objects allocated by
182	one. Despite the name, you can call it anytime, but it makes most sense	661	C<ev_loop_new>, but it can also be used on the default loop returned by
183	after forking, in either the parent or child process (or both, but that	662	C<ev_default_loop>, in which case it is not thread-safe.
184	again makes little sense).
185		663
186	You I<must> call this function after forking if and only if you want to	664	Note that it is not advisable to call this function on the default loop
187	use the event library in both processes. If you just fork+exec, you don't	665	except in the rare occasion where you really need to free its resources.
188	have to call it.	666	If you need dynamically allocated loops it is better to use C<ev_loop_new>
189		667	and C<ev_loop_destroy>.
190	The function itself is quite fast and its usually not a problem to call
191	it just in case after a fork. To make this easy, the function will fit in
192	quite nicely into a call to C<pthread_atfork>:
193
194	pthread_atfork (0, 0, ev_default_fork);
195		668
196	=item ev_loop_fork (loop)	669	=item ev_loop_fork (loop)
197		670
198	Like C<ev_default_fork>, but acts on an event loop created by	671	This function sets a flag that causes subsequent C<ev_run> iterations to
199	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	672	reinitialise the kernel state for backends that have one. Despite the
200	after fork, and how you do this is entirely your own problem.	673	name, you can call it anytime, but it makes most sense after forking, in
		674	the child process. You I<must> call it (or use C<EVFLAG_FORKCHECK>) in the
		675	child before resuming or calling C<ev_run>.
201		676
		677	Again, you I<have> to call it on I<any> loop that you want to re-use after
		678	a fork, I<even if you do not plan to use the loop in the parent>. This is
		679	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		680	during fork.
		681
		682	On the other hand, you only need to call this function in the child
		683	process if and only if you want to use the event loop in the child. If
		684	you just fork+exec or create a new loop in the child, you don't have to
		685	call it at all (in fact, C<epoll> is so badly broken that it makes a
		686	difference, but libev will usually detect this case on its own and do a
		687	costly reset of the backend).
		688
		689	The function itself is quite fast and it's usually not a problem to call
		690	it just in case after a fork.
		691
		692	Example: Automate calling C<ev_loop_fork> on the default loop when
		693	using pthreads.
		694
		695	static void
		696	post_fork_child (void)
		697	{
		698	ev_loop_fork (EV_DEFAULT);
		699	}
		700
		701	...
		702	pthread_atfork (0, 0, post_fork_child);
		703
		704	=item int ev_is_default_loop (loop)
		705
		706	Returns true when the given loop is, in fact, the default loop, and false
		707	otherwise.
		708
		709	=item unsigned int ev_iteration (loop)
		710
		711	Returns the current iteration count for the event loop, which is identical
		712	to the number of times libev did poll for new events. It starts at C<0>
		713	and happily wraps around with enough iterations.
		714
		715	This value can sometimes be useful as a generation counter of sorts (it
		716	"ticks" the number of loop iterations), as it roughly corresponds with
		717	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		718	prepare and check phases.
		719
202	=item unsigned int ev_method (loop)	720	=item unsigned int ev_depth (loop)
203		721
		722	Returns the number of times C<ev_run> was entered minus the number of
		723	times C<ev_run> was exited normally, in other words, the recursion depth.
		724
		725	Outside C<ev_run>, this number is zero. In a callback, this number is
		726	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		727	in which case it is higher.
		728
		729	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
		730	throwing an exception etc.), doesn't count as "exit" - consider this
		731	as a hint to avoid such ungentleman-like behaviour unless it's really
		732	convenient, in which case it is fully supported.
		733
		734	=item unsigned int ev_backend (loop)
		735
204	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	736	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
205	use.	737	use.
206		738
207	=item ev_tstamp = ev_now (loop)	739	=item ev_tstamp ev_now (loop)
208		740
209	Returns the current "event loop time", which is the time the event loop	741	Returns the current "event loop time", which is the time the event loop
210	got events and started processing them. This timestamp does not change	742	received events and started processing them. This timestamp does not
211	as long as callbacks are being processed, and this is also the base time	743	change as long as callbacks are being processed, and this is also the base
212	used for relative timers. You can treat it as the timestamp of the event	744	time used for relative timers. You can treat it as the timestamp of the
213	occuring (or more correctly, the mainloop finding out about it).	745	event occurring (or more correctly, libev finding out about it).
214		746
		747	=item ev_now_update (loop)
		748
		749	Establishes the current time by querying the kernel, updating the time
		750	returned by C<ev_now ()> in the progress. This is a costly operation and
		751	is usually done automatically within C<ev_run ()>.
		752
		753	This function is rarely useful, but when some event callback runs for a
		754	very long time without entering the event loop, updating libev's idea of
		755	the current time is a good idea.
		756
		757	See also L<The special problem of time updates> in the C<ev_timer> section.
		758
		759	=item ev_suspend (loop)
		760
		761	=item ev_resume (loop)
		762
		763	These two functions suspend and resume an event loop, for use when the
		764	loop is not used for a while and timeouts should not be processed.
		765
		766	A typical use case would be an interactive program such as a game: When
		767	the user presses C<^Z> to suspend the game and resumes it an hour later it
		768	would be best to handle timeouts as if no time had actually passed while
		769	the program was suspended. This can be achieved by calling C<ev_suspend>
		770	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		771	C<ev_resume> directly afterwards to resume timer processing.
		772
		773	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		774	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		775	will be rescheduled (that is, they will lose any events that would have
		776	occurred while suspended).
		777
		778	After calling C<ev_suspend> you B<must not> call I<any> function on the
		779	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		780	without a previous call to C<ev_suspend>.
		781
		782	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		783	event loop time (see C<ev_now_update>).
		784
215	=item ev_loop (loop, int flags)	785	=item ev_run (loop, int flags)
216		786
217	Finally, this is it, the event handler. This function usually is called	787	Finally, this is it, the event handler. This function usually is called
218	after you initialised all your watchers and you want to start handling	788	after you have initialised all your watchers and you want to start
219	events.	789	handling events. It will ask the operating system for any new events, call
		790	the watcher callbacks, an then repeat the whole process indefinitely: This
		791	is why event loops are called I<loops>.
220		792
221	If the flags argument is specified as 0, it will not return until either	793	If the flags argument is specified as C<0>, it will keep handling events
222	no event watchers are active anymore or C<ev_unloop> was called.	794	until either no event watchers are active anymore or C<ev_break> was
		795	called.
223		796
		797	Please note that an explicit C<ev_break> is usually better than
		798	relying on all watchers to be stopped when deciding when a program has
		799	finished (especially in interactive programs), but having a program
		800	that automatically loops as long as it has to and no longer by virtue
		801	of relying on its watchers stopping correctly, that is truly a thing of
		802	beauty.
		803
		804	This function is also I<mostly> exception-safe - you can break out of
		805	a C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		806	exception and so on. This does not decrement the C<ev_depth> value, nor
		807	will it clear any outstanding C<EVBREAK_ONE> breaks.
		808
224	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	809	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
225	those events and any outstanding ones, but will not block your process in	810	those events and any already outstanding ones, but will not wait and
226	case there are no events.	811	block your process in case there are no events and will return after one
		812	iteration of the loop. This is sometimes useful to poll and handle new
		813	events while doing lengthy calculations, to keep the program responsive.
227		814
228	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	815	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
229	neccessary) and will handle those and any outstanding ones. It will block	816	necessary) and will handle those and any already outstanding ones. It
230	your process until at least one new event arrives.	817	will block your process until at least one new event arrives (which could
		818	be an event internal to libev itself, so there is no guarantee that a
		819	user-registered callback will be called), and will return after one
		820	iteration of the loop.
231		821
232	This flags value could be used to implement alternative looping	822	This is useful if you are waiting for some external event in conjunction
233	constructs, but the C<prepare> and C<check> watchers provide a better and	823	with something not expressible using other libev watchers (i.e. "roll your
234	more generic mechanism.	824	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		825	usually a better approach for this kind of thing.
235		826
		827	Here are the gory details of what C<ev_run> does:
		828
		829	- Increment loop depth.
		830	- Reset the ev_break status.
		831	- Before the first iteration, call any pending watchers.
		832	LOOP:
		833	- If EVFLAG_FORKCHECK was used, check for a fork.
		834	- If a fork was detected (by any means), queue and call all fork watchers.
		835	- Queue and call all prepare watchers.
		836	- If ev_break was called, goto FINISH.
		837	- If we have been forked, detach and recreate the kernel state
		838	as to not disturb the other process.
		839	- Update the kernel state with all outstanding changes.
		840	- Update the "event loop time" (ev_now ()).
		841	- Calculate for how long to sleep or block, if at all
		842	(active idle watchers, EVRUN_NOWAIT or not having
		843	any active watchers at all will result in not sleeping).
		844	- Sleep if the I/O and timer collect interval say so.
		845	- Increment loop iteration counter.
		846	- Block the process, waiting for any events.
		847	- Queue all outstanding I/O (fd) events.
		848	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
		849	- Queue all expired timers.
		850	- Queue all expired periodics.
		851	- Queue all idle watchers with priority higher than that of pending events.
		852	- Queue all check watchers.
		853	- Call all queued watchers in reverse order (i.e. check watchers first).
		854	Signals and child watchers are implemented as I/O watchers, and will
		855	be handled here by queueing them when their watcher gets executed.
		856	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
		857	were used, or there are no active watchers, goto FINISH, otherwise
		858	continue with step LOOP.
		859	FINISH:
		860	- Reset the ev_break status iff it was EVBREAK_ONE.
		861	- Decrement the loop depth.
		862	- Return.
		863
		864	Example: Queue some jobs and then loop until no events are outstanding
		865	anymore.
		866
		867	... queue jobs here, make sure they register event watchers as long
		868	... as they still have work to do (even an idle watcher will do..)
		869	ev_run (my_loop, 0);
		870	... jobs done or somebody called unloop. yeah!
		871
236	=item ev_unloop (loop, how)	872	=item ev_break (loop, how)
237		873
238	Can be used to make a call to C<ev_loop> return early. The C<how> argument	874	Can be used to make a call to C<ev_run> return early (but only after it
239	must be either C<EVUNLOOP_ONCE>, which will make the innermost C<ev_loop>	875	has processed all outstanding events). The C<how> argument must be either
240	call return, or C<EVUNLOOP_ALL>, which will make all nested C<ev_loop>	876	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
241	calls return.	877	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
		878
		879	This "break state" will be cleared on the next call to C<ev_run>.
		880
		881	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		882	which case it will have no effect.
242		883
243	=item ev_ref (loop)	884	=item ev_ref (loop)
244		885
245	=item ev_unref (loop)	886	=item ev_unref (loop)
246		887
247	Ref/unref can be used to add or remove a refcount on the event loop: Every	888	Ref/unref can be used to add or remove a reference count on the event
248	watcher keeps one reference. If you have a long-runing watcher you never	889	loop: Every watcher keeps one reference, and as long as the reference
249	unregister that should not keep ev_loop from running, ev_unref() after	890	count is nonzero, C<ev_run> will not return on its own.
250	starting, and ev_ref() before stopping it. Libev itself uses this for	891
251	example for its internal signal pipe: It is not visible to you as a user	892	This is useful when you have a watcher that you never intend to
252	and should not keep C<ev_loop> from exiting if the work is done. It is	893	unregister, but that nevertheless should not keep C<ev_run> from
		894	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		895	before stopping it.
		896
		897	As an example, libev itself uses this for its internal signal pipe: It
		898	is not visible to the libev user and should not keep C<ev_run> from
		899	exiting if no event watchers registered by it are active. It is also an
253	also an excellent way to do this for generic recurring timers or from	900	excellent way to do this for generic recurring timers or from within
254	within third-party libraries. Just remember to unref after start and ref	901	third-party libraries. Just remember to I<unref after start> and I<ref
255	before stop.	902	before stop> (but only if the watcher wasn't active before, or was active
		903	before, respectively. Note also that libev might stop watchers itself
		904	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		905	in the callback).
		906
		907	Example: Create a signal watcher, but keep it from keeping C<ev_run>
		908	running when nothing else is active.
		909
		910	ev_signal exitsig;
		911	ev_signal_init (&exitsig, sig_cb, SIGINT);
		912	ev_signal_start (loop, &exitsig);
		913	ev_unref (loop);
		914
		915	Example: For some weird reason, unregister the above signal handler again.
		916
		917	ev_ref (loop);
		918	ev_signal_stop (loop, &exitsig);
		919
		920	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		921
		922	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		923
		924	These advanced functions influence the time that libev will spend waiting
		925	for events. Both time intervals are by default C<0>, meaning that libev
		926	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		927	latency.
		928
		929	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		930	allows libev to delay invocation of I/O and timer/periodic callbacks
		931	to increase efficiency of loop iterations (or to increase power-saving
		932	opportunities).
		933
		934	The idea is that sometimes your program runs just fast enough to handle
		935	one (or very few) event(s) per loop iteration. While this makes the
		936	program responsive, it also wastes a lot of CPU time to poll for new
		937	events, especially with backends like C<select ()> which have a high
		938	overhead for the actual polling but can deliver many events at once.
		939
		940	By setting a higher I<io collect interval> you allow libev to spend more
		941	time collecting I/O events, so you can handle more events per iteration,
		942	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		943	C<ev_timer>) will be not affected. Setting this to a non-null value will
		944	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		945	sleep time ensures that libev will not poll for I/O events more often then
		946	once per this interval, on average.
		947
		948	Likewise, by setting a higher I<timeout collect interval> you allow libev
		949	to spend more time collecting timeouts, at the expense of increased
		950	latency/jitter/inexactness (the watcher callback will be called
		951	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		952	value will not introduce any overhead in libev.
		953
		954	Many (busy) programs can usually benefit by setting the I/O collect
		955	interval to a value near C<0.1> or so, which is often enough for
		956	interactive servers (of course not for games), likewise for timeouts. It
		957	usually doesn't make much sense to set it to a lower value than C<0.01>,
		958	as this approaches the timing granularity of most systems. Note that if
		959	you do transactions with the outside world and you can't increase the
		960	parallelity, then this setting will limit your transaction rate (if you
		961	need to poll once per transaction and the I/O collect interval is 0.01,
		962	then you can't do more than 100 transactions per second).
		963
		964	Setting the I<timeout collect interval> can improve the opportunity for
		965	saving power, as the program will "bundle" timer callback invocations that
		966	are "near" in time together, by delaying some, thus reducing the number of
		967	times the process sleeps and wakes up again. Another useful technique to
		968	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		969	they fire on, say, one-second boundaries only.
		970
		971	Example: we only need 0.1s timeout granularity, and we wish not to poll
		972	more often than 100 times per second:
		973
		974	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		975	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		976
		977	=item ev_invoke_pending (loop)
		978
		979	This call will simply invoke all pending watchers while resetting their
		980	pending state. Normally, C<ev_run> does this automatically when required,
		981	but when overriding the invoke callback this call comes handy. This
		982	function can be invoked from a watcher - this can be useful for example
		983	when you want to do some lengthy calculation and want to pass further
		984	event handling to another thread (you still have to make sure only one
		985	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		986
		987	=item int ev_pending_count (loop)
		988
		989	Returns the number of pending watchers - zero indicates that no watchers
		990	are pending.
		991
		992	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		993
		994	This overrides the invoke pending functionality of the loop: Instead of
		995	invoking all pending watchers when there are any, C<ev_run> will call
		996	this callback instead. This is useful, for example, when you want to
		997	invoke the actual watchers inside another context (another thread etc.).
		998
		999	If you want to reset the callback, use C<ev_invoke_pending> as new
		1000	callback.
		1001
		1002	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		1003
		1004	Sometimes you want to share the same loop between multiple threads. This
		1005	can be done relatively simply by putting mutex_lock/unlock calls around
		1006	each call to a libev function.
		1007
		1008	However, C<ev_run> can run an indefinite time, so it is not feasible
		1009	to wait for it to return. One way around this is to wake up the event
		1010	loop via C<ev_break> and C<av_async_send>, another way is to set these
		1011	I<release> and I<acquire> callbacks on the loop.
		1012
		1013	When set, then C<release> will be called just before the thread is
		1014	suspended waiting for new events, and C<acquire> is called just
		1015	afterwards.
		1016
		1017	Ideally, C<release> will just call your mutex_unlock function, and
		1018	C<acquire> will just call the mutex_lock function again.
		1019
		1020	While event loop modifications are allowed between invocations of
		1021	C<release> and C<acquire> (that's their only purpose after all), no
		1022	modifications done will affect the event loop, i.e. adding watchers will
		1023	have no effect on the set of file descriptors being watched, or the time
		1024	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1025	to take note of any changes you made.
		1026
		1027	In theory, threads executing C<ev_run> will be async-cancel safe between
		1028	invocations of C<release> and C<acquire>.
		1029
		1030	See also the locking example in the C<THREADS> section later in this
		1031	document.
		1032
		1033	=item ev_set_userdata (loop, void *data)
		1034
		1035	=item void *ev_userdata (loop)
		1036
		1037	Set and retrieve a single C<void *> associated with a loop. When
		1038	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1039	C<0>.
		1040
		1041	These two functions can be used to associate arbitrary data with a loop,
		1042	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1043	C<acquire> callbacks described above, but of course can be (ab-)used for
		1044	any other purpose as well.
		1045
		1046	=item ev_verify (loop)
		1047
		1048	This function only does something when C<EV_VERIFY> support has been
		1049	compiled in, which is the default for non-minimal builds. It tries to go
		1050	through all internal structures and checks them for validity. If anything
		1051	is found to be inconsistent, it will print an error message to standard
		1052	error and call C<abort ()>.
		1053
		1054	This can be used to catch bugs inside libev itself: under normal
		1055	circumstances, this function will never abort as of course libev keeps its
		1056	data structures consistent.
256		1057
257	=back	1058	=back
258		1059
		1060
259	=head1 ANATOMY OF A WATCHER	1061	=head1 ANATOMY OF A WATCHER
260		1062
		1063	In the following description, uppercase C<TYPE> in names stands for the
		1064	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1065	watchers and C<ev_io_start> for I/O watchers.
		1066
261	A watcher is a structure that you create and register to record your	1067	A watcher is an opaque structure that you allocate and register to record
262	interest in some event. For instance, if you want to wait for STDIN to	1068	your interest in some event. To make a concrete example, imagine you want
263	become readable, you would create an ev_io watcher for that:	1069	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1070	for that:
264		1071
265	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1072	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
266	{	1073	{
267	ev_io_stop (w);	1074	ev_io_stop (w);
268	ev_unloop (loop, EVUNLOOP_ALL);	1075	ev_break (loop, EVBREAK_ALL);
269	}	1076	}
270		1077
271	struct ev_loop *loop = ev_default_loop (0);	1078	struct ev_loop *loop = ev_default_loop (0);
		1079
272	struct ev_io stdin_watcher;	1080	ev_io stdin_watcher;
		1081
273	ev_init (&stdin_watcher, my_cb);	1082	ev_init (&stdin_watcher, my_cb);
274	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1083	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
275	ev_io_start (loop, &stdin_watcher);	1084	ev_io_start (loop, &stdin_watcher);
		1085
276	ev_loop (loop, 0);	1086	ev_run (loop, 0);
277		1087
278	As you can see, you are responsible for allocating the memory for your	1088	As you can see, you are responsible for allocating the memory for your
279	watcher structures (and it is usually a bad idea to do this on the stack,	1089	watcher structures (and it is I<usually> a bad idea to do this on the
280	although this can sometimes be quite valid).	1090	stack).
281		1091
		1092	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1093	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1094
282	Each watcher structure must be initialised by a call to C<ev_init	1095	Each watcher structure must be initialised by a call to C<ev_init (watcher
283	(watcher *, callback)>, which expects a callback to be provided. This	1096	*, callback)>, which expects a callback to be provided. This callback is
284	callback gets invoked each time the event occurs (or, in the case of io	1097	invoked each time the event occurs (or, in the case of I/O watchers, each
285	watchers, each time the event loop detects that the file descriptor given	1098	time the event loop detects that the file descriptor given is readable
286	is readable and/or writable).	1099	and/or writable).
287		1100
288	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1101	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
289	with arguments specific to this watcher type. There is also a macro	1102	macro to configure it, with arguments specific to the watcher type. There
290	to combine initialisation and setting in one call: C<< ev_<type>_init	1103	is also a macro to combine initialisation and setting in one call: C<<
291	(watcher *, callback, ...) >>.	1104	ev_TYPE_init (watcher *, callback, ...) >>.
292		1105
293	To make the watcher actually watch out for events, you have to start it	1106	To make the watcher actually watch out for events, you have to start it
294	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1107	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
295	*) >>), and you can stop watching for events at any time by calling the	1108	*) >>), and you can stop watching for events at any time by calling the
296	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1109	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
297		1110
298	As long as your watcher is active (has been started but not stopped) you	1111	As long as your watcher is active (has been started but not stopped) you
299	must not touch the values stored in it. Most specifically you must never	1112	must not touch the values stored in it. Most specifically you must never
300	reinitialise it or call its set method.	1113	reinitialise it or call its C<ev_TYPE_set> macro.
301
302	You cna check whether an event is active by calling the C<ev_is_active
303	(watcher *)> macro. To see whether an event is outstanding (but the
304	callback for it has not been called yet) you cna use the C<ev_is_pending
305	(watcher *)> macro.
306		1114
307	Each and every callback receives the event loop pointer as first, the	1115	Each and every callback receives the event loop pointer as first, the
308	registered watcher structure as second, and a bitset of received events as	1116	registered watcher structure as second, and a bitset of received events as
309	third argument.	1117	third argument.
310		1118
311	The rceeived events usually include a single bit per event type received	1119	The received events usually include a single bit per event type received
312	(you can receive multiple events at the same time). The possible bit masks	1120	(you can receive multiple events at the same time). The possible bit masks
313	are:	1121	are:
314		1122
315	=over 4	1123	=over 4
316		1124
317	=item EV_READ	1125	=item C<EV_READ>
318		1126
319	=item EV_WRITE	1127	=item C<EV_WRITE>
320		1128
321	The file descriptor in the ev_io watcher has become readable and/or	1129	The file descriptor in the C<ev_io> watcher has become readable and/or
322	writable.	1130	writable.
323		1131
324	=item EV_TIMEOUT	1132	=item C<EV_TIMER>
325		1133
326	The ev_timer watcher has timed out.	1134	The C<ev_timer> watcher has timed out.
327		1135
328	=item EV_PERIODIC	1136	=item C<EV_PERIODIC>
329		1137
330	The ev_periodic watcher has timed out.	1138	The C<ev_periodic> watcher has timed out.
331		1139
332	=item EV_SIGNAL	1140	=item C<EV_SIGNAL>
333		1141
334	The signal specified in the ev_signal watcher has been received by a thread.	1142	The signal specified in the C<ev_signal> watcher has been received by a thread.
335		1143
336	=item EV_CHILD	1144	=item C<EV_CHILD>
337		1145
338	The pid specified in the ev_child watcher has received a status change.	1146	The pid specified in the C<ev_child> watcher has received a status change.
339		1147
		1148	=item C<EV_STAT>
		1149
		1150	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		1151
340	=item EV_IDLE	1152	=item C<EV_IDLE>
341		1153
342	The ev_idle watcher has determined that you have nothing better to do.	1154	The C<ev_idle> watcher has determined that you have nothing better to do.
343		1155
344	=item EV_PREPARE	1156	=item C<EV_PREPARE>
345		1157
346	=item EV_CHECK	1158	=item C<EV_CHECK>
347		1159
348	All ev_prepare watchers are invoked just I<before> C<ev_loop> starts	1160	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts
349	to gather new events, and all ev_check watchers are invoked just after	1161	to gather new events, and all C<ev_check> watchers are invoked just after
350	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1162	C<ev_run> has gathered them, but before it invokes any callbacks for any
351	received events. Callbacks of both watcher types can start and stop as	1163	received events. Callbacks of both watcher types can start and stop as
352	many watchers as they want, and all of them will be taken into account	1164	many watchers as they want, and all of them will be taken into account
353	(for example, a ev_prepare watcher might start an idle watcher to keep	1165	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
354	C<ev_loop> from blocking).	1166	C<ev_run> from blocking).
355		1167
		1168	=item C<EV_EMBED>
		1169
		1170	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1171
		1172	=item C<EV_FORK>
		1173
		1174	The event loop has been resumed in the child process after fork (see
		1175	C<ev_fork>).
		1176
		1177	=item C<EV_CLEANUP>
		1178
		1179	The event loop is about to be destroyed (see C<ev_cleanup>).
		1180
		1181	=item C<EV_ASYNC>
		1182
		1183	The given async watcher has been asynchronously notified (see C<ev_async>).
		1184
		1185	=item C<EV_CUSTOM>
		1186
		1187	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1188	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1189
356	=item EV_ERROR	1190	=item C<EV_ERROR>
357		1191
358	An unspecified error has occured, the watcher has been stopped. This might	1192	An unspecified error has occurred, the watcher has been stopped. This might
359	happen because the watcher could not be properly started because libev	1193	happen because the watcher could not be properly started because libev
360	ran out of memory, a file descriptor was found to be closed or any other	1194	ran out of memory, a file descriptor was found to be closed or any other
		1195	problem. Libev considers these application bugs.
		1196
361	problem. You best act on it by reporting the problem and somehow coping	1197	You best act on it by reporting the problem and somehow coping with the
362	with the watcher being stopped.	1198	watcher being stopped. Note that well-written programs should not receive
		1199	an error ever, so when your watcher receives it, this usually indicates a
		1200	bug in your program.
363		1201
364	Libev will usually signal a few "dummy" events together with an error,	1202	Libev will usually signal a few "dummy" events together with an error, for
365	for example it might indicate that a fd is readable or writable, and if	1203	example it might indicate that a fd is readable or writable, and if your
366	your callbacks is well-written it can just attempt the operation and cope	1204	callbacks is well-written it can just attempt the operation and cope with
367	with the error from read() or write(). This will not work in multithreaded	1205	the error from read() or write(). This will not work in multi-threaded
368	programs, though, so beware.	1206	programs, though, as the fd could already be closed and reused for another
		1207	thing, so beware.
369		1208
370	=back	1209	=back
371		1210
		1211	=head2 GENERIC WATCHER FUNCTIONS
		1212
		1213	=over 4
		1214
		1215	=item C<ev_init> (ev_TYPE *watcher, callback)
		1216
		1217	This macro initialises the generic portion of a watcher. The contents
		1218	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1219	the generic parts of the watcher are initialised, you I<need> to call
		1220	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1221	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1222	which rolls both calls into one.
		1223
		1224	You can reinitialise a watcher at any time as long as it has been stopped
		1225	(or never started) and there are no pending events outstanding.
		1226
		1227	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1228	int revents)>.
		1229
		1230	Example: Initialise an C<ev_io> watcher in two steps.
		1231
		1232	ev_io w;
		1233	ev_init (&w, my_cb);
		1234	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1235
		1236	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1237
		1238	This macro initialises the type-specific parts of a watcher. You need to
		1239	call C<ev_init> at least once before you call this macro, but you can
		1240	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1241	macro on a watcher that is active (it can be pending, however, which is a
		1242	difference to the C<ev_init> macro).
		1243
		1244	Although some watcher types do not have type-specific arguments
		1245	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1246
		1247	See C<ev_init>, above, for an example.
		1248
		1249	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1250
		1251	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1252	calls into a single call. This is the most convenient method to initialise
		1253	a watcher. The same limitations apply, of course.
		1254
		1255	Example: Initialise and set an C<ev_io> watcher in one step.
		1256
		1257	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1258
		1259	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1260
		1261	Starts (activates) the given watcher. Only active watchers will receive
		1262	events. If the watcher is already active nothing will happen.
		1263
		1264	Example: Start the C<ev_io> watcher that is being abused as example in this
		1265	whole section.
		1266
		1267	ev_io_start (EV_DEFAULT_UC, &w);
		1268
		1269	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1270
		1271	Stops the given watcher if active, and clears the pending status (whether
		1272	the watcher was active or not).
		1273
		1274	It is possible that stopped watchers are pending - for example,
		1275	non-repeating timers are being stopped when they become pending - but
		1276	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1277	pending. If you want to free or reuse the memory used by the watcher it is
		1278	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1279
		1280	=item bool ev_is_active (ev_TYPE *watcher)
		1281
		1282	Returns a true value iff the watcher is active (i.e. it has been started
		1283	and not yet been stopped). As long as a watcher is active you must not modify
		1284	it.
		1285
		1286	=item bool ev_is_pending (ev_TYPE *watcher)
		1287
		1288	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1289	events but its callback has not yet been invoked). As long as a watcher
		1290	is pending (but not active) you must not call an init function on it (but
		1291	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1292	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1293	it).
		1294
		1295	=item callback ev_cb (ev_TYPE *watcher)
		1296
		1297	Returns the callback currently set on the watcher.
		1298
		1299	=item ev_cb_set (ev_TYPE *watcher, callback)
		1300
		1301	Change the callback. You can change the callback at virtually any time
		1302	(modulo threads).
		1303
		1304	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1305
		1306	=item int ev_priority (ev_TYPE *watcher)
		1307
		1308	Set and query the priority of the watcher. The priority is a small
		1309	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1310	(default: C<-2>). Pending watchers with higher priority will be invoked
		1311	before watchers with lower priority, but priority will not keep watchers
		1312	from being executed (except for C<ev_idle> watchers).
		1313
		1314	If you need to suppress invocation when higher priority events are pending
		1315	you need to look at C<ev_idle> watchers, which provide this functionality.
		1316
		1317	You I<must not> change the priority of a watcher as long as it is active or
		1318	pending.
		1319
		1320	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1321	fine, as long as you do not mind that the priority value you query might
		1322	or might not have been clamped to the valid range.
		1323
		1324	The default priority used by watchers when no priority has been set is
		1325	always C<0>, which is supposed to not be too high and not be too low :).
		1326
		1327	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1328	priorities.
		1329
		1330	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1331
		1332	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1333	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1334	can deal with that fact, as both are simply passed through to the
		1335	callback.
		1336
		1337	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1338
		1339	If the watcher is pending, this function clears its pending status and
		1340	returns its C<revents> bitset (as if its callback was invoked). If the
		1341	watcher isn't pending it does nothing and returns C<0>.
		1342
		1343	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1344	callback to be invoked, which can be accomplished with this function.
		1345
		1346	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1347
		1348	Feeds the given event set into the event loop, as if the specified event
		1349	had happened for the specified watcher (which must be a pointer to an
		1350	initialised but not necessarily started event watcher). Obviously you must
		1351	not free the watcher as long as it has pending events.
		1352
		1353	Stopping the watcher, letting libev invoke it, or calling
		1354	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1355	not started in the first place.
		1356
		1357	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1358	functions that do not need a watcher.
		1359
		1360	=back
		1361
372	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1362	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
373		1363
374	Each watcher has, by default, a member C<void *data> that you can change	1364	Each watcher has, by default, a member C<void *data> that you can change
375	and read at any time, libev will completely ignore it. This cna be used	1365	and read at any time: libev will completely ignore it. This can be used
376	to associate arbitrary data with your watcher. If you need more data and	1366	to associate arbitrary data with your watcher. If you need more data and
377	don't want to allocate memory and store a pointer to it in that data	1367	don't want to allocate memory and store a pointer to it in that data
378	member, you can also "subclass" the watcher type and provide your own	1368	member, you can also "subclass" the watcher type and provide your own
379	data:	1369	data:
380		1370
381	struct my_io	1371	struct my_io
382	{	1372	{
383	struct ev_io io;	1373	ev_io io;
384	int otherfd;	1374	int otherfd;
385	void *somedata;	1375	void *somedata;
386	struct whatever *mostinteresting;	1376	struct whatever *mostinteresting;
387	}	1377	};
		1378
		1379	...
		1380	struct my_io w;
		1381	ev_io_init (&w.io, my_cb, fd, EV_READ);
388		1382
389	And since your callback will be called with a pointer to the watcher, you	1383	And since your callback will be called with a pointer to the watcher, you
390	can cast it back to your own type:	1384	can cast it back to your own type:
391		1385
392	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1386	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
393	{	1387	{
394	struct my_io *w = (struct my_io *)w_;	1388	struct my_io *w = (struct my_io *)w_;
395	...	1389	...
396	}	1390	}
397		1391
398	More interesting and less C-conformant ways of catsing your callback type	1392	More interesting and less C-conformant ways of casting your callback type
399	have been omitted....	1393	instead have been omitted.
		1394
		1395	Another common scenario is to use some data structure with multiple
		1396	embedded watchers:
		1397
		1398	struct my_biggy
		1399	{
		1400	int some_data;
		1401	ev_timer t1;
		1402	ev_timer t2;
		1403	}
		1404
		1405	In this case getting the pointer to C<my_biggy> is a bit more
		1406	complicated: Either you store the address of your C<my_biggy> struct
		1407	in the C<data> member of the watcher (for woozies), or you need to use
		1408	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1409	programmers):
		1410
		1411	#include <stddef.h>
		1412
		1413	static void
		1414	t1_cb (EV_P_ ev_timer *w, int revents)
		1415	{
		1416	struct my_biggy big = (struct my_biggy *)
		1417	(((char *)w) - offsetof (struct my_biggy, t1));
		1418	}
		1419
		1420	static void
		1421	t2_cb (EV_P_ ev_timer *w, int revents)
		1422	{
		1423	struct my_biggy big = (struct my_biggy *)
		1424	(((char *)w) - offsetof (struct my_biggy, t2));
		1425	}
		1426
		1427	=head2 WATCHER STATES
		1428
		1429	There are various watcher states mentioned throughout this manual -
		1430	active, pending and so on. In this section these states and the rules to
		1431	transition between them will be described in more detail - and while these
		1432	rules might look complicated, they usually do "the right thing".
		1433
		1434	=over 4
		1435
		1436	=item initialiased
		1437
		1438	Before a watcher can be registered with the event looop it has to be
		1439	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1440	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1441
		1442	In this state it is simply some block of memory that is suitable for use
		1443	in an event loop. It can be moved around, freed, reused etc. at will.
		1444
		1445	=item started/running/active
		1446
		1447	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1448	property of the event loop, and is actively waiting for events. While in
		1449	this state it cannot be accessed (except in a few documented ways), moved,
		1450	freed or anything else - the only legal thing is to keep a pointer to it,
		1451	and call libev functions on it that are documented to work on active watchers.
		1452
		1453	=item pending
		1454
		1455	If a watcher is active and libev determines that an event it is interested
		1456	in has occurred (such as a timer expiring), it will become pending. It will
		1457	stay in this pending state until either it is stopped or its callback is
		1458	about to be invoked, so it is not normally pending inside the watcher
		1459	callback.
		1460
		1461	The watcher might or might not be active while it is pending (for example,
		1462	an expired non-repeating timer can be pending but no longer active). If it
		1463	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1464	but it is still property of the event loop at this time, so cannot be
		1465	moved, freed or reused. And if it is active the rules described in the
		1466	previous item still apply.
		1467
		1468	It is also possible to feed an event on a watcher that is not active (e.g.
		1469	via C<ev_feed_event>), in which case it becomes pending without being
		1470	active.
		1471
		1472	=item stopped
		1473
		1474	A watcher can be stopped implicitly by libev (in which case it might still
		1475	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1476	latter will clear any pending state the watcher might be in, regardless
		1477	of whether it was active or not, so stopping a watcher explicitly before
		1478	freeing it is often a good idea.
		1479
		1480	While stopped (and not pending) the watcher is essentially in the
		1481	initialised state, that is it can be reused, moved, modified in any way
		1482	you wish.
		1483
		1484	=back
		1485
		1486	=head2 WATCHER PRIORITY MODELS
		1487
		1488	Many event loops support I<watcher priorities>, which are usually small
		1489	integers that influence the ordering of event callback invocation
		1490	between watchers in some way, all else being equal.
		1491
		1492	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1493	description for the more technical details such as the actual priority
		1494	range.
		1495
		1496	There are two common ways how these these priorities are being interpreted
		1497	by event loops:
		1498
		1499	In the more common lock-out model, higher priorities "lock out" invocation
		1500	of lower priority watchers, which means as long as higher priority
		1501	watchers receive events, lower priority watchers are not being invoked.
		1502
		1503	The less common only-for-ordering model uses priorities solely to order
		1504	callback invocation within a single event loop iteration: Higher priority
		1505	watchers are invoked before lower priority ones, but they all get invoked
		1506	before polling for new events.
		1507
		1508	Libev uses the second (only-for-ordering) model for all its watchers
		1509	except for idle watchers (which use the lock-out model).
		1510
		1511	The rationale behind this is that implementing the lock-out model for
		1512	watchers is not well supported by most kernel interfaces, and most event
		1513	libraries will just poll for the same events again and again as long as
		1514	their callbacks have not been executed, which is very inefficient in the
		1515	common case of one high-priority watcher locking out a mass of lower
		1516	priority ones.
		1517
		1518	Static (ordering) priorities are most useful when you have two or more
		1519	watchers handling the same resource: a typical usage example is having an
		1520	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1521	timeouts. Under load, data might be received while the program handles
		1522	other jobs, but since timers normally get invoked first, the timeout
		1523	handler will be executed before checking for data. In that case, giving
		1524	the timer a lower priority than the I/O watcher ensures that I/O will be
		1525	handled first even under adverse conditions (which is usually, but not
		1526	always, what you want).
		1527
		1528	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1529	will only be executed when no same or higher priority watchers have
		1530	received events, they can be used to implement the "lock-out" model when
		1531	required.
		1532
		1533	For example, to emulate how many other event libraries handle priorities,
		1534	you can associate an C<ev_idle> watcher to each such watcher, and in
		1535	the normal watcher callback, you just start the idle watcher. The real
		1536	processing is done in the idle watcher callback. This causes libev to
		1537	continuously poll and process kernel event data for the watcher, but when
		1538	the lock-out case is known to be rare (which in turn is rare :), this is
		1539	workable.
		1540
		1541	Usually, however, the lock-out model implemented that way will perform
		1542	miserably under the type of load it was designed to handle. In that case,
		1543	it might be preferable to stop the real watcher before starting the
		1544	idle watcher, so the kernel will not have to process the event in case
		1545	the actual processing will be delayed for considerable time.
		1546
		1547	Here is an example of an I/O watcher that should run at a strictly lower
		1548	priority than the default, and which should only process data when no
		1549	other events are pending:
		1550
		1551	ev_idle idle; // actual processing watcher
		1552	ev_io io; // actual event watcher
		1553
		1554	static void
		1555	io_cb (EV_P_ ev_io *w, int revents)
		1556	{
		1557	// stop the I/O watcher, we received the event, but
		1558	// are not yet ready to handle it.
		1559	ev_io_stop (EV_A_ w);
		1560
		1561	// start the idle watcher to handle the actual event.
		1562	// it will not be executed as long as other watchers
		1563	// with the default priority are receiving events.
		1564	ev_idle_start (EV_A_ &idle);
		1565	}
		1566
		1567	static void
		1568	idle_cb (EV_P_ ev_idle *w, int revents)
		1569	{
		1570	// actual processing
		1571	read (STDIN_FILENO, ...);
		1572
		1573	// have to start the I/O watcher again, as
		1574	// we have handled the event
		1575	ev_io_start (EV_P_ &io);
		1576	}
		1577
		1578	// initialisation
		1579	ev_idle_init (&idle, idle_cb);
		1580	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1581	ev_io_start (EV_DEFAULT_ &io);
		1582
		1583	In the "real" world, it might also be beneficial to start a timer, so that
		1584	low-priority connections can not be locked out forever under load. This
		1585	enables your program to keep a lower latency for important connections
		1586	during short periods of high load, while not completely locking out less
		1587	important ones.
400		1588
401		1589
402	=head1 WATCHER TYPES	1590	=head1 WATCHER TYPES
403		1591
404	This section describes each watcher in detail, but will not repeat	1592	This section describes each watcher in detail, but will not repeat
405	information given in the last section.	1593	information given in the last section. Any initialisation/set macros,
		1594	functions and members specific to the watcher type are explained.
406		1595
		1596	Members are additionally marked with either I<[read-only]>, meaning that,
		1597	while the watcher is active, you can look at the member and expect some
		1598	sensible content, but you must not modify it (you can modify it while the
		1599	watcher is stopped to your hearts content), or I<[read-write]>, which
		1600	means you can expect it to have some sensible content while the watcher
		1601	is active, but you can also modify it. Modifying it may not do something
		1602	sensible or take immediate effect (or do anything at all), but libev will
		1603	not crash or malfunction in any way.
		1604
		1605
407	=head2 struct ev_io - is my file descriptor readable or writable	1606	=head2 C<ev_io> - is this file descriptor readable or writable?
408		1607
409	I/O watchers check whether a file descriptor is readable or writable	1608	I/O watchers check whether a file descriptor is readable or writable
410	in each iteration of the event loop (This behaviour is called	1609	in each iteration of the event loop, or, more precisely, when reading
411	level-triggering because you keep receiving events as long as the	1610	would not block the process and writing would at least be able to write
412	condition persists. Remember you cna stop the watcher if you don't want to	1611	some data. This behaviour is called level-triggering because you keep
413	act on the event and neither want to receive future events).	1612	receiving events as long as the condition persists. Remember you can stop
		1613	the watcher if you don't want to act on the event and neither want to
		1614	receive future events.
414		1615
415	In general you can register as many read and/or write event watchers oer	1616	In general you can register as many read and/or write event watchers per
416	fd as you want (as long as you don't confuse yourself). Setting all file	1617	fd as you want (as long as you don't confuse yourself). Setting all file
417	descriptors to non-blocking mode is also usually a good idea (but not	1618	descriptors to non-blocking mode is also usually a good idea (but not
418	required if you know what you are doing).	1619	required if you know what you are doing).
419		1620
420	You have to be careful with dup'ed file descriptors, though. Some backends	1621	Another thing you have to watch out for is that it is quite easy to
421	(the linux epoll backend is a notable example) cannot handle dup'ed file	1622	receive "spurious" readiness notifications, that is, your callback might
422	descriptors correctly if you register interest in two or more fds pointing	1623	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
423	to the same file/socket etc. description.	1624	because there is no data. It is very easy to get into this situation even
		1625	with a relatively standard program structure. Thus it is best to always
		1626	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
		1627	preferable to a program hanging until some data arrives.
424		1628
425	If you must do this, then force the use of a known-to-be-good backend	1629	If you cannot run the fd in non-blocking mode (for example you should
426	(at the time of this writing, this includes only EVMETHOD_SELECT and	1630	not play around with an Xlib connection), then you have to separately
427	EVMETHOD_POLL).	1631	re-test whether a file descriptor is really ready with a known-to-be good
		1632	interface such as poll (fortunately in the case of Xlib, it already does
		1633	this on its own, so its quite safe to use). Some people additionally
		1634	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1635	indefinitely.
		1636
		1637	But really, best use non-blocking mode.
		1638
		1639	=head3 The special problem of disappearing file descriptors
		1640
		1641	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1642	descriptor (either due to calling C<close> explicitly or any other means,
		1643	such as C<dup2>). The reason is that you register interest in some file
		1644	descriptor, but when it goes away, the operating system will silently drop
		1645	this interest. If another file descriptor with the same number then is
		1646	registered with libev, there is no efficient way to see that this is, in
		1647	fact, a different file descriptor.
		1648
		1649	To avoid having to explicitly tell libev about such cases, libev follows
		1650	the following policy: Each time C<ev_io_set> is being called, libev
		1651	will assume that this is potentially a new file descriptor, otherwise
		1652	it is assumed that the file descriptor stays the same. That means that
		1653	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1654	descriptor even if the file descriptor number itself did not change.
		1655
		1656	This is how one would do it normally anyway, the important point is that
		1657	the libev application should not optimise around libev but should leave
		1658	optimisations to libev.
		1659
		1660	=head3 The special problem of dup'ed file descriptors
		1661
		1662	Some backends (e.g. epoll), cannot register events for file descriptors,
		1663	but only events for the underlying file descriptions. That means when you
		1664	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1665	events for them, only one file descriptor might actually receive events.
		1666
		1667	There is no workaround possible except not registering events
		1668	for potentially C<dup ()>'ed file descriptors, or to resort to
		1669	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1670
		1671	=head3 The special problem of files
		1672
		1673	Many people try to use C<select> (or libev) on file descriptors
		1674	representing files, and expect it to become ready when their program
		1675	doesn't block on disk accesses (which can take a long time on their own).
		1676
		1677	However, this cannot ever work in the "expected" way - you get a readiness
		1678	notification as soon as the kernel knows whether and how much data is
		1679	there, and in the case of open files, that's always the case, so you
		1680	always get a readiness notification instantly, and your read (or possibly
		1681	write) will still block on the disk I/O.
		1682
		1683	Another way to view it is that in the case of sockets, pipes, character
		1684	devices and so on, there is another party (the sender) that delivers data
		1685	on it's own, but in the case of files, there is no such thing: the disk
		1686	will not send data on it's own, simply because it doesn't know what you
		1687	wish to read - you would first have to request some data.
		1688
		1689	Since files are typically not-so-well supported by advanced notification
		1690	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1691	to files, even though you should not use it. The reason for this is
		1692	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1693	usually a tty, often a pipe, but also sometimes files or special devices
		1694	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1695	F</dev/urandom>), and even though the file might better be served with
		1696	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1697	it "just works" instead of freezing.
		1698
		1699	So avoid file descriptors pointing to files when you know it (e.g. use
		1700	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1701	when you rarely read from a file instead of from a socket, and want to
		1702	reuse the same code path.
		1703
		1704	=head3 The special problem of fork
		1705
		1706	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1707	useless behaviour. Libev fully supports fork, but needs to be told about
		1708	it in the child if you want to continue to use it in the child.
		1709
		1710	To support fork in your child processes, you have to call C<ev_loop_fork
		1711	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
		1712	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1713
		1714	=head3 The special problem of SIGPIPE
		1715
		1716	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1717	when writing to a pipe whose other end has been closed, your program gets
		1718	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1719	this is sensible behaviour, for daemons, this is usually undesirable.
		1720
		1721	So when you encounter spurious, unexplained daemon exits, make sure you
		1722	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1723	somewhere, as that would have given you a big clue).
		1724
		1725	=head3 The special problem of accept()ing when you can't
		1726
		1727	Many implementations of the POSIX C<accept> function (for example,
		1728	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1729	connection from the pending queue in all error cases.
		1730
		1731	For example, larger servers often run out of file descriptors (because
		1732	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1733	rejecting the connection, leading to libev signalling readiness on
		1734	the next iteration again (the connection still exists after all), and
		1735	typically causing the program to loop at 100% CPU usage.
		1736
		1737	Unfortunately, the set of errors that cause this issue differs between
		1738	operating systems, there is usually little the app can do to remedy the
		1739	situation, and no known thread-safe method of removing the connection to
		1740	cope with overload is known (to me).
		1741
		1742	One of the easiest ways to handle this situation is to just ignore it
		1743	- when the program encounters an overload, it will just loop until the
		1744	situation is over. While this is a form of busy waiting, no OS offers an
		1745	event-based way to handle this situation, so it's the best one can do.
		1746
		1747	A better way to handle the situation is to log any errors other than
		1748	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1749	messages, and continue as usual, which at least gives the user an idea of
		1750	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1751	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1752	usage.
		1753
		1754	If your program is single-threaded, then you could also keep a dummy file
		1755	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1756	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1757	close that fd, and create a new dummy fd. This will gracefully refuse
		1758	clients under typical overload conditions.
		1759
		1760	The last way to handle it is to simply log the error and C<exit>, as
		1761	is often done with C<malloc> failures, but this results in an easy
		1762	opportunity for a DoS attack.
		1763
		1764	=head3 Watcher-Specific Functions
428		1765
429	=over 4	1766	=over 4
430		1767
431	=item ev_io_init (ev_io *, callback, int fd, int events)	1768	=item ev_io_init (ev_io *, callback, int fd, int events)
432		1769
433	=item ev_io_set (ev_io *, int fd, int events)	1770	=item ev_io_set (ev_io *, int fd, int events)
434		1771
435	Configures an ev_io watcher. The fd is the file descriptor to rceeive	1772	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
436	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1773	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
437	EV_WRITE> to receive the given events.	1774	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1775
		1776	=item int fd [read-only]
		1777
		1778	The file descriptor being watched.
		1779
		1780	=item int events [read-only]
		1781
		1782	The events being watched.
438		1783
439	=back	1784	=back
440		1785
		1786	=head3 Examples
		1787
		1788	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1789	readable, but only once. Since it is likely line-buffered, you could
		1790	attempt to read a whole line in the callback.
		1791
		1792	static void
		1793	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1794	{
		1795	ev_io_stop (loop, w);
		1796	.. read from stdin here (or from w->fd) and handle any I/O errors
		1797	}
		1798
		1799	...
		1800	struct ev_loop *loop = ev_default_init (0);
		1801	ev_io stdin_readable;
		1802	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1803	ev_io_start (loop, &stdin_readable);
		1804	ev_run (loop, 0);
		1805
		1806
441	=head2 struct ev_timer - relative and optionally recurring timeouts	1807	=head2 C<ev_timer> - relative and optionally repeating timeouts
442		1808
443	Timer watchers are simple relative timers that generate an event after a	1809	Timer watchers are simple relative timers that generate an event after a
444	given time, and optionally repeating in regular intervals after that.	1810	given time, and optionally repeating in regular intervals after that.
445		1811
446	The timers are based on real time, that is, if you register an event that	1812	The timers are based on real time, that is, if you register an event that
447	times out after an hour and youreset your system clock to last years	1813	times out after an hour and you reset your system clock to January last
448	time, it will still time out after (roughly) and hour. "Roughly" because	1814	year, it will still time out after (roughly) one hour. "Roughly" because
449	detecting time jumps is hard, and soem inaccuracies are unavoidable (the	1815	detecting time jumps is hard, and some inaccuracies are unavoidable (the
450	monotonic clock option helps a lot here).	1816	monotonic clock option helps a lot here).
451		1817
		1818	The callback is guaranteed to be invoked only I<after> its timeout has
		1819	passed (not I<at>, so on systems with very low-resolution clocks this
		1820	might introduce a small delay). If multiple timers become ready during the
		1821	same loop iteration then the ones with earlier time-out values are invoked
		1822	before ones of the same priority with later time-out values (but this is
		1823	no longer true when a callback calls C<ev_run> recursively).
		1824
		1825	=head3 Be smart about timeouts
		1826
		1827	Many real-world problems involve some kind of timeout, usually for error
		1828	recovery. A typical example is an HTTP request - if the other side hangs,
		1829	you want to raise some error after a while.
		1830
		1831	What follows are some ways to handle this problem, from obvious and
		1832	inefficient to smart and efficient.
		1833
		1834	In the following, a 60 second activity timeout is assumed - a timeout that
		1835	gets reset to 60 seconds each time there is activity (e.g. each time some
		1836	data or other life sign was received).
		1837
452	=over 4	1838	=over 4
453		1839
		1840	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1841
		1842	This is the most obvious, but not the most simple way: In the beginning,
		1843	start the watcher:
		1844
		1845	ev_timer_init (timer, callback, 60., 0.);
		1846	ev_timer_start (loop, timer);
		1847
		1848	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1849	and start it again:
		1850
		1851	ev_timer_stop (loop, timer);
		1852	ev_timer_set (timer, 60., 0.);
		1853	ev_timer_start (loop, timer);
		1854
		1855	This is relatively simple to implement, but means that each time there is
		1856	some activity, libev will first have to remove the timer from its internal
		1857	data structure and then add it again. Libev tries to be fast, but it's
		1858	still not a constant-time operation.
		1859
		1860	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1861
		1862	This is the easiest way, and involves using C<ev_timer_again> instead of
		1863	C<ev_timer_start>.
		1864
		1865	To implement this, configure an C<ev_timer> with a C<repeat> value
		1866	of C<60> and then call C<ev_timer_again> at start and each time you
		1867	successfully read or write some data. If you go into an idle state where
		1868	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1869	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1870
		1871	That means you can ignore both the C<ev_timer_start> function and the
		1872	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1873	member and C<ev_timer_again>.
		1874
		1875	At start:
		1876
		1877	ev_init (timer, callback);
		1878	timer->repeat = 60.;
		1879	ev_timer_again (loop, timer);
		1880
		1881	Each time there is some activity:
		1882
		1883	ev_timer_again (loop, timer);
		1884
		1885	It is even possible to change the time-out on the fly, regardless of
		1886	whether the watcher is active or not:
		1887
		1888	timer->repeat = 30.;
		1889	ev_timer_again (loop, timer);
		1890
		1891	This is slightly more efficient then stopping/starting the timer each time
		1892	you want to modify its timeout value, as libev does not have to completely
		1893	remove and re-insert the timer from/into its internal data structure.
		1894
		1895	It is, however, even simpler than the "obvious" way to do it.
		1896
		1897	=item 3. Let the timer time out, but then re-arm it as required.
		1898
		1899	This method is more tricky, but usually most efficient: Most timeouts are
		1900	relatively long compared to the intervals between other activity - in
		1901	our example, within 60 seconds, there are usually many I/O events with
		1902	associated activity resets.
		1903
		1904	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1905	but remember the time of last activity, and check for a real timeout only
		1906	within the callback:
		1907
		1908	ev_tstamp last_activity; // time of last activity
		1909
		1910	static void
		1911	callback (EV_P_ ev_timer *w, int revents)
		1912	{
		1913	ev_tstamp now = ev_now (EV_A);
		1914	ev_tstamp timeout = last_activity + 60.;
		1915
		1916	// if last_activity + 60. is older than now, we did time out
		1917	if (timeout < now)
		1918	{
		1919	// timeout occurred, take action
		1920	}
		1921	else
		1922	{
		1923	// callback was invoked, but there was some activity, re-arm
		1924	// the watcher to fire in last_activity + 60, which is
		1925	// guaranteed to be in the future, so "again" is positive:
		1926	w->repeat = timeout - now;
		1927	ev_timer_again (EV_A_ w);
		1928	}
		1929	}
		1930
		1931	To summarise the callback: first calculate the real timeout (defined
		1932	as "60 seconds after the last activity"), then check if that time has
		1933	been reached, which means something I<did>, in fact, time out. Otherwise
		1934	the callback was invoked too early (C<timeout> is in the future), so
		1935	re-schedule the timer to fire at that future time, to see if maybe we have
		1936	a timeout then.
		1937
		1938	Note how C<ev_timer_again> is used, taking advantage of the
		1939	C<ev_timer_again> optimisation when the timer is already running.
		1940
		1941	This scheme causes more callback invocations (about one every 60 seconds
		1942	minus half the average time between activity), but virtually no calls to
		1943	libev to change the timeout.
		1944
		1945	To start the timer, simply initialise the watcher and set C<last_activity>
		1946	to the current time (meaning we just have some activity :), then call the
		1947	callback, which will "do the right thing" and start the timer:
		1948
		1949	ev_init (timer, callback);
		1950	last_activity = ev_now (loop);
		1951	callback (loop, timer, EV_TIMER);
		1952
		1953	And when there is some activity, simply store the current time in
		1954	C<last_activity>, no libev calls at all:
		1955
		1956	last_activity = ev_now (loop);
		1957
		1958	This technique is slightly more complex, but in most cases where the
		1959	time-out is unlikely to be triggered, much more efficient.
		1960
		1961	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1962	callback :) - just change the timeout and invoke the callback, which will
		1963	fix things for you.
		1964
		1965	=item 4. Wee, just use a double-linked list for your timeouts.
		1966
		1967	If there is not one request, but many thousands (millions...), all
		1968	employing some kind of timeout with the same timeout value, then one can
		1969	do even better:
		1970
		1971	When starting the timeout, calculate the timeout value and put the timeout
		1972	at the I<end> of the list.
		1973
		1974	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1975	the list is expected to fire (for example, using the technique #3).
		1976
		1977	When there is some activity, remove the timer from the list, recalculate
		1978	the timeout, append it to the end of the list again, and make sure to
		1979	update the C<ev_timer> if it was taken from the beginning of the list.
		1980
		1981	This way, one can manage an unlimited number of timeouts in O(1) time for
		1982	starting, stopping and updating the timers, at the expense of a major
		1983	complication, and having to use a constant timeout. The constant timeout
		1984	ensures that the list stays sorted.
		1985
		1986	=back
		1987
		1988	So which method the best?
		1989
		1990	Method #2 is a simple no-brain-required solution that is adequate in most
		1991	situations. Method #3 requires a bit more thinking, but handles many cases
		1992	better, and isn't very complicated either. In most case, choosing either
		1993	one is fine, with #3 being better in typical situations.
		1994
		1995	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1996	rather complicated, but extremely efficient, something that really pays
		1997	off after the first million or so of active timers, i.e. it's usually
		1998	overkill :)
		1999
		2000	=head3 The special problem of time updates
		2001
		2002	Establishing the current time is a costly operation (it usually takes at
		2003	least two system calls): EV therefore updates its idea of the current
		2004	time only before and after C<ev_run> collects new events, which causes a
		2005	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		2006	lots of events in one iteration.
		2007
		2008	The relative timeouts are calculated relative to the C<ev_now ()>
		2009	time. This is usually the right thing as this timestamp refers to the time
		2010	of the event triggering whatever timeout you are modifying/starting. If
		2011	you suspect event processing to be delayed and you I<need> to base the
		2012	timeout on the current time, use something like this to adjust for this:
		2013
		2014	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
		2015
		2016	If the event loop is suspended for a long time, you can also force an
		2017	update of the time returned by C<ev_now ()> by calling C<ev_now_update
		2018	()>.
		2019
		2020	=head3 The special problems of suspended animation
		2021
		2022	When you leave the server world it is quite customary to hit machines that
		2023	can suspend/hibernate - what happens to the clocks during such a suspend?
		2024
		2025	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		2026	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		2027	to run until the system is suspended, but they will not advance while the
		2028	system is suspended. That means, on resume, it will be as if the program
		2029	was frozen for a few seconds, but the suspend time will not be counted
		2030	towards C<ev_timer> when a monotonic clock source is used. The real time
		2031	clock advanced as expected, but if it is used as sole clocksource, then a
		2032	long suspend would be detected as a time jump by libev, and timers would
		2033	be adjusted accordingly.
		2034
		2035	I would not be surprised to see different behaviour in different between
		2036	operating systems, OS versions or even different hardware.
		2037
		2038	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2039	time jump in the monotonic clocks and the realtime clock. If the program
		2040	is suspended for a very long time, and monotonic clock sources are in use,
		2041	then you can expect C<ev_timer>s to expire as the full suspension time
		2042	will be counted towards the timers. When no monotonic clock source is in
		2043	use, then libev will again assume a timejump and adjust accordingly.
		2044
		2045	It might be beneficial for this latter case to call C<ev_suspend>
		2046	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2047	deterministic behaviour in this case (you can do nothing against
		2048	C<SIGSTOP>).
		2049
		2050	=head3 Watcher-Specific Functions and Data Members
		2051
		2052	=over 4
		2053
454	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2054	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
455		2055
456	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2056	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
457		2057
458	Configure the timer to trigger after C<after> seconds. If C<repeat> is	2058	Configure the timer to trigger after C<after> seconds. If C<repeat>
459	C<0.>, then it will automatically be stopped. If it is positive, then the	2059	is C<0.>, then it will automatically be stopped once the timeout is
460	timer will automatically be configured to trigger again C<repeat> seconds	2060	reached. If it is positive, then the timer will automatically be
461	later, again, and again, until stopped manually.	2061	configured to trigger again C<repeat> seconds later, again, and again,
		2062	until stopped manually.
462		2063
463	The timer itself will do a best-effort at avoiding drift, that is, if you	2064	The timer itself will do a best-effort at avoiding drift, that is, if
464	configure a timer to trigger every 10 seconds, then it will trigger at	2065	you configure a timer to trigger every 10 seconds, then it will normally
465	exactly 10 second intervals. If, however, your program cannot keep up with	2066	trigger at exactly 10 second intervals. If, however, your program cannot
466	the timer (ecause it takes longer than those 10 seconds to do stuff) the	2067	keep up with the timer (because it takes longer than those 10 seconds to
467	timer will not fire more than once per event loop iteration.	2068	do stuff) the timer will not fire more than once per event loop iteration.
468		2069
469	=item ev_timer_again (loop)	2070	=item ev_timer_again (loop, ev_timer *)
470		2071
471	This will act as if the timer timed out and restart it again if it is	2072	This will act as if the timer timed out and restart it again if it is
472	repeating. The exact semantics are:	2073	repeating. The exact semantics are:
473		2074
		2075	If the timer is pending, its pending status is cleared.
		2076
474	If the timer is started but nonrepeating, stop it.	2077	If the timer is started but non-repeating, stop it (as if it timed out).
475		2078
476	If the timer is repeating, either start it if necessary (with the repeat	2079	If the timer is repeating, either start it if necessary (with the
477	value), or reset the running timer to the repeat value.	2080	C<repeat> value), or reset the running timer to the C<repeat> value.
478		2081
479	This sounds a bit complicated, but here is a useful and typical	2082	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
480	example: Imagine you have a tcp connection and you want a so-called idle	2083	usage example.
481	timeout, that is, you want to be called when there have been, say, 60	2084
482	seconds of inactivity on the socket. The easiest way to do this is to	2085	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
483	configure an ev_timer with after=repeat=60 and calling ev_timer_again each	2086
484	time you successfully read or write some data. If you go into an idle	2087	Returns the remaining time until a timer fires. If the timer is active,
485	state where you do not expect data to travel on the socket, you can stop	2088	then this time is relative to the current event loop time, otherwise it's
486	the timer, and again will automatically restart it if need be.	2089	the timeout value currently configured.
		2090
		2091	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2092	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2093	will return C<4>. When the timer expires and is restarted, it will return
		2094	roughly C<7> (likely slightly less as callback invocation takes some time,
		2095	too), and so on.
		2096
		2097	=item ev_tstamp repeat [read-write]
		2098
		2099	The current C<repeat> value. Will be used each time the watcher times out
		2100	or C<ev_timer_again> is called, and determines the next timeout (if any),
		2101	which is also when any modifications are taken into account.
487		2102
488	=back	2103	=back
489		2104
		2105	=head3 Examples
		2106
		2107	Example: Create a timer that fires after 60 seconds.
		2108
		2109	static void
		2110	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2111	{
		2112	.. one minute over, w is actually stopped right here
		2113	}
		2114
		2115	ev_timer mytimer;
		2116	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		2117	ev_timer_start (loop, &mytimer);
		2118
		2119	Example: Create a timeout timer that times out after 10 seconds of
		2120	inactivity.
		2121
		2122	static void
		2123	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2124	{
		2125	.. ten seconds without any activity
		2126	}
		2127
		2128	ev_timer mytimer;
		2129	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		2130	ev_timer_again (&mytimer); /* start timer */
		2131	ev_run (loop, 0);
		2132
		2133	// and in some piece of code that gets executed on any "activity":
		2134	// reset the timeout to start ticking again at 10 seconds
		2135	ev_timer_again (&mytimer);
		2136
		2137
490	=head2 ev_periodic - to cron or not to cron it	2138	=head2 C<ev_periodic> - to cron or not to cron?
491		2139
492	Periodic watchers are also timers of a kind, but they are very versatile	2140	Periodic watchers are also timers of a kind, but they are very versatile
493	(and unfortunately a bit complex).	2141	(and unfortunately a bit complex).
494		2142
495	Unlike ev_timer's, they are not based on real time (or relative time)	2143	Unlike C<ev_timer>, periodic watchers are not based on real time (or
496	but on wallclock time (absolute time). You can tell a periodic watcher	2144	relative time, the physical time that passes) but on wall clock time
497	to trigger "at" some specific point in time. For example, if you tell a	2145	(absolute time, the thing you can read on your calender or clock). The
498	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	2146	difference is that wall clock time can run faster or slower than real
499	+ 10.>) and then reset your system clock to the last year, then it will	2147	time, and time jumps are not uncommon (e.g. when you adjust your
500	take a year to trigger the event (unlike an ev_timer, which would trigger	2148	wrist-watch).
501	roughly 10 seconds later and of course not if you reset your system time
502	again).
503		2149
504	They can also be used to implement vastly more complex timers, such as	2150	You can tell a periodic watcher to trigger after some specific point
		2151	in time: for example, if you tell a periodic watcher to trigger "in 10
		2152	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2153	not a delay) and then reset your system clock to January of the previous
		2154	year, then it will take a year or more to trigger the event (unlike an
		2155	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2156	it, as it uses a relative timeout).
		2157
		2158	C<ev_periodic> watchers can also be used to implement vastly more complex
505	triggering an event on eahc midnight, local time.	2159	timers, such as triggering an event on each "midnight, local time", or
		2160	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2161	those cannot react to time jumps.
		2162
		2163	As with timers, the callback is guaranteed to be invoked only when the
		2164	point in time where it is supposed to trigger has passed. If multiple
		2165	timers become ready during the same loop iteration then the ones with
		2166	earlier time-out values are invoked before ones with later time-out values
		2167	(but this is no longer true when a callback calls C<ev_run> recursively).
		2168
		2169	=head3 Watcher-Specific Functions and Data Members
506		2170
507	=over 4	2171	=over 4
508		2172
509	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2173	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
510		2174
511	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2175	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
512		2176
513	Lots of arguments, lets sort it out... There are basically three modes of	2177	Lots of arguments, let's sort it out... There are basically three modes of
514	operation, and we will explain them from simplest to complex:	2178	operation, and we will explain them from simplest to most complex:
515
516		2179
517	=over 4	2180	=over 4
518		2181
519	=item * absolute timer (interval = reschedule_cb = 0)	2182	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
520		2183
521	In this configuration the watcher triggers an event at the wallclock time	2184	In this configuration the watcher triggers an event after the wall clock
522	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2185	time C<offset> has passed. It will not repeat and will not adjust when a
523	that is, if it is to be run at January 1st 2011 then it will run when the	2186	time jump occurs, that is, if it is to be run at January 1st 2011 then it
524	system time reaches or surpasses this time.	2187	will be stopped and invoked when the system clock reaches or surpasses
		2188	this point in time.
525		2189
526	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	2190	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
527		2191
528	In this mode the watcher will always be scheduled to time out at the next	2192	In this mode the watcher will always be scheduled to time out at the next
529	C<at + N * interval> time (for some integer N) and then repeat, regardless	2193	C<offset + N * interval> time (for some integer N, which can also be
530	of any time jumps.	2194	negative) and then repeat, regardless of any time jumps. The C<offset>
		2195	argument is merely an offset into the C<interval> periods.
531		2196
532	This can be used to create timers that do not drift with respect to system	2197	This can be used to create timers that do not drift with respect to the
533	time:	2198	system clock, for example, here is an C<ev_periodic> that triggers each
		2199	hour, on the hour (with respect to UTC):
534		2200
535	ev_periodic_set (&periodic, 0., 3600., 0);	2201	ev_periodic_set (&periodic, 0., 3600., 0);
536		2202
537	This doesn't mean there will always be 3600 seconds in between triggers,	2203	This doesn't mean there will always be 3600 seconds in between triggers,
538	but only that the the callback will be called when the system time shows a	2204	but only that the callback will be called when the system time shows a
539	full hour (UTC), or more correct, when the system time is evenly divisible	2205	full hour (UTC), or more correctly, when the system time is evenly divisible
540	by 3600.	2206	by 3600.
541		2207
542	Another way to think about it (for the mathematically inclined) is that	2208	Another way to think about it (for the mathematically inclined) is that
543	ev_periodic will try to run the callback in this mode at the next possible	2209	C<ev_periodic> will try to run the callback in this mode at the next possible
544	time where C<time = at (mod interval)>, regardless of any time jumps.	2210	time where C<time = offset (mod interval)>, regardless of any time jumps.
545		2211
		2212	For numerical stability it is preferable that the C<offset> value is near
		2213	C<ev_now ()> (the current time), but there is no range requirement for
		2214	this value, and in fact is often specified as zero.
		2215
		2216	Note also that there is an upper limit to how often a timer can fire (CPU
		2217	speed for example), so if C<interval> is very small then timing stability
		2218	will of course deteriorate. Libev itself tries to be exact to be about one
		2219	millisecond (if the OS supports it and the machine is fast enough).
		2220
546	=item * manual reschedule mode (reschedule_cb = callback)	2221	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
547		2222
548	In this mode the values for C<interval> and C<at> are both being	2223	In this mode the values for C<interval> and C<offset> are both being
549	ignored. Instead, each time the periodic watcher gets scheduled, the	2224	ignored. Instead, each time the periodic watcher gets scheduled, the
550	reschedule callback will be called with the watcher as first, and the	2225	reschedule callback will be called with the watcher as first, and the
551	current time as second argument.	2226	current time as second argument.
552		2227
553	NOTE: I<This callback MUST NOT stop or destroy the periodic or any other	2228	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
554	periodic watcher, ever, or make any event loop modificstions>. If you need	2229	or make ANY other event loop modifications whatsoever, unless explicitly
555	to stop it, return 1e30 (or so, fudge fudge) and stop it afterwards.	2230	allowed by documentation here>.
556		2231
		2232	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2233	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2234	only event loop modification you are allowed to do).
		2235
557	Its prototype is c<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2236	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
558	ev_tstamp now)>, e.g.:	2237	*w, ev_tstamp now)>, e.g.:
559		2238
		2239	static ev_tstamp
560	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2240	my_rescheduler (ev_periodic *w, ev_tstamp now)
561	{	2241	{
562	return now + 60.;	2242	return now + 60.;
563	}	2243	}
564		2244
565	It must return the next time to trigger, based on the passed time value	2245	It must return the next time to trigger, based on the passed time value
566	(that is, the lowest time value larger than to the second argument). It	2246	(that is, the lowest time value larger than to the second argument). It
567	will usually be called just before the callback will be triggered, but	2247	will usually be called just before the callback will be triggered, but
568	might be called at other times, too.	2248	might be called at other times, too.
569		2249
		2250	NOTE: I<< This callback must always return a time that is higher than or
		2251	equal to the passed C<now> value >>.
		2252
570	This can be used to create very complex timers, such as a timer that	2253	This can be used to create very complex timers, such as a timer that
571	triggers on each midnight, local time. To do this, you would calculate the	2254	triggers on "next midnight, local time". To do this, you would calculate the
572	next midnight after C<now> and return the timestamp value for this. How you do this	2255	next midnight after C<now> and return the timestamp value for this. How
573	is, again, up to you (but it is not trivial).	2256	you do this is, again, up to you (but it is not trivial, which is the main
		2257	reason I omitted it as an example).
574		2258
575	=back	2259	=back
576		2260
577	=item ev_periodic_again (loop, ev_periodic *)	2261	=item ev_periodic_again (loop, ev_periodic *)
578		2262
579	Simply stops and restarts the periodic watcher again. This is only useful	2263	Simply stops and restarts the periodic watcher again. This is only useful
580	when you changed some parameters or the reschedule callback would return	2264	when you changed some parameters or the reschedule callback would return
581	a different time than the last time it was called (e.g. in a crond like	2265	a different time than the last time it was called (e.g. in a crond like
582	program when the crontabs have changed).	2266	program when the crontabs have changed).
583		2267
		2268	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2269
		2270	When active, returns the absolute time that the watcher is supposed
		2271	to trigger next. This is not the same as the C<offset> argument to
		2272	C<ev_periodic_set>, but indeed works even in interval and manual
		2273	rescheduling modes.
		2274
		2275	=item ev_tstamp offset [read-write]
		2276
		2277	When repeating, this contains the offset value, otherwise this is the
		2278	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2279	although libev might modify this value for better numerical stability).
		2280
		2281	Can be modified any time, but changes only take effect when the periodic
		2282	timer fires or C<ev_periodic_again> is being called.
		2283
		2284	=item ev_tstamp interval [read-write]
		2285
		2286	The current interval value. Can be modified any time, but changes only
		2287	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2288	called.
		2289
		2290	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2291
		2292	The current reschedule callback, or C<0>, if this functionality is
		2293	switched off. Can be changed any time, but changes only take effect when
		2294	the periodic timer fires or C<ev_periodic_again> is being called.
		2295
584	=back	2296	=back
585		2297
		2298	=head3 Examples
		2299
		2300	Example: Call a callback every hour, or, more precisely, whenever the
		2301	system time is divisible by 3600. The callback invocation times have
		2302	potentially a lot of jitter, but good long-term stability.
		2303
		2304	static void
		2305	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
		2306	{
		2307	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2308	}
		2309
		2310	ev_periodic hourly_tick;
		2311	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2312	ev_periodic_start (loop, &hourly_tick);
		2313
		2314	Example: The same as above, but use a reschedule callback to do it:
		2315
		2316	#include <math.h>
		2317
		2318	static ev_tstamp
		2319	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2320	{
		2321	return now + (3600. - fmod (now, 3600.));
		2322	}
		2323
		2324	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2325
		2326	Example: Call a callback every hour, starting now:
		2327
		2328	ev_periodic hourly_tick;
		2329	ev_periodic_init (&hourly_tick, clock_cb,
		2330	fmod (ev_now (loop), 3600.), 3600., 0);
		2331	ev_periodic_start (loop, &hourly_tick);
		2332
		2333
586	=head2 ev_signal - signal me when a signal gets signalled	2334	=head2 C<ev_signal> - signal me when a signal gets signalled!
587		2335
588	Signal watchers will trigger an event when the process receives a specific	2336	Signal watchers will trigger an event when the process receives a specific
589	signal one or more times. Even though signals are very asynchronous, libev	2337	signal one or more times. Even though signals are very asynchronous, libev
590	will try its best to deliver signals synchronously, i.e. as part of the	2338	will try its best to deliver signals synchronously, i.e. as part of the
591	normal event processing, like any other event.	2339	normal event processing, like any other event.
592		2340
		2341	If you want signals to be delivered truly asynchronously, just use
		2342	C<sigaction> as you would do without libev and forget about sharing
		2343	the signal. You can even use C<ev_async> from a signal handler to
		2344	synchronously wake up an event loop.
		2345
593	You cna configure as many watchers as you like per signal. Only when the	2346	You can configure as many watchers as you like for the same signal, but
		2347	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2348	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2349	C<SIGINT> in both the default loop and another loop at the same time. At
		2350	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2351
594	first watcher gets started will libev actually register a signal watcher	2352	When the first watcher gets started will libev actually register something
595	with the kernel (thus it coexists with your own signal handlers as long	2353	with the kernel (thus it coexists with your own signal handlers as long as
596	as you don't register any with libev). Similarly, when the last signal	2354	you don't register any with libev for the same signal).
597	watcher for a signal is stopped libev will reset the signal handler to	2355
598	SIG_DFL (regardless of what it was set to before).	2356	If possible and supported, libev will install its handlers with
		2357	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2358	not be unduly interrupted. If you have a problem with system calls getting
		2359	interrupted by signals you can block all signals in an C<ev_check> watcher
		2360	and unblock them in an C<ev_prepare> watcher.
		2361
		2362	=head3 The special problem of inheritance over fork/execve/pthread_create
		2363
		2364	Both the signal mask (C<sigprocmask>) and the signal disposition
		2365	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2366	stopping it again), that is, libev might or might not block the signal,
		2367	and might or might not set or restore the installed signal handler.
		2368
		2369	While this does not matter for the signal disposition (libev never
		2370	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2371	C<execve>), this matters for the signal mask: many programs do not expect
		2372	certain signals to be blocked.
		2373
		2374	This means that before calling C<exec> (from the child) you should reset
		2375	the signal mask to whatever "default" you expect (all clear is a good
		2376	choice usually).
		2377
		2378	The simplest way to ensure that the signal mask is reset in the child is
		2379	to install a fork handler with C<pthread_atfork> that resets it. That will
		2380	catch fork calls done by libraries (such as the libc) as well.
		2381
		2382	In current versions of libev, the signal will not be blocked indefinitely
		2383	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2384	the window of opportunity for problems, it will not go away, as libev
		2385	I<has> to modify the signal mask, at least temporarily.
		2386
		2387	So I can't stress this enough: I<If you do not reset your signal mask when
		2388	you expect it to be empty, you have a race condition in your code>. This
		2389	is not a libev-specific thing, this is true for most event libraries.
		2390
		2391	=head3 The special problem of threads signal handling
		2392
		2393	POSIX threads has problematic signal handling semantics, specifically,
		2394	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2395	threads in a process block signals, which is hard to achieve.
		2396
		2397	When you want to use sigwait (or mix libev signal handling with your own
		2398	for the same signals), you can tackle this problem by globally blocking
		2399	all signals before creating any threads (or creating them with a fully set
		2400	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2401	loops. Then designate one thread as "signal receiver thread" which handles
		2402	these signals. You can pass on any signals that libev might be interested
		2403	in by calling C<ev_feed_signal>.
		2404
		2405	=head3 Watcher-Specific Functions and Data Members
599		2406
600	=over 4	2407	=over 4
601		2408
602	=item ev_signal_init (ev_signal *, callback, int signum)	2409	=item ev_signal_init (ev_signal *, callback, int signum)
603		2410
604	=item ev_signal_set (ev_signal *, int signum)	2411	=item ev_signal_set (ev_signal *, int signum)
605		2412
606	Configures the watcher to trigger on the given signal number (usually one	2413	Configures the watcher to trigger on the given signal number (usually one
607	of the C<SIGxxx> constants).	2414	of the C<SIGxxx> constants).
608		2415
		2416	=item int signum [read-only]
		2417
		2418	The signal the watcher watches out for.
		2419
609	=back	2420	=back
610		2421
		2422	=head3 Examples
		2423
		2424	Example: Try to exit cleanly on SIGINT.
		2425
		2426	static void
		2427	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2428	{
		2429	ev_break (loop, EVBREAK_ALL);
		2430	}
		2431
		2432	ev_signal signal_watcher;
		2433	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2434	ev_signal_start (loop, &signal_watcher);
		2435
		2436
611	=head2 ev_child - wait for pid status changes	2437	=head2 C<ev_child> - watch out for process status changes
612		2438
613	Child watchers trigger when your process receives a SIGCHLD in response to	2439	Child watchers trigger when your process receives a SIGCHLD in response to
614	some child status changes (most typically when a child of yours dies).	2440	some child status changes (most typically when a child of yours dies or
		2441	exits). It is permissible to install a child watcher I<after> the child
		2442	has been forked (which implies it might have already exited), as long
		2443	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2444	forking and then immediately registering a watcher for the child is fine,
		2445	but forking and registering a watcher a few event loop iterations later or
		2446	in the next callback invocation is not.
		2447
		2448	Only the default event loop is capable of handling signals, and therefore
		2449	you can only register child watchers in the default event loop.
		2450
		2451	Due to some design glitches inside libev, child watchers will always be
		2452	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2453	libev)
		2454
		2455	=head3 Process Interaction
		2456
		2457	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2458	initialised. This is necessary to guarantee proper behaviour even if the
		2459	first child watcher is started after the child exits. The occurrence
		2460	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2461	synchronously as part of the event loop processing. Libev always reaps all
		2462	children, even ones not watched.
		2463
		2464	=head3 Overriding the Built-In Processing
		2465
		2466	Libev offers no special support for overriding the built-in child
		2467	processing, but if your application collides with libev's default child
		2468	handler, you can override it easily by installing your own handler for
		2469	C<SIGCHLD> after initialising the default loop, and making sure the
		2470	default loop never gets destroyed. You are encouraged, however, to use an
		2471	event-based approach to child reaping and thus use libev's support for
		2472	that, so other libev users can use C<ev_child> watchers freely.
		2473
		2474	=head3 Stopping the Child Watcher
		2475
		2476	Currently, the child watcher never gets stopped, even when the
		2477	child terminates, so normally one needs to stop the watcher in the
		2478	callback. Future versions of libev might stop the watcher automatically
		2479	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2480	problem).
		2481
		2482	=head3 Watcher-Specific Functions and Data Members
615		2483
616	=over 4	2484	=over 4
617		2485
618	=item ev_child_init (ev_child *, callback, int pid)	2486	=item ev_child_init (ev_child *, callback, int pid, int trace)
619		2487
620	=item ev_child_set (ev_child *, int pid)	2488	=item ev_child_set (ev_child *, int pid, int trace)
621		2489
622	Configures the watcher to wait for status changes of process C<pid> (or	2490	Configures the watcher to wait for status changes of process C<pid> (or
623	I<any> process if C<pid> is specified as C<0>). The callback can look	2491	I<any> process if C<pid> is specified as C<0>). The callback can look
624	at the C<rstatus> member of the C<ev_child> watcher structure to see	2492	at the C<rstatus> member of the C<ev_child> watcher structure to see
625	the status word (use the macros from C<sys/wait.h>). The C<rpid> member	2493	the status word (use the macros from C<sys/wait.h> and see your systems
626	contains the pid of the process causing the status change.	2494	C<waitpid> documentation). The C<rpid> member contains the pid of the
		2495	process causing the status change. C<trace> must be either C<0> (only
		2496	activate the watcher when the process terminates) or C<1> (additionally
		2497	activate the watcher when the process is stopped or continued).
		2498
		2499	=item int pid [read-only]
		2500
		2501	The process id this watcher watches out for, or C<0>, meaning any process id.
		2502
		2503	=item int rpid [read-write]
		2504
		2505	The process id that detected a status change.
		2506
		2507	=item int rstatus [read-write]
		2508
		2509	The process exit/trace status caused by C<rpid> (see your systems
		2510	C<waitpid> and C<sys/wait.h> documentation for details).
627		2511
628	=back	2512	=back
629		2513
		2514	=head3 Examples
		2515
		2516	Example: C<fork()> a new process and install a child handler to wait for
		2517	its completion.
		2518
		2519	ev_child cw;
		2520
		2521	static void
		2522	child_cb (EV_P_ ev_child *w, int revents)
		2523	{
		2524	ev_child_stop (EV_A_ w);
		2525	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2526	}
		2527
		2528	pid_t pid = fork ();
		2529
		2530	if (pid < 0)
		2531	// error
		2532	else if (pid == 0)
		2533	{
		2534	// the forked child executes here
		2535	exit (1);
		2536	}
		2537	else
		2538	{
		2539	ev_child_init (&cw, child_cb, pid, 0);
		2540	ev_child_start (EV_DEFAULT_ &cw);
		2541	}
		2542
		2543
		2544	=head2 C<ev_stat> - did the file attributes just change?
		2545
		2546	This watches a file system path for attribute changes. That is, it calls
		2547	C<stat> on that path in regular intervals (or when the OS says it changed)
		2548	and sees if it changed compared to the last time, invoking the callback if
		2549	it did.
		2550
		2551	The path does not need to exist: changing from "path exists" to "path does
		2552	not exist" is a status change like any other. The condition "path does not
		2553	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2554	C<st_nlink> field being zero (which is otherwise always forced to be at
		2555	least one) and all the other fields of the stat buffer having unspecified
		2556	contents.
		2557
		2558	The path I<must not> end in a slash or contain special components such as
		2559	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2560	your working directory changes, then the behaviour is undefined.
		2561
		2562	Since there is no portable change notification interface available, the
		2563	portable implementation simply calls C<stat(2)> regularly on the path
		2564	to see if it changed somehow. You can specify a recommended polling
		2565	interval for this case. If you specify a polling interval of C<0> (highly
		2566	recommended!) then a I<suitable, unspecified default> value will be used
		2567	(which you can expect to be around five seconds, although this might
		2568	change dynamically). Libev will also impose a minimum interval which is
		2569	currently around C<0.1>, but that's usually overkill.
		2570
		2571	This watcher type is not meant for massive numbers of stat watchers,
		2572	as even with OS-supported change notifications, this can be
		2573	resource-intensive.
		2574
		2575	At the time of this writing, the only OS-specific interface implemented
		2576	is the Linux inotify interface (implementing kqueue support is left as an
		2577	exercise for the reader. Note, however, that the author sees no way of
		2578	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2579
		2580	=head3 ABI Issues (Largefile Support)
		2581
		2582	Libev by default (unless the user overrides this) uses the default
		2583	compilation environment, which means that on systems with large file
		2584	support disabled by default, you get the 32 bit version of the stat
		2585	structure. When using the library from programs that change the ABI to
		2586	use 64 bit file offsets the programs will fail. In that case you have to
		2587	compile libev with the same flags to get binary compatibility. This is
		2588	obviously the case with any flags that change the ABI, but the problem is
		2589	most noticeably displayed with ev_stat and large file support.
		2590
		2591	The solution for this is to lobby your distribution maker to make large
		2592	file interfaces available by default (as e.g. FreeBSD does) and not
		2593	optional. Libev cannot simply switch on large file support because it has
		2594	to exchange stat structures with application programs compiled using the
		2595	default compilation environment.
		2596
		2597	=head3 Inotify and Kqueue
		2598
		2599	When C<inotify (7)> support has been compiled into libev and present at
		2600	runtime, it will be used to speed up change detection where possible. The
		2601	inotify descriptor will be created lazily when the first C<ev_stat>
		2602	watcher is being started.
		2603
		2604	Inotify presence does not change the semantics of C<ev_stat> watchers
		2605	except that changes might be detected earlier, and in some cases, to avoid
		2606	making regular C<stat> calls. Even in the presence of inotify support
		2607	there are many cases where libev has to resort to regular C<stat> polling,
		2608	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2609	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2610	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2611	xfs are fully working) libev usually gets away without polling.
		2612
		2613	There is no support for kqueue, as apparently it cannot be used to
		2614	implement this functionality, due to the requirement of having a file
		2615	descriptor open on the object at all times, and detecting renames, unlinks
		2616	etc. is difficult.
		2617
		2618	=head3 C<stat ()> is a synchronous operation
		2619
		2620	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2621	the process. The exception are C<ev_stat> watchers - those call C<stat
		2622	()>, which is a synchronous operation.
		2623
		2624	For local paths, this usually doesn't matter: unless the system is very
		2625	busy or the intervals between stat's are large, a stat call will be fast,
		2626	as the path data is usually in memory already (except when starting the
		2627	watcher).
		2628
		2629	For networked file systems, calling C<stat ()> can block an indefinite
		2630	time due to network issues, and even under good conditions, a stat call
		2631	often takes multiple milliseconds.
		2632
		2633	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2634	paths, although this is fully supported by libev.
		2635
		2636	=head3 The special problem of stat time resolution
		2637
		2638	The C<stat ()> system call only supports full-second resolution portably,
		2639	and even on systems where the resolution is higher, most file systems
		2640	still only support whole seconds.
		2641
		2642	That means that, if the time is the only thing that changes, you can
		2643	easily miss updates: on the first update, C<ev_stat> detects a change and
		2644	calls your callback, which does something. When there is another update
		2645	within the same second, C<ev_stat> will be unable to detect unless the
		2646	stat data does change in other ways (e.g. file size).
		2647
		2648	The solution to this is to delay acting on a change for slightly more
		2649	than a second (or till slightly after the next full second boundary), using
		2650	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2651	ev_timer_again (loop, w)>).
		2652
		2653	The C<.02> offset is added to work around small timing inconsistencies
		2654	of some operating systems (where the second counter of the current time
		2655	might be be delayed. One such system is the Linux kernel, where a call to
		2656	C<gettimeofday> might return a timestamp with a full second later than
		2657	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2658	update file times then there will be a small window where the kernel uses
		2659	the previous second to update file times but libev might already execute
		2660	the timer callback).
		2661
		2662	=head3 Watcher-Specific Functions and Data Members
		2663
		2664	=over 4
		2665
		2666	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2667
		2668	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2669
		2670	Configures the watcher to wait for status changes of the given
		2671	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2672	be detected and should normally be specified as C<0> to let libev choose
		2673	a suitable value. The memory pointed to by C<path> must point to the same
		2674	path for as long as the watcher is active.
		2675
		2676	The callback will receive an C<EV_STAT> event when a change was detected,
		2677	relative to the attributes at the time the watcher was started (or the
		2678	last change was detected).
		2679
		2680	=item ev_stat_stat (loop, ev_stat *)
		2681
		2682	Updates the stat buffer immediately with new values. If you change the
		2683	watched path in your callback, you could call this function to avoid
		2684	detecting this change (while introducing a race condition if you are not
		2685	the only one changing the path). Can also be useful simply to find out the
		2686	new values.
		2687
		2688	=item ev_statdata attr [read-only]
		2689
		2690	The most-recently detected attributes of the file. Although the type is
		2691	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2692	suitable for your system, but you can only rely on the POSIX-standardised
		2693	members to be present. If the C<st_nlink> member is C<0>, then there was
		2694	some error while C<stat>ing the file.
		2695
		2696	=item ev_statdata prev [read-only]
		2697
		2698	The previous attributes of the file. The callback gets invoked whenever
		2699	C<prev> != C<attr>, or, more precisely, one or more of these members
		2700	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2701	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2702
		2703	=item ev_tstamp interval [read-only]
		2704
		2705	The specified interval.
		2706
		2707	=item const char *path [read-only]
		2708
		2709	The file system path that is being watched.
		2710
		2711	=back
		2712
		2713	=head3 Examples
		2714
		2715	Example: Watch C</etc/passwd> for attribute changes.
		2716
		2717	static void
		2718	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2719	{
		2720	/* /etc/passwd changed in some way */
		2721	if (w->attr.st_nlink)
		2722	{
		2723	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2724	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2725	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2726	}
		2727	else
		2728	/* you shalt not abuse printf for puts */
		2729	puts ("wow, /etc/passwd is not there, expect problems. "
		2730	"if this is windows, they already arrived\n");
		2731	}
		2732
		2733	...
		2734	ev_stat passwd;
		2735
		2736	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2737	ev_stat_start (loop, &passwd);
		2738
		2739	Example: Like above, but additionally use a one-second delay so we do not
		2740	miss updates (however, frequent updates will delay processing, too, so
		2741	one might do the work both on C<ev_stat> callback invocation I<and> on
		2742	C<ev_timer> callback invocation).
		2743
		2744	static ev_stat passwd;
		2745	static ev_timer timer;
		2746
		2747	static void
		2748	timer_cb (EV_P_ ev_timer *w, int revents)
		2749	{
		2750	ev_timer_stop (EV_A_ w);
		2751
		2752	/* now it's one second after the most recent passwd change */
		2753	}
		2754
		2755	static void
		2756	stat_cb (EV_P_ ev_stat *w, int revents)
		2757	{
		2758	/* reset the one-second timer */
		2759	ev_timer_again (EV_A_ &timer);
		2760	}
		2761
		2762	...
		2763	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2764	ev_stat_start (loop, &passwd);
		2765	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2766
		2767
630	=head2 ev_idle - when you've got nothing better to do	2768	=head2 C<ev_idle> - when you've got nothing better to do...
631		2769
632	Idle watchers trigger events when there are no other I/O or timer (or	2770	Idle watchers trigger events when no other events of the same or higher
633	periodic) events pending. That is, as long as your process is busy	2771	priority are pending (prepare, check and other idle watchers do not count
634	handling sockets or timeouts it will not be called. But when your process	2772	as receiving "events").
635	is idle all idle watchers are being called again and again - until	2773
		2774	That is, as long as your process is busy handling sockets or timeouts
		2775	(or even signals, imagine) of the same or higher priority it will not be
		2776	triggered. But when your process is idle (or only lower-priority watchers
		2777	are pending), the idle watchers are being called once per event loop
636	stopped, that is, or your process receives more events.	2778	iteration - until stopped, that is, or your process receives more events
		2779	and becomes busy again with higher priority stuff.
637		2780
638	The most noteworthy effect is that as long as any idle watchers are	2781	The most noteworthy effect is that as long as any idle watchers are
639	active, the process will not block when waiting for new events.	2782	active, the process will not block when waiting for new events.
640		2783
641	Apart from keeping your process non-blocking (which is a useful	2784	Apart from keeping your process non-blocking (which is a useful
642	effect on its own sometimes), idle watchers are a good place to do	2785	effect on its own sometimes), idle watchers are a good place to do
643	"pseudo-background processing", or delay processing stuff to after the	2786	"pseudo-background processing", or delay processing stuff to after the
644	event loop has handled all outstanding events.	2787	event loop has handled all outstanding events.
645		2788
		2789	=head3 Watcher-Specific Functions and Data Members
		2790
646	=over 4	2791	=over 4
647		2792
648	=item ev_idle_init (ev_signal *, callback)	2793	=item ev_idle_init (ev_idle *, callback)
649		2794
650	Initialises and configures the idle watcher - it has no parameters of any	2795	Initialises and configures the idle watcher - it has no parameters of any
651	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2796	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
652	believe me.	2797	believe me.
653		2798
654	=back	2799	=back
655		2800
656	=head2 prepare and check - your hooks into the event loop	2801	=head3 Examples
657		2802
		2803	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2804	callback, free it. Also, use no error checking, as usual.
		2805
		2806	static void
		2807	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2808	{
		2809	free (w);
		2810	// now do something you wanted to do when the program has
		2811	// no longer anything immediate to do.
		2812	}
		2813
		2814	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2815	ev_idle_init (idle_watcher, idle_cb);
		2816	ev_idle_start (loop, idle_watcher);
		2817
		2818
		2819	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
		2820
658	Prepare and check watchers usually (but not always) are used in	2821	Prepare and check watchers are usually (but not always) used in pairs:
659	tandom. Prepare watchers get invoked before the process blocks and check	2822	prepare watchers get invoked before the process blocks and check watchers
660	watchers afterwards.	2823	afterwards.
661		2824
		2825	You I<must not> call C<ev_run> or similar functions that enter
		2826	the current event loop from either C<ev_prepare> or C<ev_check>
		2827	watchers. Other loops than the current one are fine, however. The
		2828	rationale behind this is that you do not need to check for recursion in
		2829	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2830	C<ev_check> so if you have one watcher of each kind they will always be
		2831	called in pairs bracketing the blocking call.
		2832
662	Their main purpose is to integrate other event mechanisms into libev. This	2833	Their main purpose is to integrate other event mechanisms into libev and
663	could be used, for example, to track variable changes, implement your own	2834	their use is somewhat advanced. They could be used, for example, to track
664	watchers, integrate net-snmp or a coroutine library and lots more.	2835	variable changes, implement your own watchers, integrate net-snmp or a
		2836	coroutine library and lots more. They are also occasionally useful if
		2837	you cache some data and want to flush it before blocking (for example,
		2838	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2839	watcher).
665		2840
666	This is done by examining in each prepare call which file descriptors need	2841	This is done by examining in each prepare call which file descriptors
667	to be watched by the other library, registering ev_io watchers for them	2842	need to be watched by the other library, registering C<ev_io> watchers
668	and starting an ev_timer watcher for any timeouts (many libraries provide	2843	for them and starting an C<ev_timer> watcher for any timeouts (many
669	just this functionality). Then, in the check watcher you check for any	2844	libraries provide exactly this functionality). Then, in the check watcher,
670	events that occured (by making your callbacks set soem flags for example)	2845	you check for any events that occurred (by checking the pending status
671	and call back into the library.	2846	of all watchers and stopping them) and call back into the library. The
		2847	I/O and timer callbacks will never actually be called (but must be valid
		2848	nevertheless, because you never know, you know?).
672		2849
673	As another example, the perl Coro module uses these hooks to integrate	2850	As another example, the Perl Coro module uses these hooks to integrate
674	coroutines into libev programs, by yielding to other active coroutines	2851	coroutines into libev programs, by yielding to other active coroutines
675	during each prepare and only letting the process block if no coroutines	2852	during each prepare and only letting the process block if no coroutines
676	are ready to run.	2853	are ready to run (it's actually more complicated: it only runs coroutines
		2854	with priority higher than or equal to the event loop and one coroutine
		2855	of lower priority, but only once, using idle watchers to keep the event
		2856	loop from blocking if lower-priority coroutines are active, thus mapping
		2857	low-priority coroutines to idle/background tasks).
		2858
		2859	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2860	priority, to ensure that they are being run before any other watchers
		2861	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2862
		2863	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2864	activate ("feed") events into libev. While libev fully supports this, they
		2865	might get executed before other C<ev_check> watchers did their job. As
		2866	C<ev_check> watchers are often used to embed other (non-libev) event
		2867	loops those other event loops might be in an unusable state until their
		2868	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2869	others).
		2870
		2871	=head3 Watcher-Specific Functions and Data Members
677		2872
678	=over 4	2873	=over 4
679		2874
680	=item ev_prepare_init (ev_prepare *, callback)	2875	=item ev_prepare_init (ev_prepare *, callback)
681		2876
682	=item ev_check_init (ev_check *, callback)	2877	=item ev_check_init (ev_check *, callback)
683		2878
684	Initialises and configures the prepare or check watcher - they have no	2879	Initialises and configures the prepare or check watcher - they have no
685	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2880	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
686	macros, but using them is utterly, utterly pointless.	2881	macros, but using them is utterly, utterly, utterly and completely
		2882	pointless.
687		2883
688	=back	2884	=back
689		2885
		2886	=head3 Examples
		2887
		2888	There are a number of principal ways to embed other event loops or modules
		2889	into libev. Here are some ideas on how to include libadns into libev
		2890	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2891	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2892	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2893	Glib event loop).
		2894
		2895	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2896	and in a check watcher, destroy them and call into libadns. What follows
		2897	is pseudo-code only of course. This requires you to either use a low
		2898	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2899	the callbacks for the IO/timeout watchers might not have been called yet.
		2900
		2901	static ev_io iow [nfd];
		2902	static ev_timer tw;
		2903
		2904	static void
		2905	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2906	{
		2907	}
		2908
		2909	// create io watchers for each fd and a timer before blocking
		2910	static void
		2911	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2912	{
		2913	int timeout = 3600000;
		2914	struct pollfd fds [nfd];
		2915	// actual code will need to loop here and realloc etc.
		2916	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2917
		2918	/* the callback is illegal, but won't be called as we stop during check */
		2919	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2920	ev_timer_start (loop, &tw);
		2921
		2922	// create one ev_io per pollfd
		2923	for (int i = 0; i < nfd; ++i)
		2924	{
		2925	ev_io_init (iow + i, io_cb, fds [i].fd,
		2926	((fds [i].events & POLLIN ? EV_READ : 0)
		2927	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2928
		2929	fds [i].revents = 0;
		2930	ev_io_start (loop, iow + i);
		2931	}
		2932	}
		2933
		2934	// stop all watchers after blocking
		2935	static void
		2936	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2937	{
		2938	ev_timer_stop (loop, &tw);
		2939
		2940	for (int i = 0; i < nfd; ++i)
		2941	{
		2942	// set the relevant poll flags
		2943	// could also call adns_processreadable etc. here
		2944	struct pollfd *fd = fds + i;
		2945	int revents = ev_clear_pending (iow + i);
		2946	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2947	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2948
		2949	// now stop the watcher
		2950	ev_io_stop (loop, iow + i);
		2951	}
		2952
		2953	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2954	}
		2955
		2956	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2957	in the prepare watcher and would dispose of the check watcher.
		2958
		2959	Method 3: If the module to be embedded supports explicit event
		2960	notification (libadns does), you can also make use of the actual watcher
		2961	callbacks, and only destroy/create the watchers in the prepare watcher.
		2962
		2963	static void
		2964	timer_cb (EV_P_ ev_timer *w, int revents)
		2965	{
		2966	adns_state ads = (adns_state)w->data;
		2967	update_now (EV_A);
		2968
		2969	adns_processtimeouts (ads, &tv_now);
		2970	}
		2971
		2972	static void
		2973	io_cb (EV_P_ ev_io *w, int revents)
		2974	{
		2975	adns_state ads = (adns_state)w->data;
		2976	update_now (EV_A);
		2977
		2978	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2979	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2980	}
		2981
		2982	// do not ever call adns_afterpoll
		2983
		2984	Method 4: Do not use a prepare or check watcher because the module you
		2985	want to embed is not flexible enough to support it. Instead, you can
		2986	override their poll function. The drawback with this solution is that the
		2987	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2988	this approach, effectively embedding EV as a client into the horrible
		2989	libglib event loop.
		2990
		2991	static gint
		2992	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2993	{
		2994	int got_events = 0;
		2995
		2996	for (n = 0; n < nfds; ++n)
		2997	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2998
		2999	if (timeout >= 0)
		3000	// create/start timer
		3001
		3002	// poll
		3003	ev_run (EV_A_ 0);
		3004
		3005	// stop timer again
		3006	if (timeout >= 0)
		3007	ev_timer_stop (EV_A_ &to);
		3008
		3009	// stop io watchers again - their callbacks should have set
		3010	for (n = 0; n < nfds; ++n)
		3011	ev_io_stop (EV_A_ iow [n]);
		3012
		3013	return got_events;
		3014	}
		3015
		3016
		3017	=head2 C<ev_embed> - when one backend isn't enough...
		3018
		3019	This is a rather advanced watcher type that lets you embed one event loop
		3020	into another (currently only C<ev_io> events are supported in the embedded
		3021	loop, other types of watchers might be handled in a delayed or incorrect
		3022	fashion and must not be used).
		3023
		3024	There are primarily two reasons you would want that: work around bugs and
		3025	prioritise I/O.
		3026
		3027	As an example for a bug workaround, the kqueue backend might only support
		3028	sockets on some platform, so it is unusable as generic backend, but you
		3029	still want to make use of it because you have many sockets and it scales
		3030	so nicely. In this case, you would create a kqueue-based loop and embed
		3031	it into your default loop (which might use e.g. poll). Overall operation
		3032	will be a bit slower because first libev has to call C<poll> and then
		3033	C<kevent>, but at least you can use both mechanisms for what they are
		3034	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		3035
		3036	As for prioritising I/O: under rare circumstances you have the case where
		3037	some fds have to be watched and handled very quickly (with low latency),
		3038	and even priorities and idle watchers might have too much overhead. In
		3039	this case you would put all the high priority stuff in one loop and all
		3040	the rest in a second one, and embed the second one in the first.
		3041
		3042	As long as the watcher is active, the callback will be invoked every
		3043	time there might be events pending in the embedded loop. The callback
		3044	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		3045	sweep and invoke their callbacks (the callback doesn't need to invoke the
		3046	C<ev_embed_sweep> function directly, it could also start an idle watcher
		3047	to give the embedded loop strictly lower priority for example).
		3048
		3049	You can also set the callback to C<0>, in which case the embed watcher
		3050	will automatically execute the embedded loop sweep whenever necessary.
		3051
		3052	Fork detection will be handled transparently while the C<ev_embed> watcher
		3053	is active, i.e., the embedded loop will automatically be forked when the
		3054	embedding loop forks. In other cases, the user is responsible for calling
		3055	C<ev_loop_fork> on the embedded loop.
		3056
		3057	Unfortunately, not all backends are embeddable: only the ones returned by
		3058	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		3059	portable one.
		3060
		3061	So when you want to use this feature you will always have to be prepared
		3062	that you cannot get an embeddable loop. The recommended way to get around
		3063	this is to have a separate variables for your embeddable loop, try to
		3064	create it, and if that fails, use the normal loop for everything.
		3065
		3066	=head3 C<ev_embed> and fork
		3067
		3068	While the C<ev_embed> watcher is running, forks in the embedding loop will
		3069	automatically be applied to the embedded loop as well, so no special
		3070	fork handling is required in that case. When the watcher is not running,
		3071	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		3072	as applicable.
		3073
		3074	=head3 Watcher-Specific Functions and Data Members
		3075
		3076	=over 4
		3077
		3078	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		3079
		3080	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		3081
		3082	Configures the watcher to embed the given loop, which must be
		3083	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		3084	invoked automatically, otherwise it is the responsibility of the callback
		3085	to invoke it (it will continue to be called until the sweep has been done,
		3086	if you do not want that, you need to temporarily stop the embed watcher).
		3087
		3088	=item ev_embed_sweep (loop, ev_embed *)
		3089
		3090	Make a single, non-blocking sweep over the embedded loop. This works
		3091	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
		3092	appropriate way for embedded loops.
		3093
		3094	=item struct ev_loop *other [read-only]
		3095
		3096	The embedded event loop.
		3097
		3098	=back
		3099
		3100	=head3 Examples
		3101
		3102	Example: Try to get an embeddable event loop and embed it into the default
		3103	event loop. If that is not possible, use the default loop. The default
		3104	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		3105	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		3106	used).
		3107
		3108	struct ev_loop *loop_hi = ev_default_init (0);
		3109	struct ev_loop *loop_lo = 0;
		3110	ev_embed embed;
		3111
		3112	// see if there is a chance of getting one that works
		3113	// (remember that a flags value of 0 means autodetection)
		3114	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3115	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3116	: 0;
		3117
		3118	// if we got one, then embed it, otherwise default to loop_hi
		3119	if (loop_lo)
		3120	{
		3121	ev_embed_init (&embed, 0, loop_lo);
		3122	ev_embed_start (loop_hi, &embed);
		3123	}
		3124	else
		3125	loop_lo = loop_hi;
		3126
		3127	Example: Check if kqueue is available but not recommended and create
		3128	a kqueue backend for use with sockets (which usually work with any
		3129	kqueue implementation). Store the kqueue/socket-only event loop in
		3130	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3131
		3132	struct ev_loop *loop = ev_default_init (0);
		3133	struct ev_loop *loop_socket = 0;
		3134	ev_embed embed;
		3135
		3136	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3137	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3138	{
		3139	ev_embed_init (&embed, 0, loop_socket);
		3140	ev_embed_start (loop, &embed);
		3141	}
		3142
		3143	if (!loop_socket)
		3144	loop_socket = loop;
		3145
		3146	// now use loop_socket for all sockets, and loop for everything else
		3147
		3148
		3149	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		3150
		3151	Fork watchers are called when a C<fork ()> was detected (usually because
		3152	whoever is a good citizen cared to tell libev about it by calling
		3153	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		3154	event loop blocks next and before C<ev_check> watchers are being called,
		3155	and only in the child after the fork. If whoever good citizen calling
		3156	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		3157	handlers will be invoked, too, of course.
		3158
		3159	=head3 The special problem of life after fork - how is it possible?
		3160
		3161	Most uses of C<fork()> consist of forking, then some simple calls to set
		3162	up/change the process environment, followed by a call to C<exec()>. This
		3163	sequence should be handled by libev without any problems.
		3164
		3165	This changes when the application actually wants to do event handling
		3166	in the child, or both parent in child, in effect "continuing" after the
		3167	fork.
		3168
		3169	The default mode of operation (for libev, with application help to detect
		3170	forks) is to duplicate all the state in the child, as would be expected
		3171	when I<either> the parent I<or> the child process continues.
		3172
		3173	When both processes want to continue using libev, then this is usually the
		3174	wrong result. In that case, usually one process (typically the parent) is
		3175	supposed to continue with all watchers in place as before, while the other
		3176	process typically wants to start fresh, i.e. without any active watchers.
		3177
		3178	The cleanest and most efficient way to achieve that with libev is to
		3179	simply create a new event loop, which of course will be "empty", and
		3180	use that for new watchers. This has the advantage of not touching more
		3181	memory than necessary, and thus avoiding the copy-on-write, and the
		3182	disadvantage of having to use multiple event loops (which do not support
		3183	signal watchers).
		3184
		3185	When this is not possible, or you want to use the default loop for
		3186	other reasons, then in the process that wants to start "fresh", call
		3187	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3188	Destroying the default loop will "orphan" (not stop) all registered
		3189	watchers, so you have to be careful not to execute code that modifies
		3190	those watchers. Note also that in that case, you have to re-register any
		3191	signal watchers.
		3192
		3193	=head3 Watcher-Specific Functions and Data Members
		3194
		3195	=over 4
		3196
		3197	=item ev_fork_init (ev_fork *, callback)
		3198
		3199	Initialises and configures the fork watcher - it has no parameters of any
		3200	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		3201	really.
		3202
		3203	=back
		3204
		3205
		3206	=head2 C<ev_cleanup> - even the best things end
		3207
		3208	Cleanup watchers are called just before the event loop is being destroyed
		3209	by a call to C<ev_loop_destroy>.
		3210
		3211	While there is no guarantee that the event loop gets destroyed, cleanup
		3212	watchers provide a convenient method to install cleanup hooks for your
		3213	program, worker threads and so on - you just to make sure to destroy the
		3214	loop when you want them to be invoked.
		3215
		3216	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3217	all other watchers, they do not keep a reference to the event loop (which
		3218	makes a lot of sense if you think about it). Like all other watchers, you
		3219	can call libev functions in the callback, except C<ev_cleanup_start>.
		3220
		3221	=head3 Watcher-Specific Functions and Data Members
		3222
		3223	=over 4
		3224
		3225	=item ev_cleanup_init (ev_cleanup *, callback)
		3226
		3227	Initialises and configures the cleanup watcher - it has no parameters of
		3228	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3229	pointless, I assure you.
		3230
		3231	=back
		3232
		3233	Example: Register an atexit handler to destroy the default loop, so any
		3234	cleanup functions are called.
		3235
		3236	static void
		3237	program_exits (void)
		3238	{
		3239	ev_loop_destroy (EV_DEFAULT_UC);
		3240	}
		3241
		3242	...
		3243	atexit (program_exits);
		3244
		3245
		3246	=head2 C<ev_async> - how to wake up an event loop
		3247
		3248	In general, you cannot use an C<ev_run> from multiple threads or other
		3249	asynchronous sources such as signal handlers (as opposed to multiple event
		3250	loops - those are of course safe to use in different threads).
		3251
		3252	Sometimes, however, you need to wake up an event loop you do not control,
		3253	for example because it belongs to another thread. This is what C<ev_async>
		3254	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3255	it by calling C<ev_async_send>, which is thread- and signal safe.
		3256
		3257	This functionality is very similar to C<ev_signal> watchers, as signals,
		3258	too, are asynchronous in nature, and signals, too, will be compressed
		3259	(i.e. the number of callback invocations may be less than the number of
		3260	C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
		3261	of "global async watchers" by using a watcher on an otherwise unused
		3262	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3263	even without knowing which loop owns the signal.
		3264
		3265	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		3266	just the default loop.
		3267
		3268	=head3 Queueing
		3269
		3270	C<ev_async> does not support queueing of data in any way. The reason
		3271	is that the author does not know of a simple (or any) algorithm for a
		3272	multiple-writer-single-reader queue that works in all cases and doesn't
		3273	need elaborate support such as pthreads or unportable memory access
		3274	semantics.
		3275
		3276	That means that if you want to queue data, you have to provide your own
		3277	queue. But at least I can tell you how to implement locking around your
		3278	queue:
		3279
		3280	=over 4
		3281
		3282	=item queueing from a signal handler context
		3283
		3284	To implement race-free queueing, you simply add to the queue in the signal
		3285	handler but you block the signal handler in the watcher callback. Here is
		3286	an example that does that for some fictitious SIGUSR1 handler:
		3287
		3288	static ev_async mysig;
		3289
		3290	static void
		3291	sigusr1_handler (void)
		3292	{
		3293	sometype data;
		3294
		3295	// no locking etc.
		3296	queue_put (data);
		3297	ev_async_send (EV_DEFAULT_ &mysig);
		3298	}
		3299
		3300	static void
		3301	mysig_cb (EV_P_ ev_async *w, int revents)
		3302	{
		3303	sometype data;
		3304	sigset_t block, prev;
		3305
		3306	sigemptyset (&block);
		3307	sigaddset (&block, SIGUSR1);
		3308	sigprocmask (SIG_BLOCK, &block, &prev);
		3309
		3310	while (queue_get (&data))
		3311	process (data);
		3312
		3313	if (sigismember (&prev, SIGUSR1)
		3314	sigprocmask (SIG_UNBLOCK, &block, 0);
		3315	}
		3316
		3317	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3318	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3319	either...).
		3320
		3321	=item queueing from a thread context
		3322
		3323	The strategy for threads is different, as you cannot (easily) block
		3324	threads but you can easily preempt them, so to queue safely you need to
		3325	employ a traditional mutex lock, such as in this pthread example:
		3326
		3327	static ev_async mysig;
		3328	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3329
		3330	static void
		3331	otherthread (void)
		3332	{
		3333	// only need to lock the actual queueing operation
		3334	pthread_mutex_lock (&mymutex);
		3335	queue_put (data);
		3336	pthread_mutex_unlock (&mymutex);
		3337
		3338	ev_async_send (EV_DEFAULT_ &mysig);
		3339	}
		3340
		3341	static void
		3342	mysig_cb (EV_P_ ev_async *w, int revents)
		3343	{
		3344	pthread_mutex_lock (&mymutex);
		3345
		3346	while (queue_get (&data))
		3347	process (data);
		3348
		3349	pthread_mutex_unlock (&mymutex);
		3350	}
		3351
		3352	=back
		3353
		3354
		3355	=head3 Watcher-Specific Functions and Data Members
		3356
		3357	=over 4
		3358
		3359	=item ev_async_init (ev_async *, callback)
		3360
		3361	Initialises and configures the async watcher - it has no parameters of any
		3362	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3363	trust me.
		3364
		3365	=item ev_async_send (loop, ev_async *)
		3366
		3367	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3368	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		3369	C<ev_feed_event>, this call is safe to do from other threads, signal or
		3370	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		3371	section below on what exactly this means).
		3372
		3373	Note that, as with other watchers in libev, multiple events might get
		3374	compressed into a single callback invocation (another way to look at this
		3375	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3376	reset when the event loop detects that).
		3377
		3378	This call incurs the overhead of a system call only once per event loop
		3379	iteration, so while the overhead might be noticeable, it doesn't apply to
		3380	repeated calls to C<ev_async_send> for the same event loop.
		3381
		3382	=item bool = ev_async_pending (ev_async *)
		3383
		3384	Returns a non-zero value when C<ev_async_send> has been called on the
		3385	watcher but the event has not yet been processed (or even noted) by the
		3386	event loop.
		3387
		3388	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3389	the loop iterates next and checks for the watcher to have become active,
		3390	it will reset the flag again. C<ev_async_pending> can be used to very
		3391	quickly check whether invoking the loop might be a good idea.
		3392
		3393	Not that this does I<not> check whether the watcher itself is pending,
		3394	only whether it has been requested to make this watcher pending: there
		3395	is a time window between the event loop checking and resetting the async
		3396	notification, and the callback being invoked.
		3397
		3398	=back
		3399
		3400
690	=head1 OTHER FUNCTIONS	3401	=head1 OTHER FUNCTIONS
691		3402
692	There are some other fucntions of possible interest. Described. Here. Now.	3403	There are some other functions of possible interest. Described. Here. Now.
693		3404
694	=over 4	3405	=over 4
695		3406
696	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3407	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
697		3408
698	This function combines a simple timer and an I/O watcher, calls your	3409	This function combines a simple timer and an I/O watcher, calls your
699	callback on whichever event happens first and automatically stop both	3410	callback on whichever event happens first and automatically stops both
700	watchers. This is useful if you want to wait for a single event on an fd	3411	watchers. This is useful if you want to wait for a single event on an fd
701	or timeout without havign to allocate/configure/start/stop/free one or	3412	or timeout without having to allocate/configure/start/stop/free one or
702	more watchers yourself.	3413	more watchers yourself.
703		3414
704	If C<fd> is less than 0, then no I/O watcher will be started and events is	3415	If C<fd> is less than 0, then no I/O watcher will be started and the
705	ignored. Otherwise, an ev_io watcher for the given C<fd> and C<events> set	3416	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
706	will be craeted and started.	3417	the given C<fd> and C<events> set will be created and started.
707		3418
708	If C<timeout> is less than 0, then no timeout watcher will be	3419	If C<timeout> is less than 0, then no timeout watcher will be
709	started. Otherwise an ev_timer watcher with after = C<timeout> (and repeat	3420	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
710	= 0) will be started.	3421	repeat = 0) will be started. C<0> is a valid timeout.
711		3422
712	The callback has the type C<void (*cb)(int revents, void *arg)> and	3423	The callback has the type C<void (*cb)(int revents, void *arg)> and is
713	gets passed an events set (normally a combination of EV_ERROR, EV_READ,	3424	passed an C<revents> set like normal event callbacks (a combination of
714	EV_WRITE or EV_TIMEOUT) and the C<arg> value passed to C<ev_once>:	3425	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
		3426	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3427	a timeout and an io event at the same time - you probably should give io
		3428	events precedence.
715		3429
		3430	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3431
716	static void stdin_ready (int revents, void *arg)	3432	static void stdin_ready (int revents, void *arg)
717	{	3433	{
718	if (revents & EV_TIMEOUT)
719	/* doh, nothing entered */
720	else if (revents & EV_READ)	3434	if (revents & EV_READ)
721	/* stdin might have data for us, joy! */	3435	/* stdin might have data for us, joy! */;
		3436	else if (revents & EV_TIMER)
		3437	/* doh, nothing entered */;
722	}	3438	}
723		3439
724	ev_once (STDIN_FILENO, EV_READm 10., stdin_ready, 0);	3440	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
725
726	=item ev_feed_event (loop, watcher, int events)
727
728	Feeds the given event set into the event loop, as if the specified event
729	has happened for the specified watcher (which must be a pointer to an
730	initialised but not necessarily active event watcher).
731		3441
732	=item ev_feed_fd_event (loop, int fd, int revents)	3442	=item ev_feed_fd_event (loop, int fd, int revents)
733		3443
734	Feed an event on the given fd, as if a file descriptor backend detected it.	3444	Feed an event on the given fd, as if a file descriptor backend detected
		3445	the given events it.
735		3446
736	=item ev_feed_signal_event (loop, int signum)	3447	=item ev_feed_signal_event (loop, int signum)
737		3448
738	Feed an event as if the given signal occured (loop must be the default loop!).	3449	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
		3450	which is async-safe.
739		3451
740	=back	3452	=back
741		3453
		3454
		3455	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3456
		3457	This section explains some common idioms that are not immediately
		3458	obvious. Note that examples are sprinkled over the whole manual, and this
		3459	section only contains stuff that wouldn't fit anywhere else.
		3460
		3461	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3462
		3463	Often (especially in GUI toolkits) there are places where you have
		3464	I<modal> interaction, which is most easily implemented by recursively
		3465	invoking C<ev_run>.
		3466
		3467	This brings the problem of exiting - a callback might want to finish the
		3468	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3469	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3470	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3471	other combination: In these cases, C<ev_break> will not work alone.
		3472
		3473	The solution is to maintain "break this loop" variable for each C<ev_run>
		3474	invocation, and use a loop around C<ev_run> until the condition is
		3475	triggered, using C<EVRUN_ONCE>:
		3476
		3477	// main loop
		3478	int exit_main_loop = 0;
		3479
		3480	while (!exit_main_loop)
		3481	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3482
		3483	// in a model watcher
		3484	int exit_nested_loop = 0;
		3485
		3486	while (!exit_nested_loop)
		3487	ev_run (EV_A_ EVRUN_ONCE);
		3488
		3489	To exit from any of these loops, just set the corresponding exit variable:
		3490
		3491	// exit modal loop
		3492	exit_nested_loop = 1;
		3493
		3494	// exit main program, after modal loop is finished
		3495	exit_main_loop = 1;
		3496
		3497	// exit both
		3498	exit_main_loop = exit_nested_loop = 1;
		3499
		3500	=head2 THREAD LOCKING EXAMPLE
		3501
		3502	Here is a fictitious example of how to run an event loop in a different
		3503	thread than where callbacks are being invoked and watchers are
		3504	created/added/removed.
		3505
		3506	For a real-world example, see the C<EV::Loop::Async> perl module,
		3507	which uses exactly this technique (which is suited for many high-level
		3508	languages).
		3509
		3510	The example uses a pthread mutex to protect the loop data, a condition
		3511	variable to wait for callback invocations, an async watcher to notify the
		3512	event loop thread and an unspecified mechanism to wake up the main thread.
		3513
		3514	First, you need to associate some data with the event loop:
		3515
		3516	typedef struct {
		3517	mutex_t lock; /* global loop lock */
		3518	ev_async async_w;
		3519	thread_t tid;
		3520	cond_t invoke_cv;
		3521	} userdata;
		3522
		3523	void prepare_loop (EV_P)
		3524	{
		3525	// for simplicity, we use a static userdata struct.
		3526	static userdata u;
		3527
		3528	ev_async_init (&u->async_w, async_cb);
		3529	ev_async_start (EV_A_ &u->async_w);
		3530
		3531	pthread_mutex_init (&u->lock, 0);
		3532	pthread_cond_init (&u->invoke_cv, 0);
		3533
		3534	// now associate this with the loop
		3535	ev_set_userdata (EV_A_ u);
		3536	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3537	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3538
		3539	// then create the thread running ev_loop
		3540	pthread_create (&u->tid, 0, l_run, EV_A);
		3541	}
		3542
		3543	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3544	solely to wake up the event loop so it takes notice of any new watchers
		3545	that might have been added:
		3546
		3547	static void
		3548	async_cb (EV_P_ ev_async *w, int revents)
		3549	{
		3550	// just used for the side effects
		3551	}
		3552
		3553	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3554	protecting the loop data, respectively.
		3555
		3556	static void
		3557	l_release (EV_P)
		3558	{
		3559	userdata *u = ev_userdata (EV_A);
		3560	pthread_mutex_unlock (&u->lock);
		3561	}
		3562
		3563	static void
		3564	l_acquire (EV_P)
		3565	{
		3566	userdata *u = ev_userdata (EV_A);
		3567	pthread_mutex_lock (&u->lock);
		3568	}
		3569
		3570	The event loop thread first acquires the mutex, and then jumps straight
		3571	into C<ev_run>:
		3572
		3573	void *
		3574	l_run (void *thr_arg)
		3575	{
		3576	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3577
		3578	l_acquire (EV_A);
		3579	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3580	ev_run (EV_A_ 0);
		3581	l_release (EV_A);
		3582
		3583	return 0;
		3584	}
		3585
		3586	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3587	signal the main thread via some unspecified mechanism (signals? pipe
		3588	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3589	have been called (in a while loop because a) spurious wakeups are possible
		3590	and b) skipping inter-thread-communication when there are no pending
		3591	watchers is very beneficial):
		3592
		3593	static void
		3594	l_invoke (EV_P)
		3595	{
		3596	userdata *u = ev_userdata (EV_A);
		3597
		3598	while (ev_pending_count (EV_A))
		3599	{
		3600	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3601	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3602	}
		3603	}
		3604
		3605	Now, whenever the main thread gets told to invoke pending watchers, it
		3606	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3607	thread to continue:
		3608
		3609	static void
		3610	real_invoke_pending (EV_P)
		3611	{
		3612	userdata *u = ev_userdata (EV_A);
		3613
		3614	pthread_mutex_lock (&u->lock);
		3615	ev_invoke_pending (EV_A);
		3616	pthread_cond_signal (&u->invoke_cv);
		3617	pthread_mutex_unlock (&u->lock);
		3618	}
		3619
		3620	Whenever you want to start/stop a watcher or do other modifications to an
		3621	event loop, you will now have to lock:
		3622
		3623	ev_timer timeout_watcher;
		3624	userdata *u = ev_userdata (EV_A);
		3625
		3626	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3627
		3628	pthread_mutex_lock (&u->lock);
		3629	ev_timer_start (EV_A_ &timeout_watcher);
		3630	ev_async_send (EV_A_ &u->async_w);
		3631	pthread_mutex_unlock (&u->lock);
		3632
		3633	Note that sending the C<ev_async> watcher is required because otherwise
		3634	an event loop currently blocking in the kernel will have no knowledge
		3635	about the newly added timer. By waking up the loop it will pick up any new
		3636	watchers in the next event loop iteration.
		3637
		3638
		3639	=head1 LIBEVENT EMULATION
		3640
		3641	Libev offers a compatibility emulation layer for libevent. It cannot
		3642	emulate the internals of libevent, so here are some usage hints:
		3643
		3644	=over 4
		3645
		3646	=item * Only the libevent-1.4.1-beta API is being emulated.
		3647
		3648	This was the newest libevent version available when libev was implemented,
		3649	and is still mostly unchanged in 2010.
		3650
		3651	=item * Use it by including <event.h>, as usual.
		3652
		3653	=item * The following members are fully supported: ev_base, ev_callback,
		3654	ev_arg, ev_fd, ev_res, ev_events.
		3655
		3656	=item * Avoid using ev_flags and the EVLIST_*-macros, while it is
		3657	maintained by libev, it does not work exactly the same way as in libevent (consider
		3658	it a private API).
		3659
		3660	=item * Priorities are not currently supported. Initialising priorities
		3661	will fail and all watchers will have the same priority, even though there
		3662	is an ev_pri field.
		3663
		3664	=item * In libevent, the last base created gets the signals, in libev, the
		3665	base that registered the signal gets the signals.
		3666
		3667	=item * Other members are not supported.
		3668
		3669	=item * The libev emulation is I<not> ABI compatible to libevent, you need
		3670	to use the libev header file and library.
		3671
		3672	=back
		3673
		3674	=head1 C++ SUPPORT
		3675
		3676	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3677	you to use some convenience methods to start/stop watchers and also change
		3678	the callback model to a model using method callbacks on objects.
		3679
		3680	To use it,
		3681
		3682	#include <ev++.h>
		3683
		3684	This automatically includes F<ev.h> and puts all of its definitions (many
		3685	of them macros) into the global namespace. All C++ specific things are
		3686	put into the C<ev> namespace. It should support all the same embedding
		3687	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3688
		3689	Care has been taken to keep the overhead low. The only data member the C++
		3690	classes add (compared to plain C-style watchers) is the event loop pointer
		3691	that the watcher is associated with (or no additional members at all if
		3692	you disable C<EV_MULTIPLICITY> when embedding libev).
		3693
		3694	Currently, functions, static and non-static member functions and classes
		3695	with C<operator ()> can be used as callbacks. Other types should be easy
		3696	to add as long as they only need one additional pointer for context. If
		3697	you need support for other types of functors please contact the author
		3698	(preferably after implementing it).
		3699
		3700	Here is a list of things available in the C<ev> namespace:
		3701
		3702	=over 4
		3703
		3704	=item C<ev::READ>, C<ev::WRITE> etc.
		3705
		3706	These are just enum values with the same values as the C<EV_READ> etc.
		3707	macros from F<ev.h>.
		3708
		3709	=item C<ev::tstamp>, C<ev::now>
		3710
		3711	Aliases to the same types/functions as with the C<ev_> prefix.
		3712
		3713	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3714
		3715	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3716	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3717	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3718	defines by many implementations.
		3719
		3720	All of those classes have these methods:
		3721
		3722	=over 4
		3723
		3724	=item ev::TYPE::TYPE ()
		3725
		3726	=item ev::TYPE::TYPE (loop)
		3727
		3728	=item ev::TYPE::~TYPE
		3729
		3730	The constructor (optionally) takes an event loop to associate the watcher
		3731	with. If it is omitted, it will use C<EV_DEFAULT>.
		3732
		3733	The constructor calls C<ev_init> for you, which means you have to call the
		3734	C<set> method before starting it.
		3735
		3736	It will not set a callback, however: You have to call the templated C<set>
		3737	method to set a callback before you can start the watcher.
		3738
		3739	(The reason why you have to use a method is a limitation in C++ which does
		3740	not allow explicit template arguments for constructors).
		3741
		3742	The destructor automatically stops the watcher if it is active.
		3743
		3744	=item w->set<class, &class::method> (object *)
		3745
		3746	This method sets the callback method to call. The method has to have a
		3747	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3748	first argument and the C<revents> as second. The object must be given as
		3749	parameter and is stored in the C<data> member of the watcher.
		3750
		3751	This method synthesizes efficient thunking code to call your method from
		3752	the C callback that libev requires. If your compiler can inline your
		3753	callback (i.e. it is visible to it at the place of the C<set> call and
		3754	your compiler is good :), then the method will be fully inlined into the
		3755	thunking function, making it as fast as a direct C callback.
		3756
		3757	Example: simple class declaration and watcher initialisation
		3758
		3759	struct myclass
		3760	{
		3761	void io_cb (ev::io &w, int revents) { }
		3762	}
		3763
		3764	myclass obj;
		3765	ev::io iow;
		3766	iow.set <myclass, &myclass::io_cb> (&obj);
		3767
		3768	=item w->set (object *)
		3769
		3770	This is a variation of a method callback - leaving out the method to call
		3771	will default the method to C<operator ()>, which makes it possible to use
		3772	functor objects without having to manually specify the C<operator ()> all
		3773	the time. Incidentally, you can then also leave out the template argument
		3774	list.
		3775
		3776	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3777	int revents)>.
		3778
		3779	See the method-C<set> above for more details.
		3780
		3781	Example: use a functor object as callback.
		3782
		3783	struct myfunctor
		3784	{
		3785	void operator() (ev::io &w, int revents)
		3786	{
		3787	...
		3788	}
		3789	}
		3790
		3791	myfunctor f;
		3792
		3793	ev::io w;
		3794	w.set (&f);
		3795
		3796	=item w->set<function> (void *data = 0)
		3797
		3798	Also sets a callback, but uses a static method or plain function as
		3799	callback. The optional C<data> argument will be stored in the watcher's
		3800	C<data> member and is free for you to use.
		3801
		3802	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3803
		3804	See the method-C<set> above for more details.
		3805
		3806	Example: Use a plain function as callback.
		3807
		3808	static void io_cb (ev::io &w, int revents) { }
		3809	iow.set <io_cb> ();
		3810
		3811	=item w->set (loop)
		3812
		3813	Associates a different C<struct ev_loop> with this watcher. You can only
		3814	do this when the watcher is inactive (and not pending either).
		3815
		3816	=item w->set ([arguments])
		3817
		3818	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this
		3819	method or a suitable start method must be called at least once. Unlike the
		3820	C counterpart, an active watcher gets automatically stopped and restarted
		3821	when reconfiguring it with this method.
		3822
		3823	=item w->start ()
		3824
		3825	Starts the watcher. Note that there is no C<loop> argument, as the
		3826	constructor already stores the event loop.
		3827
		3828	=item w->start ([arguments])
		3829
		3830	Instead of calling C<set> and C<start> methods separately, it is often
		3831	convenient to wrap them in one call. Uses the same type of arguments as
		3832	the configure C<set> method of the watcher.
		3833
		3834	=item w->stop ()
		3835
		3836	Stops the watcher if it is active. Again, no C<loop> argument.
		3837
		3838	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3839
		3840	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3841	C<ev_TYPE_again> function.
		3842
		3843	=item w->sweep () (C<ev::embed> only)
		3844
		3845	Invokes C<ev_embed_sweep>.
		3846
		3847	=item w->update () (C<ev::stat> only)
		3848
		3849	Invokes C<ev_stat_stat>.
		3850
		3851	=back
		3852
		3853	=back
		3854
		3855	Example: Define a class with two I/O and idle watchers, start the I/O
		3856	watchers in the constructor.
		3857
		3858	class myclass
		3859	{
		3860	ev::io io ; void io_cb (ev::io &w, int revents);
		3861	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);
		3862	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3863
		3864	myclass (int fd)
		3865	{
		3866	io .set <myclass, &myclass::io_cb > (this);
		3867	io2 .set <myclass, &myclass::io2_cb > (this);
		3868	idle.set <myclass, &myclass::idle_cb> (this);
		3869
		3870	io.set (fd, ev::WRITE); // configure the watcher
		3871	io.start (); // start it whenever convenient
		3872
		3873	io2.start (fd, ev::READ); // set + start in one call
		3874	}
		3875	};
		3876
		3877
		3878	=head1 OTHER LANGUAGE BINDINGS
		3879
		3880	Libev does not offer other language bindings itself, but bindings for a
		3881	number of languages exist in the form of third-party packages. If you know
		3882	any interesting language binding in addition to the ones listed here, drop
		3883	me a note.
		3884
		3885	=over 4
		3886
		3887	=item Perl
		3888
		3889	The EV module implements the full libev API and is actually used to test
		3890	libev. EV is developed together with libev. Apart from the EV core module,
		3891	there are additional modules that implement libev-compatible interfaces
		3892	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3893	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3894	and C<EV::Glib>).
		3895
		3896	It can be found and installed via CPAN, its homepage is at
		3897	L<http://software.schmorp.de/pkg/EV>.
		3898
		3899	=item Python
		3900
		3901	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3902	seems to be quite complete and well-documented.
		3903
		3904	=item Ruby
		3905
		3906	Tony Arcieri has written a ruby extension that offers access to a subset
		3907	of the libev API and adds file handle abstractions, asynchronous DNS and
		3908	more on top of it. It can be found via gem servers. Its homepage is at
		3909	L<http://rev.rubyforge.org/>.
		3910
		3911	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3912	makes rev work even on mingw.
		3913
		3914	=item Haskell
		3915
		3916	A haskell binding to libev is available at
		3917	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3918
		3919	=item D
		3920
		3921	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3922	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3923
		3924	=item Ocaml
		3925
		3926	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3927	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3928
		3929	=item Lua
		3930
		3931	Brian Maher has written a partial interface to libev for lua (at the
		3932	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		3933	L<http://github.com/brimworks/lua-ev>.
		3934
		3935	=back
		3936
		3937
		3938	=head1 MACRO MAGIC
		3939
		3940	Libev can be compiled with a variety of options, the most fundamental
		3941	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3942	functions and callbacks have an initial C<struct ev_loop *> argument.
		3943
		3944	To make it easier to write programs that cope with either variant, the
		3945	following macros are defined:
		3946
		3947	=over 4
		3948
		3949	=item C<EV_A>, C<EV_A_>
		3950
		3951	This provides the loop I<argument> for functions, if one is required ("ev
		3952	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3953	C<EV_A_> is used when other arguments are following. Example:
		3954
		3955	ev_unref (EV_A);
		3956	ev_timer_add (EV_A_ watcher);
		3957	ev_run (EV_A_ 0);
		3958
		3959	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3960	which is often provided by the following macro.
		3961
		3962	=item C<EV_P>, C<EV_P_>
		3963
		3964	This provides the loop I<parameter> for functions, if one is required ("ev
		3965	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3966	C<EV_P_> is used when other parameters are following. Example:
		3967
		3968	// this is how ev_unref is being declared
		3969	static void ev_unref (EV_P);
		3970
		3971	// this is how you can declare your typical callback
		3972	static void cb (EV_P_ ev_timer *w, int revents)
		3973
		3974	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3975	suitable for use with C<EV_A>.
		3976
		3977	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3978
		3979	Similar to the other two macros, this gives you the value of the default
		3980	loop, if multiple loops are supported ("ev loop default").
		3981
		3982	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3983
		3984	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3985	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3986	is undefined when the default loop has not been initialised by a previous
		3987	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3988
		3989	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3990	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3991
		3992	=back
		3993
		3994	Example: Declare and initialise a check watcher, utilising the above
		3995	macros so it will work regardless of whether multiple loops are supported
		3996	or not.
		3997
		3998	static void
		3999	check_cb (EV_P_ ev_timer *w, int revents)
		4000	{
		4001	ev_check_stop (EV_A_ w);
		4002	}
		4003
		4004	ev_check check;
		4005	ev_check_init (&check, check_cb);
		4006	ev_check_start (EV_DEFAULT_ &check);
		4007	ev_run (EV_DEFAULT_ 0);
		4008
		4009	=head1 EMBEDDING
		4010
		4011	Libev can (and often is) directly embedded into host
		4012	applications. Examples of applications that embed it include the Deliantra
		4013	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		4014	and rxvt-unicode.
		4015
		4016	The goal is to enable you to just copy the necessary files into your
		4017	source directory without having to change even a single line in them, so
		4018	you can easily upgrade by simply copying (or having a checked-out copy of
		4019	libev somewhere in your source tree).
		4020
		4021	=head2 FILESETS
		4022
		4023	Depending on what features you need you need to include one or more sets of files
		4024	in your application.
		4025
		4026	=head3 CORE EVENT LOOP
		4027
		4028	To include only the libev core (all the C<ev_*> functions), with manual
		4029	configuration (no autoconf):
		4030
		4031	#define EV_STANDALONE 1
		4032	#include "ev.c"
		4033
		4034	This will automatically include F<ev.h>, too, and should be done in a
		4035	single C source file only to provide the function implementations. To use
		4036	it, do the same for F<ev.h> in all files wishing to use this API (best
		4037	done by writing a wrapper around F<ev.h> that you can include instead and
		4038	where you can put other configuration options):
		4039
		4040	#define EV_STANDALONE 1
		4041	#include "ev.h"
		4042
		4043	Both header files and implementation files can be compiled with a C++
		4044	compiler (at least, that's a stated goal, and breakage will be treated
		4045	as a bug).
		4046
		4047	You need the following files in your source tree, or in a directory
		4048	in your include path (e.g. in libev/ when using -Ilibev):
		4049
		4050	ev.h
		4051	ev.c
		4052	ev_vars.h
		4053	ev_wrap.h
		4054
		4055	ev_win32.c required on win32 platforms only
		4056
		4057	ev_select.c only when select backend is enabled (which is enabled by default)
		4058	ev_poll.c only when poll backend is enabled (disabled by default)
		4059	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		4060	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		4061	ev_port.c only when the solaris port backend is enabled (disabled by default)
		4062
		4063	F<ev.c> includes the backend files directly when enabled, so you only need
		4064	to compile this single file.
		4065
		4066	=head3 LIBEVENT COMPATIBILITY API
		4067
		4068	To include the libevent compatibility API, also include:
		4069
		4070	#include "event.c"
		4071
		4072	in the file including F<ev.c>, and:
		4073
		4074	#include "event.h"
		4075
		4076	in the files that want to use the libevent API. This also includes F<ev.h>.
		4077
		4078	You need the following additional files for this:
		4079
		4080	event.h
		4081	event.c
		4082
		4083	=head3 AUTOCONF SUPPORT
		4084
		4085	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		4086	whatever way you want, you can also C<m4_include([libev.m4])> in your
		4087	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		4088	include F<config.h> and configure itself accordingly.
		4089
		4090	For this of course you need the m4 file:
		4091
		4092	libev.m4
		4093
		4094	=head2 PREPROCESSOR SYMBOLS/MACROS
		4095
		4096	Libev can be configured via a variety of preprocessor symbols you have to
		4097	define before including (or compiling) any of its files. The default in
		4098	the absence of autoconf is documented for every option.
		4099
		4100	Symbols marked with "(h)" do not change the ABI, and can have different
		4101	values when compiling libev vs. including F<ev.h>, so it is permissible
		4102	to redefine them before including F<ev.h> without breaking compatibility
		4103	to a compiled library. All other symbols change the ABI, which means all
		4104	users of libev and the libev code itself must be compiled with compatible
		4105	settings.
		4106
		4107	=over 4
		4108
		4109	=item EV_COMPAT3 (h)
		4110
		4111	Backwards compatibility is a major concern for libev. This is why this
		4112	release of libev comes with wrappers for the functions and symbols that
		4113	have been renamed between libev version 3 and 4.
		4114
		4115	You can disable these wrappers (to test compatibility with future
		4116	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4117	sources. This has the additional advantage that you can drop the C<struct>
		4118	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4119	typedef in that case.
		4120
		4121	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4122	and in some even more future version the compatibility code will be
		4123	removed completely.
		4124
		4125	=item EV_STANDALONE (h)
		4126
		4127	Must always be C<1> if you do not use autoconf configuration, which
		4128	keeps libev from including F<config.h>, and it also defines dummy
		4129	implementations for some libevent functions (such as logging, which is not
		4130	supported). It will also not define any of the structs usually found in
		4131	F<event.h> that are not directly supported by the libev core alone.
		4132
		4133	In standalone mode, libev will still try to automatically deduce the
		4134	configuration, but has to be more conservative.
		4135
		4136	=item EV_USE_MONOTONIC
		4137
		4138	If defined to be C<1>, libev will try to detect the availability of the
		4139	monotonic clock option at both compile time and runtime. Otherwise no
		4140	use of the monotonic clock option will be attempted. If you enable this,
		4141	you usually have to link against librt or something similar. Enabling it
		4142	when the functionality isn't available is safe, though, although you have
		4143	to make sure you link against any libraries where the C<clock_gettime>
		4144	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		4145
		4146	=item EV_USE_REALTIME
		4147
		4148	If defined to be C<1>, libev will try to detect the availability of the
		4149	real-time clock option at compile time (and assume its availability
		4150	at runtime if successful). Otherwise no use of the real-time clock
		4151	option will be attempted. This effectively replaces C<gettimeofday>
		4152	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		4153	correctness. See the note about libraries in the description of
		4154	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4155	C<EV_USE_CLOCK_SYSCALL>.
		4156
		4157	=item EV_USE_CLOCK_SYSCALL
		4158
		4159	If defined to be C<1>, libev will try to use a direct syscall instead
		4160	of calling the system-provided C<clock_gettime> function. This option
		4161	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4162	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4163	programs needlessly. Using a direct syscall is slightly slower (in
		4164	theory), because no optimised vdso implementation can be used, but avoids
		4165	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4166	higher, as it simplifies linking (no need for C<-lrt>).
		4167
		4168	=item EV_USE_NANOSLEEP
		4169
		4170	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		4171	and will use it for delays. Otherwise it will use C<select ()>.
		4172
		4173	=item EV_USE_EVENTFD
		4174
		4175	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4176	available and will probe for kernel support at runtime. This will improve
		4177	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4178	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4179	2.7 or newer, otherwise disabled.
		4180
		4181	=item EV_USE_SELECT
		4182
		4183	If undefined or defined to be C<1>, libev will compile in support for the
		4184	C<select>(2) backend. No attempt at auto-detection will be done: if no
		4185	other method takes over, select will be it. Otherwise the select backend
		4186	will not be compiled in.
		4187
		4188	=item EV_SELECT_USE_FD_SET
		4189
		4190	If defined to C<1>, then the select backend will use the system C<fd_set>
		4191	structure. This is useful if libev doesn't compile due to a missing
		4192	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		4193	on exotic systems. This usually limits the range of file descriptors to
		4194	some low limit such as 1024 or might have other limitations (winsocket
		4195	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		4196	configures the maximum size of the C<fd_set>.
		4197
		4198	=item EV_SELECT_IS_WINSOCKET
		4199
		4200	When defined to C<1>, the select backend will assume that
		4201	select/socket/connect etc. don't understand file descriptors but
		4202	wants osf handles on win32 (this is the case when the select to
		4203	be used is the winsock select). This means that it will call
		4204	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		4205	it is assumed that all these functions actually work on fds, even
		4206	on win32. Should not be defined on non-win32 platforms.
		4207
		4208	=item EV_FD_TO_WIN32_HANDLE(fd)
		4209
		4210	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		4211	file descriptors to socket handles. When not defining this symbol (the
		4212	default), then libev will call C<_get_osfhandle>, which is usually
		4213	correct. In some cases, programs use their own file descriptor management,
		4214	in which case they can provide this function to map fds to socket handles.
		4215
		4216	=item EV_WIN32_HANDLE_TO_FD(handle)
		4217
		4218	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4219	using the standard C<_open_osfhandle> function. For programs implementing
		4220	their own fd to handle mapping, overwriting this function makes it easier
		4221	to do so. This can be done by defining this macro to an appropriate value.
		4222
		4223	=item EV_WIN32_CLOSE_FD(fd)
		4224
		4225	If programs implement their own fd to handle mapping on win32, then this
		4226	macro can be used to override the C<close> function, useful to unregister
		4227	file descriptors again. Note that the replacement function has to close
		4228	the underlying OS handle.
		4229
		4230	=item EV_USE_POLL
		4231
		4232	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		4233	backend. Otherwise it will be enabled on non-win32 platforms. It
		4234	takes precedence over select.
		4235
		4236	=item EV_USE_EPOLL
		4237
		4238	If defined to be C<1>, libev will compile in support for the Linux
		4239	C<epoll>(7) backend. Its availability will be detected at runtime,
		4240	otherwise another method will be used as fallback. This is the preferred
		4241	backend for GNU/Linux systems. If undefined, it will be enabled if the
		4242	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4243
		4244	=item EV_USE_KQUEUE
		4245
		4246	If defined to be C<1>, libev will compile in support for the BSD style
		4247	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		4248	otherwise another method will be used as fallback. This is the preferred
		4249	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		4250	supports some types of fds correctly (the only platform we found that
		4251	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		4252	not be used unless explicitly requested. The best way to use it is to find
		4253	out whether kqueue supports your type of fd properly and use an embedded
		4254	kqueue loop.
		4255
		4256	=item EV_USE_PORT
		4257
		4258	If defined to be C<1>, libev will compile in support for the Solaris
		4259	10 port style backend. Its availability will be detected at runtime,
		4260	otherwise another method will be used as fallback. This is the preferred
		4261	backend for Solaris 10 systems.
		4262
		4263	=item EV_USE_DEVPOLL
		4264
		4265	Reserved for future expansion, works like the USE symbols above.
		4266
		4267	=item EV_USE_INOTIFY
		4268
		4269	If defined to be C<1>, libev will compile in support for the Linux inotify
		4270	interface to speed up C<ev_stat> watchers. Its actual availability will
		4271	be detected at runtime. If undefined, it will be enabled if the headers
		4272	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4273
		4274	=item EV_ATOMIC_T
		4275
		4276	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		4277	access is atomic with respect to other threads or signal contexts. No such
		4278	type is easily found in the C language, so you can provide your own type
		4279	that you know is safe for your purposes. It is used both for signal handler "locking"
		4280	as well as for signal and thread safety in C<ev_async> watchers.
		4281
		4282	In the absence of this define, libev will use C<sig_atomic_t volatile>
		4283	(from F<signal.h>), which is usually good enough on most platforms.
		4284
		4285	=item EV_H (h)
		4286
		4287	The name of the F<ev.h> header file used to include it. The default if
		4288	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		4289	used to virtually rename the F<ev.h> header file in case of conflicts.
		4290
		4291	=item EV_CONFIG_H (h)
		4292
		4293	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		4294	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		4295	C<EV_H>, above.
		4296
		4297	=item EV_EVENT_H (h)
		4298
		4299	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		4300	of how the F<event.h> header can be found, the default is C<"event.h">.
		4301
		4302	=item EV_PROTOTYPES (h)
		4303
		4304	If defined to be C<0>, then F<ev.h> will not define any function
		4305	prototypes, but still define all the structs and other symbols. This is
		4306	occasionally useful if you want to provide your own wrapper functions
		4307	around libev functions.
		4308
		4309	=item EV_MULTIPLICITY
		4310
		4311	If undefined or defined to C<1>, then all event-loop-specific functions
		4312	will have the C<struct ev_loop *> as first argument, and you can create
		4313	additional independent event loops. Otherwise there will be no support
		4314	for multiple event loops and there is no first event loop pointer
		4315	argument. Instead, all functions act on the single default loop.
		4316
		4317	=item EV_MINPRI
		4318
		4319	=item EV_MAXPRI
		4320
		4321	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		4322	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		4323	provide for more priorities by overriding those symbols (usually defined
		4324	to be C<-2> and C<2>, respectively).
		4325
		4326	When doing priority-based operations, libev usually has to linearly search
		4327	all the priorities, so having many of them (hundreds) uses a lot of space
		4328	and time, so using the defaults of five priorities (-2 .. +2) is usually
		4329	fine.
		4330
		4331	If your embedding application does not need any priorities, defining these
		4332	both to C<0> will save some memory and CPU.
		4333
		4334	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4335	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4336	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
		4337
		4338	If undefined or defined to be C<1> (and the platform supports it), then
		4339	the respective watcher type is supported. If defined to be C<0>, then it
		4340	is not. Disabling watcher types mainly saves code size.
		4341
		4342	=item EV_FEATURES
		4343
		4344	If you need to shave off some kilobytes of code at the expense of some
		4345	speed (but with the full API), you can define this symbol to request
		4346	certain subsets of functionality. The default is to enable all features
		4347	that can be enabled on the platform.
		4348
		4349	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4350	with some broad features you want) and then selectively re-enable
		4351	additional parts you want, for example if you want everything minimal,
		4352	but multiple event loop support, async and child watchers and the poll
		4353	backend, use this:
		4354
		4355	#define EV_FEATURES 0
		4356	#define EV_MULTIPLICITY 1
		4357	#define EV_USE_POLL 1
		4358	#define EV_CHILD_ENABLE 1
		4359	#define EV_ASYNC_ENABLE 1
		4360
		4361	The actual value is a bitset, it can be a combination of the following
		4362	values:
		4363
		4364	=over 4
		4365
		4366	=item C<1> - faster/larger code
		4367
		4368	Use larger code to speed up some operations.
		4369
		4370	Currently this is used to override some inlining decisions (enlarging the
		4371	code size by roughly 30% on amd64).
		4372
		4373	When optimising for size, use of compiler flags such as C<-Os> with
		4374	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4375	assertions.
		4376
		4377	=item C<2> - faster/larger data structures
		4378
		4379	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4380	hash table sizes and so on. This will usually further increase code size
		4381	and can additionally have an effect on the size of data structures at
		4382	runtime.
		4383
		4384	=item C<4> - full API configuration
		4385
		4386	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4387	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4388
		4389	=item C<8> - full API
		4390
		4391	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4392	details on which parts of the API are still available without this
		4393	feature, and do not complain if this subset changes over time.
		4394
		4395	=item C<16> - enable all optional watcher types
		4396
		4397	Enables all optional watcher types. If you want to selectively enable
		4398	only some watcher types other than I/O and timers (e.g. prepare,
		4399	embed, async, child...) you can enable them manually by defining
		4400	C<EV_watchertype_ENABLE> to C<1> instead.
		4401
		4402	=item C<32> - enable all backends
		4403
		4404	This enables all backends - without this feature, you need to enable at
		4405	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4406
		4407	=item C<64> - enable OS-specific "helper" APIs
		4408
		4409	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4410	default.
		4411
		4412	=back
		4413
		4414	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4415	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4416	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4417	watchers, timers and monotonic clock support.
		4418
		4419	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4420	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4421	your program might be left out as well - a binary starting a timer and an
		4422	I/O watcher then might come out at only 5Kb.
		4423
		4424	=item EV_AVOID_STDIO
		4425
		4426	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4427	functions (printf, scanf, perror etc.). This will increase the code size
		4428	somewhat, but if your program doesn't otherwise depend on stdio and your
		4429	libc allows it, this avoids linking in the stdio library which is quite
		4430	big.
		4431
		4432	Note that error messages might become less precise when this option is
		4433	enabled.
		4434
		4435	=item EV_NSIG
		4436
		4437	The highest supported signal number, +1 (or, the number of
		4438	signals): Normally, libev tries to deduce the maximum number of signals
		4439	automatically, but sometimes this fails, in which case it can be
		4440	specified. Also, using a lower number than detected (C<32> should be
		4441	good for about any system in existence) can save some memory, as libev
		4442	statically allocates some 12-24 bytes per signal number.
		4443
		4444	=item EV_PID_HASHSIZE
		4445
		4446	C<ev_child> watchers use a small hash table to distribute workload by
		4447	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
		4448	usually more than enough. If you need to manage thousands of children you
		4449	might want to increase this value (I<must> be a power of two).
		4450
		4451	=item EV_INOTIFY_HASHSIZE
		4452
		4453	C<ev_stat> watchers use a small hash table to distribute workload by
		4454	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
		4455	disabled), usually more than enough. If you need to manage thousands of
		4456	C<ev_stat> watchers you might want to increase this value (I<must> be a
		4457	power of two).
		4458
		4459	=item EV_USE_4HEAP
		4460
		4461	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4462	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4463	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4464	faster performance with many (thousands) of watchers.
		4465
		4466	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4467	will be C<0>.
		4468
		4469	=item EV_HEAP_CACHE_AT
		4470
		4471	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4472	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4473	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4474	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4475	but avoids random read accesses on heap changes. This improves performance
		4476	noticeably with many (hundreds) of watchers.
		4477
		4478	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4479	will be C<0>.
		4480
		4481	=item EV_VERIFY
		4482
		4483	Controls how much internal verification (see C<ev_verify ()>) will
		4484	be done: If set to C<0>, no internal verification code will be compiled
		4485	in. If set to C<1>, then verification code will be compiled in, but not
		4486	called. If set to C<2>, then the internal verification code will be
		4487	called once per loop, which can slow down libev. If set to C<3>, then the
		4488	verification code will be called very frequently, which will slow down
		4489	libev considerably.
		4490
		4491	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4492	will be C<0>.
		4493
		4494	=item EV_COMMON
		4495
		4496	By default, all watchers have a C<void *data> member. By redefining
		4497	this macro to something else you can include more and other types of
		4498	members. You have to define it each time you include one of the files,
		4499	though, and it must be identical each time.
		4500
		4501	For example, the perl EV module uses something like this:
		4502
		4503	#define EV_COMMON \
		4504	SV *self; /* contains this struct */ \
		4505	SV *cb_sv, *fh /* note no trailing ";" */
		4506
		4507	=item EV_CB_DECLARE (type)
		4508
		4509	=item EV_CB_INVOKE (watcher, revents)
		4510
		4511	=item ev_set_cb (ev, cb)
		4512
		4513	Can be used to change the callback member declaration in each watcher,
		4514	and the way callbacks are invoked and set. Must expand to a struct member
		4515	definition and a statement, respectively. See the F<ev.h> header file for
		4516	their default definitions. One possible use for overriding these is to
		4517	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		4518	method calls instead of plain function calls in C++.
		4519
		4520	=back
		4521
		4522	=head2 EXPORTED API SYMBOLS
		4523
		4524	If you need to re-export the API (e.g. via a DLL) and you need a list of
		4525	exported symbols, you can use the provided F<Symbol.*> files which list
		4526	all public symbols, one per line:
		4527
		4528	Symbols.ev for libev proper
		4529	Symbols.event for the libevent emulation
		4530
		4531	This can also be used to rename all public symbols to avoid clashes with
		4532	multiple versions of libev linked together (which is obviously bad in
		4533	itself, but sometimes it is inconvenient to avoid this).
		4534
		4535	A sed command like this will create wrapper C<#define>'s that you need to
		4536	include before including F<ev.h>:
		4537
		4538	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		4539
		4540	This would create a file F<wrap.h> which essentially looks like this:
		4541
		4542	#define ev_backend myprefix_ev_backend
		4543	#define ev_check_start myprefix_ev_check_start
		4544	#define ev_check_stop myprefix_ev_check_stop
		4545	...
		4546
		4547	=head2 EXAMPLES
		4548
		4549	For a real-world example of a program the includes libev
		4550	verbatim, you can have a look at the EV perl module
		4551	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		4552	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		4553	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		4554	will be compiled. It is pretty complex because it provides its own header
		4555	file.
		4556
		4557	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		4558	that everybody includes and which overrides some configure choices:
		4559
		4560	#define EV_FEATURES 8
		4561	#define EV_USE_SELECT 1
		4562	#define EV_PREPARE_ENABLE 1
		4563	#define EV_IDLE_ENABLE 1
		4564	#define EV_SIGNAL_ENABLE 1
		4565	#define EV_CHILD_ENABLE 1
		4566	#define EV_USE_STDEXCEPT 0
		4567	#define EV_CONFIG_H <config.h>
		4568
		4569	#include "ev++.h"
		4570
		4571	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4572
		4573	#include "ev_cpp.h"
		4574	#include "ev.c"
		4575
		4576	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
		4577
		4578	=head2 THREADS AND COROUTINES
		4579
		4580	=head3 THREADS
		4581
		4582	All libev functions are reentrant and thread-safe unless explicitly
		4583	documented otherwise, but libev implements no locking itself. This means
		4584	that you can use as many loops as you want in parallel, as long as there
		4585	are no concurrent calls into any libev function with the same loop
		4586	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4587	of course): libev guarantees that different event loops share no data
		4588	structures that need any locking.
		4589
		4590	Or to put it differently: calls with different loop parameters can be done
		4591	concurrently from multiple threads, calls with the same loop parameter
		4592	must be done serially (but can be done from different threads, as long as
		4593	only one thread ever is inside a call at any point in time, e.g. by using
		4594	a mutex per loop).
		4595
		4596	Specifically to support threads (and signal handlers), libev implements
		4597	so-called C<ev_async> watchers, which allow some limited form of
		4598	concurrency on the same event loop, namely waking it up "from the
		4599	outside".
		4600
		4601	If you want to know which design (one loop, locking, or multiple loops
		4602	without or something else still) is best for your problem, then I cannot
		4603	help you, but here is some generic advice:
		4604
		4605	=over 4
		4606
		4607	=item * most applications have a main thread: use the default libev loop
		4608	in that thread, or create a separate thread running only the default loop.
		4609
		4610	This helps integrating other libraries or software modules that use libev
		4611	themselves and don't care/know about threading.
		4612
		4613	=item * one loop per thread is usually a good model.
		4614
		4615	Doing this is almost never wrong, sometimes a better-performance model
		4616	exists, but it is always a good start.
		4617
		4618	=item * other models exist, such as the leader/follower pattern, where one
		4619	loop is handed through multiple threads in a kind of round-robin fashion.
		4620
		4621	Choosing a model is hard - look around, learn, know that usually you can do
		4622	better than you currently do :-)
		4623
		4624	=item * often you need to talk to some other thread which blocks in the
		4625	event loop.
		4626
		4627	C<ev_async> watchers can be used to wake them up from other threads safely
		4628	(or from signal contexts...).
		4629
		4630	An example use would be to communicate signals or other events that only
		4631	work in the default loop by registering the signal watcher with the
		4632	default loop and triggering an C<ev_async> watcher from the default loop
		4633	watcher callback into the event loop interested in the signal.
		4634
		4635	=back
		4636
		4637	See also L<THREAD LOCKING EXAMPLE>.
		4638
		4639	=head3 COROUTINES
		4640
		4641	Libev is very accommodating to coroutines ("cooperative threads"):
		4642	libev fully supports nesting calls to its functions from different
		4643	coroutines (e.g. you can call C<ev_run> on the same loop from two
		4644	different coroutines, and switch freely between both coroutines running
		4645	the loop, as long as you don't confuse yourself). The only exception is
		4646	that you must not do this from C<ev_periodic> reschedule callbacks.
		4647
		4648	Care has been taken to ensure that libev does not keep local state inside
		4649	C<ev_run>, and other calls do not usually allow for coroutine switches as
		4650	they do not call any callbacks.
		4651
		4652	=head2 COMPILER WARNINGS
		4653
		4654	Depending on your compiler and compiler settings, you might get no or a
		4655	lot of warnings when compiling libev code. Some people are apparently
		4656	scared by this.
		4657
		4658	However, these are unavoidable for many reasons. For one, each compiler
		4659	has different warnings, and each user has different tastes regarding
		4660	warning options. "Warn-free" code therefore cannot be a goal except when
		4661	targeting a specific compiler and compiler-version.
		4662
		4663	Another reason is that some compiler warnings require elaborate
		4664	workarounds, or other changes to the code that make it less clear and less
		4665	maintainable.
		4666
		4667	And of course, some compiler warnings are just plain stupid, or simply
		4668	wrong (because they don't actually warn about the condition their message
		4669	seems to warn about). For example, certain older gcc versions had some
		4670	warnings that resulted in an extreme number of false positives. These have
		4671	been fixed, but some people still insist on making code warn-free with
		4672	such buggy versions.
		4673
		4674	While libev is written to generate as few warnings as possible,
		4675	"warn-free" code is not a goal, and it is recommended not to build libev
		4676	with any compiler warnings enabled unless you are prepared to cope with
		4677	them (e.g. by ignoring them). Remember that warnings are just that:
		4678	warnings, not errors, or proof of bugs.
		4679
		4680
		4681	=head2 VALGRIND
		4682
		4683	Valgrind has a special section here because it is a popular tool that is
		4684	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4685
		4686	If you think you found a bug (memory leak, uninitialised data access etc.)
		4687	in libev, then check twice: If valgrind reports something like:
		4688
		4689	==2274== definitely lost: 0 bytes in 0 blocks.
		4690	==2274== possibly lost: 0 bytes in 0 blocks.
		4691	==2274== still reachable: 256 bytes in 1 blocks.
		4692
		4693	Then there is no memory leak, just as memory accounted to global variables
		4694	is not a memleak - the memory is still being referenced, and didn't leak.
		4695
		4696	Similarly, under some circumstances, valgrind might report kernel bugs
		4697	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4698	although an acceptable workaround has been found here), or it might be
		4699	confused.
		4700
		4701	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4702	make it into some kind of religion.
		4703
		4704	If you are unsure about something, feel free to contact the mailing list
		4705	with the full valgrind report and an explanation on why you think this
		4706	is a bug in libev (best check the archives, too :). However, don't be
		4707	annoyed when you get a brisk "this is no bug" answer and take the chance
		4708	of learning how to interpret valgrind properly.
		4709
		4710	If you need, for some reason, empty reports from valgrind for your project
		4711	I suggest using suppression lists.
		4712
		4713
		4714	=head1 PORTABILITY NOTES
		4715
		4716	=head2 GNU/LINUX 32 BIT LIMITATIONS
		4717
		4718	GNU/Linux is the only common platform that supports 64 bit file/large file
		4719	interfaces but I<disables> them by default.
		4720
		4721	That means that libev compiled in the default environment doesn't support
		4722	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		4723
		4724	Unfortunately, many programs try to work around this GNU/Linux issue
		4725	by enabling the large file API, which makes them incompatible with the
		4726	standard libev compiled for their system.
		4727
		4728	Likewise, libev cannot enable the large file API itself as this would
		4729	suddenly make it incompatible to the default compile time environment,
		4730	i.e. all programs not using special compile switches.
		4731
		4732	=head2 OS/X AND DARWIN BUGS
		4733
		4734	The whole thing is a bug if you ask me - basically any system interface
		4735	you touch is broken, whether it is locales, poll, kqueue or even the
		4736	OpenGL drivers.
		4737
		4738	=head3 C<kqueue> is buggy
		4739
		4740	The kqueue syscall is broken in all known versions - most versions support
		4741	only sockets, many support pipes.
		4742
		4743	Libev tries to work around this by not using C<kqueue> by default on this
		4744	rotten platform, but of course you can still ask for it when creating a
		4745	loop - embedding a socket-only kqueue loop into a select-based one is
		4746	probably going to work well.
		4747
		4748	=head3 C<poll> is buggy
		4749
		4750	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		4751	implementation by something calling C<kqueue> internally around the 10.5.6
		4752	release, so now C<kqueue> I<and> C<poll> are broken.
		4753
		4754	Libev tries to work around this by not using C<poll> by default on
		4755	this rotten platform, but of course you can still ask for it when creating
		4756	a loop.
		4757
		4758	=head3 C<select> is buggy
		4759
		4760	All that's left is C<select>, and of course Apple found a way to fuck this
		4761	one up as well: On OS/X, C<select> actively limits the number of file
		4762	descriptors you can pass in to 1024 - your program suddenly crashes when
		4763	you use more.
		4764
		4765	There is an undocumented "workaround" for this - defining
		4766	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		4767	work on OS/X.
		4768
		4769	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		4770
		4771	=head3 C<errno> reentrancy
		4772
		4773	The default compile environment on Solaris is unfortunately so
		4774	thread-unsafe that you can't even use components/libraries compiled
		4775	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		4776	defined by default. A valid, if stupid, implementation choice.
		4777
		4778	If you want to use libev in threaded environments you have to make sure
		4779	it's compiled with C<_REENTRANT> defined.
		4780
		4781	=head3 Event port backend
		4782
		4783	The scalable event interface for Solaris is called "event
		4784	ports". Unfortunately, this mechanism is very buggy in all major
		4785	releases. If you run into high CPU usage, your program freezes or you get
		4786	a large number of spurious wakeups, make sure you have all the relevant
		4787	and latest kernel patches applied. No, I don't know which ones, but there
		4788	are multiple ones to apply, and afterwards, event ports actually work
		4789	great.
		4790
		4791	If you can't get it to work, you can try running the program by setting
		4792	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		4793	C<select> backends.
		4794
		4795	=head2 AIX POLL BUG
		4796
		4797	AIX unfortunately has a broken C<poll.h> header. Libev works around
		4798	this by trying to avoid the poll backend altogether (i.e. it's not even
		4799	compiled in), which normally isn't a big problem as C<select> works fine
		4800	with large bitsets on AIX, and AIX is dead anyway.
		4801
		4802	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4803
		4804	=head3 General issues
		4805
		4806	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		4807	requires, and its I/O model is fundamentally incompatible with the POSIX
		4808	model. Libev still offers limited functionality on this platform in
		4809	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		4810	descriptors. This only applies when using Win32 natively, not when using
		4811	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		4812	as every compielr comes with a slightly differently broken/incompatible
		4813	environment.
		4814
		4815	Lifting these limitations would basically require the full
		4816	re-implementation of the I/O system. If you are into this kind of thing,
		4817	then note that glib does exactly that for you in a very portable way (note
		4818	also that glib is the slowest event library known to man).
		4819
		4820	There is no supported compilation method available on windows except
		4821	embedding it into other applications.
		4822
		4823	Sensible signal handling is officially unsupported by Microsoft - libev
		4824	tries its best, but under most conditions, signals will simply not work.
		4825
		4826	Not a libev limitation but worth mentioning: windows apparently doesn't
		4827	accept large writes: instead of resulting in a partial write, windows will
		4828	either accept everything or return C<ENOBUFS> if the buffer is too large,
		4829	so make sure you only write small amounts into your sockets (less than a
		4830	megabyte seems safe, but this apparently depends on the amount of memory
		4831	available).
		4832
		4833	Due to the many, low, and arbitrary limits on the win32 platform and
		4834	the abysmal performance of winsockets, using a large number of sockets
		4835	is not recommended (and not reasonable). If your program needs to use
		4836	more than a hundred or so sockets, then likely it needs to use a totally
		4837	different implementation for windows, as libev offers the POSIX readiness
		4838	notification model, which cannot be implemented efficiently on windows
		4839	(due to Microsoft monopoly games).
		4840
		4841	A typical way to use libev under windows is to embed it (see the embedding
		4842	section for details) and use the following F<evwrap.h> header file instead
		4843	of F<ev.h>:
		4844
		4845	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4846	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4847
		4848	#include "ev.h"
		4849
		4850	And compile the following F<evwrap.c> file into your project (make sure
		4851	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4852
		4853	#include "evwrap.h"
		4854	#include "ev.c"
		4855
		4856	=head3 The winsocket C<select> function
		4857
		4858	The winsocket C<select> function doesn't follow POSIX in that it
		4859	requires socket I<handles> and not socket I<file descriptors> (it is
		4860	also extremely buggy). This makes select very inefficient, and also
		4861	requires a mapping from file descriptors to socket handles (the Microsoft
		4862	C runtime provides the function C<_open_osfhandle> for this). See the
		4863	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		4864	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		4865
		4866	The configuration for a "naked" win32 using the Microsoft runtime
		4867	libraries and raw winsocket select is:
		4868
		4869	#define EV_USE_SELECT 1
		4870	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		4871
		4872	Note that winsockets handling of fd sets is O(n), so you can easily get a
		4873	complexity in the O(n²) range when using win32.
		4874
		4875	=head3 Limited number of file descriptors
		4876
		4877	Windows has numerous arbitrary (and low) limits on things.
		4878
		4879	Early versions of winsocket's select only supported waiting for a maximum
		4880	of C<64> handles (probably owning to the fact that all windows kernels
		4881	can only wait for C<64> things at the same time internally; Microsoft
		4882	recommends spawning a chain of threads and wait for 63 handles and the
		4883	previous thread in each. Sounds great!).
		4884
		4885	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		4886	to some high number (e.g. C<2048>) before compiling the winsocket select
		4887	call (which might be in libev or elsewhere, for example, perl and many
		4888	other interpreters do their own select emulation on windows).
		4889
		4890	Another limit is the number of file descriptors in the Microsoft runtime
		4891	libraries, which by default is C<64> (there must be a hidden I<64>
		4892	fetish or something like this inside Microsoft). You can increase this
		4893	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		4894	(another arbitrary limit), but is broken in many versions of the Microsoft
		4895	runtime libraries. This might get you to about C<512> or C<2048> sockets
		4896	(depending on windows version and/or the phase of the moon). To get more,
		4897	you need to wrap all I/O functions and provide your own fd management, but
		4898	the cost of calling select (O(n²)) will likely make this unworkable.
		4899
		4900	=head2 PORTABILITY REQUIREMENTS
		4901
		4902	In addition to a working ISO-C implementation and of course the
		4903	backend-specific APIs, libev relies on a few additional extensions:
		4904
		4905	=over 4
		4906
		4907	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4908	calling conventions regardless of C<ev_watcher_type *>.
		4909
		4910	Libev assumes not only that all watcher pointers have the same internal
		4911	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4912	assumes that the same (machine) code can be used to call any watcher
		4913	callback: The watcher callbacks have different type signatures, but libev
		4914	calls them using an C<ev_watcher *> internally.
		4915
		4916	=item pointer accesses must be thread-atomic
		4917
		4918	Accessing a pointer value must be atomic, it must both be readable and
		4919	writable in one piece - this is the case on all current architectures.
		4920
		4921	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4922
		4923	The type C<sig_atomic_t volatile> (or whatever is defined as
		4924	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4925	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4926	believed to be sufficiently portable.
		4927
		4928	=item C<sigprocmask> must work in a threaded environment
		4929
		4930	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4931	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		4932	pthread implementations will either allow C<sigprocmask> in the "main
		4933	thread" or will block signals process-wide, both behaviours would
		4934	be compatible with libev. Interaction between C<sigprocmask> and
		4935	C<pthread_sigmask> could complicate things, however.
		4936
		4937	The most portable way to handle signals is to block signals in all threads
		4938	except the initial one, and run the default loop in the initial thread as
		4939	well.
		4940
		4941	=item C<long> must be large enough for common memory allocation sizes
		4942
		4943	To improve portability and simplify its API, libev uses C<long> internally
		4944	instead of C<size_t> when allocating its data structures. On non-POSIX
		4945	systems (Microsoft...) this might be unexpectedly low, but is still at
		4946	least 31 bits everywhere, which is enough for hundreds of millions of
		4947	watchers.
		4948
		4949	=item C<double> must hold a time value in seconds with enough accuracy
		4950
		4951	The type C<double> is used to represent timestamps. It is required to
		4952	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		4953	good enough for at least into the year 4000 with millisecond accuracy
		4954	(the design goal for libev). This requirement is overfulfilled by
		4955	implementations using IEEE 754, which is basically all existing ones. With
		4956	IEEE 754 doubles, you get microsecond accuracy until at least 2200.
		4957
		4958	=back
		4959
		4960	If you know of other additional requirements drop me a note.
		4961
		4962
		4963	=head1 ALGORITHMIC COMPLEXITIES
		4964
		4965	In this section the complexities of (many of) the algorithms used inside
		4966	libev will be documented. For complexity discussions about backends see
		4967	the documentation for C<ev_default_init>.
		4968
		4969	All of the following are about amortised time: If an array needs to be
		4970	extended, libev needs to realloc and move the whole array, but this
		4971	happens asymptotically rarer with higher number of elements, so O(1) might
		4972	mean that libev does a lengthy realloc operation in rare cases, but on
		4973	average it is much faster and asymptotically approaches constant time.
		4974
		4975	=over 4
		4976
		4977	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		4978
		4979	This means that, when you have a watcher that triggers in one hour and
		4980	there are 100 watchers that would trigger before that, then inserting will
		4981	have to skip roughly seven (C<ld 100>) of these watchers.
		4982
		4983	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		4984
		4985	That means that changing a timer costs less than removing/adding them,
		4986	as only the relative motion in the event queue has to be paid for.
		4987
		4988	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		4989
		4990	These just add the watcher into an array or at the head of a list.
		4991
		4992	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		4993
		4994	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		4995
		4996	These watchers are stored in lists, so they need to be walked to find the
		4997	correct watcher to remove. The lists are usually short (you don't usually
		4998	have many watchers waiting for the same fd or signal: one is typical, two
		4999	is rare).
		5000
		5001	=item Finding the next timer in each loop iteration: O(1)
		5002
		5003	By virtue of using a binary or 4-heap, the next timer is always found at a
		5004	fixed position in the storage array.
		5005
		5006	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		5007
		5008	A change means an I/O watcher gets started or stopped, which requires
		5009	libev to recalculate its status (and possibly tell the kernel, depending
		5010	on backend and whether C<ev_io_set> was used).
		5011
		5012	=item Activating one watcher (putting it into the pending state): O(1)
		5013
		5014	=item Priority handling: O(number_of_priorities)
		5015
		5016	Priorities are implemented by allocating some space for each
		5017	priority. When doing priority-based operations, libev usually has to
		5018	linearly search all the priorities, but starting/stopping and activating
		5019	watchers becomes O(1) with respect to priority handling.
		5020
		5021	=item Sending an ev_async: O(1)
		5022
		5023	=item Processing ev_async_send: O(number_of_async_watchers)
		5024
		5025	=item Processing signals: O(max_signal_number)
		5026
		5027	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5028	calls in the current loop iteration. Checking for async and signal events
		5029	involves iterating over all running async watchers or all signal numbers.
		5030
		5031	=back
		5032
		5033
		5034	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5035
		5036	The major version 4 introduced some incompatible changes to the API.
		5037
		5038	At the moment, the C<ev.h> header file provides compatibility definitions
		5039	for all changes, so most programs should still compile. The compatibility
		5040	layer might be removed in later versions of libev, so better update to the
		5041	new API early than late.
		5042
		5043	=over 4
		5044
		5045	=item C<EV_COMPAT3> backwards compatibility mechanism
		5046
		5047	The backward compatibility mechanism can be controlled by
		5048	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
		5049	section.
		5050
		5051	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5052
		5053	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5054
		5055	ev_loop_destroy (EV_DEFAULT_UC);
		5056	ev_loop_fork (EV_DEFAULT);
		5057
		5058	=item function/symbol renames
		5059
		5060	A number of functions and symbols have been renamed:
		5061
		5062	ev_loop => ev_run
		5063	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5064	EVLOOP_ONESHOT => EVRUN_ONCE
		5065
		5066	ev_unloop => ev_break
		5067	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5068	EVUNLOOP_ONE => EVBREAK_ONE
		5069	EVUNLOOP_ALL => EVBREAK_ALL
		5070
		5071	EV_TIMEOUT => EV_TIMER
		5072
		5073	ev_loop_count => ev_iteration
		5074	ev_loop_depth => ev_depth
		5075	ev_loop_verify => ev_verify
		5076
		5077	Most functions working on C<struct ev_loop> objects don't have an
		5078	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5079	associated constants have been renamed to not collide with the C<struct
		5080	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5081	as all other watcher types. Note that C<ev_loop_fork> is still called
		5082	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5083	typedef.
		5084
		5085	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5086
		5087	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5088	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5089	and work, but the library code will of course be larger.
		5090
		5091	=back
		5092
		5093
		5094	=head1 GLOSSARY
		5095
		5096	=over 4
		5097
		5098	=item active
		5099
		5100	A watcher is active as long as it has been started and not yet stopped.
		5101	See L<WATCHER STATES> for details.
		5102
		5103	=item application
		5104
		5105	In this document, an application is whatever is using libev.
		5106
		5107	=item backend
		5108
		5109	The part of the code dealing with the operating system interfaces.
		5110
		5111	=item callback
		5112
		5113	The address of a function that is called when some event has been
		5114	detected. Callbacks are being passed the event loop, the watcher that
		5115	received the event, and the actual event bitset.
		5116
		5117	=item callback/watcher invocation
		5118
		5119	The act of calling the callback associated with a watcher.
		5120
		5121	=item event
		5122
		5123	A change of state of some external event, such as data now being available
		5124	for reading on a file descriptor, time having passed or simply not having
		5125	any other events happening anymore.
		5126
		5127	In libev, events are represented as single bits (such as C<EV_READ> or
		5128	C<EV_TIMER>).
		5129
		5130	=item event library
		5131
		5132	A software package implementing an event model and loop.
		5133
		5134	=item event loop
		5135
		5136	An entity that handles and processes external events and converts them
		5137	into callback invocations.
		5138
		5139	=item event model
		5140
		5141	The model used to describe how an event loop handles and processes
		5142	watchers and events.
		5143
		5144	=item pending
		5145
		5146	A watcher is pending as soon as the corresponding event has been
		5147	detected. See L<WATCHER STATES> for details.
		5148
		5149	=item real time
		5150
		5151	The physical time that is observed. It is apparently strictly monotonic :)
		5152
		5153	=item wall-clock time
		5154
		5155	The time and date as shown on clocks. Unlike real time, it can actually
		5156	be wrong and jump forwards and backwards, e.g. when the you adjust your
		5157	clock.
		5158
		5159	=item watcher
		5160
		5161	A data structure that describes interest in certain events. Watchers need
		5162	to be started (attached to an event loop) before they can receive events.
		5163
		5164	=back
		5165
742	=head1 AUTHOR	5166	=head1 AUTHOR
743		5167
744	Marc Lehmann <libev@schmorp.de>.	5168	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5169	Magnusson and Emanuele Giaquinta.
745		5170

Diff Legend

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