ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.84 by root, Wed Dec 12 22:26:37 2007 UTC vs.
Revision 1.424 by root, Sun Nov 25 01:38:51 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 EXAMPLE PROGRAM	9	=head2 EXAMPLE PROGRAM
10		10
		11	// a single header file is required
11	#include <ev.h>	12	#include <ev.h>
12		13
		14	#include <stdio.h> // for puts
		15
		16	// every watcher type has its own typedef'd struct
		17	// with the name ev_TYPE
13	ev_io stdin_watcher;	18	ev_io stdin_watcher;
14	ev_timer timeout_watcher;	19	ev_timer timeout_watcher;
15		20
16	/* called when data readable on stdin */	21	// all watcher callbacks have a similar signature
		22	// this callback is called when data is readable on stdin
17	static void	23	static void
18	stdin_cb (EV_P_ struct ev_io *w, int revents)	24	stdin_cb (EV_P_ ev_io *w, int revents)
19	{	25	{
20	/* puts ("stdin ready"); */	26	puts ("stdin ready");
21	ev_io_stop (EV_A_ w); /* just a syntax example */	27	// for one-shot events, one must manually stop the watcher
22	ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */	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);
23	}	33	}
24		34
		35	// another callback, this time for a time-out
25	static void	36	static void
26	timeout_cb (EV_P_ struct ev_timer *w, int revents)	37	timeout_cb (EV_P_ ev_timer *w, int revents)
27	{	38	{
28	/* puts ("timeout"); */	39	puts ("timeout");
29	ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */	40	// this causes the innermost ev_run to stop iterating
		41	ev_break (EV_A_ EVBREAK_ONE);
30	}	42	}
31		43
32	int	44	int
33	main (void)	45	main (void)
34	{	46	{
35	struct ev_loop *loop = ev_default_loop (0);	47	// use the default event loop unless you have special needs
		48	struct ev_loop *loop = EV_DEFAULT;
36		49
37	/* initialise an io watcher, then start it */	50	// initialise an io watcher, then start it
		51	// this one will watch for stdin to become readable
38	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	52	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
39	ev_io_start (loop, &stdin_watcher);	53	ev_io_start (loop, &stdin_watcher);
40		54
		55	// initialise a timer watcher, then start it
41	/* simple non-repeating 5.5 second timeout */	56	// simple non-repeating 5.5 second timeout
42	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);	57	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
43	ev_timer_start (loop, &timeout_watcher);	58	ev_timer_start (loop, &timeout_watcher);
44		59
45	/* loop till timeout or data ready */	60	// now wait for events to arrive
46	ev_loop (loop, 0);	61	ev_run (loop, 0);
47		62
		63	// break was called, so exit
48	return 0;	64	return 0;
49	}	65	}
50		66
51	=head1 DESCRIPTION	67	=head1 ABOUT THIS DOCUMENT
52		68
		69	This document documents the libev software package.
		70
53	The newest version of this document is also available as a html-formatted	71	The newest version of this document is also available as an html-formatted
54	web page you might find easier to navigate when reading it for the first	72	web page you might find easier to navigate when reading it for the first
55	time: L<http://cvs.schmorp.de/libev/ev.html>.	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
56		92
57	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
58	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
59	these event sources and provide your program with events.	95	these event sources and provide your program with events.
60		96
61	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
62	(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
63	communicate events via a callback mechanism.	99	communicate events via a callback mechanism.
…		…
65	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
66	watchers>, which are relatively small C structures you initialise with the	102	watchers>, which are relatively small C structures you initialise with the
67	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
68	watcher.	104	watcher.
69		105
70	=head1 FEATURES	106	=head2 FEATURES
71		107
72	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	108	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	109	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	110	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers	111	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
76	with customised rescheduling (C<ev_periodic>), synchronous signals	112	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
77	(C<ev_signal>), process status change events (C<ev_child>), and event	113	timers (C<ev_timer>), absolute timers with customised rescheduling
78	watchers dealing with the event loop mechanism itself (C<ev_idle>,	114	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
79	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as	115	change events (C<ev_child>), and event watchers dealing with the event
80	file watchers (C<ev_stat>) and even limited support for fork events	116	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
81	(C<ev_fork>).	117	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		118	limited support for fork events (C<ev_fork>).
82		119
83	It also is quite fast (see this	120	It also is quite fast (see this
84	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent	121	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85	for example).	122	for example).
86		123
87	=head1 CONVENTIONS	124	=head2 CONVENTIONS
88		125
89	Libev is very configurable. In this manual the default configuration will	126	Libev is very configurable. In this manual the default (and most common)
90	be described, which supports multiple event loops. For more info about	127	configuration will be described, which supports multiple event loops. For
91	various configuration options please have a look at B<EMBED> section in	128	more info about various configuration options please have a look at
92	this manual. If libev was configured without support for multiple event	129	B<EMBED> section in this manual. If libev was configured without support
93	loops, then all functions taking an initial argument of name C<loop>	130	for multiple event loops, then all functions taking an initial argument of
94	(which is always of type C<struct ev_loop *>) will not have this argument.	131	name C<loop> (which is always of type C<struct ev_loop *>) will not have
		132	this argument.
95		133
96	=head1 TIME REPRESENTATION	134	=head2 TIME REPRESENTATION
97		135
98	Libev represents time as a single floating point number, representing the	136	Libev represents time as a single floating point number, representing
99	(fractional) number of seconds since the (POSIX) epoch (somewhere near	137	the (fractional) number of seconds since the (POSIX) epoch (in practice
100	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
101	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
102	to the C<double> type in C, and when you need to do any calculations on	140	too. It usually aliases to the C<double> type in C. When you need to do
103	it, you should treat it as such.	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
104		166
105	=head1 GLOBAL FUNCTIONS	167	=head1 GLOBAL FUNCTIONS
106		168
107	These functions can be called anytime, even before initialising the	169	These functions can be called anytime, even before initialising the
108	library in any way.	170	library in any way.
…		…
111		173
112	=item ev_tstamp ev_time ()	174	=item ev_tstamp ev_time ()
113		175
114	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
115	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
116	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 >>).
117		192
118	=item int ev_version_major ()	193	=item int ev_version_major ()
119		194
120	=item int ev_version_minor ()	195	=item int ev_version_minor ()
121		196
…		…
132	as this indicates an incompatible change. Minor versions are usually	207	as this indicates an incompatible change. Minor versions are usually
133	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
134	not a problem.	209	not a problem.
135		210
136	Example: Make sure we haven't accidentally been linked against the wrong	211	Example: Make sure we haven't accidentally been linked against the wrong
137	version.	212	version (note, however, that this will not detect other ABI mismatches,
		213	such as LFS or reentrancy).
138		214
139	assert (("libev version mismatch",	215	assert (("libev version mismatch",
140	ev_version_major () == EV_VERSION_MAJOR	216	ev_version_major () == EV_VERSION_MAJOR
141	&& ev_version_minor () >= EV_VERSION_MINOR));	217	&& ev_version_minor () >= EV_VERSION_MINOR));
142		218
143	=item unsigned int ev_supported_backends ()	219	=item unsigned int ev_supported_backends ()
144		220
145	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	221	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
146	value) compiled into this binary of libev (independent of their	222	value) compiled into this binary of libev (independent of their
…		…
148	a description of the set values.	224	a description of the set values.
149		225
150	Example: make sure we have the epoll method, because yeah this is cool and	226	Example: make sure we have the epoll method, because yeah this is cool and
151	a must have and can we have a torrent of it please!!!11	227	a must have and can we have a torrent of it please!!!11
152		228
153	assert (("sorry, no epoll, no sex",	229	assert (("sorry, no epoll, no sex",
154	ev_supported_backends () & EVBACKEND_EPOLL));	230	ev_supported_backends () & EVBACKEND_EPOLL));
155		231
156	=item unsigned int ev_recommended_backends ()	232	=item unsigned int ev_recommended_backends ()
157		233
158	Return the set of all backends compiled into this binary of libev and also	234	Return the set of all backends compiled into this binary of libev and
159	recommended for this platform. This set is often smaller than the one	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
160	returned by C<ev_supported_backends>, as for example kqueue is broken on	237	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
161	most BSDs and will not be autodetected unless you explicitly request it	238	and will not be auto-detected unless you explicitly request it (assuming
162	(assuming you know what you are doing). This is the set of backends that	239	you know what you are doing). This is the set of backends that libev will
163	libev will probe for if you specify no backends explicitly.	240	probe for if you specify no backends explicitly.
164		241
165	=item unsigned int ev_embeddable_backends ()	242	=item unsigned int ev_embeddable_backends ()
166		243
167	Returns the set of backends that are embeddable in other event loops. This	244	Returns the set of backends that are embeddable in other event loops. This
168	is the theoretical, all-platform, value. To find which backends	245	value is platform-specific but can include backends not available on the
169	might be supported on the current system, you would need to look at	246	current system. To find which embeddable backends might be supported on
170	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	247	the current system, you would need to look at C<ev_embeddable_backends ()
171	recommended ones.	248	& ev_supported_backends ()>, likewise for recommended ones.
172		249
173	See the description of C<ev_embed> watchers for more info.	250	See the description of C<ev_embed> watchers for more info.
174		251
175	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	252	=item ev_set_allocator (void *(*cb)(void *ptr, long size) throw ())
176		253
177	Sets the allocation function to use (the prototype is similar - the	254	Sets the allocation function to use (the prototype is similar - the
178	semantics is identical - to the realloc C function). It is used to	255	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
179	allocate and free memory (no surprises here). If it returns zero when	256	used to allocate and free memory (no surprises here). If it returns zero
180	memory needs to be allocated, the library might abort or take some	257	when memory needs to be allocated (C<size != 0>), the library might abort
181	potentially destructive action. The default is your system realloc	258	or take some potentially destructive action.
182	function.	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.
183		263
184	You could override this function in high-availability programs to, say,	264	You could override this function in high-availability programs to, say,
185	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,
186	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.
187		267
188	Example: Replace the libev allocator with one that waits a bit and then	268	Example: Replace the libev allocator with one that waits a bit and then
189	retries).	269	retries (example requires a standards-compliant C<realloc>).
190		270
191	static void *	271	static void *
192	persistent_realloc (void *ptr, size_t size)	272	persistent_realloc (void *ptr, size_t size)
193	{	273	{
194	for (;;)	274	for (;;)
…		…
203	}	283	}
204		284
205	...	285	...
206	ev_set_allocator (persistent_realloc);	286	ev_set_allocator (persistent_realloc);
207		287
208	=item ev_set_syserr_cb (void (*cb)(const char *msg));	288	=item ev_set_syserr_cb (void (*cb)(const char *msg) throw ())
209		289
210	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
211	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
212	indicating the system call or subsystem causing the problem. If this	292	indicating the system call or subsystem causing the problem. If this
213	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
214	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
215	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
216	(such as abort).	296	(such as abort).
217		297
218	Example: This is basically the same thing that libev does internally, too.	298	Example: This is basically the same thing that libev does internally, too.
…		…
225	}	305	}
226		306
227	...	307	...
228	ev_set_syserr_cb (fatal_error);	308	ev_set_syserr_cb (fatal_error);
229		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
230	=back	323	=back
231		324
232	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	325	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
233		326
234	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
235	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
236	events, and dynamically created loops which do not.	329	libev 3 had an C<ev_loop> function colliding with the struct name).
237		330
238	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
239	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
240	create, you also create another event loop. Libev itself does no locking	333	do not.
241	whatsoever, so if you mix calls to the same event loop in different
242	threads, make sure you lock (this is usually a bad idea, though, even if
243	done correctly, because it's hideous and inefficient).
244		334
245	=over 4	335	=over 4
246		336
247	=item struct ev_loop *ev_default_loop (unsigned int flags)	337	=item struct ev_loop *ev_default_loop (unsigned int flags)
248		338
249	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
250	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
251	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
252	flags. If that is troubling you, check C<ev_backend ()> afterwards).	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".
253		349
254	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
255	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.
256		382
257	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
258	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	384	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
259		385
260	The following flags are supported:	386	The following flags are supported:
…		…
266	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
267	thing, believe me).	393	thing, believe me).
268		394
269	=item C<EVFLAG_NOENV>	395	=item C<EVFLAG_NOENV>
270		396
271	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
272	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
273	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	399	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
274	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
275	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
276	around bugs.	402	around bugs.
277		403
278	=item C<EVFLAG_FORKCHECK>	404	=item C<EVFLAG_FORKCHECK>
279		405
280	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after	406	Instead of calling C<ev_loop_fork> manually after a fork, you can also
281	a fork, you can also make libev check for a fork in each iteration by	407	make libev check for a fork in each iteration by enabling this flag.
282	enabling this flag.
283		408
284	This works by calling C<getpid ()> on every iteration of the loop,	409	This works by calling C<getpid ()> on every iteration of the loop,
285	and thus this might slow down your event loop if you do a lot of loop	410	and thus this might slow down your event loop if you do a lot of loop
286	iterations and little real work, but is usually not noticeable (on my	411	iterations and little real work, but is usually not noticeable (on my
287	Linux system for example, C<getpid> is actually a simple 5-insn sequence	412	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
288	without a syscall and thus I<very> fast, but my Linux system also has	413	without a system call and thus I<very> fast, but my GNU/Linux system also has
289	C<pthread_atfork> which is even faster).	414	C<pthread_atfork> which is even faster).
290		415
291	The big advantage of this flag is that you can forget about fork (and	416	The big advantage of this flag is that you can forget about fork (and
292	forget about forgetting to tell libev about forking) when you use this	417	forget about forgetting to tell libev about forking) when you use this
293	flag.	418	flag.
294		419
295	This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS>	420	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
296	environment variable.	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.
297		457
298	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	458	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
299		459
300	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
301	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,
302	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
303	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
304	the fastest backend for a low number of fds.	464	usually the fastest backend for a low number of (low-numbered :) fds.
		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).
305		476
306	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	477	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
307		478
308	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
309	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
310	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
311	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.
		485
		486	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		487	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
312		488
313	=item C<EVBACKEND_EPOLL> (value 4, Linux)	489	=item C<EVBACKEND_EPOLL> (value 4, Linux)
314		490
		491	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		492	kernels).
		493
315	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
316	but it scales phenomenally better. While poll and select usually scale like	495	it scales phenomenally better. While poll and select usually scale like
317	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
318	either O(1) or O(active_fds).	497	fd), epoll scales either O(1) or O(active_fds).
319		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
320	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
321	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
322	(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
323	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
324	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.
325		533
326	Please note that epoll sometimes generates spurious notifications, so you	534	Best performance from this backend is achieved by not unregistering all
327	need to use non-blocking I/O or other means to avoid blocking when no data	535	watchers for a file descriptor until it has been closed, if possible,
328	(or space) is available.	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>.
329		551
330	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	552	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
331		553
332	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
333	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
334	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
335	completely useless). For this reason its not being "autodetected"	557	it's completely useless). Unlike epoll, however, whose brokenness
		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
336	unless you explicitly specify it explicitly in the flags (i.e. using	560	"auto-detected" unless you explicitly specify it in the flags (i.e. using
337	C<EVBACKEND_KQUEUE>).	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.
338		567
339	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
340	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
341	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
342	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
343	incident, so its best to avoid that.	572	two event changes per incident. Support for C<fork ()> is very bad (you
		573	might have to leak fd's on fork, but it's more sane than epoll) and it
		574	drops fds silently in similarly hard-to-detect cases.
		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>.
344		588
345	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	589	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
346		590
347	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.
348		595
349	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	596	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
350		597
351	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,
352	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)).
353		600
354	Please note that solaris ports can result in a lot of spurious	601	While this backend scales well, it requires one system call per active
355	notifications, so you need to use non-blocking I/O or other means to avoid	602	file descriptor per loop iteration. For small and medium numbers of file
356	blocking when no data (or space) is available.	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>.
357		623
358	=item C<EVBACKEND_ALL>	624	=item C<EVBACKEND_ALL>
359		625
360	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
361	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
362	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	628	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
363		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).
		639
364	=back	640	=back
365		641
366	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,
367	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
368	specified, most compiled-in backend will be tried, usually in reverse	644	here). If none are specified, all backends in C<ev_recommended_backends
369	order of their flag values :)	645	()> will be tried.
370
371	The most typical usage is like this:
372
373	if (!ev_default_loop (0))
374	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
375
376	Restrict libev to the select and poll backends, and do not allow
377	environment settings to be taken into account:
378
379	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
380
381	Use whatever libev has to offer, but make sure that kqueue is used if
382	available (warning, breaks stuff, best use only with your own private
383	event loop and only if you know the OS supports your types of fds):
384
385	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
386
387	=item struct ev_loop *ev_loop_new (unsigned int flags)
388
389	Similar to C<ev_default_loop>, but always creates a new event loop that is
390	always distinct from the default loop. Unlike the default loop, it cannot
391	handle signal and child watchers, and attempts to do so will be greeted by
392	undefined behaviour (or a failed assertion if assertions are enabled).
393		646
394	Example: Try to create a event loop that uses epoll and nothing else.	647	Example: Try to create a event loop that uses epoll and nothing else.
395		648
396	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	649	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
397	if (!epoller)	650	if (!epoller)
398	fatal ("no epoll found here, maybe it hides under your chair");	651	fatal ("no epoll found here, maybe it hides under your chair");
399		652
		653	Example: Use whatever libev has to offer, but make sure that kqueue is
		654	used if available.
		655
		656	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
		657
400	=item ev_default_destroy ()	658	=item ev_loop_destroy (loop)
401		659
402	Destroys the default loop again (frees all memory and kernel state	660	Destroys an event loop object (frees all memory and kernel state
403	etc.). None of the active event watchers will be stopped in the normal	661	etc.). None of the active event watchers will be stopped in the normal
404	sense, so e.g. C<ev_is_active> might still return true. It is your	662	sense, so e.g. C<ev_is_active> might still return true. It is your
405	responsibility to either stop all watchers cleanly yoursef I<before>	663	responsibility to either stop all watchers cleanly yourself I<before>
406	calling this function, or cope with the fact afterwards (which is usually	664	calling this function, or cope with the fact afterwards (which is usually
407	the easiest thing, youc na just ignore the watchers and/or C<free ()> them	665	the easiest thing, you can just ignore the watchers and/or C<free ()> them
408	for example).	666	for example).
409		667
		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.
		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
410	=item ev_loop_destroy (loop)	681	=item ev_loop_fork (loop)
411		682
412	Like C<ev_default_destroy>, but destroys an event loop created by an	683	This function sets a flag that causes subsequent C<ev_run> iterations to
413	earlier call to C<ev_loop_new>.
414
415	=item ev_default_fork ()
416
417	This function reinitialises the kernel state for backends that have	684	reinitialise the kernel state for backends that have one. Despite the
418	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
419	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
420	again makes little sense).	687	child before resuming or calling C<ev_run>.
421		688
422	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
423	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
424	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).
425		700
426	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
427	it just in case after a fork. To make this easy, the function will fit in	702	it just in case after a fork.
428	quite nicely into a call to C<pthread_atfork>:
429		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	...
430	pthread_atfork (0, 0, ev_default_fork);	714	pthread_atfork (0, 0, post_fork_child);
431		715
432	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use	716	=item int ev_is_default_loop (loop)
433	without calling this function, so if you force one of those backends you
434	do not need to care.
435		717
436	=item ev_loop_fork (loop)	718	Returns true when the given loop is, in fact, the default loop, and false
		719	otherwise.
437		720
438	Like C<ev_default_fork>, but acts on an event loop created by
439	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
440	after fork, and how you do this is entirely your own problem.
441
442	=item unsigned int ev_loop_count (loop)	721	=item unsigned int ev_iteration (loop)
443		722
444	Returns the count of loop iterations for the loop, which is identical to	723	Returns the current iteration count for the event loop, which is identical
445	the number of times libev did poll for new events. It starts at C<0> and	724	to the number of times libev did poll for new events. It starts at C<0>
446	happily wraps around with enough iterations.	725	and happily wraps around with enough iterations.
447		726
448	This value can sometimes be useful as a generation counter of sorts (it	727	This value can sometimes be useful as a generation counter of sorts (it
449	"ticks" the number of loop iterations), as it roughly corresponds with	728	"ticks" the number of loop iterations), as it roughly corresponds with
450	C<ev_prepare> and C<ev_check> calls.	729	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		730	prepare and check phases.
		731
		732	=item unsigned int ev_depth (loop)
		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.
451		745
452	=item unsigned int ev_backend (loop)	746	=item unsigned int ev_backend (loop)
453		747
454	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	748	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
455	use.	749	use.
…		…
458		752
459	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
460	received events and started processing them. This timestamp does not	754	received events and started processing them. This timestamp does not
461	change as long as callbacks are being processed, and this is also the base	755	change as long as callbacks are being processed, and this is also the base
462	time used for relative timers. You can treat it as the timestamp of the	756	time used for relative timers. You can treat it as the timestamp of the
463	event occuring (or more correctly, libev finding out about it).	757	event occurring (or more correctly, libev finding out about it).
464		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
465	=item ev_loop (loop, int flags)	797	=item bool ev_run (loop, int flags)
466		798
467	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
468	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
469	events.	801	handling events. It will ask the operating system for any new events, call
		802	the watcher callbacks, and then repeat the whole process indefinitely: This
		803	is why event loops are called I<loops>.
470		804
471	If the flags argument is specified as C<0>, it will not return until	805	If the flags argument is specified as C<0>, it will keep handling events
472	either 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.
473		808
		809	The return value is false if there are no more active watchers (which
		810	usually means "all jobs done" or "deadlock"), and true in all other cases
		811	(which usually means " you should call C<ev_run> again").
		812
474	Please note that an explicit C<ev_unloop> is usually better than	813	Please note that an explicit C<ev_break> is usually better than
475	relying on all watchers to be stopped when deciding when a program has	814	relying on all watchers to be stopped when deciding when a program has
476	finished (especially in interactive programs), but having a program that	815	finished (especially in interactive programs), but having a program
477	automatically loops as long as it has to and no longer by virtue of	816	that automatically loops as long as it has to and no longer by virtue
478	relying on its watchers stopping correctly is a thing of beauty.	817	of relying on its watchers stopping correctly, that is truly a thing of
		818	beauty.
479		819
		820	This function is I<mostly> exception-safe - you can break out of a
		821	C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		822	exception and so on. This does not decrement the C<ev_depth> value, nor
		823	will it clear any outstanding C<EVBREAK_ONE> breaks.
		824
480	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	825	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
481	those events and any outstanding ones, but will not block your process in	826	those events and any already outstanding ones, but will not wait and
482	case there are no events and will return after one iteration of the loop.	827	block your process in case there are no events and will return after one
		828	iteration of the loop. This is sometimes useful to poll and handle new
		829	events while doing lengthy calculations, to keep the program responsive.
483		830
484	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	831	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
485	neccessary) and will handle those and any outstanding ones. It will block	832	necessary) and will handle those and any already outstanding ones. It
486	your process until at least one new event arrives, and will return after	833	will block your process until at least one new event arrives (which could
487	one iteration of the loop. This is useful if you are waiting for some	834	be an event internal to libev itself, so there is no guarantee that a
488	external event in conjunction with something not expressible using other	835	user-registered callback will be called), and will return after one
		836	iteration of the loop.
		837
		838	This is useful if you are waiting for some external event in conjunction
		839	with something not expressible using other libev watchers (i.e. "roll your
489	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	840	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
490	usually a better approach for this kind of thing.	841	usually a better approach for this kind of thing.
491		842
492	Here are the gory details of what C<ev_loop> does:	843	Here are the gory details of what C<ev_run> does (this is for your
		844	understanding, not a guarantee that things will work exactly like this in
		845	future versions):
493		846
		847	- Increment loop depth.
		848	- Reset the ev_break status.
494	- Before the first iteration, call any pending watchers.	849	- Before the first iteration, call any pending watchers.
495	* If there are no active watchers (reference count is zero), return.	850	LOOP:
496	- Queue all prepare watchers and then call all outstanding watchers.	851	- If EVFLAG_FORKCHECK was used, check for a fork.
		852	- If a fork was detected (by any means), queue and call all fork watchers.
		853	- Queue and call all prepare watchers.
		854	- If ev_break was called, goto FINISH.
497	- If we have been forked, recreate the kernel state.	855	- If we have been forked, detach and recreate the kernel state
		856	as to not disturb the other process.
498	- Update the kernel state with all outstanding changes.	857	- Update the kernel state with all outstanding changes.
499	- Update the "event loop time".	858	- Update the "event loop time" (ev_now ()).
500	- Calculate for how long to block.	859	- Calculate for how long to sleep or block, if at all
		860	(active idle watchers, EVRUN_NOWAIT or not having
		861	any active watchers at all will result in not sleeping).
		862	- Sleep if the I/O and timer collect interval say so.
		863	- Increment loop iteration counter.
501	- Block the process, waiting for any events.	864	- Block the process, waiting for any events.
502	- Queue all outstanding I/O (fd) events.	865	- Queue all outstanding I/O (fd) events.
503	- Update the "event loop time" and do time jump handling.	866	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
504	- Queue all outstanding timers.	867	- Queue all expired timers.
505	- Queue all outstanding periodics.	868	- Queue all expired periodics.
506	- If no events are pending now, queue all idle watchers.	869	- Queue all idle watchers with priority higher than that of pending events.
507	- Queue all check watchers.	870	- Queue all check watchers.
508	- Call all queued watchers in reverse order (i.e. check watchers first).	871	- Call all queued watchers in reverse order (i.e. check watchers first).
509	Signals and child watchers are implemented as I/O watchers, and will	872	Signals and child watchers are implemented as I/O watchers, and will
510	be handled here by queueing them when their watcher gets executed.	873	be handled here by queueing them when their watcher gets executed.
511	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	874	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
512	were used, return, otherwise continue with step *.	875	were used, or there are no active watchers, goto FINISH, otherwise
		876	continue with step LOOP.
		877	FINISH:
		878	- Reset the ev_break status iff it was EVBREAK_ONE.
		879	- Decrement the loop depth.
		880	- Return.
513		881
514	Example: Queue some jobs and then loop until no events are outsanding	882	Example: Queue some jobs and then loop until no events are outstanding
515	anymore.	883	anymore.
516		884
517	... queue jobs here, make sure they register event watchers as long	885	... queue jobs here, make sure they register event watchers as long
518	... as they still have work to do (even an idle watcher will do..)	886	... as they still have work to do (even an idle watcher will do..)
519	ev_loop (my_loop, 0);	887	ev_run (my_loop, 0);
520	... jobs done. yeah!	888	... jobs done or somebody called break. yeah!
521		889
522	=item ev_unloop (loop, how)	890	=item ev_break (loop, how)
523		891
524	Can be used to make a call to C<ev_loop> return early (but only after it	892	Can be used to make a call to C<ev_run> return early (but only after it
525	has processed all outstanding events). The C<how> argument must be either	893	has processed all outstanding events). The C<how> argument must be either
526	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	894	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
527	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	895	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
		896
		897	This "break state" will be cleared on the next call to C<ev_run>.
		898
		899	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		900	which case it will have no effect.
528		901
529	=item ev_ref (loop)	902	=item ev_ref (loop)
530		903
531	=item ev_unref (loop)	904	=item ev_unref (loop)
532		905
533	Ref/unref can be used to add or remove a reference count on the event	906	Ref/unref can be used to add or remove a reference count on the event
534	loop: Every watcher keeps one reference, and as long as the reference	907	loop: Every watcher keeps one reference, and as long as the reference
535	count is nonzero, C<ev_loop> will not return on its own. If you have	908	count is nonzero, C<ev_run> will not return on its own.
536	a watcher you never unregister that should not keep C<ev_loop> from	909
537	returning, ev_unref() after starting, and ev_ref() before stopping it. For	910	This is useful when you have a watcher that you never intend to
		911	unregister, but that nevertheless should not keep C<ev_run> from
		912	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		913	before stopping it.
		914
538	example, libev itself uses this for its internal signal pipe: It is not	915	As an example, libev itself uses this for its internal signal pipe: It
539	visible to the libev user and should not keep C<ev_loop> from exiting if	916	is not visible to the libev user and should not keep C<ev_run> from
540	no event watchers registered by it are active. It is also an excellent	917	exiting if no event watchers registered by it are active. It is also an
541	way to do this for generic recurring timers or from within third-party	918	excellent way to do this for generic recurring timers or from within
542	libraries. Just remember to I<unref after start> and I<ref before stop>.	919	third-party libraries. Just remember to I<unref after start> and I<ref
		920	before stop> (but only if the watcher wasn't active before, or was active
		921	before, respectively. Note also that libev might stop watchers itself
		922	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		923	in the callback).
543		924
544	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	925	Example: Create a signal watcher, but keep it from keeping C<ev_run>
545	running when nothing else is active.	926	running when nothing else is active.
546		927
547	struct ev_signal exitsig;	928	ev_signal exitsig;
548	ev_signal_init (&exitsig, sig_cb, SIGINT);	929	ev_signal_init (&exitsig, sig_cb, SIGINT);
549	ev_signal_start (loop, &exitsig);	930	ev_signal_start (loop, &exitsig);
550	evf_unref (loop);	931	ev_unref (loop);
551		932
552	Example: For some weird reason, unregister the above signal handler again.	933	Example: For some weird reason, unregister the above signal handler again.
553		934
554	ev_ref (loop);	935	ev_ref (loop);
555	ev_signal_stop (loop, &exitsig);	936	ev_signal_stop (loop, &exitsig);
		937
		938	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		939
		940	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		941
		942	These advanced functions influence the time that libev will spend waiting
		943	for events. Both time intervals are by default C<0>, meaning that libev
		944	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		945	latency.
		946
		947	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		948	allows libev to delay invocation of I/O and timer/periodic callbacks
		949	to increase efficiency of loop iterations (or to increase power-saving
		950	opportunities).
		951
		952	The idea is that sometimes your program runs just fast enough to handle
		953	one (or very few) event(s) per loop iteration. While this makes the
		954	program responsive, it also wastes a lot of CPU time to poll for new
		955	events, especially with backends like C<select ()> which have a high
		956	overhead for the actual polling but can deliver many events at once.
		957
		958	By setting a higher I<io collect interval> you allow libev to spend more
		959	time collecting I/O events, so you can handle more events per iteration,
		960	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		961	C<ev_timer>) will not be affected. Setting this to a non-null value will
		962	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		963	sleep time ensures that libev will not poll for I/O events more often then
		964	once per this interval, on average (as long as the host time resolution is
		965	good enough).
		966
		967	Likewise, by setting a higher I<timeout collect interval> you allow libev
		968	to spend more time collecting timeouts, at the expense of increased
		969	latency/jitter/inexactness (the watcher callback will be called
		970	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		971	value will not introduce any overhead in libev.
		972
		973	Many (busy) programs can usually benefit by setting the I/O collect
		974	interval to a value near C<0.1> or so, which is often enough for
		975	interactive servers (of course not for games), likewise for timeouts. It
		976	usually doesn't make much sense to set it to a lower value than C<0.01>,
		977	as this approaches the timing granularity of most systems. Note that if
		978	you do transactions with the outside world and you can't increase the
		979	parallelity, then this setting will limit your transaction rate (if you
		980	need to poll once per transaction and the I/O collect interval is 0.01,
		981	then you can't do more than 100 transactions per second).
		982
		983	Setting the I<timeout collect interval> can improve the opportunity for
		984	saving power, as the program will "bundle" timer callback invocations that
		985	are "near" in time together, by delaying some, thus reducing the number of
		986	times the process sleeps and wakes up again. Another useful technique to
		987	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		988	they fire on, say, one-second boundaries only.
		989
		990	Example: we only need 0.1s timeout granularity, and we wish not to poll
		991	more often than 100 times per second:
		992
		993	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		994	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		995
		996	=item ev_invoke_pending (loop)
		997
		998	This call will simply invoke all pending watchers while resetting their
		999	pending state. Normally, C<ev_run> does this automatically when required,
		1000	but when overriding the invoke callback this call comes handy. This
		1001	function can be invoked from a watcher - this can be useful for example
		1002	when you want to do some lengthy calculation and want to pass further
		1003	event handling to another thread (you still have to make sure only one
		1004	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		1005
		1006	=item int ev_pending_count (loop)
		1007
		1008	Returns the number of pending watchers - zero indicates that no watchers
		1009	are pending.
		1010
		1011	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		1012
		1013	This overrides the invoke pending functionality of the loop: Instead of
		1014	invoking all pending watchers when there are any, C<ev_run> will call
		1015	this callback instead. This is useful, for example, when you want to
		1016	invoke the actual watchers inside another context (another thread etc.).
		1017
		1018	If you want to reset the callback, use C<ev_invoke_pending> as new
		1019	callback.
		1020
		1021	=item ev_set_loop_release_cb (loop, void (*release)(EV_P) throw (), void (*acquire)(EV_P) throw ())
		1022
		1023	Sometimes you want to share the same loop between multiple threads. This
		1024	can be done relatively simply by putting mutex_lock/unlock calls around
		1025	each call to a libev function.
		1026
		1027	However, C<ev_run> can run an indefinite time, so it is not feasible
		1028	to wait for it to return. One way around this is to wake up the event
		1029	loop via C<ev_break> and C<ev_async_send>, another way is to set these
		1030	I<release> and I<acquire> callbacks on the loop.
		1031
		1032	When set, then C<release> will be called just before the thread is
		1033	suspended waiting for new events, and C<acquire> is called just
		1034	afterwards.
		1035
		1036	Ideally, C<release> will just call your mutex_unlock function, and
		1037	C<acquire> will just call the mutex_lock function again.
		1038
		1039	While event loop modifications are allowed between invocations of
		1040	C<release> and C<acquire> (that's their only purpose after all), no
		1041	modifications done will affect the event loop, i.e. adding watchers will
		1042	have no effect on the set of file descriptors being watched, or the time
		1043	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1044	to take note of any changes you made.
		1045
		1046	In theory, threads executing C<ev_run> will be async-cancel safe between
		1047	invocations of C<release> and C<acquire>.
		1048
		1049	See also the locking example in the C<THREADS> section later in this
		1050	document.
		1051
		1052	=item ev_set_userdata (loop, void *data)
		1053
		1054	=item void *ev_userdata (loop)
		1055
		1056	Set and retrieve a single C<void *> associated with a loop. When
		1057	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1058	C<0>.
		1059
		1060	These two functions can be used to associate arbitrary data with a loop,
		1061	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1062	C<acquire> callbacks described above, but of course can be (ab-)used for
		1063	any other purpose as well.
		1064
		1065	=item ev_verify (loop)
		1066
		1067	This function only does something when C<EV_VERIFY> support has been
		1068	compiled in, which is the default for non-minimal builds. It tries to go
		1069	through all internal structures and checks them for validity. If anything
		1070	is found to be inconsistent, it will print an error message to standard
		1071	error and call C<abort ()>.
		1072
		1073	This can be used to catch bugs inside libev itself: under normal
		1074	circumstances, this function will never abort as of course libev keeps its
		1075	data structures consistent.
556		1076
557	=back	1077	=back
558		1078
559		1079
560	=head1 ANATOMY OF A WATCHER	1080	=head1 ANATOMY OF A WATCHER
561		1081
		1082	In the following description, uppercase C<TYPE> in names stands for the
		1083	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1084	watchers and C<ev_io_start> for I/O watchers.
		1085
562	A watcher is a structure that you create and register to record your	1086	A watcher is an opaque structure that you allocate and register to record
563	interest in some event. For instance, if you want to wait for STDIN to	1087	your interest in some event. To make a concrete example, imagine you want
564	become readable, you would create an C<ev_io> watcher for that:	1088	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1089	for that:
565		1090
566	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1091	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
567	{	1092	{
568	ev_io_stop (w);	1093	ev_io_stop (w);
569	ev_unloop (loop, EVUNLOOP_ALL);	1094	ev_break (loop, EVBREAK_ALL);
570	}	1095	}
571		1096
572	struct ev_loop *loop = ev_default_loop (0);	1097	struct ev_loop *loop = ev_default_loop (0);
		1098
573	struct ev_io stdin_watcher;	1099	ev_io stdin_watcher;
		1100
574	ev_init (&stdin_watcher, my_cb);	1101	ev_init (&stdin_watcher, my_cb);
575	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1102	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
576	ev_io_start (loop, &stdin_watcher);	1103	ev_io_start (loop, &stdin_watcher);
		1104
577	ev_loop (loop, 0);	1105	ev_run (loop, 0);
578		1106
579	As you can see, you are responsible for allocating the memory for your	1107	As you can see, you are responsible for allocating the memory for your
580	watcher structures (and it is usually a bad idea to do this on the stack,	1108	watcher structures (and it is I<usually> a bad idea to do this on the
581	although this can sometimes be quite valid).	1109	stack).
582		1110
		1111	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1112	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1113
583	Each watcher structure must be initialised by a call to C<ev_init	1114	Each watcher structure must be initialised by a call to C<ev_init (watcher
584	(watcher *, callback)>, which expects a callback to be provided. This	1115	*, callback)>, which expects a callback to be provided. This callback is
585	callback gets invoked each time the event occurs (or, in the case of io	1116	invoked each time the event occurs (or, in the case of I/O watchers, each
586	watchers, each time the event loop detects that the file descriptor given	1117	time the event loop detects that the file descriptor given is readable
587	is readable and/or writable).	1118	and/or writable).
588		1119
589	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1120	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
590	with arguments specific to this watcher type. There is also a macro	1121	macro to configure it, with arguments specific to the watcher type. There
591	to combine initialisation and setting in one call: C<< ev_<type>_init	1122	is also a macro to combine initialisation and setting in one call: C<<
592	(watcher *, callback, ...) >>.	1123	ev_TYPE_init (watcher *, callback, ...) >>.
593		1124
594	To make the watcher actually watch out for events, you have to start it	1125	To make the watcher actually watch out for events, you have to start it
595	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1126	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
596	*) >>), and you can stop watching for events at any time by calling the	1127	*) >>), and you can stop watching for events at any time by calling the
597	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1128	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
598		1129
599	As long as your watcher is active (has been started but not stopped) you	1130	As long as your watcher is active (has been started but not stopped) you
600	must not touch the values stored in it. Most specifically you must never	1131	must not touch the values stored in it. Most specifically you must never
601	reinitialise it or call its C<set> macro.	1132	reinitialise it or call its C<ev_TYPE_set> macro.
602		1133
603	Each and every callback receives the event loop pointer as first, the	1134	Each and every callback receives the event loop pointer as first, the
604	registered watcher structure as second, and a bitset of received events as	1135	registered watcher structure as second, and a bitset of received events as
605	third argument.	1136	third argument.
606		1137
…		…
615	=item C<EV_WRITE>	1146	=item C<EV_WRITE>
616		1147
617	The file descriptor in the C<ev_io> watcher has become readable and/or	1148	The file descriptor in the C<ev_io> watcher has become readable and/or
618	writable.	1149	writable.
619		1150
620	=item C<EV_TIMEOUT>	1151	=item C<EV_TIMER>
621		1152
622	The C<ev_timer> watcher has timed out.	1153	The C<ev_timer> watcher has timed out.
623		1154
624	=item C<EV_PERIODIC>	1155	=item C<EV_PERIODIC>
625		1156
…		…
643		1174
644	=item C<EV_PREPARE>	1175	=item C<EV_PREPARE>
645		1176
646	=item C<EV_CHECK>	1177	=item C<EV_CHECK>
647		1178
648	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1179	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts to
649	to gather new events, and all C<ev_check> watchers are invoked just after	1180	gather new events, and all C<ev_check> watchers are queued (not invoked)
650	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1181	just after C<ev_run> has gathered them, but before it queues any callbacks
		1182	for any received events. That means C<ev_prepare> watchers are the last
		1183	watchers invoked before the event loop sleeps or polls for new events, and
		1184	C<ev_check> watchers will be invoked before any other watchers of the same
		1185	or lower priority within an event loop iteration.
		1186
651	received events. Callbacks of both watcher types can start and stop as	1187	Callbacks of both watcher types can start and stop as many watchers as
652	many watchers as they want, and all of them will be taken into account	1188	they want, and all of them will be taken into account (for example, a
653	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1189	C<ev_prepare> watcher might start an idle watcher to keep C<ev_run> from
654	C<ev_loop> from blocking).	1190	blocking).
655		1191
656	=item C<EV_EMBED>	1192	=item C<EV_EMBED>
657		1193
658	The embedded event loop specified in the C<ev_embed> watcher needs attention.	1194	The embedded event loop specified in the C<ev_embed> watcher needs attention.
659		1195
660	=item C<EV_FORK>	1196	=item C<EV_FORK>
661		1197
662	The event loop has been resumed in the child process after fork (see	1198	The event loop has been resumed in the child process after fork (see
663	C<ev_fork>).	1199	C<ev_fork>).
664		1200
		1201	=item C<EV_CLEANUP>
		1202
		1203	The event loop is about to be destroyed (see C<ev_cleanup>).
		1204
		1205	=item C<EV_ASYNC>
		1206
		1207	The given async watcher has been asynchronously notified (see C<ev_async>).
		1208
		1209	=item C<EV_CUSTOM>
		1210
		1211	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1212	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1213
665	=item C<EV_ERROR>	1214	=item C<EV_ERROR>
666		1215
667	An unspecified error has occured, the watcher has been stopped. This might	1216	An unspecified error has occurred, the watcher has been stopped. This might
668	happen because the watcher could not be properly started because libev	1217	happen because the watcher could not be properly started because libev
669	ran out of memory, a file descriptor was found to be closed or any other	1218	ran out of memory, a file descriptor was found to be closed or any other
		1219	problem. Libev considers these application bugs.
		1220
670	problem. You best act on it by reporting the problem and somehow coping	1221	You best act on it by reporting the problem and somehow coping with the
671	with the watcher being stopped.	1222	watcher being stopped. Note that well-written programs should not receive
		1223	an error ever, so when your watcher receives it, this usually indicates a
		1224	bug in your program.
672		1225
673	Libev will usually signal a few "dummy" events together with an error,	1226	Libev will usually signal a few "dummy" events together with an error, for
674	for example it might indicate that a fd is readable or writable, and if	1227	example it might indicate that a fd is readable or writable, and if your
675	your callbacks is well-written it can just attempt the operation and cope	1228	callbacks is well-written it can just attempt the operation and cope with
676	with the error from read() or write(). This will not work in multithreaded	1229	the error from read() or write(). This will not work in multi-threaded
677	programs, though, so beware.	1230	programs, though, as the fd could already be closed and reused for another
		1231	thing, so beware.
678		1232
679	=back	1233	=back
680		1234
681	=head2 GENERIC WATCHER FUNCTIONS	1235	=head2 GENERIC WATCHER FUNCTIONS
682
683	In the following description, C<TYPE> stands for the watcher type,
684	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
685		1236
686	=over 4	1237	=over 4
687		1238
688	=item C<ev_init> (ev_TYPE *watcher, callback)	1239	=item C<ev_init> (ev_TYPE *watcher, callback)
689		1240
…		…
695	which rolls both calls into one.	1246	which rolls both calls into one.
696		1247
697	You can reinitialise a watcher at any time as long as it has been stopped	1248	You can reinitialise a watcher at any time as long as it has been stopped
698	(or never started) and there are no pending events outstanding.	1249	(or never started) and there are no pending events outstanding.
699		1250
700	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	1251	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
701	int revents)>.	1252	int revents)>.
702		1253
		1254	Example: Initialise an C<ev_io> watcher in two steps.
		1255
		1256	ev_io w;
		1257	ev_init (&w, my_cb);
		1258	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1259
703	=item C<ev_TYPE_set> (ev_TYPE *, [args])	1260	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
704		1261
705	This macro initialises the type-specific parts of a watcher. You need to	1262	This macro initialises the type-specific parts of a watcher. You need to
706	call C<ev_init> at least once before you call this macro, but you can	1263	call C<ev_init> at least once before you call this macro, but you can
707	call C<ev_TYPE_set> any number of times. You must not, however, call this	1264	call C<ev_TYPE_set> any number of times. You must not, however, call this
708	macro on a watcher that is active (it can be pending, however, which is a	1265	macro on a watcher that is active (it can be pending, however, which is a
709	difference to the C<ev_init> macro).	1266	difference to the C<ev_init> macro).
710		1267
711	Although some watcher types do not have type-specific arguments	1268	Although some watcher types do not have type-specific arguments
712	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	1269	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
713		1270
		1271	See C<ev_init>, above, for an example.
		1272
714	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	1273	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
715		1274
716	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	1275	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
717	calls into a single call. This is the most convinient method to initialise	1276	calls into a single call. This is the most convenient method to initialise
718	a watcher. The same limitations apply, of course.	1277	a watcher. The same limitations apply, of course.
719		1278
		1279	Example: Initialise and set an C<ev_io> watcher in one step.
		1280
		1281	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1282
720	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	1283	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
721		1284
722	Starts (activates) the given watcher. Only active watchers will receive	1285	Starts (activates) the given watcher. Only active watchers will receive
723	events. If the watcher is already active nothing will happen.	1286	events. If the watcher is already active nothing will happen.
724		1287
		1288	Example: Start the C<ev_io> watcher that is being abused as example in this
		1289	whole section.
		1290
		1291	ev_io_start (EV_DEFAULT_UC, &w);
		1292
725	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	1293	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
726		1294
727	Stops the given watcher again (if active) and clears the pending	1295	Stops the given watcher if active, and clears the pending status (whether
		1296	the watcher was active or not).
		1297
728	status. It is possible that stopped watchers are pending (for example,	1298	It is possible that stopped watchers are pending - for example,
729	non-repeating timers are being stopped when they become pending), but	1299	non-repeating timers are being stopped when they become pending - but
730	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1300	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
731	you want to free or reuse the memory used by the watcher it is therefore a	1301	pending. If you want to free or reuse the memory used by the watcher it is
732	good idea to always call its C<ev_TYPE_stop> function.	1302	therefore a good idea to always call its C<ev_TYPE_stop> function.
733		1303
734	=item bool ev_is_active (ev_TYPE *watcher)	1304	=item bool ev_is_active (ev_TYPE *watcher)
735		1305
736	Returns a true value iff the watcher is active (i.e. it has been started	1306	Returns a true value iff the watcher is active (i.e. it has been started
737	and not yet been stopped). As long as a watcher is active you must not modify	1307	and not yet been stopped). As long as a watcher is active you must not modify
…		…
748		1318
749	=item callback ev_cb (ev_TYPE *watcher)	1319	=item callback ev_cb (ev_TYPE *watcher)
750		1320
751	Returns the callback currently set on the watcher.	1321	Returns the callback currently set on the watcher.
752		1322
753	=item ev_cb_set (ev_TYPE *watcher, callback)	1323	=item ev_set_cb (ev_TYPE *watcher, callback)
754		1324
755	Change the callback. You can change the callback at virtually any time	1325	Change the callback. You can change the callback at virtually any time
756	(modulo threads).	1326	(modulo threads).
757		1327
758	=item ev_set_priority (ev_TYPE *watcher, priority)	1328	=item ev_set_priority (ev_TYPE *watcher, int priority)
759		1329
760	=item int ev_priority (ev_TYPE *watcher)	1330	=item int ev_priority (ev_TYPE *watcher)
761		1331
762	Set and query the priority of the watcher. The priority is a small	1332	Set and query the priority of the watcher. The priority is a small
763	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>	1333	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
764	(default: C<-2>). Pending watchers with higher priority will be invoked	1334	(default: C<-2>). Pending watchers with higher priority will be invoked
765	before watchers with lower priority, but priority will not keep watchers	1335	before watchers with lower priority, but priority will not keep watchers
766	from being executed (except for C<ev_idle> watchers).	1336	from being executed (except for C<ev_idle> watchers).
767		1337
768	This means that priorities are I<only> used for ordering callback
769	invocation after new events have been received. This is useful, for
770	example, to reduce latency after idling, or more often, to bind two
771	watchers on the same event and make sure one is called first.
772
773	If you need to suppress invocation when higher priority events are pending	1338	If you need to suppress invocation when higher priority events are pending
774	you need to look at C<ev_idle> watchers, which provide this functionality.	1339	you need to look at C<ev_idle> watchers, which provide this functionality.
775		1340
776	You I<must not> change the priority of a watcher as long as it is active or	1341	You I<must not> change the priority of a watcher as long as it is active or
777	pending.	1342	pending.
778		1343
		1344	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1345	fine, as long as you do not mind that the priority value you query might
		1346	or might not have been clamped to the valid range.
		1347
779	The default priority used by watchers when no priority has been set is	1348	The default priority used by watchers when no priority has been set is
780	always C<0>, which is supposed to not be too high and not be too low :).	1349	always C<0>, which is supposed to not be too high and not be too low :).
781		1350
782	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is	1351	See L</WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
783	fine, as long as you do not mind that the priority value you query might	1352	priorities.
784	or might not have been adjusted to be within valid range.
785		1353
786	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1354	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
787		1355
788	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1356	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
789	C<loop> nor C<revents> need to be valid as long as the watcher callback	1357	C<loop> nor C<revents> need to be valid as long as the watcher callback
790	can deal with that fact.	1358	can deal with that fact, as both are simply passed through to the
		1359	callback.
791		1360
792	=item int ev_clear_pending (loop, ev_TYPE *watcher)	1361	=item int ev_clear_pending (loop, ev_TYPE *watcher)
793		1362
794	If the watcher is pending, this function returns clears its pending status	1363	If the watcher is pending, this function clears its pending status and
795	and returns its C<revents> bitset (as if its callback was invoked). If the	1364	returns its C<revents> bitset (as if its callback was invoked). If the
796	watcher isn't pending it does nothing and returns C<0>.	1365	watcher isn't pending it does nothing and returns C<0>.
797		1366
		1367	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1368	callback to be invoked, which can be accomplished with this function.
		1369
		1370	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1371
		1372	Feeds the given event set into the event loop, as if the specified event
		1373	had happened for the specified watcher (which must be a pointer to an
		1374	initialised but not necessarily started event watcher). Obviously you must
		1375	not free the watcher as long as it has pending events.
		1376
		1377	Stopping the watcher, letting libev invoke it, or calling
		1378	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1379	not started in the first place.
		1380
		1381	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1382	functions that do not need a watcher.
		1383
798	=back	1384	=back
799		1385
		1386	See also the L</ASSOCIATING CUSTOM DATA WITH A WATCHER> and L</BUILDING YOUR
		1387	OWN COMPOSITE WATCHERS> idioms.
800		1388
801	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1389	=head2 WATCHER STATES
802		1390
803	Each watcher has, by default, a member C<void *data> that you can change	1391	There are various watcher states mentioned throughout this manual -
804	and read at any time, libev will completely ignore it. This can be used	1392	active, pending and so on. In this section these states and the rules to
805	to associate arbitrary data with your watcher. If you need more data and	1393	transition between them will be described in more detail - and while these
806	don't want to allocate memory and store a pointer to it in that data	1394	rules might look complicated, they usually do "the right thing".
807	member, you can also "subclass" the watcher type and provide your own
808	data:
809		1395
810	struct my_io	1396	=over 4
811	{
812	struct ev_io io;
813	int otherfd;
814	void *somedata;
815	struct whatever *mostinteresting;
816	}
817		1397
818	And since your callback will be called with a pointer to the watcher, you	1398	=item initialised
819	can cast it back to your own type:
820		1399
821	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1400	Before a watcher can be registered with the event loop it has to be
822	{	1401	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
823	struct my_io *w = (struct my_io *)w_;	1402	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
824	...
825	}
826		1403
827	More interesting and less C-conformant ways of casting your callback type	1404	In this state it is simply some block of memory that is suitable for
828	instead have been omitted.	1405	use in an event loop. It can be moved around, freed, reused etc. at
		1406	will - as long as you either keep the memory contents intact, or call
		1407	C<ev_TYPE_init> again.
829		1408
830	Another common scenario is having some data structure with multiple	1409	=item started/running/active
831	watchers:
832		1410
833	struct my_biggy	1411	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
834	{	1412	property of the event loop, and is actively waiting for events. While in
835	int some_data;	1413	this state it cannot be accessed (except in a few documented ways), moved,
836	ev_timer t1;	1414	freed or anything else - the only legal thing is to keep a pointer to it,
837	ev_timer t2;	1415	and call libev functions on it that are documented to work on active watchers.
838	}
839		1416
840	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1417	=item pending
841	you need to use C<offsetof>:
842		1418
843	#include <stddef.h>	1419	If a watcher is active and libev determines that an event it is interested
		1420	in has occurred (such as a timer expiring), it will become pending. It will
		1421	stay in this pending state until either it is stopped or its callback is
		1422	about to be invoked, so it is not normally pending inside the watcher
		1423	callback.
844		1424
		1425	The watcher might or might not be active while it is pending (for example,
		1426	an expired non-repeating timer can be pending but no longer active). If it
		1427	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1428	but it is still property of the event loop at this time, so cannot be
		1429	moved, freed or reused. And if it is active the rules described in the
		1430	previous item still apply.
		1431
		1432	It is also possible to feed an event on a watcher that is not active (e.g.
		1433	via C<ev_feed_event>), in which case it becomes pending without being
		1434	active.
		1435
		1436	=item stopped
		1437
		1438	A watcher can be stopped implicitly by libev (in which case it might still
		1439	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1440	latter will clear any pending state the watcher might be in, regardless
		1441	of whether it was active or not, so stopping a watcher explicitly before
		1442	freeing it is often a good idea.
		1443
		1444	While stopped (and not pending) the watcher is essentially in the
		1445	initialised state, that is, it can be reused, moved, modified in any way
		1446	you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
		1447	it again).
		1448
		1449	=back
		1450
		1451	=head2 WATCHER PRIORITY MODELS
		1452
		1453	Many event loops support I<watcher priorities>, which are usually small
		1454	integers that influence the ordering of event callback invocation
		1455	between watchers in some way, all else being equal.
		1456
		1457	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1458	description for the more technical details such as the actual priority
		1459	range.
		1460
		1461	There are two common ways how these these priorities are being interpreted
		1462	by event loops:
		1463
		1464	In the more common lock-out model, higher priorities "lock out" invocation
		1465	of lower priority watchers, which means as long as higher priority
		1466	watchers receive events, lower priority watchers are not being invoked.
		1467
		1468	The less common only-for-ordering model uses priorities solely to order
		1469	callback invocation within a single event loop iteration: Higher priority
		1470	watchers are invoked before lower priority ones, but they all get invoked
		1471	before polling for new events.
		1472
		1473	Libev uses the second (only-for-ordering) model for all its watchers
		1474	except for idle watchers (which use the lock-out model).
		1475
		1476	The rationale behind this is that implementing the lock-out model for
		1477	watchers is not well supported by most kernel interfaces, and most event
		1478	libraries will just poll for the same events again and again as long as
		1479	their callbacks have not been executed, which is very inefficient in the
		1480	common case of one high-priority watcher locking out a mass of lower
		1481	priority ones.
		1482
		1483	Static (ordering) priorities are most useful when you have two or more
		1484	watchers handling the same resource: a typical usage example is having an
		1485	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1486	timeouts. Under load, data might be received while the program handles
		1487	other jobs, but since timers normally get invoked first, the timeout
		1488	handler will be executed before checking for data. In that case, giving
		1489	the timer a lower priority than the I/O watcher ensures that I/O will be
		1490	handled first even under adverse conditions (which is usually, but not
		1491	always, what you want).
		1492
		1493	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1494	will only be executed when no same or higher priority watchers have
		1495	received events, they can be used to implement the "lock-out" model when
		1496	required.
		1497
		1498	For example, to emulate how many other event libraries handle priorities,
		1499	you can associate an C<ev_idle> watcher to each such watcher, and in
		1500	the normal watcher callback, you just start the idle watcher. The real
		1501	processing is done in the idle watcher callback. This causes libev to
		1502	continuously poll and process kernel event data for the watcher, but when
		1503	the lock-out case is known to be rare (which in turn is rare :), this is
		1504	workable.
		1505
		1506	Usually, however, the lock-out model implemented that way will perform
		1507	miserably under the type of load it was designed to handle. In that case,
		1508	it might be preferable to stop the real watcher before starting the
		1509	idle watcher, so the kernel will not have to process the event in case
		1510	the actual processing will be delayed for considerable time.
		1511
		1512	Here is an example of an I/O watcher that should run at a strictly lower
		1513	priority than the default, and which should only process data when no
		1514	other events are pending:
		1515
		1516	ev_idle idle; // actual processing watcher
		1517	ev_io io; // actual event watcher
		1518
845	static void	1519	static void
846	t1_cb (EV_P_ struct ev_timer *w, int revents)	1520	io_cb (EV_P_ ev_io *w, int revents)
847	{	1521	{
848	struct my_biggy big = (struct my_biggy *	1522	// stop the I/O watcher, we received the event, but
849	(((char *)w) - offsetof (struct my_biggy, t1));	1523	// are not yet ready to handle it.
		1524	ev_io_stop (EV_A_ w);
		1525
		1526	// start the idle watcher to handle the actual event.
		1527	// it will not be executed as long as other watchers
		1528	// with the default priority are receiving events.
		1529	ev_idle_start (EV_A_ &idle);
850	}	1530	}
851		1531
852	static void	1532	static void
853	t2_cb (EV_P_ struct ev_timer *w, int revents)	1533	idle_cb (EV_P_ ev_idle *w, int revents)
854	{	1534	{
855	struct my_biggy big = (struct my_biggy *	1535	// actual processing
856	(((char *)w) - offsetof (struct my_biggy, t2));	1536	read (STDIN_FILENO, ...);
		1537
		1538	// have to start the I/O watcher again, as
		1539	// we have handled the event
		1540	ev_io_start (EV_P_ &io);
857	}	1541	}
		1542
		1543	// initialisation
		1544	ev_idle_init (&idle, idle_cb);
		1545	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1546	ev_io_start (EV_DEFAULT_ &io);
		1547
		1548	In the "real" world, it might also be beneficial to start a timer, so that
		1549	low-priority connections can not be locked out forever under load. This
		1550	enables your program to keep a lower latency for important connections
		1551	during short periods of high load, while not completely locking out less
		1552	important ones.
858		1553
859		1554
860	=head1 WATCHER TYPES	1555	=head1 WATCHER TYPES
861		1556
862	This section describes each watcher in detail, but will not repeat	1557	This section describes each watcher in detail, but will not repeat
…		…
886	In general you can register as many read and/or write event watchers per	1581	In general you can register as many read and/or write event watchers per
887	fd as you want (as long as you don't confuse yourself). Setting all file	1582	fd as you want (as long as you don't confuse yourself). Setting all file
888	descriptors to non-blocking mode is also usually a good idea (but not	1583	descriptors to non-blocking mode is also usually a good idea (but not
889	required if you know what you are doing).	1584	required if you know what you are doing).
890		1585
891	You have to be careful with dup'ed file descriptors, though. Some backends
892	(the linux epoll backend is a notable example) cannot handle dup'ed file
893	descriptors correctly if you register interest in two or more fds pointing
894	to the same underlying file/socket/etc. description (that is, they share
895	the same underlying "file open").
896
897	If you must do this, then force the use of a known-to-be-good backend
898	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
899	C<EVBACKEND_POLL>).
900
901	Another thing you have to watch out for is that it is quite easy to	1586	Another thing you have to watch out for is that it is quite easy to
902	receive "spurious" readyness notifications, that is your callback might	1587	receive "spurious" readiness notifications, that is, your callback might
903	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1588	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
904	because there is no data. Not only are some backends known to create a	1589	because there is no data. It is very easy to get into this situation even
905	lot of those (for example solaris ports), it is very easy to get into	1590	with a relatively standard program structure. Thus it is best to always
906	this situation even with a relatively standard program structure. Thus	1591	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
907	it is best to always use non-blocking I/O: An extra C<read>(2) returning
908	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1592	preferable to a program hanging until some data arrives.
909		1593
910	If you cannot run the fd in non-blocking mode (for example you should not	1594	If you cannot run the fd in non-blocking mode (for example you should
911	play around with an Xlib connection), then you have to seperately re-test	1595	not play around with an Xlib connection), then you have to separately
912	whether a file descriptor is really ready with a known-to-be good interface	1596	re-test whether a file descriptor is really ready with a known-to-be good
913	such as poll (fortunately in our Xlib example, Xlib already does this on	1597	interface such as poll (fortunately in the case of Xlib, it already does
914	its own, so its quite safe to use).	1598	this on its own, so its quite safe to use). Some people additionally
		1599	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1600	indefinitely.
		1601
		1602	But really, best use non-blocking mode.
915		1603
916	=head3 The special problem of disappearing file descriptors	1604	=head3 The special problem of disappearing file descriptors
917		1605
918	Some backends (e.g kqueue, epoll) need to be told about closing a file	1606	Some backends (e.g. kqueue, epoll) need to be told about closing a file
919	descriptor (either by calling C<close> explicitly or by any other means,	1607	descriptor (either due to calling C<close> explicitly or any other means,
920	such as C<dup>). The reason is that you register interest in some file	1608	such as C<dup2>). The reason is that you register interest in some file
921	descriptor, but when it goes away, the operating system will silently drop	1609	descriptor, but when it goes away, the operating system will silently drop
922	this interest. If another file descriptor with the same number then is	1610	this interest. If another file descriptor with the same number then is
923	registered with libev, there is no efficient way to see that this is, in	1611	registered with libev, there is no efficient way to see that this is, in
924	fact, a different file descriptor.	1612	fact, a different file descriptor.
925		1613
…		…
932		1620
933	This is how one would do it normally anyway, the important point is that	1621	This is how one would do it normally anyway, the important point is that
934	the libev application should not optimise around libev but should leave	1622	the libev application should not optimise around libev but should leave
935	optimisations to libev.	1623	optimisations to libev.
936		1624
		1625	=head3 The special problem of dup'ed file descriptors
		1626
		1627	Some backends (e.g. epoll), cannot register events for file descriptors,
		1628	but only events for the underlying file descriptions. That means when you
		1629	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1630	events for them, only one file descriptor might actually receive events.
		1631
		1632	There is no workaround possible except not registering events
		1633	for potentially C<dup ()>'ed file descriptors, or to resort to
		1634	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1635
		1636	=head3 The special problem of files
		1637
		1638	Many people try to use C<select> (or libev) on file descriptors
		1639	representing files, and expect it to become ready when their program
		1640	doesn't block on disk accesses (which can take a long time on their own).
		1641
		1642	However, this cannot ever work in the "expected" way - you get a readiness
		1643	notification as soon as the kernel knows whether and how much data is
		1644	there, and in the case of open files, that's always the case, so you
		1645	always get a readiness notification instantly, and your read (or possibly
		1646	write) will still block on the disk I/O.
		1647
		1648	Another way to view it is that in the case of sockets, pipes, character
		1649	devices and so on, there is another party (the sender) that delivers data
		1650	on its own, but in the case of files, there is no such thing: the disk
		1651	will not send data on its own, simply because it doesn't know what you
		1652	wish to read - you would first have to request some data.
		1653
		1654	Since files are typically not-so-well supported by advanced notification
		1655	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1656	to files, even though you should not use it. The reason for this is
		1657	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1658	usually a tty, often a pipe, but also sometimes files or special devices
		1659	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1660	F</dev/urandom>), and even though the file might better be served with
		1661	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1662	it "just works" instead of freezing.
		1663
		1664	So avoid file descriptors pointing to files when you know it (e.g. use
		1665	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1666	when you rarely read from a file instead of from a socket, and want to
		1667	reuse the same code path.
		1668
		1669	=head3 The special problem of fork
		1670
		1671	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1672	useless behaviour. Libev fully supports fork, but needs to be told about
		1673	it in the child if you want to continue to use it in the child.
		1674
		1675	To support fork in your child processes, you have to call C<ev_loop_fork
		1676	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
		1677	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1678
		1679	=head3 The special problem of SIGPIPE
		1680
		1681	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1682	when writing to a pipe whose other end has been closed, your program gets
		1683	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1684	this is sensible behaviour, for daemons, this is usually undesirable.
		1685
		1686	So when you encounter spurious, unexplained daemon exits, make sure you
		1687	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1688	somewhere, as that would have given you a big clue).
		1689
		1690	=head3 The special problem of accept()ing when you can't
		1691
		1692	Many implementations of the POSIX C<accept> function (for example,
		1693	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1694	connection from the pending queue in all error cases.
		1695
		1696	For example, larger servers often run out of file descriptors (because
		1697	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1698	rejecting the connection, leading to libev signalling readiness on
		1699	the next iteration again (the connection still exists after all), and
		1700	typically causing the program to loop at 100% CPU usage.
		1701
		1702	Unfortunately, the set of errors that cause this issue differs between
		1703	operating systems, there is usually little the app can do to remedy the
		1704	situation, and no known thread-safe method of removing the connection to
		1705	cope with overload is known (to me).
		1706
		1707	One of the easiest ways to handle this situation is to just ignore it
		1708	- when the program encounters an overload, it will just loop until the
		1709	situation is over. While this is a form of busy waiting, no OS offers an
		1710	event-based way to handle this situation, so it's the best one can do.
		1711
		1712	A better way to handle the situation is to log any errors other than
		1713	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1714	messages, and continue as usual, which at least gives the user an idea of
		1715	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1716	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1717	usage.
		1718
		1719	If your program is single-threaded, then you could also keep a dummy file
		1720	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1721	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1722	close that fd, and create a new dummy fd. This will gracefully refuse
		1723	clients under typical overload conditions.
		1724
		1725	The last way to handle it is to simply log the error and C<exit>, as
		1726	is often done with C<malloc> failures, but this results in an easy
		1727	opportunity for a DoS attack.
937		1728
938	=head3 Watcher-Specific Functions	1729	=head3 Watcher-Specific Functions
939		1730
940	=over 4	1731	=over 4
941		1732
942	=item ev_io_init (ev_io *, callback, int fd, int events)	1733	=item ev_io_init (ev_io *, callback, int fd, int events)
943		1734
944	=item ev_io_set (ev_io *, int fd, int events)	1735	=item ev_io_set (ev_io *, int fd, int events)
945		1736
946	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1737	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
947	rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or	1738	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
948	C<EV_READ | EV_WRITE> to receive the given events.	1739	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
949		1740
950	=item int fd [read-only]	1741	=item int fd [read-only]
951		1742
952	The file descriptor being watched.	1743	The file descriptor being watched.
953		1744
954	=item int events [read-only]	1745	=item int events [read-only]
955		1746
956	The events being watched.	1747	The events being watched.
957		1748
958	=back	1749	=back
		1750
		1751	=head3 Examples
959		1752
960	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1753	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
961	readable, but only once. Since it is likely line-buffered, you could	1754	readable, but only once. Since it is likely line-buffered, you could
962	attempt to read a whole line in the callback.	1755	attempt to read a whole line in the callback.
963		1756
964	static void	1757	static void
965	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1758	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
966	{	1759	{
967	ev_io_stop (loop, w);	1760	ev_io_stop (loop, w);
968	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1761	.. read from stdin here (or from w->fd) and handle any I/O errors
969	}	1762	}
970		1763
971	...	1764	...
972	struct ev_loop *loop = ev_default_init (0);	1765	struct ev_loop *loop = ev_default_init (0);
973	struct ev_io stdin_readable;	1766	ev_io stdin_readable;
974	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1767	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
975	ev_io_start (loop, &stdin_readable);	1768	ev_io_start (loop, &stdin_readable);
976	ev_loop (loop, 0);	1769	ev_run (loop, 0);
977		1770
978		1771
979	=head2 C<ev_timer> - relative and optionally repeating timeouts	1772	=head2 C<ev_timer> - relative and optionally repeating timeouts
980		1773
981	Timer watchers are simple relative timers that generate an event after a	1774	Timer watchers are simple relative timers that generate an event after a
982	given time, and optionally repeating in regular intervals after that.	1775	given time, and optionally repeating in regular intervals after that.
983		1776
984	The timers are based on real time, that is, if you register an event that	1777	The timers are based on real time, that is, if you register an event that
985	times out after an hour and you reset your system clock to last years	1778	times out after an hour and you reset your system clock to January last
986	time, it will still time out after (roughly) and hour. "Roughly" because	1779	year, it will still time out after (roughly) one hour. "Roughly" because
987	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1780	detecting time jumps is hard, and some inaccuracies are unavoidable (the
988	monotonic clock option helps a lot here).	1781	monotonic clock option helps a lot here).
		1782
		1783	The callback is guaranteed to be invoked only I<after> its timeout has
		1784	passed (not I<at>, so on systems with very low-resolution clocks this
		1785	might introduce a small delay, see "the special problem of being too
		1786	early", below). If multiple timers become ready during the same loop
		1787	iteration then the ones with earlier time-out values are invoked before
		1788	ones of the same priority with later time-out values (but this is no
		1789	longer true when a callback calls C<ev_run> recursively).
		1790
		1791	=head3 Be smart about timeouts
		1792
		1793	Many real-world problems involve some kind of timeout, usually for error
		1794	recovery. A typical example is an HTTP request - if the other side hangs,
		1795	you want to raise some error after a while.
		1796
		1797	What follows are some ways to handle this problem, from obvious and
		1798	inefficient to smart and efficient.
		1799
		1800	In the following, a 60 second activity timeout is assumed - a timeout that
		1801	gets reset to 60 seconds each time there is activity (e.g. each time some
		1802	data or other life sign was received).
		1803
		1804	=over 4
		1805
		1806	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1807
		1808	This is the most obvious, but not the most simple way: In the beginning,
		1809	start the watcher:
		1810
		1811	ev_timer_init (timer, callback, 60., 0.);
		1812	ev_timer_start (loop, timer);
		1813
		1814	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1815	and start it again:
		1816
		1817	ev_timer_stop (loop, timer);
		1818	ev_timer_set (timer, 60., 0.);
		1819	ev_timer_start (loop, timer);
		1820
		1821	This is relatively simple to implement, but means that each time there is
		1822	some activity, libev will first have to remove the timer from its internal
		1823	data structure and then add it again. Libev tries to be fast, but it's
		1824	still not a constant-time operation.
		1825
		1826	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1827
		1828	This is the easiest way, and involves using C<ev_timer_again> instead of
		1829	C<ev_timer_start>.
		1830
		1831	To implement this, configure an C<ev_timer> with a C<repeat> value
		1832	of C<60> and then call C<ev_timer_again> at start and each time you
		1833	successfully read or write some data. If you go into an idle state where
		1834	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1835	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1836
		1837	That means you can ignore both the C<ev_timer_start> function and the
		1838	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1839	member and C<ev_timer_again>.
		1840
		1841	At start:
		1842
		1843	ev_init (timer, callback);
		1844	timer->repeat = 60.;
		1845	ev_timer_again (loop, timer);
		1846
		1847	Each time there is some activity:
		1848
		1849	ev_timer_again (loop, timer);
		1850
		1851	It is even possible to change the time-out on the fly, regardless of
		1852	whether the watcher is active or not:
		1853
		1854	timer->repeat = 30.;
		1855	ev_timer_again (loop, timer);
		1856
		1857	This is slightly more efficient then stopping/starting the timer each time
		1858	you want to modify its timeout value, as libev does not have to completely
		1859	remove and re-insert the timer from/into its internal data structure.
		1860
		1861	It is, however, even simpler than the "obvious" way to do it.
		1862
		1863	=item 3. Let the timer time out, but then re-arm it as required.
		1864
		1865	This method is more tricky, but usually most efficient: Most timeouts are
		1866	relatively long compared to the intervals between other activity - in
		1867	our example, within 60 seconds, there are usually many I/O events with
		1868	associated activity resets.
		1869
		1870	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1871	but remember the time of last activity, and check for a real timeout only
		1872	within the callback:
		1873
		1874	ev_tstamp timeout = 60.;
		1875	ev_tstamp last_activity; // time of last activity
		1876	ev_timer timer;
		1877
		1878	static void
		1879	callback (EV_P_ ev_timer *w, int revents)
		1880	{
		1881	// calculate when the timeout would happen
		1882	ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
		1883
		1884	// if negative, it means we the timeout already occurred
		1885	if (after < 0.)
		1886	{
		1887	// timeout occurred, take action
		1888	}
		1889	else
		1890	{
		1891	// callback was invoked, but there was some recent
		1892	// activity. simply restart the timer to time out
		1893	// after "after" seconds, which is the earliest time
		1894	// the timeout can occur.
		1895	ev_timer_set (w, after, 0.);
		1896	ev_timer_start (EV_A_ w);
		1897	}
		1898	}
		1899
		1900	To summarise the callback: first calculate in how many seconds the
		1901	timeout will occur (by calculating the absolute time when it would occur,
		1902	C<last_activity + timeout>, and subtracting the current time, C<ev_now
		1903	(EV_A)> from that).
		1904
		1905	If this value is negative, then we are already past the timeout, i.e. we
		1906	timed out, and need to do whatever is needed in this case.
		1907
		1908	Otherwise, we now the earliest time at which the timeout would trigger,
		1909	and simply start the timer with this timeout value.
		1910
		1911	In other words, each time the callback is invoked it will check whether
		1912	the timeout occurred. If not, it will simply reschedule itself to check
		1913	again at the earliest time it could time out. Rinse. Repeat.
		1914
		1915	This scheme causes more callback invocations (about one every 60 seconds
		1916	minus half the average time between activity), but virtually no calls to
		1917	libev to change the timeout.
		1918
		1919	To start the machinery, simply initialise the watcher and set
		1920	C<last_activity> to the current time (meaning there was some activity just
		1921	now), then call the callback, which will "do the right thing" and start
		1922	the timer:
		1923
		1924	last_activity = ev_now (EV_A);
		1925	ev_init (&timer, callback);
		1926	callback (EV_A_ &timer, 0);
		1927
		1928	When there is some activity, simply store the current time in
		1929	C<last_activity>, no libev calls at all:
		1930
		1931	if (activity detected)
		1932	last_activity = ev_now (EV_A);
		1933
		1934	When your timeout value changes, then the timeout can be changed by simply
		1935	providing a new value, stopping the timer and calling the callback, which
		1936	will again do the right thing (for example, time out immediately :).
		1937
		1938	timeout = new_value;
		1939	ev_timer_stop (EV_A_ &timer);
		1940	callback (EV_A_ &timer, 0);
		1941
		1942	This technique is slightly more complex, but in most cases where the
		1943	time-out is unlikely to be triggered, much more efficient.
		1944
		1945	=item 4. Wee, just use a double-linked list for your timeouts.
		1946
		1947	If there is not one request, but many thousands (millions...), all
		1948	employing some kind of timeout with the same timeout value, then one can
		1949	do even better:
		1950
		1951	When starting the timeout, calculate the timeout value and put the timeout
		1952	at the I<end> of the list.
		1953
		1954	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1955	the list is expected to fire (for example, using the technique #3).
		1956
		1957	When there is some activity, remove the timer from the list, recalculate
		1958	the timeout, append it to the end of the list again, and make sure to
		1959	update the C<ev_timer> if it was taken from the beginning of the list.
		1960
		1961	This way, one can manage an unlimited number of timeouts in O(1) time for
		1962	starting, stopping and updating the timers, at the expense of a major
		1963	complication, and having to use a constant timeout. The constant timeout
		1964	ensures that the list stays sorted.
		1965
		1966	=back
		1967
		1968	So which method the best?
		1969
		1970	Method #2 is a simple no-brain-required solution that is adequate in most
		1971	situations. Method #3 requires a bit more thinking, but handles many cases
		1972	better, and isn't very complicated either. In most case, choosing either
		1973	one is fine, with #3 being better in typical situations.
		1974
		1975	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1976	rather complicated, but extremely efficient, something that really pays
		1977	off after the first million or so of active timers, i.e. it's usually
		1978	overkill :)
		1979
		1980	=head3 The special problem of being too early
		1981
		1982	If you ask a timer to call your callback after three seconds, then
		1983	you expect it to be invoked after three seconds - but of course, this
		1984	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		1985	guaranteed to any precision by libev - imagine somebody suspending the
		1986	process with a STOP signal for a few hours for example.
		1987
		1988	So, libev tries to invoke your callback as soon as possible I<after> the
		1989	delay has occurred, but cannot guarantee this.
		1990
		1991	A less obvious failure mode is calling your callback too early: many event
		1992	loops compare timestamps with a "elapsed delay >= requested delay", but
		1993	this can cause your callback to be invoked much earlier than you would
		1994	expect.
		1995
		1996	To see why, imagine a system with a clock that only offers full second
		1997	resolution (think windows if you can't come up with a broken enough OS
		1998	yourself). If you schedule a one-second timer at the time 500.9, then the
		1999	event loop will schedule your timeout to elapse at a system time of 500
		2000	(500.9 truncated to the resolution) + 1, or 501.
		2001
		2002	If an event library looks at the timeout 0.1s later, it will see "501 >=
		2003	501" and invoke the callback 0.1s after it was started, even though a
		2004	one-second delay was requested - this is being "too early", despite best
		2005	intentions.
		2006
		2007	This is the reason why libev will never invoke the callback if the elapsed
		2008	delay equals the requested delay, but only when the elapsed delay is
		2009	larger than the requested delay. In the example above, libev would only invoke
		2010	the callback at system time 502, or 1.1s after the timer was started.
		2011
		2012	So, while libev cannot guarantee that your callback will be invoked
		2013	exactly when requested, it I<can> and I<does> guarantee that the requested
		2014	delay has actually elapsed, or in other words, it always errs on the "too
		2015	late" side of things.
		2016
		2017	=head3 The special problem of time updates
		2018
		2019	Establishing the current time is a costly operation (it usually takes
		2020	at least one system call): EV therefore updates its idea of the current
		2021	time only before and after C<ev_run> collects new events, which causes a
		2022	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		2023	lots of events in one iteration.
989		2024
990	The relative timeouts are calculated relative to the C<ev_now ()>	2025	The relative timeouts are calculated relative to the C<ev_now ()>
991	time. This is usually the right thing as this timestamp refers to the time	2026	time. This is usually the right thing as this timestamp refers to the time
992	of the event triggering whatever timeout you are modifying/starting. If	2027	of the event triggering whatever timeout you are modifying/starting. If
993	you suspect event processing to be delayed and you I<need> to base the timeout	2028	you suspect event processing to be delayed and you I<need> to base the
994	on the current time, use something like this to adjust for this:	2029	timeout on the current time, use something like this to adjust for this:
995		2030
996	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2031	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
997		2032
998	The callback is guarenteed to be invoked only when its timeout has passed,	2033	If the event loop is suspended for a long time, you can also force an
999	but if multiple timers become ready during the same loop iteration then	2034	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1000	order of execution is undefined.	2035	()>.
		2036
		2037	=head3 The special problem of unsynchronised clocks
		2038
		2039	Modern systems have a variety of clocks - libev itself uses the normal
		2040	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2041	jumps).
		2042
		2043	Neither of these clocks is synchronised with each other or any other clock
		2044	on the system, so C<ev_time ()> might return a considerably different time
		2045	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2046	a call to C<gettimeofday> might return a second count that is one higher
		2047	than a directly following call to C<time>.
		2048
		2049	The moral of this is to only compare libev-related timestamps with
		2050	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2051	a second or so.
		2052
		2053	One more problem arises due to this lack of synchronisation: if libev uses
		2054	the system monotonic clock and you compare timestamps from C<ev_time>
		2055	or C<ev_now> from when you started your timer and when your callback is
		2056	invoked, you will find that sometimes the callback is a bit "early".
		2057
		2058	This is because C<ev_timer>s work in real time, not wall clock time, so
		2059	libev makes sure your callback is not invoked before the delay happened,
		2060	I<measured according to the real time>, not the system clock.
		2061
		2062	If your timeouts are based on a physical timescale (e.g. "time out this
		2063	connection after 100 seconds") then this shouldn't bother you as it is
		2064	exactly the right behaviour.
		2065
		2066	If you want to compare wall clock/system timestamps to your timers, then
		2067	you need to use C<ev_periodic>s, as these are based on the wall clock
		2068	time, where your comparisons will always generate correct results.
		2069
		2070	=head3 The special problems of suspended animation
		2071
		2072	When you leave the server world it is quite customary to hit machines that
		2073	can suspend/hibernate - what happens to the clocks during such a suspend?
		2074
		2075	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		2076	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		2077	to run until the system is suspended, but they will not advance while the
		2078	system is suspended. That means, on resume, it will be as if the program
		2079	was frozen for a few seconds, but the suspend time will not be counted
		2080	towards C<ev_timer> when a monotonic clock source is used. The real time
		2081	clock advanced as expected, but if it is used as sole clocksource, then a
		2082	long suspend would be detected as a time jump by libev, and timers would
		2083	be adjusted accordingly.
		2084
		2085	I would not be surprised to see different behaviour in different between
		2086	operating systems, OS versions or even different hardware.
		2087
		2088	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2089	time jump in the monotonic clocks and the realtime clock. If the program
		2090	is suspended for a very long time, and monotonic clock sources are in use,
		2091	then you can expect C<ev_timer>s to expire as the full suspension time
		2092	will be counted towards the timers. When no monotonic clock source is in
		2093	use, then libev will again assume a timejump and adjust accordingly.
		2094
		2095	It might be beneficial for this latter case to call C<ev_suspend>
		2096	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2097	deterministic behaviour in this case (you can do nothing against
		2098	C<SIGSTOP>).
1001		2099
1002	=head3 Watcher-Specific Functions and Data Members	2100	=head3 Watcher-Specific Functions and Data Members
1003		2101
1004	=over 4	2102	=over 4
1005		2103
1006	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2104	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
1007		2105
1008	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2106	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
1009		2107
1010	Configure the timer to trigger after C<after> seconds. If C<repeat> is	2108	Configure the timer to trigger after C<after> seconds. If C<repeat>
1011	C<0.>, then it will automatically be stopped. If it is positive, then the	2109	is C<0.>, then it will automatically be stopped once the timeout is
1012	timer will automatically be configured to trigger again C<repeat> seconds	2110	reached. If it is positive, then the timer will automatically be
1013	later, again, and again, until stopped manually.	2111	configured to trigger again C<repeat> seconds later, again, and again,
		2112	until stopped manually.
1014		2113
1015	The timer itself will do a best-effort at avoiding drift, that is, if you	2114	The timer itself will do a best-effort at avoiding drift, that is, if
1016	configure a timer to trigger every 10 seconds, then it will trigger at	2115	you configure a timer to trigger every 10 seconds, then it will normally
1017	exactly 10 second intervals. If, however, your program cannot keep up with	2116	trigger at exactly 10 second intervals. If, however, your program cannot
1018	the timer (because it takes longer than those 10 seconds to do stuff) the	2117	keep up with the timer (because it takes longer than those 10 seconds to
1019	timer will not fire more than once per event loop iteration.	2118	do stuff) the timer will not fire more than once per event loop iteration.
1020		2119
1021	=item ev_timer_again (loop)	2120	=item ev_timer_again (loop, ev_timer *)
1022		2121
1023	This will act as if the timer timed out and restart it again if it is	2122	This will act as if the timer timed out, and restarts it again if it is
1024	repeating. The exact semantics are:	2123	repeating. It basically works like calling C<ev_timer_stop>, updating the
		2124	timeout to the C<repeat> value and calling C<ev_timer_start>.
1025		2125
		2126	The exact semantics are as in the following rules, all of which will be
		2127	applied to the watcher:
		2128
		2129	=over 4
		2130
1026	If the timer is pending, its pending status is cleared.	2131	=item If the timer is pending, the pending status is always cleared.
1027		2132
1028	If the timer is started but nonrepeating, stop it (as if it timed out).	2133	=item If the timer is started but non-repeating, stop it (as if it timed
		2134	out, without invoking it).
1029		2135
1030	If the timer is repeating, either start it if necessary (with the	2136	=item If the timer is repeating, make the C<repeat> value the new timeout
1031	C<repeat> value), or reset the running timer to the C<repeat> value.	2137	and start the timer, if necessary.
1032		2138
1033	This sounds a bit complicated, but here is a useful and typical	2139	=back
1034	example: Imagine you have a tcp connection and you want a so-called idle
1035	timeout, that is, you want to be called when there have been, say, 60
1036	seconds of inactivity on the socket. The easiest way to do this is to
1037	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
1038	C<ev_timer_again> each time you successfully read or write some data. If
1039	you go into an idle state where you do not expect data to travel on the
1040	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
1041	automatically restart it if need be.
1042		2140
1043	That means you can ignore the C<after> value and C<ev_timer_start>	2141	This sounds a bit complicated, see L</Be smart about timeouts>, above, for a
1044	altogether and only ever use the C<repeat> value and C<ev_timer_again>:	2142	usage example.
1045		2143
1046	ev_timer_init (timer, callback, 0., 5.);	2144	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
1047	ev_timer_again (loop, timer);
1048	...
1049	timer->again = 17.;
1050	ev_timer_again (loop, timer);
1051	...
1052	timer->again = 10.;
1053	ev_timer_again (loop, timer);
1054		2145
1055	This is more slightly efficient then stopping/starting the timer each time	2146	Returns the remaining time until a timer fires. If the timer is active,
1056	you want to modify its timeout value.	2147	then this time is relative to the current event loop time, otherwise it's
		2148	the timeout value currently configured.
		2149
		2150	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2151	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2152	will return C<4>. When the timer expires and is restarted, it will return
		2153	roughly C<7> (likely slightly less as callback invocation takes some time,
		2154	too), and so on.
1057		2155
1058	=item ev_tstamp repeat [read-write]	2156	=item ev_tstamp repeat [read-write]
1059		2157
1060	The current C<repeat> value. Will be used each time the watcher times out	2158	The current C<repeat> value. Will be used each time the watcher times out
1061	or C<ev_timer_again> is called and determines the next timeout (if any),	2159	or C<ev_timer_again> is called, and determines the next timeout (if any),
1062	which is also when any modifications are taken into account.	2160	which is also when any modifications are taken into account.
1063		2161
1064	=back	2162	=back
1065		2163
		2164	=head3 Examples
		2165
1066	Example: Create a timer that fires after 60 seconds.	2166	Example: Create a timer that fires after 60 seconds.
1067		2167
1068	static void	2168	static void
1069	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2169	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1070	{	2170	{
1071	.. one minute over, w is actually stopped right here	2171	.. one minute over, w is actually stopped right here
1072	}	2172	}
1073		2173
1074	struct ev_timer mytimer;	2174	ev_timer mytimer;
1075	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	2175	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1076	ev_timer_start (loop, &mytimer);	2176	ev_timer_start (loop, &mytimer);
1077		2177
1078	Example: Create a timeout timer that times out after 10 seconds of	2178	Example: Create a timeout timer that times out after 10 seconds of
1079	inactivity.	2179	inactivity.
1080		2180
1081	static void	2181	static void
1082	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2182	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1083	{	2183	{
1084	.. ten seconds without any activity	2184	.. ten seconds without any activity
1085	}	2185	}
1086		2186
1087	struct ev_timer mytimer;	2187	ev_timer mytimer;
1088	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	2188	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1089	ev_timer_again (&mytimer); /* start timer */	2189	ev_timer_again (&mytimer); /* start timer */
1090	ev_loop (loop, 0);	2190	ev_run (loop, 0);
1091		2191
1092	// and in some piece of code that gets executed on any "activity":	2192	// and in some piece of code that gets executed on any "activity":
1093	// reset the timeout to start ticking again at 10 seconds	2193	// reset the timeout to start ticking again at 10 seconds
1094	ev_timer_again (&mytimer);	2194	ev_timer_again (&mytimer);
1095		2195
1096		2196
1097	=head2 C<ev_periodic> - to cron or not to cron?	2197	=head2 C<ev_periodic> - to cron or not to cron?
1098		2198
1099	Periodic watchers are also timers of a kind, but they are very versatile	2199	Periodic watchers are also timers of a kind, but they are very versatile
1100	(and unfortunately a bit complex).	2200	(and unfortunately a bit complex).
1101		2201
1102	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2202	Unlike C<ev_timer>, periodic watchers are not based on real time (or
1103	but on wallclock time (absolute time). You can tell a periodic watcher	2203	relative time, the physical time that passes) but on wall clock time
1104	to trigger "at" some specific point in time. For example, if you tell a	2204	(absolute time, the thing you can read on your calender or clock). The
1105	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	2205	difference is that wall clock time can run faster or slower than real
1106	+ 10.>) and then reset your system clock to the last year, then it will	2206	time, and time jumps are not uncommon (e.g. when you adjust your
1107	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2207	wrist-watch).
1108	roughly 10 seconds later).
1109		2208
1110	They can also be used to implement vastly more complex timers, such as	2209	You can tell a periodic watcher to trigger after some specific point
1111	triggering an event on each midnight, local time or other, complicated,	2210	in time: for example, if you tell a periodic watcher to trigger "in 10
1112	rules.	2211	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2212	not a delay) and then reset your system clock to January of the previous
		2213	year, then it will take a year or more to trigger the event (unlike an
		2214	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2215	it, as it uses a relative timeout).
1113		2216
		2217	C<ev_periodic> watchers can also be used to implement vastly more complex
		2218	timers, such as triggering an event on each "midnight, local time", or
		2219	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2220	those cannot react to time jumps.
		2221
1114	As with timers, the callback is guarenteed to be invoked only when the	2222	As with timers, the callback is guaranteed to be invoked only when the
1115	time (C<at>) has been passed, but if multiple periodic timers become ready	2223	point in time where it is supposed to trigger has passed. If multiple
1116	during the same loop iteration then order of execution is undefined.	2224	timers become ready during the same loop iteration then the ones with
		2225	earlier time-out values are invoked before ones with later time-out values
		2226	(but this is no longer true when a callback calls C<ev_run> recursively).
1117		2227
1118	=head3 Watcher-Specific Functions and Data Members	2228	=head3 Watcher-Specific Functions and Data Members
1119		2229
1120	=over 4	2230	=over 4
1121		2231
1122	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2232	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1123		2233
1124	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2234	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1125		2235
1126	Lots of arguments, lets sort it out... There are basically three modes of	2236	Lots of arguments, let's sort it out... There are basically three modes of
1127	operation, and we will explain them from simplest to complex:	2237	operation, and we will explain them from simplest to most complex:
1128		2238
1129	=over 4	2239	=over 4
1130		2240
1131	=item * absolute timer (at = time, interval = reschedule_cb = 0)	2241	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
1132		2242
1133	In this configuration the watcher triggers an event at the wallclock time	2243	In this configuration the watcher triggers an event after the wall clock
1134	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2244	time C<offset> has passed. It will not repeat and will not adjust when a
1135	that is, if it is to be run at January 1st 2011 then it will run when the	2245	time jump occurs, that is, if it is to be run at January 1st 2011 then it
1136	system time reaches or surpasses this time.	2246	will be stopped and invoked when the system clock reaches or surpasses
		2247	this point in time.
1137		2248
1138	=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	2249	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
1139		2250
1140	In this mode the watcher will always be scheduled to time out at the next	2251	In this mode the watcher will always be scheduled to time out at the next
1141	C<at + N * interval> time (for some integer N, which can also be negative)	2252	C<offset + N * interval> time (for some integer N, which can also be
1142	and then repeat, regardless of any time jumps.	2253	negative) and then repeat, regardless of any time jumps. The C<offset>
		2254	argument is merely an offset into the C<interval> periods.
1143		2255
1144	This can be used to create timers that do not drift with respect to system	2256	This can be used to create timers that do not drift with respect to the
1145	time:	2257	system clock, for example, here is an C<ev_periodic> that triggers each
		2258	hour, on the hour (with respect to UTC):
1146		2259
1147	ev_periodic_set (&periodic, 0., 3600., 0);	2260	ev_periodic_set (&periodic, 0., 3600., 0);
1148		2261
1149	This doesn't mean there will always be 3600 seconds in between triggers,	2262	This doesn't mean there will always be 3600 seconds in between triggers,
1150	but only that the the callback will be called when the system time shows a	2263	but only that the callback will be called when the system time shows a
1151	full hour (UTC), or more correctly, when the system time is evenly divisible	2264	full hour (UTC), or more correctly, when the system time is evenly divisible
1152	by 3600.	2265	by 3600.
1153		2266
1154	Another way to think about it (for the mathematically inclined) is that	2267	Another way to think about it (for the mathematically inclined) is that
1155	C<ev_periodic> will try to run the callback in this mode at the next possible	2268	C<ev_periodic> will try to run the callback in this mode at the next possible
1156	time where C<time = at (mod interval)>, regardless of any time jumps.	2269	time where C<time = offset (mod interval)>, regardless of any time jumps.
1157		2270
1158	For numerical stability it is preferable that the C<at> value is near	2271	The C<interval> I<MUST> be positive, and for numerical stability, the
1159	C<ev_now ()> (the current time), but there is no range requirement for	2272	interval value should be higher than C<1/8192> (which is around 100
1160	this value.	2273	microseconds) and C<offset> should be higher than C<0> and should have
		2274	at most a similar magnitude as the current time (say, within a factor of
		2275	ten). Typical values for offset are, in fact, C<0> or something between
		2276	C<0> and C<interval>, which is also the recommended range.
1161		2277
		2278	Note also that there is an upper limit to how often a timer can fire (CPU
		2279	speed for example), so if C<interval> is very small then timing stability
		2280	will of course deteriorate. Libev itself tries to be exact to be about one
		2281	millisecond (if the OS supports it and the machine is fast enough).
		2282
1162	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)	2283	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
1163		2284
1164	In this mode the values for C<interval> and C<at> are both being	2285	In this mode the values for C<interval> and C<offset> are both being
1165	ignored. Instead, each time the periodic watcher gets scheduled, the	2286	ignored. Instead, each time the periodic watcher gets scheduled, the
1166	reschedule callback will be called with the watcher as first, and the	2287	reschedule callback will be called with the watcher as first, and the
1167	current time as second argument.	2288	current time as second argument.
1168		2289
1169	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2290	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
1170	ever, or make any event loop modifications>. If you need to stop it,	2291	or make ANY other event loop modifications whatsoever, unless explicitly
1171	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2292	allowed by documentation here>.
1172	starting an C<ev_prepare> watcher, which is legal).
1173		2293
		2294	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2295	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2296	only event loop modification you are allowed to do).
		2297
1174	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2298	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1175	ev_tstamp now)>, e.g.:	2299	*w, ev_tstamp now)>, e.g.:
1176		2300
		2301	static ev_tstamp
1177	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2302	my_rescheduler (ev_periodic *w, ev_tstamp now)
1178	{	2303	{
1179	return now + 60.;	2304	return now + 60.;
1180	}	2305	}
1181		2306
1182	It must return the next time to trigger, based on the passed time value	2307	It must return the next time to trigger, based on the passed time value
1183	(that is, the lowest time value larger than to the second argument). It	2308	(that is, the lowest time value larger than to the second argument). It
1184	will usually be called just before the callback will be triggered, but	2309	will usually be called just before the callback will be triggered, but
1185	might be called at other times, too.	2310	might be called at other times, too.
1186		2311
1187	NOTE: I<< This callback must always return a time that is later than the	2312	NOTE: I<< This callback must always return a time that is higher than or
1188	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2313	equal to the passed C<now> value >>.
1189		2314
1190	This can be used to create very complex timers, such as a timer that	2315	This can be used to create very complex timers, such as a timer that
1191	triggers on each midnight, local time. To do this, you would calculate the	2316	triggers on "next midnight, local time". To do this, you would calculate the
1192	next midnight after C<now> and return the timestamp value for this. How	2317	next midnight after C<now> and return the timestamp value for this. How
1193	you do this is, again, up to you (but it is not trivial, which is the main	2318	you do this is, again, up to you (but it is not trivial, which is the main
1194	reason I omitted it as an example).	2319	reason I omitted it as an example).
1195		2320
1196	=back	2321	=back
…		…
1200	Simply stops and restarts the periodic watcher again. This is only useful	2325	Simply stops and restarts the periodic watcher again. This is only useful
1201	when you changed some parameters or the reschedule callback would return	2326	when you changed some parameters or the reschedule callback would return
1202	a different time than the last time it was called (e.g. in a crond like	2327	a different time than the last time it was called (e.g. in a crond like
1203	program when the crontabs have changed).	2328	program when the crontabs have changed).
1204		2329
		2330	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2331
		2332	When active, returns the absolute time that the watcher is supposed
		2333	to trigger next. This is not the same as the C<offset> argument to
		2334	C<ev_periodic_set>, but indeed works even in interval and manual
		2335	rescheduling modes.
		2336
1205	=item ev_tstamp offset [read-write]	2337	=item ev_tstamp offset [read-write]
1206		2338
1207	When repeating, this contains the offset value, otherwise this is the	2339	When repeating, this contains the offset value, otherwise this is the
1208	absolute point in time (the C<at> value passed to C<ev_periodic_set>).	2340	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2341	although libev might modify this value for better numerical stability).
1209		2342
1210	Can be modified any time, but changes only take effect when the periodic	2343	Can be modified any time, but changes only take effect when the periodic
1211	timer fires or C<ev_periodic_again> is being called.	2344	timer fires or C<ev_periodic_again> is being called.
1212		2345
1213	=item ev_tstamp interval [read-write]	2346	=item ev_tstamp interval [read-write]
1214		2347
1215	The current interval value. Can be modified any time, but changes only	2348	The current interval value. Can be modified any time, but changes only
1216	take effect when the periodic timer fires or C<ev_periodic_again> is being	2349	take effect when the periodic timer fires or C<ev_periodic_again> is being
1217	called.	2350	called.
1218		2351
1219	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	2352	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1220		2353
1221	The current reschedule callback, or C<0>, if this functionality is	2354	The current reschedule callback, or C<0>, if this functionality is
1222	switched off. Can be changed any time, but changes only take effect when	2355	switched off. Can be changed any time, but changes only take effect when
1223	the periodic timer fires or C<ev_periodic_again> is being called.	2356	the periodic timer fires or C<ev_periodic_again> is being called.
1224		2357
1225	=back	2358	=back
1226		2359
		2360	=head3 Examples
		2361
1227	Example: Call a callback every hour, or, more precisely, whenever the	2362	Example: Call a callback every hour, or, more precisely, whenever the
1228	system clock is divisible by 3600. The callback invocation times have	2363	system time is divisible by 3600. The callback invocation times have
1229	potentially a lot of jittering, but good long-term stability.	2364	potentially a lot of jitter, but good long-term stability.
1230		2365
1231	static void	2366	static void
1232	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	2367	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
1233	{	2368	{
1234	... its now a full hour (UTC, or TAI or whatever your clock follows)	2369	... its now a full hour (UTC, or TAI or whatever your clock follows)
1235	}	2370	}
1236		2371
1237	struct ev_periodic hourly_tick;	2372	ev_periodic hourly_tick;
1238	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	2373	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1239	ev_periodic_start (loop, &hourly_tick);	2374	ev_periodic_start (loop, &hourly_tick);
1240		2375
1241	Example: The same as above, but use a reschedule callback to do it:	2376	Example: The same as above, but use a reschedule callback to do it:
1242		2377
1243	#include <math.h>	2378	#include <math.h>
1244		2379
1245	static ev_tstamp	2380	static ev_tstamp
1246	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	2381	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1247	{	2382	{
1248	return fmod (now, 3600.) + 3600.;	2383	return now + (3600. - fmod (now, 3600.));
1249	}	2384	}
1250		2385
1251	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	2386	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1252		2387
1253	Example: Call a callback every hour, starting now:	2388	Example: Call a callback every hour, starting now:
1254		2389
1255	struct ev_periodic hourly_tick;	2390	ev_periodic hourly_tick;
1256	ev_periodic_init (&hourly_tick, clock_cb,	2391	ev_periodic_init (&hourly_tick, clock_cb,
1257	fmod (ev_now (loop), 3600.), 3600., 0);	2392	fmod (ev_now (loop), 3600.), 3600., 0);
1258	ev_periodic_start (loop, &hourly_tick);	2393	ev_periodic_start (loop, &hourly_tick);
1259		2394
1260		2395
1261	=head2 C<ev_signal> - signal me when a signal gets signalled!	2396	=head2 C<ev_signal> - signal me when a signal gets signalled!
1262		2397
1263	Signal watchers will trigger an event when the process receives a specific	2398	Signal watchers will trigger an event when the process receives a specific
1264	signal one or more times. Even though signals are very asynchronous, libev	2399	signal one or more times. Even though signals are very asynchronous, libev
1265	will try it's best to deliver signals synchronously, i.e. as part of the	2400	will try its best to deliver signals synchronously, i.e. as part of the
1266	normal event processing, like any other event.	2401	normal event processing, like any other event.
1267		2402
		2403	If you want signals to be delivered truly asynchronously, just use
		2404	C<sigaction> as you would do without libev and forget about sharing
		2405	the signal. You can even use C<ev_async> from a signal handler to
		2406	synchronously wake up an event loop.
		2407
1268	You can configure as many watchers as you like per signal. Only when the	2408	You can configure as many watchers as you like for the same signal, but
		2409	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2410	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2411	C<SIGINT> in both the default loop and another loop at the same time. At
		2412	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2413
1269	first watcher gets started will libev actually register a signal watcher	2414	When the first watcher gets started will libev actually register something
1270	with the kernel (thus it coexists with your own signal handlers as long	2415	with the kernel (thus it coexists with your own signal handlers as long as
1271	as you don't register any with libev). Similarly, when the last signal	2416	you don't register any with libev for the same signal).
1272	watcher for a signal is stopped libev will reset the signal handler to	2417
1273	SIG_DFL (regardless of what it was set to before).	2418	If possible and supported, libev will install its handlers with
		2419	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2420	not be unduly interrupted. If you have a problem with system calls getting
		2421	interrupted by signals you can block all signals in an C<ev_check> watcher
		2422	and unblock them in an C<ev_prepare> watcher.
		2423
		2424	=head3 The special problem of inheritance over fork/execve/pthread_create
		2425
		2426	Both the signal mask (C<sigprocmask>) and the signal disposition
		2427	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2428	stopping it again), that is, libev might or might not block the signal,
		2429	and might or might not set or restore the installed signal handler (but
		2430	see C<EVFLAG_NOSIGMASK>).
		2431
		2432	While this does not matter for the signal disposition (libev never
		2433	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2434	C<execve>), this matters for the signal mask: many programs do not expect
		2435	certain signals to be blocked.
		2436
		2437	This means that before calling C<exec> (from the child) you should reset
		2438	the signal mask to whatever "default" you expect (all clear is a good
		2439	choice usually).
		2440
		2441	The simplest way to ensure that the signal mask is reset in the child is
		2442	to install a fork handler with C<pthread_atfork> that resets it. That will
		2443	catch fork calls done by libraries (such as the libc) as well.
		2444
		2445	In current versions of libev, the signal will not be blocked indefinitely
		2446	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2447	the window of opportunity for problems, it will not go away, as libev
		2448	I<has> to modify the signal mask, at least temporarily.
		2449
		2450	So I can't stress this enough: I<If you do not reset your signal mask when
		2451	you expect it to be empty, you have a race condition in your code>. This
		2452	is not a libev-specific thing, this is true for most event libraries.
		2453
		2454	=head3 The special problem of threads signal handling
		2455
		2456	POSIX threads has problematic signal handling semantics, specifically,
		2457	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2458	threads in a process block signals, which is hard to achieve.
		2459
		2460	When you want to use sigwait (or mix libev signal handling with your own
		2461	for the same signals), you can tackle this problem by globally blocking
		2462	all signals before creating any threads (or creating them with a fully set
		2463	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2464	loops. Then designate one thread as "signal receiver thread" which handles
		2465	these signals. You can pass on any signals that libev might be interested
		2466	in by calling C<ev_feed_signal>.
1274		2467
1275	=head3 Watcher-Specific Functions and Data Members	2468	=head3 Watcher-Specific Functions and Data Members
1276		2469
1277	=over 4	2470	=over 4
1278		2471
…		…
1287		2480
1288	The signal the watcher watches out for.	2481	The signal the watcher watches out for.
1289		2482
1290	=back	2483	=back
1291		2484
		2485	=head3 Examples
		2486
		2487	Example: Try to exit cleanly on SIGINT.
		2488
		2489	static void
		2490	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2491	{
		2492	ev_break (loop, EVBREAK_ALL);
		2493	}
		2494
		2495	ev_signal signal_watcher;
		2496	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2497	ev_signal_start (loop, &signal_watcher);
		2498
1292		2499
1293	=head2 C<ev_child> - watch out for process status changes	2500	=head2 C<ev_child> - watch out for process status changes
1294		2501
1295	Child watchers trigger when your process receives a SIGCHLD in response to	2502	Child watchers trigger when your process receives a SIGCHLD in response to
1296	some child status changes (most typically when a child of yours dies).	2503	some child status changes (most typically when a child of yours dies or
		2504	exits). It is permissible to install a child watcher I<after> the child
		2505	has been forked (which implies it might have already exited), as long
		2506	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2507	forking and then immediately registering a watcher for the child is fine,
		2508	but forking and registering a watcher a few event loop iterations later or
		2509	in the next callback invocation is not.
		2510
		2511	Only the default event loop is capable of handling signals, and therefore
		2512	you can only register child watchers in the default event loop.
		2513
		2514	Due to some design glitches inside libev, child watchers will always be
		2515	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2516	libev)
		2517
		2518	=head3 Process Interaction
		2519
		2520	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2521	initialised. This is necessary to guarantee proper behaviour even if the
		2522	first child watcher is started after the child exits. The occurrence
		2523	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2524	synchronously as part of the event loop processing. Libev always reaps all
		2525	children, even ones not watched.
		2526
		2527	=head3 Overriding the Built-In Processing
		2528
		2529	Libev offers no special support for overriding the built-in child
		2530	processing, but if your application collides with libev's default child
		2531	handler, you can override it easily by installing your own handler for
		2532	C<SIGCHLD> after initialising the default loop, and making sure the
		2533	default loop never gets destroyed. You are encouraged, however, to use an
		2534	event-based approach to child reaping and thus use libev's support for
		2535	that, so other libev users can use C<ev_child> watchers freely.
		2536
		2537	=head3 Stopping the Child Watcher
		2538
		2539	Currently, the child watcher never gets stopped, even when the
		2540	child terminates, so normally one needs to stop the watcher in the
		2541	callback. Future versions of libev might stop the watcher automatically
		2542	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2543	problem).
1297		2544
1298	=head3 Watcher-Specific Functions and Data Members	2545	=head3 Watcher-Specific Functions and Data Members
1299		2546
1300	=over 4	2547	=over 4
1301		2548
1302	=item ev_child_init (ev_child *, callback, int pid)	2549	=item ev_child_init (ev_child *, callback, int pid, int trace)
1303		2550
1304	=item ev_child_set (ev_child *, int pid)	2551	=item ev_child_set (ev_child *, int pid, int trace)
1305		2552
1306	Configures the watcher to wait for status changes of process C<pid> (or	2553	Configures the watcher to wait for status changes of process C<pid> (or
1307	I<any> process if C<pid> is specified as C<0>). The callback can look	2554	I<any> process if C<pid> is specified as C<0>). The callback can look
1308	at the C<rstatus> member of the C<ev_child> watcher structure to see	2555	at the C<rstatus> member of the C<ev_child> watcher structure to see
1309	the status word (use the macros from C<sys/wait.h> and see your systems	2556	the status word (use the macros from C<sys/wait.h> and see your systems
1310	C<waitpid> documentation). The C<rpid> member contains the pid of the	2557	C<waitpid> documentation). The C<rpid> member contains the pid of the
1311	process causing the status change.	2558	process causing the status change. C<trace> must be either C<0> (only
		2559	activate the watcher when the process terminates) or C<1> (additionally
		2560	activate the watcher when the process is stopped or continued).
1312		2561
1313	=item int pid [read-only]	2562	=item int pid [read-only]
1314		2563
1315	The process id this watcher watches out for, or C<0>, meaning any process id.	2564	The process id this watcher watches out for, or C<0>, meaning any process id.
1316		2565
…		…
1323	The process exit/trace status caused by C<rpid> (see your systems	2572	The process exit/trace status caused by C<rpid> (see your systems
1324	C<waitpid> and C<sys/wait.h> documentation for details).	2573	C<waitpid> and C<sys/wait.h> documentation for details).
1325		2574
1326	=back	2575	=back
1327		2576
1328	Example: Try to exit cleanly on SIGINT and SIGTERM.	2577	=head3 Examples
1329		2578
		2579	Example: C<fork()> a new process and install a child handler to wait for
		2580	its completion.
		2581
		2582	ev_child cw;
		2583
1330	static void	2584	static void
1331	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	2585	child_cb (EV_P_ ev_child *w, int revents)
1332	{	2586	{
1333	ev_unloop (loop, EVUNLOOP_ALL);	2587	ev_child_stop (EV_A_ w);
		2588	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1334	}	2589	}
1335		2590
1336	struct ev_signal signal_watcher;	2591	pid_t pid = fork ();
1337	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	2592
1338	ev_signal_start (loop, &sigint_cb);	2593	if (pid < 0)
		2594	// error
		2595	else if (pid == 0)
		2596	{
		2597	// the forked child executes here
		2598	exit (1);
		2599	}
		2600	else
		2601	{
		2602	ev_child_init (&cw, child_cb, pid, 0);
		2603	ev_child_start (EV_DEFAULT_ &cw);
		2604	}
1339		2605
1340		2606
1341	=head2 C<ev_stat> - did the file attributes just change?	2607	=head2 C<ev_stat> - did the file attributes just change?
1342		2608
1343	This watches a filesystem path for attribute changes. That is, it calls	2609	This watches a file system path for attribute changes. That is, it calls
1344	C<stat> regularly (or when the OS says it changed) and sees if it changed	2610	C<stat> on that path in regular intervals (or when the OS says it changed)
1345	compared to the last time, invoking the callback if it did.	2611	and sees if it changed compared to the last time, invoking the callback if
		2612	it did.
1346		2613
1347	The path does not need to exist: changing from "path exists" to "path does	2614	The path does not need to exist: changing from "path exists" to "path does
1348	not exist" is a status change like any other. The condition "path does	2615	not exist" is a status change like any other. The condition "path does not
1349	not exist" is signified by the C<st_nlink> field being zero (which is	2616	exist" (or more correctly "path cannot be stat'ed") is signified by the
1350	otherwise always forced to be at least one) and all the other fields of	2617	C<st_nlink> field being zero (which is otherwise always forced to be at
1351	the stat buffer having unspecified contents.	2618	least one) and all the other fields of the stat buffer having unspecified
		2619	contents.
1352		2620
1353	The path I<should> be absolute and I<must not> end in a slash. If it is	2621	The path I<must not> end in a slash or contain special components such as
		2622	C<.> or C<..>. The path I<should> be absolute: If it is relative and
1354	relative and your working directory changes, the behaviour is undefined.	2623	your working directory changes, then the behaviour is undefined.
1355		2624
1356	Since there is no standard to do this, the portable implementation simply	2625	Since there is no portable change notification interface available, the
1357	calls C<stat (2)> regularly on the path to see if it changed somehow. You	2626	portable implementation simply calls C<stat(2)> regularly on the path
1358	can specify a recommended polling interval for this case. If you specify	2627	to see if it changed somehow. You can specify a recommended polling
1359	a polling interval of C<0> (highly recommended!) then a I<suitable,	2628	interval for this case. If you specify a polling interval of C<0> (highly
1360	unspecified default> value will be used (which you can expect to be around	2629	recommended!) then a I<suitable, unspecified default> value will be used
1361	five seconds, although this might change dynamically). Libev will also	2630	(which you can expect to be around five seconds, although this might
1362	impose a minimum interval which is currently around C<0.1>, but thats	2631	change dynamically). Libev will also impose a minimum interval which is
1363	usually overkill.	2632	currently around C<0.1>, but that's usually overkill.
1364		2633
1365	This watcher type is not meant for massive numbers of stat watchers,	2634	This watcher type is not meant for massive numbers of stat watchers,
1366	as even with OS-supported change notifications, this can be	2635	as even with OS-supported change notifications, this can be
1367	resource-intensive.	2636	resource-intensive.
1368		2637
1369	At the time of this writing, only the Linux inotify interface is	2638	At the time of this writing, the only OS-specific interface implemented
1370	implemented (implementing kqueue support is left as an exercise for the	2639	is the Linux inotify interface (implementing kqueue support is left as an
1371	reader). Inotify will be used to give hints only and should not change the	2640	exercise for the reader. Note, however, that the author sees no way of
1372	semantics of C<ev_stat> watchers, which means that libev sometimes needs	2641	implementing C<ev_stat> semantics with kqueue, except as a hint).
1373	to fall back to regular polling again even with inotify, but changes are	2642
1374	usually detected immediately, and if the file exists there will be no	2643	=head3 ABI Issues (Largefile Support)
1375	polling.	2644
		2645	Libev by default (unless the user overrides this) uses the default
		2646	compilation environment, which means that on systems with large file
		2647	support disabled by default, you get the 32 bit version of the stat
		2648	structure. When using the library from programs that change the ABI to
		2649	use 64 bit file offsets the programs will fail. In that case you have to
		2650	compile libev with the same flags to get binary compatibility. This is
		2651	obviously the case with any flags that change the ABI, but the problem is
		2652	most noticeably displayed with ev_stat and large file support.
		2653
		2654	The solution for this is to lobby your distribution maker to make large
		2655	file interfaces available by default (as e.g. FreeBSD does) and not
		2656	optional. Libev cannot simply switch on large file support because it has
		2657	to exchange stat structures with application programs compiled using the
		2658	default compilation environment.
		2659
		2660	=head3 Inotify and Kqueue
		2661
		2662	When C<inotify (7)> support has been compiled into libev and present at
		2663	runtime, it will be used to speed up change detection where possible. The
		2664	inotify descriptor will be created lazily when the first C<ev_stat>
		2665	watcher is being started.
		2666
		2667	Inotify presence does not change the semantics of C<ev_stat> watchers
		2668	except that changes might be detected earlier, and in some cases, to avoid
		2669	making regular C<stat> calls. Even in the presence of inotify support
		2670	there are many cases where libev has to resort to regular C<stat> polling,
		2671	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2672	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2673	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2674	xfs are fully working) libev usually gets away without polling.
		2675
		2676	There is no support for kqueue, as apparently it cannot be used to
		2677	implement this functionality, due to the requirement of having a file
		2678	descriptor open on the object at all times, and detecting renames, unlinks
		2679	etc. is difficult.
		2680
		2681	=head3 C<stat ()> is a synchronous operation
		2682
		2683	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2684	the process. The exception are C<ev_stat> watchers - those call C<stat
		2685	()>, which is a synchronous operation.
		2686
		2687	For local paths, this usually doesn't matter: unless the system is very
		2688	busy or the intervals between stat's are large, a stat call will be fast,
		2689	as the path data is usually in memory already (except when starting the
		2690	watcher).
		2691
		2692	For networked file systems, calling C<stat ()> can block an indefinite
		2693	time due to network issues, and even under good conditions, a stat call
		2694	often takes multiple milliseconds.
		2695
		2696	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2697	paths, although this is fully supported by libev.
		2698
		2699	=head3 The special problem of stat time resolution
		2700
		2701	The C<stat ()> system call only supports full-second resolution portably,
		2702	and even on systems where the resolution is higher, most file systems
		2703	still only support whole seconds.
		2704
		2705	That means that, if the time is the only thing that changes, you can
		2706	easily miss updates: on the first update, C<ev_stat> detects a change and
		2707	calls your callback, which does something. When there is another update
		2708	within the same second, C<ev_stat> will be unable to detect unless the
		2709	stat data does change in other ways (e.g. file size).
		2710
		2711	The solution to this is to delay acting on a change for slightly more
		2712	than a second (or till slightly after the next full second boundary), using
		2713	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2714	ev_timer_again (loop, w)>).
		2715
		2716	The C<.02> offset is added to work around small timing inconsistencies
		2717	of some operating systems (where the second counter of the current time
		2718	might be be delayed. One such system is the Linux kernel, where a call to
		2719	C<gettimeofday> might return a timestamp with a full second later than
		2720	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2721	update file times then there will be a small window where the kernel uses
		2722	the previous second to update file times but libev might already execute
		2723	the timer callback).
1376		2724
1377	=head3 Watcher-Specific Functions and Data Members	2725	=head3 Watcher-Specific Functions and Data Members
1378		2726
1379	=over 4	2727	=over 4
1380		2728
…		…
1386	C<path>. The C<interval> is a hint on how quickly a change is expected to	2734	C<path>. The C<interval> is a hint on how quickly a change is expected to
1387	be detected and should normally be specified as C<0> to let libev choose	2735	be detected and should normally be specified as C<0> to let libev choose
1388	a suitable value. The memory pointed to by C<path> must point to the same	2736	a suitable value. The memory pointed to by C<path> must point to the same
1389	path for as long as the watcher is active.	2737	path for as long as the watcher is active.
1390		2738
1391	The callback will be receive C<EV_STAT> when a change was detected,	2739	The callback will receive an C<EV_STAT> event when a change was detected,
1392	relative to the attributes at the time the watcher was started (or the	2740	relative to the attributes at the time the watcher was started (or the
1393	last change was detected).	2741	last change was detected).
1394		2742
1395	=item ev_stat_stat (ev_stat *)	2743	=item ev_stat_stat (loop, ev_stat *)
1396		2744
1397	Updates the stat buffer immediately with new values. If you change the	2745	Updates the stat buffer immediately with new values. If you change the
1398	watched path in your callback, you could call this fucntion to avoid	2746	watched path in your callback, you could call this function to avoid
1399	detecting this change (while introducing a race condition). Can also be	2747	detecting this change (while introducing a race condition if you are not
1400	useful simply to find out the new values.	2748	the only one changing the path). Can also be useful simply to find out the
		2749	new values.
1401		2750
1402	=item ev_statdata attr [read-only]	2751	=item ev_statdata attr [read-only]
1403		2752
1404	The most-recently detected attributes of the file. Although the type is of	2753	The most-recently detected attributes of the file. Although the type is
1405	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types	2754	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1406	suitable for your system. If the C<st_nlink> member is C<0>, then there	2755	suitable for your system, but you can only rely on the POSIX-standardised
		2756	members to be present. If the C<st_nlink> member is C<0>, then there was
1407	was some error while C<stat>ing the file.	2757	some error while C<stat>ing the file.
1408		2758
1409	=item ev_statdata prev [read-only]	2759	=item ev_statdata prev [read-only]
1410		2760
1411	The previous attributes of the file. The callback gets invoked whenever	2761	The previous attributes of the file. The callback gets invoked whenever
1412	C<prev> != C<attr>.	2762	C<prev> != C<attr>, or, more precisely, one or more of these members
		2763	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2764	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
1413		2765
1414	=item ev_tstamp interval [read-only]	2766	=item ev_tstamp interval [read-only]
1415		2767
1416	The specified interval.	2768	The specified interval.
1417		2769
1418	=item const char *path [read-only]	2770	=item const char *path [read-only]
1419		2771
1420	The filesystem path that is being watched.	2772	The file system path that is being watched.
1421		2773
1422	=back	2774	=back
1423		2775
		2776	=head3 Examples
		2777
1424	Example: Watch C</etc/passwd> for attribute changes.	2778	Example: Watch C</etc/passwd> for attribute changes.
1425		2779
1426	static void	2780	static void
1427	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2781	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1428	{	2782	{
1429	/* /etc/passwd changed in some way */	2783	/* /etc/passwd changed in some way */
1430	if (w->attr.st_nlink)	2784	if (w->attr.st_nlink)
1431	{	2785	{
1432	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2786	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1433	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2787	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1434	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2788	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1435	}	2789	}
1436	else	2790	else
1437	/* you shalt not abuse printf for puts */	2791	/* you shalt not abuse printf for puts */
1438	puts ("wow, /etc/passwd is not there, expect problems. "	2792	puts ("wow, /etc/passwd is not there, expect problems. "
1439	"if this is windows, they already arrived\n");	2793	"if this is windows, they already arrived\n");
1440	}	2794	}
1441		2795
1442	...	2796	...
1443	ev_stat passwd;	2797	ev_stat passwd;
1444		2798
1445	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	2799	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1446	ev_stat_start (loop, &passwd);	2800	ev_stat_start (loop, &passwd);
		2801
		2802	Example: Like above, but additionally use a one-second delay so we do not
		2803	miss updates (however, frequent updates will delay processing, too, so
		2804	one might do the work both on C<ev_stat> callback invocation I<and> on
		2805	C<ev_timer> callback invocation).
		2806
		2807	static ev_stat passwd;
		2808	static ev_timer timer;
		2809
		2810	static void
		2811	timer_cb (EV_P_ ev_timer *w, int revents)
		2812	{
		2813	ev_timer_stop (EV_A_ w);
		2814
		2815	/* now it's one second after the most recent passwd change */
		2816	}
		2817
		2818	static void
		2819	stat_cb (EV_P_ ev_stat *w, int revents)
		2820	{
		2821	/* reset the one-second timer */
		2822	ev_timer_again (EV_A_ &timer);
		2823	}
		2824
		2825	...
		2826	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2827	ev_stat_start (loop, &passwd);
		2828	ev_timer_init (&timer, timer_cb, 0., 1.02);
1447		2829
1448		2830
1449	=head2 C<ev_idle> - when you've got nothing better to do...	2831	=head2 C<ev_idle> - when you've got nothing better to do...
1450		2832
1451	Idle watchers trigger events when no other events of the same or higher	2833	Idle watchers trigger events when no other events of the same or higher
1452	priority are pending (prepare, check and other idle watchers do not	2834	priority are pending (prepare, check and other idle watchers do not count
1453	count).	2835	as receiving "events").
1454		2836
1455	That is, as long as your process is busy handling sockets or timeouts	2837	That is, as long as your process is busy handling sockets or timeouts
1456	(or even signals, imagine) of the same or higher priority it will not be	2838	(or even signals, imagine) of the same or higher priority it will not be
1457	triggered. But when your process is idle (or only lower-priority watchers	2839	triggered. But when your process is idle (or only lower-priority watchers
1458	are pending), the idle watchers are being called once per event loop	2840	are pending), the idle watchers are being called once per event loop
…		…
1465	Apart from keeping your process non-blocking (which is a useful	2847	Apart from keeping your process non-blocking (which is a useful
1466	effect on its own sometimes), idle watchers are a good place to do	2848	effect on its own sometimes), idle watchers are a good place to do
1467	"pseudo-background processing", or delay processing stuff to after the	2849	"pseudo-background processing", or delay processing stuff to after the
1468	event loop has handled all outstanding events.	2850	event loop has handled all outstanding events.
1469		2851
		2852	=head3 Abusing an C<ev_idle> watcher for its side-effect
		2853
		2854	As long as there is at least one active idle watcher, libev will never
		2855	sleep unnecessarily. Or in other words, it will loop as fast as possible.
		2856	For this to work, the idle watcher doesn't need to be invoked at all - the
		2857	lowest priority will do.
		2858
		2859	This mode of operation can be useful together with an C<ev_check> watcher,
		2860	to do something on each event loop iteration - for example to balance load
		2861	between different connections.
		2862
		2863	See L</Abusing an ev_check watcher for its side-effect> for a longer
		2864	example.
		2865
1470	=head3 Watcher-Specific Functions and Data Members	2866	=head3 Watcher-Specific Functions and Data Members
1471		2867
1472	=over 4	2868	=over 4
1473		2869
1474	=item ev_idle_init (ev_signal *, callback)	2870	=item ev_idle_init (ev_idle *, callback)
1475		2871
1476	Initialises and configures the idle watcher - it has no parameters of any	2872	Initialises and configures the idle watcher - it has no parameters of any
1477	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2873	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1478	believe me.	2874	believe me.
1479		2875
1480	=back	2876	=back
1481		2877
		2878	=head3 Examples
		2879
1482	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	2880	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1483	callback, free it. Also, use no error checking, as usual.	2881	callback, free it. Also, use no error checking, as usual.
1484		2882
1485	static void	2883	static void
1486	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2884	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1487	{	2885	{
		2886	// stop the watcher
		2887	ev_idle_stop (loop, w);
		2888
		2889	// now we can free it
1488	free (w);	2890	free (w);
		2891
1489	// now do something you wanted to do when the program has	2892	// now do something you wanted to do when the program has
1490	// no longer asnything immediate to do.	2893	// no longer anything immediate to do.
1491	}	2894	}
1492		2895
1493	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2896	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1494	ev_idle_init (idle_watcher, idle_cb);	2897	ev_idle_init (idle_watcher, idle_cb);
1495	ev_idle_start (loop, idle_cb);	2898	ev_idle_start (loop, idle_watcher);
1496		2899
1497		2900
1498	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	2901	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1499		2902
1500	Prepare and check watchers are usually (but not always) used in tandem:	2903	Prepare and check watchers are often (but not always) used in pairs:
1501	prepare watchers get invoked before the process blocks and check watchers	2904	prepare watchers get invoked before the process blocks and check watchers
1502	afterwards.	2905	afterwards.
1503		2906
1504	You I<must not> call C<ev_loop> or similar functions that enter	2907	You I<must not> call C<ev_run> or similar functions that enter
1505	the current event loop from either C<ev_prepare> or C<ev_check>	2908	the current event loop from either C<ev_prepare> or C<ev_check>
1506	watchers. Other loops than the current one are fine, however. The	2909	watchers. Other loops than the current one are fine, however. The
1507	rationale behind this is that you do not need to check for recursion in	2910	rationale behind this is that you do not need to check for recursion in
1508	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	2911	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
1509	C<ev_check> so if you have one watcher of each kind they will always be	2912	C<ev_check> so if you have one watcher of each kind they will always be
1510	called in pairs bracketing the blocking call.	2913	called in pairs bracketing the blocking call.
1511		2914
1512	Their main purpose is to integrate other event mechanisms into libev and	2915	Their main purpose is to integrate other event mechanisms into libev and
1513	their use is somewhat advanced. This could be used, for example, to track	2916	their use is somewhat advanced. They could be used, for example, to track
1514	variable changes, implement your own watchers, integrate net-snmp or a	2917	variable changes, implement your own watchers, integrate net-snmp or a
1515	coroutine library and lots more. They are also occasionally useful if	2918	coroutine library and lots more. They are also occasionally useful if
1516	you cache some data and want to flush it before blocking (for example,	2919	you cache some data and want to flush it before blocking (for example,
1517	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	2920	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1518	watcher).	2921	watcher).
1519		2922
1520	This is done by examining in each prepare call which file descriptors need	2923	This is done by examining in each prepare call which file descriptors
1521	to be watched by the other library, registering C<ev_io> watchers for	2924	need to be watched by the other library, registering C<ev_io> watchers
1522	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2925	for them and starting an C<ev_timer> watcher for any timeouts (many
1523	provide just this functionality). Then, in the check watcher you check for	2926	libraries provide exactly this functionality). Then, in the check watcher,
1524	any events that occured (by checking the pending status of all watchers	2927	you check for any events that occurred (by checking the pending status
1525	and stopping them) and call back into the library. The I/O and timer	2928	of all watchers and stopping them) and call back into the library. The
1526	callbacks will never actually be called (but must be valid nevertheless,	2929	I/O and timer callbacks will never actually be called (but must be valid
1527	because you never know, you know?).	2930	nevertheless, because you never know, you know?).
1528		2931
1529	As another example, the Perl Coro module uses these hooks to integrate	2932	As another example, the Perl Coro module uses these hooks to integrate
1530	coroutines into libev programs, by yielding to other active coroutines	2933	coroutines into libev programs, by yielding to other active coroutines
1531	during each prepare and only letting the process block if no coroutines	2934	during each prepare and only letting the process block if no coroutines
1532	are ready to run (it's actually more complicated: it only runs coroutines	2935	are ready to run (it's actually more complicated: it only runs coroutines
1533	with priority higher than or equal to the event loop and one coroutine	2936	with priority higher than or equal to the event loop and one coroutine
1534	of lower priority, but only once, using idle watchers to keep the event	2937	of lower priority, but only once, using idle watchers to keep the event
1535	loop from blocking if lower-priority coroutines are active, thus mapping	2938	loop from blocking if lower-priority coroutines are active, thus mapping
1536	low-priority coroutines to idle/background tasks).	2939	low-priority coroutines to idle/background tasks).
1537		2940
1538	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	2941	When used for this purpose, it is recommended to give C<ev_check> watchers
1539	priority, to ensure that they are being run before any other watchers	2942	highest (C<EV_MAXPRI>) priority, to ensure that they are being run before
		2943	any other watchers after the poll (this doesn't matter for C<ev_prepare>
		2944	watchers).
		2945
1540	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	2946	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
1541	too) should not activate ("feed") events into libev. While libev fully	2947	activate ("feed") events into libev. While libev fully supports this, they
1542	supports this, they will be called before other C<ev_check> watchers did	2948	might get executed before other C<ev_check> watchers did their job. As
1543	their job. As C<ev_check> watchers are often used to embed other event	2949	C<ev_check> watchers are often used to embed other (non-libev) event
1544	loops those other event loops might be in an unusable state until their	2950	loops those other event loops might be in an unusable state until their
1545	C<ev_check> watcher ran (always remind yourself to coexist peacefully with	2951	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
1546	others).	2952	others).
1547		2953
		2954	=head3 Abusing an C<ev_check> watcher for its side-effect
		2955
		2956	C<ev_check> (and less often also C<ev_prepare>) watchers can also be
		2957	useful because they are called once per event loop iteration. For
		2958	example, if you want to handle a large number of connections fairly, you
		2959	normally only do a bit of work for each active connection, and if there
		2960	is more work to do, you wait for the next event loop iteration, so other
		2961	connections have a chance of making progress.
		2962
		2963	Using an C<ev_check> watcher is almost enough: it will be called on the
		2964	next event loop iteration. However, that isn't as soon as possible -
		2965	without external events, your C<ev_check> watcher will not be invoked.
		2966
		2967	This is where C<ev_idle> watchers come in handy - all you need is a
		2968	single global idle watcher that is active as long as you have one active
		2969	C<ev_check> watcher. The C<ev_idle> watcher makes sure the event loop
		2970	will not sleep, and the C<ev_check> watcher makes sure a callback gets
		2971	invoked. Neither watcher alone can do that.
		2972
1548	=head3 Watcher-Specific Functions and Data Members	2973	=head3 Watcher-Specific Functions and Data Members
1549		2974
1550	=over 4	2975	=over 4
1551		2976
1552	=item ev_prepare_init (ev_prepare *, callback)	2977	=item ev_prepare_init (ev_prepare *, callback)
1553		2978
1554	=item ev_check_init (ev_check *, callback)	2979	=item ev_check_init (ev_check *, callback)
1555		2980
1556	Initialises and configures the prepare or check watcher - they have no	2981	Initialises and configures the prepare or check watcher - they have no
1557	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2982	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1558	macros, but using them is utterly, utterly and completely pointless.	2983	macros, but using them is utterly, utterly, utterly and completely
		2984	pointless.
1559		2985
1560	=back	2986	=back
		2987
		2988	=head3 Examples
1561		2989
1562	There are a number of principal ways to embed other event loops or modules	2990	There are a number of principal ways to embed other event loops or modules
1563	into libev. Here are some ideas on how to include libadns into libev	2991	into libev. Here are some ideas on how to include libadns into libev
1564	(there is a Perl module named C<EV::ADNS> that does this, which you could	2992	(there is a Perl module named C<EV::ADNS> that does this, which you could
1565	use for an actually working example. Another Perl module named C<EV::Glib>	2993	use as a working example. Another Perl module named C<EV::Glib> embeds a
1566	embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV	2994	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
1567	into the Glib event loop).	2995	Glib event loop).
1568		2996
1569	Method 1: Add IO watchers and a timeout watcher in a prepare handler,	2997	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1570	and in a check watcher, destroy them and call into libadns. What follows	2998	and in a check watcher, destroy them and call into libadns. What follows
1571	is pseudo-code only of course. This requires you to either use a low	2999	is pseudo-code only of course. This requires you to either use a low
1572	priority for the check watcher or use C<ev_clear_pending> explicitly, as	3000	priority for the check watcher or use C<ev_clear_pending> explicitly, as
1573	the callbacks for the IO/timeout watchers might not have been called yet.	3001	the callbacks for the IO/timeout watchers might not have been called yet.
1574		3002
1575	static ev_io iow [nfd];	3003	static ev_io iow [nfd];
1576	static ev_timer tw;	3004	static ev_timer tw;
1577		3005
1578	static void	3006	static void
1579	io_cb (ev_loop *loop, ev_io *w, int revents)	3007	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1580	{	3008	{
1581	}	3009	}
1582		3010
1583	// create io watchers for each fd and a timer before blocking	3011	// create io watchers for each fd and a timer before blocking
1584	static void	3012	static void
1585	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	3013	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1586	{	3014	{
1587	int timeout = 3600000;	3015	int timeout = 3600000;
1588	struct pollfd fds [nfd];	3016	struct pollfd fds [nfd];
1589	// actual code will need to loop here and realloc etc.	3017	// actual code will need to loop here and realloc etc.
1590	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	3018	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1591		3019
1592	/* the callback is illegal, but won't be called as we stop during check */	3020	/* the callback is illegal, but won't be called as we stop during check */
1593	ev_timer_init (&tw, 0, timeout * 1e-3);	3021	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
1594	ev_timer_start (loop, &tw);	3022	ev_timer_start (loop, &tw);
1595		3023
1596	// create one ev_io per pollfd	3024	// create one ev_io per pollfd
1597	for (int i = 0; i < nfd; ++i)	3025	for (int i = 0; i < nfd; ++i)
1598	{	3026	{
1599	ev_io_init (iow + i, io_cb, fds [i].fd,	3027	ev_io_init (iow + i, io_cb, fds [i].fd,
1600	((fds [i].events & POLLIN ? EV_READ : 0)	3028	((fds [i].events & POLLIN ? EV_READ : 0)
1601	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	3029	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1602		3030
1603	fds [i].revents = 0;	3031	fds [i].revents = 0;
1604	ev_io_start (loop, iow + i);	3032	ev_io_start (loop, iow + i);
1605	}	3033	}
1606	}	3034	}
1607		3035
1608	// stop all watchers after blocking	3036	// stop all watchers after blocking
1609	static void	3037	static void
1610	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	3038	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1611	{	3039	{
1612	ev_timer_stop (loop, &tw);	3040	ev_timer_stop (loop, &tw);
1613		3041
1614	for (int i = 0; i < nfd; ++i)	3042	for (int i = 0; i < nfd; ++i)
1615	{	3043	{
1616	// set the relevant poll flags	3044	// set the relevant poll flags
1617	// could also call adns_processreadable etc. here	3045	// could also call adns_processreadable etc. here
1618	struct pollfd *fd = fds + i;	3046	struct pollfd *fd = fds + i;
1619	int revents = ev_clear_pending (iow + i);	3047	int revents = ev_clear_pending (iow + i);
1620	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;	3048	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1621	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;	3049	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1622		3050
1623	// now stop the watcher	3051	// now stop the watcher
1624	ev_io_stop (loop, iow + i);	3052	ev_io_stop (loop, iow + i);
1625	}	3053	}
1626		3054
1627	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	3055	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1628	}	3056	}
1629		3057
1630	Method 2: This would be just like method 1, but you run C<adns_afterpoll>	3058	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
1631	in the prepare watcher and would dispose of the check watcher.	3059	in the prepare watcher and would dispose of the check watcher.
1632		3060
1633	Method 3: If the module to be embedded supports explicit event	3061	Method 3: If the module to be embedded supports explicit event
1634	notification (adns does), you can also make use of the actual watcher	3062	notification (libadns does), you can also make use of the actual watcher
1635	callbacks, and only destroy/create the watchers in the prepare watcher.	3063	callbacks, and only destroy/create the watchers in the prepare watcher.
1636		3064
1637	static void	3065	static void
1638	timer_cb (EV_P_ ev_timer *w, int revents)	3066	timer_cb (EV_P_ ev_timer *w, int revents)
1639	{	3067	{
1640	adns_state ads = (adns_state)w->data;	3068	adns_state ads = (adns_state)w->data;
1641	update_now (EV_A);	3069	update_now (EV_A);
1642		3070
1643	adns_processtimeouts (ads, &tv_now);	3071	adns_processtimeouts (ads, &tv_now);
1644	}	3072	}
1645		3073
1646	static void	3074	static void
1647	io_cb (EV_P_ ev_io *w, int revents)	3075	io_cb (EV_P_ ev_io *w, int revents)
1648	{	3076	{
1649	adns_state ads = (adns_state)w->data;	3077	adns_state ads = (adns_state)w->data;
1650	update_now (EV_A);	3078	update_now (EV_A);
1651		3079
1652	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);	3080	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1653	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);	3081	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1654	}	3082	}
1655		3083
1656	// do not ever call adns_afterpoll	3084	// do not ever call adns_afterpoll
1657		3085
1658	Method 4: Do not use a prepare or check watcher because the module you	3086	Method 4: Do not use a prepare or check watcher because the module you
1659	want to embed is too inflexible to support it. Instead, youc na override	3087	want to embed is not flexible enough to support it. Instead, you can
1660	their poll function. The drawback with this solution is that the main	3088	override their poll function. The drawback with this solution is that the
1661	loop is now no longer controllable by EV. The C<Glib::EV> module does	3089	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
1662	this.	3090	this approach, effectively embedding EV as a client into the horrible
		3091	libglib event loop.
1663		3092
1664	static gint	3093	static gint
1665	event_poll_func (GPollFD *fds, guint nfds, gint timeout)	3094	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1666	{	3095	{
1667	int got_events = 0;	3096	int got_events = 0;
1668		3097
1669	for (n = 0; n < nfds; ++n)	3098	for (n = 0; n < nfds; ++n)
1670	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events	3099	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1671		3100
1672	if (timeout >= 0)	3101	if (timeout >= 0)
1673	// create/start timer	3102	// create/start timer
1674		3103
1675	// poll	3104	// poll
1676	ev_loop (EV_A_ 0);	3105	ev_run (EV_A_ 0);
1677		3106
1678	// stop timer again	3107	// stop timer again
1679	if (timeout >= 0)	3108	if (timeout >= 0)
1680	ev_timer_stop (EV_A_ &to);	3109	ev_timer_stop (EV_A_ &to);
1681		3110
1682	// stop io watchers again - their callbacks should have set	3111	// stop io watchers again - their callbacks should have set
1683	for (n = 0; n < nfds; ++n)	3112	for (n = 0; n < nfds; ++n)
1684	ev_io_stop (EV_A_ iow [n]);	3113	ev_io_stop (EV_A_ iow [n]);
1685		3114
1686	return got_events;	3115	return got_events;
1687	}	3116	}
1688		3117
1689		3118
1690	=head2 C<ev_embed> - when one backend isn't enough...	3119	=head2 C<ev_embed> - when one backend isn't enough...
1691		3120
1692	This is a rather advanced watcher type that lets you embed one event loop	3121	This is a rather advanced watcher type that lets you embed one event loop
…		…
1698	prioritise I/O.	3127	prioritise I/O.
1699		3128
1700	As an example for a bug workaround, the kqueue backend might only support	3129	As an example for a bug workaround, the kqueue backend might only support
1701	sockets on some platform, so it is unusable as generic backend, but you	3130	sockets on some platform, so it is unusable as generic backend, but you
1702	still want to make use of it because you have many sockets and it scales	3131	still want to make use of it because you have many sockets and it scales
1703	so nicely. In this case, you would create a kqueue-based loop and embed it	3132	so nicely. In this case, you would create a kqueue-based loop and embed
1704	into your default loop (which might use e.g. poll). Overall operation will	3133	it into your default loop (which might use e.g. poll). Overall operation
1705	be a bit slower because first libev has to poll and then call kevent, but	3134	will be a bit slower because first libev has to call C<poll> and then
1706	at least you can use both at what they are best.	3135	C<kevent>, but at least you can use both mechanisms for what they are
		3136	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
1707		3137
1708	As for prioritising I/O: rarely you have the case where some fds have	3138	As for prioritising I/O: under rare circumstances you have the case where
1709	to be watched and handled very quickly (with low latency), and even	3139	some fds have to be watched and handled very quickly (with low latency),
1710	priorities and idle watchers might have too much overhead. In this case	3140	and even priorities and idle watchers might have too much overhead. In
1711	you would put all the high priority stuff in one loop and all the rest in	3141	this case you would put all the high priority stuff in one loop and all
1712	a second one, and embed the second one in the first.	3142	the rest in a second one, and embed the second one in the first.
1713		3143
1714	As long as the watcher is active, the callback will be invoked every time	3144	As long as the watcher is active, the callback will be invoked every
1715	there might be events pending in the embedded loop. The callback must then	3145	time there might be events pending in the embedded loop. The callback
1716	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	3146	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
1717	their callbacks (you could also start an idle watcher to give the embedded	3147	sweep and invoke their callbacks (the callback doesn't need to invoke the
1718	loop strictly lower priority for example). You can also set the callback	3148	C<ev_embed_sweep> function directly, it could also start an idle watcher
1719	to C<0>, in which case the embed watcher will automatically execute the	3149	to give the embedded loop strictly lower priority for example).
1720	embedded loop sweep.
1721		3150
1722	As long as the watcher is started it will automatically handle events. The	3151	You can also set the callback to C<0>, in which case the embed watcher
1723	callback will be invoked whenever some events have been handled. You can	3152	will automatically execute the embedded loop sweep whenever necessary.
1724	set the callback to C<0> to avoid having to specify one if you are not
1725	interested in that.
1726		3153
1727	Also, there have not currently been made special provisions for forking:	3154	Fork detection will be handled transparently while the C<ev_embed> watcher
1728	when you fork, you not only have to call C<ev_loop_fork> on both loops,	3155	is active, i.e., the embedded loop will automatically be forked when the
1729	but you will also have to stop and restart any C<ev_embed> watchers	3156	embedding loop forks. In other cases, the user is responsible for calling
1730	yourself.	3157	C<ev_loop_fork> on the embedded loop.
1731		3158
1732	Unfortunately, not all backends are embeddable, only the ones returned by	3159	Unfortunately, not all backends are embeddable: only the ones returned by
1733	C<ev_embeddable_backends> are, which, unfortunately, does not include any	3160	C<ev_embeddable_backends> are, which, unfortunately, does not include any
1734	portable one.	3161	portable one.
1735		3162
1736	So when you want to use this feature you will always have to be prepared	3163	So when you want to use this feature you will always have to be prepared
1737	that you cannot get an embeddable loop. The recommended way to get around	3164	that you cannot get an embeddable loop. The recommended way to get around
1738	this is to have a separate variables for your embeddable loop, try to	3165	this is to have a separate variables for your embeddable loop, try to
1739	create it, and if that fails, use the normal loop for everything:	3166	create it, and if that fails, use the normal loop for everything.
1740		3167
1741	struct ev_loop *loop_hi = ev_default_init (0);	3168	=head3 C<ev_embed> and fork
1742	struct ev_loop *loop_lo = 0;
1743	struct ev_embed embed;
1744
1745	// see if there is a chance of getting one that works
1746	// (remember that a flags value of 0 means autodetection)
1747	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1748	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1749	: 0;
1750		3169
1751	// if we got one, then embed it, otherwise default to loop_hi	3170	While the C<ev_embed> watcher is running, forks in the embedding loop will
1752	if (loop_lo)	3171	automatically be applied to the embedded loop as well, so no special
1753	{	3172	fork handling is required in that case. When the watcher is not running,
1754	ev_embed_init (&embed, 0, loop_lo);	3173	however, it is still the task of the libev user to call C<ev_loop_fork ()>
1755	ev_embed_start (loop_hi, &embed);	3174	as applicable.
1756	}
1757	else
1758	loop_lo = loop_hi;
1759		3175
1760	=head3 Watcher-Specific Functions and Data Members	3176	=head3 Watcher-Specific Functions and Data Members
1761		3177
1762	=over 4	3178	=over 4
1763		3179
1764	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	3180	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1765		3181
1766	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)	3182	=item ev_embed_set (ev_embed *, struct ev_loop *embedded_loop)
1767		3183
1768	Configures the watcher to embed the given loop, which must be	3184	Configures the watcher to embed the given loop, which must be
1769	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	3185	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1770	invoked automatically, otherwise it is the responsibility of the callback	3186	invoked automatically, otherwise it is the responsibility of the callback
1771	to invoke it (it will continue to be called until the sweep has been done,	3187	to invoke it (it will continue to be called until the sweep has been done,
1772	if you do not want thta, you need to temporarily stop the embed watcher).	3188	if you do not want that, you need to temporarily stop the embed watcher).
1773		3189
1774	=item ev_embed_sweep (loop, ev_embed *)	3190	=item ev_embed_sweep (loop, ev_embed *)
1775		3191
1776	Make a single, non-blocking sweep over the embedded loop. This works	3192	Make a single, non-blocking sweep over the embedded loop. This works
1777	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	3193	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
1778	apropriate way for embedded loops.	3194	appropriate way for embedded loops.
1779		3195
1780	=item struct ev_loop *loop [read-only]	3196	=item struct ev_loop *other [read-only]
1781		3197
1782	The embedded event loop.	3198	The embedded event loop.
1783		3199
1784	=back	3200	=back
		3201
		3202	=head3 Examples
		3203
		3204	Example: Try to get an embeddable event loop and embed it into the default
		3205	event loop. If that is not possible, use the default loop. The default
		3206	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		3207	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		3208	used).
		3209
		3210	struct ev_loop *loop_hi = ev_default_init (0);
		3211	struct ev_loop *loop_lo = 0;
		3212	ev_embed embed;
		3213
		3214	// see if there is a chance of getting one that works
		3215	// (remember that a flags value of 0 means autodetection)
		3216	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3217	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3218	: 0;
		3219
		3220	// if we got one, then embed it, otherwise default to loop_hi
		3221	if (loop_lo)
		3222	{
		3223	ev_embed_init (&embed, 0, loop_lo);
		3224	ev_embed_start (loop_hi, &embed);
		3225	}
		3226	else
		3227	loop_lo = loop_hi;
		3228
		3229	Example: Check if kqueue is available but not recommended and create
		3230	a kqueue backend for use with sockets (which usually work with any
		3231	kqueue implementation). Store the kqueue/socket-only event loop in
		3232	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3233
		3234	struct ev_loop *loop = ev_default_init (0);
		3235	struct ev_loop *loop_socket = 0;
		3236	ev_embed embed;
		3237
		3238	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3239	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3240	{
		3241	ev_embed_init (&embed, 0, loop_socket);
		3242	ev_embed_start (loop, &embed);
		3243	}
		3244
		3245	if (!loop_socket)
		3246	loop_socket = loop;
		3247
		3248	// now use loop_socket for all sockets, and loop for everything else
1785		3249
1786		3250
1787	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	3251	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1788		3252
1789	Fork watchers are called when a C<fork ()> was detected (usually because	3253	Fork watchers are called when a C<fork ()> was detected (usually because
1790	whoever is a good citizen cared to tell libev about it by calling	3254	whoever is a good citizen cared to tell libev about it by calling
1791	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the	3255	C<ev_loop_fork>). The invocation is done before the event loop blocks next
1792	event loop blocks next and before C<ev_check> watchers are being called,	3256	and before C<ev_check> watchers are being called, and only in the child
1793	and only in the child after the fork. If whoever good citizen calling	3257	after the fork. If whoever good citizen calling C<ev_default_fork> cheats
1794	C<ev_default_fork> cheats and calls it in the wrong process, the fork	3258	and calls it in the wrong process, the fork handlers will be invoked, too,
1795	handlers will be invoked, too, of course.	3259	of course.
		3260
		3261	=head3 The special problem of life after fork - how is it possible?
		3262
		3263	Most uses of C<fork()> consist of forking, then some simple calls to set
		3264	up/change the process environment, followed by a call to C<exec()>. This
		3265	sequence should be handled by libev without any problems.
		3266
		3267	This changes when the application actually wants to do event handling
		3268	in the child, or both parent in child, in effect "continuing" after the
		3269	fork.
		3270
		3271	The default mode of operation (for libev, with application help to detect
		3272	forks) is to duplicate all the state in the child, as would be expected
		3273	when I<either> the parent I<or> the child process continues.
		3274
		3275	When both processes want to continue using libev, then this is usually the
		3276	wrong result. In that case, usually one process (typically the parent) is
		3277	supposed to continue with all watchers in place as before, while the other
		3278	process typically wants to start fresh, i.e. without any active watchers.
		3279
		3280	The cleanest and most efficient way to achieve that with libev is to
		3281	simply create a new event loop, which of course will be "empty", and
		3282	use that for new watchers. This has the advantage of not touching more
		3283	memory than necessary, and thus avoiding the copy-on-write, and the
		3284	disadvantage of having to use multiple event loops (which do not support
		3285	signal watchers).
		3286
		3287	When this is not possible, or you want to use the default loop for
		3288	other reasons, then in the process that wants to start "fresh", call
		3289	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3290	Destroying the default loop will "orphan" (not stop) all registered
		3291	watchers, so you have to be careful not to execute code that modifies
		3292	those watchers. Note also that in that case, you have to re-register any
		3293	signal watchers.
1796		3294
1797	=head3 Watcher-Specific Functions and Data Members	3295	=head3 Watcher-Specific Functions and Data Members
1798		3296
1799	=over 4	3297	=over 4
1800		3298
1801	=item ev_fork_init (ev_signal *, callback)	3299	=item ev_fork_init (ev_fork *, callback)
1802		3300
1803	Initialises and configures the fork watcher - it has no parameters of any	3301	Initialises and configures the fork watcher - it has no parameters of any
1804	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,	3302	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
1805	believe me.	3303	really.
1806		3304
1807	=back	3305	=back
1808		3306
1809		3307
		3308	=head2 C<ev_cleanup> - even the best things end
		3309
		3310	Cleanup watchers are called just before the event loop is being destroyed
		3311	by a call to C<ev_loop_destroy>.
		3312
		3313	While there is no guarantee that the event loop gets destroyed, cleanup
		3314	watchers provide a convenient method to install cleanup hooks for your
		3315	program, worker threads and so on - you just to make sure to destroy the
		3316	loop when you want them to be invoked.
		3317
		3318	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3319	all other watchers, they do not keep a reference to the event loop (which
		3320	makes a lot of sense if you think about it). Like all other watchers, you
		3321	can call libev functions in the callback, except C<ev_cleanup_start>.
		3322
		3323	=head3 Watcher-Specific Functions and Data Members
		3324
		3325	=over 4
		3326
		3327	=item ev_cleanup_init (ev_cleanup *, callback)
		3328
		3329	Initialises and configures the cleanup watcher - it has no parameters of
		3330	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3331	pointless, I assure you.
		3332
		3333	=back
		3334
		3335	Example: Register an atexit handler to destroy the default loop, so any
		3336	cleanup functions are called.
		3337
		3338	static void
		3339	program_exits (void)
		3340	{
		3341	ev_loop_destroy (EV_DEFAULT_UC);
		3342	}
		3343
		3344	...
		3345	atexit (program_exits);
		3346
		3347
		3348	=head2 C<ev_async> - how to wake up an event loop
		3349
		3350	In general, you cannot use an C<ev_loop> from multiple threads or other
		3351	asynchronous sources such as signal handlers (as opposed to multiple event
		3352	loops - those are of course safe to use in different threads).
		3353
		3354	Sometimes, however, you need to wake up an event loop you do not control,
		3355	for example because it belongs to another thread. This is what C<ev_async>
		3356	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3357	it by calling C<ev_async_send>, which is thread- and signal safe.
		3358
		3359	This functionality is very similar to C<ev_signal> watchers, as signals,
		3360	too, are asynchronous in nature, and signals, too, will be compressed
		3361	(i.e. the number of callback invocations may be less than the number of
		3362	C<ev_async_send> calls). In fact, you could use signal watchers as a kind
		3363	of "global async watchers" by using a watcher on an otherwise unused
		3364	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3365	even without knowing which loop owns the signal.
		3366
		3367	=head3 Queueing
		3368
		3369	C<ev_async> does not support queueing of data in any way. The reason
		3370	is that the author does not know of a simple (or any) algorithm for a
		3371	multiple-writer-single-reader queue that works in all cases and doesn't
		3372	need elaborate support such as pthreads or unportable memory access
		3373	semantics.
		3374
		3375	That means that if you want to queue data, you have to provide your own
		3376	queue. But at least I can tell you how to implement locking around your
		3377	queue:
		3378
		3379	=over 4
		3380
		3381	=item queueing from a signal handler context
		3382
		3383	To implement race-free queueing, you simply add to the queue in the signal
		3384	handler but you block the signal handler in the watcher callback. Here is
		3385	an example that does that for some fictitious SIGUSR1 handler:
		3386
		3387	static ev_async mysig;
		3388
		3389	static void
		3390	sigusr1_handler (void)
		3391	{
		3392	sometype data;
		3393
		3394	// no locking etc.
		3395	queue_put (data);
		3396	ev_async_send (EV_DEFAULT_ &mysig);
		3397	}
		3398
		3399	static void
		3400	mysig_cb (EV_P_ ev_async *w, int revents)
		3401	{
		3402	sometype data;
		3403	sigset_t block, prev;
		3404
		3405	sigemptyset (&block);
		3406	sigaddset (&block, SIGUSR1);
		3407	sigprocmask (SIG_BLOCK, &block, &prev);
		3408
		3409	while (queue_get (&data))
		3410	process (data);
		3411
		3412	if (sigismember (&prev, SIGUSR1)
		3413	sigprocmask (SIG_UNBLOCK, &block, 0);
		3414	}
		3415
		3416	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3417	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3418	either...).
		3419
		3420	=item queueing from a thread context
		3421
		3422	The strategy for threads is different, as you cannot (easily) block
		3423	threads but you can easily preempt them, so to queue safely you need to
		3424	employ a traditional mutex lock, such as in this pthread example:
		3425
		3426	static ev_async mysig;
		3427	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3428
		3429	static void
		3430	otherthread (void)
		3431	{
		3432	// only need to lock the actual queueing operation
		3433	pthread_mutex_lock (&mymutex);
		3434	queue_put (data);
		3435	pthread_mutex_unlock (&mymutex);
		3436
		3437	ev_async_send (EV_DEFAULT_ &mysig);
		3438	}
		3439
		3440	static void
		3441	mysig_cb (EV_P_ ev_async *w, int revents)
		3442	{
		3443	pthread_mutex_lock (&mymutex);
		3444
		3445	while (queue_get (&data))
		3446	process (data);
		3447
		3448	pthread_mutex_unlock (&mymutex);
		3449	}
		3450
		3451	=back
		3452
		3453
		3454	=head3 Watcher-Specific Functions and Data Members
		3455
		3456	=over 4
		3457
		3458	=item ev_async_init (ev_async *, callback)
		3459
		3460	Initialises and configures the async watcher - it has no parameters of any
		3461	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3462	trust me.
		3463
		3464	=item ev_async_send (loop, ev_async *)
		3465
		3466	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3467	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3468	returns.
		3469
		3470	Unlike C<ev_feed_event>, this call is safe to do from other threads,
		3471	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
		3472	embedding section below on what exactly this means).
		3473
		3474	Note that, as with other watchers in libev, multiple events might get
		3475	compressed into a single callback invocation (another way to look at
		3476	this is that C<ev_async> watchers are level-triggered: they are set on
		3477	C<ev_async_send>, reset when the event loop detects that).
		3478
		3479	This call incurs the overhead of at most one extra system call per event
		3480	loop iteration, if the event loop is blocked, and no syscall at all if
		3481	the event loop (or your program) is processing events. That means that
		3482	repeated calls are basically free (there is no need to avoid calls for
		3483	performance reasons) and that the overhead becomes smaller (typically
		3484	zero) under load.
		3485
		3486	=item bool = ev_async_pending (ev_async *)
		3487
		3488	Returns a non-zero value when C<ev_async_send> has been called on the
		3489	watcher but the event has not yet been processed (or even noted) by the
		3490	event loop.
		3491
		3492	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3493	the loop iterates next and checks for the watcher to have become active,
		3494	it will reset the flag again. C<ev_async_pending> can be used to very
		3495	quickly check whether invoking the loop might be a good idea.
		3496
		3497	Not that this does I<not> check whether the watcher itself is pending,
		3498	only whether it has been requested to make this watcher pending: there
		3499	is a time window between the event loop checking and resetting the async
		3500	notification, and the callback being invoked.
		3501
		3502	=back
		3503
		3504
1810	=head1 OTHER FUNCTIONS	3505	=head1 OTHER FUNCTIONS
1811		3506
1812	There are some other functions of possible interest. Described. Here. Now.	3507	There are some other functions of possible interest. Described. Here. Now.
1813		3508
1814	=over 4	3509	=over 4
1815		3510
1816	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3511	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
1817		3512
1818	This function combines a simple timer and an I/O watcher, calls your	3513	This function combines a simple timer and an I/O watcher, calls your
1819	callback on whichever event happens first and automatically stop both	3514	callback on whichever event happens first and automatically stops both
1820	watchers. This is useful if you want to wait for a single event on an fd	3515	watchers. This is useful if you want to wait for a single event on an fd
1821	or timeout without having to allocate/configure/start/stop/free one or	3516	or timeout without having to allocate/configure/start/stop/free one or
1822	more watchers yourself.	3517	more watchers yourself.
1823		3518
1824	If C<fd> is less than 0, then no I/O watcher will be started and events	3519	If C<fd> is less than 0, then no I/O watcher will be started and the
1825	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3520	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
1826	C<events> set will be craeted and started.	3521	the given C<fd> and C<events> set will be created and started.
1827		3522
1828	If C<timeout> is less than 0, then no timeout watcher will be	3523	If C<timeout> is less than 0, then no timeout watcher will be
1829	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3524	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
1830	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3525	repeat = 0) will be started. C<0> is a valid timeout.
1831	dubious value.
1832		3526
1833	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3527	The callback has the type C<void (*cb)(int revents, void *arg)> and is
1834	passed an C<revents> set like normal event callbacks (a combination of	3528	passed an C<revents> set like normal event callbacks (a combination of
1835	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3529	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
1836	value passed to C<ev_once>:	3530	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3531	a timeout and an io event at the same time - you probably should give io
		3532	events precedence.
1837		3533
		3534	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3535
1838	static void stdin_ready (int revents, void *arg)	3536	static void stdin_ready (int revents, void *arg)
		3537	{
		3538	if (revents & EV_READ)
		3539	/* stdin might have data for us, joy! */;
		3540	else if (revents & EV_TIMER)
		3541	/* doh, nothing entered */;
		3542	}
		3543
		3544	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
		3545
		3546	=item ev_feed_fd_event (loop, int fd, int revents)
		3547
		3548	Feed an event on the given fd, as if a file descriptor backend detected
		3549	the given events.
		3550
		3551	=item ev_feed_signal_event (loop, int signum)
		3552
		3553	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
		3554	which is async-safe.
		3555
		3556	=back
		3557
		3558
		3559	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3560
		3561	This section explains some common idioms that are not immediately
		3562	obvious. Note that examples are sprinkled over the whole manual, and this
		3563	section only contains stuff that wouldn't fit anywhere else.
		3564
		3565	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3566
		3567	Each watcher has, by default, a C<void *data> member that you can read
		3568	or modify at any time: libev will completely ignore it. This can be used
		3569	to associate arbitrary data with your watcher. If you need more data and
		3570	don't want to allocate memory separately and store a pointer to it in that
		3571	data member, you can also "subclass" the watcher type and provide your own
		3572	data:
		3573
		3574	struct my_io
		3575	{
		3576	ev_io io;
		3577	int otherfd;
		3578	void *somedata;
		3579	struct whatever *mostinteresting;
		3580	};
		3581
		3582	...
		3583	struct my_io w;
		3584	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3585
		3586	And since your callback will be called with a pointer to the watcher, you
		3587	can cast it back to your own type:
		3588
		3589	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3590	{
		3591	struct my_io *w = (struct my_io *)w_;
		3592	...
		3593	}
		3594
		3595	More interesting and less C-conformant ways of casting your callback
		3596	function type instead have been omitted.
		3597
		3598	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3599
		3600	Another common scenario is to use some data structure with multiple
		3601	embedded watchers, in effect creating your own watcher that combines
		3602	multiple libev event sources into one "super-watcher":
		3603
		3604	struct my_biggy
		3605	{
		3606	int some_data;
		3607	ev_timer t1;
		3608	ev_timer t2;
		3609	}
		3610
		3611	In this case getting the pointer to C<my_biggy> is a bit more
		3612	complicated: Either you store the address of your C<my_biggy> struct in
		3613	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3614	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3615	real programmers):
		3616
		3617	#include <stddef.h>
		3618
		3619	static void
		3620	t1_cb (EV_P_ ev_timer *w, int revents)
		3621	{
		3622	struct my_biggy big = (struct my_biggy *)
		3623	(((char *)w) - offsetof (struct my_biggy, t1));
		3624	}
		3625
		3626	static void
		3627	t2_cb (EV_P_ ev_timer *w, int revents)
		3628	{
		3629	struct my_biggy big = (struct my_biggy *)
		3630	(((char *)w) - offsetof (struct my_biggy, t2));
		3631	}
		3632
		3633	=head2 AVOIDING FINISHING BEFORE RETURNING
		3634
		3635	Often you have structures like this in event-based programs:
		3636
		3637	callback ()
1839	{	3638	{
1840	if (revents & EV_TIMEOUT)	3639	free (request);
1841	/* doh, nothing entered */;
1842	else if (revents & EV_READ)
1843	/* stdin might have data for us, joy! */;
1844	}	3640	}
1845		3641
1846	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3642	request = start_new_request (..., callback);
1847		3643
1848	=item ev_feed_event (ev_loop *, watcher *, int revents)	3644	The intent is to start some "lengthy" operation. The C<request> could be
		3645	used to cancel the operation, or do other things with it.
1849		3646
1850	Feeds the given event set into the event loop, as if the specified event	3647	It's not uncommon to have code paths in C<start_new_request> that
1851	had happened for the specified watcher (which must be a pointer to an	3648	immediately invoke the callback, for example, to report errors. Or you add
1852	initialised but not necessarily started event watcher).	3649	some caching layer that finds that it can skip the lengthy aspects of the
		3650	operation and simply invoke the callback with the result.
1853		3651
1854	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	3652	The problem here is that this will happen I<before> C<start_new_request>
		3653	has returned, so C<request> is not set.
1855		3654
1856	Feed an event on the given fd, as if a file descriptor backend detected	3655	Even if you pass the request by some safer means to the callback, you
1857	the given events it.	3656	might want to do something to the request after starting it, such as
		3657	canceling it, which probably isn't working so well when the callback has
		3658	already been invoked.
1858		3659
1859	=item ev_feed_signal_event (ev_loop *loop, int signum)	3660	A common way around all these issues is to make sure that
		3661	C<start_new_request> I<always> returns before the callback is invoked. If
		3662	C<start_new_request> immediately knows the result, it can artificially
		3663	delay invoking the callback by using a C<prepare> or C<idle> watcher for
		3664	example, or more sneakily, by reusing an existing (stopped) watcher and
		3665	pushing it into the pending queue:
1860		3666
1861	Feed an event as if the given signal occured (C<loop> must be the default	3667	ev_set_cb (watcher, callback);
1862	loop!).	3668	ev_feed_event (EV_A_ watcher, 0);
1863		3669
1864	=back	3670	This way, C<start_new_request> can safely return before the callback is
		3671	invoked, while not delaying callback invocation too much.
		3672
		3673	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3674
		3675	Often (especially in GUI toolkits) there are places where you have
		3676	I<modal> interaction, which is most easily implemented by recursively
		3677	invoking C<ev_run>.
		3678
		3679	This brings the problem of exiting - a callback might want to finish the
		3680	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3681	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3682	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3683	other combination: In these cases, a simple C<ev_break> will not work.
		3684
		3685	The solution is to maintain "break this loop" variable for each C<ev_run>
		3686	invocation, and use a loop around C<ev_run> until the condition is
		3687	triggered, using C<EVRUN_ONCE>:
		3688
		3689	// main loop
		3690	int exit_main_loop = 0;
		3691
		3692	while (!exit_main_loop)
		3693	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3694
		3695	// in a modal watcher
		3696	int exit_nested_loop = 0;
		3697
		3698	while (!exit_nested_loop)
		3699	ev_run (EV_A_ EVRUN_ONCE);
		3700
		3701	To exit from any of these loops, just set the corresponding exit variable:
		3702
		3703	// exit modal loop
		3704	exit_nested_loop = 1;
		3705
		3706	// exit main program, after modal loop is finished
		3707	exit_main_loop = 1;
		3708
		3709	// exit both
		3710	exit_main_loop = exit_nested_loop = 1;
		3711
		3712	=head2 THREAD LOCKING EXAMPLE
		3713
		3714	Here is a fictitious example of how to run an event loop in a different
		3715	thread from where callbacks are being invoked and watchers are
		3716	created/added/removed.
		3717
		3718	For a real-world example, see the C<EV::Loop::Async> perl module,
		3719	which uses exactly this technique (which is suited for many high-level
		3720	languages).
		3721
		3722	The example uses a pthread mutex to protect the loop data, a condition
		3723	variable to wait for callback invocations, an async watcher to notify the
		3724	event loop thread and an unspecified mechanism to wake up the main thread.
		3725
		3726	First, you need to associate some data with the event loop:
		3727
		3728	typedef struct {
		3729	mutex_t lock; /* global loop lock */
		3730	ev_async async_w;
		3731	thread_t tid;
		3732	cond_t invoke_cv;
		3733	} userdata;
		3734
		3735	void prepare_loop (EV_P)
		3736	{
		3737	// for simplicity, we use a static userdata struct.
		3738	static userdata u;
		3739
		3740	ev_async_init (&u->async_w, async_cb);
		3741	ev_async_start (EV_A_ &u->async_w);
		3742
		3743	pthread_mutex_init (&u->lock, 0);
		3744	pthread_cond_init (&u->invoke_cv, 0);
		3745
		3746	// now associate this with the loop
		3747	ev_set_userdata (EV_A_ u);
		3748	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3749	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3750
		3751	// then create the thread running ev_run
		3752	pthread_create (&u->tid, 0, l_run, EV_A);
		3753	}
		3754
		3755	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3756	solely to wake up the event loop so it takes notice of any new watchers
		3757	that might have been added:
		3758
		3759	static void
		3760	async_cb (EV_P_ ev_async *w, int revents)
		3761	{
		3762	// just used for the side effects
		3763	}
		3764
		3765	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3766	protecting the loop data, respectively.
		3767
		3768	static void
		3769	l_release (EV_P)
		3770	{
		3771	userdata *u = ev_userdata (EV_A);
		3772	pthread_mutex_unlock (&u->lock);
		3773	}
		3774
		3775	static void
		3776	l_acquire (EV_P)
		3777	{
		3778	userdata *u = ev_userdata (EV_A);
		3779	pthread_mutex_lock (&u->lock);
		3780	}
		3781
		3782	The event loop thread first acquires the mutex, and then jumps straight
		3783	into C<ev_run>:
		3784
		3785	void *
		3786	l_run (void *thr_arg)
		3787	{
		3788	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3789
		3790	l_acquire (EV_A);
		3791	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3792	ev_run (EV_A_ 0);
		3793	l_release (EV_A);
		3794
		3795	return 0;
		3796	}
		3797
		3798	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3799	signal the main thread via some unspecified mechanism (signals? pipe
		3800	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3801	have been called (in a while loop because a) spurious wakeups are possible
		3802	and b) skipping inter-thread-communication when there are no pending
		3803	watchers is very beneficial):
		3804
		3805	static void
		3806	l_invoke (EV_P)
		3807	{
		3808	userdata *u = ev_userdata (EV_A);
		3809
		3810	while (ev_pending_count (EV_A))
		3811	{
		3812	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3813	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3814	}
		3815	}
		3816
		3817	Now, whenever the main thread gets told to invoke pending watchers, it
		3818	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3819	thread to continue:
		3820
		3821	static void
		3822	real_invoke_pending (EV_P)
		3823	{
		3824	userdata *u = ev_userdata (EV_A);
		3825
		3826	pthread_mutex_lock (&u->lock);
		3827	ev_invoke_pending (EV_A);
		3828	pthread_cond_signal (&u->invoke_cv);
		3829	pthread_mutex_unlock (&u->lock);
		3830	}
		3831
		3832	Whenever you want to start/stop a watcher or do other modifications to an
		3833	event loop, you will now have to lock:
		3834
		3835	ev_timer timeout_watcher;
		3836	userdata *u = ev_userdata (EV_A);
		3837
		3838	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3839
		3840	pthread_mutex_lock (&u->lock);
		3841	ev_timer_start (EV_A_ &timeout_watcher);
		3842	ev_async_send (EV_A_ &u->async_w);
		3843	pthread_mutex_unlock (&u->lock);
		3844
		3845	Note that sending the C<ev_async> watcher is required because otherwise
		3846	an event loop currently blocking in the kernel will have no knowledge
		3847	about the newly added timer. By waking up the loop it will pick up any new
		3848	watchers in the next event loop iteration.
		3849
		3850	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3851
		3852	While the overhead of a callback that e.g. schedules a thread is small, it
		3853	is still an overhead. If you embed libev, and your main usage is with some
		3854	kind of threads or coroutines, you might want to customise libev so that
		3855	doesn't need callbacks anymore.
		3856
		3857	Imagine you have coroutines that you can switch to using a function
		3858	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3859	and that due to some magic, the currently active coroutine is stored in a
		3860	global called C<current_coro>. Then you can build your own "wait for libev
		3861	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3862	the differing C<;> conventions):
		3863
		3864	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3865	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3866
		3867	That means instead of having a C callback function, you store the
		3868	coroutine to switch to in each watcher, and instead of having libev call
		3869	your callback, you instead have it switch to that coroutine.
		3870
		3871	A coroutine might now wait for an event with a function called
		3872	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		3873	matter when, or whether the watcher is active or not when this function is
		3874	called):
		3875
		3876	void
		3877	wait_for_event (ev_watcher *w)
		3878	{
		3879	ev_set_cb (w, current_coro);
		3880	switch_to (libev_coro);
		3881	}
		3882
		3883	That basically suspends the coroutine inside C<wait_for_event> and
		3884	continues the libev coroutine, which, when appropriate, switches back to
		3885	this or any other coroutine.
		3886
		3887	You can do similar tricks if you have, say, threads with an event queue -
		3888	instead of storing a coroutine, you store the queue object and instead of
		3889	switching to a coroutine, you push the watcher onto the queue and notify
		3890	any waiters.
		3891
		3892	To embed libev, see L</EMBEDDING>, but in short, it's easiest to create two
		3893	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		3894
		3895	// my_ev.h
		3896	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3897	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb);
		3898	#include "../libev/ev.h"
		3899
		3900	// my_ev.c
		3901	#define EV_H "my_ev.h"
		3902	#include "../libev/ev.c"
		3903
		3904	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		3905	F<my_ev.c> into your project. When properly specifying include paths, you
		3906	can even use F<ev.h> as header file name directly.
1865		3907
1866		3908
1867	=head1 LIBEVENT EMULATION	3909	=head1 LIBEVENT EMULATION
1868		3910
1869	Libev offers a compatibility emulation layer for libevent. It cannot	3911	Libev offers a compatibility emulation layer for libevent. It cannot
1870	emulate the internals of libevent, so here are some usage hints:	3912	emulate the internals of libevent, so here are some usage hints:
1871		3913
1872	=over 4	3914	=over 4
		3915
		3916	=item * Only the libevent-1.4.1-beta API is being emulated.
		3917
		3918	This was the newest libevent version available when libev was implemented,
		3919	and is still mostly unchanged in 2010.
1873		3920
1874	=item * Use it by including <event.h>, as usual.	3921	=item * Use it by including <event.h>, as usual.
1875		3922
1876	=item * The following members are fully supported: ev_base, ev_callback,	3923	=item * The following members are fully supported: ev_base, ev_callback,
1877	ev_arg, ev_fd, ev_res, ev_events.	3924	ev_arg, ev_fd, ev_res, ev_events.
…		…
1882		3929
1883	=item * Priorities are not currently supported. Initialising priorities	3930	=item * Priorities are not currently supported. Initialising priorities
1884	will fail and all watchers will have the same priority, even though there	3931	will fail and all watchers will have the same priority, even though there
1885	is an ev_pri field.	3932	is an ev_pri field.
1886		3933
		3934	=item * In libevent, the last base created gets the signals, in libev, the
		3935	base that registered the signal gets the signals.
		3936
1887	=item * Other members are not supported.	3937	=item * Other members are not supported.
1888		3938
1889	=item * The libev emulation is I<not> ABI compatible to libevent, you need	3939	=item * The libev emulation is I<not> ABI compatible to libevent, you need
1890	to use the libev header file and library.	3940	to use the libev header file and library.
1891		3941
1892	=back	3942	=back
1893		3943
1894	=head1 C++ SUPPORT	3944	=head1 C++ SUPPORT
1895		3945
		3946	=head2 C API
		3947
		3948	The normal C API should work fine when used from C++: both ev.h and the
		3949	libev sources can be compiled as C++. Therefore, code that uses the C API
		3950	will work fine.
		3951
		3952	Proper exception specifications might have to be added to callbacks passed
		3953	to libev: exceptions may be thrown only from watcher callbacks, all
		3954	other callbacks (allocator, syserr, loop acquire/release and periodic
		3955	reschedule callbacks) must not throw exceptions, and might need a C<throw
		3956	()> specification. If you have code that needs to be compiled as both C
		3957	and C++ you can use the C<EV_THROW> macro for this:
		3958
		3959	static void
		3960	fatal_error (const char *msg) EV_THROW
		3961	{
		3962	perror (msg);
		3963	abort ();
		3964	}
		3965
		3966	...
		3967	ev_set_syserr_cb (fatal_error);
		3968
		3969	The only API functions that can currently throw exceptions are C<ev_run>,
		3970	C<ev_invoke>, C<ev_invoke_pending> and C<ev_loop_destroy> (the latter
		3971	because it runs cleanup watchers).
		3972
		3973	Throwing exceptions in watcher callbacks is only supported if libev itself
		3974	is compiled with a C++ compiler or your C and C++ environments allow
		3975	throwing exceptions through C libraries (most do).
		3976
		3977	=head2 C++ API
		3978
1896	Libev comes with some simplistic wrapper classes for C++ that mainly allow	3979	Libev comes with some simplistic wrapper classes for C++ that mainly allow
1897	you to use some convinience methods to start/stop watchers and also change	3980	you to use some convenience methods to start/stop watchers and also change
1898	the callback model to a model using method callbacks on objects.	3981	the callback model to a model using method callbacks on objects.
1899		3982
1900	To use it,	3983	To use it,
1901		3984
1902	#include <ev++.h>	3985	#include <ev++.h>
1903		3986
1904	This automatically includes F<ev.h> and puts all of its definitions (many	3987	This automatically includes F<ev.h> and puts all of its definitions (many
1905	of them macros) into the global namespace. All C++ specific things are	3988	of them macros) into the global namespace. All C++ specific things are
1906	put into the C<ev> namespace. It should support all the same embedding	3989	put into the C<ev> namespace. It should support all the same embedding
1907	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.	3990	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
…		…
1909	Care has been taken to keep the overhead low. The only data member the C++	3992	Care has been taken to keep the overhead low. The only data member the C++
1910	classes add (compared to plain C-style watchers) is the event loop pointer	3993	classes add (compared to plain C-style watchers) is the event loop pointer
1911	that the watcher is associated with (or no additional members at all if	3994	that the watcher is associated with (or no additional members at all if
1912	you disable C<EV_MULTIPLICITY> when embedding libev).	3995	you disable C<EV_MULTIPLICITY> when embedding libev).
1913		3996
1914	Currently, functions, and static and non-static member functions can be	3997	Currently, functions, static and non-static member functions and classes
1915	used as callbacks. Other types should be easy to add as long as they only	3998	with C<operator ()> can be used as callbacks. Other types should be easy
1916	need one additional pointer for context. If you need support for other	3999	to add as long as they only need one additional pointer for context. If
1917	types of functors please contact the author (preferably after implementing	4000	you need support for other types of functors please contact the author
1918	it).	4001	(preferably after implementing it).
		4002
		4003	For all this to work, your C++ compiler either has to use the same calling
		4004	conventions as your C compiler (for static member functions), or you have
		4005	to embed libev and compile libev itself as C++.
1919		4006
1920	Here is a list of things available in the C<ev> namespace:	4007	Here is a list of things available in the C<ev> namespace:
1921		4008
1922	=over 4	4009	=over 4
1923		4010
…		…
1933	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.	4020	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
1934		4021
1935	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of	4022	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
1936	the same name in the C<ev> namespace, with the exception of C<ev_signal>	4023	the same name in the C<ev> namespace, with the exception of C<ev_signal>
1937	which is called C<ev::sig> to avoid clashes with the C<signal> macro	4024	which is called C<ev::sig> to avoid clashes with the C<signal> macro
1938	defines by many implementations.	4025	defined by many implementations.
1939		4026
1940	All of those classes have these methods:	4027	All of those classes have these methods:
1941		4028
1942	=over 4	4029	=over 4
1943		4030
1944	=item ev::TYPE::TYPE ()	4031	=item ev::TYPE::TYPE ()
1945		4032
1946	=item ev::TYPE::TYPE (struct ev_loop *)	4033	=item ev::TYPE::TYPE (loop)
1947		4034
1948	=item ev::TYPE::~TYPE	4035	=item ev::TYPE::~TYPE
1949		4036
1950	The constructor (optionally) takes an event loop to associate the watcher	4037	The constructor (optionally) takes an event loop to associate the watcher
1951	with. If it is omitted, it will use C<EV_DEFAULT>.	4038	with. If it is omitted, it will use C<EV_DEFAULT>.
…		…
1974	your compiler is good :), then the method will be fully inlined into the	4061	your compiler is good :), then the method will be fully inlined into the
1975	thunking function, making it as fast as a direct C callback.	4062	thunking function, making it as fast as a direct C callback.
1976		4063
1977	Example: simple class declaration and watcher initialisation	4064	Example: simple class declaration and watcher initialisation
1978		4065
1979	struct myclass	4066	struct myclass
1980	{	4067	{
1981	void io_cb (ev::io &w, int revents) { }	4068	void io_cb (ev::io &w, int revents) { }
1982	}	4069	}
1983		4070
1984	myclass obj;	4071	myclass obj;
1985	ev::io iow;	4072	ev::io iow;
1986	iow.set <myclass, &myclass::io_cb> (&obj);	4073	iow.set <myclass, &myclass::io_cb> (&obj);
		4074
		4075	=item w->set (object *)
		4076
		4077	This is a variation of a method callback - leaving out the method to call
		4078	will default the method to C<operator ()>, which makes it possible to use
		4079	functor objects without having to manually specify the C<operator ()> all
		4080	the time. Incidentally, you can then also leave out the template argument
		4081	list.
		4082
		4083	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		4084	int revents)>.
		4085
		4086	See the method-C<set> above for more details.
		4087
		4088	Example: use a functor object as callback.
		4089
		4090	struct myfunctor
		4091	{
		4092	void operator() (ev::io &w, int revents)
		4093	{
		4094	...
		4095	}
		4096	}
		4097
		4098	myfunctor f;
		4099
		4100	ev::io w;
		4101	w.set (&f);
1987		4102
1988	=item w->set<function> (void *data = 0)	4103	=item w->set<function> (void *data = 0)
1989		4104
1990	Also sets a callback, but uses a static method or plain function as	4105	Also sets a callback, but uses a static method or plain function as
1991	callback. The optional C<data> argument will be stored in the watcher's	4106	callback. The optional C<data> argument will be stored in the watcher's
…		…
1993		4108
1994	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.	4109	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
1995		4110
1996	See the method-C<set> above for more details.	4111	See the method-C<set> above for more details.
1997		4112
1998	Example:	4113	Example: Use a plain function as callback.
1999		4114
2000	static void io_cb (ev::io &w, int revents) { }	4115	static void io_cb (ev::io &w, int revents) { }
2001	iow.set <io_cb> ();	4116	iow.set <io_cb> ();
2002		4117
2003	=item w->set (struct ev_loop *)	4118	=item w->set (loop)
2004		4119
2005	Associates a different C<struct ev_loop> with this watcher. You can only	4120	Associates a different C<struct ev_loop> with this watcher. You can only
2006	do this when the watcher is inactive (and not pending either).	4121	do this when the watcher is inactive (and not pending either).
2007		4122
2008	=item w->set ([args])	4123	=item w->set ([arguments])
2009		4124
2010	Basically the same as C<ev_TYPE_set>, with the same args. Must be	4125	Basically the same as C<ev_TYPE_set> (except for C<ev::embed> watchers>),
		4126	with the same arguments. Either this method or a suitable start method
2011	called at least once. Unlike the C counterpart, an active watcher gets	4127	must be called at least once. Unlike the C counterpart, an active watcher
2012	automatically stopped and restarted when reconfiguring it with this	4128	gets automatically stopped and restarted when reconfiguring it with this
2013	method.	4129	method.
		4130
		4131	For C<ev::embed> watchers this method is called C<set_embed>, to avoid
		4132	clashing with the C<set (loop)> method.
2014		4133
2015	=item w->start ()	4134	=item w->start ()
2016		4135
2017	Starts the watcher. Note that there is no C<loop> argument, as the	4136	Starts the watcher. Note that there is no C<loop> argument, as the
2018	constructor already stores the event loop.	4137	constructor already stores the event loop.
2019		4138
		4139	=item w->start ([arguments])
		4140
		4141	Instead of calling C<set> and C<start> methods separately, it is often
		4142	convenient to wrap them in one call. Uses the same type of arguments as
		4143	the configure C<set> method of the watcher.
		4144
2020	=item w->stop ()	4145	=item w->stop ()
2021		4146
2022	Stops the watcher if it is active. Again, no C<loop> argument.	4147	Stops the watcher if it is active. Again, no C<loop> argument.
2023		4148
2024	=item w->again () (C<ev::timer>, C<ev::periodic> only)	4149	=item w->again () (C<ev::timer>, C<ev::periodic> only)
…		…
2036		4161
2037	=back	4162	=back
2038		4163
2039	=back	4164	=back
2040		4165
2041	Example: Define a class with an IO and idle watcher, start one of them in	4166	Example: Define a class with two I/O and idle watchers, start the I/O
2042	the constructor.	4167	watchers in the constructor.
2043		4168
2044	class myclass	4169	class myclass
2045	{	4170	{
2046	ev_io io; void io_cb (ev::io &w, int revents);	4171	ev::io io ; void io_cb (ev::io &w, int revents);
		4172	ev::io io2 ; void io2_cb (ev::io &w, int revents);
2047	ev_idle idle void idle_cb (ev::idle &w, int revents);	4173	ev::idle idle; void idle_cb (ev::idle &w, int revents);
2048		4174
2049	myclass ();	4175	myclass (int fd)
2050	}	4176	{
2051
2052	myclass::myclass (int fd)
2053	{
2054	io .set <myclass, &myclass::io_cb > (this);	4177	io .set <myclass, &myclass::io_cb > (this);
		4178	io2 .set <myclass, &myclass::io2_cb > (this);
2055	idle.set <myclass, &myclass::idle_cb> (this);	4179	idle.set <myclass, &myclass::idle_cb> (this);
2056		4180
2057	io.start (fd, ev::READ);	4181	io.set (fd, ev::WRITE); // configure the watcher
		4182	io.start (); // start it whenever convenient
		4183
		4184	io2.start (fd, ev::READ); // set + start in one call
		4185	}
2058	}	4186	};
		4187
		4188
		4189	=head1 OTHER LANGUAGE BINDINGS
		4190
		4191	Libev does not offer other language bindings itself, but bindings for a
		4192	number of languages exist in the form of third-party packages. If you know
		4193	any interesting language binding in addition to the ones listed here, drop
		4194	me a note.
		4195
		4196	=over 4
		4197
		4198	=item Perl
		4199
		4200	The EV module implements the full libev API and is actually used to test
		4201	libev. EV is developed together with libev. Apart from the EV core module,
		4202	there are additional modules that implement libev-compatible interfaces
		4203	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		4204	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		4205	and C<EV::Glib>).
		4206
		4207	It can be found and installed via CPAN, its homepage is at
		4208	L<http://software.schmorp.de/pkg/EV>.
		4209
		4210	=item Python
		4211
		4212	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		4213	seems to be quite complete and well-documented.
		4214
		4215	=item Ruby
		4216
		4217	Tony Arcieri has written a ruby extension that offers access to a subset
		4218	of the libev API and adds file handle abstractions, asynchronous DNS and
		4219	more on top of it. It can be found via gem servers. Its homepage is at
		4220	L<http://rev.rubyforge.org/>.
		4221
		4222	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		4223	makes rev work even on mingw.
		4224
		4225	=item Haskell
		4226
		4227	A haskell binding to libev is available at
		4228	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		4229
		4230	=item D
		4231
		4232	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		4233	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
		4234
		4235	=item Ocaml
		4236
		4237	Erkki Seppala has written Ocaml bindings for libev, to be found at
		4238	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		4239
		4240	=item Lua
		4241
		4242	Brian Maher has written a partial interface to libev for lua (at the
		4243	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		4244	L<http://github.com/brimworks/lua-ev>.
		4245
		4246	=item Javascript
		4247
		4248	Node.js (L<http://nodejs.org>) uses libev as the underlying event library.
		4249
		4250	=item Others
		4251
		4252	There are others, and I stopped counting.
		4253
		4254	=back
2059		4255
2060		4256
2061	=head1 MACRO MAGIC	4257	=head1 MACRO MAGIC
2062		4258
2063	Libev can be compiled with a variety of options, the most fundamantal	4259	Libev can be compiled with a variety of options, the most fundamental
2064	of which is C<EV_MULTIPLICITY>. This option determines whether (most)	4260	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
2065	functions and callbacks have an initial C<struct ev_loop *> argument.	4261	functions and callbacks have an initial C<struct ev_loop *> argument.
2066		4262
2067	To make it easier to write programs that cope with either variant, the	4263	To make it easier to write programs that cope with either variant, the
2068	following macros are defined:	4264	following macros are defined:
…		…
2073		4269
2074	This provides the loop I<argument> for functions, if one is required ("ev	4270	This provides the loop I<argument> for functions, if one is required ("ev
2075	loop argument"). The C<EV_A> form is used when this is the sole argument,	4271	loop argument"). The C<EV_A> form is used when this is the sole argument,
2076	C<EV_A_> is used when other arguments are following. Example:	4272	C<EV_A_> is used when other arguments are following. Example:
2077		4273
2078	ev_unref (EV_A);	4274	ev_unref (EV_A);
2079	ev_timer_add (EV_A_ watcher);	4275	ev_timer_add (EV_A_ watcher);
2080	ev_loop (EV_A_ 0);	4276	ev_run (EV_A_ 0);
2081		4277
2082	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	4278	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
2083	which is often provided by the following macro.	4279	which is often provided by the following macro.
2084		4280
2085	=item C<EV_P>, C<EV_P_>	4281	=item C<EV_P>, C<EV_P_>
2086		4282
2087	This provides the loop I<parameter> for functions, if one is required ("ev	4283	This provides the loop I<parameter> for functions, if one is required ("ev
2088	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	4284	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
2089	C<EV_P_> is used when other parameters are following. Example:	4285	C<EV_P_> is used when other parameters are following. Example:
2090		4286
2091	// this is how ev_unref is being declared	4287	// this is how ev_unref is being declared
2092	static void ev_unref (EV_P);	4288	static void ev_unref (EV_P);
2093		4289
2094	// this is how you can declare your typical callback	4290	// this is how you can declare your typical callback
2095	static void cb (EV_P_ ev_timer *w, int revents)	4291	static void cb (EV_P_ ev_timer *w, int revents)
2096		4292
2097	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	4293	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
2098	suitable for use with C<EV_A>.	4294	suitable for use with C<EV_A>.
2099		4295
2100	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	4296	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
2101		4297
2102	Similar to the other two macros, this gives you the value of the default	4298	Similar to the other two macros, this gives you the value of the default
2103	loop, if multiple loops are supported ("ev loop default").	4299	loop, if multiple loops are supported ("ev loop default"). The default loop
		4300	will be initialised if it isn't already initialised.
		4301
		4302	For non-multiplicity builds, these macros do nothing, so you always have
		4303	to initialise the loop somewhere.
		4304
		4305	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		4306
		4307	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		4308	default loop has been initialised (C<UC> == unchecked). Their behaviour
		4309	is undefined when the default loop has not been initialised by a previous
		4310	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		4311
		4312	It is often prudent to use C<EV_DEFAULT> when initialising the first
		4313	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
2104		4314
2105	=back	4315	=back
2106		4316
2107	Example: Declare and initialise a check watcher, utilising the above	4317	Example: Declare and initialise a check watcher, utilising the above
2108	macros so it will work regardless of whether multiple loops are supported	4318	macros so it will work regardless of whether multiple loops are supported
2109	or not.	4319	or not.
2110		4320
2111	static void	4321	static void
2112	check_cb (EV_P_ ev_timer *w, int revents)	4322	check_cb (EV_P_ ev_timer *w, int revents)
2113	{	4323	{
2114	ev_check_stop (EV_A_ w);	4324	ev_check_stop (EV_A_ w);
2115	}	4325	}
2116		4326
2117	ev_check check;	4327	ev_check check;
2118	ev_check_init (&check, check_cb);	4328	ev_check_init (&check, check_cb);
2119	ev_check_start (EV_DEFAULT_ &check);	4329	ev_check_start (EV_DEFAULT_ &check);
2120	ev_loop (EV_DEFAULT_ 0);	4330	ev_run (EV_DEFAULT_ 0);
2121		4331
2122	=head1 EMBEDDING	4332	=head1 EMBEDDING
2123		4333
2124	Libev can (and often is) directly embedded into host	4334	Libev can (and often is) directly embedded into host
2125	applications. Examples of applications that embed it include the Deliantra	4335	applications. Examples of applications that embed it include the Deliantra
2126	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)	4336	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2127	and rxvt-unicode.	4337	and rxvt-unicode.
2128		4338
2129	The goal is to enable you to just copy the neecssary files into your	4339	The goal is to enable you to just copy the necessary files into your
2130	source directory without having to change even a single line in them, so	4340	source directory without having to change even a single line in them, so
2131	you can easily upgrade by simply copying (or having a checked-out copy of	4341	you can easily upgrade by simply copying (or having a checked-out copy of
2132	libev somewhere in your source tree).	4342	libev somewhere in your source tree).
2133		4343
2134	=head2 FILESETS	4344	=head2 FILESETS
2135		4345
2136	Depending on what features you need you need to include one or more sets of files	4346	Depending on what features you need you need to include one or more sets of files
2137	in your app.	4347	in your application.
2138		4348
2139	=head3 CORE EVENT LOOP	4349	=head3 CORE EVENT LOOP
2140		4350
2141	To include only the libev core (all the C<ev_*> functions), with manual	4351	To include only the libev core (all the C<ev_*> functions), with manual
2142	configuration (no autoconf):	4352	configuration (no autoconf):
2143		4353
2144	#define EV_STANDALONE 1	4354	#define EV_STANDALONE 1
2145	#include "ev.c"	4355	#include "ev.c"
2146		4356
2147	This will automatically include F<ev.h>, too, and should be done in a	4357	This will automatically include F<ev.h>, too, and should be done in a
2148	single C source file only to provide the function implementations. To use	4358	single C source file only to provide the function implementations. To use
2149	it, do the same for F<ev.h> in all files wishing to use this API (best	4359	it, do the same for F<ev.h> in all files wishing to use this API (best
2150	done by writing a wrapper around F<ev.h> that you can include instead and	4360	done by writing a wrapper around F<ev.h> that you can include instead and
2151	where you can put other configuration options):	4361	where you can put other configuration options):
2152		4362
2153	#define EV_STANDALONE 1	4363	#define EV_STANDALONE 1
2154	#include "ev.h"	4364	#include "ev.h"
2155		4365
2156	Both header files and implementation files can be compiled with a C++	4366	Both header files and implementation files can be compiled with a C++
2157	compiler (at least, thats a stated goal, and breakage will be treated	4367	compiler (at least, that's a stated goal, and breakage will be treated
2158	as a bug).	4368	as a bug).
2159		4369
2160	You need the following files in your source tree, or in a directory	4370	You need the following files in your source tree, or in a directory
2161	in your include path (e.g. in libev/ when using -Ilibev):	4371	in your include path (e.g. in libev/ when using -Ilibev):
2162		4372
2163	ev.h	4373	ev.h
2164	ev.c	4374	ev.c
2165	ev_vars.h	4375	ev_vars.h
2166	ev_wrap.h	4376	ev_wrap.h
2167		4377
2168	ev_win32.c required on win32 platforms only	4378	ev_win32.c required on win32 platforms only
2169		4379
2170	ev_select.c only when select backend is enabled (which is enabled by default)	4380	ev_select.c only when select backend is enabled (which is enabled by default)
2171	ev_poll.c only when poll backend is enabled (disabled by default)	4381	ev_poll.c only when poll backend is enabled (disabled by default)
2172	ev_epoll.c only when the epoll backend is enabled (disabled by default)	4382	ev_epoll.c only when the epoll backend is enabled (disabled by default)
2173	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	4383	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2174	ev_port.c only when the solaris port backend is enabled (disabled by default)	4384	ev_port.c only when the solaris port backend is enabled (disabled by default)
2175		4385
2176	F<ev.c> includes the backend files directly when enabled, so you only need	4386	F<ev.c> includes the backend files directly when enabled, so you only need
2177	to compile this single file.	4387	to compile this single file.
2178		4388
2179	=head3 LIBEVENT COMPATIBILITY API	4389	=head3 LIBEVENT COMPATIBILITY API
2180		4390
2181	To include the libevent compatibility API, also include:	4391	To include the libevent compatibility API, also include:
2182		4392
2183	#include "event.c"	4393	#include "event.c"
2184		4394
2185	in the file including F<ev.c>, and:	4395	in the file including F<ev.c>, and:
2186		4396
2187	#include "event.h"	4397	#include "event.h"
2188		4398
2189	in the files that want to use the libevent API. This also includes F<ev.h>.	4399	in the files that want to use the libevent API. This also includes F<ev.h>.
2190		4400
2191	You need the following additional files for this:	4401	You need the following additional files for this:
2192		4402
2193	event.h	4403	event.h
2194	event.c	4404	event.c
2195		4405
2196	=head3 AUTOCONF SUPPORT	4406	=head3 AUTOCONF SUPPORT
2197		4407
2198	Instead of using C<EV_STANDALONE=1> and providing your config in	4408	Instead of using C<EV_STANDALONE=1> and providing your configuration in
2199	whatever way you want, you can also C<m4_include([libev.m4])> in your	4409	whatever way you want, you can also C<m4_include([libev.m4])> in your
2200	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	4410	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
2201	include F<config.h> and configure itself accordingly.	4411	include F<config.h> and configure itself accordingly.
2202		4412
2203	For this of course you need the m4 file:	4413	For this of course you need the m4 file:
2204		4414
2205	libev.m4	4415	libev.m4
2206		4416
2207	=head2 PREPROCESSOR SYMBOLS/MACROS	4417	=head2 PREPROCESSOR SYMBOLS/MACROS
2208		4418
2209	Libev can be configured via a variety of preprocessor symbols you have to define	4419	Libev can be configured via a variety of preprocessor symbols you have to
2210	before including any of its files. The default is not to build for multiplicity	4420	define before including (or compiling) any of its files. The default in
2211	and only include the select backend.	4421	the absence of autoconf is documented for every option.
		4422
		4423	Symbols marked with "(h)" do not change the ABI, and can have different
		4424	values when compiling libev vs. including F<ev.h>, so it is permissible
		4425	to redefine them before including F<ev.h> without breaking compatibility
		4426	to a compiled library. All other symbols change the ABI, which means all
		4427	users of libev and the libev code itself must be compiled with compatible
		4428	settings.
2212		4429
2213	=over 4	4430	=over 4
2214		4431
		4432	=item EV_COMPAT3 (h)
		4433
		4434	Backwards compatibility is a major concern for libev. This is why this
		4435	release of libev comes with wrappers for the functions and symbols that
		4436	have been renamed between libev version 3 and 4.
		4437
		4438	You can disable these wrappers (to test compatibility with future
		4439	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4440	sources. This has the additional advantage that you can drop the C<struct>
		4441	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4442	typedef in that case.
		4443
		4444	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4445	and in some even more future version the compatibility code will be
		4446	removed completely.
		4447
2215	=item EV_STANDALONE	4448	=item EV_STANDALONE (h)
2216		4449
2217	Must always be C<1> if you do not use autoconf configuration, which	4450	Must always be C<1> if you do not use autoconf configuration, which
2218	keeps libev from including F<config.h>, and it also defines dummy	4451	keeps libev from including F<config.h>, and it also defines dummy
2219	implementations for some libevent functions (such as logging, which is not	4452	implementations for some libevent functions (such as logging, which is not
2220	supported). It will also not define any of the structs usually found in	4453	supported). It will also not define any of the structs usually found in
2221	F<event.h> that are not directly supported by the libev core alone.	4454	F<event.h> that are not directly supported by the libev core alone.
2222		4455
		4456	In standalone mode, libev will still try to automatically deduce the
		4457	configuration, but has to be more conservative.
		4458
		4459	=item EV_USE_FLOOR
		4460
		4461	If defined to be C<1>, libev will use the C<floor ()> function for its
		4462	periodic reschedule calculations, otherwise libev will fall back on a
		4463	portable (slower) implementation. If you enable this, you usually have to
		4464	link against libm or something equivalent. Enabling this when the C<floor>
		4465	function is not available will fail, so the safe default is to not enable
		4466	this.
		4467
2223	=item EV_USE_MONOTONIC	4468	=item EV_USE_MONOTONIC
2224		4469
2225	If defined to be C<1>, libev will try to detect the availability of the	4470	If defined to be C<1>, libev will try to detect the availability of the
2226	monotonic clock option at both compiletime and runtime. Otherwise no use	4471	monotonic clock option at both compile time and runtime. Otherwise no
2227	of the monotonic clock option will be attempted. If you enable this, you	4472	use of the monotonic clock option will be attempted. If you enable this,
2228	usually have to link against librt or something similar. Enabling it when	4473	you usually have to link against librt or something similar. Enabling it
2229	the functionality isn't available is safe, though, althoguh you have	4474	when the functionality isn't available is safe, though, although you have
2230	to make sure you link against any libraries where the C<clock_gettime>	4475	to make sure you link against any libraries where the C<clock_gettime>
2231	function is hiding in (often F<-lrt>).	4476	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
2232		4477
2233	=item EV_USE_REALTIME	4478	=item EV_USE_REALTIME
2234		4479
2235	If defined to be C<1>, libev will try to detect the availability of the	4480	If defined to be C<1>, libev will try to detect the availability of the
2236	realtime clock option at compiletime (and assume its availability at	4481	real-time clock option at compile time (and assume its availability
2237	runtime if successful). Otherwise no use of the realtime clock option will	4482	at runtime if successful). Otherwise no use of the real-time clock
2238	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	4483	option will be attempted. This effectively replaces C<gettimeofday>
2239	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries	4484	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
2240	in the description of C<EV_USE_MONOTONIC>, though.	4485	correctness. See the note about libraries in the description of
		4486	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4487	C<EV_USE_CLOCK_SYSCALL>.
		4488
		4489	=item EV_USE_CLOCK_SYSCALL
		4490
		4491	If defined to be C<1>, libev will try to use a direct syscall instead
		4492	of calling the system-provided C<clock_gettime> function. This option
		4493	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4494	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4495	programs needlessly. Using a direct syscall is slightly slower (in
		4496	theory), because no optimised vdso implementation can be used, but avoids
		4497	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4498	higher, as it simplifies linking (no need for C<-lrt>).
		4499
		4500	=item EV_USE_NANOSLEEP
		4501
		4502	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		4503	and will use it for delays. Otherwise it will use C<select ()>.
		4504
		4505	=item EV_USE_EVENTFD
		4506
		4507	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4508	available and will probe for kernel support at runtime. This will improve
		4509	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4510	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4511	2.7 or newer, otherwise disabled.
2241		4512
2242	=item EV_USE_SELECT	4513	=item EV_USE_SELECT
2243		4514
2244	If undefined or defined to be C<1>, libev will compile in support for the	4515	If undefined or defined to be C<1>, libev will compile in support for the
2245	C<select>(2) backend. No attempt at autodetection will be done: if no	4516	C<select>(2) backend. No attempt at auto-detection will be done: if no
2246	other method takes over, select will be it. Otherwise the select backend	4517	other method takes over, select will be it. Otherwise the select backend
2247	will not be compiled in.	4518	will not be compiled in.
2248		4519
2249	=item EV_SELECT_USE_FD_SET	4520	=item EV_SELECT_USE_FD_SET
2250		4521
2251	If defined to C<1>, then the select backend will use the system C<fd_set>	4522	If defined to C<1>, then the select backend will use the system C<fd_set>
2252	structure. This is useful if libev doesn't compile due to a missing	4523	structure. This is useful if libev doesn't compile due to a missing
2253	C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on	4524	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
2254	exotic systems. This usually limits the range of file descriptors to some	4525	on exotic systems. This usually limits the range of file descriptors to
2255	low limit such as 1024 or might have other limitations (winsocket only	4526	some low limit such as 1024 or might have other limitations (winsocket
2256	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	4527	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
2257	influence the size of the C<fd_set> used.	4528	configures the maximum size of the C<fd_set>.
2258		4529
2259	=item EV_SELECT_IS_WINSOCKET	4530	=item EV_SELECT_IS_WINSOCKET
2260		4531
2261	When defined to C<1>, the select backend will assume that	4532	When defined to C<1>, the select backend will assume that
2262	select/socket/connect etc. don't understand file descriptors but	4533	select/socket/connect etc. don't understand file descriptors but
…		…
2264	be used is the winsock select). This means that it will call	4535	be used is the winsock select). This means that it will call
2265	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	4536	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2266	it is assumed that all these functions actually work on fds, even	4537	it is assumed that all these functions actually work on fds, even
2267	on win32. Should not be defined on non-win32 platforms.	4538	on win32. Should not be defined on non-win32 platforms.
2268		4539
		4540	=item EV_FD_TO_WIN32_HANDLE(fd)
		4541
		4542	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		4543	file descriptors to socket handles. When not defining this symbol (the
		4544	default), then libev will call C<_get_osfhandle>, which is usually
		4545	correct. In some cases, programs use their own file descriptor management,
		4546	in which case they can provide this function to map fds to socket handles.
		4547
		4548	=item EV_WIN32_HANDLE_TO_FD(handle)
		4549
		4550	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4551	using the standard C<_open_osfhandle> function. For programs implementing
		4552	their own fd to handle mapping, overwriting this function makes it easier
		4553	to do so. This can be done by defining this macro to an appropriate value.
		4554
		4555	=item EV_WIN32_CLOSE_FD(fd)
		4556
		4557	If programs implement their own fd to handle mapping on win32, then this
		4558	macro can be used to override the C<close> function, useful to unregister
		4559	file descriptors again. Note that the replacement function has to close
		4560	the underlying OS handle.
		4561
		4562	=item EV_USE_WSASOCKET
		4563
		4564	If defined to be C<1>, libev will use C<WSASocket> to create its internal
		4565	communication socket, which works better in some environments. Otherwise,
		4566	the normal C<socket> function will be used, which works better in other
		4567	environments.
		4568
2269	=item EV_USE_POLL	4569	=item EV_USE_POLL
2270		4570
2271	If defined to be C<1>, libev will compile in support for the C<poll>(2)	4571	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2272	backend. Otherwise it will be enabled on non-win32 platforms. It	4572	backend. Otherwise it will be enabled on non-win32 platforms. It
2273	takes precedence over select.	4573	takes precedence over select.
2274		4574
2275	=item EV_USE_EPOLL	4575	=item EV_USE_EPOLL
2276		4576
2277	If defined to be C<1>, libev will compile in support for the Linux	4577	If defined to be C<1>, libev will compile in support for the Linux
2278	C<epoll>(7) backend. Its availability will be detected at runtime,	4578	C<epoll>(7) backend. Its availability will be detected at runtime,
2279	otherwise another method will be used as fallback. This is the	4579	otherwise another method will be used as fallback. This is the preferred
2280	preferred backend for GNU/Linux systems.	4580	backend for GNU/Linux systems. If undefined, it will be enabled if the
		4581	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
2281		4582
2282	=item EV_USE_KQUEUE	4583	=item EV_USE_KQUEUE
2283		4584
2284	If defined to be C<1>, libev will compile in support for the BSD style	4585	If defined to be C<1>, libev will compile in support for the BSD style
2285	C<kqueue>(2) backend. Its actual availability will be detected at runtime,	4586	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
…		…
2298	otherwise another method will be used as fallback. This is the preferred	4599	otherwise another method will be used as fallback. This is the preferred
2299	backend for Solaris 10 systems.	4600	backend for Solaris 10 systems.
2300		4601
2301	=item EV_USE_DEVPOLL	4602	=item EV_USE_DEVPOLL
2302		4603
2303	reserved for future expansion, works like the USE symbols above.	4604	Reserved for future expansion, works like the USE symbols above.
2304		4605
2305	=item EV_USE_INOTIFY	4606	=item EV_USE_INOTIFY
2306		4607
2307	If defined to be C<1>, libev will compile in support for the Linux inotify	4608	If defined to be C<1>, libev will compile in support for the Linux inotify
2308	interface to speed up C<ev_stat> watchers. Its actual availability will	4609	interface to speed up C<ev_stat> watchers. Its actual availability will
2309	be detected at runtime.	4610	be detected at runtime. If undefined, it will be enabled if the headers
		4611	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
2310		4612
		4613	=item EV_NO_SMP
		4614
		4615	If defined to be C<1>, libev will assume that memory is always coherent
		4616	between threads, that is, threads can be used, but threads never run on
		4617	different cpus (or different cpu cores). This reduces dependencies
		4618	and makes libev faster.
		4619
		4620	=item EV_NO_THREADS
		4621
		4622	If defined to be C<1>, libev will assume that it will never be called
		4623	from different threads, which is a stronger assumption than C<EV_NO_SMP>,
		4624	above. This reduces dependencies and makes libev faster.
		4625
		4626	=item EV_ATOMIC_T
		4627
		4628	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		4629	access is atomic with respect to other threads or signal contexts. No
		4630	such type is easily found in the C language, so you can provide your own
		4631	type that you know is safe for your purposes. It is used both for signal
		4632	handler "locking" as well as for signal and thread safety in C<ev_async>
		4633	watchers.
		4634
		4635	In the absence of this define, libev will use C<sig_atomic_t volatile>
		4636	(from F<signal.h>), which is usually good enough on most platforms.
		4637
2311	=item EV_H	4638	=item EV_H (h)
2312		4639
2313	The name of the F<ev.h> header file used to include it. The default if	4640	The name of the F<ev.h> header file used to include it. The default if
2314	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	4641	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2315	can be used to virtually rename the F<ev.h> header file in case of conflicts.	4642	used to virtually rename the F<ev.h> header file in case of conflicts.
2316		4643
2317	=item EV_CONFIG_H	4644	=item EV_CONFIG_H (h)
2318		4645
2319	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	4646	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2320	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	4647	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2321	C<EV_H>, above.	4648	C<EV_H>, above.
2322		4649
2323	=item EV_EVENT_H	4650	=item EV_EVENT_H (h)
2324		4651
2325	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	4652	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2326	of how the F<event.h> header can be found.	4653	of how the F<event.h> header can be found, the default is C<"event.h">.
2327		4654
2328	=item EV_PROTOTYPES	4655	=item EV_PROTOTYPES (h)
2329		4656
2330	If defined to be C<0>, then F<ev.h> will not define any function	4657	If defined to be C<0>, then F<ev.h> will not define any function
2331	prototypes, but still define all the structs and other symbols. This is	4658	prototypes, but still define all the structs and other symbols. This is
2332	occasionally useful if you want to provide your own wrapper functions	4659	occasionally useful if you want to provide your own wrapper functions
2333	around libev functions.	4660	around libev functions.
…		…
2338	will have the C<struct ev_loop *> as first argument, and you can create	4665	will have the C<struct ev_loop *> as first argument, and you can create
2339	additional independent event loops. Otherwise there will be no support	4666	additional independent event loops. Otherwise there will be no support
2340	for multiple event loops and there is no first event loop pointer	4667	for multiple event loops and there is no first event loop pointer
2341	argument. Instead, all functions act on the single default loop.	4668	argument. Instead, all functions act on the single default loop.
2342		4669
		4670	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4671	default loop when multiplicity is switched off - you always have to
		4672	initialise the loop manually in this case.
		4673
2343	=item EV_MINPRI	4674	=item EV_MINPRI
2344		4675
2345	=item EV_MAXPRI	4676	=item EV_MAXPRI
2346		4677
2347	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to	4678	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
…		…
2352	When doing priority-based operations, libev usually has to linearly search	4683	When doing priority-based operations, libev usually has to linearly search
2353	all the priorities, so having many of them (hundreds) uses a lot of space	4684	all the priorities, so having many of them (hundreds) uses a lot of space
2354	and time, so using the defaults of five priorities (-2 .. +2) is usually	4685	and time, so using the defaults of five priorities (-2 .. +2) is usually
2355	fine.	4686	fine.
2356		4687
2357	If your embedding app does not need any priorities, defining these both to	4688	If your embedding application does not need any priorities, defining these
2358	C<0> will save some memory and cpu.	4689	both to C<0> will save some memory and CPU.
2359		4690
2360	=item EV_PERIODIC_ENABLE	4691	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4692	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4693	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
2361		4694
2362	If undefined or defined to be C<1>, then periodic timers are supported. If	4695	If undefined or defined to be C<1> (and the platform supports it), then
2363	defined to be C<0>, then they are not. Disabling them saves a few kB of	4696	the respective watcher type is supported. If defined to be C<0>, then it
2364	code.	4697	is not. Disabling watcher types mainly saves code size.
2365		4698
2366	=item EV_IDLE_ENABLE	4699	=item EV_FEATURES
2367
2368	If undefined or defined to be C<1>, then idle watchers are supported. If
2369	defined to be C<0>, then they are not. Disabling them saves a few kB of
2370	code.
2371
2372	=item EV_EMBED_ENABLE
2373
2374	If undefined or defined to be C<1>, then embed watchers are supported. If
2375	defined to be C<0>, then they are not.
2376
2377	=item EV_STAT_ENABLE
2378
2379	If undefined or defined to be C<1>, then stat watchers are supported. If
2380	defined to be C<0>, then they are not.
2381
2382	=item EV_FORK_ENABLE
2383
2384	If undefined or defined to be C<1>, then fork watchers are supported. If
2385	defined to be C<0>, then they are not.
2386
2387	=item EV_MINIMAL
2388		4700
2389	If you need to shave off some kilobytes of code at the expense of some	4701	If you need to shave off some kilobytes of code at the expense of some
2390	speed, define this symbol to C<1>. Currently only used for gcc to override	4702	speed (but with the full API), you can define this symbol to request
2391	some inlining decisions, saves roughly 30% codesize of amd64.	4703	certain subsets of functionality. The default is to enable all features
		4704	that can be enabled on the platform.
		4705
		4706	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4707	with some broad features you want) and then selectively re-enable
		4708	additional parts you want, for example if you want everything minimal,
		4709	but multiple event loop support, async and child watchers and the poll
		4710	backend, use this:
		4711
		4712	#define EV_FEATURES 0
		4713	#define EV_MULTIPLICITY 1
		4714	#define EV_USE_POLL 1
		4715	#define EV_CHILD_ENABLE 1
		4716	#define EV_ASYNC_ENABLE 1
		4717
		4718	The actual value is a bitset, it can be a combination of the following
		4719	values (by default, all of these are enabled):
		4720
		4721	=over 4
		4722
		4723	=item C<1> - faster/larger code
		4724
		4725	Use larger code to speed up some operations.
		4726
		4727	Currently this is used to override some inlining decisions (enlarging the
		4728	code size by roughly 30% on amd64).
		4729
		4730	When optimising for size, use of compiler flags such as C<-Os> with
		4731	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4732	assertions.
		4733
		4734	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4735	(e.g. gcc with C<-Os>).
		4736
		4737	=item C<2> - faster/larger data structures
		4738
		4739	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4740	hash table sizes and so on. This will usually further increase code size
		4741	and can additionally have an effect on the size of data structures at
		4742	runtime.
		4743
		4744	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4745	(e.g. gcc with C<-Os>).
		4746
		4747	=item C<4> - full API configuration
		4748
		4749	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4750	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4751
		4752	=item C<8> - full API
		4753
		4754	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4755	details on which parts of the API are still available without this
		4756	feature, and do not complain if this subset changes over time.
		4757
		4758	=item C<16> - enable all optional watcher types
		4759
		4760	Enables all optional watcher types. If you want to selectively enable
		4761	only some watcher types other than I/O and timers (e.g. prepare,
		4762	embed, async, child...) you can enable them manually by defining
		4763	C<EV_watchertype_ENABLE> to C<1> instead.
		4764
		4765	=item C<32> - enable all backends
		4766
		4767	This enables all backends - without this feature, you need to enable at
		4768	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4769
		4770	=item C<64> - enable OS-specific "helper" APIs
		4771
		4772	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4773	default.
		4774
		4775	=back
		4776
		4777	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4778	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4779	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4780	watchers, timers and monotonic clock support.
		4781
		4782	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4783	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4784	your program might be left out as well - a binary starting a timer and an
		4785	I/O watcher then might come out at only 5Kb.
		4786
		4787	=item EV_API_STATIC
		4788
		4789	If this symbol is defined (by default it is not), then all identifiers
		4790	will have static linkage. This means that libev will not export any
		4791	identifiers, and you cannot link against libev anymore. This can be useful
		4792	when you embed libev, only want to use libev functions in a single file,
		4793	and do not want its identifiers to be visible.
		4794
		4795	To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
		4796	wants to use libev.
		4797
		4798	This option only works when libev is compiled with a C compiler, as C++
		4799	doesn't support the required declaration syntax.
		4800
		4801	=item EV_AVOID_STDIO
		4802
		4803	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4804	functions (printf, scanf, perror etc.). This will increase the code size
		4805	somewhat, but if your program doesn't otherwise depend on stdio and your
		4806	libc allows it, this avoids linking in the stdio library which is quite
		4807	big.
		4808
		4809	Note that error messages might become less precise when this option is
		4810	enabled.
		4811
		4812	=item EV_NSIG
		4813
		4814	The highest supported signal number, +1 (or, the number of
		4815	signals): Normally, libev tries to deduce the maximum number of signals
		4816	automatically, but sometimes this fails, in which case it can be
		4817	specified. Also, using a lower number than detected (C<32> should be
		4818	good for about any system in existence) can save some memory, as libev
		4819	statically allocates some 12-24 bytes per signal number.
2392		4820
2393	=item EV_PID_HASHSIZE	4821	=item EV_PID_HASHSIZE
2394		4822
2395	C<ev_child> watchers use a small hash table to distribute workload by	4823	C<ev_child> watchers use a small hash table to distribute workload by
2396	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	4824	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
2397	than enough. If you need to manage thousands of children you might want to	4825	usually more than enough. If you need to manage thousands of children you
2398	increase this value (I<must> be a power of two).	4826	might want to increase this value (I<must> be a power of two).
2399		4827
2400	=item EV_INOTIFY_HASHSIZE	4828	=item EV_INOTIFY_HASHSIZE
2401		4829
2402	C<ev_staz> watchers use a small hash table to distribute workload by	4830	C<ev_stat> watchers use a small hash table to distribute workload by
2403	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	4831	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
2404	usually more than enough. If you need to manage thousands of C<ev_stat>	4832	disabled), usually more than enough. If you need to manage thousands of
2405	watchers you might want to increase this value (I<must> be a power of	4833	C<ev_stat> watchers you might want to increase this value (I<must> be a
2406	two).	4834	power of two).
		4835
		4836	=item EV_USE_4HEAP
		4837
		4838	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4839	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4840	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4841	faster performance with many (thousands) of watchers.
		4842
		4843	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4844	will be C<0>.
		4845
		4846	=item EV_HEAP_CACHE_AT
		4847
		4848	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4849	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4850	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4851	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4852	but avoids random read accesses on heap changes. This improves performance
		4853	noticeably with many (hundreds) of watchers.
		4854
		4855	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4856	will be C<0>.
		4857
		4858	=item EV_VERIFY
		4859
		4860	Controls how much internal verification (see C<ev_verify ()>) will
		4861	be done: If set to C<0>, no internal verification code will be compiled
		4862	in. If set to C<1>, then verification code will be compiled in, but not
		4863	called. If set to C<2>, then the internal verification code will be
		4864	called once per loop, which can slow down libev. If set to C<3>, then the
		4865	verification code will be called very frequently, which will slow down
		4866	libev considerably.
		4867
		4868	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4869	will be C<0>.
2407		4870
2408	=item EV_COMMON	4871	=item EV_COMMON
2409		4872
2410	By default, all watchers have a C<void *data> member. By redefining	4873	By default, all watchers have a C<void *data> member. By redefining
2411	this macro to a something else you can include more and other types of	4874	this macro to something else you can include more and other types of
2412	members. You have to define it each time you include one of the files,	4875	members. You have to define it each time you include one of the files,
2413	though, and it must be identical each time.	4876	though, and it must be identical each time.
2414		4877
2415	For example, the perl EV module uses something like this:	4878	For example, the perl EV module uses something like this:
2416		4879
2417	#define EV_COMMON \	4880	#define EV_COMMON \
2418	SV *self; /* contains this struct */ \	4881	SV *self; /* contains this struct */ \
2419	SV *cb_sv, *fh /* note no trailing ";" */	4882	SV *cb_sv, *fh /* note no trailing ";" */
2420		4883
2421	=item EV_CB_DECLARE (type)	4884	=item EV_CB_DECLARE (type)
2422		4885
2423	=item EV_CB_INVOKE (watcher, revents)	4886	=item EV_CB_INVOKE (watcher, revents)
2424		4887
2425	=item ev_set_cb (ev, cb)	4888	=item ev_set_cb (ev, cb)
2426		4889
2427	Can be used to change the callback member declaration in each watcher,	4890	Can be used to change the callback member declaration in each watcher,
2428	and the way callbacks are invoked and set. Must expand to a struct member	4891	and the way callbacks are invoked and set. Must expand to a struct member
2429	definition and a statement, respectively. See the F<ev.v> header file for	4892	definition and a statement, respectively. See the F<ev.h> header file for
2430	their default definitions. One possible use for overriding these is to	4893	their default definitions. One possible use for overriding these is to
2431	avoid the C<struct ev_loop *> as first argument in all cases, or to use	4894	avoid the C<struct ev_loop *> as first argument in all cases, or to use
2432	method calls instead of plain function calls in C++.	4895	method calls instead of plain function calls in C++.
		4896
		4897	=back
		4898
		4899	=head2 EXPORTED API SYMBOLS
		4900
		4901	If you need to re-export the API (e.g. via a DLL) and you need a list of
		4902	exported symbols, you can use the provided F<Symbol.*> files which list
		4903	all public symbols, one per line:
		4904
		4905	Symbols.ev for libev proper
		4906	Symbols.event for the libevent emulation
		4907
		4908	This can also be used to rename all public symbols to avoid clashes with
		4909	multiple versions of libev linked together (which is obviously bad in
		4910	itself, but sometimes it is inconvenient to avoid this).
		4911
		4912	A sed command like this will create wrapper C<#define>'s that you need to
		4913	include before including F<ev.h>:
		4914
		4915	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		4916
		4917	This would create a file F<wrap.h> which essentially looks like this:
		4918
		4919	#define ev_backend myprefix_ev_backend
		4920	#define ev_check_start myprefix_ev_check_start
		4921	#define ev_check_stop myprefix_ev_check_stop
		4922	...
2433		4923
2434	=head2 EXAMPLES	4924	=head2 EXAMPLES
2435		4925
2436	For a real-world example of a program the includes libev	4926	For a real-world example of a program the includes libev
2437	verbatim, you can have a look at the EV perl module	4927	verbatim, you can have a look at the EV perl module
…		…
2442	file.	4932	file.
2443		4933
2444	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	4934	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
2445	that everybody includes and which overrides some configure choices:	4935	that everybody includes and which overrides some configure choices:
2446		4936
2447	#define EV_MINIMAL 1	4937	#define EV_FEATURES 8
2448	#define EV_USE_POLL 0	4938	#define EV_USE_SELECT 1
2449	#define EV_MULTIPLICITY 0
2450	#define EV_PERIODIC_ENABLE 0	4939	#define EV_PREPARE_ENABLE 1
		4940	#define EV_IDLE_ENABLE 1
2451	#define EV_STAT_ENABLE 0	4941	#define EV_SIGNAL_ENABLE 1
2452	#define EV_FORK_ENABLE 0	4942	#define EV_CHILD_ENABLE 1
		4943	#define EV_USE_STDEXCEPT 0
2453	#define EV_CONFIG_H <config.h>	4944	#define EV_CONFIG_H <config.h>
2454	#define EV_MINPRI 0
2455	#define EV_MAXPRI 0
2456		4945
2457	#include "ev++.h"	4946	#include "ev++.h"
2458		4947
2459	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	4948	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
2460		4949
2461	#include "ev_cpp.h"	4950	#include "ev_cpp.h"
2462	#include "ev.c"	4951	#include "ev.c"
2463		4952
		4953	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
2464		4954
		4955	=head2 THREADS AND COROUTINES
		4956
		4957	=head3 THREADS
		4958
		4959	All libev functions are reentrant and thread-safe unless explicitly
		4960	documented otherwise, but libev implements no locking itself. This means
		4961	that you can use as many loops as you want in parallel, as long as there
		4962	are no concurrent calls into any libev function with the same loop
		4963	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4964	of course): libev guarantees that different event loops share no data
		4965	structures that need any locking.
		4966
		4967	Or to put it differently: calls with different loop parameters can be done
		4968	concurrently from multiple threads, calls with the same loop parameter
		4969	must be done serially (but can be done from different threads, as long as
		4970	only one thread ever is inside a call at any point in time, e.g. by using
		4971	a mutex per loop).
		4972
		4973	Specifically to support threads (and signal handlers), libev implements
		4974	so-called C<ev_async> watchers, which allow some limited form of
		4975	concurrency on the same event loop, namely waking it up "from the
		4976	outside".
		4977
		4978	If you want to know which design (one loop, locking, or multiple loops
		4979	without or something else still) is best for your problem, then I cannot
		4980	help you, but here is some generic advice:
		4981
		4982	=over 4
		4983
		4984	=item * most applications have a main thread: use the default libev loop
		4985	in that thread, or create a separate thread running only the default loop.
		4986
		4987	This helps integrating other libraries or software modules that use libev
		4988	themselves and don't care/know about threading.
		4989
		4990	=item * one loop per thread is usually a good model.
		4991
		4992	Doing this is almost never wrong, sometimes a better-performance model
		4993	exists, but it is always a good start.
		4994
		4995	=item * other models exist, such as the leader/follower pattern, where one
		4996	loop is handed through multiple threads in a kind of round-robin fashion.
		4997
		4998	Choosing a model is hard - look around, learn, know that usually you can do
		4999	better than you currently do :-)
		5000
		5001	=item * often you need to talk to some other thread which blocks in the
		5002	event loop.
		5003
		5004	C<ev_async> watchers can be used to wake them up from other threads safely
		5005	(or from signal contexts...).
		5006
		5007	An example use would be to communicate signals or other events that only
		5008	work in the default loop by registering the signal watcher with the
		5009	default loop and triggering an C<ev_async> watcher from the default loop
		5010	watcher callback into the event loop interested in the signal.
		5011
		5012	=back
		5013
		5014	See also L</THREAD LOCKING EXAMPLE>.
		5015
		5016	=head3 COROUTINES
		5017
		5018	Libev is very accommodating to coroutines ("cooperative threads"):
		5019	libev fully supports nesting calls to its functions from different
		5020	coroutines (e.g. you can call C<ev_run> on the same loop from two
		5021	different coroutines, and switch freely between both coroutines running
		5022	the loop, as long as you don't confuse yourself). The only exception is
		5023	that you must not do this from C<ev_periodic> reschedule callbacks.
		5024
		5025	Care has been taken to ensure that libev does not keep local state inside
		5026	C<ev_run>, and other calls do not usually allow for coroutine switches as
		5027	they do not call any callbacks.
		5028
		5029	=head2 COMPILER WARNINGS
		5030
		5031	Depending on your compiler and compiler settings, you might get no or a
		5032	lot of warnings when compiling libev code. Some people are apparently
		5033	scared by this.
		5034
		5035	However, these are unavoidable for many reasons. For one, each compiler
		5036	has different warnings, and each user has different tastes regarding
		5037	warning options. "Warn-free" code therefore cannot be a goal except when
		5038	targeting a specific compiler and compiler-version.
		5039
		5040	Another reason is that some compiler warnings require elaborate
		5041	workarounds, or other changes to the code that make it less clear and less
		5042	maintainable.
		5043
		5044	And of course, some compiler warnings are just plain stupid, or simply
		5045	wrong (because they don't actually warn about the condition their message
		5046	seems to warn about). For example, certain older gcc versions had some
		5047	warnings that resulted in an extreme number of false positives. These have
		5048	been fixed, but some people still insist on making code warn-free with
		5049	such buggy versions.
		5050
		5051	While libev is written to generate as few warnings as possible,
		5052	"warn-free" code is not a goal, and it is recommended not to build libev
		5053	with any compiler warnings enabled unless you are prepared to cope with
		5054	them (e.g. by ignoring them). Remember that warnings are just that:
		5055	warnings, not errors, or proof of bugs.
		5056
		5057
		5058	=head2 VALGRIND
		5059
		5060	Valgrind has a special section here because it is a popular tool that is
		5061	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		5062
		5063	If you think you found a bug (memory leak, uninitialised data access etc.)
		5064	in libev, then check twice: If valgrind reports something like:
		5065
		5066	==2274== definitely lost: 0 bytes in 0 blocks.
		5067	==2274== possibly lost: 0 bytes in 0 blocks.
		5068	==2274== still reachable: 256 bytes in 1 blocks.
		5069
		5070	Then there is no memory leak, just as memory accounted to global variables
		5071	is not a memleak - the memory is still being referenced, and didn't leak.
		5072
		5073	Similarly, under some circumstances, valgrind might report kernel bugs
		5074	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		5075	although an acceptable workaround has been found here), or it might be
		5076	confused.
		5077
		5078	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		5079	make it into some kind of religion.
		5080
		5081	If you are unsure about something, feel free to contact the mailing list
		5082	with the full valgrind report and an explanation on why you think this
		5083	is a bug in libev (best check the archives, too :). However, don't be
		5084	annoyed when you get a brisk "this is no bug" answer and take the chance
		5085	of learning how to interpret valgrind properly.
		5086
		5087	If you need, for some reason, empty reports from valgrind for your project
		5088	I suggest using suppression lists.
		5089
		5090
		5091	=head1 PORTABILITY NOTES
		5092
		5093	=head2 GNU/LINUX 32 BIT LIMITATIONS
		5094
		5095	GNU/Linux is the only common platform that supports 64 bit file/large file
		5096	interfaces but I<disables> them by default.
		5097
		5098	That means that libev compiled in the default environment doesn't support
		5099	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		5100
		5101	Unfortunately, many programs try to work around this GNU/Linux issue
		5102	by enabling the large file API, which makes them incompatible with the
		5103	standard libev compiled for their system.
		5104
		5105	Likewise, libev cannot enable the large file API itself as this would
		5106	suddenly make it incompatible to the default compile time environment,
		5107	i.e. all programs not using special compile switches.
		5108
		5109	=head2 OS/X AND DARWIN BUGS
		5110
		5111	The whole thing is a bug if you ask me - basically any system interface
		5112	you touch is broken, whether it is locales, poll, kqueue or even the
		5113	OpenGL drivers.
		5114
		5115	=head3 C<kqueue> is buggy
		5116
		5117	The kqueue syscall is broken in all known versions - most versions support
		5118	only sockets, many support pipes.
		5119
		5120	Libev tries to work around this by not using C<kqueue> by default on this
		5121	rotten platform, but of course you can still ask for it when creating a
		5122	loop - embedding a socket-only kqueue loop into a select-based one is
		5123	probably going to work well.
		5124
		5125	=head3 C<poll> is buggy
		5126
		5127	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		5128	implementation by something calling C<kqueue> internally around the 10.5.6
		5129	release, so now C<kqueue> I<and> C<poll> are broken.
		5130
		5131	Libev tries to work around this by not using C<poll> by default on
		5132	this rotten platform, but of course you can still ask for it when creating
		5133	a loop.
		5134
		5135	=head3 C<select> is buggy
		5136
		5137	All that's left is C<select>, and of course Apple found a way to fuck this
		5138	one up as well: On OS/X, C<select> actively limits the number of file
		5139	descriptors you can pass in to 1024 - your program suddenly crashes when
		5140	you use more.
		5141
		5142	There is an undocumented "workaround" for this - defining
		5143	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		5144	work on OS/X.
		5145
		5146	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		5147
		5148	=head3 C<errno> reentrancy
		5149
		5150	The default compile environment on Solaris is unfortunately so
		5151	thread-unsafe that you can't even use components/libraries compiled
		5152	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		5153	defined by default. A valid, if stupid, implementation choice.
		5154
		5155	If you want to use libev in threaded environments you have to make sure
		5156	it's compiled with C<_REENTRANT> defined.
		5157
		5158	=head3 Event port backend
		5159
		5160	The scalable event interface for Solaris is called "event
		5161	ports". Unfortunately, this mechanism is very buggy in all major
		5162	releases. If you run into high CPU usage, your program freezes or you get
		5163	a large number of spurious wakeups, make sure you have all the relevant
		5164	and latest kernel patches applied. No, I don't know which ones, but there
		5165	are multiple ones to apply, and afterwards, event ports actually work
		5166	great.
		5167
		5168	If you can't get it to work, you can try running the program by setting
		5169	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		5170	C<select> backends.
		5171
		5172	=head2 AIX POLL BUG
		5173
		5174	AIX unfortunately has a broken C<poll.h> header. Libev works around
		5175	this by trying to avoid the poll backend altogether (i.e. it's not even
		5176	compiled in), which normally isn't a big problem as C<select> works fine
		5177	with large bitsets on AIX, and AIX is dead anyway.
		5178
		5179	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		5180
		5181	=head3 General issues
		5182
		5183	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		5184	requires, and its I/O model is fundamentally incompatible with the POSIX
		5185	model. Libev still offers limited functionality on this platform in
		5186	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		5187	descriptors. This only applies when using Win32 natively, not when using
		5188	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		5189	as every compiler comes with a slightly differently broken/incompatible
		5190	environment.
		5191
		5192	Lifting these limitations would basically require the full
		5193	re-implementation of the I/O system. If you are into this kind of thing,
		5194	then note that glib does exactly that for you in a very portable way (note
		5195	also that glib is the slowest event library known to man).
		5196
		5197	There is no supported compilation method available on windows except
		5198	embedding it into other applications.
		5199
		5200	Sensible signal handling is officially unsupported by Microsoft - libev
		5201	tries its best, but under most conditions, signals will simply not work.
		5202
		5203	Not a libev limitation but worth mentioning: windows apparently doesn't
		5204	accept large writes: instead of resulting in a partial write, windows will
		5205	either accept everything or return C<ENOBUFS> if the buffer is too large,
		5206	so make sure you only write small amounts into your sockets (less than a
		5207	megabyte seems safe, but this apparently depends on the amount of memory
		5208	available).
		5209
		5210	Due to the many, low, and arbitrary limits on the win32 platform and
		5211	the abysmal performance of winsockets, using a large number of sockets
		5212	is not recommended (and not reasonable). If your program needs to use
		5213	more than a hundred or so sockets, then likely it needs to use a totally
		5214	different implementation for windows, as libev offers the POSIX readiness
		5215	notification model, which cannot be implemented efficiently on windows
		5216	(due to Microsoft monopoly games).
		5217
		5218	A typical way to use libev under windows is to embed it (see the embedding
		5219	section for details) and use the following F<evwrap.h> header file instead
		5220	of F<ev.h>:
		5221
		5222	#define EV_STANDALONE /* keeps ev from requiring config.h */
		5223	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		5224
		5225	#include "ev.h"
		5226
		5227	And compile the following F<evwrap.c> file into your project (make sure
		5228	you do I<not> compile the F<ev.c> or any other embedded source files!):
		5229
		5230	#include "evwrap.h"
		5231	#include "ev.c"
		5232
		5233	=head3 The winsocket C<select> function
		5234
		5235	The winsocket C<select> function doesn't follow POSIX in that it
		5236	requires socket I<handles> and not socket I<file descriptors> (it is
		5237	also extremely buggy). This makes select very inefficient, and also
		5238	requires a mapping from file descriptors to socket handles (the Microsoft
		5239	C runtime provides the function C<_open_osfhandle> for this). See the
		5240	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		5241	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		5242
		5243	The configuration for a "naked" win32 using the Microsoft runtime
		5244	libraries and raw winsocket select is:
		5245
		5246	#define EV_USE_SELECT 1
		5247	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		5248
		5249	Note that winsockets handling of fd sets is O(n), so you can easily get a
		5250	complexity in the O(n²) range when using win32.
		5251
		5252	=head3 Limited number of file descriptors
		5253
		5254	Windows has numerous arbitrary (and low) limits on things.
		5255
		5256	Early versions of winsocket's select only supported waiting for a maximum
		5257	of C<64> handles (probably owning to the fact that all windows kernels
		5258	can only wait for C<64> things at the same time internally; Microsoft
		5259	recommends spawning a chain of threads and wait for 63 handles and the
		5260	previous thread in each. Sounds great!).
		5261
		5262	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		5263	to some high number (e.g. C<2048>) before compiling the winsocket select
		5264	call (which might be in libev or elsewhere, for example, perl and many
		5265	other interpreters do their own select emulation on windows).
		5266
		5267	Another limit is the number of file descriptors in the Microsoft runtime
		5268	libraries, which by default is C<64> (there must be a hidden I<64>
		5269	fetish or something like this inside Microsoft). You can increase this
		5270	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		5271	(another arbitrary limit), but is broken in many versions of the Microsoft
		5272	runtime libraries. This might get you to about C<512> or C<2048> sockets
		5273	(depending on windows version and/or the phase of the moon). To get more,
		5274	you need to wrap all I/O functions and provide your own fd management, but
		5275	the cost of calling select (O(n²)) will likely make this unworkable.
		5276
		5277	=head2 PORTABILITY REQUIREMENTS
		5278
		5279	In addition to a working ISO-C implementation and of course the
		5280	backend-specific APIs, libev relies on a few additional extensions:
		5281
		5282	=over 4
		5283
		5284	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		5285	calling conventions regardless of C<ev_watcher_type *>.
		5286
		5287	Libev assumes not only that all watcher pointers have the same internal
		5288	structure (guaranteed by POSIX but not by ISO C for example), but it also
		5289	assumes that the same (machine) code can be used to call any watcher
		5290	callback: The watcher callbacks have different type signatures, but libev
		5291	calls them using an C<ev_watcher *> internally.
		5292
		5293	=item pointer accesses must be thread-atomic
		5294
		5295	Accessing a pointer value must be atomic, it must both be readable and
		5296	writable in one piece - this is the case on all current architectures.
		5297
		5298	=item C<sig_atomic_t volatile> must be thread-atomic as well
		5299
		5300	The type C<sig_atomic_t volatile> (or whatever is defined as
		5301	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		5302	threads. This is not part of the specification for C<sig_atomic_t>, but is
		5303	believed to be sufficiently portable.
		5304
		5305	=item C<sigprocmask> must work in a threaded environment
		5306
		5307	Libev uses C<sigprocmask> to temporarily block signals. This is not
		5308	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		5309	pthread implementations will either allow C<sigprocmask> in the "main
		5310	thread" or will block signals process-wide, both behaviours would
		5311	be compatible with libev. Interaction between C<sigprocmask> and
		5312	C<pthread_sigmask> could complicate things, however.
		5313
		5314	The most portable way to handle signals is to block signals in all threads
		5315	except the initial one, and run the signal handling loop in the initial
		5316	thread as well.
		5317
		5318	=item C<long> must be large enough for common memory allocation sizes
		5319
		5320	To improve portability and simplify its API, libev uses C<long> internally
		5321	instead of C<size_t> when allocating its data structures. On non-POSIX
		5322	systems (Microsoft...) this might be unexpectedly low, but is still at
		5323	least 31 bits everywhere, which is enough for hundreds of millions of
		5324	watchers.
		5325
		5326	=item C<double> must hold a time value in seconds with enough accuracy
		5327
		5328	The type C<double> is used to represent timestamps. It is required to
		5329	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		5330	good enough for at least into the year 4000 with millisecond accuracy
		5331	(the design goal for libev). This requirement is overfulfilled by
		5332	implementations using IEEE 754, which is basically all existing ones.
		5333
		5334	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5335	year 2255 (and millisecond accuracy till the year 287396 - by then, libev
		5336	is either obsolete or somebody patched it to use C<long double> or
		5337	something like that, just kidding).
		5338
		5339	=back
		5340
		5341	If you know of other additional requirements drop me a note.
		5342
		5343
2465	=head1 COMPLEXITIES	5344	=head1 ALGORITHMIC COMPLEXITIES
2466		5345
2467	In this section the complexities of (many of) the algorithms used inside	5346	In this section the complexities of (many of) the algorithms used inside
2468	libev will be explained. For complexity discussions about backends see the	5347	libev will be documented. For complexity discussions about backends see
2469	documentation for C<ev_default_init>.	5348	the documentation for C<ev_default_init>.
2470		5349
2471	All of the following are about amortised time: If an array needs to be	5350	All of the following are about amortised time: If an array needs to be
2472	extended, libev needs to realloc and move the whole array, but this	5351	extended, libev needs to realloc and move the whole array, but this
2473	happens asymptotically never with higher number of elements, so O(1) might	5352	happens asymptotically rarer with higher number of elements, so O(1) might
2474	mean it might do a lengthy realloc operation in rare cases, but on average	5353	mean that libev does a lengthy realloc operation in rare cases, but on
2475	it is much faster and asymptotically approaches constant time.	5354	average it is much faster and asymptotically approaches constant time.
2476		5355
2477	=over 4	5356	=over 4
2478		5357
2479	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	5358	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2480		5359
2481	This means that, when you have a watcher that triggers in one hour and	5360	This means that, when you have a watcher that triggers in one hour and
2482	there are 100 watchers that would trigger before that then inserting will	5361	there are 100 watchers that would trigger before that, then inserting will
2483	have to skip those 100 watchers.	5362	have to skip roughly seven (C<ld 100>) of these watchers.
2484		5363
2485	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	5364	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2486		5365
2487	That means that for changing a timer costs less than removing/adding them	5366	That means that changing a timer costs less than removing/adding them,
2488	as only the relative motion in the event queue has to be paid for.	5367	as only the relative motion in the event queue has to be paid for.
2489		5368
2490	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	5369	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2491		5370
2492	These just add the watcher into an array or at the head of a list.	5371	These just add the watcher into an array or at the head of a list.
		5372
2493	=item Stopping check/prepare/idle watchers: O(1)	5373	=item Stopping check/prepare/idle/fork/async watchers: O(1)
2494		5374
2495	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	5375	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2496		5376
2497	These watchers are stored in lists then need to be walked to find the	5377	These watchers are stored in lists, so they need to be walked to find the
2498	correct watcher to remove. The lists are usually short (you don't usually	5378	correct watcher to remove. The lists are usually short (you don't usually
2499	have many watchers waiting for the same fd or signal).	5379	have many watchers waiting for the same fd or signal: one is typical, two
		5380	is rare).
2500		5381
2501	=item Finding the next timer per loop iteration: O(1)	5382	=item Finding the next timer in each loop iteration: O(1)
		5383
		5384	By virtue of using a binary or 4-heap, the next timer is always found at a
		5385	fixed position in the storage array.
2502		5386
2503	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	5387	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2504		5388
2505	A change means an I/O watcher gets started or stopped, which requires	5389	A change means an I/O watcher gets started or stopped, which requires
2506	libev to recalculate its status (and possibly tell the kernel).	5390	libev to recalculate its status (and possibly tell the kernel, depending
		5391	on backend and whether C<ev_io_set> was used).
2507		5392
2508	=item Activating one watcher: O(1)	5393	=item Activating one watcher (putting it into the pending state): O(1)
2509		5394
2510	=item Priority handling: O(number_of_priorities)	5395	=item Priority handling: O(number_of_priorities)
2511		5396
2512	Priorities are implemented by allocating some space for each	5397	Priorities are implemented by allocating some space for each
2513	priority. When doing priority-based operations, libev usually has to	5398	priority. When doing priority-based operations, libev usually has to
2514	linearly search all the priorities.	5399	linearly search all the priorities, but starting/stopping and activating
		5400	watchers becomes O(1) with respect to priority handling.
		5401
		5402	=item Sending an ev_async: O(1)
		5403
		5404	=item Processing ev_async_send: O(number_of_async_watchers)
		5405
		5406	=item Processing signals: O(max_signal_number)
		5407
		5408	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5409	calls in the current loop iteration and the loop is currently
		5410	blocked. Checking for async and signal events involves iterating over all
		5411	running async watchers or all signal numbers.
2515		5412
2516	=back	5413	=back
2517		5414
2518		5415
		5416	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5417
		5418	The major version 4 introduced some incompatible changes to the API.
		5419
		5420	At the moment, the C<ev.h> header file provides compatibility definitions
		5421	for all changes, so most programs should still compile. The compatibility
		5422	layer might be removed in later versions of libev, so better update to the
		5423	new API early than late.
		5424
		5425	=over 4
		5426
		5427	=item C<EV_COMPAT3> backwards compatibility mechanism
		5428
		5429	The backward compatibility mechanism can be controlled by
		5430	C<EV_COMPAT3>. See L</PREPROCESSOR SYMBOLS/MACROS> in the L</EMBEDDING>
		5431	section.
		5432
		5433	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5434
		5435	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5436
		5437	ev_loop_destroy (EV_DEFAULT_UC);
		5438	ev_loop_fork (EV_DEFAULT);
		5439
		5440	=item function/symbol renames
		5441
		5442	A number of functions and symbols have been renamed:
		5443
		5444	ev_loop => ev_run
		5445	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5446	EVLOOP_ONESHOT => EVRUN_ONCE
		5447
		5448	ev_unloop => ev_break
		5449	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5450	EVUNLOOP_ONE => EVBREAK_ONE
		5451	EVUNLOOP_ALL => EVBREAK_ALL
		5452
		5453	EV_TIMEOUT => EV_TIMER
		5454
		5455	ev_loop_count => ev_iteration
		5456	ev_loop_depth => ev_depth
		5457	ev_loop_verify => ev_verify
		5458
		5459	Most functions working on C<struct ev_loop> objects don't have an
		5460	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5461	associated constants have been renamed to not collide with the C<struct
		5462	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5463	as all other watcher types. Note that C<ev_loop_fork> is still called
		5464	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5465	typedef.
		5466
		5467	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5468
		5469	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5470	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5471	and work, but the library code will of course be larger.
		5472
		5473	=back
		5474
		5475
		5476	=head1 GLOSSARY
		5477
		5478	=over 4
		5479
		5480	=item active
		5481
		5482	A watcher is active as long as it has been started and not yet stopped.
		5483	See L</WATCHER STATES> for details.
		5484
		5485	=item application
		5486
		5487	In this document, an application is whatever is using libev.
		5488
		5489	=item backend
		5490
		5491	The part of the code dealing with the operating system interfaces.
		5492
		5493	=item callback
		5494
		5495	The address of a function that is called when some event has been
		5496	detected. Callbacks are being passed the event loop, the watcher that
		5497	received the event, and the actual event bitset.
		5498
		5499	=item callback/watcher invocation
		5500
		5501	The act of calling the callback associated with a watcher.
		5502
		5503	=item event
		5504
		5505	A change of state of some external event, such as data now being available
		5506	for reading on a file descriptor, time having passed or simply not having
		5507	any other events happening anymore.
		5508
		5509	In libev, events are represented as single bits (such as C<EV_READ> or
		5510	C<EV_TIMER>).
		5511
		5512	=item event library
		5513
		5514	A software package implementing an event model and loop.
		5515
		5516	=item event loop
		5517
		5518	An entity that handles and processes external events and converts them
		5519	into callback invocations.
		5520
		5521	=item event model
		5522
		5523	The model used to describe how an event loop handles and processes
		5524	watchers and events.
		5525
		5526	=item pending
		5527
		5528	A watcher is pending as soon as the corresponding event has been
		5529	detected. See L</WATCHER STATES> for details.
		5530
		5531	=item real time
		5532
		5533	The physical time that is observed. It is apparently strictly monotonic :)
		5534
		5535	=item wall-clock time
		5536
		5537	The time and date as shown on clocks. Unlike real time, it can actually
		5538	be wrong and jump forwards and backwards, e.g. when you adjust your
		5539	clock.
		5540
		5541	=item watcher
		5542
		5543	A data structure that describes interest in certain events. Watchers need
		5544	to be started (attached to an event loop) before they can receive events.
		5545
		5546	=back
		5547
2519	=head1 AUTHOR	5548	=head1 AUTHOR
2520		5549
2521	Marc Lehmann <libev@schmorp.de>.	5550	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5551	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
2522		5552

Diff Legend

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