ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.30 by root, Fri Nov 23 04:36:03 2007 UTC vs.
Revision 1.396 by root, Sat Feb 4 17:57:55 2012 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	// break 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 REPRESENTATION	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
51		166
52	=head1 GLOBAL FUNCTIONS	167	=head1 GLOBAL FUNCTIONS
53		168
54	These functions can be called anytime, even before initialising the	169	These functions can be called anytime, even before initialising the
55	library in any way.	170	library in any way.
…		…
58		173
59	=item ev_tstamp ev_time ()	174	=item ev_tstamp ev_time ()
60		175
61	Returns the current time as libev would use it. Please note that the	176	Returns the current time as libev would use it. Please note that the
62	C<ev_now> function is usually faster and also often returns the timestamp	177	C<ev_now> function is usually faster and also often returns the timestamp
63	you actually want to know.	178	you actually want to know. Also interesting is the combination of
		179	C<ev_now_update> 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
		184	until either it is interrupted or the given time interval has
		185	passed (approximately - it might return a bit earlier even if not
		186	interrupted). Returns immediately if C<< interval <= 0 >>.
		187
		188	Basically this is a sub-second-resolution C<sleep ()>.
		189
		190	The range of the C<interval> is limited - libev only guarantees to work
		191	with sleep times of up to one day (C<< interval <= 86400 >>).
64		192
65	=item int ev_version_major ()	193	=item int ev_version_major ()
66		194
67	=item int ev_version_minor ()	195	=item int ev_version_minor ()
68		196
69	You can find out the major and minor version numbers of the library	197	You can find out the major and minor ABI version numbers of the library
70	you linked against by calling the functions C<ev_version_major> and	198	you linked against by calling the functions C<ev_version_major> and
71	C<ev_version_minor>. If you want, you can compare against the global	199	C<ev_version_minor>. If you want, you can compare against the global
72	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	200	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
73	version of the library your program was compiled against.	201	version of the library your program was compiled against.
74		202
		203	These version numbers refer to the ABI version of the library, not the
		204	release version.
		205
75	Usually, it's a good idea to terminate if the major versions mismatch,	206	Usually, it's a good idea to terminate if the major versions mismatch,
76	as this indicates an incompatible change. Minor versions are usually	207	as this indicates an incompatible change. Minor versions are usually
77	compatible to older versions, so a larger minor version alone is usually	208	compatible to older versions, so a larger minor version alone is usually
78	not a problem.	209	not a problem.
79		210
		211	Example: Make sure we haven't accidentally been linked against the wrong
		212	version (note, however, that this will not detect other ABI mismatches,
		213	such as LFS or reentrancy).
		214
		215	assert (("libev version mismatch",
		216	ev_version_major () == EV_VERSION_MAJOR
		217	&& ev_version_minor () >= EV_VERSION_MINOR));
		218
		219	=item unsigned int ev_supported_backends ()
		220
		221	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		222	value) compiled into this binary of libev (independent of their
		223	availability on the system you are running on). See C<ev_default_loop> for
		224	a description of the set values.
		225
		226	Example: make sure we have the epoll method, because yeah this is cool and
		227	a must have and can we have a torrent of it please!!!11
		228
		229	assert (("sorry, no epoll, no sex",
		230	ev_supported_backends () & EVBACKEND_EPOLL));
		231
		232	=item unsigned int ev_recommended_backends ()
		233
		234	Return the set of all backends compiled into this binary of libev and
		235	also recommended for this platform, meaning it will work for most file
		236	descriptor types. This set is often smaller than the one returned by
		237	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
		238	and will not be auto-detected unless you explicitly request it (assuming
		239	you know what you are doing). This is the set of backends that libev will
		240	probe for if you specify no backends explicitly.
		241
		242	=item unsigned int ev_embeddable_backends ()
		243
		244	Returns the set of backends that are embeddable in other event loops. This
		245	value is platform-specific but can include backends not available on the
		246	current system. To find which embeddable backends might be supported on
		247	the current system, you would need to look at C<ev_embeddable_backends ()
		248	& ev_supported_backends ()>, likewise for recommended ones.
		249
		250	See the description of C<ev_embed> watchers for more info.
		251
80	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	252	=item ev_set_allocator (void *(*cb)(void *ptr, long size))
81		253
82	Sets the allocation function to use (the prototype is similar to the	254	Sets the allocation function to use (the prototype is similar - the
83	realloc C function, the semantics are identical). It is used to allocate	255	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
84	and free memory (no surprises here). If it returns zero when memory	256	used to allocate and free memory (no surprises here). If it returns zero
85	needs to be allocated, the library might abort or take some potentially	257	when memory needs to be allocated (C<size != 0>), the library might abort
86	destructive action. The default is your system realloc function.	258	or take some potentially destructive action.
		259
		260	Since some systems (at least OpenBSD and Darwin) fail to implement
		261	correct C<realloc> semantics, libev will use a wrapper around the system
		262	C<realloc> and C<free> functions by default.
87		263
88	You could override this function in high-availability programs to, say,	264	You could override this function in high-availability programs to, say,
89	free some memory if it cannot allocate memory, to use a special allocator,	265	free some memory if it cannot allocate memory, to use a special allocator,
90	or even to sleep a while and retry until some memory is available.	266	or even to sleep a while and retry until some memory is available.
91		267
		268	Example: Replace the libev allocator with one that waits a bit and then
		269	retries (example requires a standards-compliant C<realloc>).
		270
		271	static void *
		272	persistent_realloc (void *ptr, size_t size)
		273	{
		274	for (;;)
		275	{
		276	void *newptr = realloc (ptr, size);
		277
		278	if (newptr)
		279	return newptr;
		280
		281	sleep (60);
		282	}
		283	}
		284
		285	...
		286	ev_set_allocator (persistent_realloc);
		287
92	=item ev_set_syserr_cb (void (*cb)(const char *msg));	288	=item ev_set_syserr_cb (void (*cb)(const char *msg))
93		289
94	Set the callback function to call on a retryable syscall error (such	290	Set the callback function to call on a retryable system call error (such
95	as failed select, poll, epoll_wait). The message is a printable string	291	as failed select, poll, epoll_wait). The message is a printable string
96	indicating the system call or subsystem causing the problem. If this	292	indicating the system call or subsystem causing the problem. If this
97	callback is set, then libev will expect it to remedy the sitution, no	293	callback is set, then libev will expect it to remedy the situation, no
98	matter what, when it returns. That is, libev will generally retry the	294	matter what, when it returns. That is, libev will generally retry the
99	requested operation, or, if the condition doesn't go away, do bad stuff	295	requested operation, or, if the condition doesn't go away, do bad stuff
100	(such as abort).	296	(such as abort).
101		297
		298	Example: This is basically the same thing that libev does internally, too.
		299
		300	static void
		301	fatal_error (const char *msg)
		302	{
		303	perror (msg);
		304	abort ();
		305	}
		306
		307	...
		308	ev_set_syserr_cb (fatal_error);
		309
		310	=item ev_feed_signal (int signum)
		311
		312	This function can be used to "simulate" a signal receive. It is completely
		313	safe to call this function at any time, from any context, including signal
		314	handlers or random threads.
		315
		316	Its main use is to customise signal handling in your process, especially
		317	in the presence of threads. For example, you could block signals
		318	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		319	creating any loops), and in one thread, use C<sigwait> or any other
		320	mechanism to wait for signals, then "deliver" them to libev by calling
		321	C<ev_feed_signal>.
		322
102	=back	323	=back
103		324
104	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	325	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
105		326
106	An event loop is described by a C<struct ev_loop *>. The library knows two	327	An event loop is described by a C<struct ev_loop *> (the C<struct> is
107	types of such loops, the I<default> loop, which supports signals and child	328	I<not> optional in this case unless libev 3 compatibility is disabled, as
108	events, and dynamically created loops which do not.	329	libev 3 had an C<ev_loop> function colliding with the struct name).
109		330
110	If you use threads, a common model is to run the default event loop	331	The library knows two types of such loops, the I<default> loop, which
111	in your main thread (or in a separate thread) and for each thread you	332	supports child process events, and dynamically created event loops which
112	create, you also create another event loop. Libev itself does no locking	333	do not.
113	whatsoever, so if you mix calls to the same event loop in different
114	threads, make sure you lock (this is usually a bad idea, though, even if
115	done correctly, because it's hideous and inefficient).
116		334
117	=over 4	335	=over 4
118		336
119	=item struct ev_loop *ev_default_loop (unsigned int flags)	337	=item struct ev_loop *ev_default_loop (unsigned int flags)
120		338
121	This will initialise the default event loop if it hasn't been initialised	339	This returns the "default" event loop object, which is what you should
122	yet and return it. If the default loop could not be initialised, returns	340	normally use when you just need "the event loop". Event loop objects and
123	false. If it already was initialised it simply returns it (and ignores the	341	the C<flags> parameter are described in more detail in the entry for
124	flags).	342	C<ev_loop_new>.
		343
		344	If the default loop is already initialised then this function simply
		345	returns it (and ignores the flags. If that is troubling you, check
		346	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		347	flags, which should almost always be C<0>, unless the caller is also the
		348	one calling C<ev_run> or otherwise qualifies as "the main program".
125		349
126	If you don't know what event loop to use, use the one returned from this	350	If you don't know what event loop to use, use the one returned from this
127	function.	351	function (or via the C<EV_DEFAULT> macro).
		352
		353	Note that this function is I<not> thread-safe, so if you want to use it
		354	from multiple threads, you have to employ some kind of mutex (note also
		355	that this case is unlikely, as loops cannot be shared easily between
		356	threads anyway).
		357
		358	The default loop is the only loop that can handle C<ev_child> watchers,
		359	and to do this, it always registers a handler for C<SIGCHLD>. If this is
		360	a problem for your application you can either create a dynamic loop with
		361	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
		362	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
		363
		364	Example: This is the most typical usage.
		365
		366	if (!ev_default_loop (0))
		367	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		368
		369	Example: Restrict libev to the select and poll backends, and do not allow
		370	environment settings to be taken into account:
		371
		372	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		373
		374	=item struct ev_loop *ev_loop_new (unsigned int flags)
		375
		376	This will create and initialise a new event loop object. If the loop
		377	could not be initialised, returns false.
		378
		379	This function is thread-safe, and one common way to use libev with
		380	threads is indeed to create one loop per thread, and using the default
		381	loop in the "main" or "initial" thread.
128		382
129	The flags argument can be used to specify special behaviour or specific	383	The flags argument can be used to specify special behaviour or specific
130	backends to use, and is usually specified as 0 (or EVFLAG_AUTO).	384	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
131		385
132	It supports the following flags:	386	The following flags are supported:
133		387
134	=over 4	388	=over 4
135		389
136	=item C<EVFLAG_AUTO>	390	=item C<EVFLAG_AUTO>
137		391
138	The default flags value. Use this if you have no clue (it's the right	392	The default flags value. Use this if you have no clue (it's the right
139	thing, believe me).	393	thing, believe me).
140		394
141	=item C<EVFLAG_NOENV>	395	=item C<EVFLAG_NOENV>
142		396
143	If this flag bit is ored into the flag value (or the program runs setuid	397	If this flag bit is or'ed into the flag value (or the program runs setuid
144	or setgid) then libev will I<not> look at the environment variable	398	or setgid) then libev will I<not> look at the environment variable
145	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	399	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
146	override the flags completely if it is found in the environment. This is	400	override the flags completely if it is found in the environment. This is
147	useful to try out specific backends to test their performance, or to work	401	useful to try out specific backends to test their performance, or to work
148	around bugs.	402	around bugs.
149		403
		404	=item C<EVFLAG_FORKCHECK>
		405
		406	Instead of calling C<ev_loop_fork> manually after a fork, you can also
		407	make libev check for a fork in each iteration by enabling this flag.
		408
		409	This works by calling C<getpid ()> on every iteration of the loop,
		410	and thus this might slow down your event loop if you do a lot of loop
		411	iterations and little real work, but is usually not noticeable (on my
		412	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		413	without a system call and thus I<very> fast, but my GNU/Linux system also has
		414	C<pthread_atfork> which is even faster).
		415
		416	The big advantage of this flag is that you can forget about fork (and
		417	forget about forgetting to tell libev about forking) when you use this
		418	flag.
		419
		420	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		421	environment variable.
		422
		423	=item C<EVFLAG_NOINOTIFY>
		424
		425	When this flag is specified, then libev will not attempt to use the
		426	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
		427	testing, this flag can be useful to conserve inotify file descriptors, as
		428	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		429
		430	=item C<EVFLAG_SIGNALFD>
		431
		432	When this flag is specified, then libev will attempt to use the
		433	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
		434	delivers signals synchronously, which makes it both faster and might make
		435	it possible to get the queued signal data. It can also simplify signal
		436	handling with threads, as long as you properly block signals in your
		437	threads that are not interested in handling them.
		438
		439	Signalfd will not be used by default as this changes your signal mask, and
		440	there are a lot of shoddy libraries and programs (glib's threadpool for
		441	example) that can't properly initialise their signal masks.
		442
		443	=item C<EVFLAG_NOSIGMASK>
		444
		445	When this flag is specified, then libev will avoid to modify the signal
		446	mask. Specifically, this means you have to make sure signals are unblocked
		447	when you want to receive them.
		448
		449	This behaviour is useful when you want to do your own signal handling, or
		450	want to handle signals only in specific threads and want to avoid libev
		451	unblocking the signals.
		452
		453	It's also required by POSIX in a threaded program, as libev calls
		454	C<sigprocmask>, whose behaviour is officially unspecified.
		455
		456	This flag's behaviour will become the default in future versions of libev.
		457
150	=item C<EVMETHOD_SELECT> (value 1, portable select backend)	458	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
151		459
152	This is your standard select(2) backend. Not I<completely> standard, as	460	This is your standard select(2) backend. Not I<completely> standard, as
153	libev tries to roll its own fd_set with no limits on the number of fds,	461	libev tries to roll its own fd_set with no limits on the number of fds,
154	but if that fails, expect a fairly low limit on the number of fds when	462	but if that fails, expect a fairly low limit on the number of fds when
155	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	463	using this backend. It doesn't scale too well (O(highest_fd)), but its
156	the fastest backend for a low number of fds.	464	usually the fastest backend for a low number of (low-numbered :) fds.
157		465
		466	To get good performance out of this backend you need a high amount of
		467	parallelism (most of the file descriptors should be busy). If you are
		468	writing a server, you should C<accept ()> in a loop to accept as many
		469	connections as possible during one iteration. You might also want to have
		470	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		471	readiness notifications you get per iteration.
		472
		473	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		474	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		475	C<exceptfds> set on that platform).
		476
158	=item C<EVMETHOD_POLL> (value 2, poll backend, available everywhere except on windows)	477	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
159		478
160	And this is your standard poll(2) backend. It's more complicated than	479	And this is your standard poll(2) backend. It's more complicated
161	select, but handles sparse fds better and has no artificial limit on the	480	than select, but handles sparse fds better and has no artificial
162	number of fds you can use (except it will slow down considerably with a	481	limit on the number of fds you can use (except it will slow down
163	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	482	considerably with a lot of inactive fds). It scales similarly to select,
		483	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		484	performance tips.
164		485
		486	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		487	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		488
165	=item C<EVMETHOD_EPOLL> (value 4, Linux)	489	=item C<EVBACKEND_EPOLL> (value 4, Linux)
166		490
		491	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		492	kernels).
		493
167	For few fds, this backend is a bit little slower than poll and select,	494	For few fds, this backend is a bit little slower than poll and select, but
168	but it scales phenomenally better. While poll and select usually scale like	495	it scales phenomenally better. While poll and select usually scale like
169	O(total_fds) where n is the total number of fds (or the highest fd), epoll scales	496	O(total_fds) where total_fds is the total number of fds (or the highest
170	either O(1) or O(active_fds).	497	fd), epoll scales either O(1) or O(active_fds).
171		498
		499	The epoll mechanism deserves honorable mention as the most misdesigned
		500	of the more advanced event mechanisms: mere annoyances include silently
		501	dropping file descriptors, requiring a system call per change per file
		502	descriptor (and unnecessary guessing of parameters), problems with dup,
		503	returning before the timeout value, resulting in additional iterations
		504	(and only giving 5ms accuracy while select on the same platform gives
		505	0.1ms) and so on. The biggest issue is fork races, however - if a program
		506	forks then I<both> parent and child process have to recreate the epoll
		507	set, which can take considerable time (one syscall per file descriptor)
		508	and is of course hard to detect.
		509
		510	Epoll is also notoriously buggy - embedding epoll fds I<should> work,
		511	but of course I<doesn't>, and epoll just loves to report events for
		512	totally I<different> file descriptors (even already closed ones, so
		513	one cannot even remove them from the set) than registered in the set
		514	(especially on SMP systems). Libev tries to counter these spurious
		515	notifications by employing an additional generation counter and comparing
		516	that against the events to filter out spurious ones, recreating the set
		517	when required. Epoll also erroneously rounds down timeouts, but gives you
		518	no way to know when and by how much, so sometimes you have to busy-wait
		519	because epoll returns immediately despite a nonzero timeout. And last
		520	not least, it also refuses to work with some file descriptors which work
		521	perfectly fine with C<select> (files, many character devices...).
		522
		523	Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
		524	cobbled together in a hurry, no thought to design or interaction with
		525	others. Oh, the pain, will it ever stop...
		526
172	While stopping and starting an I/O watcher in the same iteration will	527	While stopping, setting and starting an I/O watcher in the same iteration
173	result in some caching, there is still a syscall per such incident	528	will result in some caching, there is still a system call per such
174	(because the fd could point to a different file description now), so its	529	incident (because the same I<file descriptor> could point to a different
175	best to avoid that. Also, dup()ed file descriptors might not work very	530	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
176	well if you register events for both fds.	531	file descriptors might not work very well if you register events for both
		532	file descriptors.
177		533
		534	Best performance from this backend is achieved by not unregistering all
		535	watchers for a file descriptor until it has been closed, if possible,
		536	i.e. keep at least one watcher active per fd at all times. Stopping and
		537	starting a watcher (without re-setting it) also usually doesn't cause
		538	extra overhead. A fork can both result in spurious notifications as well
		539	as in libev having to destroy and recreate the epoll object, which can
		540	take considerable time and thus should be avoided.
		541
		542	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		543	faster than epoll for maybe up to a hundred file descriptors, depending on
		544	the usage. So sad.
		545
		546	While nominally embeddable in other event loops, this feature is broken in
		547	all kernel versions tested so far.
		548
		549	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		550	C<EVBACKEND_POLL>.
		551
178	=item C<EVMETHOD_KQUEUE> (value 8, most BSD clones)	552	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
179		553
180	Kqueue deserves special mention, as at the time of this writing, it	554	Kqueue deserves special mention, as at the time of this writing, it
181	was broken on all BSDs except NetBSD (usually it doesn't work with	555	was broken on all BSDs except NetBSD (usually it doesn't work reliably
182	anything but sockets and pipes, except on Darwin, where of course its	556	with anything but sockets and pipes, except on Darwin, where of course
183	completely useless). For this reason its not being "autodetected" unless	557	it's completely useless). Unlike epoll, however, whose brokenness
184	you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO).	558	is by design, these kqueue bugs can (and eventually will) be fixed
		559	without API changes to existing programs. For this reason it's not being
		560	"auto-detected" unless you explicitly specify it in the flags (i.e. using
		561	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		562	system like NetBSD.
		563
		564	You still can embed kqueue into a normal poll or select backend and use it
		565	only for sockets (after having made sure that sockets work with kqueue on
		566	the target platform). See C<ev_embed> watchers for more info.
185		567
186	It scales in the same way as the epoll backend, but the interface to the	568	It scales in the same way as the epoll backend, but the interface to the
187	kernel is more efficient (which says nothing about its actual speed, of	569	kernel is more efficient (which says nothing about its actual speed, of
188	course). While starting and stopping an I/O watcher does not cause an	570	course). While stopping, setting and starting an I/O watcher does never
189	extra syscall as with epoll, it still adds up to four event changes per	571	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
190	incident, so its best to avoid that.	572	two event changes per incident. Support for C<fork ()> is very bad (but
		573	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		574	cases
191		575
		576	This backend usually performs well under most conditions.
		577
		578	While nominally embeddable in other event loops, this doesn't work
		579	everywhere, so you might need to test for this. And since it is broken
		580	almost everywhere, you should only use it when you have a lot of sockets
		581	(for which it usually works), by embedding it into another event loop
		582	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		583	also broken on OS X)) and, did I mention it, using it only for sockets.
		584
		585	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		586	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		587	C<NOTE_EOF>.
		588
192	=item C<EVMETHOD_DEVPOLL> (value 16, Solaris 8)	589	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
193		590
194	This is not implemented yet (and might never be).	591	This is not implemented yet (and might never be, unless you send me an
		592	implementation). According to reports, C</dev/poll> only supports sockets
		593	and is not embeddable, which would limit the usefulness of this backend
		594	immensely.
195		595
196	=item C<EVMETHOD_PORT> (value 32, Solaris 10)	596	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
197		597
198	This uses the Solaris 10 port mechanism. As with everything on Solaris,	598	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
199	it's really slow, but it still scales very well (O(active_fds)).	599	it's really slow, but it still scales very well (O(active_fds)).
200		600
		601	While this backend scales well, it requires one system call per active
		602	file descriptor per loop iteration. For small and medium numbers of file
		603	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		604	might perform better.
		605
		606	On the positive side, this backend actually performed fully to
		607	specification in all tests and is fully embeddable, which is a rare feat
		608	among the OS-specific backends (I vastly prefer correctness over speed
		609	hacks).
		610
		611	On the negative side, the interface is I<bizarre> - so bizarre that
		612	even sun itself gets it wrong in their code examples: The event polling
		613	function sometimes returns events to the caller even though an error
		614	occurred, but with no indication whether it has done so or not (yes, it's
		615	even documented that way) - deadly for edge-triggered interfaces where you
		616	absolutely have to know whether an event occurred or not because you have
		617	to re-arm the watcher.
		618
		619	Fortunately libev seems to be able to work around these idiocies.
		620
		621	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		622	C<EVBACKEND_POLL>.
		623
201	=item C<EVMETHOD_ALL>	624	=item C<EVBACKEND_ALL>
202		625
203	Try all backends (even potentially broken ones that wouldn't be tried	626	Try all backends (even potentially broken ones that wouldn't be tried
204	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	627	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
205	C<EVMETHOD_ALL & ~EVMETHOD_KQUEUE>.	628	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		629
		630	It is definitely not recommended to use this flag, use whatever
		631	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		632	at all.
		633
		634	=item C<EVBACKEND_MASK>
		635
		636	Not a backend at all, but a mask to select all backend bits from a
		637	C<flags> value, in case you want to mask out any backends from a flags
		638	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
206		639
207	=back	640	=back
208		641
209	If one or more of these are ored into the flags value, then only these	642	If one or more of the backend flags are or'ed into the flags value,
210	backends will be tried (in the reverse order as given here). If none are	643	then only these backends will be tried (in the reverse order as listed
211	specified, most compiled-in backend will be tried, usually in reverse	644	here). If none are specified, all backends in C<ev_recommended_backends
212	order of their flag values :)	645	()> will be tried.
213		646
214	=item struct ev_loop *ev_loop_new (unsigned int flags)	647	Example: Try to create a event loop that uses epoll and nothing else.
215		648
216	Similar to C<ev_default_loop>, but always creates a new event loop that is	649	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
217	always distinct from the default loop. Unlike the default loop, it cannot	650	if (!epoller)
218	handle signal and child watchers, and attempts to do so will be greeted by	651	fatal ("no epoll found here, maybe it hides under your chair");
219	undefined behaviour (or a failed assertion if assertions are enabled).
220		652
221	=item ev_default_destroy ()	653	Example: Use whatever libev has to offer, but make sure that kqueue is
		654	used if available.
222		655
223	Destroys the default loop again (frees all memory and kernel state	656	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
224	etc.). This stops all registered event watchers (by not touching them in
225	any way whatsoever, although you cannot rely on this :).
226		657
227	=item ev_loop_destroy (loop)	658	=item ev_loop_destroy (loop)
228		659
229	Like C<ev_default_destroy>, but destroys an event loop created by an	660	Destroys an event loop object (frees all memory and kernel state
230	earlier call to C<ev_loop_new>.	661	etc.). None of the active event watchers will be stopped in the normal
		662	sense, so e.g. C<ev_is_active> might still return true. It is your
		663	responsibility to either stop all watchers cleanly yourself I<before>
		664	calling this function, or cope with the fact afterwards (which is usually
		665	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		666	for example).
231		667
232	=item ev_default_fork ()	668	Note that certain global state, such as signal state (and installed signal
		669	handlers), will not be freed by this function, and related watchers (such
		670	as signal and child watchers) would need to be stopped manually.
233		671
		672	This function is normally used on loop objects allocated by
		673	C<ev_loop_new>, but it can also be used on the default loop returned by
		674	C<ev_default_loop>, in which case it is not thread-safe.
		675
		676	Note that it is not advisable to call this function on the default loop
		677	except in the rare occasion where you really need to free its resources.
		678	If you need dynamically allocated loops it is better to use C<ev_loop_new>
		679	and C<ev_loop_destroy>.
		680
		681	=item ev_loop_fork (loop)
		682
		683	This function sets a flag that causes subsequent C<ev_run> iterations to
234	This function reinitialises the kernel state for backends that have	684	reinitialise the kernel state for backends that have one. Despite the
235	one. Despite the name, you can call it anytime, but it makes most sense	685	name, you can call it anytime, but it makes most sense after forking, in
236	after forking, in either the parent or child process (or both, but that	686	the child process. You I<must> call it (or use C<EVFLAG_FORKCHECK>) in the
237	again makes little sense).	687	child before resuming or calling C<ev_run>.
238		688
239	You I<must> call this function in the child process after forking if and	689	Again, you I<have> to call it on I<any> loop that you want to re-use after
240	only if you want to use the event library in both processes. If you just	690	a fork, I<even if you do not plan to use the loop in the parent>. This is
241	fork+exec, you don't have to call it.	691	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		692	during fork.
		693
		694	On the other hand, you only need to call this function in the child
		695	process if and only if you want to use the event loop in the child. If
		696	you just fork+exec or create a new loop in the child, you don't have to
		697	call it at all (in fact, C<epoll> is so badly broken that it makes a
		698	difference, but libev will usually detect this case on its own and do a
		699	costly reset of the backend).
242		700
243	The function itself is quite fast and it's usually not a problem to call	701	The function itself is quite fast and it's usually not a problem to call
244	it just in case after a fork. To make this easy, the function will fit in	702	it just in case after a fork.
245	quite nicely into a call to C<pthread_atfork>:
246		703
		704	Example: Automate calling C<ev_loop_fork> on the default loop when
		705	using pthreads.
		706
		707	static void
		708	post_fork_child (void)
		709	{
		710	ev_loop_fork (EV_DEFAULT);
		711	}
		712
		713	...
247	pthread_atfork (0, 0, ev_default_fork);	714	pthread_atfork (0, 0, post_fork_child);
248		715
249	=item ev_loop_fork (loop)	716	=item int ev_is_default_loop (loop)
250		717
251	Like C<ev_default_fork>, but acts on an event loop created by	718	Returns true when the given loop is, in fact, the default loop, and false
252	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	719	otherwise.
253	after fork, and how you do this is entirely your own problem.
254		720
		721	=item unsigned int ev_iteration (loop)
		722
		723	Returns the current iteration count for the event loop, which is identical
		724	to the number of times libev did poll for new events. It starts at C<0>
		725	and happily wraps around with enough iterations.
		726
		727	This value can sometimes be useful as a generation counter of sorts (it
		728	"ticks" the number of loop iterations), as it roughly corresponds with
		729	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		730	prepare and check phases.
		731
255	=item unsigned int ev_method (loop)	732	=item unsigned int ev_depth (loop)
256		733
		734	Returns the number of times C<ev_run> was entered minus the number of
		735	times C<ev_run> was exited normally, in other words, the recursion depth.
		736
		737	Outside C<ev_run>, this number is zero. In a callback, this number is
		738	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		739	in which case it is higher.
		740
		741	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
		742	throwing an exception etc.), doesn't count as "exit" - consider this
		743	as a hint to avoid such ungentleman-like behaviour unless it's really
		744	convenient, in which case it is fully supported.
		745
		746	=item unsigned int ev_backend (loop)
		747
257	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	748	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
258	use.	749	use.
259		750
260	=item ev_tstamp ev_now (loop)	751	=item ev_tstamp ev_now (loop)
261		752
262	Returns the current "event loop time", which is the time the event loop	753	Returns the current "event loop time", which is the time the event loop
263	got events and started processing them. This timestamp does not change	754	received events and started processing them. This timestamp does not
264	as long as callbacks are being processed, and this is also the base time	755	change as long as callbacks are being processed, and this is also the base
265	used for relative timers. You can treat it as the timestamp of the event	756	time used for relative timers. You can treat it as the timestamp of the
266	occuring (or more correctly, the mainloop finding out about it).	757	event occurring (or more correctly, libev finding out about it).
267		758
		759	=item ev_now_update (loop)
		760
		761	Establishes the current time by querying the kernel, updating the time
		762	returned by C<ev_now ()> in the progress. This is a costly operation and
		763	is usually done automatically within C<ev_run ()>.
		764
		765	This function is rarely useful, but when some event callback runs for a
		766	very long time without entering the event loop, updating libev's idea of
		767	the current time is a good idea.
		768
		769	See also L<The special problem of time updates> in the C<ev_timer> section.
		770
		771	=item ev_suspend (loop)
		772
		773	=item ev_resume (loop)
		774
		775	These two functions suspend and resume an event loop, for use when the
		776	loop is not used for a while and timeouts should not be processed.
		777
		778	A typical use case would be an interactive program such as a game: When
		779	the user presses C<^Z> to suspend the game and resumes it an hour later it
		780	would be best to handle timeouts as if no time had actually passed while
		781	the program was suspended. This can be achieved by calling C<ev_suspend>
		782	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		783	C<ev_resume> directly afterwards to resume timer processing.
		784
		785	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		786	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		787	will be rescheduled (that is, they will lose any events that would have
		788	occurred while suspended).
		789
		790	After calling C<ev_suspend> you B<must not> call I<any> function on the
		791	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		792	without a previous call to C<ev_suspend>.
		793
		794	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		795	event loop time (see C<ev_now_update>).
		796
268	=item ev_loop (loop, int flags)	797	=item ev_run (loop, int flags)
269		798
270	Finally, this is it, the event handler. This function usually is called	799	Finally, this is it, the event handler. This function usually is called
271	after you initialised all your watchers and you want to start handling	800	after you have initialised all your watchers and you want to start
272	events.	801	handling events. It will ask the operating system for any new events, call
		802	the watcher callbacks, an then repeat the whole process indefinitely: This
		803	is why event loops are called I<loops>.
273		804
274	If the flags argument is specified as 0, it will not return until either	805	If the flags argument is specified as C<0>, it will keep handling events
275	no event watchers are active anymore or C<ev_unloop> was called.	806	until either no event watchers are active anymore or C<ev_break> was
		807	called.
276		808
		809	Please note that an explicit C<ev_break> is usually better than
		810	relying on all watchers to be stopped when deciding when a program has
		811	finished (especially in interactive programs), but having a program
		812	that automatically loops as long as it has to and no longer by virtue
		813	of relying on its watchers stopping correctly, that is truly a thing of
		814	beauty.
		815
		816	This function is also I<mostly> exception-safe - you can break out of
		817	a C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		818	exception and so on. This does not decrement the C<ev_depth> value, nor
		819	will it clear any outstanding C<EVBREAK_ONE> breaks.
		820
277	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	821	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
278	those events and any outstanding ones, but will not block your process in	822	those events and any already outstanding ones, but will not wait and
279	case there are no events and will return after one iteration of the loop.	823	block your process in case there are no events and will return after one
		824	iteration of the loop. This is sometimes useful to poll and handle new
		825	events while doing lengthy calculations, to keep the program responsive.
280		826
281	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	827	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
282	neccessary) and will handle those and any outstanding ones. It will block	828	necessary) and will handle those and any already outstanding ones. It
283	your process until at least one new event arrives, and will return after	829	will block your process until at least one new event arrives (which could
		830	be an event internal to libev itself, so there is no guarantee that a
		831	user-registered callback will be called), and will return after one
284	one iteration of the loop.	832	iteration of the loop.
285		833
286	This flags value could be used to implement alternative looping	834	This is useful if you are waiting for some external event in conjunction
287	constructs, but the C<prepare> and C<check> watchers provide a better and	835	with something not expressible using other libev watchers (i.e. "roll your
288	more generic mechanism.	836	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		837	usually a better approach for this kind of thing.
289		838
290	Here are the gory details of what ev_loop does:	839	Here are the gory details of what C<ev_run> does (this is for your
		840	understanding, not a guarantee that things will work exactly like this in
		841	future versions):
291		842
292	1. If there are no active watchers (reference count is zero), return.	843	- Increment loop depth.
		844	- Reset the ev_break status.
		845	- Before the first iteration, call any pending watchers.
		846	LOOP:
		847	- If EVFLAG_FORKCHECK was used, check for a fork.
		848	- If a fork was detected (by any means), queue and call all fork watchers.
293	2. Queue and immediately call all prepare watchers.	849	- Queue and call all prepare watchers.
		850	- If ev_break was called, goto FINISH.
294	3. If we have been forked, recreate the kernel state.	851	- If we have been forked, detach and recreate the kernel state
		852	as to not disturb the other process.
295	4. Update the kernel state with all outstanding changes.	853	- Update the kernel state with all outstanding changes.
296	5. Update the "event loop time".	854	- Update the "event loop time" (ev_now ()).
297	6. Calculate for how long to block.	855	- Calculate for how long to sleep or block, if at all
		856	(active idle watchers, EVRUN_NOWAIT or not having
		857	any active watchers at all will result in not sleeping).
		858	- Sleep if the I/O and timer collect interval say so.
		859	- Increment loop iteration counter.
298	7. Block the process, waiting for events.	860	- Block the process, waiting for any events.
		861	- Queue all outstanding I/O (fd) events.
299	8. Update the "event loop time" and do time jump handling.	862	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
300	9. Queue all outstanding timers.	863	- Queue all expired timers.
301	10. Queue all outstanding periodics.	864	- Queue all expired periodics.
302	11. If no events are pending now, queue all idle watchers.	865	- Queue all idle watchers with priority higher than that of pending events.
303	12. Queue all check watchers.	866	- Queue all check watchers.
304	13. Call all queued watchers in reverse order (i.e. check watchers first).	867	- Call all queued watchers in reverse order (i.e. check watchers first).
305	14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	868	Signals and child watchers are implemented as I/O watchers, and will
306	was used, return, otherwise continue with step #1.	869	be handled here by queueing them when their watcher gets executed.
		870	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
		871	were used, or there are no active watchers, goto FINISH, otherwise
		872	continue with step LOOP.
		873	FINISH:
		874	- Reset the ev_break status iff it was EVBREAK_ONE.
		875	- Decrement the loop depth.
		876	- Return.
307		877
		878	Example: Queue some jobs and then loop until no events are outstanding
		879	anymore.
		880
		881	... queue jobs here, make sure they register event watchers as long
		882	... as they still have work to do (even an idle watcher will do..)
		883	ev_run (my_loop, 0);
		884	... jobs done or somebody called break. yeah!
		885
308	=item ev_unloop (loop, how)	886	=item ev_break (loop, how)
309		887
310	Can be used to make a call to C<ev_loop> return early (but only after it	888	Can be used to make a call to C<ev_run> return early (but only after it
311	has processed all outstanding events). The C<how> argument must be either	889	has processed all outstanding events). The C<how> argument must be either
312	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	890	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
313	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	891	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
		892
		893	This "break state" will be cleared on the next call to C<ev_run>.
		894
		895	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		896	which case it will have no effect.
314		897
315	=item ev_ref (loop)	898	=item ev_ref (loop)
316		899
317	=item ev_unref (loop)	900	=item ev_unref (loop)
318		901
319	Ref/unref can be used to add or remove a reference count on the event	902	Ref/unref can be used to add or remove a reference count on the event
320	loop: Every watcher keeps one reference, and as long as the reference	903	loop: Every watcher keeps one reference, and as long as the reference
321	count is nonzero, C<ev_loop> will not return on its own. If you have	904	count is nonzero, C<ev_run> will not return on its own.
322	a watcher you never unregister that should not keep C<ev_loop> from	905
323	returning, ev_unref() after starting, and ev_ref() before stopping it. For	906	This is useful when you have a watcher that you never intend to
		907	unregister, but that nevertheless should not keep C<ev_run> from
		908	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		909	before stopping it.
		910
324	example, libev itself uses this for its internal signal pipe: It is not	911	As an example, libev itself uses this for its internal signal pipe: It
325	visible to the libev user and should not keep C<ev_loop> from exiting if	912	is not visible to the libev user and should not keep C<ev_run> from
326	no event watchers registered by it are active. It is also an excellent	913	exiting if no event watchers registered by it are active. It is also an
327	way to do this for generic recurring timers or from within third-party	914	excellent way to do this for generic recurring timers or from within
328	libraries. Just remember to I<unref after start> and I<ref before stop>.	915	third-party libraries. Just remember to I<unref after start> and I<ref
		916	before stop> (but only if the watcher wasn't active before, or was active
		917	before, respectively. Note also that libev might stop watchers itself
		918	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		919	in the callback).
		920
		921	Example: Create a signal watcher, but keep it from keeping C<ev_run>
		922	running when nothing else is active.
		923
		924	ev_signal exitsig;
		925	ev_signal_init (&exitsig, sig_cb, SIGINT);
		926	ev_signal_start (loop, &exitsig);
		927	ev_unref (loop);
		928
		929	Example: For some weird reason, unregister the above signal handler again.
		930
		931	ev_ref (loop);
		932	ev_signal_stop (loop, &exitsig);
		933
		934	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		935
		936	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		937
		938	These advanced functions influence the time that libev will spend waiting
		939	for events. Both time intervals are by default C<0>, meaning that libev
		940	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		941	latency.
		942
		943	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		944	allows libev to delay invocation of I/O and timer/periodic callbacks
		945	to increase efficiency of loop iterations (or to increase power-saving
		946	opportunities).
		947
		948	The idea is that sometimes your program runs just fast enough to handle
		949	one (or very few) event(s) per loop iteration. While this makes the
		950	program responsive, it also wastes a lot of CPU time to poll for new
		951	events, especially with backends like C<select ()> which have a high
		952	overhead for the actual polling but can deliver many events at once.
		953
		954	By setting a higher I<io collect interval> you allow libev to spend more
		955	time collecting I/O events, so you can handle more events per iteration,
		956	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		957	C<ev_timer>) will not be affected. Setting this to a non-null value will
		958	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		959	sleep time ensures that libev will not poll for I/O events more often then
		960	once per this interval, on average (as long as the host time resolution is
		961	good enough).
		962
		963	Likewise, by setting a higher I<timeout collect interval> you allow libev
		964	to spend more time collecting timeouts, at the expense of increased
		965	latency/jitter/inexactness (the watcher callback will be called
		966	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		967	value will not introduce any overhead in libev.
		968
		969	Many (busy) programs can usually benefit by setting the I/O collect
		970	interval to a value near C<0.1> or so, which is often enough for
		971	interactive servers (of course not for games), likewise for timeouts. It
		972	usually doesn't make much sense to set it to a lower value than C<0.01>,
		973	as this approaches the timing granularity of most systems. Note that if
		974	you do transactions with the outside world and you can't increase the
		975	parallelity, then this setting will limit your transaction rate (if you
		976	need to poll once per transaction and the I/O collect interval is 0.01,
		977	then you can't do more than 100 transactions per second).
		978
		979	Setting the I<timeout collect interval> can improve the opportunity for
		980	saving power, as the program will "bundle" timer callback invocations that
		981	are "near" in time together, by delaying some, thus reducing the number of
		982	times the process sleeps and wakes up again. Another useful technique to
		983	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		984	they fire on, say, one-second boundaries only.
		985
		986	Example: we only need 0.1s timeout granularity, and we wish not to poll
		987	more often than 100 times per second:
		988
		989	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		990	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		991
		992	=item ev_invoke_pending (loop)
		993
		994	This call will simply invoke all pending watchers while resetting their
		995	pending state. Normally, C<ev_run> does this automatically when required,
		996	but when overriding the invoke callback this call comes handy. This
		997	function can be invoked from a watcher - this can be useful for example
		998	when you want to do some lengthy calculation and want to pass further
		999	event handling to another thread (you still have to make sure only one
		1000	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		1001
		1002	=item int ev_pending_count (loop)
		1003
		1004	Returns the number of pending watchers - zero indicates that no watchers
		1005	are pending.
		1006
		1007	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		1008
		1009	This overrides the invoke pending functionality of the loop: Instead of
		1010	invoking all pending watchers when there are any, C<ev_run> will call
		1011	this callback instead. This is useful, for example, when you want to
		1012	invoke the actual watchers inside another context (another thread etc.).
		1013
		1014	If you want to reset the callback, use C<ev_invoke_pending> as new
		1015	callback.
		1016
		1017	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		1018
		1019	Sometimes you want to share the same loop between multiple threads. This
		1020	can be done relatively simply by putting mutex_lock/unlock calls around
		1021	each call to a libev function.
		1022
		1023	However, C<ev_run> can run an indefinite time, so it is not feasible
		1024	to wait for it to return. One way around this is to wake up the event
		1025	loop via C<ev_break> and C<ev_async_send>, another way is to set these
		1026	I<release> and I<acquire> callbacks on the loop.
		1027
		1028	When set, then C<release> will be called just before the thread is
		1029	suspended waiting for new events, and C<acquire> is called just
		1030	afterwards.
		1031
		1032	Ideally, C<release> will just call your mutex_unlock function, and
		1033	C<acquire> will just call the mutex_lock function again.
		1034
		1035	While event loop modifications are allowed between invocations of
		1036	C<release> and C<acquire> (that's their only purpose after all), no
		1037	modifications done will affect the event loop, i.e. adding watchers will
		1038	have no effect on the set of file descriptors being watched, or the time
		1039	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1040	to take note of any changes you made.
		1041
		1042	In theory, threads executing C<ev_run> will be async-cancel safe between
		1043	invocations of C<release> and C<acquire>.
		1044
		1045	See also the locking example in the C<THREADS> section later in this
		1046	document.
		1047
		1048	=item ev_set_userdata (loop, void *data)
		1049
		1050	=item void *ev_userdata (loop)
		1051
		1052	Set and retrieve a single C<void *> associated with a loop. When
		1053	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1054	C<0>.
		1055
		1056	These two functions can be used to associate arbitrary data with a loop,
		1057	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1058	C<acquire> callbacks described above, but of course can be (ab-)used for
		1059	any other purpose as well.
		1060
		1061	=item ev_verify (loop)
		1062
		1063	This function only does something when C<EV_VERIFY> support has been
		1064	compiled in, which is the default for non-minimal builds. It tries to go
		1065	through all internal structures and checks them for validity. If anything
		1066	is found to be inconsistent, it will print an error message to standard
		1067	error and call C<abort ()>.
		1068
		1069	This can be used to catch bugs inside libev itself: under normal
		1070	circumstances, this function will never abort as of course libev keeps its
		1071	data structures consistent.
329		1072
330	=back	1073	=back
331		1074
		1075
332	=head1 ANATOMY OF A WATCHER	1076	=head1 ANATOMY OF A WATCHER
333		1077
		1078	In the following description, uppercase C<TYPE> in names stands for the
		1079	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1080	watchers and C<ev_io_start> for I/O watchers.
		1081
334	A watcher is a structure that you create and register to record your	1082	A watcher is an opaque structure that you allocate and register to record
335	interest in some event. For instance, if you want to wait for STDIN to	1083	your interest in some event. To make a concrete example, imagine you want
336	become readable, you would create an C<ev_io> watcher for that:	1084	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1085	for that:
337		1086
338	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1087	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
339	{	1088	{
340	ev_io_stop (w);	1089	ev_io_stop (w);
341	ev_unloop (loop, EVUNLOOP_ALL);	1090	ev_break (loop, EVBREAK_ALL);
342	}	1091	}
343		1092
344	struct ev_loop *loop = ev_default_loop (0);	1093	struct ev_loop *loop = ev_default_loop (0);
		1094
345	struct ev_io stdin_watcher;	1095	ev_io stdin_watcher;
		1096
346	ev_init (&stdin_watcher, my_cb);	1097	ev_init (&stdin_watcher, my_cb);
347	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1098	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
348	ev_io_start (loop, &stdin_watcher);	1099	ev_io_start (loop, &stdin_watcher);
		1100
349	ev_loop (loop, 0);	1101	ev_run (loop, 0);
350		1102
351	As you can see, you are responsible for allocating the memory for your	1103	As you can see, you are responsible for allocating the memory for your
352	watcher structures (and it is usually a bad idea to do this on the stack,	1104	watcher structures (and it is I<usually> a bad idea to do this on the
353	although this can sometimes be quite valid).	1105	stack).
354		1106
		1107	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1108	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1109
355	Each watcher structure must be initialised by a call to C<ev_init	1110	Each watcher structure must be initialised by a call to C<ev_init (watcher
356	(watcher *, callback)>, which expects a callback to be provided. This	1111	*, callback)>, which expects a callback to be provided. This callback is
357	callback gets invoked each time the event occurs (or, in the case of io	1112	invoked each time the event occurs (or, in the case of I/O watchers, each
358	watchers, each time the event loop detects that the file descriptor given	1113	time the event loop detects that the file descriptor given is readable
359	is readable and/or writable).	1114	and/or writable).
360		1115
361	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1116	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
362	with arguments specific to this watcher type. There is also a macro	1117	macro to configure it, with arguments specific to the watcher type. There
363	to combine initialisation and setting in one call: C<< ev_<type>_init	1118	is also a macro to combine initialisation and setting in one call: C<<
364	(watcher *, callback, ...) >>.	1119	ev_TYPE_init (watcher *, callback, ...) >>.
365		1120
366	To make the watcher actually watch out for events, you have to start it	1121	To make the watcher actually watch out for events, you have to start it
367	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1122	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
368	*) >>), and you can stop watching for events at any time by calling the	1123	*) >>), and you can stop watching for events at any time by calling the
369	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1124	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
370		1125
371	As long as your watcher is active (has been started but not stopped) you	1126	As long as your watcher is active (has been started but not stopped) you
372	must not touch the values stored in it. Most specifically you must never	1127	must not touch the values stored in it. Most specifically you must never
373	reinitialise it or call its set method.	1128	reinitialise it or call its C<ev_TYPE_set> macro.
374
375	You can check whether an event is active by calling the C<ev_is_active
376	(watcher *)> macro. To see whether an event is outstanding (but the
377	callback for it has not been called yet) you can use the C<ev_is_pending
378	(watcher *)> macro.
379		1129
380	Each and every callback receives the event loop pointer as first, the	1130	Each and every callback receives the event loop pointer as first, the
381	registered watcher structure as second, and a bitset of received events as	1131	registered watcher structure as second, and a bitset of received events as
382	third argument.	1132	third argument.
383		1133
…		…
392	=item C<EV_WRITE>	1142	=item C<EV_WRITE>
393		1143
394	The file descriptor in the C<ev_io> watcher has become readable and/or	1144	The file descriptor in the C<ev_io> watcher has become readable and/or
395	writable.	1145	writable.
396		1146
397	=item C<EV_TIMEOUT>	1147	=item C<EV_TIMER>
398		1148
399	The C<ev_timer> watcher has timed out.	1149	The C<ev_timer> watcher has timed out.
400		1150
401	=item C<EV_PERIODIC>	1151	=item C<EV_PERIODIC>
402		1152
…		…
408		1158
409	=item C<EV_CHILD>	1159	=item C<EV_CHILD>
410		1160
411	The pid specified in the C<ev_child> watcher has received a status change.	1161	The pid specified in the C<ev_child> watcher has received a status change.
412		1162
		1163	=item C<EV_STAT>
		1164
		1165	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		1166
413	=item C<EV_IDLE>	1167	=item C<EV_IDLE>
414		1168
415	The C<ev_idle> watcher has determined that you have nothing better to do.	1169	The C<ev_idle> watcher has determined that you have nothing better to do.
416		1170
417	=item C<EV_PREPARE>	1171	=item C<EV_PREPARE>
418		1172
419	=item C<EV_CHECK>	1173	=item C<EV_CHECK>
420		1174
421	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1175	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts
422	to gather new events, and all C<ev_check> watchers are invoked just after	1176	to gather new events, and all C<ev_check> watchers are invoked just after
423	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1177	C<ev_run> has gathered them, but before it invokes any callbacks for any
424	received events. Callbacks of both watcher types can start and stop as	1178	received events. Callbacks of both watcher types can start and stop as
425	many watchers as they want, and all of them will be taken into account	1179	many watchers as they want, and all of them will be taken into account
426	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1180	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
427	C<ev_loop> from blocking).	1181	C<ev_run> from blocking).
		1182
		1183	=item C<EV_EMBED>
		1184
		1185	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1186
		1187	=item C<EV_FORK>
		1188
		1189	The event loop has been resumed in the child process after fork (see
		1190	C<ev_fork>).
		1191
		1192	=item C<EV_CLEANUP>
		1193
		1194	The event loop is about to be destroyed (see C<ev_cleanup>).
		1195
		1196	=item C<EV_ASYNC>
		1197
		1198	The given async watcher has been asynchronously notified (see C<ev_async>).
		1199
		1200	=item C<EV_CUSTOM>
		1201
		1202	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1203	by libev users to signal watchers (e.g. via C<ev_feed_event>).
428		1204
429	=item C<EV_ERROR>	1205	=item C<EV_ERROR>
430		1206
431	An unspecified error has occured, the watcher has been stopped. This might	1207	An unspecified error has occurred, the watcher has been stopped. This might
432	happen because the watcher could not be properly started because libev	1208	happen because the watcher could not be properly started because libev
433	ran out of memory, a file descriptor was found to be closed or any other	1209	ran out of memory, a file descriptor was found to be closed or any other
		1210	problem. Libev considers these application bugs.
		1211
434	problem. You best act on it by reporting the problem and somehow coping	1212	You best act on it by reporting the problem and somehow coping with the
435	with the watcher being stopped.	1213	watcher being stopped. Note that well-written programs should not receive
		1214	an error ever, so when your watcher receives it, this usually indicates a
		1215	bug in your program.
436		1216
437	Libev will usually signal a few "dummy" events together with an error,	1217	Libev will usually signal a few "dummy" events together with an error, for
438	for example it might indicate that a fd is readable or writable, and if	1218	example it might indicate that a fd is readable or writable, and if your
439	your callbacks is well-written it can just attempt the operation and cope	1219	callbacks is well-written it can just attempt the operation and cope with
440	with the error from read() or write(). This will not work in multithreaded	1220	the error from read() or write(). This will not work in multi-threaded
441	programs, though, so beware.	1221	programs, though, as the fd could already be closed and reused for another
		1222	thing, so beware.
442		1223
443	=back	1224	=back
444		1225
445	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1226	=head2 GENERIC WATCHER FUNCTIONS
446		1227
447	Each watcher has, by default, a member C<void *data> that you can change	1228	=over 4
448	and read at any time, libev will completely ignore it. This can be used
449	to associate arbitrary data with your watcher. If you need more data and
450	don't want to allocate memory and store a pointer to it in that data
451	member, you can also "subclass" the watcher type and provide your own
452	data:
453		1229
454	struct my_io	1230	=item C<ev_init> (ev_TYPE *watcher, callback)
		1231
		1232	This macro initialises the generic portion of a watcher. The contents
		1233	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1234	the generic parts of the watcher are initialised, you I<need> to call
		1235	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1236	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1237	which rolls both calls into one.
		1238
		1239	You can reinitialise a watcher at any time as long as it has been stopped
		1240	(or never started) and there are no pending events outstanding.
		1241
		1242	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1243	int revents)>.
		1244
		1245	Example: Initialise an C<ev_io> watcher in two steps.
		1246
		1247	ev_io w;
		1248	ev_init (&w, my_cb);
		1249	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1250
		1251	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1252
		1253	This macro initialises the type-specific parts of a watcher. You need to
		1254	call C<ev_init> at least once before you call this macro, but you can
		1255	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1256	macro on a watcher that is active (it can be pending, however, which is a
		1257	difference to the C<ev_init> macro).
		1258
		1259	Although some watcher types do not have type-specific arguments
		1260	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1261
		1262	See C<ev_init>, above, for an example.
		1263
		1264	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1265
		1266	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1267	calls into a single call. This is the most convenient method to initialise
		1268	a watcher. The same limitations apply, of course.
		1269
		1270	Example: Initialise and set an C<ev_io> watcher in one step.
		1271
		1272	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1273
		1274	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1275
		1276	Starts (activates) the given watcher. Only active watchers will receive
		1277	events. If the watcher is already active nothing will happen.
		1278
		1279	Example: Start the C<ev_io> watcher that is being abused as example in this
		1280	whole section.
		1281
		1282	ev_io_start (EV_DEFAULT_UC, &w);
		1283
		1284	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1285
		1286	Stops the given watcher if active, and clears the pending status (whether
		1287	the watcher was active or not).
		1288
		1289	It is possible that stopped watchers are pending - for example,
		1290	non-repeating timers are being stopped when they become pending - but
		1291	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1292	pending. If you want to free or reuse the memory used by the watcher it is
		1293	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1294
		1295	=item bool ev_is_active (ev_TYPE *watcher)
		1296
		1297	Returns a true value iff the watcher is active (i.e. it has been started
		1298	and not yet been stopped). As long as a watcher is active you must not modify
		1299	it.
		1300
		1301	=item bool ev_is_pending (ev_TYPE *watcher)
		1302
		1303	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1304	events but its callback has not yet been invoked). As long as a watcher
		1305	is pending (but not active) you must not call an init function on it (but
		1306	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1307	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1308	it).
		1309
		1310	=item callback ev_cb (ev_TYPE *watcher)
		1311
		1312	Returns the callback currently set on the watcher.
		1313
		1314	=item ev_cb_set (ev_TYPE *watcher, callback)
		1315
		1316	Change the callback. You can change the callback at virtually any time
		1317	(modulo threads).
		1318
		1319	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1320
		1321	=item int ev_priority (ev_TYPE *watcher)
		1322
		1323	Set and query the priority of the watcher. The priority is a small
		1324	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1325	(default: C<-2>). Pending watchers with higher priority will be invoked
		1326	before watchers with lower priority, but priority will not keep watchers
		1327	from being executed (except for C<ev_idle> watchers).
		1328
		1329	If you need to suppress invocation when higher priority events are pending
		1330	you need to look at C<ev_idle> watchers, which provide this functionality.
		1331
		1332	You I<must not> change the priority of a watcher as long as it is active or
		1333	pending.
		1334
		1335	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1336	fine, as long as you do not mind that the priority value you query might
		1337	or might not have been clamped to the valid range.
		1338
		1339	The default priority used by watchers when no priority has been set is
		1340	always C<0>, which is supposed to not be too high and not be too low :).
		1341
		1342	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1343	priorities.
		1344
		1345	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1346
		1347	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1348	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1349	can deal with that fact, as both are simply passed through to the
		1350	callback.
		1351
		1352	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1353
		1354	If the watcher is pending, this function clears its pending status and
		1355	returns its C<revents> bitset (as if its callback was invoked). If the
		1356	watcher isn't pending it does nothing and returns C<0>.
		1357
		1358	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1359	callback to be invoked, which can be accomplished with this function.
		1360
		1361	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1362
		1363	Feeds the given event set into the event loop, as if the specified event
		1364	had happened for the specified watcher (which must be a pointer to an
		1365	initialised but not necessarily started event watcher). Obviously you must
		1366	not free the watcher as long as it has pending events.
		1367
		1368	Stopping the watcher, letting libev invoke it, or calling
		1369	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1370	not started in the first place.
		1371
		1372	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1373	functions that do not need a watcher.
		1374
		1375	=back
		1376
		1377	See also the L<ASSOCIATING CUSTOM DATA WITH A WATCHER> and L<BUILDING YOUR
		1378	OWN COMPOSITE WATCHERS> idioms.
		1379
		1380	=head2 WATCHER STATES
		1381
		1382	There are various watcher states mentioned throughout this manual -
		1383	active, pending and so on. In this section these states and the rules to
		1384	transition between them will be described in more detail - and while these
		1385	rules might look complicated, they usually do "the right thing".
		1386
		1387	=over 4
		1388
		1389	=item initialiased
		1390
		1391	Before a watcher can be registered with the event loop it has to be
		1392	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1393	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1394
		1395	In this state it is simply some block of memory that is suitable for
		1396	use in an event loop. It can be moved around, freed, reused etc. at
		1397	will - as long as you either keep the memory contents intact, or call
		1398	C<ev_TYPE_init> again.
		1399
		1400	=item started/running/active
		1401
		1402	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1403	property of the event loop, and is actively waiting for events. While in
		1404	this state it cannot be accessed (except in a few documented ways), moved,
		1405	freed or anything else - the only legal thing is to keep a pointer to it,
		1406	and call libev functions on it that are documented to work on active watchers.
		1407
		1408	=item pending
		1409
		1410	If a watcher is active and libev determines that an event it is interested
		1411	in has occurred (such as a timer expiring), it will become pending. It will
		1412	stay in this pending state until either it is stopped or its callback is
		1413	about to be invoked, so it is not normally pending inside the watcher
		1414	callback.
		1415
		1416	The watcher might or might not be active while it is pending (for example,
		1417	an expired non-repeating timer can be pending but no longer active). If it
		1418	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1419	but it is still property of the event loop at this time, so cannot be
		1420	moved, freed or reused. And if it is active the rules described in the
		1421	previous item still apply.
		1422
		1423	It is also possible to feed an event on a watcher that is not active (e.g.
		1424	via C<ev_feed_event>), in which case it becomes pending without being
		1425	active.
		1426
		1427	=item stopped
		1428
		1429	A watcher can be stopped implicitly by libev (in which case it might still
		1430	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1431	latter will clear any pending state the watcher might be in, regardless
		1432	of whether it was active or not, so stopping a watcher explicitly before
		1433	freeing it is often a good idea.
		1434
		1435	While stopped (and not pending) the watcher is essentially in the
		1436	initialised state, that is, it can be reused, moved, modified in any way
		1437	you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
		1438	it again).
		1439
		1440	=back
		1441
		1442	=head2 WATCHER PRIORITY MODELS
		1443
		1444	Many event loops support I<watcher priorities>, which are usually small
		1445	integers that influence the ordering of event callback invocation
		1446	between watchers in some way, all else being equal.
		1447
		1448	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1449	description for the more technical details such as the actual priority
		1450	range.
		1451
		1452	There are two common ways how these these priorities are being interpreted
		1453	by event loops:
		1454
		1455	In the more common lock-out model, higher priorities "lock out" invocation
		1456	of lower priority watchers, which means as long as higher priority
		1457	watchers receive events, lower priority watchers are not being invoked.
		1458
		1459	The less common only-for-ordering model uses priorities solely to order
		1460	callback invocation within a single event loop iteration: Higher priority
		1461	watchers are invoked before lower priority ones, but they all get invoked
		1462	before polling for new events.
		1463
		1464	Libev uses the second (only-for-ordering) model for all its watchers
		1465	except for idle watchers (which use the lock-out model).
		1466
		1467	The rationale behind this is that implementing the lock-out model for
		1468	watchers is not well supported by most kernel interfaces, and most event
		1469	libraries will just poll for the same events again and again as long as
		1470	their callbacks have not been executed, which is very inefficient in the
		1471	common case of one high-priority watcher locking out a mass of lower
		1472	priority ones.
		1473
		1474	Static (ordering) priorities are most useful when you have two or more
		1475	watchers handling the same resource: a typical usage example is having an
		1476	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1477	timeouts. Under load, data might be received while the program handles
		1478	other jobs, but since timers normally get invoked first, the timeout
		1479	handler will be executed before checking for data. In that case, giving
		1480	the timer a lower priority than the I/O watcher ensures that I/O will be
		1481	handled first even under adverse conditions (which is usually, but not
		1482	always, what you want).
		1483
		1484	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1485	will only be executed when no same or higher priority watchers have
		1486	received events, they can be used to implement the "lock-out" model when
		1487	required.
		1488
		1489	For example, to emulate how many other event libraries handle priorities,
		1490	you can associate an C<ev_idle> watcher to each such watcher, and in
		1491	the normal watcher callback, you just start the idle watcher. The real
		1492	processing is done in the idle watcher callback. This causes libev to
		1493	continuously poll and process kernel event data for the watcher, but when
		1494	the lock-out case is known to be rare (which in turn is rare :), this is
		1495	workable.
		1496
		1497	Usually, however, the lock-out model implemented that way will perform
		1498	miserably under the type of load it was designed to handle. In that case,
		1499	it might be preferable to stop the real watcher before starting the
		1500	idle watcher, so the kernel will not have to process the event in case
		1501	the actual processing will be delayed for considerable time.
		1502
		1503	Here is an example of an I/O watcher that should run at a strictly lower
		1504	priority than the default, and which should only process data when no
		1505	other events are pending:
		1506
		1507	ev_idle idle; // actual processing watcher
		1508	ev_io io; // actual event watcher
		1509
		1510	static void
		1511	io_cb (EV_P_ ev_io *w, int revents)
455	{	1512	{
456	struct ev_io io;	1513	// stop the I/O watcher, we received the event, but
457	int otherfd;	1514	// are not yet ready to handle it.
458	void *somedata;	1515	ev_io_stop (EV_A_ w);
459	struct whatever *mostinteresting;	1516
		1517	// start the idle watcher to handle the actual event.
		1518	// it will not be executed as long as other watchers
		1519	// with the default priority are receiving events.
		1520	ev_idle_start (EV_A_ &idle);
460	}	1521	}
461		1522
462	And since your callback will be called with a pointer to the watcher, you	1523	static void
463	can cast it back to your own type:	1524	idle_cb (EV_P_ ev_idle *w, int revents)
464
465	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
466	{	1525	{
467	struct my_io *w = (struct my_io *)w_;	1526	// actual processing
468	...	1527	read (STDIN_FILENO, ...);
		1528
		1529	// have to start the I/O watcher again, as
		1530	// we have handled the event
		1531	ev_io_start (EV_P_ &io);
469	}	1532	}
470		1533
471	More interesting and less C-conformant ways of catsing your callback type	1534	// initialisation
472	have been omitted....	1535	ev_idle_init (&idle, idle_cb);
		1536	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1537	ev_io_start (EV_DEFAULT_ &io);
		1538
		1539	In the "real" world, it might also be beneficial to start a timer, so that
		1540	low-priority connections can not be locked out forever under load. This
		1541	enables your program to keep a lower latency for important connections
		1542	during short periods of high load, while not completely locking out less
		1543	important ones.
473		1544
474		1545
475	=head1 WATCHER TYPES	1546	=head1 WATCHER TYPES
476		1547
477	This section describes each watcher in detail, but will not repeat	1548	This section describes each watcher in detail, but will not repeat
478	information given in the last section.	1549	information given in the last section. Any initialisation/set macros,
		1550	functions and members specific to the watcher type are explained.
479		1551
		1552	Members are additionally marked with either I<[read-only]>, meaning that,
		1553	while the watcher is active, you can look at the member and expect some
		1554	sensible content, but you must not modify it (you can modify it while the
		1555	watcher is stopped to your hearts content), or I<[read-write]>, which
		1556	means you can expect it to have some sensible content while the watcher
		1557	is active, but you can also modify it. Modifying it may not do something
		1558	sensible or take immediate effect (or do anything at all), but libev will
		1559	not crash or malfunction in any way.
		1560
		1561
480	=head2 C<ev_io> - is this file descriptor readable or writable	1562	=head2 C<ev_io> - is this file descriptor readable or writable?
481		1563
482	I/O watchers check whether a file descriptor is readable or writable	1564	I/O watchers check whether a file descriptor is readable or writable
483	in each iteration of the event loop (This behaviour is called	1565	in each iteration of the event loop, or, more precisely, when reading
484	level-triggering because you keep receiving events as long as the	1566	would not block the process and writing would at least be able to write
485	condition persists. Remember you can stop the watcher if you don't want to	1567	some data. This behaviour is called level-triggering because you keep
486	act on the event and neither want to receive future events).	1568	receiving events as long as the condition persists. Remember you can stop
		1569	the watcher if you don't want to act on the event and neither want to
		1570	receive future events.
487		1571
488	In general you can register as many read and/or write event watchers per	1572	In general you can register as many read and/or write event watchers per
489	fd as you want (as long as you don't confuse yourself). Setting all file	1573	fd as you want (as long as you don't confuse yourself). Setting all file
490	descriptors to non-blocking mode is also usually a good idea (but not	1574	descriptors to non-blocking mode is also usually a good idea (but not
491	required if you know what you are doing).	1575	required if you know what you are doing).
492		1576
493	You have to be careful with dup'ed file descriptors, though. Some backends	1577	Another thing you have to watch out for is that it is quite easy to
494	(the linux epoll backend is a notable example) cannot handle dup'ed file	1578	receive "spurious" readiness notifications, that is, your callback might
495	descriptors correctly if you register interest in two or more fds pointing	1579	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
496	to the same underlying file/socket etc. description (that is, they share	1580	because there is no data. It is very easy to get into this situation even
497	the same underlying "file open").	1581	with a relatively standard program structure. Thus it is best to always
		1582	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
		1583	preferable to a program hanging until some data arrives.
498		1584
499	If you must do this, then force the use of a known-to-be-good backend	1585	If you cannot run the fd in non-blocking mode (for example you should
500	(at the time of this writing, this includes only EVMETHOD_SELECT and	1586	not play around with an Xlib connection), then you have to separately
501	EVMETHOD_POLL).	1587	re-test whether a file descriptor is really ready with a known-to-be good
		1588	interface such as poll (fortunately in the case of Xlib, it already does
		1589	this on its own, so its quite safe to use). Some people additionally
		1590	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1591	indefinitely.
		1592
		1593	But really, best use non-blocking mode.
		1594
		1595	=head3 The special problem of disappearing file descriptors
		1596
		1597	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1598	descriptor (either due to calling C<close> explicitly or any other means,
		1599	such as C<dup2>). The reason is that you register interest in some file
		1600	descriptor, but when it goes away, the operating system will silently drop
		1601	this interest. If another file descriptor with the same number then is
		1602	registered with libev, there is no efficient way to see that this is, in
		1603	fact, a different file descriptor.
		1604
		1605	To avoid having to explicitly tell libev about such cases, libev follows
		1606	the following policy: Each time C<ev_io_set> is being called, libev
		1607	will assume that this is potentially a new file descriptor, otherwise
		1608	it is assumed that the file descriptor stays the same. That means that
		1609	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1610	descriptor even if the file descriptor number itself did not change.
		1611
		1612	This is how one would do it normally anyway, the important point is that
		1613	the libev application should not optimise around libev but should leave
		1614	optimisations to libev.
		1615
		1616	=head3 The special problem of dup'ed file descriptors
		1617
		1618	Some backends (e.g. epoll), cannot register events for file descriptors,
		1619	but only events for the underlying file descriptions. That means when you
		1620	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1621	events for them, only one file descriptor might actually receive events.
		1622
		1623	There is no workaround possible except not registering events
		1624	for potentially C<dup ()>'ed file descriptors, or to resort to
		1625	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1626
		1627	=head3 The special problem of files
		1628
		1629	Many people try to use C<select> (or libev) on file descriptors
		1630	representing files, and expect it to become ready when their program
		1631	doesn't block on disk accesses (which can take a long time on their own).
		1632
		1633	However, this cannot ever work in the "expected" way - you get a readiness
		1634	notification as soon as the kernel knows whether and how much data is
		1635	there, and in the case of open files, that's always the case, so you
		1636	always get a readiness notification instantly, and your read (or possibly
		1637	write) will still block on the disk I/O.
		1638
		1639	Another way to view it is that in the case of sockets, pipes, character
		1640	devices and so on, there is another party (the sender) that delivers data
		1641	on its own, but in the case of files, there is no such thing: the disk
		1642	will not send data on its own, simply because it doesn't know what you
		1643	wish to read - you would first have to request some data.
		1644
		1645	Since files are typically not-so-well supported by advanced notification
		1646	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1647	to files, even though you should not use it. The reason for this is
		1648	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1649	usually a tty, often a pipe, but also sometimes files or special devices
		1650	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1651	F</dev/urandom>), and even though the file might better be served with
		1652	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1653	it "just works" instead of freezing.
		1654
		1655	So avoid file descriptors pointing to files when you know it (e.g. use
		1656	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1657	when you rarely read from a file instead of from a socket, and want to
		1658	reuse the same code path.
		1659
		1660	=head3 The special problem of fork
		1661
		1662	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1663	useless behaviour. Libev fully supports fork, but needs to be told about
		1664	it in the child if you want to continue to use it in the child.
		1665
		1666	To support fork in your child processes, you have to call C<ev_loop_fork
		1667	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
		1668	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1669
		1670	=head3 The special problem of SIGPIPE
		1671
		1672	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1673	when writing to a pipe whose other end has been closed, your program gets
		1674	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1675	this is sensible behaviour, for daemons, this is usually undesirable.
		1676
		1677	So when you encounter spurious, unexplained daemon exits, make sure you
		1678	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1679	somewhere, as that would have given you a big clue).
		1680
		1681	=head3 The special problem of accept()ing when you can't
		1682
		1683	Many implementations of the POSIX C<accept> function (for example,
		1684	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1685	connection from the pending queue in all error cases.
		1686
		1687	For example, larger servers often run out of file descriptors (because
		1688	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1689	rejecting the connection, leading to libev signalling readiness on
		1690	the next iteration again (the connection still exists after all), and
		1691	typically causing the program to loop at 100% CPU usage.
		1692
		1693	Unfortunately, the set of errors that cause this issue differs between
		1694	operating systems, there is usually little the app can do to remedy the
		1695	situation, and no known thread-safe method of removing the connection to
		1696	cope with overload is known (to me).
		1697
		1698	One of the easiest ways to handle this situation is to just ignore it
		1699	- when the program encounters an overload, it will just loop until the
		1700	situation is over. While this is a form of busy waiting, no OS offers an
		1701	event-based way to handle this situation, so it's the best one can do.
		1702
		1703	A better way to handle the situation is to log any errors other than
		1704	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1705	messages, and continue as usual, which at least gives the user an idea of
		1706	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1707	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1708	usage.
		1709
		1710	If your program is single-threaded, then you could also keep a dummy file
		1711	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1712	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1713	close that fd, and create a new dummy fd. This will gracefully refuse
		1714	clients under typical overload conditions.
		1715
		1716	The last way to handle it is to simply log the error and C<exit>, as
		1717	is often done with C<malloc> failures, but this results in an easy
		1718	opportunity for a DoS attack.
		1719
		1720	=head3 Watcher-Specific Functions
502		1721
503	=over 4	1722	=over 4
504		1723
505	=item ev_io_init (ev_io *, callback, int fd, int events)	1724	=item ev_io_init (ev_io *, callback, int fd, int events)
506		1725
507	=item ev_io_set (ev_io *, int fd, int events)	1726	=item ev_io_set (ev_io *, int fd, int events)
508		1727
509	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1728	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
510	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1729	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
511	EV_WRITE> to receive the given events.	1730	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1731
		1732	=item int fd [read-only]
		1733
		1734	The file descriptor being watched.
		1735
		1736	=item int events [read-only]
		1737
		1738	The events being watched.
512		1739
513	=back	1740	=back
514		1741
		1742	=head3 Examples
		1743
		1744	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1745	readable, but only once. Since it is likely line-buffered, you could
		1746	attempt to read a whole line in the callback.
		1747
		1748	static void
		1749	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1750	{
		1751	ev_io_stop (loop, w);
		1752	.. read from stdin here (or from w->fd) and handle any I/O errors
		1753	}
		1754
		1755	...
		1756	struct ev_loop *loop = ev_default_init (0);
		1757	ev_io stdin_readable;
		1758	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1759	ev_io_start (loop, &stdin_readable);
		1760	ev_run (loop, 0);
		1761
		1762
515	=head2 C<ev_timer> - relative and optionally recurring timeouts	1763	=head2 C<ev_timer> - relative and optionally repeating timeouts
516		1764
517	Timer watchers are simple relative timers that generate an event after a	1765	Timer watchers are simple relative timers that generate an event after a
518	given time, and optionally repeating in regular intervals after that.	1766	given time, and optionally repeating in regular intervals after that.
519		1767
520	The timers are based on real time, that is, if you register an event that	1768	The timers are based on real time, that is, if you register an event that
521	times out after an hour and you reset your system clock to last years	1769	times out after an hour and you reset your system clock to January last
522	time, it will still time out after (roughly) and hour. "Roughly" because	1770	year, it will still time out after (roughly) one hour. "Roughly" because
523	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1771	detecting time jumps is hard, and some inaccuracies are unavoidable (the
524	monotonic clock option helps a lot here).	1772	monotonic clock option helps a lot here).
		1773
		1774	The callback is guaranteed to be invoked only I<after> its timeout has
		1775	passed (not I<at>, so on systems with very low-resolution clocks this
		1776	might introduce a small delay, see "the special problem of being too
		1777	early", below). If multiple timers become ready during the same loop
		1778	iteration then the ones with earlier time-out values are invoked before
		1779	ones of the same priority with later time-out values (but this is no
		1780	longer true when a callback calls C<ev_run> recursively).
		1781
		1782	=head3 Be smart about timeouts
		1783
		1784	Many real-world problems involve some kind of timeout, usually for error
		1785	recovery. A typical example is an HTTP request - if the other side hangs,
		1786	you want to raise some error after a while.
		1787
		1788	What follows are some ways to handle this problem, from obvious and
		1789	inefficient to smart and efficient.
		1790
		1791	In the following, a 60 second activity timeout is assumed - a timeout that
		1792	gets reset to 60 seconds each time there is activity (e.g. each time some
		1793	data or other life sign was received).
		1794
		1795	=over 4
		1796
		1797	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1798
		1799	This is the most obvious, but not the most simple way: In the beginning,
		1800	start the watcher:
		1801
		1802	ev_timer_init (timer, callback, 60., 0.);
		1803	ev_timer_start (loop, timer);
		1804
		1805	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1806	and start it again:
		1807
		1808	ev_timer_stop (loop, timer);
		1809	ev_timer_set (timer, 60., 0.);
		1810	ev_timer_start (loop, timer);
		1811
		1812	This is relatively simple to implement, but means that each time there is
		1813	some activity, libev will first have to remove the timer from its internal
		1814	data structure and then add it again. Libev tries to be fast, but it's
		1815	still not a constant-time operation.
		1816
		1817	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1818
		1819	This is the easiest way, and involves using C<ev_timer_again> instead of
		1820	C<ev_timer_start>.
		1821
		1822	To implement this, configure an C<ev_timer> with a C<repeat> value
		1823	of C<60> and then call C<ev_timer_again> at start and each time you
		1824	successfully read or write some data. If you go into an idle state where
		1825	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1826	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1827
		1828	That means you can ignore both the C<ev_timer_start> function and the
		1829	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1830	member and C<ev_timer_again>.
		1831
		1832	At start:
		1833
		1834	ev_init (timer, callback);
		1835	timer->repeat = 60.;
		1836	ev_timer_again (loop, timer);
		1837
		1838	Each time there is some activity:
		1839
		1840	ev_timer_again (loop, timer);
		1841
		1842	It is even possible to change the time-out on the fly, regardless of
		1843	whether the watcher is active or not:
		1844
		1845	timer->repeat = 30.;
		1846	ev_timer_again (loop, timer);
		1847
		1848	This is slightly more efficient then stopping/starting the timer each time
		1849	you want to modify its timeout value, as libev does not have to completely
		1850	remove and re-insert the timer from/into its internal data structure.
		1851
		1852	It is, however, even simpler than the "obvious" way to do it.
		1853
		1854	=item 3. Let the timer time out, but then re-arm it as required.
		1855
		1856	This method is more tricky, but usually most efficient: Most timeouts are
		1857	relatively long compared to the intervals between other activity - in
		1858	our example, within 60 seconds, there are usually many I/O events with
		1859	associated activity resets.
		1860
		1861	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1862	but remember the time of last activity, and check for a real timeout only
		1863	within the callback:
		1864
		1865	ev_tstamp timeout = 60.;
		1866	ev_tstamp last_activity; // time of last activity
		1867	ev_timer timer;
		1868
		1869	static void
		1870	callback (EV_P_ ev_timer *w, int revents)
		1871	{
		1872	// calculate when the timeout would happen
		1873	ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
		1874
		1875	// if negative, it means we the timeout already occured
		1876	if (after < 0.)
		1877	{
		1878	// timeout occurred, take action
		1879	}
		1880	else
		1881	{
		1882	// callback was invoked, but there was some recent
		1883	// activity. simply restart the timer to time out
		1884	// after "after" seconds, which is the earliest time
		1885	// the timeout can occur.
		1886	ev_timer_set (w, after, 0.);
		1887	ev_timer_start (EV_A_ w);
		1888	}
		1889	}
		1890
		1891	To summarise the callback: first calculate in how many seconds the
		1892	timeout will occur (by calculating the absolute time when it would occur,
		1893	C<last_activity + timeout>, and subtracting the current time, C<ev_now
		1894	(EV_A)> from that).
		1895
		1896	If this value is negative, then we are already past the timeout, i.e. we
		1897	timed out, and need to do whatever is needed in this case.
		1898
		1899	Otherwise, we now the earliest time at which the timeout would trigger,
		1900	and simply start the timer with this timeout value.
		1901
		1902	In other words, each time the callback is invoked it will check whether
		1903	the timeout cocured. If not, it will simply reschedule itself to check
		1904	again at the earliest time it could time out. Rinse. Repeat.
		1905
		1906	This scheme causes more callback invocations (about one every 60 seconds
		1907	minus half the average time between activity), but virtually no calls to
		1908	libev to change the timeout.
		1909
		1910	To start the machinery, simply initialise the watcher and set
		1911	C<last_activity> to the current time (meaning there was some activity just
		1912	now), then call the callback, which will "do the right thing" and start
		1913	the timer:
		1914
		1915	last_activity = ev_now (EV_A);
		1916	ev_init (&timer, callback);
		1917	callback (EV_A_ &timer, 0);
		1918
		1919	When there is some activity, simply store the current time in
		1920	C<last_activity>, no libev calls at all:
		1921
		1922	if (activity detected)
		1923	last_activity = ev_now (EV_A);
		1924
		1925	When your timeout value changes, then the timeout can be changed by simply
		1926	providing a new value, stopping the timer and calling the callback, which
		1927	will agaion do the right thing (for example, time out immediately :).
		1928
		1929	timeout = new_value;
		1930	ev_timer_stop (EV_A_ &timer);
		1931	callback (EV_A_ &timer, 0);
		1932
		1933	This technique is slightly more complex, but in most cases where the
		1934	time-out is unlikely to be triggered, much more efficient.
		1935
		1936	=item 4. Wee, just use a double-linked list for your timeouts.
		1937
		1938	If there is not one request, but many thousands (millions...), all
		1939	employing some kind of timeout with the same timeout value, then one can
		1940	do even better:
		1941
		1942	When starting the timeout, calculate the timeout value and put the timeout
		1943	at the I<end> of the list.
		1944
		1945	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1946	the list is expected to fire (for example, using the technique #3).
		1947
		1948	When there is some activity, remove the timer from the list, recalculate
		1949	the timeout, append it to the end of the list again, and make sure to
		1950	update the C<ev_timer> if it was taken from the beginning of the list.
		1951
		1952	This way, one can manage an unlimited number of timeouts in O(1) time for
		1953	starting, stopping and updating the timers, at the expense of a major
		1954	complication, and having to use a constant timeout. The constant timeout
		1955	ensures that the list stays sorted.
		1956
		1957	=back
		1958
		1959	So which method the best?
		1960
		1961	Method #2 is a simple no-brain-required solution that is adequate in most
		1962	situations. Method #3 requires a bit more thinking, but handles many cases
		1963	better, and isn't very complicated either. In most case, choosing either
		1964	one is fine, with #3 being better in typical situations.
		1965
		1966	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1967	rather complicated, but extremely efficient, something that really pays
		1968	off after the first million or so of active timers, i.e. it's usually
		1969	overkill :)
		1970
		1971	=head3 The special problem of being too early
		1972
		1973	If you ask a timer to call your callback after three seconds, then
		1974	you expect it to be invoked after three seconds - but of course, this
		1975	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		1976	guaranteed to any precision by libev - imagine somebody suspending the
		1977	process with a STOP signal for a few hours for example.
		1978
		1979	So, libev tries to invoke your callback as soon as possible I<after> the
		1980	delay has occurred, but cannot guarantee this.
		1981
		1982	A less obvious failure mode is calling your callback too early: many event
		1983	loops compare timestamps with a "elapsed delay >= requested delay", but
		1984	this can cause your callback to be invoked much earlier than you would
		1985	expect.
		1986
		1987	To see why, imagine a system with a clock that only offers full second
		1988	resolution (think windows if you can't come up with a broken enough OS
		1989	yourself). If you schedule a one-second timer at the time 500.9, then the
		1990	event loop will schedule your timeout to elapse at a system time of 500
		1991	(500.9 truncated to the resolution) + 1, or 501.
		1992
		1993	If an event library looks at the timeout 0.1s later, it will see "501 >=
		1994	501" and invoke the callback 0.1s after it was started, even though a
		1995	one-second delay was requested - this is being "too early", despite best
		1996	intentions.
		1997
		1998	This is the reason why libev will never invoke the callback if the elapsed
		1999	delay equals the requested delay, but only when the elapsed delay is
		2000	larger than the requested delay. In the example above, libev would only invoke
		2001	the callback at system time 502, or 1.1s after the timer was started.
		2002
		2003	So, while libev cannot guarantee that your callback will be invoked
		2004	exactly when requested, it I<can> and I<does> guarantee that the requested
		2005	delay has actually elapsed, or in other words, it always errs on the "too
		2006	late" side of things.
		2007
		2008	=head3 The special problem of time updates
		2009
		2010	Establishing the current time is a costly operation (it usually takes
		2011	at least one system call): EV therefore updates its idea of the current
		2012	time only before and after C<ev_run> collects new events, which causes a
		2013	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		2014	lots of events in one iteration.
525		2015
526	The relative timeouts are calculated relative to the C<ev_now ()>	2016	The relative timeouts are calculated relative to the C<ev_now ()>
527	time. This is usually the right thing as this timestamp refers to the time	2017	time. This is usually the right thing as this timestamp refers to the time
528	of the event triggering whatever timeout you are modifying/starting. If	2018	of the event triggering whatever timeout you are modifying/starting. If
529	you suspect event processing to be delayed and you I<need> to base the timeout	2019	you suspect event processing to be delayed and you I<need> to base the
530	on the current time, use something like this to adjust for this:	2020	timeout on the current time, use something like this to adjust for this:
531		2021
532	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2022	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
533		2023
534	The callback is guarenteed to be invoked only when its timeout has passed,	2024	If the event loop is suspended for a long time, you can also force an
535	but if multiple timers become ready during the same loop iteration then	2025	update of the time returned by C<ev_now ()> by calling C<ev_now_update
536	order of execution is undefined.	2026	()>.
		2027
		2028	=head3 The special problem of unsynchronised clocks
		2029
		2030	Modern systems have a variety of clocks - libev itself uses the normal
		2031	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2032	jumps).
		2033
		2034	Neither of these clocks is synchronised with each other or any other clock
		2035	on the system, so C<ev_time ()> might return a considerably different time
		2036	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2037	a call to C<gettimeofday> might return a second count that is one higher
		2038	than a directly following call to C<time>.
		2039
		2040	The moral of this is to only compare libev-related timestamps with
		2041	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2042	a second or so.
		2043
		2044	One more problem arises due to this lack of synchronisation: if libev uses
		2045	the system monotonic clock and you compare timestamps from C<ev_time>
		2046	or C<ev_now> from when you started your timer and when your callback is
		2047	invoked, you will find that sometimes the callback is a bit "early".
		2048
		2049	This is because C<ev_timer>s work in real time, not wall clock time, so
		2050	libev makes sure your callback is not invoked before the delay happened,
		2051	I<measured according to the real time>, not the system clock.
		2052
		2053	If your timeouts are based on a physical timescale (e.g. "time out this
		2054	connection after 100 seconds") then this shouldn't bother you as it is
		2055	exactly the right behaviour.
		2056
		2057	If you want to compare wall clock/system timestamps to your timers, then
		2058	you need to use C<ev_periodic>s, as these are based on the wall clock
		2059	time, where your comparisons will always generate correct results.
		2060
		2061	=head3 The special problems of suspended animation
		2062
		2063	When you leave the server world it is quite customary to hit machines that
		2064	can suspend/hibernate - what happens to the clocks during such a suspend?
		2065
		2066	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		2067	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		2068	to run until the system is suspended, but they will not advance while the
		2069	system is suspended. That means, on resume, it will be as if the program
		2070	was frozen for a few seconds, but the suspend time will not be counted
		2071	towards C<ev_timer> when a monotonic clock source is used. The real time
		2072	clock advanced as expected, but if it is used as sole clocksource, then a
		2073	long suspend would be detected as a time jump by libev, and timers would
		2074	be adjusted accordingly.
		2075
		2076	I would not be surprised to see different behaviour in different between
		2077	operating systems, OS versions or even different hardware.
		2078
		2079	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2080	time jump in the monotonic clocks and the realtime clock. If the program
		2081	is suspended for a very long time, and monotonic clock sources are in use,
		2082	then you can expect C<ev_timer>s to expire as the full suspension time
		2083	will be counted towards the timers. When no monotonic clock source is in
		2084	use, then libev will again assume a timejump and adjust accordingly.
		2085
		2086	It might be beneficial for this latter case to call C<ev_suspend>
		2087	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2088	deterministic behaviour in this case (you can do nothing against
		2089	C<SIGSTOP>).
		2090
		2091	=head3 Watcher-Specific Functions and Data Members
537		2092
538	=over 4	2093	=over 4
539		2094
540	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2095	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
541		2096
542	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2097	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
543		2098
544	Configure the timer to trigger after C<after> seconds. If C<repeat> is	2099	Configure the timer to trigger after C<after> seconds. If C<repeat>
545	C<0.>, then it will automatically be stopped. If it is positive, then the	2100	is C<0.>, then it will automatically be stopped once the timeout is
546	timer will automatically be configured to trigger again C<repeat> seconds	2101	reached. If it is positive, then the timer will automatically be
547	later, again, and again, until stopped manually.	2102	configured to trigger again C<repeat> seconds later, again, and again,
		2103	until stopped manually.
548		2104
549	The timer itself will do a best-effort at avoiding drift, that is, if you	2105	The timer itself will do a best-effort at avoiding drift, that is, if
550	configure a timer to trigger every 10 seconds, then it will trigger at	2106	you configure a timer to trigger every 10 seconds, then it will normally
551	exactly 10 second intervals. If, however, your program cannot keep up with	2107	trigger at exactly 10 second intervals. If, however, your program cannot
552	the timer (because it takes longer than those 10 seconds to do stuff) the	2108	keep up with the timer (because it takes longer than those 10 seconds to
553	timer will not fire more than once per event loop iteration.	2109	do stuff) the timer will not fire more than once per event loop iteration.
554		2110
555	=item ev_timer_again (loop)	2111	=item ev_timer_again (loop, ev_timer *)
556		2112
557	This will act as if the timer timed out and restart it again if it is	2113	This will act as if the timer timed out, and restarts it again if it is
558	repeating. The exact semantics are:	2114	repeating. It basically works like calling C<ev_timer_stop>, updating the
		2115	timeout to the C<repeat> value and calling C<ev_timer_start>.
559		2116
		2117	The exact semantics are as in the following rules, all of which will be
		2118	applied to the watcher:
		2119
		2120	=over 4
		2121
		2122	=item If the timer is pending, the pending status is always cleared.
		2123
560	If the timer is started but nonrepeating, stop it.	2124	=item If the timer is started but non-repeating, stop it (as if it timed
		2125	out, without invoking it).
561		2126
562	If the timer is repeating, either start it if necessary (with the repeat	2127	=item If the timer is repeating, make the C<repeat> value the new timeout
563	value), or reset the running timer to the repeat value.	2128	and start the timer, if necessary.
564
565	This sounds a bit complicated, but here is a useful and typical
566	example: Imagine you have a tcp connection and you want a so-called idle
567	timeout, that is, you want to be called when there have been, say, 60
568	seconds of inactivity on the socket. The easiest way to do this is to
569	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each
570	time you successfully read or write some data. If you go into an idle
571	state where you do not expect data to travel on the socket, you can stop
572	the timer, and again will automatically restart it if need be.
573		2129
574	=back	2130	=back
575		2131
		2132	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
		2133	usage example.
		2134
		2135	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
		2136
		2137	Returns the remaining time until a timer fires. If the timer is active,
		2138	then this time is relative to the current event loop time, otherwise it's
		2139	the timeout value currently configured.
		2140
		2141	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2142	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2143	will return C<4>. When the timer expires and is restarted, it will return
		2144	roughly C<7> (likely slightly less as callback invocation takes some time,
		2145	too), and so on.
		2146
		2147	=item ev_tstamp repeat [read-write]
		2148
		2149	The current C<repeat> value. Will be used each time the watcher times out
		2150	or C<ev_timer_again> is called, and determines the next timeout (if any),
		2151	which is also when any modifications are taken into account.
		2152
		2153	=back
		2154
		2155	=head3 Examples
		2156
		2157	Example: Create a timer that fires after 60 seconds.
		2158
		2159	static void
		2160	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2161	{
		2162	.. one minute over, w is actually stopped right here
		2163	}
		2164
		2165	ev_timer mytimer;
		2166	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		2167	ev_timer_start (loop, &mytimer);
		2168
		2169	Example: Create a timeout timer that times out after 10 seconds of
		2170	inactivity.
		2171
		2172	static void
		2173	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2174	{
		2175	.. ten seconds without any activity
		2176	}
		2177
		2178	ev_timer mytimer;
		2179	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		2180	ev_timer_again (&mytimer); /* start timer */
		2181	ev_run (loop, 0);
		2182
		2183	// and in some piece of code that gets executed on any "activity":
		2184	// reset the timeout to start ticking again at 10 seconds
		2185	ev_timer_again (&mytimer);
		2186
		2187
576	=head2 C<ev_periodic> - to cron or not to cron	2188	=head2 C<ev_periodic> - to cron or not to cron?
577		2189
578	Periodic watchers are also timers of a kind, but they are very versatile	2190	Periodic watchers are also timers of a kind, but they are very versatile
579	(and unfortunately a bit complex).	2191	(and unfortunately a bit complex).
580		2192
581	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2193	Unlike C<ev_timer>, periodic watchers are not based on real time (or
582	but on wallclock time (absolute time). You can tell a periodic watcher	2194	relative time, the physical time that passes) but on wall clock time
583	to trigger "at" some specific point in time. For example, if you tell a	2195	(absolute time, the thing you can read on your calender or clock). The
584	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	2196	difference is that wall clock time can run faster or slower than real
585	+ 10.>) and then reset your system clock to the last year, then it will	2197	time, and time jumps are not uncommon (e.g. when you adjust your
586	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2198	wrist-watch).
587	roughly 10 seconds later and of course not if you reset your system time
588	again).
589		2199
590	They can also be used to implement vastly more complex timers, such as	2200	You can tell a periodic watcher to trigger after some specific point
		2201	in time: for example, if you tell a periodic watcher to trigger "in 10
		2202	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2203	not a delay) and then reset your system clock to January of the previous
		2204	year, then it will take a year or more to trigger the event (unlike an
		2205	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2206	it, as it uses a relative timeout).
		2207
		2208	C<ev_periodic> watchers can also be used to implement vastly more complex
591	triggering an event on eahc midnight, local time.	2209	timers, such as triggering an event on each "midnight, local time", or
		2210	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2211	those cannot react to time jumps.
592		2212
593	As with timers, the callback is guarenteed to be invoked only when the	2213	As with timers, the callback is guaranteed to be invoked only when the
594	time (C<at>) has been passed, but if multiple periodic timers become ready	2214	point in time where it is supposed to trigger has passed. If multiple
595	during the same loop iteration then order of execution is undefined.	2215	timers become ready during the same loop iteration then the ones with
		2216	earlier time-out values are invoked before ones with later time-out values
		2217	(but this is no longer true when a callback calls C<ev_run> recursively).
		2218
		2219	=head3 Watcher-Specific Functions and Data Members
596		2220
597	=over 4	2221	=over 4
598		2222
599	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2223	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
600		2224
601	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2225	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
602		2226
603	Lots of arguments, lets sort it out... There are basically three modes of	2227	Lots of arguments, let's sort it out... There are basically three modes of
604	operation, and we will explain them from simplest to complex:	2228	operation, and we will explain them from simplest to most complex:
605		2229
606	=over 4	2230	=over 4
607		2231
608	=item * absolute timer (interval = reschedule_cb = 0)	2232	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
609		2233
610	In this configuration the watcher triggers an event at the wallclock time	2234	In this configuration the watcher triggers an event after the wall clock
611	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2235	time C<offset> has passed. It will not repeat and will not adjust when a
612	that is, if it is to be run at January 1st 2011 then it will run when the	2236	time jump occurs, that is, if it is to be run at January 1st 2011 then it
613	system time reaches or surpasses this time.	2237	will be stopped and invoked when the system clock reaches or surpasses
		2238	this point in time.
614		2239
615	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	2240	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
616		2241
617	In this mode the watcher will always be scheduled to time out at the next	2242	In this mode the watcher will always be scheduled to time out at the next
618	C<at + N * interval> time (for some integer N) and then repeat, regardless	2243	C<offset + N * interval> time (for some integer N, which can also be
619	of any time jumps.	2244	negative) and then repeat, regardless of any time jumps. The C<offset>
		2245	argument is merely an offset into the C<interval> periods.
620		2246
621	This can be used to create timers that do not drift with respect to system	2247	This can be used to create timers that do not drift with respect to the
622	time:	2248	system clock, for example, here is an C<ev_periodic> that triggers each
		2249	hour, on the hour (with respect to UTC):
623		2250
624	ev_periodic_set (&periodic, 0., 3600., 0);	2251	ev_periodic_set (&periodic, 0., 3600., 0);
625		2252
626	This doesn't mean there will always be 3600 seconds in between triggers,	2253	This doesn't mean there will always be 3600 seconds in between triggers,
627	but only that the the callback will be called when the system time shows a	2254	but only that the callback will be called when the system time shows a
628	full hour (UTC), or more correctly, when the system time is evenly divisible	2255	full hour (UTC), or more correctly, when the system time is evenly divisible
629	by 3600.	2256	by 3600.
630		2257
631	Another way to think about it (for the mathematically inclined) is that	2258	Another way to think about it (for the mathematically inclined) is that
632	C<ev_periodic> will try to run the callback in this mode at the next possible	2259	C<ev_periodic> will try to run the callback in this mode at the next possible
633	time where C<time = at (mod interval)>, regardless of any time jumps.	2260	time where C<time = offset (mod interval)>, regardless of any time jumps.
634		2261
		2262	The C<interval> I<MUST> be positive, and for numerical stability, the
		2263	interval value should be higher than C<1/8192> (which is around 100
		2264	microseconds) and C<offset> should be higher than C<0> and should have
		2265	at most a similar magnitude as the current time (say, within a factor of
		2266	ten). Typical values for offset are, in fact, C<0> or something between
		2267	C<0> and C<interval>, which is also the recommended range.
		2268
		2269	Note also that there is an upper limit to how often a timer can fire (CPU
		2270	speed for example), so if C<interval> is very small then timing stability
		2271	will of course deteriorate. Libev itself tries to be exact to be about one
		2272	millisecond (if the OS supports it and the machine is fast enough).
		2273
635	=item * manual reschedule mode (reschedule_cb = callback)	2274	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
636		2275
637	In this mode the values for C<interval> and C<at> are both being	2276	In this mode the values for C<interval> and C<offset> are both being
638	ignored. Instead, each time the periodic watcher gets scheduled, the	2277	ignored. Instead, each time the periodic watcher gets scheduled, the
639	reschedule callback will be called with the watcher as first, and the	2278	reschedule callback will be called with the watcher as first, and the
640	current time as second argument.	2279	current time as second argument.
641		2280
642	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2281	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
643	ever, or make any event loop modifications>. If you need to stop it,	2282	or make ANY other event loop modifications whatsoever, unless explicitly
644	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2283	allowed by documentation here>.
645	starting a prepare watcher).
646		2284
		2285	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2286	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2287	only event loop modification you are allowed to do).
		2288
647	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2289	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
648	ev_tstamp now)>, e.g.:	2290	*w, ev_tstamp now)>, e.g.:
649		2291
		2292	static ev_tstamp
650	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2293	my_rescheduler (ev_periodic *w, ev_tstamp now)
651	{	2294	{
652	return now + 60.;	2295	return now + 60.;
653	}	2296	}
654		2297
655	It must return the next time to trigger, based on the passed time value	2298	It must return the next time to trigger, based on the passed time value
656	(that is, the lowest time value larger than to the second argument). It	2299	(that is, the lowest time value larger than to the second argument). It
657	will usually be called just before the callback will be triggered, but	2300	will usually be called just before the callback will be triggered, but
658	might be called at other times, too.	2301	might be called at other times, too.
659		2302
660	NOTE: I<< This callback must always return a time that is later than the	2303	NOTE: I<< This callback must always return a time that is higher than or
661	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2304	equal to the passed C<now> value >>.
662		2305
663	This can be used to create very complex timers, such as a timer that	2306	This can be used to create very complex timers, such as a timer that
664	triggers on each midnight, local time. To do this, you would calculate the	2307	triggers on "next midnight, local time". To do this, you would calculate the
665	next midnight after C<now> and return the timestamp value for this. How	2308	next midnight after C<now> and return the timestamp value for this. How
666	you do this is, again, up to you (but it is not trivial, which is the main	2309	you do this is, again, up to you (but it is not trivial, which is the main
667	reason I omitted it as an example).	2310	reason I omitted it as an example).
668		2311
669	=back	2312	=back
…		…
673	Simply stops and restarts the periodic watcher again. This is only useful	2316	Simply stops and restarts the periodic watcher again. This is only useful
674	when you changed some parameters or the reschedule callback would return	2317	when you changed some parameters or the reschedule callback would return
675	a different time than the last time it was called (e.g. in a crond like	2318	a different time than the last time it was called (e.g. in a crond like
676	program when the crontabs have changed).	2319	program when the crontabs have changed).
677		2320
		2321	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2322
		2323	When active, returns the absolute time that the watcher is supposed
		2324	to trigger next. This is not the same as the C<offset> argument to
		2325	C<ev_periodic_set>, but indeed works even in interval and manual
		2326	rescheduling modes.
		2327
		2328	=item ev_tstamp offset [read-write]
		2329
		2330	When repeating, this contains the offset value, otherwise this is the
		2331	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2332	although libev might modify this value for better numerical stability).
		2333
		2334	Can be modified any time, but changes only take effect when the periodic
		2335	timer fires or C<ev_periodic_again> is being called.
		2336
		2337	=item ev_tstamp interval [read-write]
		2338
		2339	The current interval value. Can be modified any time, but changes only
		2340	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2341	called.
		2342
		2343	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2344
		2345	The current reschedule callback, or C<0>, if this functionality is
		2346	switched off. Can be changed any time, but changes only take effect when
		2347	the periodic timer fires or C<ev_periodic_again> is being called.
		2348
678	=back	2349	=back
679		2350
		2351	=head3 Examples
		2352
		2353	Example: Call a callback every hour, or, more precisely, whenever the
		2354	system time is divisible by 3600. The callback invocation times have
		2355	potentially a lot of jitter, but good long-term stability.
		2356
		2357	static void
		2358	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
		2359	{
		2360	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2361	}
		2362
		2363	ev_periodic hourly_tick;
		2364	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2365	ev_periodic_start (loop, &hourly_tick);
		2366
		2367	Example: The same as above, but use a reschedule callback to do it:
		2368
		2369	#include <math.h>
		2370
		2371	static ev_tstamp
		2372	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2373	{
		2374	return now + (3600. - fmod (now, 3600.));
		2375	}
		2376
		2377	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2378
		2379	Example: Call a callback every hour, starting now:
		2380
		2381	ev_periodic hourly_tick;
		2382	ev_periodic_init (&hourly_tick, clock_cb,
		2383	fmod (ev_now (loop), 3600.), 3600., 0);
		2384	ev_periodic_start (loop, &hourly_tick);
		2385
		2386
680	=head2 C<ev_signal> - signal me when a signal gets signalled	2387	=head2 C<ev_signal> - signal me when a signal gets signalled!
681		2388
682	Signal watchers will trigger an event when the process receives a specific	2389	Signal watchers will trigger an event when the process receives a specific
683	signal one or more times. Even though signals are very asynchronous, libev	2390	signal one or more times. Even though signals are very asynchronous, libev
684	will try it's best to deliver signals synchronously, i.e. as part of the	2391	will try its best to deliver signals synchronously, i.e. as part of the
685	normal event processing, like any other event.	2392	normal event processing, like any other event.
686		2393
		2394	If you want signals to be delivered truly asynchronously, just use
		2395	C<sigaction> as you would do without libev and forget about sharing
		2396	the signal. You can even use C<ev_async> from a signal handler to
		2397	synchronously wake up an event loop.
		2398
687	You can configure as many watchers as you like per signal. Only when the	2399	You can configure as many watchers as you like for the same signal, but
		2400	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2401	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2402	C<SIGINT> in both the default loop and another loop at the same time. At
		2403	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2404
688	first watcher gets started will libev actually register a signal watcher	2405	When the first watcher gets started will libev actually register something
689	with the kernel (thus it coexists with your own signal handlers as long	2406	with the kernel (thus it coexists with your own signal handlers as long as
690	as you don't register any with libev). Similarly, when the last signal	2407	you don't register any with libev for the same signal).
691	watcher for a signal is stopped libev will reset the signal handler to	2408
692	SIG_DFL (regardless of what it was set to before).	2409	If possible and supported, libev will install its handlers with
		2410	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2411	not be unduly interrupted. If you have a problem with system calls getting
		2412	interrupted by signals you can block all signals in an C<ev_check> watcher
		2413	and unblock them in an C<ev_prepare> watcher.
		2414
		2415	=head3 The special problem of inheritance over fork/execve/pthread_create
		2416
		2417	Both the signal mask (C<sigprocmask>) and the signal disposition
		2418	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2419	stopping it again), that is, libev might or might not block the signal,
		2420	and might or might not set or restore the installed signal handler (but
		2421	see C<EVFLAG_NOSIGMASK>).
		2422
		2423	While this does not matter for the signal disposition (libev never
		2424	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2425	C<execve>), this matters for the signal mask: many programs do not expect
		2426	certain signals to be blocked.
		2427
		2428	This means that before calling C<exec> (from the child) you should reset
		2429	the signal mask to whatever "default" you expect (all clear is a good
		2430	choice usually).
		2431
		2432	The simplest way to ensure that the signal mask is reset in the child is
		2433	to install a fork handler with C<pthread_atfork> that resets it. That will
		2434	catch fork calls done by libraries (such as the libc) as well.
		2435
		2436	In current versions of libev, the signal will not be blocked indefinitely
		2437	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2438	the window of opportunity for problems, it will not go away, as libev
		2439	I<has> to modify the signal mask, at least temporarily.
		2440
		2441	So I can't stress this enough: I<If you do not reset your signal mask when
		2442	you expect it to be empty, you have a race condition in your code>. This
		2443	is not a libev-specific thing, this is true for most event libraries.
		2444
		2445	=head3 The special problem of threads signal handling
		2446
		2447	POSIX threads has problematic signal handling semantics, specifically,
		2448	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2449	threads in a process block signals, which is hard to achieve.
		2450
		2451	When you want to use sigwait (or mix libev signal handling with your own
		2452	for the same signals), you can tackle this problem by globally blocking
		2453	all signals before creating any threads (or creating them with a fully set
		2454	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2455	loops. Then designate one thread as "signal receiver thread" which handles
		2456	these signals. You can pass on any signals that libev might be interested
		2457	in by calling C<ev_feed_signal>.
		2458
		2459	=head3 Watcher-Specific Functions and Data Members
693		2460
694	=over 4	2461	=over 4
695		2462
696	=item ev_signal_init (ev_signal *, callback, int signum)	2463	=item ev_signal_init (ev_signal *, callback, int signum)
697		2464
698	=item ev_signal_set (ev_signal *, int signum)	2465	=item ev_signal_set (ev_signal *, int signum)
699		2466
700	Configures the watcher to trigger on the given signal number (usually one	2467	Configures the watcher to trigger on the given signal number (usually one
701	of the C<SIGxxx> constants).	2468	of the C<SIGxxx> constants).
702		2469
		2470	=item int signum [read-only]
		2471
		2472	The signal the watcher watches out for.
		2473
703	=back	2474	=back
704		2475
		2476	=head3 Examples
		2477
		2478	Example: Try to exit cleanly on SIGINT.
		2479
		2480	static void
		2481	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2482	{
		2483	ev_break (loop, EVBREAK_ALL);
		2484	}
		2485
		2486	ev_signal signal_watcher;
		2487	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2488	ev_signal_start (loop, &signal_watcher);
		2489
		2490
705	=head2 C<ev_child> - wait for pid status changes	2491	=head2 C<ev_child> - watch out for process status changes
706		2492
707	Child watchers trigger when your process receives a SIGCHLD in response to	2493	Child watchers trigger when your process receives a SIGCHLD in response to
708	some child status changes (most typically when a child of yours dies).	2494	some child status changes (most typically when a child of yours dies or
		2495	exits). It is permissible to install a child watcher I<after> the child
		2496	has been forked (which implies it might have already exited), as long
		2497	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2498	forking and then immediately registering a watcher for the child is fine,
		2499	but forking and registering a watcher a few event loop iterations later or
		2500	in the next callback invocation is not.
		2501
		2502	Only the default event loop is capable of handling signals, and therefore
		2503	you can only register child watchers in the default event loop.
		2504
		2505	Due to some design glitches inside libev, child watchers will always be
		2506	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2507	libev)
		2508
		2509	=head3 Process Interaction
		2510
		2511	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2512	initialised. This is necessary to guarantee proper behaviour even if the
		2513	first child watcher is started after the child exits. The occurrence
		2514	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2515	synchronously as part of the event loop processing. Libev always reaps all
		2516	children, even ones not watched.
		2517
		2518	=head3 Overriding the Built-In Processing
		2519
		2520	Libev offers no special support for overriding the built-in child
		2521	processing, but if your application collides with libev's default child
		2522	handler, you can override it easily by installing your own handler for
		2523	C<SIGCHLD> after initialising the default loop, and making sure the
		2524	default loop never gets destroyed. You are encouraged, however, to use an
		2525	event-based approach to child reaping and thus use libev's support for
		2526	that, so other libev users can use C<ev_child> watchers freely.
		2527
		2528	=head3 Stopping the Child Watcher
		2529
		2530	Currently, the child watcher never gets stopped, even when the
		2531	child terminates, so normally one needs to stop the watcher in the
		2532	callback. Future versions of libev might stop the watcher automatically
		2533	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2534	problem).
		2535
		2536	=head3 Watcher-Specific Functions and Data Members
709		2537
710	=over 4	2538	=over 4
711		2539
712	=item ev_child_init (ev_child *, callback, int pid)	2540	=item ev_child_init (ev_child *, callback, int pid, int trace)
713		2541
714	=item ev_child_set (ev_child *, int pid)	2542	=item ev_child_set (ev_child *, int pid, int trace)
715		2543
716	Configures the watcher to wait for status changes of process C<pid> (or	2544	Configures the watcher to wait for status changes of process C<pid> (or
717	I<any> process if C<pid> is specified as C<0>). The callback can look	2545	I<any> process if C<pid> is specified as C<0>). The callback can look
718	at the C<rstatus> member of the C<ev_child> watcher structure to see	2546	at the C<rstatus> member of the C<ev_child> watcher structure to see
719	the status word (use the macros from C<sys/wait.h> and see your systems	2547	the status word (use the macros from C<sys/wait.h> and see your systems
720	C<waitpid> documentation). The C<rpid> member contains the pid of the	2548	C<waitpid> documentation). The C<rpid> member contains the pid of the
721	process causing the status change.	2549	process causing the status change. C<trace> must be either C<0> (only
		2550	activate the watcher when the process terminates) or C<1> (additionally
		2551	activate the watcher when the process is stopped or continued).
		2552
		2553	=item int pid [read-only]
		2554
		2555	The process id this watcher watches out for, or C<0>, meaning any process id.
		2556
		2557	=item int rpid [read-write]
		2558
		2559	The process id that detected a status change.
		2560
		2561	=item int rstatus [read-write]
		2562
		2563	The process exit/trace status caused by C<rpid> (see your systems
		2564	C<waitpid> and C<sys/wait.h> documentation for details).
722		2565
723	=back	2566	=back
724		2567
		2568	=head3 Examples
		2569
		2570	Example: C<fork()> a new process and install a child handler to wait for
		2571	its completion.
		2572
		2573	ev_child cw;
		2574
		2575	static void
		2576	child_cb (EV_P_ ev_child *w, int revents)
		2577	{
		2578	ev_child_stop (EV_A_ w);
		2579	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2580	}
		2581
		2582	pid_t pid = fork ();
		2583
		2584	if (pid < 0)
		2585	// error
		2586	else if (pid == 0)
		2587	{
		2588	// the forked child executes here
		2589	exit (1);
		2590	}
		2591	else
		2592	{
		2593	ev_child_init (&cw, child_cb, pid, 0);
		2594	ev_child_start (EV_DEFAULT_ &cw);
		2595	}
		2596
		2597
		2598	=head2 C<ev_stat> - did the file attributes just change?
		2599
		2600	This watches a file system path for attribute changes. That is, it calls
		2601	C<stat> on that path in regular intervals (or when the OS says it changed)
		2602	and sees if it changed compared to the last time, invoking the callback if
		2603	it did.
		2604
		2605	The path does not need to exist: changing from "path exists" to "path does
		2606	not exist" is a status change like any other. The condition "path does not
		2607	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2608	C<st_nlink> field being zero (which is otherwise always forced to be at
		2609	least one) and all the other fields of the stat buffer having unspecified
		2610	contents.
		2611
		2612	The path I<must not> end in a slash or contain special components such as
		2613	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2614	your working directory changes, then the behaviour is undefined.
		2615
		2616	Since there is no portable change notification interface available, the
		2617	portable implementation simply calls C<stat(2)> regularly on the path
		2618	to see if it changed somehow. You can specify a recommended polling
		2619	interval for this case. If you specify a polling interval of C<0> (highly
		2620	recommended!) then a I<suitable, unspecified default> value will be used
		2621	(which you can expect to be around five seconds, although this might
		2622	change dynamically). Libev will also impose a minimum interval which is
		2623	currently around C<0.1>, but that's usually overkill.
		2624
		2625	This watcher type is not meant for massive numbers of stat watchers,
		2626	as even with OS-supported change notifications, this can be
		2627	resource-intensive.
		2628
		2629	At the time of this writing, the only OS-specific interface implemented
		2630	is the Linux inotify interface (implementing kqueue support is left as an
		2631	exercise for the reader. Note, however, that the author sees no way of
		2632	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2633
		2634	=head3 ABI Issues (Largefile Support)
		2635
		2636	Libev by default (unless the user overrides this) uses the default
		2637	compilation environment, which means that on systems with large file
		2638	support disabled by default, you get the 32 bit version of the stat
		2639	structure. When using the library from programs that change the ABI to
		2640	use 64 bit file offsets the programs will fail. In that case you have to
		2641	compile libev with the same flags to get binary compatibility. This is
		2642	obviously the case with any flags that change the ABI, but the problem is
		2643	most noticeably displayed with ev_stat and large file support.
		2644
		2645	The solution for this is to lobby your distribution maker to make large
		2646	file interfaces available by default (as e.g. FreeBSD does) and not
		2647	optional. Libev cannot simply switch on large file support because it has
		2648	to exchange stat structures with application programs compiled using the
		2649	default compilation environment.
		2650
		2651	=head3 Inotify and Kqueue
		2652
		2653	When C<inotify (7)> support has been compiled into libev and present at
		2654	runtime, it will be used to speed up change detection where possible. The
		2655	inotify descriptor will be created lazily when the first C<ev_stat>
		2656	watcher is being started.
		2657
		2658	Inotify presence does not change the semantics of C<ev_stat> watchers
		2659	except that changes might be detected earlier, and in some cases, to avoid
		2660	making regular C<stat> calls. Even in the presence of inotify support
		2661	there are many cases where libev has to resort to regular C<stat> polling,
		2662	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2663	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2664	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2665	xfs are fully working) libev usually gets away without polling.
		2666
		2667	There is no support for kqueue, as apparently it cannot be used to
		2668	implement this functionality, due to the requirement of having a file
		2669	descriptor open on the object at all times, and detecting renames, unlinks
		2670	etc. is difficult.
		2671
		2672	=head3 C<stat ()> is a synchronous operation
		2673
		2674	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2675	the process. The exception are C<ev_stat> watchers - those call C<stat
		2676	()>, which is a synchronous operation.
		2677
		2678	For local paths, this usually doesn't matter: unless the system is very
		2679	busy or the intervals between stat's are large, a stat call will be fast,
		2680	as the path data is usually in memory already (except when starting the
		2681	watcher).
		2682
		2683	For networked file systems, calling C<stat ()> can block an indefinite
		2684	time due to network issues, and even under good conditions, a stat call
		2685	often takes multiple milliseconds.
		2686
		2687	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2688	paths, although this is fully supported by libev.
		2689
		2690	=head3 The special problem of stat time resolution
		2691
		2692	The C<stat ()> system call only supports full-second resolution portably,
		2693	and even on systems where the resolution is higher, most file systems
		2694	still only support whole seconds.
		2695
		2696	That means that, if the time is the only thing that changes, you can
		2697	easily miss updates: on the first update, C<ev_stat> detects a change and
		2698	calls your callback, which does something. When there is another update
		2699	within the same second, C<ev_stat> will be unable to detect unless the
		2700	stat data does change in other ways (e.g. file size).
		2701
		2702	The solution to this is to delay acting on a change for slightly more
		2703	than a second (or till slightly after the next full second boundary), using
		2704	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2705	ev_timer_again (loop, w)>).
		2706
		2707	The C<.02> offset is added to work around small timing inconsistencies
		2708	of some operating systems (where the second counter of the current time
		2709	might be be delayed. One such system is the Linux kernel, where a call to
		2710	C<gettimeofday> might return a timestamp with a full second later than
		2711	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2712	update file times then there will be a small window where the kernel uses
		2713	the previous second to update file times but libev might already execute
		2714	the timer callback).
		2715
		2716	=head3 Watcher-Specific Functions and Data Members
		2717
		2718	=over 4
		2719
		2720	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2721
		2722	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2723
		2724	Configures the watcher to wait for status changes of the given
		2725	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2726	be detected and should normally be specified as C<0> to let libev choose
		2727	a suitable value. The memory pointed to by C<path> must point to the same
		2728	path for as long as the watcher is active.
		2729
		2730	The callback will receive an C<EV_STAT> event when a change was detected,
		2731	relative to the attributes at the time the watcher was started (or the
		2732	last change was detected).
		2733
		2734	=item ev_stat_stat (loop, ev_stat *)
		2735
		2736	Updates the stat buffer immediately with new values. If you change the
		2737	watched path in your callback, you could call this function to avoid
		2738	detecting this change (while introducing a race condition if you are not
		2739	the only one changing the path). Can also be useful simply to find out the
		2740	new values.
		2741
		2742	=item ev_statdata attr [read-only]
		2743
		2744	The most-recently detected attributes of the file. Although the type is
		2745	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2746	suitable for your system, but you can only rely on the POSIX-standardised
		2747	members to be present. If the C<st_nlink> member is C<0>, then there was
		2748	some error while C<stat>ing the file.
		2749
		2750	=item ev_statdata prev [read-only]
		2751
		2752	The previous attributes of the file. The callback gets invoked whenever
		2753	C<prev> != C<attr>, or, more precisely, one or more of these members
		2754	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2755	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2756
		2757	=item ev_tstamp interval [read-only]
		2758
		2759	The specified interval.
		2760
		2761	=item const char *path [read-only]
		2762
		2763	The file system path that is being watched.
		2764
		2765	=back
		2766
		2767	=head3 Examples
		2768
		2769	Example: Watch C</etc/passwd> for attribute changes.
		2770
		2771	static void
		2772	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2773	{
		2774	/* /etc/passwd changed in some way */
		2775	if (w->attr.st_nlink)
		2776	{
		2777	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2778	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2779	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2780	}
		2781	else
		2782	/* you shalt not abuse printf for puts */
		2783	puts ("wow, /etc/passwd is not there, expect problems. "
		2784	"if this is windows, they already arrived\n");
		2785	}
		2786
		2787	...
		2788	ev_stat passwd;
		2789
		2790	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2791	ev_stat_start (loop, &passwd);
		2792
		2793	Example: Like above, but additionally use a one-second delay so we do not
		2794	miss updates (however, frequent updates will delay processing, too, so
		2795	one might do the work both on C<ev_stat> callback invocation I<and> on
		2796	C<ev_timer> callback invocation).
		2797
		2798	static ev_stat passwd;
		2799	static ev_timer timer;
		2800
		2801	static void
		2802	timer_cb (EV_P_ ev_timer *w, int revents)
		2803	{
		2804	ev_timer_stop (EV_A_ w);
		2805
		2806	/* now it's one second after the most recent passwd change */
		2807	}
		2808
		2809	static void
		2810	stat_cb (EV_P_ ev_stat *w, int revents)
		2811	{
		2812	/* reset the one-second timer */
		2813	ev_timer_again (EV_A_ &timer);
		2814	}
		2815
		2816	...
		2817	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2818	ev_stat_start (loop, &passwd);
		2819	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2820
		2821
725	=head2 C<ev_idle> - when you've got nothing better to do	2822	=head2 C<ev_idle> - when you've got nothing better to do...
726		2823
727	Idle watchers trigger events when there are no other events are pending	2824	Idle watchers trigger events when no other events of the same or higher
728	(prepare, check and other idle watchers do not count). That is, as long	2825	priority are pending (prepare, check and other idle watchers do not count
729	as your process is busy handling sockets or timeouts (or even signals,	2826	as receiving "events").
730	imagine) it will not be triggered. But when your process is idle all idle	2827
731	watchers are being called again and again, once per event loop iteration -	2828	That is, as long as your process is busy handling sockets or timeouts
		2829	(or even signals, imagine) of the same or higher priority it will not be
		2830	triggered. But when your process is idle (or only lower-priority watchers
		2831	are pending), the idle watchers are being called once per event loop
732	until stopped, that is, or your process receives more events and becomes	2832	iteration - until stopped, that is, or your process receives more events
733	busy.	2833	and becomes busy again with higher priority stuff.
734		2834
735	The most noteworthy effect is that as long as any idle watchers are	2835	The most noteworthy effect is that as long as any idle watchers are
736	active, the process will not block when waiting for new events.	2836	active, the process will not block when waiting for new events.
737		2837
738	Apart from keeping your process non-blocking (which is a useful	2838	Apart from keeping your process non-blocking (which is a useful
739	effect on its own sometimes), idle watchers are a good place to do	2839	effect on its own sometimes), idle watchers are a good place to do
740	"pseudo-background processing", or delay processing stuff to after the	2840	"pseudo-background processing", or delay processing stuff to after the
741	event loop has handled all outstanding events.	2841	event loop has handled all outstanding events.
742		2842
		2843	=head3 Watcher-Specific Functions and Data Members
		2844
743	=over 4	2845	=over 4
744		2846
745	=item ev_idle_init (ev_signal *, callback)	2847	=item ev_idle_init (ev_idle *, callback)
746		2848
747	Initialises and configures the idle watcher - it has no parameters of any	2849	Initialises and configures the idle watcher - it has no parameters of any
748	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2850	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
749	believe me.	2851	believe me.
750		2852
751	=back	2853	=back
752		2854
		2855	=head3 Examples
		2856
		2857	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2858	callback, free it. Also, use no error checking, as usual.
		2859
		2860	static void
		2861	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2862	{
		2863	free (w);
		2864	// now do something you wanted to do when the program has
		2865	// no longer anything immediate to do.
		2866	}
		2867
		2868	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2869	ev_idle_init (idle_watcher, idle_cb);
		2870	ev_idle_start (loop, idle_watcher);
		2871
		2872
753	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2873	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
754		2874
755	Prepare and check watchers are usually (but not always) used in tandem:	2875	Prepare and check watchers are usually (but not always) used in pairs:
756	prepare watchers get invoked before the process blocks and check watchers	2876	prepare watchers get invoked before the process blocks and check watchers
757	afterwards.	2877	afterwards.
758		2878
		2879	You I<must not> call C<ev_run> or similar functions that enter
		2880	the current event loop from either C<ev_prepare> or C<ev_check>
		2881	watchers. Other loops than the current one are fine, however. The
		2882	rationale behind this is that you do not need to check for recursion in
		2883	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2884	C<ev_check> so if you have one watcher of each kind they will always be
		2885	called in pairs bracketing the blocking call.
		2886
759	Their main purpose is to integrate other event mechanisms into libev. This	2887	Their main purpose is to integrate other event mechanisms into libev and
760	could be used, for example, to track variable changes, implement your own	2888	their use is somewhat advanced. They could be used, for example, to track
761	watchers, integrate net-snmp or a coroutine library and lots more.	2889	variable changes, implement your own watchers, integrate net-snmp or a
		2890	coroutine library and lots more. They are also occasionally useful if
		2891	you cache some data and want to flush it before blocking (for example,
		2892	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2893	watcher).
762		2894
763	This is done by examining in each prepare call which file descriptors need	2895	This is done by examining in each prepare call which file descriptors
764	to be watched by the other library, registering C<ev_io> watchers for	2896	need to be watched by the other library, registering C<ev_io> watchers
765	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2897	for them and starting an C<ev_timer> watcher for any timeouts (many
766	provide just this functionality). Then, in the check watcher you check for	2898	libraries provide exactly this functionality). Then, in the check watcher,
767	any events that occured (by checking the pending status of all watchers	2899	you check for any events that occurred (by checking the pending status
768	and stopping them) and call back into the library. The I/O and timer	2900	of all watchers and stopping them) and call back into the library. The
769	callbacks will never actually be called (but must be valid nevertheless,	2901	I/O and timer callbacks will never actually be called (but must be valid
770	because you never know, you know?).	2902	nevertheless, because you never know, you know?).
771		2903
772	As another example, the Perl Coro module uses these hooks to integrate	2904	As another example, the Perl Coro module uses these hooks to integrate
773	coroutines into libev programs, by yielding to other active coroutines	2905	coroutines into libev programs, by yielding to other active coroutines
774	during each prepare and only letting the process block if no coroutines	2906	during each prepare and only letting the process block if no coroutines
775	are ready to run (it's actually more complicated: it only runs coroutines	2907	are ready to run (it's actually more complicated: it only runs coroutines
776	with priority higher than or equal to the event loop and one coroutine	2908	with priority higher than or equal to the event loop and one coroutine
777	of lower priority, but only once, using idle watchers to keep the event	2909	of lower priority, but only once, using idle watchers to keep the event
778	loop from blocking if lower-priority coroutines are active, thus mapping	2910	loop from blocking if lower-priority coroutines are active, thus mapping
779	low-priority coroutines to idle/background tasks).	2911	low-priority coroutines to idle/background tasks).
780		2912
		2913	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2914	priority, to ensure that they are being run before any other watchers
		2915	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2916
		2917	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2918	activate ("feed") events into libev. While libev fully supports this, they
		2919	might get executed before other C<ev_check> watchers did their job. As
		2920	C<ev_check> watchers are often used to embed other (non-libev) event
		2921	loops those other event loops might be in an unusable state until their
		2922	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2923	others).
		2924
		2925	=head3 Watcher-Specific Functions and Data Members
		2926
781	=over 4	2927	=over 4
782		2928
783	=item ev_prepare_init (ev_prepare *, callback)	2929	=item ev_prepare_init (ev_prepare *, callback)
784		2930
785	=item ev_check_init (ev_check *, callback)	2931	=item ev_check_init (ev_check *, callback)
786		2932
787	Initialises and configures the prepare or check watcher - they have no	2933	Initialises and configures the prepare or check watcher - they have no
788	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2934	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
789	macros, but using them is utterly, utterly and completely pointless.	2935	macros, but using them is utterly, utterly, utterly and completely
		2936	pointless.
790		2937
791	=back	2938	=back
792		2939
		2940	=head3 Examples
		2941
		2942	There are a number of principal ways to embed other event loops or modules
		2943	into libev. Here are some ideas on how to include libadns into libev
		2944	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2945	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2946	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2947	Glib event loop).
		2948
		2949	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2950	and in a check watcher, destroy them and call into libadns. What follows
		2951	is pseudo-code only of course. This requires you to either use a low
		2952	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2953	the callbacks for the IO/timeout watchers might not have been called yet.
		2954
		2955	static ev_io iow [nfd];
		2956	static ev_timer tw;
		2957
		2958	static void
		2959	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2960	{
		2961	}
		2962
		2963	// create io watchers for each fd and a timer before blocking
		2964	static void
		2965	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2966	{
		2967	int timeout = 3600000;
		2968	struct pollfd fds [nfd];
		2969	// actual code will need to loop here and realloc etc.
		2970	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2971
		2972	/* the callback is illegal, but won't be called as we stop during check */
		2973	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2974	ev_timer_start (loop, &tw);
		2975
		2976	// create one ev_io per pollfd
		2977	for (int i = 0; i < nfd; ++i)
		2978	{
		2979	ev_io_init (iow + i, io_cb, fds [i].fd,
		2980	((fds [i].events & POLLIN ? EV_READ : 0)
		2981	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2982
		2983	fds [i].revents = 0;
		2984	ev_io_start (loop, iow + i);
		2985	}
		2986	}
		2987
		2988	// stop all watchers after blocking
		2989	static void
		2990	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2991	{
		2992	ev_timer_stop (loop, &tw);
		2993
		2994	for (int i = 0; i < nfd; ++i)
		2995	{
		2996	// set the relevant poll flags
		2997	// could also call adns_processreadable etc. here
		2998	struct pollfd *fd = fds + i;
		2999	int revents = ev_clear_pending (iow + i);
		3000	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		3001	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		3002
		3003	// now stop the watcher
		3004	ev_io_stop (loop, iow + i);
		3005	}
		3006
		3007	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		3008	}
		3009
		3010	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		3011	in the prepare watcher and would dispose of the check watcher.
		3012
		3013	Method 3: If the module to be embedded supports explicit event
		3014	notification (libadns does), you can also make use of the actual watcher
		3015	callbacks, and only destroy/create the watchers in the prepare watcher.
		3016
		3017	static void
		3018	timer_cb (EV_P_ ev_timer *w, int revents)
		3019	{
		3020	adns_state ads = (adns_state)w->data;
		3021	update_now (EV_A);
		3022
		3023	adns_processtimeouts (ads, &tv_now);
		3024	}
		3025
		3026	static void
		3027	io_cb (EV_P_ ev_io *w, int revents)
		3028	{
		3029	adns_state ads = (adns_state)w->data;
		3030	update_now (EV_A);
		3031
		3032	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		3033	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		3034	}
		3035
		3036	// do not ever call adns_afterpoll
		3037
		3038	Method 4: Do not use a prepare or check watcher because the module you
		3039	want to embed is not flexible enough to support it. Instead, you can
		3040	override their poll function. The drawback with this solution is that the
		3041	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		3042	this approach, effectively embedding EV as a client into the horrible
		3043	libglib event loop.
		3044
		3045	static gint
		3046	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		3047	{
		3048	int got_events = 0;
		3049
		3050	for (n = 0; n < nfds; ++n)
		3051	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		3052
		3053	if (timeout >= 0)
		3054	// create/start timer
		3055
		3056	// poll
		3057	ev_run (EV_A_ 0);
		3058
		3059	// stop timer again
		3060	if (timeout >= 0)
		3061	ev_timer_stop (EV_A_ &to);
		3062
		3063	// stop io watchers again - their callbacks should have set
		3064	for (n = 0; n < nfds; ++n)
		3065	ev_io_stop (EV_A_ iow [n]);
		3066
		3067	return got_events;
		3068	}
		3069
		3070
		3071	=head2 C<ev_embed> - when one backend isn't enough...
		3072
		3073	This is a rather advanced watcher type that lets you embed one event loop
		3074	into another (currently only C<ev_io> events are supported in the embedded
		3075	loop, other types of watchers might be handled in a delayed or incorrect
		3076	fashion and must not be used).
		3077
		3078	There are primarily two reasons you would want that: work around bugs and
		3079	prioritise I/O.
		3080
		3081	As an example for a bug workaround, the kqueue backend might only support
		3082	sockets on some platform, so it is unusable as generic backend, but you
		3083	still want to make use of it because you have many sockets and it scales
		3084	so nicely. In this case, you would create a kqueue-based loop and embed
		3085	it into your default loop (which might use e.g. poll). Overall operation
		3086	will be a bit slower because first libev has to call C<poll> and then
		3087	C<kevent>, but at least you can use both mechanisms for what they are
		3088	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		3089
		3090	As for prioritising I/O: under rare circumstances you have the case where
		3091	some fds have to be watched and handled very quickly (with low latency),
		3092	and even priorities and idle watchers might have too much overhead. In
		3093	this case you would put all the high priority stuff in one loop and all
		3094	the rest in a second one, and embed the second one in the first.
		3095
		3096	As long as the watcher is active, the callback will be invoked every
		3097	time there might be events pending in the embedded loop. The callback
		3098	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		3099	sweep and invoke their callbacks (the callback doesn't need to invoke the
		3100	C<ev_embed_sweep> function directly, it could also start an idle watcher
		3101	to give the embedded loop strictly lower priority for example).
		3102
		3103	You can also set the callback to C<0>, in which case the embed watcher
		3104	will automatically execute the embedded loop sweep whenever necessary.
		3105
		3106	Fork detection will be handled transparently while the C<ev_embed> watcher
		3107	is active, i.e., the embedded loop will automatically be forked when the
		3108	embedding loop forks. In other cases, the user is responsible for calling
		3109	C<ev_loop_fork> on the embedded loop.
		3110
		3111	Unfortunately, not all backends are embeddable: only the ones returned by
		3112	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		3113	portable one.
		3114
		3115	So when you want to use this feature you will always have to be prepared
		3116	that you cannot get an embeddable loop. The recommended way to get around
		3117	this is to have a separate variables for your embeddable loop, try to
		3118	create it, and if that fails, use the normal loop for everything.
		3119
		3120	=head3 C<ev_embed> and fork
		3121
		3122	While the C<ev_embed> watcher is running, forks in the embedding loop will
		3123	automatically be applied to the embedded loop as well, so no special
		3124	fork handling is required in that case. When the watcher is not running,
		3125	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		3126	as applicable.
		3127
		3128	=head3 Watcher-Specific Functions and Data Members
		3129
		3130	=over 4
		3131
		3132	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		3133
		3134	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		3135
		3136	Configures the watcher to embed the given loop, which must be
		3137	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		3138	invoked automatically, otherwise it is the responsibility of the callback
		3139	to invoke it (it will continue to be called until the sweep has been done,
		3140	if you do not want that, you need to temporarily stop the embed watcher).
		3141
		3142	=item ev_embed_sweep (loop, ev_embed *)
		3143
		3144	Make a single, non-blocking sweep over the embedded loop. This works
		3145	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
		3146	appropriate way for embedded loops.
		3147
		3148	=item struct ev_loop *other [read-only]
		3149
		3150	The embedded event loop.
		3151
		3152	=back
		3153
		3154	=head3 Examples
		3155
		3156	Example: Try to get an embeddable event loop and embed it into the default
		3157	event loop. If that is not possible, use the default loop. The default
		3158	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		3159	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		3160	used).
		3161
		3162	struct ev_loop *loop_hi = ev_default_init (0);
		3163	struct ev_loop *loop_lo = 0;
		3164	ev_embed embed;
		3165
		3166	// see if there is a chance of getting one that works
		3167	// (remember that a flags value of 0 means autodetection)
		3168	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3169	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3170	: 0;
		3171
		3172	// if we got one, then embed it, otherwise default to loop_hi
		3173	if (loop_lo)
		3174	{
		3175	ev_embed_init (&embed, 0, loop_lo);
		3176	ev_embed_start (loop_hi, &embed);
		3177	}
		3178	else
		3179	loop_lo = loop_hi;
		3180
		3181	Example: Check if kqueue is available but not recommended and create
		3182	a kqueue backend for use with sockets (which usually work with any
		3183	kqueue implementation). Store the kqueue/socket-only event loop in
		3184	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3185
		3186	struct ev_loop *loop = ev_default_init (0);
		3187	struct ev_loop *loop_socket = 0;
		3188	ev_embed embed;
		3189
		3190	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3191	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3192	{
		3193	ev_embed_init (&embed, 0, loop_socket);
		3194	ev_embed_start (loop, &embed);
		3195	}
		3196
		3197	if (!loop_socket)
		3198	loop_socket = loop;
		3199
		3200	// now use loop_socket for all sockets, and loop for everything else
		3201
		3202
		3203	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		3204
		3205	Fork watchers are called when a C<fork ()> was detected (usually because
		3206	whoever is a good citizen cared to tell libev about it by calling
		3207	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		3208	event loop blocks next and before C<ev_check> watchers are being called,
		3209	and only in the child after the fork. If whoever good citizen calling
		3210	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		3211	handlers will be invoked, too, of course.
		3212
		3213	=head3 The special problem of life after fork - how is it possible?
		3214
		3215	Most uses of C<fork()> consist of forking, then some simple calls to set
		3216	up/change the process environment, followed by a call to C<exec()>. This
		3217	sequence should be handled by libev without any problems.
		3218
		3219	This changes when the application actually wants to do event handling
		3220	in the child, or both parent in child, in effect "continuing" after the
		3221	fork.
		3222
		3223	The default mode of operation (for libev, with application help to detect
		3224	forks) is to duplicate all the state in the child, as would be expected
		3225	when I<either> the parent I<or> the child process continues.
		3226
		3227	When both processes want to continue using libev, then this is usually the
		3228	wrong result. In that case, usually one process (typically the parent) is
		3229	supposed to continue with all watchers in place as before, while the other
		3230	process typically wants to start fresh, i.e. without any active watchers.
		3231
		3232	The cleanest and most efficient way to achieve that with libev is to
		3233	simply create a new event loop, which of course will be "empty", and
		3234	use that for new watchers. This has the advantage of not touching more
		3235	memory than necessary, and thus avoiding the copy-on-write, and the
		3236	disadvantage of having to use multiple event loops (which do not support
		3237	signal watchers).
		3238
		3239	When this is not possible, or you want to use the default loop for
		3240	other reasons, then in the process that wants to start "fresh", call
		3241	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3242	Destroying the default loop will "orphan" (not stop) all registered
		3243	watchers, so you have to be careful not to execute code that modifies
		3244	those watchers. Note also that in that case, you have to re-register any
		3245	signal watchers.
		3246
		3247	=head3 Watcher-Specific Functions and Data Members
		3248
		3249	=over 4
		3250
		3251	=item ev_fork_init (ev_fork *, callback)
		3252
		3253	Initialises and configures the fork watcher - it has no parameters of any
		3254	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		3255	really.
		3256
		3257	=back
		3258
		3259
		3260	=head2 C<ev_cleanup> - even the best things end
		3261
		3262	Cleanup watchers are called just before the event loop is being destroyed
		3263	by a call to C<ev_loop_destroy>.
		3264
		3265	While there is no guarantee that the event loop gets destroyed, cleanup
		3266	watchers provide a convenient method to install cleanup hooks for your
		3267	program, worker threads and so on - you just to make sure to destroy the
		3268	loop when you want them to be invoked.
		3269
		3270	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3271	all other watchers, they do not keep a reference to the event loop (which
		3272	makes a lot of sense if you think about it). Like all other watchers, you
		3273	can call libev functions in the callback, except C<ev_cleanup_start>.
		3274
		3275	=head3 Watcher-Specific Functions and Data Members
		3276
		3277	=over 4
		3278
		3279	=item ev_cleanup_init (ev_cleanup *, callback)
		3280
		3281	Initialises and configures the cleanup watcher - it has no parameters of
		3282	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3283	pointless, I assure you.
		3284
		3285	=back
		3286
		3287	Example: Register an atexit handler to destroy the default loop, so any
		3288	cleanup functions are called.
		3289
		3290	static void
		3291	program_exits (void)
		3292	{
		3293	ev_loop_destroy (EV_DEFAULT_UC);
		3294	}
		3295
		3296	...
		3297	atexit (program_exits);
		3298
		3299
		3300	=head2 C<ev_async> - how to wake up an event loop
		3301
		3302	In general, you cannot use an C<ev_loop> from multiple threads or other
		3303	asynchronous sources such as signal handlers (as opposed to multiple event
		3304	loops - those are of course safe to use in different threads).
		3305
		3306	Sometimes, however, you need to wake up an event loop you do not control,
		3307	for example because it belongs to another thread. This is what C<ev_async>
		3308	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3309	it by calling C<ev_async_send>, which is thread- and signal safe.
		3310
		3311	This functionality is very similar to C<ev_signal> watchers, as signals,
		3312	too, are asynchronous in nature, and signals, too, will be compressed
		3313	(i.e. the number of callback invocations may be less than the number of
		3314	C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
		3315	of "global async watchers" by using a watcher on an otherwise unused
		3316	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3317	even without knowing which loop owns the signal.
		3318
		3319	=head3 Queueing
		3320
		3321	C<ev_async> does not support queueing of data in any way. The reason
		3322	is that the author does not know of a simple (or any) algorithm for a
		3323	multiple-writer-single-reader queue that works in all cases and doesn't
		3324	need elaborate support such as pthreads or unportable memory access
		3325	semantics.
		3326
		3327	That means that if you want to queue data, you have to provide your own
		3328	queue. But at least I can tell you how to implement locking around your
		3329	queue:
		3330
		3331	=over 4
		3332
		3333	=item queueing from a signal handler context
		3334
		3335	To implement race-free queueing, you simply add to the queue in the signal
		3336	handler but you block the signal handler in the watcher callback. Here is
		3337	an example that does that for some fictitious SIGUSR1 handler:
		3338
		3339	static ev_async mysig;
		3340
		3341	static void
		3342	sigusr1_handler (void)
		3343	{
		3344	sometype data;
		3345
		3346	// no locking etc.
		3347	queue_put (data);
		3348	ev_async_send (EV_DEFAULT_ &mysig);
		3349	}
		3350
		3351	static void
		3352	mysig_cb (EV_P_ ev_async *w, int revents)
		3353	{
		3354	sometype data;
		3355	sigset_t block, prev;
		3356
		3357	sigemptyset (&block);
		3358	sigaddset (&block, SIGUSR1);
		3359	sigprocmask (SIG_BLOCK, &block, &prev);
		3360
		3361	while (queue_get (&data))
		3362	process (data);
		3363
		3364	if (sigismember (&prev, SIGUSR1)
		3365	sigprocmask (SIG_UNBLOCK, &block, 0);
		3366	}
		3367
		3368	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3369	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3370	either...).
		3371
		3372	=item queueing from a thread context
		3373
		3374	The strategy for threads is different, as you cannot (easily) block
		3375	threads but you can easily preempt them, so to queue safely you need to
		3376	employ a traditional mutex lock, such as in this pthread example:
		3377
		3378	static ev_async mysig;
		3379	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3380
		3381	static void
		3382	otherthread (void)
		3383	{
		3384	// only need to lock the actual queueing operation
		3385	pthread_mutex_lock (&mymutex);
		3386	queue_put (data);
		3387	pthread_mutex_unlock (&mymutex);
		3388
		3389	ev_async_send (EV_DEFAULT_ &mysig);
		3390	}
		3391
		3392	static void
		3393	mysig_cb (EV_P_ ev_async *w, int revents)
		3394	{
		3395	pthread_mutex_lock (&mymutex);
		3396
		3397	while (queue_get (&data))
		3398	process (data);
		3399
		3400	pthread_mutex_unlock (&mymutex);
		3401	}
		3402
		3403	=back
		3404
		3405
		3406	=head3 Watcher-Specific Functions and Data Members
		3407
		3408	=over 4
		3409
		3410	=item ev_async_init (ev_async *, callback)
		3411
		3412	Initialises and configures the async watcher - it has no parameters of any
		3413	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3414	trust me.
		3415
		3416	=item ev_async_send (loop, ev_async *)
		3417
		3418	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3419	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3420	returns.
		3421
		3422	Unlike C<ev_feed_event>, this call is safe to do from other threads,
		3423	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
		3424	embedding section below on what exactly this means).
		3425
		3426	Note that, as with other watchers in libev, multiple events might get
		3427	compressed into a single callback invocation (another way to look at
		3428	this is that C<ev_async> watchers are level-triggered: they are set on
		3429	C<ev_async_send>, reset when the event loop detects that).
		3430
		3431	This call incurs the overhead of at most one extra system call per event
		3432	loop iteration, if the event loop is blocked, and no syscall at all if
		3433	the event loop (or your program) is processing events. That means that
		3434	repeated calls are basically free (there is no need to avoid calls for
		3435	performance reasons) and that the overhead becomes smaller (typically
		3436	zero) under load.
		3437
		3438	=item bool = ev_async_pending (ev_async *)
		3439
		3440	Returns a non-zero value when C<ev_async_send> has been called on the
		3441	watcher but the event has not yet been processed (or even noted) by the
		3442	event loop.
		3443
		3444	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3445	the loop iterates next and checks for the watcher to have become active,
		3446	it will reset the flag again. C<ev_async_pending> can be used to very
		3447	quickly check whether invoking the loop might be a good idea.
		3448
		3449	Not that this does I<not> check whether the watcher itself is pending,
		3450	only whether it has been requested to make this watcher pending: there
		3451	is a time window between the event loop checking and resetting the async
		3452	notification, and the callback being invoked.
		3453
		3454	=back
		3455
		3456
793	=head1 OTHER FUNCTIONS	3457	=head1 OTHER FUNCTIONS
794		3458
795	There are some other functions of possible interest. Described. Here. Now.	3459	There are some other functions of possible interest. Described. Here. Now.
796		3460
797	=over 4	3461	=over 4
798		3462
799	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3463	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
800		3464
801	This function combines a simple timer and an I/O watcher, calls your	3465	This function combines a simple timer and an I/O watcher, calls your
802	callback on whichever event happens first and automatically stop both	3466	callback on whichever event happens first and automatically stops both
803	watchers. This is useful if you want to wait for a single event on an fd	3467	watchers. This is useful if you want to wait for a single event on an fd
804	or timeout without having to allocate/configure/start/stop/free one or	3468	or timeout without having to allocate/configure/start/stop/free one or
805	more watchers yourself.	3469	more watchers yourself.
806		3470
807	If C<fd> is less than 0, then no I/O watcher will be started and events	3471	If C<fd> is less than 0, then no I/O watcher will be started and the
808	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3472	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
809	C<events> set will be craeted and started.	3473	the given C<fd> and C<events> set will be created and started.
810		3474
811	If C<timeout> is less than 0, then no timeout watcher will be	3475	If C<timeout> is less than 0, then no timeout watcher will be
812	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3476	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
813	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3477	repeat = 0) will be started. C<0> is a valid timeout.
814	dubious value.
815		3478
816	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3479	The callback has the type C<void (*cb)(int revents, void *arg)> and is
817	passed an C<revents> set like normal event callbacks (a combination of	3480	passed an C<revents> set like normal event callbacks (a combination of
818	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3481	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
819	value passed to C<ev_once>:	3482	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3483	a timeout and an io event at the same time - you probably should give io
		3484	events precedence.
820		3485
		3486	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3487
821	static void stdin_ready (int revents, void *arg)	3488	static void stdin_ready (int revents, void *arg)
		3489	{
		3490	if (revents & EV_READ)
		3491	/* stdin might have data for us, joy! */;
		3492	else if (revents & EV_TIMER)
		3493	/* doh, nothing entered */;
		3494	}
		3495
		3496	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
		3497
		3498	=item ev_feed_fd_event (loop, int fd, int revents)
		3499
		3500	Feed an event on the given fd, as if a file descriptor backend detected
		3501	the given events.
		3502
		3503	=item ev_feed_signal_event (loop, int signum)
		3504
		3505	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
		3506	which is async-safe.
		3507
		3508	=back
		3509
		3510
		3511	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3512
		3513	This section explains some common idioms that are not immediately
		3514	obvious. Note that examples are sprinkled over the whole manual, and this
		3515	section only contains stuff that wouldn't fit anywhere else.
		3516
		3517	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3518
		3519	Each watcher has, by default, a C<void *data> member that you can read
		3520	or modify at any time: libev will completely ignore it. This can be used
		3521	to associate arbitrary data with your watcher. If you need more data and
		3522	don't want to allocate memory separately and store a pointer to it in that
		3523	data member, you can also "subclass" the watcher type and provide your own
		3524	data:
		3525
		3526	struct my_io
		3527	{
		3528	ev_io io;
		3529	int otherfd;
		3530	void *somedata;
		3531	struct whatever *mostinteresting;
		3532	};
		3533
		3534	...
		3535	struct my_io w;
		3536	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3537
		3538	And since your callback will be called with a pointer to the watcher, you
		3539	can cast it back to your own type:
		3540
		3541	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3542	{
		3543	struct my_io *w = (struct my_io *)w_;
		3544	...
		3545	}
		3546
		3547	More interesting and less C-conformant ways of casting your callback
		3548	function type instead have been omitted.
		3549
		3550	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3551
		3552	Another common scenario is to use some data structure with multiple
		3553	embedded watchers, in effect creating your own watcher that combines
		3554	multiple libev event sources into one "super-watcher":
		3555
		3556	struct my_biggy
		3557	{
		3558	int some_data;
		3559	ev_timer t1;
		3560	ev_timer t2;
		3561	}
		3562
		3563	In this case getting the pointer to C<my_biggy> is a bit more
		3564	complicated: Either you store the address of your C<my_biggy> struct in
		3565	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3566	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3567	real programmers):
		3568
		3569	#include <stddef.h>
		3570
		3571	static void
		3572	t1_cb (EV_P_ ev_timer *w, int revents)
		3573	{
		3574	struct my_biggy big = (struct my_biggy *)
		3575	(((char *)w) - offsetof (struct my_biggy, t1));
		3576	}
		3577
		3578	static void
		3579	t2_cb (EV_P_ ev_timer *w, int revents)
		3580	{
		3581	struct my_biggy big = (struct my_biggy *)
		3582	(((char *)w) - offsetof (struct my_biggy, t2));
		3583	}
		3584
		3585	=head2 AVOIDING FINISHING BEFORE RETURNING
		3586
		3587	Often you have structures like this in event-based programs:
		3588
		3589	callback ()
822	{	3590	{
823	if (revents & EV_TIMEOUT)	3591	free (request);
824	/* doh, nothing entered */;
825	else if (revents & EV_READ)
826	/* stdin might have data for us, joy! */;
827	}	3592	}
828		3593
829	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3594	request = start_new_request (..., callback);
830		3595
831	=item ev_feed_event (loop, watcher, int events)	3596	The intent is to start some "lengthy" operation. The C<request> could be
		3597	used to cancel the operation, or do other things with it.
832		3598
833	Feeds the given event set into the event loop, as if the specified event	3599	It's not uncommon to have code paths in C<start_new_request> that
834	had happened for the specified watcher (which must be a pointer to an	3600	immediately invoke the callback, for example, to report errors. Or you add
835	initialised but not necessarily started event watcher).	3601	some caching layer that finds that it can skip the lengthy aspects of the
		3602	operation and simply invoke the callback with the result.
836		3603
837	=item ev_feed_fd_event (loop, int fd, int revents)	3604	The problem here is that this will happen I<before> C<start_new_request>
		3605	has returned, so C<request> is not set.
838		3606
839	Feed an event on the given fd, as if a file descriptor backend detected	3607	Even if you pass the request by some safer means to the callback, you
840	the given events it.	3608	might want to do something to the request after starting it, such as
		3609	canceling it, which probably isn't working so well when the callback has
		3610	already been invoked.
841		3611
842	=item ev_feed_signal_event (loop, int signum)	3612	A common way around all these issues is to make sure that
		3613	C<start_new_request> I<always> returns before the callback is invoked. If
		3614	C<start_new_request> immediately knows the result, it can artificially
		3615	delay invoking the callback by e.g. using a C<prepare> or C<idle> watcher
		3616	for example, or more sneakily, by reusing an existing (stopped) watcher
		3617	and pushing it into the pending queue:
843		3618
844	Feed an event as if the given signal occured (loop must be the default loop!).	3619	ev_set_cb (watcher, callback);
		3620	ev_feed_event (EV_A_ watcher, 0);
845		3621
846	=back	3622	This way, C<start_new_request> can safely return before the callback is
		3623	invoked, while not delaying callback invocation too much.
		3624
		3625	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3626
		3627	Often (especially in GUI toolkits) there are places where you have
		3628	I<modal> interaction, which is most easily implemented by recursively
		3629	invoking C<ev_run>.
		3630
		3631	This brings the problem of exiting - a callback might want to finish the
		3632	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3633	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3634	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3635	other combination: In these cases, C<ev_break> will not work alone.
		3636
		3637	The solution is to maintain "break this loop" variable for each C<ev_run>
		3638	invocation, and use a loop around C<ev_run> until the condition is
		3639	triggered, using C<EVRUN_ONCE>:
		3640
		3641	// main loop
		3642	int exit_main_loop = 0;
		3643
		3644	while (!exit_main_loop)
		3645	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3646
		3647	// in a modal watcher
		3648	int exit_nested_loop = 0;
		3649
		3650	while (!exit_nested_loop)
		3651	ev_run (EV_A_ EVRUN_ONCE);
		3652
		3653	To exit from any of these loops, just set the corresponding exit variable:
		3654
		3655	// exit modal loop
		3656	exit_nested_loop = 1;
		3657
		3658	// exit main program, after modal loop is finished
		3659	exit_main_loop = 1;
		3660
		3661	// exit both
		3662	exit_main_loop = exit_nested_loop = 1;
		3663
		3664	=head2 THREAD LOCKING EXAMPLE
		3665
		3666	Here is a fictitious example of how to run an event loop in a different
		3667	thread from where callbacks are being invoked and watchers are
		3668	created/added/removed.
		3669
		3670	For a real-world example, see the C<EV::Loop::Async> perl module,
		3671	which uses exactly this technique (which is suited for many high-level
		3672	languages).
		3673
		3674	The example uses a pthread mutex to protect the loop data, a condition
		3675	variable to wait for callback invocations, an async watcher to notify the
		3676	event loop thread and an unspecified mechanism to wake up the main thread.
		3677
		3678	First, you need to associate some data with the event loop:
		3679
		3680	typedef struct {
		3681	mutex_t lock; /* global loop lock */
		3682	ev_async async_w;
		3683	thread_t tid;
		3684	cond_t invoke_cv;
		3685	} userdata;
		3686
		3687	void prepare_loop (EV_P)
		3688	{
		3689	// for simplicity, we use a static userdata struct.
		3690	static userdata u;
		3691
		3692	ev_async_init (&u->async_w, async_cb);
		3693	ev_async_start (EV_A_ &u->async_w);
		3694
		3695	pthread_mutex_init (&u->lock, 0);
		3696	pthread_cond_init (&u->invoke_cv, 0);
		3697
		3698	// now associate this with the loop
		3699	ev_set_userdata (EV_A_ u);
		3700	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3701	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3702
		3703	// then create the thread running ev_run
		3704	pthread_create (&u->tid, 0, l_run, EV_A);
		3705	}
		3706
		3707	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3708	solely to wake up the event loop so it takes notice of any new watchers
		3709	that might have been added:
		3710
		3711	static void
		3712	async_cb (EV_P_ ev_async *w, int revents)
		3713	{
		3714	// just used for the side effects
		3715	}
		3716
		3717	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3718	protecting the loop data, respectively.
		3719
		3720	static void
		3721	l_release (EV_P)
		3722	{
		3723	userdata *u = ev_userdata (EV_A);
		3724	pthread_mutex_unlock (&u->lock);
		3725	}
		3726
		3727	static void
		3728	l_acquire (EV_P)
		3729	{
		3730	userdata *u = ev_userdata (EV_A);
		3731	pthread_mutex_lock (&u->lock);
		3732	}
		3733
		3734	The event loop thread first acquires the mutex, and then jumps straight
		3735	into C<ev_run>:
		3736
		3737	void *
		3738	l_run (void *thr_arg)
		3739	{
		3740	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3741
		3742	l_acquire (EV_A);
		3743	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3744	ev_run (EV_A_ 0);
		3745	l_release (EV_A);
		3746
		3747	return 0;
		3748	}
		3749
		3750	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3751	signal the main thread via some unspecified mechanism (signals? pipe
		3752	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3753	have been called (in a while loop because a) spurious wakeups are possible
		3754	and b) skipping inter-thread-communication when there are no pending
		3755	watchers is very beneficial):
		3756
		3757	static void
		3758	l_invoke (EV_P)
		3759	{
		3760	userdata *u = ev_userdata (EV_A);
		3761
		3762	while (ev_pending_count (EV_A))
		3763	{
		3764	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3765	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3766	}
		3767	}
		3768
		3769	Now, whenever the main thread gets told to invoke pending watchers, it
		3770	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3771	thread to continue:
		3772
		3773	static void
		3774	real_invoke_pending (EV_P)
		3775	{
		3776	userdata *u = ev_userdata (EV_A);
		3777
		3778	pthread_mutex_lock (&u->lock);
		3779	ev_invoke_pending (EV_A);
		3780	pthread_cond_signal (&u->invoke_cv);
		3781	pthread_mutex_unlock (&u->lock);
		3782	}
		3783
		3784	Whenever you want to start/stop a watcher or do other modifications to an
		3785	event loop, you will now have to lock:
		3786
		3787	ev_timer timeout_watcher;
		3788	userdata *u = ev_userdata (EV_A);
		3789
		3790	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3791
		3792	pthread_mutex_lock (&u->lock);
		3793	ev_timer_start (EV_A_ &timeout_watcher);
		3794	ev_async_send (EV_A_ &u->async_w);
		3795	pthread_mutex_unlock (&u->lock);
		3796
		3797	Note that sending the C<ev_async> watcher is required because otherwise
		3798	an event loop currently blocking in the kernel will have no knowledge
		3799	about the newly added timer. By waking up the loop it will pick up any new
		3800	watchers in the next event loop iteration.
		3801
		3802	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3803
		3804	While the overhead of a callback that e.g. schedules a thread is small, it
		3805	is still an overhead. If you embed libev, and your main usage is with some
		3806	kind of threads or coroutines, you might want to customise libev so that
		3807	doesn't need callbacks anymore.
		3808
		3809	Imagine you have coroutines that you can switch to using a function
		3810	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3811	and that due to some magic, the currently active coroutine is stored in a
		3812	global called C<current_coro>. Then you can build your own "wait for libev
		3813	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3814	the differing C<;> conventions):
		3815
		3816	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3817	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3818
		3819	That means instead of having a C callback function, you store the
		3820	coroutine to switch to in each watcher, and instead of having libev call
		3821	your callback, you instead have it switch to that coroutine.
		3822
		3823	A coroutine might now wait for an event with a function called
		3824	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		3825	matter when, or whether the watcher is active or not when this function is
		3826	called):
		3827
		3828	void
		3829	wait_for_event (ev_watcher *w)
		3830	{
		3831	ev_cb_set (w) = current_coro;
		3832	switch_to (libev_coro);
		3833	}
		3834
		3835	That basically suspends the coroutine inside C<wait_for_event> and
		3836	continues the libev coroutine, which, when appropriate, switches back to
		3837	this or any other coroutine.
		3838
		3839	You can do similar tricks if you have, say, threads with an event queue -
		3840	instead of storing a coroutine, you store the queue object and instead of
		3841	switching to a coroutine, you push the watcher onto the queue and notify
		3842	any waiters.
		3843
		3844	To embed libev, see L<EMBEDDING>, but in short, it's easiest to create two
		3845	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		3846
		3847	// my_ev.h
		3848	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3849	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb);
		3850	#include "../libev/ev.h"
		3851
		3852	// my_ev.c
		3853	#define EV_H "my_ev.h"
		3854	#include "../libev/ev.c"
		3855
		3856	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		3857	F<my_ev.c> into your project. When properly specifying include paths, you
		3858	can even use F<ev.h> as header file name directly.
		3859
847		3860
848	=head1 LIBEVENT EMULATION	3861	=head1 LIBEVENT EMULATION
849		3862
850	Libev offers a compatibility emulation layer for libevent. It cannot	3863	Libev offers a compatibility emulation layer for libevent. It cannot
851	emulate the internals of libevent, so here are some usage hints:	3864	emulate the internals of libevent, so here are some usage hints:
852		3865
853	=over 4	3866	=over 4
		3867
		3868	=item * Only the libevent-1.4.1-beta API is being emulated.
		3869
		3870	This was the newest libevent version available when libev was implemented,
		3871	and is still mostly unchanged in 2010.
854		3872
855	=item * Use it by including <event.h>, as usual.	3873	=item * Use it by including <event.h>, as usual.
856		3874
857	=item * The following members are fully supported: ev_base, ev_callback,	3875	=item * The following members are fully supported: ev_base, ev_callback,
858	ev_arg, ev_fd, ev_res, ev_events.	3876	ev_arg, ev_fd, ev_res, ev_events.
…		…
863		3881
864	=item * Priorities are not currently supported. Initialising priorities	3882	=item * Priorities are not currently supported. Initialising priorities
865	will fail and all watchers will have the same priority, even though there	3883	will fail and all watchers will have the same priority, even though there
866	is an ev_pri field.	3884	is an ev_pri field.
867		3885
		3886	=item * In libevent, the last base created gets the signals, in libev, the
		3887	base that registered the signal gets the signals.
		3888
868	=item * Other members are not supported.	3889	=item * Other members are not supported.
869		3890
870	=item * The libev emulation is I<not> ABI compatible to libevent, you need	3891	=item * The libev emulation is I<not> ABI compatible to libevent, you need
871	to use the libev header file and library.	3892	to use the libev header file and library.
872		3893
873	=back	3894	=back
874		3895
875	=head1 C++ SUPPORT	3896	=head1 C++ SUPPORT
876		3897
877	TBD.	3898	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3899	you to use some convenience methods to start/stop watchers and also change
		3900	the callback model to a model using method callbacks on objects.
		3901
		3902	To use it,
		3903
		3904	#include <ev++.h>
		3905
		3906	This automatically includes F<ev.h> and puts all of its definitions (many
		3907	of them macros) into the global namespace. All C++ specific things are
		3908	put into the C<ev> namespace. It should support all the same embedding
		3909	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3910
		3911	Care has been taken to keep the overhead low. The only data member the C++
		3912	classes add (compared to plain C-style watchers) is the event loop pointer
		3913	that the watcher is associated with (or no additional members at all if
		3914	you disable C<EV_MULTIPLICITY> when embedding libev).
		3915
		3916	Currently, functions, static and non-static member functions and classes
		3917	with C<operator ()> can be used as callbacks. Other types should be easy
		3918	to add as long as they only need one additional pointer for context. If
		3919	you need support for other types of functors please contact the author
		3920	(preferably after implementing it).
		3921
		3922	Here is a list of things available in the C<ev> namespace:
		3923
		3924	=over 4
		3925
		3926	=item C<ev::READ>, C<ev::WRITE> etc.
		3927
		3928	These are just enum values with the same values as the C<EV_READ> etc.
		3929	macros from F<ev.h>.
		3930
		3931	=item C<ev::tstamp>, C<ev::now>
		3932
		3933	Aliases to the same types/functions as with the C<ev_> prefix.
		3934
		3935	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3936
		3937	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3938	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3939	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3940	defined by many implementations.
		3941
		3942	All of those classes have these methods:
		3943
		3944	=over 4
		3945
		3946	=item ev::TYPE::TYPE ()
		3947
		3948	=item ev::TYPE::TYPE (loop)
		3949
		3950	=item ev::TYPE::~TYPE
		3951
		3952	The constructor (optionally) takes an event loop to associate the watcher
		3953	with. If it is omitted, it will use C<EV_DEFAULT>.
		3954
		3955	The constructor calls C<ev_init> for you, which means you have to call the
		3956	C<set> method before starting it.
		3957
		3958	It will not set a callback, however: You have to call the templated C<set>
		3959	method to set a callback before you can start the watcher.
		3960
		3961	(The reason why you have to use a method is a limitation in C++ which does
		3962	not allow explicit template arguments for constructors).
		3963
		3964	The destructor automatically stops the watcher if it is active.
		3965
		3966	=item w->set<class, &class::method> (object *)
		3967
		3968	This method sets the callback method to call. The method has to have a
		3969	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3970	first argument and the C<revents> as second. The object must be given as
		3971	parameter and is stored in the C<data> member of the watcher.
		3972
		3973	This method synthesizes efficient thunking code to call your method from
		3974	the C callback that libev requires. If your compiler can inline your
		3975	callback (i.e. it is visible to it at the place of the C<set> call and
		3976	your compiler is good :), then the method will be fully inlined into the
		3977	thunking function, making it as fast as a direct C callback.
		3978
		3979	Example: simple class declaration and watcher initialisation
		3980
		3981	struct myclass
		3982	{
		3983	void io_cb (ev::io &w, int revents) { }
		3984	}
		3985
		3986	myclass obj;
		3987	ev::io iow;
		3988	iow.set <myclass, &myclass::io_cb> (&obj);
		3989
		3990	=item w->set (object *)
		3991
		3992	This is a variation of a method callback - leaving out the method to call
		3993	will default the method to C<operator ()>, which makes it possible to use
		3994	functor objects without having to manually specify the C<operator ()> all
		3995	the time. Incidentally, you can then also leave out the template argument
		3996	list.
		3997
		3998	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3999	int revents)>.
		4000
		4001	See the method-C<set> above for more details.
		4002
		4003	Example: use a functor object as callback.
		4004
		4005	struct myfunctor
		4006	{
		4007	void operator() (ev::io &w, int revents)
		4008	{
		4009	...
		4010	}
		4011	}
		4012
		4013	myfunctor f;
		4014
		4015	ev::io w;
		4016	w.set (&f);
		4017
		4018	=item w->set<function> (void *data = 0)
		4019
		4020	Also sets a callback, but uses a static method or plain function as
		4021	callback. The optional C<data> argument will be stored in the watcher's
		4022	C<data> member and is free for you to use.
		4023
		4024	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		4025
		4026	See the method-C<set> above for more details.
		4027
		4028	Example: Use a plain function as callback.
		4029
		4030	static void io_cb (ev::io &w, int revents) { }
		4031	iow.set <io_cb> ();
		4032
		4033	=item w->set (loop)
		4034
		4035	Associates a different C<struct ev_loop> with this watcher. You can only
		4036	do this when the watcher is inactive (and not pending either).
		4037
		4038	=item w->set ([arguments])
		4039
		4040	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this
		4041	method or a suitable start method must be called at least once. Unlike the
		4042	C counterpart, an active watcher gets automatically stopped and restarted
		4043	when reconfiguring it with this method.
		4044
		4045	=item w->start ()
		4046
		4047	Starts the watcher. Note that there is no C<loop> argument, as the
		4048	constructor already stores the event loop.
		4049
		4050	=item w->start ([arguments])
		4051
		4052	Instead of calling C<set> and C<start> methods separately, it is often
		4053	convenient to wrap them in one call. Uses the same type of arguments as
		4054	the configure C<set> method of the watcher.
		4055
		4056	=item w->stop ()
		4057
		4058	Stops the watcher if it is active. Again, no C<loop> argument.
		4059
		4060	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		4061
		4062	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		4063	C<ev_TYPE_again> function.
		4064
		4065	=item w->sweep () (C<ev::embed> only)
		4066
		4067	Invokes C<ev_embed_sweep>.
		4068
		4069	=item w->update () (C<ev::stat> only)
		4070
		4071	Invokes C<ev_stat_stat>.
		4072
		4073	=back
		4074
		4075	=back
		4076
		4077	Example: Define a class with two I/O and idle watchers, start the I/O
		4078	watchers in the constructor.
		4079
		4080	class myclass
		4081	{
		4082	ev::io io ; void io_cb (ev::io &w, int revents);
		4083	ev::io io2 ; void io2_cb (ev::io &w, int revents);
		4084	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		4085
		4086	myclass (int fd)
		4087	{
		4088	io .set <myclass, &myclass::io_cb > (this);
		4089	io2 .set <myclass, &myclass::io2_cb > (this);
		4090	idle.set <myclass, &myclass::idle_cb> (this);
		4091
		4092	io.set (fd, ev::WRITE); // configure the watcher
		4093	io.start (); // start it whenever convenient
		4094
		4095	io2.start (fd, ev::READ); // set + start in one call
		4096	}
		4097	};
		4098
		4099
		4100	=head1 OTHER LANGUAGE BINDINGS
		4101
		4102	Libev does not offer other language bindings itself, but bindings for a
		4103	number of languages exist in the form of third-party packages. If you know
		4104	any interesting language binding in addition to the ones listed here, drop
		4105	me a note.
		4106
		4107	=over 4
		4108
		4109	=item Perl
		4110
		4111	The EV module implements the full libev API and is actually used to test
		4112	libev. EV is developed together with libev. Apart from the EV core module,
		4113	there are additional modules that implement libev-compatible interfaces
		4114	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		4115	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		4116	and C<EV::Glib>).
		4117
		4118	It can be found and installed via CPAN, its homepage is at
		4119	L<http://software.schmorp.de/pkg/EV>.
		4120
		4121	=item Python
		4122
		4123	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		4124	seems to be quite complete and well-documented.
		4125
		4126	=item Ruby
		4127
		4128	Tony Arcieri has written a ruby extension that offers access to a subset
		4129	of the libev API and adds file handle abstractions, asynchronous DNS and
		4130	more on top of it. It can be found via gem servers. Its homepage is at
		4131	L<http://rev.rubyforge.org/>.
		4132
		4133	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		4134	makes rev work even on mingw.
		4135
		4136	=item Haskell
		4137
		4138	A haskell binding to libev is available at
		4139	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		4140
		4141	=item D
		4142
		4143	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		4144	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
		4145
		4146	=item Ocaml
		4147
		4148	Erkki Seppala has written Ocaml bindings for libev, to be found at
		4149	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		4150
		4151	=item Lua
		4152
		4153	Brian Maher has written a partial interface to libev for lua (at the
		4154	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		4155	L<http://github.com/brimworks/lua-ev>.
		4156
		4157	=back
		4158
		4159
		4160	=head1 MACRO MAGIC
		4161
		4162	Libev can be compiled with a variety of options, the most fundamental
		4163	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		4164	functions and callbacks have an initial C<struct ev_loop *> argument.
		4165
		4166	To make it easier to write programs that cope with either variant, the
		4167	following macros are defined:
		4168
		4169	=over 4
		4170
		4171	=item C<EV_A>, C<EV_A_>
		4172
		4173	This provides the loop I<argument> for functions, if one is required ("ev
		4174	loop argument"). The C<EV_A> form is used when this is the sole argument,
		4175	C<EV_A_> is used when other arguments are following. Example:
		4176
		4177	ev_unref (EV_A);
		4178	ev_timer_add (EV_A_ watcher);
		4179	ev_run (EV_A_ 0);
		4180
		4181	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		4182	which is often provided by the following macro.
		4183
		4184	=item C<EV_P>, C<EV_P_>
		4185
		4186	This provides the loop I<parameter> for functions, if one is required ("ev
		4187	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		4188	C<EV_P_> is used when other parameters are following. Example:
		4189
		4190	// this is how ev_unref is being declared
		4191	static void ev_unref (EV_P);
		4192
		4193	// this is how you can declare your typical callback
		4194	static void cb (EV_P_ ev_timer *w, int revents)
		4195
		4196	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		4197	suitable for use with C<EV_A>.
		4198
		4199	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		4200
		4201	Similar to the other two macros, this gives you the value of the default
		4202	loop, if multiple loops are supported ("ev loop default"). The default loop
		4203	will be initialised if it isn't already initialised.
		4204
		4205	For non-multiplicity builds, these macros do nothing, so you always have
		4206	to initialise the loop somewhere.
		4207
		4208	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		4209
		4210	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		4211	default loop has been initialised (C<UC> == unchecked). Their behaviour
		4212	is undefined when the default loop has not been initialised by a previous
		4213	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		4214
		4215	It is often prudent to use C<EV_DEFAULT> when initialising the first
		4216	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		4217
		4218	=back
		4219
		4220	Example: Declare and initialise a check watcher, utilising the above
		4221	macros so it will work regardless of whether multiple loops are supported
		4222	or not.
		4223
		4224	static void
		4225	check_cb (EV_P_ ev_timer *w, int revents)
		4226	{
		4227	ev_check_stop (EV_A_ w);
		4228	}
		4229
		4230	ev_check check;
		4231	ev_check_init (&check, check_cb);
		4232	ev_check_start (EV_DEFAULT_ &check);
		4233	ev_run (EV_DEFAULT_ 0);
		4234
		4235	=head1 EMBEDDING
		4236
		4237	Libev can (and often is) directly embedded into host
		4238	applications. Examples of applications that embed it include the Deliantra
		4239	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		4240	and rxvt-unicode.
		4241
		4242	The goal is to enable you to just copy the necessary files into your
		4243	source directory without having to change even a single line in them, so
		4244	you can easily upgrade by simply copying (or having a checked-out copy of
		4245	libev somewhere in your source tree).
		4246
		4247	=head2 FILESETS
		4248
		4249	Depending on what features you need you need to include one or more sets of files
		4250	in your application.
		4251
		4252	=head3 CORE EVENT LOOP
		4253
		4254	To include only the libev core (all the C<ev_*> functions), with manual
		4255	configuration (no autoconf):
		4256
		4257	#define EV_STANDALONE 1
		4258	#include "ev.c"
		4259
		4260	This will automatically include F<ev.h>, too, and should be done in a
		4261	single C source file only to provide the function implementations. To use
		4262	it, do the same for F<ev.h> in all files wishing to use this API (best
		4263	done by writing a wrapper around F<ev.h> that you can include instead and
		4264	where you can put other configuration options):
		4265
		4266	#define EV_STANDALONE 1
		4267	#include "ev.h"
		4268
		4269	Both header files and implementation files can be compiled with a C++
		4270	compiler (at least, that's a stated goal, and breakage will be treated
		4271	as a bug).
		4272
		4273	You need the following files in your source tree, or in a directory
		4274	in your include path (e.g. in libev/ when using -Ilibev):
		4275
		4276	ev.h
		4277	ev.c
		4278	ev_vars.h
		4279	ev_wrap.h
		4280
		4281	ev_win32.c required on win32 platforms only
		4282
		4283	ev_select.c only when select backend is enabled (which is enabled by default)
		4284	ev_poll.c only when poll backend is enabled (disabled by default)
		4285	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		4286	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		4287	ev_port.c only when the solaris port backend is enabled (disabled by default)
		4288
		4289	F<ev.c> includes the backend files directly when enabled, so you only need
		4290	to compile this single file.
		4291
		4292	=head3 LIBEVENT COMPATIBILITY API
		4293
		4294	To include the libevent compatibility API, also include:
		4295
		4296	#include "event.c"
		4297
		4298	in the file including F<ev.c>, and:
		4299
		4300	#include "event.h"
		4301
		4302	in the files that want to use the libevent API. This also includes F<ev.h>.
		4303
		4304	You need the following additional files for this:
		4305
		4306	event.h
		4307	event.c
		4308
		4309	=head3 AUTOCONF SUPPORT
		4310
		4311	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		4312	whatever way you want, you can also C<m4_include([libev.m4])> in your
		4313	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		4314	include F<config.h> and configure itself accordingly.
		4315
		4316	For this of course you need the m4 file:
		4317
		4318	libev.m4
		4319
		4320	=head2 PREPROCESSOR SYMBOLS/MACROS
		4321
		4322	Libev can be configured via a variety of preprocessor symbols you have to
		4323	define before including (or compiling) any of its files. The default in
		4324	the absence of autoconf is documented for every option.
		4325
		4326	Symbols marked with "(h)" do not change the ABI, and can have different
		4327	values when compiling libev vs. including F<ev.h>, so it is permissible
		4328	to redefine them before including F<ev.h> without breaking compatibility
		4329	to a compiled library. All other symbols change the ABI, which means all
		4330	users of libev and the libev code itself must be compiled with compatible
		4331	settings.
		4332
		4333	=over 4
		4334
		4335	=item EV_COMPAT3 (h)
		4336
		4337	Backwards compatibility is a major concern for libev. This is why this
		4338	release of libev comes with wrappers for the functions and symbols that
		4339	have been renamed between libev version 3 and 4.
		4340
		4341	You can disable these wrappers (to test compatibility with future
		4342	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4343	sources. This has the additional advantage that you can drop the C<struct>
		4344	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4345	typedef in that case.
		4346
		4347	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4348	and in some even more future version the compatibility code will be
		4349	removed completely.
		4350
		4351	=item EV_STANDALONE (h)
		4352
		4353	Must always be C<1> if you do not use autoconf configuration, which
		4354	keeps libev from including F<config.h>, and it also defines dummy
		4355	implementations for some libevent functions (such as logging, which is not
		4356	supported). It will also not define any of the structs usually found in
		4357	F<event.h> that are not directly supported by the libev core alone.
		4358
		4359	In standalone mode, libev will still try to automatically deduce the
		4360	configuration, but has to be more conservative.
		4361
		4362	=item EV_USE_FLOOR
		4363
		4364	If defined to be C<1>, libev will use the C<floor ()> function for its
		4365	periodic reschedule calculations, otherwise libev will fall back on a
		4366	portable (slower) implementation. If you enable this, you usually have to
		4367	link against libm or something equivalent. Enabling this when the C<floor>
		4368	function is not available will fail, so the safe default is to not enable
		4369	this.
		4370
		4371	=item EV_USE_MONOTONIC
		4372
		4373	If defined to be C<1>, libev will try to detect the availability of the
		4374	monotonic clock option at both compile time and runtime. Otherwise no
		4375	use of the monotonic clock option will be attempted. If you enable this,
		4376	you usually have to link against librt or something similar. Enabling it
		4377	when the functionality isn't available is safe, though, although you have
		4378	to make sure you link against any libraries where the C<clock_gettime>
		4379	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		4380
		4381	=item EV_USE_REALTIME
		4382
		4383	If defined to be C<1>, libev will try to detect the availability of the
		4384	real-time clock option at compile time (and assume its availability
		4385	at runtime if successful). Otherwise no use of the real-time clock
		4386	option will be attempted. This effectively replaces C<gettimeofday>
		4387	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		4388	correctness. See the note about libraries in the description of
		4389	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4390	C<EV_USE_CLOCK_SYSCALL>.
		4391
		4392	=item EV_USE_CLOCK_SYSCALL
		4393
		4394	If defined to be C<1>, libev will try to use a direct syscall instead
		4395	of calling the system-provided C<clock_gettime> function. This option
		4396	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4397	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4398	programs needlessly. Using a direct syscall is slightly slower (in
		4399	theory), because no optimised vdso implementation can be used, but avoids
		4400	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4401	higher, as it simplifies linking (no need for C<-lrt>).
		4402
		4403	=item EV_USE_NANOSLEEP
		4404
		4405	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		4406	and will use it for delays. Otherwise it will use C<select ()>.
		4407
		4408	=item EV_USE_EVENTFD
		4409
		4410	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4411	available and will probe for kernel support at runtime. This will improve
		4412	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4413	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4414	2.7 or newer, otherwise disabled.
		4415
		4416	=item EV_USE_SELECT
		4417
		4418	If undefined or defined to be C<1>, libev will compile in support for the
		4419	C<select>(2) backend. No attempt at auto-detection will be done: if no
		4420	other method takes over, select will be it. Otherwise the select backend
		4421	will not be compiled in.
		4422
		4423	=item EV_SELECT_USE_FD_SET
		4424
		4425	If defined to C<1>, then the select backend will use the system C<fd_set>
		4426	structure. This is useful if libev doesn't compile due to a missing
		4427	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		4428	on exotic systems. This usually limits the range of file descriptors to
		4429	some low limit such as 1024 or might have other limitations (winsocket
		4430	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		4431	configures the maximum size of the C<fd_set>.
		4432
		4433	=item EV_SELECT_IS_WINSOCKET
		4434
		4435	When defined to C<1>, the select backend will assume that
		4436	select/socket/connect etc. don't understand file descriptors but
		4437	wants osf handles on win32 (this is the case when the select to
		4438	be used is the winsock select). This means that it will call
		4439	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		4440	it is assumed that all these functions actually work on fds, even
		4441	on win32. Should not be defined on non-win32 platforms.
		4442
		4443	=item EV_FD_TO_WIN32_HANDLE(fd)
		4444
		4445	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		4446	file descriptors to socket handles. When not defining this symbol (the
		4447	default), then libev will call C<_get_osfhandle>, which is usually
		4448	correct. In some cases, programs use their own file descriptor management,
		4449	in which case they can provide this function to map fds to socket handles.
		4450
		4451	=item EV_WIN32_HANDLE_TO_FD(handle)
		4452
		4453	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4454	using the standard C<_open_osfhandle> function. For programs implementing
		4455	their own fd to handle mapping, overwriting this function makes it easier
		4456	to do so. This can be done by defining this macro to an appropriate value.
		4457
		4458	=item EV_WIN32_CLOSE_FD(fd)
		4459
		4460	If programs implement their own fd to handle mapping on win32, then this
		4461	macro can be used to override the C<close> function, useful to unregister
		4462	file descriptors again. Note that the replacement function has to close
		4463	the underlying OS handle.
		4464
		4465	=item EV_USE_POLL
		4466
		4467	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		4468	backend. Otherwise it will be enabled on non-win32 platforms. It
		4469	takes precedence over select.
		4470
		4471	=item EV_USE_EPOLL
		4472
		4473	If defined to be C<1>, libev will compile in support for the Linux
		4474	C<epoll>(7) backend. Its availability will be detected at runtime,
		4475	otherwise another method will be used as fallback. This is the preferred
		4476	backend for GNU/Linux systems. If undefined, it will be enabled if the
		4477	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4478
		4479	=item EV_USE_KQUEUE
		4480
		4481	If defined to be C<1>, libev will compile in support for the BSD style
		4482	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		4483	otherwise another method will be used as fallback. This is the preferred
		4484	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		4485	supports some types of fds correctly (the only platform we found that
		4486	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		4487	not be used unless explicitly requested. The best way to use it is to find
		4488	out whether kqueue supports your type of fd properly and use an embedded
		4489	kqueue loop.
		4490
		4491	=item EV_USE_PORT
		4492
		4493	If defined to be C<1>, libev will compile in support for the Solaris
		4494	10 port style backend. Its availability will be detected at runtime,
		4495	otherwise another method will be used as fallback. This is the preferred
		4496	backend for Solaris 10 systems.
		4497
		4498	=item EV_USE_DEVPOLL
		4499
		4500	Reserved for future expansion, works like the USE symbols above.
		4501
		4502	=item EV_USE_INOTIFY
		4503
		4504	If defined to be C<1>, libev will compile in support for the Linux inotify
		4505	interface to speed up C<ev_stat> watchers. Its actual availability will
		4506	be detected at runtime. If undefined, it will be enabled if the headers
		4507	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4508
		4509	=item EV_NO_SMP
		4510
		4511	If defined to be C<1>, libev will assume that memory is always coherent
		4512	between threads, that is, threads can be used, but threads never run on
		4513	different cpus (or different cpu cores). This reduces dependencies
		4514	and makes libev faster.
		4515
		4516	=item EV_NO_THREADS
		4517
		4518	If defined to be C<1>, libev will assume that it will never be called
		4519	from different threads, which is a stronger assumption than C<EV_NO_SMP>,
		4520	above. This reduces dependencies and makes libev faster.
		4521
		4522	=item EV_ATOMIC_T
		4523
		4524	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		4525	access is atomic and serialised with respect to other threads or signal
		4526	contexts. No such type is easily found in the C language, so you can
		4527	provide your own type that you know is safe for your purposes. It is used
		4528	both for signal handler "locking" as well as for signal and thread safety
		4529	in C<ev_async> watchers.
		4530
		4531	In the absence of this define, libev will use C<sig_atomic_t volatile>
		4532	(from F<signal.h>), which is usually good enough on most platforms,
		4533	although strictly speaking using a type that also implies a memory fence
		4534	is required.
		4535
		4536	=item EV_H (h)
		4537
		4538	The name of the F<ev.h> header file used to include it. The default if
		4539	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		4540	used to virtually rename the F<ev.h> header file in case of conflicts.
		4541
		4542	=item EV_CONFIG_H (h)
		4543
		4544	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		4545	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		4546	C<EV_H>, above.
		4547
		4548	=item EV_EVENT_H (h)
		4549
		4550	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		4551	of how the F<event.h> header can be found, the default is C<"event.h">.
		4552
		4553	=item EV_PROTOTYPES (h)
		4554
		4555	If defined to be C<0>, then F<ev.h> will not define any function
		4556	prototypes, but still define all the structs and other symbols. This is
		4557	occasionally useful if you want to provide your own wrapper functions
		4558	around libev functions.
		4559
		4560	=item EV_MULTIPLICITY
		4561
		4562	If undefined or defined to C<1>, then all event-loop-specific functions
		4563	will have the C<struct ev_loop *> as first argument, and you can create
		4564	additional independent event loops. Otherwise there will be no support
		4565	for multiple event loops and there is no first event loop pointer
		4566	argument. Instead, all functions act on the single default loop.
		4567
		4568	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4569	default loop when multiplicity is switched off - you always have to
		4570	initialise the loop manually in this case.
		4571
		4572	=item EV_MINPRI
		4573
		4574	=item EV_MAXPRI
		4575
		4576	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		4577	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		4578	provide for more priorities by overriding those symbols (usually defined
		4579	to be C<-2> and C<2>, respectively).
		4580
		4581	When doing priority-based operations, libev usually has to linearly search
		4582	all the priorities, so having many of them (hundreds) uses a lot of space
		4583	and time, so using the defaults of five priorities (-2 .. +2) is usually
		4584	fine.
		4585
		4586	If your embedding application does not need any priorities, defining these
		4587	both to C<0> will save some memory and CPU.
		4588
		4589	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4590	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4591	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
		4592
		4593	If undefined or defined to be C<1> (and the platform supports it), then
		4594	the respective watcher type is supported. If defined to be C<0>, then it
		4595	is not. Disabling watcher types mainly saves code size.
		4596
		4597	=item EV_FEATURES
		4598
		4599	If you need to shave off some kilobytes of code at the expense of some
		4600	speed (but with the full API), you can define this symbol to request
		4601	certain subsets of functionality. The default is to enable all features
		4602	that can be enabled on the platform.
		4603
		4604	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4605	with some broad features you want) and then selectively re-enable
		4606	additional parts you want, for example if you want everything minimal,
		4607	but multiple event loop support, async and child watchers and the poll
		4608	backend, use this:
		4609
		4610	#define EV_FEATURES 0
		4611	#define EV_MULTIPLICITY 1
		4612	#define EV_USE_POLL 1
		4613	#define EV_CHILD_ENABLE 1
		4614	#define EV_ASYNC_ENABLE 1
		4615
		4616	The actual value is a bitset, it can be a combination of the following
		4617	values:
		4618
		4619	=over 4
		4620
		4621	=item C<1> - faster/larger code
		4622
		4623	Use larger code to speed up some operations.
		4624
		4625	Currently this is used to override some inlining decisions (enlarging the
		4626	code size by roughly 30% on amd64).
		4627
		4628	When optimising for size, use of compiler flags such as C<-Os> with
		4629	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4630	assertions.
		4631
		4632	=item C<2> - faster/larger data structures
		4633
		4634	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4635	hash table sizes and so on. This will usually further increase code size
		4636	and can additionally have an effect on the size of data structures at
		4637	runtime.
		4638
		4639	=item C<4> - full API configuration
		4640
		4641	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4642	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4643
		4644	=item C<8> - full API
		4645
		4646	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4647	details on which parts of the API are still available without this
		4648	feature, and do not complain if this subset changes over time.
		4649
		4650	=item C<16> - enable all optional watcher types
		4651
		4652	Enables all optional watcher types. If you want to selectively enable
		4653	only some watcher types other than I/O and timers (e.g. prepare,
		4654	embed, async, child...) you can enable them manually by defining
		4655	C<EV_watchertype_ENABLE> to C<1> instead.
		4656
		4657	=item C<32> - enable all backends
		4658
		4659	This enables all backends - without this feature, you need to enable at
		4660	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4661
		4662	=item C<64> - enable OS-specific "helper" APIs
		4663
		4664	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4665	default.
		4666
		4667	=back
		4668
		4669	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4670	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4671	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4672	watchers, timers and monotonic clock support.
		4673
		4674	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4675	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4676	your program might be left out as well - a binary starting a timer and an
		4677	I/O watcher then might come out at only 5Kb.
		4678
		4679	=item EV_API_STATIC
		4680
		4681	If this symbol is defined (by default it is not), then all identifiers
		4682	will have static linkage. This means that libev will not export any
		4683	identifiers, and you cannot link against libev anymore. This can be useful
		4684	when you embed libev, only want to use libev functions in a single file,
		4685	and do not want its identifiers to be visible.
		4686
		4687	To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
		4688	wants to use libev.
		4689
		4690	This option only works when libev is compiled with a C compiler, as C++
		4691	doesn't support the required declaration syntax.
		4692
		4693	=item EV_AVOID_STDIO
		4694
		4695	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4696	functions (printf, scanf, perror etc.). This will increase the code size
		4697	somewhat, but if your program doesn't otherwise depend on stdio and your
		4698	libc allows it, this avoids linking in the stdio library which is quite
		4699	big.
		4700
		4701	Note that error messages might become less precise when this option is
		4702	enabled.
		4703
		4704	=item EV_NSIG
		4705
		4706	The highest supported signal number, +1 (or, the number of
		4707	signals): Normally, libev tries to deduce the maximum number of signals
		4708	automatically, but sometimes this fails, in which case it can be
		4709	specified. Also, using a lower number than detected (C<32> should be
		4710	good for about any system in existence) can save some memory, as libev
		4711	statically allocates some 12-24 bytes per signal number.
		4712
		4713	=item EV_PID_HASHSIZE
		4714
		4715	C<ev_child> watchers use a small hash table to distribute workload by
		4716	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
		4717	usually more than enough. If you need to manage thousands of children you
		4718	might want to increase this value (I<must> be a power of two).
		4719
		4720	=item EV_INOTIFY_HASHSIZE
		4721
		4722	C<ev_stat> watchers use a small hash table to distribute workload by
		4723	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
		4724	disabled), usually more than enough. If you need to manage thousands of
		4725	C<ev_stat> watchers you might want to increase this value (I<must> be a
		4726	power of two).
		4727
		4728	=item EV_USE_4HEAP
		4729
		4730	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4731	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4732	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4733	faster performance with many (thousands) of watchers.
		4734
		4735	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4736	will be C<0>.
		4737
		4738	=item EV_HEAP_CACHE_AT
		4739
		4740	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4741	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4742	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4743	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4744	but avoids random read accesses on heap changes. This improves performance
		4745	noticeably with many (hundreds) of watchers.
		4746
		4747	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4748	will be C<0>.
		4749
		4750	=item EV_VERIFY
		4751
		4752	Controls how much internal verification (see C<ev_verify ()>) will
		4753	be done: If set to C<0>, no internal verification code will be compiled
		4754	in. If set to C<1>, then verification code will be compiled in, but not
		4755	called. If set to C<2>, then the internal verification code will be
		4756	called once per loop, which can slow down libev. If set to C<3>, then the
		4757	verification code will be called very frequently, which will slow down
		4758	libev considerably.
		4759
		4760	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4761	will be C<0>.
		4762
		4763	=item EV_COMMON
		4764
		4765	By default, all watchers have a C<void *data> member. By redefining
		4766	this macro to something else you can include more and other types of
		4767	members. You have to define it each time you include one of the files,
		4768	though, and it must be identical each time.
		4769
		4770	For example, the perl EV module uses something like this:
		4771
		4772	#define EV_COMMON \
		4773	SV *self; /* contains this struct */ \
		4774	SV *cb_sv, *fh /* note no trailing ";" */
		4775
		4776	=item EV_CB_DECLARE (type)
		4777
		4778	=item EV_CB_INVOKE (watcher, revents)
		4779
		4780	=item ev_set_cb (ev, cb)
		4781
		4782	Can be used to change the callback member declaration in each watcher,
		4783	and the way callbacks are invoked and set. Must expand to a struct member
		4784	definition and a statement, respectively. See the F<ev.h> header file for
		4785	their default definitions. One possible use for overriding these is to
		4786	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		4787	method calls instead of plain function calls in C++.
		4788
		4789	=back
		4790
		4791	=head2 EXPORTED API SYMBOLS
		4792
		4793	If you need to re-export the API (e.g. via a DLL) and you need a list of
		4794	exported symbols, you can use the provided F<Symbol.*> files which list
		4795	all public symbols, one per line:
		4796
		4797	Symbols.ev for libev proper
		4798	Symbols.event for the libevent emulation
		4799
		4800	This can also be used to rename all public symbols to avoid clashes with
		4801	multiple versions of libev linked together (which is obviously bad in
		4802	itself, but sometimes it is inconvenient to avoid this).
		4803
		4804	A sed command like this will create wrapper C<#define>'s that you need to
		4805	include before including F<ev.h>:
		4806
		4807	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		4808
		4809	This would create a file F<wrap.h> which essentially looks like this:
		4810
		4811	#define ev_backend myprefix_ev_backend
		4812	#define ev_check_start myprefix_ev_check_start
		4813	#define ev_check_stop myprefix_ev_check_stop
		4814	...
		4815
		4816	=head2 EXAMPLES
		4817
		4818	For a real-world example of a program the includes libev
		4819	verbatim, you can have a look at the EV perl module
		4820	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		4821	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		4822	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		4823	will be compiled. It is pretty complex because it provides its own header
		4824	file.
		4825
		4826	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		4827	that everybody includes and which overrides some configure choices:
		4828
		4829	#define EV_FEATURES 8
		4830	#define EV_USE_SELECT 1
		4831	#define EV_PREPARE_ENABLE 1
		4832	#define EV_IDLE_ENABLE 1
		4833	#define EV_SIGNAL_ENABLE 1
		4834	#define EV_CHILD_ENABLE 1
		4835	#define EV_USE_STDEXCEPT 0
		4836	#define EV_CONFIG_H <config.h>
		4837
		4838	#include "ev++.h"
		4839
		4840	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4841
		4842	#include "ev_cpp.h"
		4843	#include "ev.c"
		4844
		4845	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
		4846
		4847	=head2 THREADS AND COROUTINES
		4848
		4849	=head3 THREADS
		4850
		4851	All libev functions are reentrant and thread-safe unless explicitly
		4852	documented otherwise, but libev implements no locking itself. This means
		4853	that you can use as many loops as you want in parallel, as long as there
		4854	are no concurrent calls into any libev function with the same loop
		4855	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4856	of course): libev guarantees that different event loops share no data
		4857	structures that need any locking.
		4858
		4859	Or to put it differently: calls with different loop parameters can be done
		4860	concurrently from multiple threads, calls with the same loop parameter
		4861	must be done serially (but can be done from different threads, as long as
		4862	only one thread ever is inside a call at any point in time, e.g. by using
		4863	a mutex per loop).
		4864
		4865	Specifically to support threads (and signal handlers), libev implements
		4866	so-called C<ev_async> watchers, which allow some limited form of
		4867	concurrency on the same event loop, namely waking it up "from the
		4868	outside".
		4869
		4870	If you want to know which design (one loop, locking, or multiple loops
		4871	without or something else still) is best for your problem, then I cannot
		4872	help you, but here is some generic advice:
		4873
		4874	=over 4
		4875
		4876	=item * most applications have a main thread: use the default libev loop
		4877	in that thread, or create a separate thread running only the default loop.
		4878
		4879	This helps integrating other libraries or software modules that use libev
		4880	themselves and don't care/know about threading.
		4881
		4882	=item * one loop per thread is usually a good model.
		4883
		4884	Doing this is almost never wrong, sometimes a better-performance model
		4885	exists, but it is always a good start.
		4886
		4887	=item * other models exist, such as the leader/follower pattern, where one
		4888	loop is handed through multiple threads in a kind of round-robin fashion.
		4889
		4890	Choosing a model is hard - look around, learn, know that usually you can do
		4891	better than you currently do :-)
		4892
		4893	=item * often you need to talk to some other thread which blocks in the
		4894	event loop.
		4895
		4896	C<ev_async> watchers can be used to wake them up from other threads safely
		4897	(or from signal contexts...).
		4898
		4899	An example use would be to communicate signals or other events that only
		4900	work in the default loop by registering the signal watcher with the
		4901	default loop and triggering an C<ev_async> watcher from the default loop
		4902	watcher callback into the event loop interested in the signal.
		4903
		4904	=back
		4905
		4906	See also L<THREAD LOCKING EXAMPLE>.
		4907
		4908	=head3 COROUTINES
		4909
		4910	Libev is very accommodating to coroutines ("cooperative threads"):
		4911	libev fully supports nesting calls to its functions from different
		4912	coroutines (e.g. you can call C<ev_run> on the same loop from two
		4913	different coroutines, and switch freely between both coroutines running
		4914	the loop, as long as you don't confuse yourself). The only exception is
		4915	that you must not do this from C<ev_periodic> reschedule callbacks.
		4916
		4917	Care has been taken to ensure that libev does not keep local state inside
		4918	C<ev_run>, and other calls do not usually allow for coroutine switches as
		4919	they do not call any callbacks.
		4920
		4921	=head2 COMPILER WARNINGS
		4922
		4923	Depending on your compiler and compiler settings, you might get no or a
		4924	lot of warnings when compiling libev code. Some people are apparently
		4925	scared by this.
		4926
		4927	However, these are unavoidable for many reasons. For one, each compiler
		4928	has different warnings, and each user has different tastes regarding
		4929	warning options. "Warn-free" code therefore cannot be a goal except when
		4930	targeting a specific compiler and compiler-version.
		4931
		4932	Another reason is that some compiler warnings require elaborate
		4933	workarounds, or other changes to the code that make it less clear and less
		4934	maintainable.
		4935
		4936	And of course, some compiler warnings are just plain stupid, or simply
		4937	wrong (because they don't actually warn about the condition their message
		4938	seems to warn about). For example, certain older gcc versions had some
		4939	warnings that resulted in an extreme number of false positives. These have
		4940	been fixed, but some people still insist on making code warn-free with
		4941	such buggy versions.
		4942
		4943	While libev is written to generate as few warnings as possible,
		4944	"warn-free" code is not a goal, and it is recommended not to build libev
		4945	with any compiler warnings enabled unless you are prepared to cope with
		4946	them (e.g. by ignoring them). Remember that warnings are just that:
		4947	warnings, not errors, or proof of bugs.
		4948
		4949
		4950	=head2 VALGRIND
		4951
		4952	Valgrind has a special section here because it is a popular tool that is
		4953	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4954
		4955	If you think you found a bug (memory leak, uninitialised data access etc.)
		4956	in libev, then check twice: If valgrind reports something like:
		4957
		4958	==2274== definitely lost: 0 bytes in 0 blocks.
		4959	==2274== possibly lost: 0 bytes in 0 blocks.
		4960	==2274== still reachable: 256 bytes in 1 blocks.
		4961
		4962	Then there is no memory leak, just as memory accounted to global variables
		4963	is not a memleak - the memory is still being referenced, and didn't leak.
		4964
		4965	Similarly, under some circumstances, valgrind might report kernel bugs
		4966	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4967	although an acceptable workaround has been found here), or it might be
		4968	confused.
		4969
		4970	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4971	make it into some kind of religion.
		4972
		4973	If you are unsure about something, feel free to contact the mailing list
		4974	with the full valgrind report and an explanation on why you think this
		4975	is a bug in libev (best check the archives, too :). However, don't be
		4976	annoyed when you get a brisk "this is no bug" answer and take the chance
		4977	of learning how to interpret valgrind properly.
		4978
		4979	If you need, for some reason, empty reports from valgrind for your project
		4980	I suggest using suppression lists.
		4981
		4982
		4983	=head1 PORTABILITY NOTES
		4984
		4985	=head2 GNU/LINUX 32 BIT LIMITATIONS
		4986
		4987	GNU/Linux is the only common platform that supports 64 bit file/large file
		4988	interfaces but I<disables> them by default.
		4989
		4990	That means that libev compiled in the default environment doesn't support
		4991	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		4992
		4993	Unfortunately, many programs try to work around this GNU/Linux issue
		4994	by enabling the large file API, which makes them incompatible with the
		4995	standard libev compiled for their system.
		4996
		4997	Likewise, libev cannot enable the large file API itself as this would
		4998	suddenly make it incompatible to the default compile time environment,
		4999	i.e. all programs not using special compile switches.
		5000
		5001	=head2 OS/X AND DARWIN BUGS
		5002
		5003	The whole thing is a bug if you ask me - basically any system interface
		5004	you touch is broken, whether it is locales, poll, kqueue or even the
		5005	OpenGL drivers.
		5006
		5007	=head3 C<kqueue> is buggy
		5008
		5009	The kqueue syscall is broken in all known versions - most versions support
		5010	only sockets, many support pipes.
		5011
		5012	Libev tries to work around this by not using C<kqueue> by default on this
		5013	rotten platform, but of course you can still ask for it when creating a
		5014	loop - embedding a socket-only kqueue loop into a select-based one is
		5015	probably going to work well.
		5016
		5017	=head3 C<poll> is buggy
		5018
		5019	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		5020	implementation by something calling C<kqueue> internally around the 10.5.6
		5021	release, so now C<kqueue> I<and> C<poll> are broken.
		5022
		5023	Libev tries to work around this by not using C<poll> by default on
		5024	this rotten platform, but of course you can still ask for it when creating
		5025	a loop.
		5026
		5027	=head3 C<select> is buggy
		5028
		5029	All that's left is C<select>, and of course Apple found a way to fuck this
		5030	one up as well: On OS/X, C<select> actively limits the number of file
		5031	descriptors you can pass in to 1024 - your program suddenly crashes when
		5032	you use more.
		5033
		5034	There is an undocumented "workaround" for this - defining
		5035	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		5036	work on OS/X.
		5037
		5038	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		5039
		5040	=head3 C<errno> reentrancy
		5041
		5042	The default compile environment on Solaris is unfortunately so
		5043	thread-unsafe that you can't even use components/libraries compiled
		5044	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		5045	defined by default. A valid, if stupid, implementation choice.
		5046
		5047	If you want to use libev in threaded environments you have to make sure
		5048	it's compiled with C<_REENTRANT> defined.
		5049
		5050	=head3 Event port backend
		5051
		5052	The scalable event interface for Solaris is called "event
		5053	ports". Unfortunately, this mechanism is very buggy in all major
		5054	releases. If you run into high CPU usage, your program freezes or you get
		5055	a large number of spurious wakeups, make sure you have all the relevant
		5056	and latest kernel patches applied. No, I don't know which ones, but there
		5057	are multiple ones to apply, and afterwards, event ports actually work
		5058	great.
		5059
		5060	If you can't get it to work, you can try running the program by setting
		5061	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		5062	C<select> backends.
		5063
		5064	=head2 AIX POLL BUG
		5065
		5066	AIX unfortunately has a broken C<poll.h> header. Libev works around
		5067	this by trying to avoid the poll backend altogether (i.e. it's not even
		5068	compiled in), which normally isn't a big problem as C<select> works fine
		5069	with large bitsets on AIX, and AIX is dead anyway.
		5070
		5071	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		5072
		5073	=head3 General issues
		5074
		5075	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		5076	requires, and its I/O model is fundamentally incompatible with the POSIX
		5077	model. Libev still offers limited functionality on this platform in
		5078	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		5079	descriptors. This only applies when using Win32 natively, not when using
		5080	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		5081	as every compiler comes with a slightly differently broken/incompatible
		5082	environment.
		5083
		5084	Lifting these limitations would basically require the full
		5085	re-implementation of the I/O system. If you are into this kind of thing,
		5086	then note that glib does exactly that for you in a very portable way (note
		5087	also that glib is the slowest event library known to man).
		5088
		5089	There is no supported compilation method available on windows except
		5090	embedding it into other applications.
		5091
		5092	Sensible signal handling is officially unsupported by Microsoft - libev
		5093	tries its best, but under most conditions, signals will simply not work.
		5094
		5095	Not a libev limitation but worth mentioning: windows apparently doesn't
		5096	accept large writes: instead of resulting in a partial write, windows will
		5097	either accept everything or return C<ENOBUFS> if the buffer is too large,
		5098	so make sure you only write small amounts into your sockets (less than a
		5099	megabyte seems safe, but this apparently depends on the amount of memory
		5100	available).
		5101
		5102	Due to the many, low, and arbitrary limits on the win32 platform and
		5103	the abysmal performance of winsockets, using a large number of sockets
		5104	is not recommended (and not reasonable). If your program needs to use
		5105	more than a hundred or so sockets, then likely it needs to use a totally
		5106	different implementation for windows, as libev offers the POSIX readiness
		5107	notification model, which cannot be implemented efficiently on windows
		5108	(due to Microsoft monopoly games).
		5109
		5110	A typical way to use libev under windows is to embed it (see the embedding
		5111	section for details) and use the following F<evwrap.h> header file instead
		5112	of F<ev.h>:
		5113
		5114	#define EV_STANDALONE /* keeps ev from requiring config.h */
		5115	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		5116
		5117	#include "ev.h"
		5118
		5119	And compile the following F<evwrap.c> file into your project (make sure
		5120	you do I<not> compile the F<ev.c> or any other embedded source files!):
		5121
		5122	#include "evwrap.h"
		5123	#include "ev.c"
		5124
		5125	=head3 The winsocket C<select> function
		5126
		5127	The winsocket C<select> function doesn't follow POSIX in that it
		5128	requires socket I<handles> and not socket I<file descriptors> (it is
		5129	also extremely buggy). This makes select very inefficient, and also
		5130	requires a mapping from file descriptors to socket handles (the Microsoft
		5131	C runtime provides the function C<_open_osfhandle> for this). See the
		5132	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		5133	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		5134
		5135	The configuration for a "naked" win32 using the Microsoft runtime
		5136	libraries and raw winsocket select is:
		5137
		5138	#define EV_USE_SELECT 1
		5139	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		5140
		5141	Note that winsockets handling of fd sets is O(n), so you can easily get a
		5142	complexity in the O(n²) range when using win32.
		5143
		5144	=head3 Limited number of file descriptors
		5145
		5146	Windows has numerous arbitrary (and low) limits on things.
		5147
		5148	Early versions of winsocket's select only supported waiting for a maximum
		5149	of C<64> handles (probably owning to the fact that all windows kernels
		5150	can only wait for C<64> things at the same time internally; Microsoft
		5151	recommends spawning a chain of threads and wait for 63 handles and the
		5152	previous thread in each. Sounds great!).
		5153
		5154	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		5155	to some high number (e.g. C<2048>) before compiling the winsocket select
		5156	call (which might be in libev or elsewhere, for example, perl and many
		5157	other interpreters do their own select emulation on windows).
		5158
		5159	Another limit is the number of file descriptors in the Microsoft runtime
		5160	libraries, which by default is C<64> (there must be a hidden I<64>
		5161	fetish or something like this inside Microsoft). You can increase this
		5162	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		5163	(another arbitrary limit), but is broken in many versions of the Microsoft
		5164	runtime libraries. This might get you to about C<512> or C<2048> sockets
		5165	(depending on windows version and/or the phase of the moon). To get more,
		5166	you need to wrap all I/O functions and provide your own fd management, but
		5167	the cost of calling select (O(n²)) will likely make this unworkable.
		5168
		5169	=head2 PORTABILITY REQUIREMENTS
		5170
		5171	In addition to a working ISO-C implementation and of course the
		5172	backend-specific APIs, libev relies on a few additional extensions:
		5173
		5174	=over 4
		5175
		5176	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		5177	calling conventions regardless of C<ev_watcher_type *>.
		5178
		5179	Libev assumes not only that all watcher pointers have the same internal
		5180	structure (guaranteed by POSIX but not by ISO C for example), but it also
		5181	assumes that the same (machine) code can be used to call any watcher
		5182	callback: The watcher callbacks have different type signatures, but libev
		5183	calls them using an C<ev_watcher *> internally.
		5184
		5185	=item pointer accesses must be thread-atomic
		5186
		5187	Accessing a pointer value must be atomic, it must both be readable and
		5188	writable in one piece - this is the case on all current architectures.
		5189
		5190	=item C<sig_atomic_t volatile> must be thread-atomic as well
		5191
		5192	The type C<sig_atomic_t volatile> (or whatever is defined as
		5193	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		5194	threads. This is not part of the specification for C<sig_atomic_t>, but is
		5195	believed to be sufficiently portable.
		5196
		5197	=item C<sigprocmask> must work in a threaded environment
		5198
		5199	Libev uses C<sigprocmask> to temporarily block signals. This is not
		5200	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		5201	pthread implementations will either allow C<sigprocmask> in the "main
		5202	thread" or will block signals process-wide, both behaviours would
		5203	be compatible with libev. Interaction between C<sigprocmask> and
		5204	C<pthread_sigmask> could complicate things, however.
		5205
		5206	The most portable way to handle signals is to block signals in all threads
		5207	except the initial one, and run the default loop in the initial thread as
		5208	well.
		5209
		5210	=item C<long> must be large enough for common memory allocation sizes
		5211
		5212	To improve portability and simplify its API, libev uses C<long> internally
		5213	instead of C<size_t> when allocating its data structures. On non-POSIX
		5214	systems (Microsoft...) this might be unexpectedly low, but is still at
		5215	least 31 bits everywhere, which is enough for hundreds of millions of
		5216	watchers.
		5217
		5218	=item C<double> must hold a time value in seconds with enough accuracy
		5219
		5220	The type C<double> is used to represent timestamps. It is required to
		5221	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		5222	good enough for at least into the year 4000 with millisecond accuracy
		5223	(the design goal for libev). This requirement is overfulfilled by
		5224	implementations using IEEE 754, which is basically all existing ones.
		5225
		5226	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5227	year 2255 (and millisecond accuracy till the year 287396 - by then, libev
		5228	is either obsolete or somebody patched it to use C<long double> or
		5229	something like that, just kidding).
		5230
		5231	=back
		5232
		5233	If you know of other additional requirements drop me a note.
		5234
		5235
		5236	=head1 ALGORITHMIC COMPLEXITIES
		5237
		5238	In this section the complexities of (many of) the algorithms used inside
		5239	libev will be documented. For complexity discussions about backends see
		5240	the documentation for C<ev_default_init>.
		5241
		5242	All of the following are about amortised time: If an array needs to be
		5243	extended, libev needs to realloc and move the whole array, but this
		5244	happens asymptotically rarer with higher number of elements, so O(1) might
		5245	mean that libev does a lengthy realloc operation in rare cases, but on
		5246	average it is much faster and asymptotically approaches constant time.
		5247
		5248	=over 4
		5249
		5250	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		5251
		5252	This means that, when you have a watcher that triggers in one hour and
		5253	there are 100 watchers that would trigger before that, then inserting will
		5254	have to skip roughly seven (C<ld 100>) of these watchers.
		5255
		5256	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		5257
		5258	That means that changing a timer costs less than removing/adding them,
		5259	as only the relative motion in the event queue has to be paid for.
		5260
		5261	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		5262
		5263	These just add the watcher into an array or at the head of a list.
		5264
		5265	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		5266
		5267	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		5268
		5269	These watchers are stored in lists, so they need to be walked to find the
		5270	correct watcher to remove. The lists are usually short (you don't usually
		5271	have many watchers waiting for the same fd or signal: one is typical, two
		5272	is rare).
		5273
		5274	=item Finding the next timer in each loop iteration: O(1)
		5275
		5276	By virtue of using a binary or 4-heap, the next timer is always found at a
		5277	fixed position in the storage array.
		5278
		5279	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		5280
		5281	A change means an I/O watcher gets started or stopped, which requires
		5282	libev to recalculate its status (and possibly tell the kernel, depending
		5283	on backend and whether C<ev_io_set> was used).
		5284
		5285	=item Activating one watcher (putting it into the pending state): O(1)
		5286
		5287	=item Priority handling: O(number_of_priorities)
		5288
		5289	Priorities are implemented by allocating some space for each
		5290	priority. When doing priority-based operations, libev usually has to
		5291	linearly search all the priorities, but starting/stopping and activating
		5292	watchers becomes O(1) with respect to priority handling.
		5293
		5294	=item Sending an ev_async: O(1)
		5295
		5296	=item Processing ev_async_send: O(number_of_async_watchers)
		5297
		5298	=item Processing signals: O(max_signal_number)
		5299
		5300	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5301	calls in the current loop iteration and the loop is currently
		5302	blocked. Checking for async and signal events involves iterating over all
		5303	running async watchers or all signal numbers.
		5304
		5305	=back
		5306
		5307
		5308	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5309
		5310	The major version 4 introduced some incompatible changes to the API.
		5311
		5312	At the moment, the C<ev.h> header file provides compatibility definitions
		5313	for all changes, so most programs should still compile. The compatibility
		5314	layer might be removed in later versions of libev, so better update to the
		5315	new API early than late.
		5316
		5317	=over 4
		5318
		5319	=item C<EV_COMPAT3> backwards compatibility mechanism
		5320
		5321	The backward compatibility mechanism can be controlled by
		5322	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
		5323	section.
		5324
		5325	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5326
		5327	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5328
		5329	ev_loop_destroy (EV_DEFAULT_UC);
		5330	ev_loop_fork (EV_DEFAULT);
		5331
		5332	=item function/symbol renames
		5333
		5334	A number of functions and symbols have been renamed:
		5335
		5336	ev_loop => ev_run
		5337	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5338	EVLOOP_ONESHOT => EVRUN_ONCE
		5339
		5340	ev_unloop => ev_break
		5341	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5342	EVUNLOOP_ONE => EVBREAK_ONE
		5343	EVUNLOOP_ALL => EVBREAK_ALL
		5344
		5345	EV_TIMEOUT => EV_TIMER
		5346
		5347	ev_loop_count => ev_iteration
		5348	ev_loop_depth => ev_depth
		5349	ev_loop_verify => ev_verify
		5350
		5351	Most functions working on C<struct ev_loop> objects don't have an
		5352	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5353	associated constants have been renamed to not collide with the C<struct
		5354	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5355	as all other watcher types. Note that C<ev_loop_fork> is still called
		5356	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5357	typedef.
		5358
		5359	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5360
		5361	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5362	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5363	and work, but the library code will of course be larger.
		5364
		5365	=back
		5366
		5367
		5368	=head1 GLOSSARY
		5369
		5370	=over 4
		5371
		5372	=item active
		5373
		5374	A watcher is active as long as it has been started and not yet stopped.
		5375	See L<WATCHER STATES> for details.
		5376
		5377	=item application
		5378
		5379	In this document, an application is whatever is using libev.
		5380
		5381	=item backend
		5382
		5383	The part of the code dealing with the operating system interfaces.
		5384
		5385	=item callback
		5386
		5387	The address of a function that is called when some event has been
		5388	detected. Callbacks are being passed the event loop, the watcher that
		5389	received the event, and the actual event bitset.
		5390
		5391	=item callback/watcher invocation
		5392
		5393	The act of calling the callback associated with a watcher.
		5394
		5395	=item event
		5396
		5397	A change of state of some external event, such as data now being available
		5398	for reading on a file descriptor, time having passed or simply not having
		5399	any other events happening anymore.
		5400
		5401	In libev, events are represented as single bits (such as C<EV_READ> or
		5402	C<EV_TIMER>).
		5403
		5404	=item event library
		5405
		5406	A software package implementing an event model and loop.
		5407
		5408	=item event loop
		5409
		5410	An entity that handles and processes external events and converts them
		5411	into callback invocations.
		5412
		5413	=item event model
		5414
		5415	The model used to describe how an event loop handles and processes
		5416	watchers and events.
		5417
		5418	=item pending
		5419
		5420	A watcher is pending as soon as the corresponding event has been
		5421	detected. See L<WATCHER STATES> for details.
		5422
		5423	=item real time
		5424
		5425	The physical time that is observed. It is apparently strictly monotonic :)
		5426
		5427	=item wall-clock time
		5428
		5429	The time and date as shown on clocks. Unlike real time, it can actually
		5430	be wrong and jump forwards and backwards, e.g. when you adjust your
		5431	clock.
		5432
		5433	=item watcher
		5434
		5435	A data structure that describes interest in certain events. Watchers need
		5436	to be started (attached to an event loop) before they can receive events.
		5437
		5438	=back
878		5439
879	=head1 AUTHOR	5440	=head1 AUTHOR
880		5441
881	Marc Lehmann <libev@schmorp.de>.	5442	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5443	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
882		5444

Diff Legend

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