ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.99 by root, Sat Dec 22 06:16:36 2007 UTC vs.
Revision 1.443 by root, Thu Aug 30 21:51:15 2018 UTC

		1	=encoding utf-8
		2
1	=head1 NAME	3	=head1 NAME
2		4
3	libev - a high performance full-featured event loop written in C	5	libev - a high performance full-featured event loop written in C
4		6
5	=head1 SYNOPSIS	7	=head1 SYNOPSIS
6		8
7	#include <ev.h>	9	#include <ev.h>
8		10
9	=head1 EXAMPLE PROGRAM	11	=head2 EXAMPLE PROGRAM
10		12
		13	// a single header file is required
11	#include <ev.h>	14	#include <ev.h>
12		15
		16	#include <stdio.h> // for puts
		17
		18	// every watcher type has its own typedef'd struct
		19	// with the name ev_TYPE
13	ev_io stdin_watcher;	20	ev_io stdin_watcher;
14	ev_timer timeout_watcher;	21	ev_timer timeout_watcher;
15		22
16	/* called when data readable on stdin */	23	// all watcher callbacks have a similar signature
		24	// this callback is called when data is readable on stdin
17	static void	25	static void
18	stdin_cb (EV_P_ struct ev_io *w, int revents)	26	stdin_cb (EV_P_ ev_io *w, int revents)
19	{	27	{
20	/* puts ("stdin ready"); */	28	puts ("stdin ready");
21	ev_io_stop (EV_A_ w); /* just a syntax example */	29	// for one-shot events, one must manually stop the watcher
22	ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */	30	// with its corresponding stop function.
		31	ev_io_stop (EV_A_ w);
		32
		33	// this causes all nested ev_run's to stop iterating
		34	ev_break (EV_A_ EVBREAK_ALL);
23	}	35	}
24		36
		37	// another callback, this time for a time-out
25	static void	38	static void
26	timeout_cb (EV_P_ struct ev_timer *w, int revents)	39	timeout_cb (EV_P_ ev_timer *w, int revents)
27	{	40	{
28	/* puts ("timeout"); */	41	puts ("timeout");
29	ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */	42	// this causes the innermost ev_run to stop iterating
		43	ev_break (EV_A_ EVBREAK_ONE);
30	}	44	}
31		45
32	int	46	int
33	main (void)	47	main (void)
34	{	48	{
35	struct ev_loop *loop = ev_default_loop (0);	49	// use the default event loop unless you have special needs
		50	struct ev_loop *loop = EV_DEFAULT;
36		51
37	/* initialise an io watcher, then start it */	52	// initialise an io watcher, then start it
		53	// this one will watch for stdin to become readable
38	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	54	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
39	ev_io_start (loop, &stdin_watcher);	55	ev_io_start (loop, &stdin_watcher);
40		56
		57	// initialise a timer watcher, then start it
41	/* simple non-repeating 5.5 second timeout */	58	// simple non-repeating 5.5 second timeout
42	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);	59	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
43	ev_timer_start (loop, &timeout_watcher);	60	ev_timer_start (loop, &timeout_watcher);
44		61
45	/* loop till timeout or data ready */	62	// now wait for events to arrive
46	ev_loop (loop, 0);	63	ev_run (loop, 0);
47		64
		65	// break was called, so exit
48	return 0;	66	return 0;
49	}	67	}
50		68
51	=head1 DESCRIPTION	69	=head1 ABOUT THIS DOCUMENT
52		70
		71	This document documents the libev software package.
		72
53	The newest version of this document is also available as a html-formatted	73	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	74	web page you might find easier to navigate when reading it for the first
55	time: L<http://cvs.schmorp.de/libev/ev.html>.	75	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		76
		77	While this document tries to be as complete as possible in documenting
		78	libev, its usage and the rationale behind its design, it is not a tutorial
		79	on event-based programming, nor will it introduce event-based programming
		80	with libev.
		81
		82	Familiarity with event based programming techniques in general is assumed
		83	throughout this document.
		84
		85	=head1 WHAT TO READ WHEN IN A HURRY
		86
		87	This manual tries to be very detailed, but unfortunately, this also makes
		88	it very long. If you just want to know the basics of libev, I suggest
		89	reading L</ANATOMY OF A WATCHER>, then the L</EXAMPLE PROGRAM> above and
		90	look up the missing functions in L</GLOBAL FUNCTIONS> and the C<ev_io> and
		91	C<ev_timer> sections in L</WATCHER TYPES>.
		92
		93	=head1 ABOUT LIBEV
56		94
57	Libev is an event loop: you register interest in certain events (such as a	95	Libev is an event loop: you register interest in certain events (such as a
58	file descriptor being readable or a timeout occurring), and it will manage	96	file descriptor being readable or a timeout occurring), and it will manage
59	these event sources and provide your program with events.	97	these event sources and provide your program with events.
60		98
…		…
65	You register interest in certain events by registering so-called I<event	103	You register interest in certain events by registering so-called I<event
66	watchers>, which are relatively small C structures you initialise with the	104	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	105	details of the event, and then hand it over to libev by I<starting> the
68	watcher.	106	watcher.
69		107
70	=head1 FEATURES	108	=head2 FEATURES
71		109
72	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	110	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	111	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	112	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers	113	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
76	with customised rescheduling (C<ev_periodic>), synchronous signals	114	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	115	timers (C<ev_timer>), absolute timers with customised rescheduling
78	watchers dealing with the event loop mechanism itself (C<ev_idle>,	116	(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	117	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	118	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
81	(C<ev_fork>).	119	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		120	limited support for fork events (C<ev_fork>).
82		121
83	It also is quite fast (see this	122	It also is quite fast (see this
84	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent	123	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85	for example).	124	for example).
86		125
87	=head1 CONVENTIONS	126	=head2 CONVENTIONS
88		127
89	Libev is very configurable. In this manual the default configuration will	128	Libev is very configurable. In this manual the default (and most common)
90	be described, which supports multiple event loops. For more info about	129	configuration will be described, which supports multiple event loops. For
91	various configuration options please have a look at B<EMBED> section in	130	more info about various configuration options please have a look at
92	this manual. If libev was configured without support for multiple event	131	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>	132	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.	133	name C<loop> (which is always of type C<struct ev_loop *>) will not have
		134	this argument.
95		135
96	=head1 TIME REPRESENTATION	136	=head2 TIME REPRESENTATION
97		137
98	Libev represents time as a single floating point number, representing the	138	Libev represents time as a single floating point number, representing
99	(fractional) number of seconds since the (POSIX) epoch (somewhere near	139	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	140	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	141	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	142	too. It usually aliases to the C<double> type in C. When you need to do
103	it, you should treat it as some floatingpoint value. Unlike the name	143	any calculations on it, you should treat it as some floating point value.
		144
104	component C<stamp> might indicate, it is also used for time differences	145	Unlike the name component C<stamp> might indicate, it is also used for
105	throughout libev.	146	time differences (e.g. delays) throughout libev.
		147
		148	=head1 ERROR HANDLING
		149
		150	Libev knows three classes of errors: operating system errors, usage errors
		151	and internal errors (bugs).
		152
		153	When libev catches an operating system error it cannot handle (for example
		154	a system call indicating a condition libev cannot fix), it calls the callback
		155	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		156	abort. The default is to print a diagnostic message and to call C<abort
		157	()>.
		158
		159	When libev detects a usage error such as a negative timer interval, then
		160	it will print a diagnostic message and abort (via the C<assert> mechanism,
		161	so C<NDEBUG> will disable this checking): these are programming errors in
		162	the libev caller and need to be fixed there.
		163
		164	Libev also has a few internal error-checking C<assert>ions, and also has
		165	extensive consistency checking code. These do not trigger under normal
		166	circumstances, as they indicate either a bug in libev or worse.
		167
106		168
107	=head1 GLOBAL FUNCTIONS	169	=head1 GLOBAL FUNCTIONS
108		170
109	These functions can be called anytime, even before initialising the	171	These functions can be called anytime, even before initialising the
110	library in any way.	172	library in any way.
…		…
113		175
114	=item ev_tstamp ev_time ()	176	=item ev_tstamp ev_time ()
115		177
116	Returns the current time as libev would use it. Please note that the	178	Returns the current time as libev would use it. Please note that the
117	C<ev_now> function is usually faster and also often returns the timestamp	179	C<ev_now> function is usually faster and also often returns the timestamp
118	you actually want to know.	180	you actually want to know. Also interesting is the combination of
		181	C<ev_now_update> and C<ev_now>.
119		182
120	=item ev_sleep (ev_tstamp interval)	183	=item ev_sleep (ev_tstamp interval)
121		184
122	Sleep for the given interval: The current thread will be blocked until	185	Sleep for the given interval: The current thread will be blocked
123	either it is interrupted or the given time interval has passed. Basically	186	until either it is interrupted or the given time interval has
		187	passed (approximately - it might return a bit earlier even if not
		188	interrupted). Returns immediately if C<< interval <= 0 >>.
		189
124	this is a subsecond-resolution C<sleep ()>.	190	Basically this is a sub-second-resolution C<sleep ()>.
		191
		192	The range of the C<interval> is limited - libev only guarantees to work
		193	with sleep times of up to one day (C<< interval <= 86400 >>).
125		194
126	=item int ev_version_major ()	195	=item int ev_version_major ()
127		196
128	=item int ev_version_minor ()	197	=item int ev_version_minor ()
129		198
…		…
140	as this indicates an incompatible change. Minor versions are usually	209	as this indicates an incompatible change. Minor versions are usually
141	compatible to older versions, so a larger minor version alone is usually	210	compatible to older versions, so a larger minor version alone is usually
142	not a problem.	211	not a problem.
143		212
144	Example: Make sure we haven't accidentally been linked against the wrong	213	Example: Make sure we haven't accidentally been linked against the wrong
145	version.	214	version (note, however, that this will not detect other ABI mismatches,
		215	such as LFS or reentrancy).
146		216
147	assert (("libev version mismatch",	217	assert (("libev version mismatch",
148	ev_version_major () == EV_VERSION_MAJOR	218	ev_version_major () == EV_VERSION_MAJOR
149	&& ev_version_minor () >= EV_VERSION_MINOR));	219	&& ev_version_minor () >= EV_VERSION_MINOR));
150		220
151	=item unsigned int ev_supported_backends ()	221	=item unsigned int ev_supported_backends ()
152		222
153	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	223	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
154	value) compiled into this binary of libev (independent of their	224	value) compiled into this binary of libev (independent of their
…		…
156	a description of the set values.	226	a description of the set values.
157		227
158	Example: make sure we have the epoll method, because yeah this is cool and	228	Example: make sure we have the epoll method, because yeah this is cool and
159	a must have and can we have a torrent of it please!!!11	229	a must have and can we have a torrent of it please!!!11
160		230
161	assert (("sorry, no epoll, no sex",	231	assert (("sorry, no epoll, no sex",
162	ev_supported_backends () & EVBACKEND_EPOLL));	232	ev_supported_backends () & EVBACKEND_EPOLL));
163		233
164	=item unsigned int ev_recommended_backends ()	234	=item unsigned int ev_recommended_backends ()
165		235
166	Return the set of all backends compiled into this binary of libev and also	236	Return the set of all backends compiled into this binary of libev and
167	recommended for this platform. This set is often smaller than the one	237	also recommended for this platform, meaning it will work for most file
		238	descriptor types. This set is often smaller than the one returned by
168	returned by C<ev_supported_backends>, as for example kqueue is broken on	239	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
169	most BSDs and will not be autodetected unless you explicitly request it	240	and will not be auto-detected unless you explicitly request it (assuming
170	(assuming you know what you are doing). This is the set of backends that	241	you know what you are doing). This is the set of backends that libev will
171	libev will probe for if you specify no backends explicitly.	242	probe for if you specify no backends explicitly.
172		243
173	=item unsigned int ev_embeddable_backends ()	244	=item unsigned int ev_embeddable_backends ()
174		245
175	Returns the set of backends that are embeddable in other event loops. This	246	Returns the set of backends that are embeddable in other event loops. This
176	is the theoretical, all-platform, value. To find which backends	247	value is platform-specific but can include backends not available on the
177	might be supported on the current system, you would need to look at	248	current system. To find which embeddable backends might be supported on
178	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	249	the current system, you would need to look at C<ev_embeddable_backends ()
179	recommended ones.	250	& ev_supported_backends ()>, likewise for recommended ones.
180		251
181	See the description of C<ev_embed> watchers for more info.	252	See the description of C<ev_embed> watchers for more info.
182		253
183	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	254	=item ev_set_allocator (void *(*cb)(void *ptr, long size) throw ())
184		255
185	Sets the allocation function to use (the prototype is similar - the	256	Sets the allocation function to use (the prototype is similar - the
186	semantics is identical - to the realloc C function). It is used to	257	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
187	allocate and free memory (no surprises here). If it returns zero when	258	used to allocate and free memory (no surprises here). If it returns zero
188	memory needs to be allocated, the library might abort or take some	259	when memory needs to be allocated (C<size != 0>), the library might abort
189	potentially destructive action. The default is your system realloc	260	or take some potentially destructive action.
190	function.	261
		262	Since some systems (at least OpenBSD and Darwin) fail to implement
		263	correct C<realloc> semantics, libev will use a wrapper around the system
		264	C<realloc> and C<free> functions by default.
191		265
192	You could override this function in high-availability programs to, say,	266	You could override this function in high-availability programs to, say,
193	free some memory if it cannot allocate memory, to use a special allocator,	267	free some memory if it cannot allocate memory, to use a special allocator,
194	or even to sleep a while and retry until some memory is available.	268	or even to sleep a while and retry until some memory is available.
195		269
196	Example: Replace the libev allocator with one that waits a bit and then	270	Example: Replace the libev allocator with one that waits a bit and then
197	retries).	271	retries (example requires a standards-compliant C<realloc>).
198		272
199	static void *	273	static void *
200	persistent_realloc (void *ptr, size_t size)	274	persistent_realloc (void *ptr, size_t size)
201	{	275	{
202	for (;;)	276	for (;;)
…		…
211	}	285	}
212		286
213	...	287	...
214	ev_set_allocator (persistent_realloc);	288	ev_set_allocator (persistent_realloc);
215		289
216	=item ev_set_syserr_cb (void (*cb)(const char *msg));	290	=item ev_set_syserr_cb (void (*cb)(const char *msg) throw ())
217		291
218	Set the callback function to call on a retryable syscall error (such	292	Set the callback function to call on a retryable system call error (such
219	as failed select, poll, epoll_wait). The message is a printable string	293	as failed select, poll, epoll_wait). The message is a printable string
220	indicating the system call or subsystem causing the problem. If this	294	indicating the system call or subsystem causing the problem. If this
221	callback is set, then libev will expect it to remedy the sitution, no	295	callback is set, then libev will expect it to remedy the situation, no
222	matter what, when it returns. That is, libev will generally retry the	296	matter what, when it returns. That is, libev will generally retry the
223	requested operation, or, if the condition doesn't go away, do bad stuff	297	requested operation, or, if the condition doesn't go away, do bad stuff
224	(such as abort).	298	(such as abort).
225		299
226	Example: This is basically the same thing that libev does internally, too.	300	Example: This is basically the same thing that libev does internally, too.
…		…
233	}	307	}
234		308
235	...	309	...
236	ev_set_syserr_cb (fatal_error);	310	ev_set_syserr_cb (fatal_error);
237		311
		312	=item ev_feed_signal (int signum)
		313
		314	This function can be used to "simulate" a signal receive. It is completely
		315	safe to call this function at any time, from any context, including signal
		316	handlers or random threads.
		317
		318	Its main use is to customise signal handling in your process, especially
		319	in the presence of threads. For example, you could block signals
		320	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		321	creating any loops), and in one thread, use C<sigwait> or any other
		322	mechanism to wait for signals, then "deliver" them to libev by calling
		323	C<ev_feed_signal>.
		324
238	=back	325	=back
239		326
240	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	327	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
241		328
242	An event loop is described by a C<struct ev_loop *>. The library knows two	329	An event loop is described by a C<struct ev_loop *> (the C<struct> is
243	types of such loops, the I<default> loop, which supports signals and child	330	I<not> optional in this case unless libev 3 compatibility is disabled, as
244	events, and dynamically created loops which do not.	331	libev 3 had an C<ev_loop> function colliding with the struct name).
245		332
246	If you use threads, a common model is to run the default event loop	333	The library knows two types of such loops, the I<default> loop, which
247	in your main thread (or in a separate thread) and for each thread you	334	supports child process events, and dynamically created event loops which
248	create, you also create another event loop. Libev itself does no locking	335	do not.
249	whatsoever, so if you mix calls to the same event loop in different
250	threads, make sure you lock (this is usually a bad idea, though, even if
251	done correctly, because it's hideous and inefficient).
252		336
253	=over 4	337	=over 4
254		338
255	=item struct ev_loop *ev_default_loop (unsigned int flags)	339	=item struct ev_loop *ev_default_loop (unsigned int flags)
256		340
257	This will initialise the default event loop if it hasn't been initialised	341	This returns the "default" event loop object, which is what you should
258	yet and return it. If the default loop could not be initialised, returns	342	normally use when you just need "the event loop". Event loop objects and
259	false. If it already was initialised it simply returns it (and ignores the	343	the C<flags> parameter are described in more detail in the entry for
260	flags. If that is troubling you, check C<ev_backend ()> afterwards).	344	C<ev_loop_new>.
		345
		346	If the default loop is already initialised then this function simply
		347	returns it (and ignores the flags. If that is troubling you, check
		348	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		349	flags, which should almost always be C<0>, unless the caller is also the
		350	one calling C<ev_run> or otherwise qualifies as "the main program".
261		351
262	If you don't know what event loop to use, use the one returned from this	352	If you don't know what event loop to use, use the one returned from this
263	function.	353	function (or via the C<EV_DEFAULT> macro).
		354
		355	Note that this function is I<not> thread-safe, so if you want to use it
		356	from multiple threads, you have to employ some kind of mutex (note also
		357	that this case is unlikely, as loops cannot be shared easily between
		358	threads anyway).
		359
		360	The default loop is the only loop that can handle C<ev_child> watchers,
		361	and to do this, it always registers a handler for C<SIGCHLD>. If this is
		362	a problem for your application you can either create a dynamic loop with
		363	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
		364	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
		365
		366	Example: This is the most typical usage.
		367
		368	if (!ev_default_loop (0))
		369	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		370
		371	Example: Restrict libev to the select and poll backends, and do not allow
		372	environment settings to be taken into account:
		373
		374	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		375
		376	=item struct ev_loop *ev_loop_new (unsigned int flags)
		377
		378	This will create and initialise a new event loop object. If the loop
		379	could not be initialised, returns false.
		380
		381	This function is thread-safe, and one common way to use libev with
		382	threads is indeed to create one loop per thread, and using the default
		383	loop in the "main" or "initial" thread.
264		384
265	The flags argument can be used to specify special behaviour or specific	385	The flags argument can be used to specify special behaviour or specific
266	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	386	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
267		387
268	The following flags are supported:	388	The following flags are supported:
…		…
274	The default flags value. Use this if you have no clue (it's the right	394	The default flags value. Use this if you have no clue (it's the right
275	thing, believe me).	395	thing, believe me).
276		396
277	=item C<EVFLAG_NOENV>	397	=item C<EVFLAG_NOENV>
278		398
279	If this flag bit is ored into the flag value (or the program runs setuid	399	If this flag bit is or'ed into the flag value (or the program runs setuid
280	or setgid) then libev will I<not> look at the environment variable	400	or setgid) then libev will I<not> look at the environment variable
281	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	401	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
282	override the flags completely if it is found in the environment. This is	402	override the flags completely if it is found in the environment. This is
283	useful to try out specific backends to test their performance, or to work	403	useful to try out specific backends to test their performance, to work
284	around bugs.	404	around bugs, or to make libev threadsafe (accessing environment variables
		405	cannot be done in a threadsafe way, but usually it works if no other
		406	thread modifies them).
285		407
286	=item C<EVFLAG_FORKCHECK>	408	=item C<EVFLAG_FORKCHECK>
287		409
288	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after	410	Instead of calling C<ev_loop_fork> manually after a fork, you can also
289	a fork, you can also make libev check for a fork in each iteration by	411	make libev check for a fork in each iteration by enabling this flag.
290	enabling this flag.
291		412
292	This works by calling C<getpid ()> on every iteration of the loop,	413	This works by calling C<getpid ()> on every iteration of the loop,
293	and thus this might slow down your event loop if you do a lot of loop	414	and thus this might slow down your event loop if you do a lot of loop
294	iterations and little real work, but is usually not noticeable (on my	415	iterations and little real work, but is usually not noticeable (on my
295	Linux system for example, C<getpid> is actually a simple 5-insn sequence	416	GNU/Linux system for example, C<getpid> is actually a simple 5-insn
296	without a syscall and thus I<very> fast, but my Linux system also has	417	sequence without a system call and thus I<very> fast, but my GNU/Linux
297	C<pthread_atfork> which is even faster).	418	system also has C<pthread_atfork> which is even faster). (Update: glibc
		419	versions 2.25 apparently removed the C<getpid> optimisation again).
298		420
299	The big advantage of this flag is that you can forget about fork (and	421	The big advantage of this flag is that you can forget about fork (and
300	forget about forgetting to tell libev about forking) when you use this	422	forget about forgetting to tell libev about forking, although you still
301	flag.	423	have to ignore C<SIGPIPE>) when you use this flag.
302		424
303	This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS>	425	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
304	environment variable.	426	environment variable.
		427
		428	=item C<EVFLAG_NOINOTIFY>
		429
		430	When this flag is specified, then libev will not attempt to use the
		431	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
		432	testing, this flag can be useful to conserve inotify file descriptors, as
		433	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		434
		435	=item C<EVFLAG_SIGNALFD>
		436
		437	When this flag is specified, then libev will attempt to use the
		438	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
		439	delivers signals synchronously, which makes it both faster and might make
		440	it possible to get the queued signal data. It can also simplify signal
		441	handling with threads, as long as you properly block signals in your
		442	threads that are not interested in handling them.
		443
		444	Signalfd will not be used by default as this changes your signal mask, and
		445	there are a lot of shoddy libraries and programs (glib's threadpool for
		446	example) that can't properly initialise their signal masks.
		447
		448	=item C<EVFLAG_NOSIGMASK>
		449
		450	When this flag is specified, then libev will avoid to modify the signal
		451	mask. Specifically, this means you have to make sure signals are unblocked
		452	when you want to receive them.
		453
		454	This behaviour is useful when you want to do your own signal handling, or
		455	want to handle signals only in specific threads and want to avoid libev
		456	unblocking the signals.
		457
		458	It's also required by POSIX in a threaded program, as libev calls
		459	C<sigprocmask>, whose behaviour is officially unspecified.
		460
		461	This flag's behaviour will become the default in future versions of libev.
305		462
306	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	463	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
307		464
308	This is your standard select(2) backend. Not I<completely> standard, as	465	This is your standard select(2) backend. Not I<completely> standard, as
309	libev tries to roll its own fd_set with no limits on the number of fds,	466	libev tries to roll its own fd_set with no limits on the number of fds,
310	but if that fails, expect a fairly low limit on the number of fds when	467	but if that fails, expect a fairly low limit on the number of fds when
311	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	468	using this backend. It doesn't scale too well (O(highest_fd)), but its
312	the fastest backend for a low number of fds.	469	usually the fastest backend for a low number of (low-numbered :) fds.
		470
		471	To get good performance out of this backend you need a high amount of
		472	parallelism (most of the file descriptors should be busy). If you are
		473	writing a server, you should C<accept ()> in a loop to accept as many
		474	connections as possible during one iteration. You might also want to have
		475	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		476	readiness notifications you get per iteration.
		477
		478	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		479	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		480	C<exceptfds> set on that platform).
313		481
314	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	482	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
315		483
316	And this is your standard poll(2) backend. It's more complicated than	484	And this is your standard poll(2) backend. It's more complicated
317	select, but handles sparse fds better and has no artificial limit on the	485	than select, but handles sparse fds better and has no artificial
318	number of fds you can use (except it will slow down considerably with a	486	limit on the number of fds you can use (except it will slow down
319	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	487	considerably with a lot of inactive fds). It scales similarly to select,
		488	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		489	performance tips.
		490
		491	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		492	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
320		493
321	=item C<EVBACKEND_EPOLL> (value 4, Linux)	494	=item C<EVBACKEND_EPOLL> (value 4, Linux)
322		495
		496	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		497	kernels).
		498
323	For few fds, this backend is a bit little slower than poll and select,	499	For few fds, this backend is a bit little slower than poll and select, but
324	but it scales phenomenally better. While poll and select usually scale	500	it scales phenomenally better. While poll and select usually scale like
325	like O(total_fds) where n is the total number of fds (or the highest fd),	501	O(total_fds) where total_fds is the total number of fds (or the highest
326	epoll scales either O(1) or O(active_fds). The epoll design has a number	502	fd), epoll scales either O(1) or O(active_fds).
327	of shortcomings, such as silently dropping events in some hard-to-detect	503
328	cases and rewiring a syscall per fd change, no fork support and bad	504	The epoll mechanism deserves honorable mention as the most misdesigned
329	support for dup:	505	of the more advanced event mechanisms: mere annoyances include silently
		506	dropping file descriptors, requiring a system call per change per file
		507	descriptor (and unnecessary guessing of parameters), problems with dup,
		508	returning before the timeout value, resulting in additional iterations
		509	(and only giving 5ms accuracy while select on the same platform gives
		510	0.1ms) and so on. The biggest issue is fork races, however - if a program
		511	forks then I<both> parent and child process have to recreate the epoll
		512	set, which can take considerable time (one syscall per file descriptor)
		513	and is of course hard to detect.
		514
		515	Epoll is also notoriously buggy - embedding epoll fds I<should> work,
		516	but of course I<doesn't>, and epoll just loves to report events for
		517	totally I<different> file descriptors (even already closed ones, so
		518	one cannot even remove them from the set) than registered in the set
		519	(especially on SMP systems). Libev tries to counter these spurious
		520	notifications by employing an additional generation counter and comparing
		521	that against the events to filter out spurious ones, recreating the set
		522	when required. Epoll also erroneously rounds down timeouts, but gives you
		523	no way to know when and by how much, so sometimes you have to busy-wait
		524	because epoll returns immediately despite a nonzero timeout. And last
		525	not least, it also refuses to work with some file descriptors which work
		526	perfectly fine with C<select> (files, many character devices...).
		527
		528	Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
		529	cobbled together in a hurry, no thought to design or interaction with
		530	others. Oh, the pain, will it ever stop...
330		531
331	While stopping, setting and starting an I/O watcher in the same iteration	532	While stopping, setting and starting an I/O watcher in the same iteration
332	will result in some caching, there is still a syscall per such incident	533	will result in some caching, there is still a system call per such
333	(because the fd could point to a different file description now), so its	534	incident (because the same I<file descriptor> could point to a different
334	best to avoid that. Also, C<dup ()>'ed file descriptors might not work	535	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
335	very well if you register events for both fds.	536	file descriptors might not work very well if you register events for both
		537	file descriptors.
336		538
337	Please note that epoll sometimes generates spurious notifications, so you	539	Best performance from this backend is achieved by not unregistering all
338	need to use non-blocking I/O or other means to avoid blocking when no data	540	watchers for a file descriptor until it has been closed, if possible,
339	(or space) is available.	541	i.e. keep at least one watcher active per fd at all times. Stopping and
		542	starting a watcher (without re-setting it) also usually doesn't cause
		543	extra overhead. A fork can both result in spurious notifications as well
		544	as in libev having to destroy and recreate the epoll object, which can
		545	take considerable time and thus should be avoided.
		546
		547	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		548	faster than epoll for maybe up to a hundred file descriptors, depending on
		549	the usage. So sad.
		550
		551	While nominally embeddable in other event loops, this feature is broken in
		552	all kernel versions tested so far.
		553
		554	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		555	C<EVBACKEND_POLL>.
340		556
341	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	557	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
342		558
343	Kqueue deserves special mention, as at the time of this writing, it	559	Kqueue deserves special mention, as at the time of this writing, it
344	was broken on I<all> BSDs (usually it doesn't work with anything but	560	was broken on all BSDs except NetBSD (usually it doesn't work reliably
345	sockets and pipes, except on Darwin, where of course it's completely	561	with anything but sockets and pipes, except on Darwin, where of course
346	useless. On NetBSD, it seems to work for all the FD types I tested, so it	562	it's completely useless). Unlike epoll, however, whose brokenness
347	is used by default there). For this reason it's not being "autodetected"	563	is by design, these kqueue bugs can (and eventually will) be fixed
		564	without API changes to existing programs. For this reason it's not being
348	unless you explicitly specify it explicitly in the flags (i.e. using	565	"auto-detected" unless you explicitly specify it in the flags (i.e. using
349	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	566	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
350	system like NetBSD.	567	system like NetBSD.
351		568
		569	You still can embed kqueue into a normal poll or select backend and use it
		570	only for sockets (after having made sure that sockets work with kqueue on
		571	the target platform). See C<ev_embed> watchers for more info.
		572
352	It scales in the same way as the epoll backend, but the interface to the	573	It scales in the same way as the epoll backend, but the interface to the
353	kernel is more efficient (which says nothing about its actual speed,	574	kernel is more efficient (which says nothing about its actual speed, of
354	of course). While stopping, setting and starting an I/O watcher does	575	course). While stopping, setting and starting an I/O watcher does never
355	never cause an extra syscall as with epoll, it still adds up to two event	576	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
356	changes per incident, support for C<fork ()> is very bad and it drops fds	577	two event changes per incident. Support for C<fork ()> is very bad (you
		578	might have to leak fd's on fork, but it's more sane than epoll) and it
357	silently in similarly hard-to-detetc cases.	579	drops fds silently in similarly hard-to-detect cases.
		580
		581	This backend usually performs well under most conditions.
		582
		583	While nominally embeddable in other event loops, this doesn't work
		584	everywhere, so you might need to test for this. And since it is broken
		585	almost everywhere, you should only use it when you have a lot of sockets
		586	(for which it usually works), by embedding it into another event loop
		587	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		588	also broken on OS X)) and, did I mention it, using it only for sockets.
		589
		590	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		591	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		592	C<NOTE_EOF>.
358		593
359	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	594	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
360		595
361	This is not implemented yet (and might never be).	596	This is not implemented yet (and might never be, unless you send me an
		597	implementation). According to reports, C</dev/poll> only supports sockets
		598	and is not embeddable, which would limit the usefulness of this backend
		599	immensely.
362		600
363	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	601	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
364		602
365	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	603	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
366	it's really slow, but it still scales very well (O(active_fds)).	604	it's really slow, but it still scales very well (O(active_fds)).
367		605
368	Please note that solaris event ports can deliver a lot of spurious	606	While this backend scales well, it requires one system call per active
369	notifications, so you need to use non-blocking I/O or other means to avoid	607	file descriptor per loop iteration. For small and medium numbers of file
370	blocking when no data (or space) is available.	608	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		609	might perform better.
		610
		611	On the positive side, this backend actually performed fully to
		612	specification in all tests and is fully embeddable, which is a rare feat
		613	among the OS-specific backends (I vastly prefer correctness over speed
		614	hacks).
		615
		616	On the negative side, the interface is I<bizarre> - so bizarre that
		617	even sun itself gets it wrong in their code examples: The event polling
		618	function sometimes returns events to the caller even though an error
		619	occurred, but with no indication whether it has done so or not (yes, it's
		620	even documented that way) - deadly for edge-triggered interfaces where you
		621	absolutely have to know whether an event occurred or not because you have
		622	to re-arm the watcher.
		623
		624	Fortunately libev seems to be able to work around these idiocies.
		625
		626	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		627	C<EVBACKEND_POLL>.
371		628
372	=item C<EVBACKEND_ALL>	629	=item C<EVBACKEND_ALL>
373		630
374	Try all backends (even potentially broken ones that wouldn't be tried	631	Try all backends (even potentially broken ones that wouldn't be tried
375	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	632	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
376	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	633	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
377		634
		635	It is definitely not recommended to use this flag, use whatever
		636	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		637	at all.
		638
		639	=item C<EVBACKEND_MASK>
		640
		641	Not a backend at all, but a mask to select all backend bits from a
		642	C<flags> value, in case you want to mask out any backends from a flags
		643	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
		644
378	=back	645	=back
379		646
380	If one or more of these are ored into the flags value, then only these	647	If one or more of the backend flags are or'ed into the flags value,
381	backends will be tried (in the reverse order as given here). If none are	648	then only these backends will be tried (in the reverse order as listed
382	specified, most compiled-in backend will be tried, usually in reverse	649	here). If none are specified, all backends in C<ev_recommended_backends
383	order of their flag values :)	650	()> will be tried.
384
385	The most typical usage is like this:
386
387	if (!ev_default_loop (0))
388	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
389
390	Restrict libev to the select and poll backends, and do not allow
391	environment settings to be taken into account:
392
393	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
394
395	Use whatever libev has to offer, but make sure that kqueue is used if
396	available (warning, breaks stuff, best use only with your own private
397	event loop and only if you know the OS supports your types of fds):
398
399	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
400
401	=item struct ev_loop *ev_loop_new (unsigned int flags)
402
403	Similar to C<ev_default_loop>, but always creates a new event loop that is
404	always distinct from the default loop. Unlike the default loop, it cannot
405	handle signal and child watchers, and attempts to do so will be greeted by
406	undefined behaviour (or a failed assertion if assertions are enabled).
407		651
408	Example: Try to create a event loop that uses epoll and nothing else.	652	Example: Try to create a event loop that uses epoll and nothing else.
409		653
410	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	654	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
411	if (!epoller)	655	if (!epoller)
412	fatal ("no epoll found here, maybe it hides under your chair");	656	fatal ("no epoll found here, maybe it hides under your chair");
413		657
		658	Example: Use whatever libev has to offer, but make sure that kqueue is
		659	used if available.
		660
		661	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
		662
414	=item ev_default_destroy ()	663	=item ev_loop_destroy (loop)
415		664
416	Destroys the default loop again (frees all memory and kernel state	665	Destroys an event loop object (frees all memory and kernel state
417	etc.). None of the active event watchers will be stopped in the normal	666	etc.). None of the active event watchers will be stopped in the normal
418	sense, so e.g. C<ev_is_active> might still return true. It is your	667	sense, so e.g. C<ev_is_active> might still return true. It is your
419	responsibility to either stop all watchers cleanly yoursef I<before>	668	responsibility to either stop all watchers cleanly yourself I<before>
420	calling this function, or cope with the fact afterwards (which is usually	669	calling this function, or cope with the fact afterwards (which is usually
421	the easiest thing, you can just ignore the watchers and/or C<free ()> them	670	the easiest thing, you can just ignore the watchers and/or C<free ()> them
422	for example).	671	for example).
423		672
424	Note that certain global state, such as signal state, will not be freed by	673	Note that certain global state, such as signal state (and installed signal
425	this function, and related watchers (such as signal and child watchers)	674	handlers), will not be freed by this function, and related watchers (such
426	would need to be stopped manually.	675	as signal and child watchers) would need to be stopped manually.
427		676
428	In general it is not advisable to call this function except in the	677	This function is normally used on loop objects allocated by
429	rare occasion where you really need to free e.g. the signal handling	678	C<ev_loop_new>, but it can also be used on the default loop returned by
		679	C<ev_default_loop>, in which case it is not thread-safe.
		680
		681	Note that it is not advisable to call this function on the default loop
		682	except in the rare occasion where you really need to free its resources.
430	pipe fds. If you need dynamically allocated loops it is better to use	683	If you need dynamically allocated loops it is better to use C<ev_loop_new>
431	C<ev_loop_new> and C<ev_loop_destroy>).	684	and C<ev_loop_destroy>.
432		685
433	=item ev_loop_destroy (loop)	686	=item ev_loop_fork (loop)
434		687
435	Like C<ev_default_destroy>, but destroys an event loop created by an	688	This function sets a flag that causes subsequent C<ev_run> iterations
436	earlier call to C<ev_loop_new>.
437
438	=item ev_default_fork ()
439
440	This function reinitialises the kernel state for backends that have	689	to reinitialise the kernel state for backends that have one. Despite
441	one. Despite the name, you can call it anytime, but it makes most sense	690	the name, you can call it anytime you are allowed to start or stop
442	after forking, in either the parent or child process (or both, but that	691	watchers (except inside an C<ev_prepare> callback), but it makes most
443	again makes little sense).	692	sense after forking, in the child process. You I<must> call it (or use
		693	C<EVFLAG_FORKCHECK>) in the child before resuming or calling C<ev_run>.
444		694
445	You I<must> call this function in the child process after forking if and	695	In addition, if you want to reuse a loop (via this function or
446	only if you want to use the event library in both processes. If you just	696	C<EVFLAG_FORKCHECK>), you I<also> have to ignore C<SIGPIPE>.
447	fork+exec, you don't have to call it.	697
		698	Again, you I<have> to call it on I<any> loop that you want to re-use after
		699	a fork, I<even if you do not plan to use the loop in the parent>. This is
		700	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		701	during fork.
		702
		703	On the other hand, you only need to call this function in the child
		704	process if and only if you want to use the event loop in the child. If
		705	you just fork+exec or create a new loop in the child, you don't have to
		706	call it at all (in fact, C<epoll> is so badly broken that it makes a
		707	difference, but libev will usually detect this case on its own and do a
		708	costly reset of the backend).
448		709
449	The function itself is quite fast and it's usually not a problem to call	710	The function itself is quite fast and it's usually not a problem to call
450	it just in case after a fork. To make this easy, the function will fit in	711	it just in case after a fork.
451	quite nicely into a call to C<pthread_atfork>:
452		712
		713	Example: Automate calling C<ev_loop_fork> on the default loop when
		714	using pthreads.
		715
		716	static void
		717	post_fork_child (void)
		718	{
		719	ev_loop_fork (EV_DEFAULT);
		720	}
		721
		722	...
453	pthread_atfork (0, 0, ev_default_fork);	723	pthread_atfork (0, 0, post_fork_child);
454		724
455	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use	725	=item int ev_is_default_loop (loop)
456	without calling this function, so if you force one of those backends you
457	do not need to care.
458		726
459	=item ev_loop_fork (loop)	727	Returns true when the given loop is, in fact, the default loop, and false
		728	otherwise.
460		729
461	Like C<ev_default_fork>, but acts on an event loop created by
462	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
463	after fork, and how you do this is entirely your own problem.
464
465	=item unsigned int ev_loop_count (loop)	730	=item unsigned int ev_iteration (loop)
466		731
467	Returns the count of loop iterations for the loop, which is identical to	732	Returns the current iteration count for the event loop, which is identical
468	the number of times libev did poll for new events. It starts at C<0> and	733	to the number of times libev did poll for new events. It starts at C<0>
469	happily wraps around with enough iterations.	734	and happily wraps around with enough iterations.
470		735
471	This value can sometimes be useful as a generation counter of sorts (it	736	This value can sometimes be useful as a generation counter of sorts (it
472	"ticks" the number of loop iterations), as it roughly corresponds with	737	"ticks" the number of loop iterations), as it roughly corresponds with
473	C<ev_prepare> and C<ev_check> calls.	738	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		739	prepare and check phases.
		740
		741	=item unsigned int ev_depth (loop)
		742
		743	Returns the number of times C<ev_run> was entered minus the number of
		744	times C<ev_run> was exited normally, in other words, the recursion depth.
		745
		746	Outside C<ev_run>, this number is zero. In a callback, this number is
		747	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		748	in which case it is higher.
		749
		750	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
		751	throwing an exception etc.), doesn't count as "exit" - consider this
		752	as a hint to avoid such ungentleman-like behaviour unless it's really
		753	convenient, in which case it is fully supported.
474		754
475	=item unsigned int ev_backend (loop)	755	=item unsigned int ev_backend (loop)
476		756
477	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	757	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
478	use.	758	use.
…		…
483	received events and started processing them. This timestamp does not	763	received events and started processing them. This timestamp does not
484	change as long as callbacks are being processed, and this is also the base	764	change as long as callbacks are being processed, and this is also the base
485	time used for relative timers. You can treat it as the timestamp of the	765	time used for relative timers. You can treat it as the timestamp of the
486	event occurring (or more correctly, libev finding out about it).	766	event occurring (or more correctly, libev finding out about it).
487		767
		768	=item ev_now_update (loop)
		769
		770	Establishes the current time by querying the kernel, updating the time
		771	returned by C<ev_now ()> in the progress. This is a costly operation and
		772	is usually done automatically within C<ev_run ()>.
		773
		774	This function is rarely useful, but when some event callback runs for a
		775	very long time without entering the event loop, updating libev's idea of
		776	the current time is a good idea.
		777
		778	See also L</The special problem of time updates> in the C<ev_timer> section.
		779
		780	=item ev_suspend (loop)
		781
		782	=item ev_resume (loop)
		783
		784	These two functions suspend and resume an event loop, for use when the
		785	loop is not used for a while and timeouts should not be processed.
		786
		787	A typical use case would be an interactive program such as a game: When
		788	the user presses C<^Z> to suspend the game and resumes it an hour later it
		789	would be best to handle timeouts as if no time had actually passed while
		790	the program was suspended. This can be achieved by calling C<ev_suspend>
		791	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		792	C<ev_resume> directly afterwards to resume timer processing.
		793
		794	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		795	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		796	will be rescheduled (that is, they will lose any events that would have
		797	occurred while suspended).
		798
		799	After calling C<ev_suspend> you B<must not> call I<any> function on the
		800	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		801	without a previous call to C<ev_suspend>.
		802
		803	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		804	event loop time (see C<ev_now_update>).
		805
488	=item ev_loop (loop, int flags)	806	=item bool ev_run (loop, int flags)
489		807
490	Finally, this is it, the event handler. This function usually is called	808	Finally, this is it, the event handler. This function usually is called
491	after you initialised all your watchers and you want to start handling	809	after you have initialised all your watchers and you want to start
492	events.	810	handling events. It will ask the operating system for any new events, call
		811	the watcher callbacks, and then repeat the whole process indefinitely: This
		812	is why event loops are called I<loops>.
493		813
494	If the flags argument is specified as C<0>, it will not return until	814	If the flags argument is specified as C<0>, it will keep handling events
495	either no event watchers are active anymore or C<ev_unloop> was called.	815	until either no event watchers are active anymore or C<ev_break> was
		816	called.
496		817
		818	The return value is false if there are no more active watchers (which
		819	usually means "all jobs done" or "deadlock"), and true in all other cases
		820	(which usually means " you should call C<ev_run> again").
		821
497	Please note that an explicit C<ev_unloop> is usually better than	822	Please note that an explicit C<ev_break> is usually better than
498	relying on all watchers to be stopped when deciding when a program has	823	relying on all watchers to be stopped when deciding when a program has
499	finished (especially in interactive programs), but having a program that	824	finished (especially in interactive programs), but having a program
500	automatically loops as long as it has to and no longer by virtue of	825	that automatically loops as long as it has to and no longer by virtue
501	relying on its watchers stopping correctly is a thing of beauty.	826	of relying on its watchers stopping correctly, that is truly a thing of
		827	beauty.
502		828
		829	This function is I<mostly> exception-safe - you can break out of a
		830	C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		831	exception and so on. This does not decrement the C<ev_depth> value, nor
		832	will it clear any outstanding C<EVBREAK_ONE> breaks.
		833
503	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	834	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
504	those events and any outstanding ones, but will not block your process in	835	those events and any already outstanding ones, but will not wait and
505	case there are no events and will return after one iteration of the loop.	836	block your process in case there are no events and will return after one
		837	iteration of the loop. This is sometimes useful to poll and handle new
		838	events while doing lengthy calculations, to keep the program responsive.
506		839
507	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	840	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
508	neccessary) and will handle those and any outstanding ones. It will block	841	necessary) and will handle those and any already outstanding ones. It
509	your process until at least one new event arrives, and will return after	842	will block your process until at least one new event arrives (which could
510	one iteration of the loop. This is useful if you are waiting for some	843	be an event internal to libev itself, so there is no guarantee that a
511	external event in conjunction with something not expressible using other	844	user-registered callback will be called), and will return after one
		845	iteration of the loop.
		846
		847	This is useful if you are waiting for some external event in conjunction
		848	with something not expressible using other libev watchers (i.e. "roll your
512	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	849	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
513	usually a better approach for this kind of thing.	850	usually a better approach for this kind of thing.
514		851
515	Here are the gory details of what C<ev_loop> does:	852	Here are the gory details of what C<ev_run> does (this is for your
		853	understanding, not a guarantee that things will work exactly like this in
		854	future versions):
516		855
		856	- Increment loop depth.
		857	- Reset the ev_break status.
517	- Before the first iteration, call any pending watchers.	858	- Before the first iteration, call any pending watchers.
518	* If there are no active watchers (reference count is zero), return.	859	LOOP:
519	- Queue all prepare watchers and then call all outstanding watchers.	860	- If EVFLAG_FORKCHECK was used, check for a fork.
		861	- If a fork was detected (by any means), queue and call all fork watchers.
		862	- Queue and call all prepare watchers.
		863	- If ev_break was called, goto FINISH.
520	- If we have been forked, recreate the kernel state.	864	- If we have been forked, detach and recreate the kernel state
		865	as to not disturb the other process.
521	- Update the kernel state with all outstanding changes.	866	- Update the kernel state with all outstanding changes.
522	- Update the "event loop time".	867	- Update the "event loop time" (ev_now ()).
523	- Calculate for how long to block.	868	- Calculate for how long to sleep or block, if at all
		869	(active idle watchers, EVRUN_NOWAIT or not having
		870	any active watchers at all will result in not sleeping).
		871	- Sleep if the I/O and timer collect interval say so.
		872	- Increment loop iteration counter.
524	- Block the process, waiting for any events.	873	- Block the process, waiting for any events.
525	- Queue all outstanding I/O (fd) events.	874	- Queue all outstanding I/O (fd) events.
526	- Update the "event loop time" and do time jump handling.	875	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
527	- Queue all outstanding timers.	876	- Queue all expired timers.
528	- Queue all outstanding periodics.	877	- Queue all expired periodics.
529	- If no events are pending now, queue all idle watchers.	878	- Queue all idle watchers with priority higher than that of pending events.
530	- Queue all check watchers.	879	- Queue all check watchers.
531	- Call all queued watchers in reverse order (i.e. check watchers first).	880	- Call all queued watchers in reverse order (i.e. check watchers first).
532	Signals and child watchers are implemented as I/O watchers, and will	881	Signals and child watchers are implemented as I/O watchers, and will
533	be handled here by queueing them when their watcher gets executed.	882	be handled here by queueing them when their watcher gets executed.
534	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	883	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
535	were used, return, otherwise continue with step *.	884	were used, or there are no active watchers, goto FINISH, otherwise
		885	continue with step LOOP.
		886	FINISH:
		887	- Reset the ev_break status iff it was EVBREAK_ONE.
		888	- Decrement the loop depth.
		889	- Return.
536		890
537	Example: Queue some jobs and then loop until no events are outsanding	891	Example: Queue some jobs and then loop until no events are outstanding
538	anymore.	892	anymore.
539		893
540	... queue jobs here, make sure they register event watchers as long	894	... queue jobs here, make sure they register event watchers as long
541	... as they still have work to do (even an idle watcher will do..)	895	... as they still have work to do (even an idle watcher will do..)
542	ev_loop (my_loop, 0);	896	ev_run (my_loop, 0);
543	... jobs done. yeah!	897	... jobs done or somebody called break. yeah!
544		898
545	=item ev_unloop (loop, how)	899	=item ev_break (loop, how)
546		900
547	Can be used to make a call to C<ev_loop> return early (but only after it	901	Can be used to make a call to C<ev_run> return early (but only after it
548	has processed all outstanding events). The C<how> argument must be either	902	has processed all outstanding events). The C<how> argument must be either
549	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	903	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
550	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	904	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
		905
		906	This "break state" will be cleared on the next call to C<ev_run>.
		907
		908	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		909	which case it will have no effect.
551		910
552	=item ev_ref (loop)	911	=item ev_ref (loop)
553		912
554	=item ev_unref (loop)	913	=item ev_unref (loop)
555		914
556	Ref/unref can be used to add or remove a reference count on the event	915	Ref/unref can be used to add or remove a reference count on the event
557	loop: Every watcher keeps one reference, and as long as the reference	916	loop: Every watcher keeps one reference, and as long as the reference
558	count is nonzero, C<ev_loop> will not return on its own. If you have	917	count is nonzero, C<ev_run> will not return on its own.
559	a watcher you never unregister that should not keep C<ev_loop> from	918
560	returning, ev_unref() after starting, and ev_ref() before stopping it. For	919	This is useful when you have a watcher that you never intend to
		920	unregister, but that nevertheless should not keep C<ev_run> from
		921	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		922	before stopping it.
		923
561	example, libev itself uses this for its internal signal pipe: It is not	924	As an example, libev itself uses this for its internal signal pipe: It
562	visible to the libev user and should not keep C<ev_loop> from exiting if	925	is not visible to the libev user and should not keep C<ev_run> from
563	no event watchers registered by it are active. It is also an excellent	926	exiting if no event watchers registered by it are active. It is also an
564	way to do this for generic recurring timers or from within third-party	927	excellent way to do this for generic recurring timers or from within
565	libraries. Just remember to I<unref after start> and I<ref before stop>.	928	third-party libraries. Just remember to I<unref after start> and I<ref
		929	before stop> (but only if the watcher wasn't active before, or was active
		930	before, respectively. Note also that libev might stop watchers itself
		931	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		932	in the callback).
566		933
567	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	934	Example: Create a signal watcher, but keep it from keeping C<ev_run>
568	running when nothing else is active.	935	running when nothing else is active.
569		936
570	struct ev_signal exitsig;	937	ev_signal exitsig;
571	ev_signal_init (&exitsig, sig_cb, SIGINT);	938	ev_signal_init (&exitsig, sig_cb, SIGINT);
572	ev_signal_start (loop, &exitsig);	939	ev_signal_start (loop, &exitsig);
573	evf_unref (loop);	940	ev_unref (loop);
574		941
575	Example: For some weird reason, unregister the above signal handler again.	942	Example: For some weird reason, unregister the above signal handler again.
576		943
577	ev_ref (loop);	944	ev_ref (loop);
578	ev_signal_stop (loop, &exitsig);	945	ev_signal_stop (loop, &exitsig);
579		946
580	=item ev_set_io_collect_interval (loop, ev_tstamp interval)	947	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
581		948
582	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)	949	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
583		950
584	These advanced functions influence the time that libev will spend waiting	951	These advanced functions influence the time that libev will spend waiting
585	for events. Both are by default C<0>, meaning that libev will try to	952	for events. Both time intervals are by default C<0>, meaning that libev
586	invoke timer/periodic callbacks and I/O callbacks with minimum latency.	953	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		954	latency.
587		955
588	Setting these to a higher value (the C<interval> I<must> be >= C<0>)	956	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
589	allows libev to delay invocation of I/O and timer/periodic callbacks to	957	allows libev to delay invocation of I/O and timer/periodic callbacks
590	increase efficiency of loop iterations.	958	to increase efficiency of loop iterations (or to increase power-saving
		959	opportunities).
591		960
592	The background is that sometimes your program runs just fast enough to	961	The idea is that sometimes your program runs just fast enough to handle
593	handle one (or very few) event(s) per loop iteration. While this makes	962	one (or very few) event(s) per loop iteration. While this makes the
594	the program responsive, it also wastes a lot of CPU time to poll for new	963	program responsive, it also wastes a lot of CPU time to poll for new
595	events, especially with backends like C<select ()> which have a high	964	events, especially with backends like C<select ()> which have a high
596	overhead for the actual polling but can deliver many events at once.	965	overhead for the actual polling but can deliver many events at once.
597		966
598	By setting a higher I<io collect interval> you allow libev to spend more	967	By setting a higher I<io collect interval> you allow libev to spend more
599	time collecting I/O events, so you can handle more events per iteration,	968	time collecting I/O events, so you can handle more events per iteration,
600	at the cost of increasing latency. Timeouts (both C<ev_periodic> and	969	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
601	C<ev_timer>) will be not affected. Setting this to a non-null bvalue will	970	C<ev_timer>) will not be affected. Setting this to a non-null value will
602	introduce an additional C<ev_sleep ()> call into most loop iterations.	971	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		972	sleep time ensures that libev will not poll for I/O events more often then
		973	once per this interval, on average (as long as the host time resolution is
		974	good enough).
603		975
604	Likewise, by setting a higher I<timeout collect interval> you allow libev	976	Likewise, by setting a higher I<timeout collect interval> you allow libev
605	to spend more time collecting timeouts, at the expense of increased	977	to spend more time collecting timeouts, at the expense of increased
606	latency (the watcher callback will be called later). C<ev_io> watchers	978	latency/jitter/inexactness (the watcher callback will be called
607	will not be affected. Setting this to a non-null value will not introduce	979	later). C<ev_io> watchers will not be affected. Setting this to a non-null
608	any overhead in libev.	980	value will not introduce any overhead in libev.
609		981
610	Many (busy) programs can usually benefit by setting the io collect	982	Many (busy) programs can usually benefit by setting the I/O collect
611	interval to a value near C<0.1> or so, which is often enough for	983	interval to a value near C<0.1> or so, which is often enough for
612	interactive servers (of course not for games), likewise for timeouts. It	984	interactive servers (of course not for games), likewise for timeouts. It
613	usually doesn't make much sense to set it to a lower value than C<0.01>,	985	usually doesn't make much sense to set it to a lower value than C<0.01>,
614	as this approsaches the timing granularity of most systems.	986	as this approaches the timing granularity of most systems. Note that if
		987	you do transactions with the outside world and you can't increase the
		988	parallelity, then this setting will limit your transaction rate (if you
		989	need to poll once per transaction and the I/O collect interval is 0.01,
		990	then you can't do more than 100 transactions per second).
		991
		992	Setting the I<timeout collect interval> can improve the opportunity for
		993	saving power, as the program will "bundle" timer callback invocations that
		994	are "near" in time together, by delaying some, thus reducing the number of
		995	times the process sleeps and wakes up again. Another useful technique to
		996	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		997	they fire on, say, one-second boundaries only.
		998
		999	Example: we only need 0.1s timeout granularity, and we wish not to poll
		1000	more often than 100 times per second:
		1001
		1002	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		1003	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		1004
		1005	=item ev_invoke_pending (loop)
		1006
		1007	This call will simply invoke all pending watchers while resetting their
		1008	pending state. Normally, C<ev_run> does this automatically when required,
		1009	but when overriding the invoke callback this call comes handy. This
		1010	function can be invoked from a watcher - this can be useful for example
		1011	when you want to do some lengthy calculation and want to pass further
		1012	event handling to another thread (you still have to make sure only one
		1013	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		1014
		1015	=item int ev_pending_count (loop)
		1016
		1017	Returns the number of pending watchers - zero indicates that no watchers
		1018	are pending.
		1019
		1020	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		1021
		1022	This overrides the invoke pending functionality of the loop: Instead of
		1023	invoking all pending watchers when there are any, C<ev_run> will call
		1024	this callback instead. This is useful, for example, when you want to
		1025	invoke the actual watchers inside another context (another thread etc.).
		1026
		1027	If you want to reset the callback, use C<ev_invoke_pending> as new
		1028	callback.
		1029
		1030	=item ev_set_loop_release_cb (loop, void (*release)(EV_P) throw (), void (*acquire)(EV_P) throw ())
		1031
		1032	Sometimes you want to share the same loop between multiple threads. This
		1033	can be done relatively simply by putting mutex_lock/unlock calls around
		1034	each call to a libev function.
		1035
		1036	However, C<ev_run> can run an indefinite time, so it is not feasible
		1037	to wait for it to return. One way around this is to wake up the event
		1038	loop via C<ev_break> and C<ev_async_send>, another way is to set these
		1039	I<release> and I<acquire> callbacks on the loop.
		1040
		1041	When set, then C<release> will be called just before the thread is
		1042	suspended waiting for new events, and C<acquire> is called just
		1043	afterwards.
		1044
		1045	Ideally, C<release> will just call your mutex_unlock function, and
		1046	C<acquire> will just call the mutex_lock function again.
		1047
		1048	While event loop modifications are allowed between invocations of
		1049	C<release> and C<acquire> (that's their only purpose after all), no
		1050	modifications done will affect the event loop, i.e. adding watchers will
		1051	have no effect on the set of file descriptors being watched, or the time
		1052	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1053	to take note of any changes you made.
		1054
		1055	In theory, threads executing C<ev_run> will be async-cancel safe between
		1056	invocations of C<release> and C<acquire>.
		1057
		1058	See also the locking example in the C<THREADS> section later in this
		1059	document.
		1060
		1061	=item ev_set_userdata (loop, void *data)
		1062
		1063	=item void *ev_userdata (loop)
		1064
		1065	Set and retrieve a single C<void *> associated with a loop. When
		1066	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1067	C<0>.
		1068
		1069	These two functions can be used to associate arbitrary data with a loop,
		1070	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1071	C<acquire> callbacks described above, but of course can be (ab-)used for
		1072	any other purpose as well.
		1073
		1074	=item ev_verify (loop)
		1075
		1076	This function only does something when C<EV_VERIFY> support has been
		1077	compiled in, which is the default for non-minimal builds. It tries to go
		1078	through all internal structures and checks them for validity. If anything
		1079	is found to be inconsistent, it will print an error message to standard
		1080	error and call C<abort ()>.
		1081
		1082	This can be used to catch bugs inside libev itself: under normal
		1083	circumstances, this function will never abort as of course libev keeps its
		1084	data structures consistent.
615		1085
616	=back	1086	=back
617		1087
618		1088
619	=head1 ANATOMY OF A WATCHER	1089	=head1 ANATOMY OF A WATCHER
620		1090
		1091	In the following description, uppercase C<TYPE> in names stands for the
		1092	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1093	watchers and C<ev_io_start> for I/O watchers.
		1094
621	A watcher is a structure that you create and register to record your	1095	A watcher is an opaque structure that you allocate and register to record
622	interest in some event. For instance, if you want to wait for STDIN to	1096	your interest in some event. To make a concrete example, imagine you want
623	become readable, you would create an C<ev_io> watcher for that:	1097	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1098	for that:
624		1099
625	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1100	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
626	{	1101	{
627	ev_io_stop (w);	1102	ev_io_stop (w);
628	ev_unloop (loop, EVUNLOOP_ALL);	1103	ev_break (loop, EVBREAK_ALL);
629	}	1104	}
630		1105
631	struct ev_loop *loop = ev_default_loop (0);	1106	struct ev_loop *loop = ev_default_loop (0);
		1107
632	struct ev_io stdin_watcher;	1108	ev_io stdin_watcher;
		1109
633	ev_init (&stdin_watcher, my_cb);	1110	ev_init (&stdin_watcher, my_cb);
634	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1111	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
635	ev_io_start (loop, &stdin_watcher);	1112	ev_io_start (loop, &stdin_watcher);
		1113
636	ev_loop (loop, 0);	1114	ev_run (loop, 0);
637		1115
638	As you can see, you are responsible for allocating the memory for your	1116	As you can see, you are responsible for allocating the memory for your
639	watcher structures (and it is usually a bad idea to do this on the stack,	1117	watcher structures (and it is I<usually> a bad idea to do this on the
640	although this can sometimes be quite valid).	1118	stack).
641		1119
		1120	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1121	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1122
642	Each watcher structure must be initialised by a call to C<ev_init	1123	Each watcher structure must be initialised by a call to C<ev_init (watcher
643	(watcher *, callback)>, which expects a callback to be provided. This	1124	*, callback)>, which expects a callback to be provided. This callback is
644	callback gets invoked each time the event occurs (or, in the case of io	1125	invoked each time the event occurs (or, in the case of I/O watchers, each
645	watchers, each time the event loop detects that the file descriptor given	1126	time the event loop detects that the file descriptor given is readable
646	is readable and/or writable).	1127	and/or writable).
647		1128
648	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1129	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
649	with arguments specific to this watcher type. There is also a macro	1130	macro to configure it, with arguments specific to the watcher type. There
650	to combine initialisation and setting in one call: C<< ev_<type>_init	1131	is also a macro to combine initialisation and setting in one call: C<<
651	(watcher *, callback, ...) >>.	1132	ev_TYPE_init (watcher *, callback, ...) >>.
652		1133
653	To make the watcher actually watch out for events, you have to start it	1134	To make the watcher actually watch out for events, you have to start it
654	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1135	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
655	*) >>), and you can stop watching for events at any time by calling the	1136	*) >>), and you can stop watching for events at any time by calling the
656	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1137	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
657		1138
658	As long as your watcher is active (has been started but not stopped) you	1139	As long as your watcher is active (has been started but not stopped) you
659	must not touch the values stored in it. Most specifically you must never	1140	must not touch the values stored in it. Most specifically you must never
660	reinitialise it or call its C<set> macro.	1141	reinitialise it or call its C<ev_TYPE_set> macro.
661		1142
662	Each and every callback receives the event loop pointer as first, the	1143	Each and every callback receives the event loop pointer as first, the
663	registered watcher structure as second, and a bitset of received events as	1144	registered watcher structure as second, and a bitset of received events as
664	third argument.	1145	third argument.
665		1146
…		…
674	=item C<EV_WRITE>	1155	=item C<EV_WRITE>
675		1156
676	The file descriptor in the C<ev_io> watcher has become readable and/or	1157	The file descriptor in the C<ev_io> watcher has become readable and/or
677	writable.	1158	writable.
678		1159
679	=item C<EV_TIMEOUT>	1160	=item C<EV_TIMER>
680		1161
681	The C<ev_timer> watcher has timed out.	1162	The C<ev_timer> watcher has timed out.
682		1163
683	=item C<EV_PERIODIC>	1164	=item C<EV_PERIODIC>
684		1165
…		…
702		1183
703	=item C<EV_PREPARE>	1184	=item C<EV_PREPARE>
704		1185
705	=item C<EV_CHECK>	1186	=item C<EV_CHECK>
706		1187
707	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1188	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts to
708	to gather new events, and all C<ev_check> watchers are invoked just after	1189	gather new events, and all C<ev_check> watchers are queued (not invoked)
709	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1190	just after C<ev_run> has gathered them, but before it queues any callbacks
		1191	for any received events. That means C<ev_prepare> watchers are the last
		1192	watchers invoked before the event loop sleeps or polls for new events, and
		1193	C<ev_check> watchers will be invoked before any other watchers of the same
		1194	or lower priority within an event loop iteration.
		1195
710	received events. Callbacks of both watcher types can start and stop as	1196	Callbacks of both watcher types can start and stop as many watchers as
711	many watchers as they want, and all of them will be taken into account	1197	they want, and all of them will be taken into account (for example, a
712	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1198	C<ev_prepare> watcher might start an idle watcher to keep C<ev_run> from
713	C<ev_loop> from blocking).	1199	blocking).
714		1200
715	=item C<EV_EMBED>	1201	=item C<EV_EMBED>
716		1202
717	The embedded event loop specified in the C<ev_embed> watcher needs attention.	1203	The embedded event loop specified in the C<ev_embed> watcher needs attention.
718		1204
719	=item C<EV_FORK>	1205	=item C<EV_FORK>
720		1206
721	The event loop has been resumed in the child process after fork (see	1207	The event loop has been resumed in the child process after fork (see
722	C<ev_fork>).	1208	C<ev_fork>).
723		1209
		1210	=item C<EV_CLEANUP>
		1211
		1212	The event loop is about to be destroyed (see C<ev_cleanup>).
		1213
		1214	=item C<EV_ASYNC>
		1215
		1216	The given async watcher has been asynchronously notified (see C<ev_async>).
		1217
		1218	=item C<EV_CUSTOM>
		1219
		1220	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1221	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1222
724	=item C<EV_ERROR>	1223	=item C<EV_ERROR>
725		1224
726	An unspecified error has occured, the watcher has been stopped. This might	1225	An unspecified error has occurred, the watcher has been stopped. This might
727	happen because the watcher could not be properly started because libev	1226	happen because the watcher could not be properly started because libev
728	ran out of memory, a file descriptor was found to be closed or any other	1227	ran out of memory, a file descriptor was found to be closed or any other
		1228	problem. Libev considers these application bugs.
		1229
729	problem. You best act on it by reporting the problem and somehow coping	1230	You best act on it by reporting the problem and somehow coping with the
730	with the watcher being stopped.	1231	watcher being stopped. Note that well-written programs should not receive
		1232	an error ever, so when your watcher receives it, this usually indicates a
		1233	bug in your program.
731		1234
732	Libev will usually signal a few "dummy" events together with an error,	1235	Libev will usually signal a few "dummy" events together with an error, for
733	for example it might indicate that a fd is readable or writable, and if	1236	example it might indicate that a fd is readable or writable, and if your
734	your callbacks is well-written it can just attempt the operation and cope	1237	callbacks is well-written it can just attempt the operation and cope with
735	with the error from read() or write(). This will not work in multithreaded	1238	the error from read() or write(). This will not work in multi-threaded
736	programs, though, so beware.	1239	programs, though, as the fd could already be closed and reused for another
		1240	thing, so beware.
737		1241
738	=back	1242	=back
739		1243
740	=head2 GENERIC WATCHER FUNCTIONS	1244	=head2 GENERIC WATCHER FUNCTIONS
741
742	In the following description, C<TYPE> stands for the watcher type,
743	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
744		1245
745	=over 4	1246	=over 4
746		1247
747	=item C<ev_init> (ev_TYPE *watcher, callback)	1248	=item C<ev_init> (ev_TYPE *watcher, callback)
748		1249
…		…
754	which rolls both calls into one.	1255	which rolls both calls into one.
755		1256
756	You can reinitialise a watcher at any time as long as it has been stopped	1257	You can reinitialise a watcher at any time as long as it has been stopped
757	(or never started) and there are no pending events outstanding.	1258	(or never started) and there are no pending events outstanding.
758		1259
759	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	1260	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
760	int revents)>.	1261	int revents)>.
761		1262
		1263	Example: Initialise an C<ev_io> watcher in two steps.
		1264
		1265	ev_io w;
		1266	ev_init (&w, my_cb);
		1267	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1268
762	=item C<ev_TYPE_set> (ev_TYPE *, [args])	1269	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
763		1270
764	This macro initialises the type-specific parts of a watcher. You need to	1271	This macro initialises the type-specific parts of a watcher. You need to
765	call C<ev_init> at least once before you call this macro, but you can	1272	call C<ev_init> at least once before you call this macro, but you can
766	call C<ev_TYPE_set> any number of times. You must not, however, call this	1273	call C<ev_TYPE_set> any number of times. You must not, however, call this
767	macro on a watcher that is active (it can be pending, however, which is a	1274	macro on a watcher that is active (it can be pending, however, which is a
768	difference to the C<ev_init> macro).	1275	difference to the C<ev_init> macro).
769		1276
770	Although some watcher types do not have type-specific arguments	1277	Although some watcher types do not have type-specific arguments
771	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	1278	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
772		1279
		1280	See C<ev_init>, above, for an example.
		1281
773	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	1282	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
774		1283
775	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	1284	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
776	calls into a single call. This is the most convinient method to initialise	1285	calls into a single call. This is the most convenient method to initialise
777	a watcher. The same limitations apply, of course.	1286	a watcher. The same limitations apply, of course.
778		1287
		1288	Example: Initialise and set an C<ev_io> watcher in one step.
		1289
		1290	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1291
779	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	1292	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
780		1293
781	Starts (activates) the given watcher. Only active watchers will receive	1294	Starts (activates) the given watcher. Only active watchers will receive
782	events. If the watcher is already active nothing will happen.	1295	events. If the watcher is already active nothing will happen.
783		1296
		1297	Example: Start the C<ev_io> watcher that is being abused as example in this
		1298	whole section.
		1299
		1300	ev_io_start (EV_DEFAULT_UC, &w);
		1301
784	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	1302	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
785		1303
786	Stops the given watcher again (if active) and clears the pending	1304	Stops the given watcher if active, and clears the pending status (whether
		1305	the watcher was active or not).
		1306
787	status. It is possible that stopped watchers are pending (for example,	1307	It is possible that stopped watchers are pending - for example,
788	non-repeating timers are being stopped when they become pending), but	1308	non-repeating timers are being stopped when they become pending - but
789	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1309	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
790	you want to free or reuse the memory used by the watcher it is therefore a	1310	pending. If you want to free or reuse the memory used by the watcher it is
791	good idea to always call its C<ev_TYPE_stop> function.	1311	therefore a good idea to always call its C<ev_TYPE_stop> function.
792		1312
793	=item bool ev_is_active (ev_TYPE *watcher)	1313	=item bool ev_is_active (ev_TYPE *watcher)
794		1314
795	Returns a true value iff the watcher is active (i.e. it has been started	1315	Returns a true value iff the watcher is active (i.e. it has been started
796	and not yet been stopped). As long as a watcher is active you must not modify	1316	and not yet been stopped). As long as a watcher is active you must not modify
…		…
807		1327
808	=item callback ev_cb (ev_TYPE *watcher)	1328	=item callback ev_cb (ev_TYPE *watcher)
809		1329
810	Returns the callback currently set on the watcher.	1330	Returns the callback currently set on the watcher.
811		1331
812	=item ev_cb_set (ev_TYPE *watcher, callback)	1332	=item ev_set_cb (ev_TYPE *watcher, callback)
813		1333
814	Change the callback. You can change the callback at virtually any time	1334	Change the callback. You can change the callback at virtually any time
815	(modulo threads).	1335	(modulo threads).
816		1336
817	=item ev_set_priority (ev_TYPE *watcher, priority)	1337	=item ev_set_priority (ev_TYPE *watcher, int priority)
818		1338
819	=item int ev_priority (ev_TYPE *watcher)	1339	=item int ev_priority (ev_TYPE *watcher)
820		1340
821	Set and query the priority of the watcher. The priority is a small	1341	Set and query the priority of the watcher. The priority is a small
822	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>	1342	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
823	(default: C<-2>). Pending watchers with higher priority will be invoked	1343	(default: C<-2>). Pending watchers with higher priority will be invoked
824	before watchers with lower priority, but priority will not keep watchers	1344	before watchers with lower priority, but priority will not keep watchers
825	from being executed (except for C<ev_idle> watchers).	1345	from being executed (except for C<ev_idle> watchers).
826		1346
827	This means that priorities are I<only> used for ordering callback
828	invocation after new events have been received. This is useful, for
829	example, to reduce latency after idling, or more often, to bind two
830	watchers on the same event and make sure one is called first.
831
832	If you need to suppress invocation when higher priority events are pending	1347	If you need to suppress invocation when higher priority events are pending
833	you need to look at C<ev_idle> watchers, which provide this functionality.	1348	you need to look at C<ev_idle> watchers, which provide this functionality.
834		1349
835	You I<must not> change the priority of a watcher as long as it is active or	1350	You I<must not> change the priority of a watcher as long as it is active or
836	pending.	1351	pending.
837		1352
		1353	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1354	fine, as long as you do not mind that the priority value you query might
		1355	or might not have been clamped to the valid range.
		1356
838	The default priority used by watchers when no priority has been set is	1357	The default priority used by watchers when no priority has been set is
839	always C<0>, which is supposed to not be too high and not be too low :).	1358	always C<0>, which is supposed to not be too high and not be too low :).
840		1359
841	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is	1360	See L</WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
842	fine, as long as you do not mind that the priority value you query might	1361	priorities.
843	or might not have been adjusted to be within valid range.
844		1362
845	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1363	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
846		1364
847	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1365	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
848	C<loop> nor C<revents> need to be valid as long as the watcher callback	1366	C<loop> nor C<revents> need to be valid as long as the watcher callback
849	can deal with that fact.	1367	can deal with that fact, as both are simply passed through to the
		1368	callback.
850		1369
851	=item int ev_clear_pending (loop, ev_TYPE *watcher)	1370	=item int ev_clear_pending (loop, ev_TYPE *watcher)
852		1371
853	If the watcher is pending, this function returns clears its pending status	1372	If the watcher is pending, this function clears its pending status and
854	and returns its C<revents> bitset (as if its callback was invoked). If the	1373	returns its C<revents> bitset (as if its callback was invoked). If the
855	watcher isn't pending it does nothing and returns C<0>.	1374	watcher isn't pending it does nothing and returns C<0>.
856		1375
		1376	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1377	callback to be invoked, which can be accomplished with this function.
		1378
		1379	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1380
		1381	Feeds the given event set into the event loop, as if the specified event
		1382	had happened for the specified watcher (which must be a pointer to an
		1383	initialised but not necessarily started event watcher). Obviously you must
		1384	not free the watcher as long as it has pending events.
		1385
		1386	Stopping the watcher, letting libev invoke it, or calling
		1387	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1388	not started in the first place.
		1389
		1390	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1391	functions that do not need a watcher.
		1392
857	=back	1393	=back
858		1394
		1395	See also the L</ASSOCIATING CUSTOM DATA WITH A WATCHER> and L</BUILDING YOUR
		1396	OWN COMPOSITE WATCHERS> idioms.
859		1397
860	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1398	=head2 WATCHER STATES
861		1399
862	Each watcher has, by default, a member C<void *data> that you can change	1400	There are various watcher states mentioned throughout this manual -
863	and read at any time, libev will completely ignore it. This can be used	1401	active, pending and so on. In this section these states and the rules to
864	to associate arbitrary data with your watcher. If you need more data and	1402	transition between them will be described in more detail - and while these
865	don't want to allocate memory and store a pointer to it in that data	1403	rules might look complicated, they usually do "the right thing".
866	member, you can also "subclass" the watcher type and provide your own
867	data:
868		1404
869	struct my_io	1405	=over 4
870	{
871	struct ev_io io;
872	int otherfd;
873	void *somedata;
874	struct whatever *mostinteresting;
875	}
876		1406
877	And since your callback will be called with a pointer to the watcher, you	1407	=item initialised
878	can cast it back to your own type:
879		1408
880	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1409	Before a watcher can be registered with the event loop it has to be
881	{	1410	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
882	struct my_io *w = (struct my_io *)w_;	1411	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
883	...
884	}
885		1412
886	More interesting and less C-conformant ways of casting your callback type	1413	In this state it is simply some block of memory that is suitable for
887	instead have been omitted.	1414	use in an event loop. It can be moved around, freed, reused etc. at
		1415	will - as long as you either keep the memory contents intact, or call
		1416	C<ev_TYPE_init> again.
888		1417
889	Another common scenario is having some data structure with multiple	1418	=item started/running/active
890	watchers:
891		1419
892	struct my_biggy	1420	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
893	{	1421	property of the event loop, and is actively waiting for events. While in
894	int some_data;	1422	this state it cannot be accessed (except in a few documented ways), moved,
895	ev_timer t1;	1423	freed or anything else - the only legal thing is to keep a pointer to it,
896	ev_timer t2;	1424	and call libev functions on it that are documented to work on active watchers.
897	}
898		1425
899	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1426	=item pending
900	you need to use C<offsetof>:
901		1427
902	#include <stddef.h>	1428	If a watcher is active and libev determines that an event it is interested
		1429	in has occurred (such as a timer expiring), it will become pending. It will
		1430	stay in this pending state until either it is stopped or its callback is
		1431	about to be invoked, so it is not normally pending inside the watcher
		1432	callback.
903		1433
		1434	The watcher might or might not be active while it is pending (for example,
		1435	an expired non-repeating timer can be pending but no longer active). If it
		1436	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1437	but it is still property of the event loop at this time, so cannot be
		1438	moved, freed or reused. And if it is active the rules described in the
		1439	previous item still apply.
		1440
		1441	It is also possible to feed an event on a watcher that is not active (e.g.
		1442	via C<ev_feed_event>), in which case it becomes pending without being
		1443	active.
		1444
		1445	=item stopped
		1446
		1447	A watcher can be stopped implicitly by libev (in which case it might still
		1448	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1449	latter will clear any pending state the watcher might be in, regardless
		1450	of whether it was active or not, so stopping a watcher explicitly before
		1451	freeing it is often a good idea.
		1452
		1453	While stopped (and not pending) the watcher is essentially in the
		1454	initialised state, that is, it can be reused, moved, modified in any way
		1455	you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
		1456	it again).
		1457
		1458	=back
		1459
		1460	=head2 WATCHER PRIORITY MODELS
		1461
		1462	Many event loops support I<watcher priorities>, which are usually small
		1463	integers that influence the ordering of event callback invocation
		1464	between watchers in some way, all else being equal.
		1465
		1466	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1467	description for the more technical details such as the actual priority
		1468	range.
		1469
		1470	There are two common ways how these these priorities are being interpreted
		1471	by event loops:
		1472
		1473	In the more common lock-out model, higher priorities "lock out" invocation
		1474	of lower priority watchers, which means as long as higher priority
		1475	watchers receive events, lower priority watchers are not being invoked.
		1476
		1477	The less common only-for-ordering model uses priorities solely to order
		1478	callback invocation within a single event loop iteration: Higher priority
		1479	watchers are invoked before lower priority ones, but they all get invoked
		1480	before polling for new events.
		1481
		1482	Libev uses the second (only-for-ordering) model for all its watchers
		1483	except for idle watchers (which use the lock-out model).
		1484
		1485	The rationale behind this is that implementing the lock-out model for
		1486	watchers is not well supported by most kernel interfaces, and most event
		1487	libraries will just poll for the same events again and again as long as
		1488	their callbacks have not been executed, which is very inefficient in the
		1489	common case of one high-priority watcher locking out a mass of lower
		1490	priority ones.
		1491
		1492	Static (ordering) priorities are most useful when you have two or more
		1493	watchers handling the same resource: a typical usage example is having an
		1494	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1495	timeouts. Under load, data might be received while the program handles
		1496	other jobs, but since timers normally get invoked first, the timeout
		1497	handler will be executed before checking for data. In that case, giving
		1498	the timer a lower priority than the I/O watcher ensures that I/O will be
		1499	handled first even under adverse conditions (which is usually, but not
		1500	always, what you want).
		1501
		1502	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1503	will only be executed when no same or higher priority watchers have
		1504	received events, they can be used to implement the "lock-out" model when
		1505	required.
		1506
		1507	For example, to emulate how many other event libraries handle priorities,
		1508	you can associate an C<ev_idle> watcher to each such watcher, and in
		1509	the normal watcher callback, you just start the idle watcher. The real
		1510	processing is done in the idle watcher callback. This causes libev to
		1511	continuously poll and process kernel event data for the watcher, but when
		1512	the lock-out case is known to be rare (which in turn is rare :), this is
		1513	workable.
		1514
		1515	Usually, however, the lock-out model implemented that way will perform
		1516	miserably under the type of load it was designed to handle. In that case,
		1517	it might be preferable to stop the real watcher before starting the
		1518	idle watcher, so the kernel will not have to process the event in case
		1519	the actual processing will be delayed for considerable time.
		1520
		1521	Here is an example of an I/O watcher that should run at a strictly lower
		1522	priority than the default, and which should only process data when no
		1523	other events are pending:
		1524
		1525	ev_idle idle; // actual processing watcher
		1526	ev_io io; // actual event watcher
		1527
904	static void	1528	static void
905	t1_cb (EV_P_ struct ev_timer *w, int revents)	1529	io_cb (EV_P_ ev_io *w, int revents)
906	{	1530	{
907	struct my_biggy big = (struct my_biggy *	1531	// stop the I/O watcher, we received the event, but
908	(((char *)w) - offsetof (struct my_biggy, t1));	1532	// are not yet ready to handle it.
		1533	ev_io_stop (EV_A_ w);
		1534
		1535	// start the idle watcher to handle the actual event.
		1536	// it will not be executed as long as other watchers
		1537	// with the default priority are receiving events.
		1538	ev_idle_start (EV_A_ &idle);
909	}	1539	}
910		1540
911	static void	1541	static void
912	t2_cb (EV_P_ struct ev_timer *w, int revents)	1542	idle_cb (EV_P_ ev_idle *w, int revents)
913	{	1543	{
914	struct my_biggy big = (struct my_biggy *	1544	// actual processing
915	(((char *)w) - offsetof (struct my_biggy, t2));	1545	read (STDIN_FILENO, ...);
		1546
		1547	// have to start the I/O watcher again, as
		1548	// we have handled the event
		1549	ev_io_start (EV_P_ &io);
916	}	1550	}
		1551
		1552	// initialisation
		1553	ev_idle_init (&idle, idle_cb);
		1554	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1555	ev_io_start (EV_DEFAULT_ &io);
		1556
		1557	In the "real" world, it might also be beneficial to start a timer, so that
		1558	low-priority connections can not be locked out forever under load. This
		1559	enables your program to keep a lower latency for important connections
		1560	during short periods of high load, while not completely locking out less
		1561	important ones.
917		1562
918		1563
919	=head1 WATCHER TYPES	1564	=head1 WATCHER TYPES
920		1565
921	This section describes each watcher in detail, but will not repeat	1566	This section describes each watcher in detail, but will not repeat
…		…
945	In general you can register as many read and/or write event watchers per	1590	In general you can register as many read and/or write event watchers per
946	fd as you want (as long as you don't confuse yourself). Setting all file	1591	fd as you want (as long as you don't confuse yourself). Setting all file
947	descriptors to non-blocking mode is also usually a good idea (but not	1592	descriptors to non-blocking mode is also usually a good idea (but not
948	required if you know what you are doing).	1593	required if you know what you are doing).
949		1594
950	You have to be careful with dup'ed file descriptors, though. Some backends
951	(the linux epoll backend is a notable example) cannot handle dup'ed file
952	descriptors correctly if you register interest in two or more fds pointing
953	to the same underlying file/socket/etc. description (that is, they share
954	the same underlying "file open").
955
956	If you must do this, then force the use of a known-to-be-good backend
957	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
958	C<EVBACKEND_POLL>).
959
960	Another thing you have to watch out for is that it is quite easy to	1595	Another thing you have to watch out for is that it is quite easy to
961	receive "spurious" readyness notifications, that is your callback might	1596	receive "spurious" readiness notifications, that is, your callback might
962	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1597	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
963	because there is no data. Not only are some backends known to create a	1598	because there is no data. It is very easy to get into this situation even
964	lot of those (for example solaris ports), it is very easy to get into	1599	with a relatively standard program structure. Thus it is best to always
965	this situation even with a relatively standard program structure. Thus	1600	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
966	it is best to always use non-blocking I/O: An extra C<read>(2) returning
967	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1601	preferable to a program hanging until some data arrives.
968		1602
969	If you cannot run the fd in non-blocking mode (for example you should not	1603	If you cannot run the fd in non-blocking mode (for example you should
970	play around with an Xlib connection), then you have to seperately re-test	1604	not play around with an Xlib connection), then you have to separately
971	whether a file descriptor is really ready with a known-to-be good interface	1605	re-test whether a file descriptor is really ready with a known-to-be good
972	such as poll (fortunately in our Xlib example, Xlib already does this on	1606	interface such as poll (fortunately in the case of Xlib, it already does
973	its own, so its quite safe to use).	1607	this on its own, so its quite safe to use). Some people additionally
		1608	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1609	indefinitely.
		1610
		1611	But really, best use non-blocking mode.
974		1612
975	=head3 The special problem of disappearing file descriptors	1613	=head3 The special problem of disappearing file descriptors
976		1614
977	Some backends (e.g. kqueue, epoll) need to be told about closing a file	1615	Some backends (e.g. kqueue, epoll) need to be told about closing a file
978	descriptor (either by calling C<close> explicitly or by any other means,	1616	descriptor (either due to calling C<close> explicitly or any other means,
979	such as C<dup>). The reason is that you register interest in some file	1617	such as C<dup2>). The reason is that you register interest in some file
980	descriptor, but when it goes away, the operating system will silently drop	1618	descriptor, but when it goes away, the operating system will silently drop
981	this interest. If another file descriptor with the same number then is	1619	this interest. If another file descriptor with the same number then is
982	registered with libev, there is no efficient way to see that this is, in	1620	registered with libev, there is no efficient way to see that this is, in
983	fact, a different file descriptor.	1621	fact, a different file descriptor.
984		1622
…		…
994	optimisations to libev.	1632	optimisations to libev.
995		1633
996	=head3 The special problem of dup'ed file descriptors	1634	=head3 The special problem of dup'ed file descriptors
997		1635
998	Some backends (e.g. epoll), cannot register events for file descriptors,	1636	Some backends (e.g. epoll), cannot register events for file descriptors,
999	but only events for the underlying file descriptions. That menas when you	1637	but only events for the underlying file descriptions. That means when you
1000	have C<dup ()>'ed file descriptors and register events for them, only one	1638	have C<dup ()>'ed file descriptors or weirder constellations, and register
1001	file descriptor might actually receive events.	1639	events for them, only one file descriptor might actually receive events.
1002		1640
1003	There is no workaorund possible except not registering events	1641	There is no workaround possible except not registering events
1004	for potentially C<dup ()>'ed file descriptors or to resort to	1642	for potentially C<dup ()>'ed file descriptors, or to resort to
1005	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.	1643	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1644
		1645	=head3 The special problem of files
		1646
		1647	Many people try to use C<select> (or libev) on file descriptors
		1648	representing files, and expect it to become ready when their program
		1649	doesn't block on disk accesses (which can take a long time on their own).
		1650
		1651	However, this cannot ever work in the "expected" way - you get a readiness
		1652	notification as soon as the kernel knows whether and how much data is
		1653	there, and in the case of open files, that's always the case, so you
		1654	always get a readiness notification instantly, and your read (or possibly
		1655	write) will still block on the disk I/O.
		1656
		1657	Another way to view it is that in the case of sockets, pipes, character
		1658	devices and so on, there is another party (the sender) that delivers data
		1659	on its own, but in the case of files, there is no such thing: the disk
		1660	will not send data on its own, simply because it doesn't know what you
		1661	wish to read - you would first have to request some data.
		1662
		1663	Since files are typically not-so-well supported by advanced notification
		1664	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1665	to files, even though you should not use it. The reason for this is
		1666	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1667	usually a tty, often a pipe, but also sometimes files or special devices
		1668	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1669	F</dev/urandom>), and even though the file might better be served with
		1670	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1671	it "just works" instead of freezing.
		1672
		1673	So avoid file descriptors pointing to files when you know it (e.g. use
		1674	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1675	when you rarely read from a file instead of from a socket, and want to
		1676	reuse the same code path.
1006		1677
1007	=head3 The special problem of fork	1678	=head3 The special problem of fork
1008		1679
1009	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit	1680	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1010	useless behaviour. Libev fully supports fork, but needs to be told about	1681	useless behaviour. Libev fully supports fork, but needs to be told about
1011	it in the child.	1682	it in the child if you want to continue to use it in the child.
1012		1683
1013	To support fork in your programs, you either have to call	1684	To support fork in your child processes, you have to call C<ev_loop_fork
1014	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,	1685	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
1015	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or	1686	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1016	C<EVBACKEND_POLL>.
1017		1687
		1688	=head3 The special problem of SIGPIPE
		1689
		1690	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1691	when writing to a pipe whose other end has been closed, your program gets
		1692	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1693	this is sensible behaviour, for daemons, this is usually undesirable.
		1694
		1695	So when you encounter spurious, unexplained daemon exits, make sure you
		1696	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1697	somewhere, as that would have given you a big clue).
		1698
		1699	=head3 The special problem of accept()ing when you can't
		1700
		1701	Many implementations of the POSIX C<accept> function (for example,
		1702	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1703	connection from the pending queue in all error cases.
		1704
		1705	For example, larger servers often run out of file descriptors (because
		1706	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1707	rejecting the connection, leading to libev signalling readiness on
		1708	the next iteration again (the connection still exists after all), and
		1709	typically causing the program to loop at 100% CPU usage.
		1710
		1711	Unfortunately, the set of errors that cause this issue differs between
		1712	operating systems, there is usually little the app can do to remedy the
		1713	situation, and no known thread-safe method of removing the connection to
		1714	cope with overload is known (to me).
		1715
		1716	One of the easiest ways to handle this situation is to just ignore it
		1717	- when the program encounters an overload, it will just loop until the
		1718	situation is over. While this is a form of busy waiting, no OS offers an
		1719	event-based way to handle this situation, so it's the best one can do.
		1720
		1721	A better way to handle the situation is to log any errors other than
		1722	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1723	messages, and continue as usual, which at least gives the user an idea of
		1724	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1725	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1726	usage.
		1727
		1728	If your program is single-threaded, then you could also keep a dummy file
		1729	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1730	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1731	close that fd, and create a new dummy fd. This will gracefully refuse
		1732	clients under typical overload conditions.
		1733
		1734	The last way to handle it is to simply log the error and C<exit>, as
		1735	is often done with C<malloc> failures, but this results in an easy
		1736	opportunity for a DoS attack.
1018		1737
1019	=head3 Watcher-Specific Functions	1738	=head3 Watcher-Specific Functions
1020		1739
1021	=over 4	1740	=over 4
1022		1741
1023	=item ev_io_init (ev_io *, callback, int fd, int events)	1742	=item ev_io_init (ev_io *, callback, int fd, int events)
1024		1743
1025	=item ev_io_set (ev_io *, int fd, int events)	1744	=item ev_io_set (ev_io *, int fd, int events)
1026		1745
1027	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1746	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
1028	rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or	1747	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
1029	C<EV_READ | EV_WRITE> to receive the given events.	1748	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
1030		1749
1031	=item int fd [read-only]	1750	=item int fd [read-only]
1032		1751
1033	The file descriptor being watched.	1752	The file descriptor being watched.
1034		1753
1035	=item int events [read-only]	1754	=item int events [read-only]
1036		1755
1037	The events being watched.	1756	The events being watched.
1038		1757
1039	=back	1758	=back
		1759
		1760	=head3 Examples
1040		1761
1041	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1762	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1042	readable, but only once. Since it is likely line-buffered, you could	1763	readable, but only once. Since it is likely line-buffered, you could
1043	attempt to read a whole line in the callback.	1764	attempt to read a whole line in the callback.
1044		1765
1045	static void	1766	static void
1046	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1767	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
1047	{	1768	{
1048	ev_io_stop (loop, w);	1769	ev_io_stop (loop, w);
1049	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1770	.. read from stdin here (or from w->fd) and handle any I/O errors
1050	}	1771	}
1051		1772
1052	...	1773	...
1053	struct ev_loop *loop = ev_default_init (0);	1774	struct ev_loop *loop = ev_default_init (0);
1054	struct ev_io stdin_readable;	1775	ev_io stdin_readable;
1055	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1776	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1056	ev_io_start (loop, &stdin_readable);	1777	ev_io_start (loop, &stdin_readable);
1057	ev_loop (loop, 0);	1778	ev_run (loop, 0);
1058		1779
1059		1780
1060	=head2 C<ev_timer> - relative and optionally repeating timeouts	1781	=head2 C<ev_timer> - relative and optionally repeating timeouts
1061		1782
1062	Timer watchers are simple relative timers that generate an event after a	1783	Timer watchers are simple relative timers that generate an event after a
1063	given time, and optionally repeating in regular intervals after that.	1784	given time, and optionally repeating in regular intervals after that.
1064		1785
1065	The timers are based on real time, that is, if you register an event that	1786	The timers are based on real time, that is, if you register an event that
1066	times out after an hour and you reset your system clock to last years	1787	times out after an hour and you reset your system clock to January last
1067	time, it will still time out after (roughly) and hour. "Roughly" because	1788	year, it will still time out after (roughly) one hour. "Roughly" because
1068	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1789	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1069	monotonic clock option helps a lot here).	1790	monotonic clock option helps a lot here).
		1791
		1792	The callback is guaranteed to be invoked only I<after> its timeout has
		1793	passed (not I<at>, so on systems with very low-resolution clocks this
		1794	might introduce a small delay, see "the special problem of being too
		1795	early", below). If multiple timers become ready during the same loop
		1796	iteration then the ones with earlier time-out values are invoked before
		1797	ones of the same priority with later time-out values (but this is no
		1798	longer true when a callback calls C<ev_run> recursively).
		1799
		1800	=head3 Be smart about timeouts
		1801
		1802	Many real-world problems involve some kind of timeout, usually for error
		1803	recovery. A typical example is an HTTP request - if the other side hangs,
		1804	you want to raise some error after a while.
		1805
		1806	What follows are some ways to handle this problem, from obvious and
		1807	inefficient to smart and efficient.
		1808
		1809	In the following, a 60 second activity timeout is assumed - a timeout that
		1810	gets reset to 60 seconds each time there is activity (e.g. each time some
		1811	data or other life sign was received).
		1812
		1813	=over 4
		1814
		1815	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1816
		1817	This is the most obvious, but not the most simple way: In the beginning,
		1818	start the watcher:
		1819
		1820	ev_timer_init (timer, callback, 60., 0.);
		1821	ev_timer_start (loop, timer);
		1822
		1823	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1824	and start it again:
		1825
		1826	ev_timer_stop (loop, timer);
		1827	ev_timer_set (timer, 60., 0.);
		1828	ev_timer_start (loop, timer);
		1829
		1830	This is relatively simple to implement, but means that each time there is
		1831	some activity, libev will first have to remove the timer from its internal
		1832	data structure and then add it again. Libev tries to be fast, but it's
		1833	still not a constant-time operation.
		1834
		1835	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1836
		1837	This is the easiest way, and involves using C<ev_timer_again> instead of
		1838	C<ev_timer_start>.
		1839
		1840	To implement this, configure an C<ev_timer> with a C<repeat> value
		1841	of C<60> and then call C<ev_timer_again> at start and each time you
		1842	successfully read or write some data. If you go into an idle state where
		1843	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1844	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1845
		1846	That means you can ignore both the C<ev_timer_start> function and the
		1847	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1848	member and C<ev_timer_again>.
		1849
		1850	At start:
		1851
		1852	ev_init (timer, callback);
		1853	timer->repeat = 60.;
		1854	ev_timer_again (loop, timer);
		1855
		1856	Each time there is some activity:
		1857
		1858	ev_timer_again (loop, timer);
		1859
		1860	It is even possible to change the time-out on the fly, regardless of
		1861	whether the watcher is active or not:
		1862
		1863	timer->repeat = 30.;
		1864	ev_timer_again (loop, timer);
		1865
		1866	This is slightly more efficient then stopping/starting the timer each time
		1867	you want to modify its timeout value, as libev does not have to completely
		1868	remove and re-insert the timer from/into its internal data structure.
		1869
		1870	It is, however, even simpler than the "obvious" way to do it.
		1871
		1872	=item 3. Let the timer time out, but then re-arm it as required.
		1873
		1874	This method is more tricky, but usually most efficient: Most timeouts are
		1875	relatively long compared to the intervals between other activity - in
		1876	our example, within 60 seconds, there are usually many I/O events with
		1877	associated activity resets.
		1878
		1879	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1880	but remember the time of last activity, and check for a real timeout only
		1881	within the callback:
		1882
		1883	ev_tstamp timeout = 60.;
		1884	ev_tstamp last_activity; // time of last activity
		1885	ev_timer timer;
		1886
		1887	static void
		1888	callback (EV_P_ ev_timer *w, int revents)
		1889	{
		1890	// calculate when the timeout would happen
		1891	ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
		1892
		1893	// if negative, it means we the timeout already occurred
		1894	if (after < 0.)
		1895	{
		1896	// timeout occurred, take action
		1897	}
		1898	else
		1899	{
		1900	// callback was invoked, but there was some recent
		1901	// activity. simply restart the timer to time out
		1902	// after "after" seconds, which is the earliest time
		1903	// the timeout can occur.
		1904	ev_timer_set (w, after, 0.);
		1905	ev_timer_start (EV_A_ w);
		1906	}
		1907	}
		1908
		1909	To summarise the callback: first calculate in how many seconds the
		1910	timeout will occur (by calculating the absolute time when it would occur,
		1911	C<last_activity + timeout>, and subtracting the current time, C<ev_now
		1912	(EV_A)> from that).
		1913
		1914	If this value is negative, then we are already past the timeout, i.e. we
		1915	timed out, and need to do whatever is needed in this case.
		1916
		1917	Otherwise, we now the earliest time at which the timeout would trigger,
		1918	and simply start the timer with this timeout value.
		1919
		1920	In other words, each time the callback is invoked it will check whether
		1921	the timeout occurred. If not, it will simply reschedule itself to check
		1922	again at the earliest time it could time out. Rinse. Repeat.
		1923
		1924	This scheme causes more callback invocations (about one every 60 seconds
		1925	minus half the average time between activity), but virtually no calls to
		1926	libev to change the timeout.
		1927
		1928	To start the machinery, simply initialise the watcher and set
		1929	C<last_activity> to the current time (meaning there was some activity just
		1930	now), then call the callback, which will "do the right thing" and start
		1931	the timer:
		1932
		1933	last_activity = ev_now (EV_A);
		1934	ev_init (&timer, callback);
		1935	callback (EV_A_ &timer, 0);
		1936
		1937	When there is some activity, simply store the current time in
		1938	C<last_activity>, no libev calls at all:
		1939
		1940	if (activity detected)
		1941	last_activity = ev_now (EV_A);
		1942
		1943	When your timeout value changes, then the timeout can be changed by simply
		1944	providing a new value, stopping the timer and calling the callback, which
		1945	will again do the right thing (for example, time out immediately :).
		1946
		1947	timeout = new_value;
		1948	ev_timer_stop (EV_A_ &timer);
		1949	callback (EV_A_ &timer, 0);
		1950
		1951	This technique is slightly more complex, but in most cases where the
		1952	time-out is unlikely to be triggered, much more efficient.
		1953
		1954	=item 4. Wee, just use a double-linked list for your timeouts.
		1955
		1956	If there is not one request, but many thousands (millions...), all
		1957	employing some kind of timeout with the same timeout value, then one can
		1958	do even better:
		1959
		1960	When starting the timeout, calculate the timeout value and put the timeout
		1961	at the I<end> of the list.
		1962
		1963	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1964	the list is expected to fire (for example, using the technique #3).
		1965
		1966	When there is some activity, remove the timer from the list, recalculate
		1967	the timeout, append it to the end of the list again, and make sure to
		1968	update the C<ev_timer> if it was taken from the beginning of the list.
		1969
		1970	This way, one can manage an unlimited number of timeouts in O(1) time for
		1971	starting, stopping and updating the timers, at the expense of a major
		1972	complication, and having to use a constant timeout. The constant timeout
		1973	ensures that the list stays sorted.
		1974
		1975	=back
		1976
		1977	So which method the best?
		1978
		1979	Method #2 is a simple no-brain-required solution that is adequate in most
		1980	situations. Method #3 requires a bit more thinking, but handles many cases
		1981	better, and isn't very complicated either. In most case, choosing either
		1982	one is fine, with #3 being better in typical situations.
		1983
		1984	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1985	rather complicated, but extremely efficient, something that really pays
		1986	off after the first million or so of active timers, i.e. it's usually
		1987	overkill :)
		1988
		1989	=head3 The special problem of being too early
		1990
		1991	If you ask a timer to call your callback after three seconds, then
		1992	you expect it to be invoked after three seconds - but of course, this
		1993	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		1994	guaranteed to any precision by libev - imagine somebody suspending the
		1995	process with a STOP signal for a few hours for example.
		1996
		1997	So, libev tries to invoke your callback as soon as possible I<after> the
		1998	delay has occurred, but cannot guarantee this.
		1999
		2000	A less obvious failure mode is calling your callback too early: many event
		2001	loops compare timestamps with a "elapsed delay >= requested delay", but
		2002	this can cause your callback to be invoked much earlier than you would
		2003	expect.
		2004
		2005	To see why, imagine a system with a clock that only offers full second
		2006	resolution (think windows if you can't come up with a broken enough OS
		2007	yourself). If you schedule a one-second timer at the time 500.9, then the
		2008	event loop will schedule your timeout to elapse at a system time of 500
		2009	(500.9 truncated to the resolution) + 1, or 501.
		2010
		2011	If an event library looks at the timeout 0.1s later, it will see "501 >=
		2012	501" and invoke the callback 0.1s after it was started, even though a
		2013	one-second delay was requested - this is being "too early", despite best
		2014	intentions.
		2015
		2016	This is the reason why libev will never invoke the callback if the elapsed
		2017	delay equals the requested delay, but only when the elapsed delay is
		2018	larger than the requested delay. In the example above, libev would only invoke
		2019	the callback at system time 502, or 1.1s after the timer was started.
		2020
		2021	So, while libev cannot guarantee that your callback will be invoked
		2022	exactly when requested, it I<can> and I<does> guarantee that the requested
		2023	delay has actually elapsed, or in other words, it always errs on the "too
		2024	late" side of things.
		2025
		2026	=head3 The special problem of time updates
		2027
		2028	Establishing the current time is a costly operation (it usually takes
		2029	at least one system call): EV therefore updates its idea of the current
		2030	time only before and after C<ev_run> collects new events, which causes a
		2031	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		2032	lots of events in one iteration.
1070		2033
1071	The relative timeouts are calculated relative to the C<ev_now ()>	2034	The relative timeouts are calculated relative to the C<ev_now ()>
1072	time. This is usually the right thing as this timestamp refers to the time	2035	time. This is usually the right thing as this timestamp refers to the time
1073	of the event triggering whatever timeout you are modifying/starting. If	2036	of the event triggering whatever timeout you are modifying/starting. If
1074	you suspect event processing to be delayed and you I<need> to base the timeout	2037	you suspect event processing to be delayed and you I<need> to base the
1075	on the current time, use something like this to adjust for this:	2038	timeout on the current time, use something like the following to adjust
		2039	for it:
1076		2040
1077	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2041	ev_timer_set (&timer, after + (ev_time () - ev_now ()), 0.);
1078		2042
1079	The callback is guarenteed to be invoked only when its timeout has passed,	2043	If the event loop is suspended for a long time, you can also force an
1080	but if multiple timers become ready during the same loop iteration then	2044	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1081	order of execution is undefined.	2045	()>, although that will push the event time of all outstanding events
		2046	further into the future.
		2047
		2048	=head3 The special problem of unsynchronised clocks
		2049
		2050	Modern systems have a variety of clocks - libev itself uses the normal
		2051	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2052	jumps).
		2053
		2054	Neither of these clocks is synchronised with each other or any other clock
		2055	on the system, so C<ev_time ()> might return a considerably different time
		2056	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2057	a call to C<gettimeofday> might return a second count that is one higher
		2058	than a directly following call to C<time>.
		2059
		2060	The moral of this is to only compare libev-related timestamps with
		2061	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2062	a second or so.
		2063
		2064	One more problem arises due to this lack of synchronisation: if libev uses
		2065	the system monotonic clock and you compare timestamps from C<ev_time>
		2066	or C<ev_now> from when you started your timer and when your callback is
		2067	invoked, you will find that sometimes the callback is a bit "early".
		2068
		2069	This is because C<ev_timer>s work in real time, not wall clock time, so
		2070	libev makes sure your callback is not invoked before the delay happened,
		2071	I<measured according to the real time>, not the system clock.
		2072
		2073	If your timeouts are based on a physical timescale (e.g. "time out this
		2074	connection after 100 seconds") then this shouldn't bother you as it is
		2075	exactly the right behaviour.
		2076
		2077	If you want to compare wall clock/system timestamps to your timers, then
		2078	you need to use C<ev_periodic>s, as these are based on the wall clock
		2079	time, where your comparisons will always generate correct results.
		2080
		2081	=head3 The special problems of suspended animation
		2082
		2083	When you leave the server world it is quite customary to hit machines that
		2084	can suspend/hibernate - what happens to the clocks during such a suspend?
		2085
		2086	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		2087	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		2088	to run until the system is suspended, but they will not advance while the
		2089	system is suspended. That means, on resume, it will be as if the program
		2090	was frozen for a few seconds, but the suspend time will not be counted
		2091	towards C<ev_timer> when a monotonic clock source is used. The real time
		2092	clock advanced as expected, but if it is used as sole clocksource, then a
		2093	long suspend would be detected as a time jump by libev, and timers would
		2094	be adjusted accordingly.
		2095
		2096	I would not be surprised to see different behaviour in different between
		2097	operating systems, OS versions or even different hardware.
		2098
		2099	The other form of suspend (job control, or sending a SIGSTOP) will see a
		2100	time jump in the monotonic clocks and the realtime clock. If the program
		2101	is suspended for a very long time, and monotonic clock sources are in use,
		2102	then you can expect C<ev_timer>s to expire as the full suspension time
		2103	will be counted towards the timers. When no monotonic clock source is in
		2104	use, then libev will again assume a timejump and adjust accordingly.
		2105
		2106	It might be beneficial for this latter case to call C<ev_suspend>
		2107	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		2108	deterministic behaviour in this case (you can do nothing against
		2109	C<SIGSTOP>).
1082		2110
1083	=head3 Watcher-Specific Functions and Data Members	2111	=head3 Watcher-Specific Functions and Data Members
1084		2112
1085	=over 4	2113	=over 4
1086		2114
1087	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2115	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
1088		2116
1089	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2117	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
1090		2118
1091	Configure the timer to trigger after C<after> seconds. If C<repeat> is	2119	Configure the timer to trigger after C<after> seconds (fractional and
1092	C<0.>, then it will automatically be stopped. If it is positive, then the	2120	negative values are supported). If C<repeat> is C<0.>, then it will
		2121	automatically be stopped once the timeout is reached. If it is positive,
1093	timer will automatically be configured to trigger again C<repeat> seconds	2122	then the timer will automatically be configured to trigger again C<repeat>
1094	later, again, and again, until stopped manually.	2123	seconds later, again, and again, until stopped manually.
1095		2124
1096	The timer itself will do a best-effort at avoiding drift, that is, if you	2125	The timer itself will do a best-effort at avoiding drift, that is, if
1097	configure a timer to trigger every 10 seconds, then it will trigger at	2126	you configure a timer to trigger every 10 seconds, then it will normally
1098	exactly 10 second intervals. If, however, your program cannot keep up with	2127	trigger at exactly 10 second intervals. If, however, your program cannot
1099	the timer (because it takes longer than those 10 seconds to do stuff) the	2128	keep up with the timer (because it takes longer than those 10 seconds to
1100	timer will not fire more than once per event loop iteration.	2129	do stuff) the timer will not fire more than once per event loop iteration.
1101		2130
1102	=item ev_timer_again (loop)	2131	=item ev_timer_again (loop, ev_timer *)
1103		2132
1104	This will act as if the timer timed out and restart it again if it is	2133	This will act as if the timer timed out, and restarts it again if it is
1105	repeating. The exact semantics are:	2134	repeating. It basically works like calling C<ev_timer_stop>, updating the
		2135	timeout to the C<repeat> value and calling C<ev_timer_start>.
1106		2136
		2137	The exact semantics are as in the following rules, all of which will be
		2138	applied to the watcher:
		2139
		2140	=over 4
		2141
1107	If the timer is pending, its pending status is cleared.	2142	=item If the timer is pending, the pending status is always cleared.
1108		2143
1109	If the timer is started but nonrepeating, stop it (as if it timed out).	2144	=item If the timer is started but non-repeating, stop it (as if it timed
		2145	out, without invoking it).
1110		2146
1111	If the timer is repeating, either start it if necessary (with the	2147	=item If the timer is repeating, make the C<repeat> value the new timeout
1112	C<repeat> value), or reset the running timer to the C<repeat> value.	2148	and start the timer, if necessary.
1113		2149
1114	This sounds a bit complicated, but here is a useful and typical	2150	=back
1115	example: Imagine you have a tcp connection and you want a so-called idle
1116	timeout, that is, you want to be called when there have been, say, 60
1117	seconds of inactivity on the socket. The easiest way to do this is to
1118	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
1119	C<ev_timer_again> each time you successfully read or write some data. If
1120	you go into an idle state where you do not expect data to travel on the
1121	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
1122	automatically restart it if need be.
1123		2151
1124	That means you can ignore the C<after> value and C<ev_timer_start>	2152	This sounds a bit complicated, see L</Be smart about timeouts>, above, for a
1125	altogether and only ever use the C<repeat> value and C<ev_timer_again>:	2153	usage example.
1126		2154
1127	ev_timer_init (timer, callback, 0., 5.);	2155	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
1128	ev_timer_again (loop, timer);
1129	...
1130	timer->again = 17.;
1131	ev_timer_again (loop, timer);
1132	...
1133	timer->again = 10.;
1134	ev_timer_again (loop, timer);
1135		2156
1136	This is more slightly efficient then stopping/starting the timer each time	2157	Returns the remaining time until a timer fires. If the timer is active,
1137	you want to modify its timeout value.	2158	then this time is relative to the current event loop time, otherwise it's
		2159	the timeout value currently configured.
		2160
		2161	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2162	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2163	will return C<4>. When the timer expires and is restarted, it will return
		2164	roughly C<7> (likely slightly less as callback invocation takes some time,
		2165	too), and so on.
1138		2166
1139	=item ev_tstamp repeat [read-write]	2167	=item ev_tstamp repeat [read-write]
1140		2168
1141	The current C<repeat> value. Will be used each time the watcher times out	2169	The current C<repeat> value. Will be used each time the watcher times out
1142	or C<ev_timer_again> is called and determines the next timeout (if any),	2170	or C<ev_timer_again> is called, and determines the next timeout (if any),
1143	which is also when any modifications are taken into account.	2171	which is also when any modifications are taken into account.
1144		2172
1145	=back	2173	=back
1146		2174
		2175	=head3 Examples
		2176
1147	Example: Create a timer that fires after 60 seconds.	2177	Example: Create a timer that fires after 60 seconds.
1148		2178
1149	static void	2179	static void
1150	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2180	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1151	{	2181	{
1152	.. one minute over, w is actually stopped right here	2182	.. one minute over, w is actually stopped right here
1153	}	2183	}
1154		2184
1155	struct ev_timer mytimer;	2185	ev_timer mytimer;
1156	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	2186	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1157	ev_timer_start (loop, &mytimer);	2187	ev_timer_start (loop, &mytimer);
1158		2188
1159	Example: Create a timeout timer that times out after 10 seconds of	2189	Example: Create a timeout timer that times out after 10 seconds of
1160	inactivity.	2190	inactivity.
1161		2191
1162	static void	2192	static void
1163	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2193	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1164	{	2194	{
1165	.. ten seconds without any activity	2195	.. ten seconds without any activity
1166	}	2196	}
1167		2197
1168	struct ev_timer mytimer;	2198	ev_timer mytimer;
1169	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	2199	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1170	ev_timer_again (&mytimer); /* start timer */	2200	ev_timer_again (&mytimer); /* start timer */
1171	ev_loop (loop, 0);	2201	ev_run (loop, 0);
1172		2202
1173	// and in some piece of code that gets executed on any "activity":	2203	// and in some piece of code that gets executed on any "activity":
1174	// reset the timeout to start ticking again at 10 seconds	2204	// reset the timeout to start ticking again at 10 seconds
1175	ev_timer_again (&mytimer);	2205	ev_timer_again (&mytimer);
1176		2206
1177		2207
1178	=head2 C<ev_periodic> - to cron or not to cron?	2208	=head2 C<ev_periodic> - to cron or not to cron?
1179		2209
1180	Periodic watchers are also timers of a kind, but they are very versatile	2210	Periodic watchers are also timers of a kind, but they are very versatile
1181	(and unfortunately a bit complex).	2211	(and unfortunately a bit complex).
1182		2212
1183	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2213	Unlike C<ev_timer>, periodic watchers are not based on real time (or
1184	but on wallclock time (absolute time). You can tell a periodic watcher	2214	relative time, the physical time that passes) but on wall clock time
1185	to trigger "at" some specific point in time. For example, if you tell a	2215	(absolute time, the thing you can read on your calendar or clock). The
1186	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	2216	difference is that wall clock time can run faster or slower than real
1187	+ 10.>) and then reset your system clock to the last year, then it will	2217	time, and time jumps are not uncommon (e.g. when you adjust your
1188	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2218	wrist-watch).
1189	roughly 10 seconds later).
1190		2219
1191	They can also be used to implement vastly more complex timers, such as	2220	You can tell a periodic watcher to trigger after some specific point
1192	triggering an event on each midnight, local time or other, complicated,	2221	in time: for example, if you tell a periodic watcher to trigger "in 10
1193	rules.	2222	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2223	not a delay) and then reset your system clock to January of the previous
		2224	year, then it will take a year or more to trigger the event (unlike an
		2225	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2226	it, as it uses a relative timeout).
1194		2227
		2228	C<ev_periodic> watchers can also be used to implement vastly more complex
		2229	timers, such as triggering an event on each "midnight, local time", or
		2230	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2231	those cannot react to time jumps.
		2232
1195	As with timers, the callback is guarenteed to be invoked only when the	2233	As with timers, the callback is guaranteed to be invoked only when the
1196	time (C<at>) has been passed, but if multiple periodic timers become ready	2234	point in time where it is supposed to trigger has passed. If multiple
1197	during the same loop iteration then order of execution is undefined.	2235	timers become ready during the same loop iteration then the ones with
		2236	earlier time-out values are invoked before ones with later time-out values
		2237	(but this is no longer true when a callback calls C<ev_run> recursively).
1198		2238
1199	=head3 Watcher-Specific Functions and Data Members	2239	=head3 Watcher-Specific Functions and Data Members
1200		2240
1201	=over 4	2241	=over 4
1202		2242
1203	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2243	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1204		2244
1205	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2245	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
1206		2246
1207	Lots of arguments, lets sort it out... There are basically three modes of	2247	Lots of arguments, let's sort it out... There are basically three modes of
1208	operation, and we will explain them from simplest to complex:	2248	operation, and we will explain them from simplest to most complex:
1209		2249
1210	=over 4	2250	=over 4
1211		2251
1212	=item * absolute timer (at = time, interval = reschedule_cb = 0)	2252	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
1213		2253
1214	In this configuration the watcher triggers an event at the wallclock time	2254	In this configuration the watcher triggers an event after the wall clock
1215	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2255	time C<offset> has passed. It will not repeat and will not adjust when a
1216	that is, if it is to be run at January 1st 2011 then it will run when the	2256	time jump occurs, that is, if it is to be run at January 1st 2011 then it
1217	system time reaches or surpasses this time.	2257	will be stopped and invoked when the system clock reaches or surpasses
		2258	this point in time.
1218		2259
1219	=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	2260	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
1220		2261
1221	In this mode the watcher will always be scheduled to time out at the next	2262	In this mode the watcher will always be scheduled to time out at the next
1222	C<at + N * interval> time (for some integer N, which can also be negative)	2263	C<offset + N * interval> time (for some integer N, which can also be
1223	and then repeat, regardless of any time jumps.	2264	negative) and then repeat, regardless of any time jumps. The C<offset>
		2265	argument is merely an offset into the C<interval> periods.
1224		2266
1225	This can be used to create timers that do not drift with respect to system	2267	This can be used to create timers that do not drift with respect to the
1226	time:	2268	system clock, for example, here is an C<ev_periodic> that triggers each
		2269	hour, on the hour (with respect to UTC):
1227		2270
1228	ev_periodic_set (&periodic, 0., 3600., 0);	2271	ev_periodic_set (&periodic, 0., 3600., 0);
1229		2272
1230	This doesn't mean there will always be 3600 seconds in between triggers,	2273	This doesn't mean there will always be 3600 seconds in between triggers,
1231	but only that the the callback will be called when the system time shows a	2274	but only that the callback will be called when the system time shows a
1232	full hour (UTC), or more correctly, when the system time is evenly divisible	2275	full hour (UTC), or more correctly, when the system time is evenly divisible
1233	by 3600.	2276	by 3600.
1234		2277
1235	Another way to think about it (for the mathematically inclined) is that	2278	Another way to think about it (for the mathematically inclined) is that
1236	C<ev_periodic> will try to run the callback in this mode at the next possible	2279	C<ev_periodic> will try to run the callback in this mode at the next possible
1237	time where C<time = at (mod interval)>, regardless of any time jumps.	2280	time where C<time = offset (mod interval)>, regardless of any time jumps.
1238		2281
1239	For numerical stability it is preferable that the C<at> value is near	2282	The C<interval> I<MUST> be positive, and for numerical stability, the
1240	C<ev_now ()> (the current time), but there is no range requirement for	2283	interval value should be higher than C<1/8192> (which is around 100
1241	this value.	2284	microseconds) and C<offset> should be higher than C<0> and should have
		2285	at most a similar magnitude as the current time (say, within a factor of
		2286	ten). Typical values for offset are, in fact, C<0> or something between
		2287	C<0> and C<interval>, which is also the recommended range.
1242		2288
		2289	Note also that there is an upper limit to how often a timer can fire (CPU
		2290	speed for example), so if C<interval> is very small then timing stability
		2291	will of course deteriorate. Libev itself tries to be exact to be about one
		2292	millisecond (if the OS supports it and the machine is fast enough).
		2293
1243	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)	2294	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
1244		2295
1245	In this mode the values for C<interval> and C<at> are both being	2296	In this mode the values for C<interval> and C<offset> are both being
1246	ignored. Instead, each time the periodic watcher gets scheduled, the	2297	ignored. Instead, each time the periodic watcher gets scheduled, the
1247	reschedule callback will be called with the watcher as first, and the	2298	reschedule callback will be called with the watcher as first, and the
1248	current time as second argument.	2299	current time as second argument.
1249		2300
1250	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2301	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
1251	ever, or make any event loop modifications>. If you need to stop it,	2302	or make ANY other event loop modifications whatsoever, unless explicitly
1252	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2303	allowed by documentation here>.
1253	starting an C<ev_prepare> watcher, which is legal).
1254		2304
		2305	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2306	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2307	only event loop modification you are allowed to do).
		2308
1255	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2309	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1256	ev_tstamp now)>, e.g.:	2310	*w, ev_tstamp now)>, e.g.:
1257		2311
		2312	static ev_tstamp
1258	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2313	my_rescheduler (ev_periodic *w, ev_tstamp now)
1259	{	2314	{
1260	return now + 60.;	2315	return now + 60.;
1261	}	2316	}
1262		2317
1263	It must return the next time to trigger, based on the passed time value	2318	It must return the next time to trigger, based on the passed time value
1264	(that is, the lowest time value larger than to the second argument). It	2319	(that is, the lowest time value larger than to the second argument). It
1265	will usually be called just before the callback will be triggered, but	2320	will usually be called just before the callback will be triggered, but
1266	might be called at other times, too.	2321	might be called at other times, too.
1267		2322
1268	NOTE: I<< This callback must always return a time that is later than the	2323	NOTE: I<< This callback must always return a time that is higher than or
1269	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2324	equal to the passed C<now> value >>.
1270		2325
1271	This can be used to create very complex timers, such as a timer that	2326	This can be used to create very complex timers, such as a timer that
1272	triggers on each midnight, local time. To do this, you would calculate the	2327	triggers on "next midnight, local time". To do this, you would calculate the
1273	next midnight after C<now> and return the timestamp value for this. How	2328	next midnight after C<now> and return the timestamp value for this. How
1274	you do this is, again, up to you (but it is not trivial, which is the main	2329	you do this is, again, up to you (but it is not trivial, which is the main
1275	reason I omitted it as an example).	2330	reason I omitted it as an example).
1276		2331
1277	=back	2332	=back
…		…
1281	Simply stops and restarts the periodic watcher again. This is only useful	2336	Simply stops and restarts the periodic watcher again. This is only useful
1282	when you changed some parameters or the reschedule callback would return	2337	when you changed some parameters or the reschedule callback would return
1283	a different time than the last time it was called (e.g. in a crond like	2338	a different time than the last time it was called (e.g. in a crond like
1284	program when the crontabs have changed).	2339	program when the crontabs have changed).
1285		2340
		2341	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2342
		2343	When active, returns the absolute time that the watcher is supposed
		2344	to trigger next. This is not the same as the C<offset> argument to
		2345	C<ev_periodic_set>, but indeed works even in interval and manual
		2346	rescheduling modes.
		2347
1286	=item ev_tstamp offset [read-write]	2348	=item ev_tstamp offset [read-write]
1287		2349
1288	When repeating, this contains the offset value, otherwise this is the	2350	When repeating, this contains the offset value, otherwise this is the
1289	absolute point in time (the C<at> value passed to C<ev_periodic_set>).	2351	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2352	although libev might modify this value for better numerical stability).
1290		2353
1291	Can be modified any time, but changes only take effect when the periodic	2354	Can be modified any time, but changes only take effect when the periodic
1292	timer fires or C<ev_periodic_again> is being called.	2355	timer fires or C<ev_periodic_again> is being called.
1293		2356
1294	=item ev_tstamp interval [read-write]	2357	=item ev_tstamp interval [read-write]
1295		2358
1296	The current interval value. Can be modified any time, but changes only	2359	The current interval value. Can be modified any time, but changes only
1297	take effect when the periodic timer fires or C<ev_periodic_again> is being	2360	take effect when the periodic timer fires or C<ev_periodic_again> is being
1298	called.	2361	called.
1299		2362
1300	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	2363	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1301		2364
1302	The current reschedule callback, or C<0>, if this functionality is	2365	The current reschedule callback, or C<0>, if this functionality is
1303	switched off. Can be changed any time, but changes only take effect when	2366	switched off. Can be changed any time, but changes only take effect when
1304	the periodic timer fires or C<ev_periodic_again> is being called.	2367	the periodic timer fires or C<ev_periodic_again> is being called.
1305		2368
1306	=item ev_tstamp at [read-only]
1307
1308	When active, contains the absolute time that the watcher is supposed to
1309	trigger next.
1310
1311	=back	2369	=back
1312		2370
		2371	=head3 Examples
		2372
1313	Example: Call a callback every hour, or, more precisely, whenever the	2373	Example: Call a callback every hour, or, more precisely, whenever the
1314	system clock is divisible by 3600. The callback invocation times have	2374	system time is divisible by 3600. The callback invocation times have
1315	potentially a lot of jittering, but good long-term stability.	2375	potentially a lot of jitter, but good long-term stability.
1316		2376
1317	static void	2377	static void
1318	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	2378	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
1319	{	2379	{
1320	... its now a full hour (UTC, or TAI or whatever your clock follows)	2380	... its now a full hour (UTC, or TAI or whatever your clock follows)
1321	}	2381	}
1322		2382
1323	struct ev_periodic hourly_tick;	2383	ev_periodic hourly_tick;
1324	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	2384	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1325	ev_periodic_start (loop, &hourly_tick);	2385	ev_periodic_start (loop, &hourly_tick);
1326		2386
1327	Example: The same as above, but use a reschedule callback to do it:	2387	Example: The same as above, but use a reschedule callback to do it:
1328		2388
1329	#include <math.h>	2389	#include <math.h>
1330		2390
1331	static ev_tstamp	2391	static ev_tstamp
1332	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	2392	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1333	{	2393	{
1334	return fmod (now, 3600.) + 3600.;	2394	return now + (3600. - fmod (now, 3600.));
1335	}	2395	}
1336		2396
1337	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	2397	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1338		2398
1339	Example: Call a callback every hour, starting now:	2399	Example: Call a callback every hour, starting now:
1340		2400
1341	struct ev_periodic hourly_tick;	2401	ev_periodic hourly_tick;
1342	ev_periodic_init (&hourly_tick, clock_cb,	2402	ev_periodic_init (&hourly_tick, clock_cb,
1343	fmod (ev_now (loop), 3600.), 3600., 0);	2403	fmod (ev_now (loop), 3600.), 3600., 0);
1344	ev_periodic_start (loop, &hourly_tick);	2404	ev_periodic_start (loop, &hourly_tick);
1345		2405
1346		2406
1347	=head2 C<ev_signal> - signal me when a signal gets signalled!	2407	=head2 C<ev_signal> - signal me when a signal gets signalled!
1348		2408
1349	Signal watchers will trigger an event when the process receives a specific	2409	Signal watchers will trigger an event when the process receives a specific
1350	signal one or more times. Even though signals are very asynchronous, libev	2410	signal one or more times. Even though signals are very asynchronous, libev
1351	will try it's best to deliver signals synchronously, i.e. as part of the	2411	will try its best to deliver signals synchronously, i.e. as part of the
1352	normal event processing, like any other event.	2412	normal event processing, like any other event.
1353		2413
		2414	If you want signals to be delivered truly asynchronously, just use
		2415	C<sigaction> as you would do without libev and forget about sharing
		2416	the signal. You can even use C<ev_async> from a signal handler to
		2417	synchronously wake up an event loop.
		2418
1354	You can configure as many watchers as you like per signal. Only when the	2419	You can configure as many watchers as you like for the same signal, but
1355	first watcher gets started will libev actually register a signal watcher	2420	only within the same loop, i.e. you can watch for C<SIGINT> in your
1356	with the kernel (thus it coexists with your own signal handlers as long	2421	default loop and for C<SIGIO> in another loop, but you cannot watch for
1357	as you don't register any with libev). Similarly, when the last signal	2422	C<SIGINT> in both the default loop and another loop at the same time. At
1358	watcher for a signal is stopped libev will reset the signal handler to	2423	the moment, C<SIGCHLD> is permanently tied to the default loop.
1359	SIG_DFL (regardless of what it was set to before).	2424
		2425	Only after the first watcher for a signal is started will libev actually
		2426	register something with the kernel. It thus coexists with your own signal
		2427	handlers as long as you don't register any with libev for the same signal.
		2428
		2429	If possible and supported, libev will install its handlers with
		2430	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2431	not be unduly interrupted. If you have a problem with system calls getting
		2432	interrupted by signals you can block all signals in an C<ev_check> watcher
		2433	and unblock them in an C<ev_prepare> watcher.
		2434
		2435	=head3 The special problem of inheritance over fork/execve/pthread_create
		2436
		2437	Both the signal mask (C<sigprocmask>) and the signal disposition
		2438	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2439	stopping it again), that is, libev might or might not block the signal,
		2440	and might or might not set or restore the installed signal handler (but
		2441	see C<EVFLAG_NOSIGMASK>).
		2442
		2443	While this does not matter for the signal disposition (libev never
		2444	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2445	C<execve>), this matters for the signal mask: many programs do not expect
		2446	certain signals to be blocked.
		2447
		2448	This means that before calling C<exec> (from the child) you should reset
		2449	the signal mask to whatever "default" you expect (all clear is a good
		2450	choice usually).
		2451
		2452	The simplest way to ensure that the signal mask is reset in the child is
		2453	to install a fork handler with C<pthread_atfork> that resets it. That will
		2454	catch fork calls done by libraries (such as the libc) as well.
		2455
		2456	In current versions of libev, the signal will not be blocked indefinitely
		2457	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2458	the window of opportunity for problems, it will not go away, as libev
		2459	I<has> to modify the signal mask, at least temporarily.
		2460
		2461	So I can't stress this enough: I<If you do not reset your signal mask when
		2462	you expect it to be empty, you have a race condition in your code>. This
		2463	is not a libev-specific thing, this is true for most event libraries.
		2464
		2465	=head3 The special problem of threads signal handling
		2466
		2467	POSIX threads has problematic signal handling semantics, specifically,
		2468	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2469	threads in a process block signals, which is hard to achieve.
		2470
		2471	When you want to use sigwait (or mix libev signal handling with your own
		2472	for the same signals), you can tackle this problem by globally blocking
		2473	all signals before creating any threads (or creating them with a fully set
		2474	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2475	loops. Then designate one thread as "signal receiver thread" which handles
		2476	these signals. You can pass on any signals that libev might be interested
		2477	in by calling C<ev_feed_signal>.
1360		2478
1361	=head3 Watcher-Specific Functions and Data Members	2479	=head3 Watcher-Specific Functions and Data Members
1362		2480
1363	=over 4	2481	=over 4
1364		2482
…		…
1373		2491
1374	The signal the watcher watches out for.	2492	The signal the watcher watches out for.
1375		2493
1376	=back	2494	=back
1377		2495
		2496	=head3 Examples
		2497
		2498	Example: Try to exit cleanly on SIGINT.
		2499
		2500	static void
		2501	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2502	{
		2503	ev_break (loop, EVBREAK_ALL);
		2504	}
		2505
		2506	ev_signal signal_watcher;
		2507	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2508	ev_signal_start (loop, &signal_watcher);
		2509
1378		2510
1379	=head2 C<ev_child> - watch out for process status changes	2511	=head2 C<ev_child> - watch out for process status changes
1380		2512
1381	Child watchers trigger when your process receives a SIGCHLD in response to	2513	Child watchers trigger when your process receives a SIGCHLD in response to
1382	some child status changes (most typically when a child of yours dies).	2514	some child status changes (most typically when a child of yours dies or
		2515	exits). It is permissible to install a child watcher I<after> the child
		2516	has been forked (which implies it might have already exited), as long
		2517	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2518	forking and then immediately registering a watcher for the child is fine,
		2519	but forking and registering a watcher a few event loop iterations later or
		2520	in the next callback invocation is not.
		2521
		2522	Only the default event loop is capable of handling signals, and therefore
		2523	you can only register child watchers in the default event loop.
		2524
		2525	Due to some design glitches inside libev, child watchers will always be
		2526	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2527	libev)
		2528
		2529	=head3 Process Interaction
		2530
		2531	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2532	initialised. This is necessary to guarantee proper behaviour even if the
		2533	first child watcher is started after the child exits. The occurrence
		2534	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2535	synchronously as part of the event loop processing. Libev always reaps all
		2536	children, even ones not watched.
		2537
		2538	=head3 Overriding the Built-In Processing
		2539
		2540	Libev offers no special support for overriding the built-in child
		2541	processing, but if your application collides with libev's default child
		2542	handler, you can override it easily by installing your own handler for
		2543	C<SIGCHLD> after initialising the default loop, and making sure the
		2544	default loop never gets destroyed. You are encouraged, however, to use an
		2545	event-based approach to child reaping and thus use libev's support for
		2546	that, so other libev users can use C<ev_child> watchers freely.
		2547
		2548	=head3 Stopping the Child Watcher
		2549
		2550	Currently, the child watcher never gets stopped, even when the
		2551	child terminates, so normally one needs to stop the watcher in the
		2552	callback. Future versions of libev might stop the watcher automatically
		2553	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2554	problem).
1383		2555
1384	=head3 Watcher-Specific Functions and Data Members	2556	=head3 Watcher-Specific Functions and Data Members
1385		2557
1386	=over 4	2558	=over 4
1387		2559
1388	=item ev_child_init (ev_child *, callback, int pid)	2560	=item ev_child_init (ev_child *, callback, int pid, int trace)
1389		2561
1390	=item ev_child_set (ev_child *, int pid)	2562	=item ev_child_set (ev_child *, int pid, int trace)
1391		2563
1392	Configures the watcher to wait for status changes of process C<pid> (or	2564	Configures the watcher to wait for status changes of process C<pid> (or
1393	I<any> process if C<pid> is specified as C<0>). The callback can look	2565	I<any> process if C<pid> is specified as C<0>). The callback can look
1394	at the C<rstatus> member of the C<ev_child> watcher structure to see	2566	at the C<rstatus> member of the C<ev_child> watcher structure to see
1395	the status word (use the macros from C<sys/wait.h> and see your systems	2567	the status word (use the macros from C<sys/wait.h> and see your systems
1396	C<waitpid> documentation). The C<rpid> member contains the pid of the	2568	C<waitpid> documentation). The C<rpid> member contains the pid of the
1397	process causing the status change.	2569	process causing the status change. C<trace> must be either C<0> (only
		2570	activate the watcher when the process terminates) or C<1> (additionally
		2571	activate the watcher when the process is stopped or continued).
1398		2572
1399	=item int pid [read-only]	2573	=item int pid [read-only]
1400		2574
1401	The process id this watcher watches out for, or C<0>, meaning any process id.	2575	The process id this watcher watches out for, or C<0>, meaning any process id.
1402		2576
…		…
1409	The process exit/trace status caused by C<rpid> (see your systems	2583	The process exit/trace status caused by C<rpid> (see your systems
1410	C<waitpid> and C<sys/wait.h> documentation for details).	2584	C<waitpid> and C<sys/wait.h> documentation for details).
1411		2585
1412	=back	2586	=back
1413		2587
1414	Example: Try to exit cleanly on SIGINT and SIGTERM.	2588	=head3 Examples
1415		2589
		2590	Example: C<fork()> a new process and install a child handler to wait for
		2591	its completion.
		2592
		2593	ev_child cw;
		2594
1416	static void	2595	static void
1417	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	2596	child_cb (EV_P_ ev_child *w, int revents)
1418	{	2597	{
1419	ev_unloop (loop, EVUNLOOP_ALL);	2598	ev_child_stop (EV_A_ w);
		2599	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1420	}	2600	}
1421		2601
1422	struct ev_signal signal_watcher;	2602	pid_t pid = fork ();
1423	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	2603
1424	ev_signal_start (loop, &sigint_cb);	2604	if (pid < 0)
		2605	// error
		2606	else if (pid == 0)
		2607	{
		2608	// the forked child executes here
		2609	exit (1);
		2610	}
		2611	else
		2612	{
		2613	ev_child_init (&cw, child_cb, pid, 0);
		2614	ev_child_start (EV_DEFAULT_ &cw);
		2615	}
1425		2616
1426		2617
1427	=head2 C<ev_stat> - did the file attributes just change?	2618	=head2 C<ev_stat> - did the file attributes just change?
1428		2619
1429	This watches a filesystem path for attribute changes. That is, it calls	2620	This watches a file system path for attribute changes. That is, it calls
1430	C<stat> regularly (or when the OS says it changed) and sees if it changed	2621	C<stat> on that path in regular intervals (or when the OS says it changed)
1431	compared to the last time, invoking the callback if it did.	2622	and sees if it changed compared to the last time, invoking the callback
		2623	if it did. Starting the watcher C<stat>'s the file, so only changes that
		2624	happen after the watcher has been started will be reported.
1432		2625
1433	The path does not need to exist: changing from "path exists" to "path does	2626	The path does not need to exist: changing from "path exists" to "path does
1434	not exist" is a status change like any other. The condition "path does	2627	not exist" is a status change like any other. The condition "path does not
1435	not exist" is signified by the C<st_nlink> field being zero (which is	2628	exist" (or more correctly "path cannot be stat'ed") is signified by the
1436	otherwise always forced to be at least one) and all the other fields of	2629	C<st_nlink> field being zero (which is otherwise always forced to be at
1437	the stat buffer having unspecified contents.	2630	least one) and all the other fields of the stat buffer having unspecified
		2631	contents.
1438		2632
1439	The path I<should> be absolute and I<must not> end in a slash. If it is	2633	The path I<must not> end in a slash or contain special components such as
		2634	C<.> or C<..>. The path I<should> be absolute: If it is relative and
1440	relative and your working directory changes, the behaviour is undefined.	2635	your working directory changes, then the behaviour is undefined.
1441		2636
1442	Since there is no standard to do this, the portable implementation simply	2637	Since there is no portable change notification interface available, the
1443	calls C<stat (2)> regularly on the path to see if it changed somehow. You	2638	portable implementation simply calls C<stat(2)> regularly on the path
1444	can specify a recommended polling interval for this case. If you specify	2639	to see if it changed somehow. You can specify a recommended polling
1445	a polling interval of C<0> (highly recommended!) then a I<suitable,	2640	interval for this case. If you specify a polling interval of C<0> (highly
1446	unspecified default> value will be used (which you can expect to be around	2641	recommended!) then a I<suitable, unspecified default> value will be used
1447	five seconds, although this might change dynamically). Libev will also	2642	(which you can expect to be around five seconds, although this might
1448	impose a minimum interval which is currently around C<0.1>, but thats	2643	change dynamically). Libev will also impose a minimum interval which is
1449	usually overkill.	2644	currently around C<0.1>, but that's usually overkill.
1450		2645
1451	This watcher type is not meant for massive numbers of stat watchers,	2646	This watcher type is not meant for massive numbers of stat watchers,
1452	as even with OS-supported change notifications, this can be	2647	as even with OS-supported change notifications, this can be
1453	resource-intensive.	2648	resource-intensive.
1454		2649
1455	At the time of this writing, only the Linux inotify interface is	2650	At the time of this writing, the only OS-specific interface implemented
1456	implemented (implementing kqueue support is left as an exercise for the	2651	is the Linux inotify interface (implementing kqueue support is left as an
1457	reader). Inotify will be used to give hints only and should not change the	2652	exercise for the reader. Note, however, that the author sees no way of
1458	semantics of C<ev_stat> watchers, which means that libev sometimes needs	2653	implementing C<ev_stat> semantics with kqueue, except as a hint).
1459	to fall back to regular polling again even with inotify, but changes are	2654
1460	usually detected immediately, and if the file exists there will be no	2655	=head3 ABI Issues (Largefile Support)
1461	polling.	2656
		2657	Libev by default (unless the user overrides this) uses the default
		2658	compilation environment, which means that on systems with large file
		2659	support disabled by default, you get the 32 bit version of the stat
		2660	structure. When using the library from programs that change the ABI to
		2661	use 64 bit file offsets the programs will fail. In that case you have to
		2662	compile libev with the same flags to get binary compatibility. This is
		2663	obviously the case with any flags that change the ABI, but the problem is
		2664	most noticeably displayed with ev_stat and large file support.
		2665
		2666	The solution for this is to lobby your distribution maker to make large
		2667	file interfaces available by default (as e.g. FreeBSD does) and not
		2668	optional. Libev cannot simply switch on large file support because it has
		2669	to exchange stat structures with application programs compiled using the
		2670	default compilation environment.
		2671
		2672	=head3 Inotify and Kqueue
		2673
		2674	When C<inotify (7)> support has been compiled into libev and present at
		2675	runtime, it will be used to speed up change detection where possible. The
		2676	inotify descriptor will be created lazily when the first C<ev_stat>
		2677	watcher is being started.
		2678
		2679	Inotify presence does not change the semantics of C<ev_stat> watchers
		2680	except that changes might be detected earlier, and in some cases, to avoid
		2681	making regular C<stat> calls. Even in the presence of inotify support
		2682	there are many cases where libev has to resort to regular C<stat> polling,
		2683	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2684	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2685	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2686	xfs are fully working) libev usually gets away without polling.
		2687
		2688	There is no support for kqueue, as apparently it cannot be used to
		2689	implement this functionality, due to the requirement of having a file
		2690	descriptor open on the object at all times, and detecting renames, unlinks
		2691	etc. is difficult.
		2692
		2693	=head3 C<stat ()> is a synchronous operation
		2694
		2695	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2696	the process. The exception are C<ev_stat> watchers - those call C<stat
		2697	()>, which is a synchronous operation.
		2698
		2699	For local paths, this usually doesn't matter: unless the system is very
		2700	busy or the intervals between stat's are large, a stat call will be fast,
		2701	as the path data is usually in memory already (except when starting the
		2702	watcher).
		2703
		2704	For networked file systems, calling C<stat ()> can block an indefinite
		2705	time due to network issues, and even under good conditions, a stat call
		2706	often takes multiple milliseconds.
		2707
		2708	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2709	paths, although this is fully supported by libev.
		2710
		2711	=head3 The special problem of stat time resolution
		2712
		2713	The C<stat ()> system call only supports full-second resolution portably,
		2714	and even on systems where the resolution is higher, most file systems
		2715	still only support whole seconds.
		2716
		2717	That means that, if the time is the only thing that changes, you can
		2718	easily miss updates: on the first update, C<ev_stat> detects a change and
		2719	calls your callback, which does something. When there is another update
		2720	within the same second, C<ev_stat> will be unable to detect unless the
		2721	stat data does change in other ways (e.g. file size).
		2722
		2723	The solution to this is to delay acting on a change for slightly more
		2724	than a second (or till slightly after the next full second boundary), using
		2725	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2726	ev_timer_again (loop, w)>).
		2727
		2728	The C<.02> offset is added to work around small timing inconsistencies
		2729	of some operating systems (where the second counter of the current time
		2730	might be be delayed. One such system is the Linux kernel, where a call to
		2731	C<gettimeofday> might return a timestamp with a full second later than
		2732	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2733	update file times then there will be a small window where the kernel uses
		2734	the previous second to update file times but libev might already execute
		2735	the timer callback).
1462		2736
1463	=head3 Watcher-Specific Functions and Data Members	2737	=head3 Watcher-Specific Functions and Data Members
1464		2738
1465	=over 4	2739	=over 4
1466		2740
…		…
1472	C<path>. The C<interval> is a hint on how quickly a change is expected to	2746	C<path>. The C<interval> is a hint on how quickly a change is expected to
1473	be detected and should normally be specified as C<0> to let libev choose	2747	be detected and should normally be specified as C<0> to let libev choose
1474	a suitable value. The memory pointed to by C<path> must point to the same	2748	a suitable value. The memory pointed to by C<path> must point to the same
1475	path for as long as the watcher is active.	2749	path for as long as the watcher is active.
1476		2750
1477	The callback will be receive C<EV_STAT> when a change was detected,	2751	The callback will receive an C<EV_STAT> event when a change was detected,
1478	relative to the attributes at the time the watcher was started (or the	2752	relative to the attributes at the time the watcher was started (or the
1479	last change was detected).	2753	last change was detected).
1480		2754
1481	=item ev_stat_stat (ev_stat *)	2755	=item ev_stat_stat (loop, ev_stat *)
1482		2756
1483	Updates the stat buffer immediately with new values. If you change the	2757	Updates the stat buffer immediately with new values. If you change the
1484	watched path in your callback, you could call this fucntion to avoid	2758	watched path in your callback, you could call this function to avoid
1485	detecting this change (while introducing a race condition). Can also be	2759	detecting this change (while introducing a race condition if you are not
1486	useful simply to find out the new values.	2760	the only one changing the path). Can also be useful simply to find out the
		2761	new values.
1487		2762
1488	=item ev_statdata attr [read-only]	2763	=item ev_statdata attr [read-only]
1489		2764
1490	The most-recently detected attributes of the file. Although the type is of	2765	The most-recently detected attributes of the file. Although the type is
1491	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types	2766	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1492	suitable for your system. If the C<st_nlink> member is C<0>, then there	2767	suitable for your system, but you can only rely on the POSIX-standardised
		2768	members to be present. If the C<st_nlink> member is C<0>, then there was
1493	was some error while C<stat>ing the file.	2769	some error while C<stat>ing the file.
1494		2770
1495	=item ev_statdata prev [read-only]	2771	=item ev_statdata prev [read-only]
1496		2772
1497	The previous attributes of the file. The callback gets invoked whenever	2773	The previous attributes of the file. The callback gets invoked whenever
1498	C<prev> != C<attr>.	2774	C<prev> != C<attr>, or, more precisely, one or more of these members
		2775	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2776	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
1499		2777
1500	=item ev_tstamp interval [read-only]	2778	=item ev_tstamp interval [read-only]
1501		2779
1502	The specified interval.	2780	The specified interval.
1503		2781
1504	=item const char *path [read-only]	2782	=item const char *path [read-only]
1505		2783
1506	The filesystem path that is being watched.	2784	The file system path that is being watched.
1507		2785
1508	=back	2786	=back
1509		2787
		2788	=head3 Examples
		2789
1510	Example: Watch C</etc/passwd> for attribute changes.	2790	Example: Watch C</etc/passwd> for attribute changes.
1511		2791
1512	static void	2792	static void
1513	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2793	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1514	{	2794	{
1515	/* /etc/passwd changed in some way */	2795	/* /etc/passwd changed in some way */
1516	if (w->attr.st_nlink)	2796	if (w->attr.st_nlink)
1517	{	2797	{
1518	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2798	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1519	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2799	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1520	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2800	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1521	}	2801	}
1522	else	2802	else
1523	/* you shalt not abuse printf for puts */	2803	/* you shalt not abuse printf for puts */
1524	puts ("wow, /etc/passwd is not there, expect problems. "	2804	puts ("wow, /etc/passwd is not there, expect problems. "
1525	"if this is windows, they already arrived\n");	2805	"if this is windows, they already arrived\n");
1526	}	2806	}
1527		2807
1528	...	2808	...
1529	ev_stat passwd;	2809	ev_stat passwd;
1530		2810
1531	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	2811	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1532	ev_stat_start (loop, &passwd);	2812	ev_stat_start (loop, &passwd);
		2813
		2814	Example: Like above, but additionally use a one-second delay so we do not
		2815	miss updates (however, frequent updates will delay processing, too, so
		2816	one might do the work both on C<ev_stat> callback invocation I<and> on
		2817	C<ev_timer> callback invocation).
		2818
		2819	static ev_stat passwd;
		2820	static ev_timer timer;
		2821
		2822	static void
		2823	timer_cb (EV_P_ ev_timer *w, int revents)
		2824	{
		2825	ev_timer_stop (EV_A_ w);
		2826
		2827	/* now it's one second after the most recent passwd change */
		2828	}
		2829
		2830	static void
		2831	stat_cb (EV_P_ ev_stat *w, int revents)
		2832	{
		2833	/* reset the one-second timer */
		2834	ev_timer_again (EV_A_ &timer);
		2835	}
		2836
		2837	...
		2838	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2839	ev_stat_start (loop, &passwd);
		2840	ev_timer_init (&timer, timer_cb, 0., 1.02);
1533		2841
1534		2842
1535	=head2 C<ev_idle> - when you've got nothing better to do...	2843	=head2 C<ev_idle> - when you've got nothing better to do...
1536		2844
1537	Idle watchers trigger events when no other events of the same or higher	2845	Idle watchers trigger events when no other events of the same or higher
1538	priority are pending (prepare, check and other idle watchers do not	2846	priority are pending (prepare, check and other idle watchers do not count
1539	count).	2847	as receiving "events").
1540		2848
1541	That is, as long as your process is busy handling sockets or timeouts	2849	That is, as long as your process is busy handling sockets or timeouts
1542	(or even signals, imagine) of the same or higher priority it will not be	2850	(or even signals, imagine) of the same or higher priority it will not be
1543	triggered. But when your process is idle (or only lower-priority watchers	2851	triggered. But when your process is idle (or only lower-priority watchers
1544	are pending), the idle watchers are being called once per event loop	2852	are pending), the idle watchers are being called once per event loop
…		…
1551	Apart from keeping your process non-blocking (which is a useful	2859	Apart from keeping your process non-blocking (which is a useful
1552	effect on its own sometimes), idle watchers are a good place to do	2860	effect on its own sometimes), idle watchers are a good place to do
1553	"pseudo-background processing", or delay processing stuff to after the	2861	"pseudo-background processing", or delay processing stuff to after the
1554	event loop has handled all outstanding events.	2862	event loop has handled all outstanding events.
1555		2863
		2864	=head3 Abusing an C<ev_idle> watcher for its side-effect
		2865
		2866	As long as there is at least one active idle watcher, libev will never
		2867	sleep unnecessarily. Or in other words, it will loop as fast as possible.
		2868	For this to work, the idle watcher doesn't need to be invoked at all - the
		2869	lowest priority will do.
		2870
		2871	This mode of operation can be useful together with an C<ev_check> watcher,
		2872	to do something on each event loop iteration - for example to balance load
		2873	between different connections.
		2874
		2875	See L</Abusing an ev_check watcher for its side-effect> for a longer
		2876	example.
		2877
1556	=head3 Watcher-Specific Functions and Data Members	2878	=head3 Watcher-Specific Functions and Data Members
1557		2879
1558	=over 4	2880	=over 4
1559		2881
1560	=item ev_idle_init (ev_signal *, callback)	2882	=item ev_idle_init (ev_idle *, callback)
1561		2883
1562	Initialises and configures the idle watcher - it has no parameters of any	2884	Initialises and configures the idle watcher - it has no parameters of any
1563	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2885	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1564	believe me.	2886	believe me.
1565		2887
1566	=back	2888	=back
1567		2889
		2890	=head3 Examples
		2891
1568	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	2892	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1569	callback, free it. Also, use no error checking, as usual.	2893	callback, free it. Also, use no error checking, as usual.
1570		2894
1571	static void	2895	static void
1572	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2896	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1573	{	2897	{
		2898	// stop the watcher
		2899	ev_idle_stop (loop, w);
		2900
		2901	// now we can free it
1574	free (w);	2902	free (w);
		2903
1575	// now do something you wanted to do when the program has	2904	// now do something you wanted to do when the program has
1576	// no longer asnything immediate to do.	2905	// no longer anything immediate to do.
1577	}	2906	}
1578		2907
1579	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2908	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1580	ev_idle_init (idle_watcher, idle_cb);	2909	ev_idle_init (idle_watcher, idle_cb);
1581	ev_idle_start (loop, idle_cb);	2910	ev_idle_start (loop, idle_watcher);
1582		2911
1583		2912
1584	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	2913	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1585		2914
1586	Prepare and check watchers are usually (but not always) used in tandem:	2915	Prepare and check watchers are often (but not always) used in pairs:
1587	prepare watchers get invoked before the process blocks and check watchers	2916	prepare watchers get invoked before the process blocks and check watchers
1588	afterwards.	2917	afterwards.
1589		2918
1590	You I<must not> call C<ev_loop> or similar functions that enter	2919	You I<must not> call C<ev_run> (or similar functions that enter the
1591	the current event loop from either C<ev_prepare> or C<ev_check>	2920	current event loop) or C<ev_loop_fork> from either C<ev_prepare> or
1592	watchers. Other loops than the current one are fine, however. The	2921	C<ev_check> watchers. Other loops than the current one are fine,
1593	rationale behind this is that you do not need to check for recursion in	2922	however. The rationale behind this is that you do not need to check
1594	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	2923	for recursion in those watchers, i.e. the sequence will always be
1595	C<ev_check> so if you have one watcher of each kind they will always be	2924	C<ev_prepare>, blocking, C<ev_check> so if you have one watcher of each
1596	called in pairs bracketing the blocking call.	2925	kind they will always be called in pairs bracketing the blocking call.
1597		2926
1598	Their main purpose is to integrate other event mechanisms into libev and	2927	Their main purpose is to integrate other event mechanisms into libev and
1599	their use is somewhat advanced. This could be used, for example, to track	2928	their use is somewhat advanced. They could be used, for example, to track
1600	variable changes, implement your own watchers, integrate net-snmp or a	2929	variable changes, implement your own watchers, integrate net-snmp or a
1601	coroutine library and lots more. They are also occasionally useful if	2930	coroutine library and lots more. They are also occasionally useful if
1602	you cache some data and want to flush it before blocking (for example,	2931	you cache some data and want to flush it before blocking (for example,
1603	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	2932	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1604	watcher).	2933	watcher).
1605		2934
1606	This is done by examining in each prepare call which file descriptors need	2935	This is done by examining in each prepare call which file descriptors
1607	to be watched by the other library, registering C<ev_io> watchers for	2936	need to be watched by the other library, registering C<ev_io> watchers
1608	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2937	for them and starting an C<ev_timer> watcher for any timeouts (many
1609	provide just this functionality). Then, in the check watcher you check for	2938	libraries provide exactly this functionality). Then, in the check watcher,
1610	any events that occured (by checking the pending status of all watchers	2939	you check for any events that occurred (by checking the pending status
1611	and stopping them) and call back into the library. The I/O and timer	2940	of all watchers and stopping them) and call back into the library. The
1612	callbacks will never actually be called (but must be valid nevertheless,	2941	I/O and timer callbacks will never actually be called (but must be valid
1613	because you never know, you know?).	2942	nevertheless, because you never know, you know?).
1614		2943
1615	As another example, the Perl Coro module uses these hooks to integrate	2944	As another example, the Perl Coro module uses these hooks to integrate
1616	coroutines into libev programs, by yielding to other active coroutines	2945	coroutines into libev programs, by yielding to other active coroutines
1617	during each prepare and only letting the process block if no coroutines	2946	during each prepare and only letting the process block if no coroutines
1618	are ready to run (it's actually more complicated: it only runs coroutines	2947	are ready to run (it's actually more complicated: it only runs coroutines
1619	with priority higher than or equal to the event loop and one coroutine	2948	with priority higher than or equal to the event loop and one coroutine
1620	of lower priority, but only once, using idle watchers to keep the event	2949	of lower priority, but only once, using idle watchers to keep the event
1621	loop from blocking if lower-priority coroutines are active, thus mapping	2950	loop from blocking if lower-priority coroutines are active, thus mapping
1622	low-priority coroutines to idle/background tasks).	2951	low-priority coroutines to idle/background tasks).
1623		2952
1624	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	2953	When used for this purpose, it is recommended to give C<ev_check> watchers
1625	priority, to ensure that they are being run before any other watchers	2954	highest (C<EV_MAXPRI>) priority, to ensure that they are being run before
		2955	any other watchers after the poll (this doesn't matter for C<ev_prepare>
		2956	watchers).
		2957
1626	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	2958	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
1627	too) should not activate ("feed") events into libev. While libev fully	2959	activate ("feed") events into libev. While libev fully supports this, they
1628	supports this, they will be called before other C<ev_check> watchers did	2960	might get executed before other C<ev_check> watchers did their job. As
1629	their job. As C<ev_check> watchers are often used to embed other event	2961	C<ev_check> watchers are often used to embed other (non-libev) event
1630	loops those other event loops might be in an unusable state until their	2962	loops those other event loops might be in an unusable state until their
1631	C<ev_check> watcher ran (always remind yourself to coexist peacefully with	2963	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
1632	others).	2964	others).
1633		2965
		2966	=head3 Abusing an C<ev_check> watcher for its side-effect
		2967
		2968	C<ev_check> (and less often also C<ev_prepare>) watchers can also be
		2969	useful because they are called once per event loop iteration. For
		2970	example, if you want to handle a large number of connections fairly, you
		2971	normally only do a bit of work for each active connection, and if there
		2972	is more work to do, you wait for the next event loop iteration, so other
		2973	connections have a chance of making progress.
		2974
		2975	Using an C<ev_check> watcher is almost enough: it will be called on the
		2976	next event loop iteration. However, that isn't as soon as possible -
		2977	without external events, your C<ev_check> watcher will not be invoked.
		2978
		2979	This is where C<ev_idle> watchers come in handy - all you need is a
		2980	single global idle watcher that is active as long as you have one active
		2981	C<ev_check> watcher. The C<ev_idle> watcher makes sure the event loop
		2982	will not sleep, and the C<ev_check> watcher makes sure a callback gets
		2983	invoked. Neither watcher alone can do that.
		2984
1634	=head3 Watcher-Specific Functions and Data Members	2985	=head3 Watcher-Specific Functions and Data Members
1635		2986
1636	=over 4	2987	=over 4
1637		2988
1638	=item ev_prepare_init (ev_prepare *, callback)	2989	=item ev_prepare_init (ev_prepare *, callback)
1639		2990
1640	=item ev_check_init (ev_check *, callback)	2991	=item ev_check_init (ev_check *, callback)
1641		2992
1642	Initialises and configures the prepare or check watcher - they have no	2993	Initialises and configures the prepare or check watcher - they have no
1643	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2994	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1644	macros, but using them is utterly, utterly and completely pointless.	2995	macros, but using them is utterly, utterly, utterly and completely
		2996	pointless.
1645		2997
1646	=back	2998	=back
		2999
		3000	=head3 Examples
1647		3001
1648	There are a number of principal ways to embed other event loops or modules	3002	There are a number of principal ways to embed other event loops or modules
1649	into libev. Here are some ideas on how to include libadns into libev	3003	into libev. Here are some ideas on how to include libadns into libev
1650	(there is a Perl module named C<EV::ADNS> that does this, which you could	3004	(there is a Perl module named C<EV::ADNS> that does this, which you could
1651	use for an actually working example. Another Perl module named C<EV::Glib>	3005	use as a working example. Another Perl module named C<EV::Glib> embeds a
1652	embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV	3006	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
1653	into the Glib event loop).	3007	Glib event loop).
1654		3008
1655	Method 1: Add IO watchers and a timeout watcher in a prepare handler,	3009	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1656	and in a check watcher, destroy them and call into libadns. What follows	3010	and in a check watcher, destroy them and call into libadns. What follows
1657	is pseudo-code only of course. This requires you to either use a low	3011	is pseudo-code only of course. This requires you to either use a low
1658	priority for the check watcher or use C<ev_clear_pending> explicitly, as	3012	priority for the check watcher or use C<ev_clear_pending> explicitly, as
1659	the callbacks for the IO/timeout watchers might not have been called yet.	3013	the callbacks for the IO/timeout watchers might not have been called yet.
1660		3014
1661	static ev_io iow [nfd];	3015	static ev_io iow [nfd];
1662	static ev_timer tw;	3016	static ev_timer tw;
1663		3017
1664	static void	3018	static void
1665	io_cb (ev_loop *loop, ev_io *w, int revents)	3019	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1666	{	3020	{
1667	}	3021	}
1668		3022
1669	// create io watchers for each fd and a timer before blocking	3023	// create io watchers for each fd and a timer before blocking
1670	static void	3024	static void
1671	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	3025	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1672	{	3026	{
1673	int timeout = 3600000;	3027	int timeout = 3600000;
1674	struct pollfd fds [nfd];	3028	struct pollfd fds [nfd];
1675	// actual code will need to loop here and realloc etc.	3029	// actual code will need to loop here and realloc etc.
1676	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	3030	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1677		3031
1678	/* the callback is illegal, but won't be called as we stop during check */	3032	/* the callback is illegal, but won't be called as we stop during check */
1679	ev_timer_init (&tw, 0, timeout * 1e-3);	3033	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
1680	ev_timer_start (loop, &tw);	3034	ev_timer_start (loop, &tw);
1681		3035
1682	// create one ev_io per pollfd	3036	// create one ev_io per pollfd
1683	for (int i = 0; i < nfd; ++i)	3037	for (int i = 0; i < nfd; ++i)
1684	{	3038	{
1685	ev_io_init (iow + i, io_cb, fds [i].fd,	3039	ev_io_init (iow + i, io_cb, fds [i].fd,
1686	((fds [i].events & POLLIN ? EV_READ : 0)	3040	((fds [i].events & POLLIN ? EV_READ : 0)
1687	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	3041	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1688		3042
1689	fds [i].revents = 0;	3043	fds [i].revents = 0;
1690	ev_io_start (loop, iow + i);	3044	ev_io_start (loop, iow + i);
1691	}	3045	}
1692	}	3046	}
1693		3047
1694	// stop all watchers after blocking	3048	// stop all watchers after blocking
1695	static void	3049	static void
1696	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	3050	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1697	{	3051	{
1698	ev_timer_stop (loop, &tw);	3052	ev_timer_stop (loop, &tw);
1699		3053
1700	for (int i = 0; i < nfd; ++i)	3054	for (int i = 0; i < nfd; ++i)
1701	{	3055	{
1702	// set the relevant poll flags	3056	// set the relevant poll flags
1703	// could also call adns_processreadable etc. here	3057	// could also call adns_processreadable etc. here
1704	struct pollfd *fd = fds + i;	3058	struct pollfd *fd = fds + i;
1705	int revents = ev_clear_pending (iow + i);	3059	int revents = ev_clear_pending (iow + i);
1706	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;	3060	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1707	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;	3061	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1708		3062
1709	// now stop the watcher	3063	// now stop the watcher
1710	ev_io_stop (loop, iow + i);	3064	ev_io_stop (loop, iow + i);
1711	}	3065	}
1712		3066
1713	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	3067	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1714	}	3068	}
1715		3069
1716	Method 2: This would be just like method 1, but you run C<adns_afterpoll>	3070	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
1717	in the prepare watcher and would dispose of the check watcher.	3071	in the prepare watcher and would dispose of the check watcher.
1718		3072
1719	Method 3: If the module to be embedded supports explicit event	3073	Method 3: If the module to be embedded supports explicit event
1720	notification (adns does), you can also make use of the actual watcher	3074	notification (libadns does), you can also make use of the actual watcher
1721	callbacks, and only destroy/create the watchers in the prepare watcher.	3075	callbacks, and only destroy/create the watchers in the prepare watcher.
1722		3076
1723	static void	3077	static void
1724	timer_cb (EV_P_ ev_timer *w, int revents)	3078	timer_cb (EV_P_ ev_timer *w, int revents)
1725	{	3079	{
1726	adns_state ads = (adns_state)w->data;	3080	adns_state ads = (adns_state)w->data;
1727	update_now (EV_A);	3081	update_now (EV_A);
1728		3082
1729	adns_processtimeouts (ads, &tv_now);	3083	adns_processtimeouts (ads, &tv_now);
1730	}	3084	}
1731		3085
1732	static void	3086	static void
1733	io_cb (EV_P_ ev_io *w, int revents)	3087	io_cb (EV_P_ ev_io *w, int revents)
1734	{	3088	{
1735	adns_state ads = (adns_state)w->data;	3089	adns_state ads = (adns_state)w->data;
1736	update_now (EV_A);	3090	update_now (EV_A);
1737		3091
1738	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);	3092	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1739	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);	3093	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1740	}	3094	}
1741		3095
1742	// do not ever call adns_afterpoll	3096	// do not ever call adns_afterpoll
1743		3097
1744	Method 4: Do not use a prepare or check watcher because the module you	3098	Method 4: Do not use a prepare or check watcher because the module you
1745	want to embed is too inflexible to support it. Instead, youc na override	3099	want to embed is not flexible enough to support it. Instead, you can
1746	their poll function. The drawback with this solution is that the main	3100	override their poll function. The drawback with this solution is that the
1747	loop is now no longer controllable by EV. The C<Glib::EV> module does	3101	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
1748	this.	3102	this approach, effectively embedding EV as a client into the horrible
		3103	libglib event loop.
1749		3104
1750	static gint	3105	static gint
1751	event_poll_func (GPollFD *fds, guint nfds, gint timeout)	3106	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1752	{	3107	{
1753	int got_events = 0;	3108	int got_events = 0;
1754		3109
1755	for (n = 0; n < nfds; ++n)	3110	for (n = 0; n < nfds; ++n)
1756	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events	3111	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1757		3112
1758	if (timeout >= 0)	3113	if (timeout >= 0)
1759	// create/start timer	3114	// create/start timer
1760		3115
1761	// poll	3116	// poll
1762	ev_loop (EV_A_ 0);	3117	ev_run (EV_A_ 0);
1763		3118
1764	// stop timer again	3119	// stop timer again
1765	if (timeout >= 0)	3120	if (timeout >= 0)
1766	ev_timer_stop (EV_A_ &to);	3121	ev_timer_stop (EV_A_ &to);
1767		3122
1768	// stop io watchers again - their callbacks should have set	3123	// stop io watchers again - their callbacks should have set
1769	for (n = 0; n < nfds; ++n)	3124	for (n = 0; n < nfds; ++n)
1770	ev_io_stop (EV_A_ iow [n]);	3125	ev_io_stop (EV_A_ iow [n]);
1771		3126
1772	return got_events;	3127	return got_events;
1773	}	3128	}
1774		3129
1775		3130
1776	=head2 C<ev_embed> - when one backend isn't enough...	3131	=head2 C<ev_embed> - when one backend isn't enough...
1777		3132
1778	This is a rather advanced watcher type that lets you embed one event loop	3133	This is a rather advanced watcher type that lets you embed one event loop
1779	into another (currently only C<ev_io> events are supported in the embedded	3134	into another (currently only C<ev_io> events are supported in the embedded
1780	loop, other types of watchers might be handled in a delayed or incorrect	3135	loop, other types of watchers might be handled in a delayed or incorrect
1781	fashion and must not be used). (See portability notes, below).	3136	fashion and must not be used).
1782		3137
1783	There are primarily two reasons you would want that: work around bugs and	3138	There are primarily two reasons you would want that: work around bugs and
1784	prioritise I/O.	3139	prioritise I/O.
1785		3140
1786	As an example for a bug workaround, the kqueue backend might only support	3141	As an example for a bug workaround, the kqueue backend might only support
1787	sockets on some platform, so it is unusable as generic backend, but you	3142	sockets on some platform, so it is unusable as generic backend, but you
1788	still want to make use of it because you have many sockets and it scales	3143	still want to make use of it because you have many sockets and it scales
1789	so nicely. In this case, you would create a kqueue-based loop and embed it	3144	so nicely. In this case, you would create a kqueue-based loop and embed
1790	into your default loop (which might use e.g. poll). Overall operation will	3145	it into your default loop (which might use e.g. poll). Overall operation
1791	be a bit slower because first libev has to poll and then call kevent, but	3146	will be a bit slower because first libev has to call C<poll> and then
1792	at least you can use both at what they are best.	3147	C<kevent>, but at least you can use both mechanisms for what they are
		3148	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
1793		3149
1794	As for prioritising I/O: rarely you have the case where some fds have	3150	As for prioritising I/O: under rare circumstances you have the case where
1795	to be watched and handled very quickly (with low latency), and even	3151	some fds have to be watched and handled very quickly (with low latency),
1796	priorities and idle watchers might have too much overhead. In this case	3152	and even priorities and idle watchers might have too much overhead. In
1797	you would put all the high priority stuff in one loop and all the rest in	3153	this case you would put all the high priority stuff in one loop and all
1798	a second one, and embed the second one in the first.	3154	the rest in a second one, and embed the second one in the first.
1799		3155
1800	As long as the watcher is active, the callback will be invoked every time	3156	As long as the watcher is active, the callback will be invoked every
1801	there might be events pending in the embedded loop. The callback must then	3157	time there might be events pending in the embedded loop. The callback
1802	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	3158	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
1803	their callbacks (you could also start an idle watcher to give the embedded	3159	sweep and invoke their callbacks (the callback doesn't need to invoke the
1804	loop strictly lower priority for example). You can also set the callback	3160	C<ev_embed_sweep> function directly, it could also start an idle watcher
1805	to C<0>, in which case the embed watcher will automatically execute the	3161	to give the embedded loop strictly lower priority for example).
1806	embedded loop sweep.
1807		3162
1808	As long as the watcher is started it will automatically handle events. The	3163	You can also set the callback to C<0>, in which case the embed watcher
1809	callback will be invoked whenever some events have been handled. You can	3164	will automatically execute the embedded loop sweep whenever necessary.
1810	set the callback to C<0> to avoid having to specify one if you are not
1811	interested in that.
1812		3165
1813	Also, there have not currently been made special provisions for forking:	3166	Fork detection will be handled transparently while the C<ev_embed> watcher
1814	when you fork, you not only have to call C<ev_loop_fork> on both loops,	3167	is active, i.e., the embedded loop will automatically be forked when the
1815	but you will also have to stop and restart any C<ev_embed> watchers	3168	embedding loop forks. In other cases, the user is responsible for calling
1816	yourself.	3169	C<ev_loop_fork> on the embedded loop.
1817		3170
1818	Unfortunately, not all backends are embeddable, only the ones returned by	3171	Unfortunately, not all backends are embeddable: only the ones returned by
1819	C<ev_embeddable_backends> are, which, unfortunately, does not include any	3172	C<ev_embeddable_backends> are, which, unfortunately, does not include any
1820	portable one.	3173	portable one.
1821		3174
1822	So when you want to use this feature you will always have to be prepared	3175	So when you want to use this feature you will always have to be prepared
1823	that you cannot get an embeddable loop. The recommended way to get around	3176	that you cannot get an embeddable loop. The recommended way to get around
1824	this is to have a separate variables for your embeddable loop, try to	3177	this is to have a separate variables for your embeddable loop, try to
1825	create it, and if that fails, use the normal loop for everything:	3178	create it, and if that fails, use the normal loop for everything.
1826		3179
1827	struct ev_loop *loop_hi = ev_default_init (0);	3180	=head3 C<ev_embed> and fork
1828	struct ev_loop *loop_lo = 0;
1829	struct ev_embed embed;
1830
1831	// see if there is a chance of getting one that works
1832	// (remember that a flags value of 0 means autodetection)
1833	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1834	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1835	: 0;
1836		3181
1837	// if we got one, then embed it, otherwise default to loop_hi	3182	While the C<ev_embed> watcher is running, forks in the embedding loop will
1838	if (loop_lo)	3183	automatically be applied to the embedded loop as well, so no special
1839	{	3184	fork handling is required in that case. When the watcher is not running,
1840	ev_embed_init (&embed, 0, loop_lo);	3185	however, it is still the task of the libev user to call C<ev_loop_fork ()>
1841	ev_embed_start (loop_hi, &embed);	3186	as applicable.
1842	}
1843	else
1844	loop_lo = loop_hi;
1845
1846	=head2 Portability notes
1847
1848	Kqueue is nominally embeddable, but this is broken on all BSDs that I
1849	tried, in various ways. Usually the embedded event loop will simply never
1850	receive events, sometimes it will only trigger a few times, sometimes in a
1851	loop. Epoll is also nominally embeddable, but many Linux kernel versions
1852	will always eport the epoll fd as ready, even when no events are pending.
1853
1854	While libev allows embedding these backends (they are contained in
1855	C<ev_embeddable_backends ()>), take extreme care that it will actually
1856	work.
1857
1858	When in doubt, create a dynamic event loop forced to use sockets (this
1859	usually works) and possibly another thread and a pipe or so to report to
1860	your main event loop.
1861		3187
1862	=head3 Watcher-Specific Functions and Data Members	3188	=head3 Watcher-Specific Functions and Data Members
1863		3189
1864	=over 4	3190	=over 4
1865		3191
1866	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	3192	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1867		3193
1868	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)	3194	=item ev_embed_set (ev_embed *, struct ev_loop *embedded_loop)
1869		3195
1870	Configures the watcher to embed the given loop, which must be	3196	Configures the watcher to embed the given loop, which must be
1871	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	3197	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1872	invoked automatically, otherwise it is the responsibility of the callback	3198	invoked automatically, otherwise it is the responsibility of the callback
1873	to invoke it (it will continue to be called until the sweep has been done,	3199	to invoke it (it will continue to be called until the sweep has been done,
1874	if you do not want thta, you need to temporarily stop the embed watcher).	3200	if you do not want that, you need to temporarily stop the embed watcher).
1875		3201
1876	=item ev_embed_sweep (loop, ev_embed *)	3202	=item ev_embed_sweep (loop, ev_embed *)
1877		3203
1878	Make a single, non-blocking sweep over the embedded loop. This works	3204	Make a single, non-blocking sweep over the embedded loop. This works
1879	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	3205	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
1880	apropriate way for embedded loops.	3206	appropriate way for embedded loops.
1881		3207
1882	=item struct ev_loop *other [read-only]	3208	=item struct ev_loop *other [read-only]
1883		3209
1884	The embedded event loop.	3210	The embedded event loop.
1885		3211
1886	=back	3212	=back
		3213
		3214	=head3 Examples
		3215
		3216	Example: Try to get an embeddable event loop and embed it into the default
		3217	event loop. If that is not possible, use the default loop. The default
		3218	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		3219	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		3220	used).
		3221
		3222	struct ev_loop *loop_hi = ev_default_init (0);
		3223	struct ev_loop *loop_lo = 0;
		3224	ev_embed embed;
		3225
		3226	// see if there is a chance of getting one that works
		3227	// (remember that a flags value of 0 means autodetection)
		3228	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3229	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3230	: 0;
		3231
		3232	// if we got one, then embed it, otherwise default to loop_hi
		3233	if (loop_lo)
		3234	{
		3235	ev_embed_init (&embed, 0, loop_lo);
		3236	ev_embed_start (loop_hi, &embed);
		3237	}
		3238	else
		3239	loop_lo = loop_hi;
		3240
		3241	Example: Check if kqueue is available but not recommended and create
		3242	a kqueue backend for use with sockets (which usually work with any
		3243	kqueue implementation). Store the kqueue/socket-only event loop in
		3244	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3245
		3246	struct ev_loop *loop = ev_default_init (0);
		3247	struct ev_loop *loop_socket = 0;
		3248	ev_embed embed;
		3249
		3250	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3251	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3252	{
		3253	ev_embed_init (&embed, 0, loop_socket);
		3254	ev_embed_start (loop, &embed);
		3255	}
		3256
		3257	if (!loop_socket)
		3258	loop_socket = loop;
		3259
		3260	// now use loop_socket for all sockets, and loop for everything else
1887		3261
1888		3262
1889	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	3263	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1890		3264
1891	Fork watchers are called when a C<fork ()> was detected (usually because	3265	Fork watchers are called when a C<fork ()> was detected (usually because
1892	whoever is a good citizen cared to tell libev about it by calling	3266	whoever is a good citizen cared to tell libev about it by calling
1893	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the	3267	C<ev_loop_fork>). The invocation is done before the event loop blocks next
1894	event loop blocks next and before C<ev_check> watchers are being called,	3268	and before C<ev_check> watchers are being called, and only in the child
1895	and only in the child after the fork. If whoever good citizen calling	3269	after the fork. If whoever good citizen calling C<ev_default_fork> cheats
1896	C<ev_default_fork> cheats and calls it in the wrong process, the fork	3270	and calls it in the wrong process, the fork handlers will be invoked, too,
1897	handlers will be invoked, too, of course.	3271	of course.
		3272
		3273	=head3 The special problem of life after fork - how is it possible?
		3274
		3275	Most uses of C<fork ()> consist of forking, then some simple calls to set
		3276	up/change the process environment, followed by a call to C<exec()>. This
		3277	sequence should be handled by libev without any problems.
		3278
		3279	This changes when the application actually wants to do event handling
		3280	in the child, or both parent in child, in effect "continuing" after the
		3281	fork.
		3282
		3283	The default mode of operation (for libev, with application help to detect
		3284	forks) is to duplicate all the state in the child, as would be expected
		3285	when I<either> the parent I<or> the child process continues.
		3286
		3287	When both processes want to continue using libev, then this is usually the
		3288	wrong result. In that case, usually one process (typically the parent) is
		3289	supposed to continue with all watchers in place as before, while the other
		3290	process typically wants to start fresh, i.e. without any active watchers.
		3291
		3292	The cleanest and most efficient way to achieve that with libev is to
		3293	simply create a new event loop, which of course will be "empty", and
		3294	use that for new watchers. This has the advantage of not touching more
		3295	memory than necessary, and thus avoiding the copy-on-write, and the
		3296	disadvantage of having to use multiple event loops (which do not support
		3297	signal watchers).
		3298
		3299	When this is not possible, or you want to use the default loop for
		3300	other reasons, then in the process that wants to start "fresh", call
		3301	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3302	Destroying the default loop will "orphan" (not stop) all registered
		3303	watchers, so you have to be careful not to execute code that modifies
		3304	those watchers. Note also that in that case, you have to re-register any
		3305	signal watchers.
1898		3306
1899	=head3 Watcher-Specific Functions and Data Members	3307	=head3 Watcher-Specific Functions and Data Members
1900		3308
1901	=over 4	3309	=over 4
1902		3310
1903	=item ev_fork_init (ev_signal *, callback)	3311	=item ev_fork_init (ev_fork *, callback)
1904		3312
1905	Initialises and configures the fork watcher - it has no parameters of any	3313	Initialises and configures the fork watcher - it has no parameters of any
1906	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,	3314	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
1907	believe me.	3315	really.
1908		3316
1909	=back	3317	=back
1910		3318
1911		3319
		3320	=head2 C<ev_cleanup> - even the best things end
		3321
		3322	Cleanup watchers are called just before the event loop is being destroyed
		3323	by a call to C<ev_loop_destroy>.
		3324
		3325	While there is no guarantee that the event loop gets destroyed, cleanup
		3326	watchers provide a convenient method to install cleanup hooks for your
		3327	program, worker threads and so on - you just to make sure to destroy the
		3328	loop when you want them to be invoked.
		3329
		3330	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3331	all other watchers, they do not keep a reference to the event loop (which
		3332	makes a lot of sense if you think about it). Like all other watchers, you
		3333	can call libev functions in the callback, except C<ev_cleanup_start>.
		3334
		3335	=head3 Watcher-Specific Functions and Data Members
		3336
		3337	=over 4
		3338
		3339	=item ev_cleanup_init (ev_cleanup *, callback)
		3340
		3341	Initialises and configures the cleanup watcher - it has no parameters of
		3342	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3343	pointless, I assure you.
		3344
		3345	=back
		3346
		3347	Example: Register an atexit handler to destroy the default loop, so any
		3348	cleanup functions are called.
		3349
		3350	static void
		3351	program_exits (void)
		3352	{
		3353	ev_loop_destroy (EV_DEFAULT_UC);
		3354	}
		3355
		3356	...
		3357	atexit (program_exits);
		3358
		3359
		3360	=head2 C<ev_async> - how to wake up an event loop
		3361
		3362	In general, you cannot use an C<ev_loop> from multiple threads or other
		3363	asynchronous sources such as signal handlers (as opposed to multiple event
		3364	loops - those are of course safe to use in different threads).
		3365
		3366	Sometimes, however, you need to wake up an event loop you do not control,
		3367	for example because it belongs to another thread. This is what C<ev_async>
		3368	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3369	it by calling C<ev_async_send>, which is thread- and signal safe.
		3370
		3371	This functionality is very similar to C<ev_signal> watchers, as signals,
		3372	too, are asynchronous in nature, and signals, too, will be compressed
		3373	(i.e. the number of callback invocations may be less than the number of
		3374	C<ev_async_send> calls). In fact, you could use signal watchers as a kind
		3375	of "global async watchers" by using a watcher on an otherwise unused
		3376	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3377	even without knowing which loop owns the signal.
		3378
		3379	=head3 Queueing
		3380
		3381	C<ev_async> does not support queueing of data in any way. The reason
		3382	is that the author does not know of a simple (or any) algorithm for a
		3383	multiple-writer-single-reader queue that works in all cases and doesn't
		3384	need elaborate support such as pthreads or unportable memory access
		3385	semantics.
		3386
		3387	That means that if you want to queue data, you have to provide your own
		3388	queue. But at least I can tell you how to implement locking around your
		3389	queue:
		3390
		3391	=over 4
		3392
		3393	=item queueing from a signal handler context
		3394
		3395	To implement race-free queueing, you simply add to the queue in the signal
		3396	handler but you block the signal handler in the watcher callback. Here is
		3397	an example that does that for some fictitious SIGUSR1 handler:
		3398
		3399	static ev_async mysig;
		3400
		3401	static void
		3402	sigusr1_handler (void)
		3403	{
		3404	sometype data;
		3405
		3406	// no locking etc.
		3407	queue_put (data);
		3408	ev_async_send (EV_DEFAULT_ &mysig);
		3409	}
		3410
		3411	static void
		3412	mysig_cb (EV_P_ ev_async *w, int revents)
		3413	{
		3414	sometype data;
		3415	sigset_t block, prev;
		3416
		3417	sigemptyset (&block);
		3418	sigaddset (&block, SIGUSR1);
		3419	sigprocmask (SIG_BLOCK, &block, &prev);
		3420
		3421	while (queue_get (&data))
		3422	process (data);
		3423
		3424	if (sigismember (&prev, SIGUSR1)
		3425	sigprocmask (SIG_UNBLOCK, &block, 0);
		3426	}
		3427
		3428	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3429	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3430	either...).
		3431
		3432	=item queueing from a thread context
		3433
		3434	The strategy for threads is different, as you cannot (easily) block
		3435	threads but you can easily preempt them, so to queue safely you need to
		3436	employ a traditional mutex lock, such as in this pthread example:
		3437
		3438	static ev_async mysig;
		3439	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3440
		3441	static void
		3442	otherthread (void)
		3443	{
		3444	// only need to lock the actual queueing operation
		3445	pthread_mutex_lock (&mymutex);
		3446	queue_put (data);
		3447	pthread_mutex_unlock (&mymutex);
		3448
		3449	ev_async_send (EV_DEFAULT_ &mysig);
		3450	}
		3451
		3452	static void
		3453	mysig_cb (EV_P_ ev_async *w, int revents)
		3454	{
		3455	pthread_mutex_lock (&mymutex);
		3456
		3457	while (queue_get (&data))
		3458	process (data);
		3459
		3460	pthread_mutex_unlock (&mymutex);
		3461	}
		3462
		3463	=back
		3464
		3465
		3466	=head3 Watcher-Specific Functions and Data Members
		3467
		3468	=over 4
		3469
		3470	=item ev_async_init (ev_async *, callback)
		3471
		3472	Initialises and configures the async watcher - it has no parameters of any
		3473	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3474	trust me.
		3475
		3476	=item ev_async_send (loop, ev_async *)
		3477
		3478	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3479	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3480	returns.
		3481
		3482	Unlike C<ev_feed_event>, this call is safe to do from other threads,
		3483	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
		3484	embedding section below on what exactly this means).
		3485
		3486	Note that, as with other watchers in libev, multiple events might get
		3487	compressed into a single callback invocation (another way to look at
		3488	this is that C<ev_async> watchers are level-triggered: they are set on
		3489	C<ev_async_send>, reset when the event loop detects that).
		3490
		3491	This call incurs the overhead of at most one extra system call per event
		3492	loop iteration, if the event loop is blocked, and no syscall at all if
		3493	the event loop (or your program) is processing events. That means that
		3494	repeated calls are basically free (there is no need to avoid calls for
		3495	performance reasons) and that the overhead becomes smaller (typically
		3496	zero) under load.
		3497
		3498	=item bool = ev_async_pending (ev_async *)
		3499
		3500	Returns a non-zero value when C<ev_async_send> has been called on the
		3501	watcher but the event has not yet been processed (or even noted) by the
		3502	event loop.
		3503
		3504	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3505	the loop iterates next and checks for the watcher to have become active,
		3506	it will reset the flag again. C<ev_async_pending> can be used to very
		3507	quickly check whether invoking the loop might be a good idea.
		3508
		3509	Not that this does I<not> check whether the watcher itself is pending,
		3510	only whether it has been requested to make this watcher pending: there
		3511	is a time window between the event loop checking and resetting the async
		3512	notification, and the callback being invoked.
		3513
		3514	=back
		3515
		3516
1912	=head1 OTHER FUNCTIONS	3517	=head1 OTHER FUNCTIONS
1913		3518
1914	There are some other functions of possible interest. Described. Here. Now.	3519	There are some other functions of possible interest. Described. Here. Now.
1915		3520
1916	=over 4	3521	=over 4
1917		3522
1918	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3523	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback, arg)
1919		3524
1920	This function combines a simple timer and an I/O watcher, calls your	3525	This function combines a simple timer and an I/O watcher, calls your
1921	callback on whichever event happens first and automatically stop both	3526	callback on whichever event happens first and automatically stops both
1922	watchers. This is useful if you want to wait for a single event on an fd	3527	watchers. This is useful if you want to wait for a single event on an fd
1923	or timeout without having to allocate/configure/start/stop/free one or	3528	or timeout without having to allocate/configure/start/stop/free one or
1924	more watchers yourself.	3529	more watchers yourself.
1925		3530
1926	If C<fd> is less than 0, then no I/O watcher will be started and events	3531	If C<fd> is less than 0, then no I/O watcher will be started and the
1927	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3532	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
1928	C<events> set will be craeted and started.	3533	the given C<fd> and C<events> set will be created and started.
1929		3534
1930	If C<timeout> is less than 0, then no timeout watcher will be	3535	If C<timeout> is less than 0, then no timeout watcher will be
1931	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3536	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
1932	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3537	repeat = 0) will be started. C<0> is a valid timeout.
1933	dubious value.
1934		3538
1935	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3539	The callback has the type C<void (*cb)(int revents, void *arg)> and is
1936	passed an C<revents> set like normal event callbacks (a combination of	3540	passed an C<revents> set like normal event callbacks (a combination of
1937	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3541	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
1938	value passed to C<ev_once>:	3542	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3543	a timeout and an io event at the same time - you probably should give io
		3544	events precedence.
1939		3545
		3546	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3547
1940	static void stdin_ready (int revents, void *arg)	3548	static void stdin_ready (int revents, void *arg)
		3549	{
		3550	if (revents & EV_READ)
		3551	/* stdin might have data for us, joy! */;
		3552	else if (revents & EV_TIMER)
		3553	/* doh, nothing entered */;
		3554	}
		3555
		3556	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
		3557
		3558	=item ev_feed_fd_event (loop, int fd, int revents)
		3559
		3560	Feed an event on the given fd, as if a file descriptor backend detected
		3561	the given events.
		3562
		3563	=item ev_feed_signal_event (loop, int signum)
		3564
		3565	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
		3566	which is async-safe.
		3567
		3568	=back
		3569
		3570
		3571	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3572
		3573	This section explains some common idioms that are not immediately
		3574	obvious. Note that examples are sprinkled over the whole manual, and this
		3575	section only contains stuff that wouldn't fit anywhere else.
		3576
		3577	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3578
		3579	Each watcher has, by default, a C<void *data> member that you can read
		3580	or modify at any time: libev will completely ignore it. This can be used
		3581	to associate arbitrary data with your watcher. If you need more data and
		3582	don't want to allocate memory separately and store a pointer to it in that
		3583	data member, you can also "subclass" the watcher type and provide your own
		3584	data:
		3585
		3586	struct my_io
		3587	{
		3588	ev_io io;
		3589	int otherfd;
		3590	void *somedata;
		3591	struct whatever *mostinteresting;
		3592	};
		3593
		3594	...
		3595	struct my_io w;
		3596	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3597
		3598	And since your callback will be called with a pointer to the watcher, you
		3599	can cast it back to your own type:
		3600
		3601	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3602	{
		3603	struct my_io *w = (struct my_io *)w_;
		3604	...
		3605	}
		3606
		3607	More interesting and less C-conformant ways of casting your callback
		3608	function type instead have been omitted.
		3609
		3610	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3611
		3612	Another common scenario is to use some data structure with multiple
		3613	embedded watchers, in effect creating your own watcher that combines
		3614	multiple libev event sources into one "super-watcher":
		3615
		3616	struct my_biggy
		3617	{
		3618	int some_data;
		3619	ev_timer t1;
		3620	ev_timer t2;
		3621	}
		3622
		3623	In this case getting the pointer to C<my_biggy> is a bit more
		3624	complicated: Either you store the address of your C<my_biggy> struct in
		3625	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3626	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3627	real programmers):
		3628
		3629	#include <stddef.h>
		3630
		3631	static void
		3632	t1_cb (EV_P_ ev_timer *w, int revents)
		3633	{
		3634	struct my_biggy big = (struct my_biggy *)
		3635	(((char *)w) - offsetof (struct my_biggy, t1));
		3636	}
		3637
		3638	static void
		3639	t2_cb (EV_P_ ev_timer *w, int revents)
		3640	{
		3641	struct my_biggy big = (struct my_biggy *)
		3642	(((char *)w) - offsetof (struct my_biggy, t2));
		3643	}
		3644
		3645	=head2 AVOIDING FINISHING BEFORE RETURNING
		3646
		3647	Often you have structures like this in event-based programs:
		3648
		3649	callback ()
1941	{	3650	{
1942	if (revents & EV_TIMEOUT)	3651	free (request);
1943	/* doh, nothing entered */;
1944	else if (revents & EV_READ)
1945	/* stdin might have data for us, joy! */;
1946	}	3652	}
1947		3653
1948	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3654	request = start_new_request (..., callback);
1949		3655
1950	=item ev_feed_event (ev_loop *, watcher *, int revents)	3656	The intent is to start some "lengthy" operation. The C<request> could be
		3657	used to cancel the operation, or do other things with it.
1951		3658
1952	Feeds the given event set into the event loop, as if the specified event	3659	It's not uncommon to have code paths in C<start_new_request> that
1953	had happened for the specified watcher (which must be a pointer to an	3660	immediately invoke the callback, for example, to report errors. Or you add
1954	initialised but not necessarily started event watcher).	3661	some caching layer that finds that it can skip the lengthy aspects of the
		3662	operation and simply invoke the callback with the result.
1955		3663
1956	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	3664	The problem here is that this will happen I<before> C<start_new_request>
		3665	has returned, so C<request> is not set.
1957		3666
1958	Feed an event on the given fd, as if a file descriptor backend detected	3667	Even if you pass the request by some safer means to the callback, you
1959	the given events it.	3668	might want to do something to the request after starting it, such as
		3669	canceling it, which probably isn't working so well when the callback has
		3670	already been invoked.
1960		3671
1961	=item ev_feed_signal_event (ev_loop *loop, int signum)	3672	A common way around all these issues is to make sure that
		3673	C<start_new_request> I<always> returns before the callback is invoked. If
		3674	C<start_new_request> immediately knows the result, it can artificially
		3675	delay invoking the callback by using a C<prepare> or C<idle> watcher for
		3676	example, or more sneakily, by reusing an existing (stopped) watcher and
		3677	pushing it into the pending queue:
1962		3678
1963	Feed an event as if the given signal occured (C<loop> must be the default	3679	ev_set_cb (watcher, callback);
1964	loop!).	3680	ev_feed_event (EV_A_ watcher, 0);
1965		3681
1966	=back	3682	This way, C<start_new_request> can safely return before the callback is
		3683	invoked, while not delaying callback invocation too much.
		3684
		3685	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3686
		3687	Often (especially in GUI toolkits) there are places where you have
		3688	I<modal> interaction, which is most easily implemented by recursively
		3689	invoking C<ev_run>.
		3690
		3691	This brings the problem of exiting - a callback might want to finish the
		3692	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3693	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3694	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3695	other combination: In these cases, a simple C<ev_break> will not work.
		3696
		3697	The solution is to maintain "break this loop" variable for each C<ev_run>
		3698	invocation, and use a loop around C<ev_run> until the condition is
		3699	triggered, using C<EVRUN_ONCE>:
		3700
		3701	// main loop
		3702	int exit_main_loop = 0;
		3703
		3704	while (!exit_main_loop)
		3705	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3706
		3707	// in a modal watcher
		3708	int exit_nested_loop = 0;
		3709
		3710	while (!exit_nested_loop)
		3711	ev_run (EV_A_ EVRUN_ONCE);
		3712
		3713	To exit from any of these loops, just set the corresponding exit variable:
		3714
		3715	// exit modal loop
		3716	exit_nested_loop = 1;
		3717
		3718	// exit main program, after modal loop is finished
		3719	exit_main_loop = 1;
		3720
		3721	// exit both
		3722	exit_main_loop = exit_nested_loop = 1;
		3723
		3724	=head2 THREAD LOCKING EXAMPLE
		3725
		3726	Here is a fictitious example of how to run an event loop in a different
		3727	thread from where callbacks are being invoked and watchers are
		3728	created/added/removed.
		3729
		3730	For a real-world example, see the C<EV::Loop::Async> perl module,
		3731	which uses exactly this technique (which is suited for many high-level
		3732	languages).
		3733
		3734	The example uses a pthread mutex to protect the loop data, a condition
		3735	variable to wait for callback invocations, an async watcher to notify the
		3736	event loop thread and an unspecified mechanism to wake up the main thread.
		3737
		3738	First, you need to associate some data with the event loop:
		3739
		3740	typedef struct {
		3741	mutex_t lock; /* global loop lock */
		3742	ev_async async_w;
		3743	thread_t tid;
		3744	cond_t invoke_cv;
		3745	} userdata;
		3746
		3747	void prepare_loop (EV_P)
		3748	{
		3749	// for simplicity, we use a static userdata struct.
		3750	static userdata u;
		3751
		3752	ev_async_init (&u->async_w, async_cb);
		3753	ev_async_start (EV_A_ &u->async_w);
		3754
		3755	pthread_mutex_init (&u->lock, 0);
		3756	pthread_cond_init (&u->invoke_cv, 0);
		3757
		3758	// now associate this with the loop
		3759	ev_set_userdata (EV_A_ u);
		3760	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3761	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3762
		3763	// then create the thread running ev_run
		3764	pthread_create (&u->tid, 0, l_run, EV_A);
		3765	}
		3766
		3767	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3768	solely to wake up the event loop so it takes notice of any new watchers
		3769	that might have been added:
		3770
		3771	static void
		3772	async_cb (EV_P_ ev_async *w, int revents)
		3773	{
		3774	// just used for the side effects
		3775	}
		3776
		3777	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3778	protecting the loop data, respectively.
		3779
		3780	static void
		3781	l_release (EV_P)
		3782	{
		3783	userdata *u = ev_userdata (EV_A);
		3784	pthread_mutex_unlock (&u->lock);
		3785	}
		3786
		3787	static void
		3788	l_acquire (EV_P)
		3789	{
		3790	userdata *u = ev_userdata (EV_A);
		3791	pthread_mutex_lock (&u->lock);
		3792	}
		3793
		3794	The event loop thread first acquires the mutex, and then jumps straight
		3795	into C<ev_run>:
		3796
		3797	void *
		3798	l_run (void *thr_arg)
		3799	{
		3800	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3801
		3802	l_acquire (EV_A);
		3803	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3804	ev_run (EV_A_ 0);
		3805	l_release (EV_A);
		3806
		3807	return 0;
		3808	}
		3809
		3810	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3811	signal the main thread via some unspecified mechanism (signals? pipe
		3812	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3813	have been called (in a while loop because a) spurious wakeups are possible
		3814	and b) skipping inter-thread-communication when there are no pending
		3815	watchers is very beneficial):
		3816
		3817	static void
		3818	l_invoke (EV_P)
		3819	{
		3820	userdata *u = ev_userdata (EV_A);
		3821
		3822	while (ev_pending_count (EV_A))
		3823	{
		3824	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3825	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3826	}
		3827	}
		3828
		3829	Now, whenever the main thread gets told to invoke pending watchers, it
		3830	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3831	thread to continue:
		3832
		3833	static void
		3834	real_invoke_pending (EV_P)
		3835	{
		3836	userdata *u = ev_userdata (EV_A);
		3837
		3838	pthread_mutex_lock (&u->lock);
		3839	ev_invoke_pending (EV_A);
		3840	pthread_cond_signal (&u->invoke_cv);
		3841	pthread_mutex_unlock (&u->lock);
		3842	}
		3843
		3844	Whenever you want to start/stop a watcher or do other modifications to an
		3845	event loop, you will now have to lock:
		3846
		3847	ev_timer timeout_watcher;
		3848	userdata *u = ev_userdata (EV_A);
		3849
		3850	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3851
		3852	pthread_mutex_lock (&u->lock);
		3853	ev_timer_start (EV_A_ &timeout_watcher);
		3854	ev_async_send (EV_A_ &u->async_w);
		3855	pthread_mutex_unlock (&u->lock);
		3856
		3857	Note that sending the C<ev_async> watcher is required because otherwise
		3858	an event loop currently blocking in the kernel will have no knowledge
		3859	about the newly added timer. By waking up the loop it will pick up any new
		3860	watchers in the next event loop iteration.
		3861
		3862	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3863
		3864	While the overhead of a callback that e.g. schedules a thread is small, it
		3865	is still an overhead. If you embed libev, and your main usage is with some
		3866	kind of threads or coroutines, you might want to customise libev so that
		3867	doesn't need callbacks anymore.
		3868
		3869	Imagine you have coroutines that you can switch to using a function
		3870	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3871	and that due to some magic, the currently active coroutine is stored in a
		3872	global called C<current_coro>. Then you can build your own "wait for libev
		3873	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3874	the differing C<;> conventions):
		3875
		3876	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3877	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3878
		3879	That means instead of having a C callback function, you store the
		3880	coroutine to switch to in each watcher, and instead of having libev call
		3881	your callback, you instead have it switch to that coroutine.
		3882
		3883	A coroutine might now wait for an event with a function called
		3884	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		3885	matter when, or whether the watcher is active or not when this function is
		3886	called):
		3887
		3888	void
		3889	wait_for_event (ev_watcher *w)
		3890	{
		3891	ev_set_cb (w, current_coro);
		3892	switch_to (libev_coro);
		3893	}
		3894
		3895	That basically suspends the coroutine inside C<wait_for_event> and
		3896	continues the libev coroutine, which, when appropriate, switches back to
		3897	this or any other coroutine.
		3898
		3899	You can do similar tricks if you have, say, threads with an event queue -
		3900	instead of storing a coroutine, you store the queue object and instead of
		3901	switching to a coroutine, you push the watcher onto the queue and notify
		3902	any waiters.
		3903
		3904	To embed libev, see L</EMBEDDING>, but in short, it's easiest to create two
		3905	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		3906
		3907	// my_ev.h
		3908	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3909	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3910	#include "../libev/ev.h"
		3911
		3912	// my_ev.c
		3913	#define EV_H "my_ev.h"
		3914	#include "../libev/ev.c"
		3915
		3916	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		3917	F<my_ev.c> into your project. When properly specifying include paths, you
		3918	can even use F<ev.h> as header file name directly.
1967		3919
1968		3920
1969	=head1 LIBEVENT EMULATION	3921	=head1 LIBEVENT EMULATION
1970		3922
1971	Libev offers a compatibility emulation layer for libevent. It cannot	3923	Libev offers a compatibility emulation layer for libevent. It cannot
1972	emulate the internals of libevent, so here are some usage hints:	3924	emulate the internals of libevent, so here are some usage hints:
1973		3925
1974	=over 4	3926	=over 4
		3927
		3928	=item * Only the libevent-1.4.1-beta API is being emulated.
		3929
		3930	This was the newest libevent version available when libev was implemented,
		3931	and is still mostly unchanged in 2010.
1975		3932
1976	=item * Use it by including <event.h>, as usual.	3933	=item * Use it by including <event.h>, as usual.
1977		3934
1978	=item * The following members are fully supported: ev_base, ev_callback,	3935	=item * The following members are fully supported: ev_base, ev_callback,
1979	ev_arg, ev_fd, ev_res, ev_events.	3936	ev_arg, ev_fd, ev_res, ev_events.
…		…
1984		3941
1985	=item * Priorities are not currently supported. Initialising priorities	3942	=item * Priorities are not currently supported. Initialising priorities
1986	will fail and all watchers will have the same priority, even though there	3943	will fail and all watchers will have the same priority, even though there
1987	is an ev_pri field.	3944	is an ev_pri field.
1988		3945
		3946	=item * In libevent, the last base created gets the signals, in libev, the
		3947	base that registered the signal gets the signals.
		3948
1989	=item * Other members are not supported.	3949	=item * Other members are not supported.
1990		3950
1991	=item * The libev emulation is I<not> ABI compatible to libevent, you need	3951	=item * The libev emulation is I<not> ABI compatible to libevent, you need
1992	to use the libev header file and library.	3952	to use the libev header file and library.
1993		3953
1994	=back	3954	=back
1995		3955
1996	=head1 C++ SUPPORT	3956	=head1 C++ SUPPORT
1997		3957
		3958	=head2 C API
		3959
		3960	The normal C API should work fine when used from C++: both ev.h and the
		3961	libev sources can be compiled as C++. Therefore, code that uses the C API
		3962	will work fine.
		3963
		3964	Proper exception specifications might have to be added to callbacks passed
		3965	to libev: exceptions may be thrown only from watcher callbacks, all
		3966	other callbacks (allocator, syserr, loop acquire/release and periodic
		3967	reschedule callbacks) must not throw exceptions, and might need a C<throw
		3968	()> specification. If you have code that needs to be compiled as both C
		3969	and C++ you can use the C<EV_THROW> macro for this:
		3970
		3971	static void
		3972	fatal_error (const char *msg) EV_THROW
		3973	{
		3974	perror (msg);
		3975	abort ();
		3976	}
		3977
		3978	...
		3979	ev_set_syserr_cb (fatal_error);
		3980
		3981	The only API functions that can currently throw exceptions are C<ev_run>,
		3982	C<ev_invoke>, C<ev_invoke_pending> and C<ev_loop_destroy> (the latter
		3983	because it runs cleanup watchers).
		3984
		3985	Throwing exceptions in watcher callbacks is only supported if libev itself
		3986	is compiled with a C++ compiler or your C and C++ environments allow
		3987	throwing exceptions through C libraries (most do).
		3988
		3989	=head2 C++ API
		3990
1998	Libev comes with some simplistic wrapper classes for C++ that mainly allow	3991	Libev comes with some simplistic wrapper classes for C++ that mainly allow
1999	you to use some convinience methods to start/stop watchers and also change	3992	you to use some convenience methods to start/stop watchers and also change
2000	the callback model to a model using method callbacks on objects.	3993	the callback model to a model using method callbacks on objects.
2001		3994
2002	To use it,	3995	To use it,
2003		3996
2004	#include <ev++.h>	3997	#include <ev++.h>
2005		3998
2006	This automatically includes F<ev.h> and puts all of its definitions (many	3999	This automatically includes F<ev.h> and puts all of its definitions (many
2007	of them macros) into the global namespace. All C++ specific things are	4000	of them macros) into the global namespace. All C++ specific things are
2008	put into the C<ev> namespace. It should support all the same embedding	4001	put into the C<ev> namespace. It should support all the same embedding
2009	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.	4002	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
…		…
2011	Care has been taken to keep the overhead low. The only data member the C++	4004	Care has been taken to keep the overhead low. The only data member the C++
2012	classes add (compared to plain C-style watchers) is the event loop pointer	4005	classes add (compared to plain C-style watchers) is the event loop pointer
2013	that the watcher is associated with (or no additional members at all if	4006	that the watcher is associated with (or no additional members at all if
2014	you disable C<EV_MULTIPLICITY> when embedding libev).	4007	you disable C<EV_MULTIPLICITY> when embedding libev).
2015		4008
2016	Currently, functions, and static and non-static member functions can be	4009	Currently, functions, static and non-static member functions and classes
2017	used as callbacks. Other types should be easy to add as long as they only	4010	with C<operator ()> can be used as callbacks. Other types should be easy
2018	need one additional pointer for context. If you need support for other	4011	to add as long as they only need one additional pointer for context. If
2019	types of functors please contact the author (preferably after implementing	4012	you need support for other types of functors please contact the author
2020	it).	4013	(preferably after implementing it).
		4014
		4015	For all this to work, your C++ compiler either has to use the same calling
		4016	conventions as your C compiler (for static member functions), or you have
		4017	to embed libev and compile libev itself as C++.
2021		4018
2022	Here is a list of things available in the C<ev> namespace:	4019	Here is a list of things available in the C<ev> namespace:
2023		4020
2024	=over 4	4021	=over 4
2025		4022
…		…
2035	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.	4032	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
2036		4033
2037	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of	4034	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
2038	the same name in the C<ev> namespace, with the exception of C<ev_signal>	4035	the same name in the C<ev> namespace, with the exception of C<ev_signal>
2039	which is called C<ev::sig> to avoid clashes with the C<signal> macro	4036	which is called C<ev::sig> to avoid clashes with the C<signal> macro
2040	defines by many implementations.	4037	defined by many implementations.
2041		4038
2042	All of those classes have these methods:	4039	All of those classes have these methods:
2043		4040
2044	=over 4	4041	=over 4
2045		4042
2046	=item ev::TYPE::TYPE ()	4043	=item ev::TYPE::TYPE ()
2047		4044
2048	=item ev::TYPE::TYPE (struct ev_loop *)	4045	=item ev::TYPE::TYPE (loop)
2049		4046
2050	=item ev::TYPE::~TYPE	4047	=item ev::TYPE::~TYPE
2051		4048
2052	The constructor (optionally) takes an event loop to associate the watcher	4049	The constructor (optionally) takes an event loop to associate the watcher
2053	with. If it is omitted, it will use C<EV_DEFAULT>.	4050	with. If it is omitted, it will use C<EV_DEFAULT>.
…		…
2076	your compiler is good :), then the method will be fully inlined into the	4073	your compiler is good :), then the method will be fully inlined into the
2077	thunking function, making it as fast as a direct C callback.	4074	thunking function, making it as fast as a direct C callback.
2078		4075
2079	Example: simple class declaration and watcher initialisation	4076	Example: simple class declaration and watcher initialisation
2080		4077
2081	struct myclass	4078	struct myclass
2082	{	4079	{
2083	void io_cb (ev::io &w, int revents) { }	4080	void io_cb (ev::io &w, int revents) { }
2084	}	4081	}
2085		4082
2086	myclass obj;	4083	myclass obj;
2087	ev::io iow;	4084	ev::io iow;
2088	iow.set <myclass, &myclass::io_cb> (&obj);	4085	iow.set <myclass, &myclass::io_cb> (&obj);
		4086
		4087	=item w->set (object *)
		4088
		4089	This is a variation of a method callback - leaving out the method to call
		4090	will default the method to C<operator ()>, which makes it possible to use
		4091	functor objects without having to manually specify the C<operator ()> all
		4092	the time. Incidentally, you can then also leave out the template argument
		4093	list.
		4094
		4095	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		4096	int revents)>.
		4097
		4098	See the method-C<set> above for more details.
		4099
		4100	Example: use a functor object as callback.
		4101
		4102	struct myfunctor
		4103	{
		4104	void operator() (ev::io &w, int revents)
		4105	{
		4106	...
		4107	}
		4108	}
		4109
		4110	myfunctor f;
		4111
		4112	ev::io w;
		4113	w.set (&f);
2089		4114
2090	=item w->set<function> (void *data = 0)	4115	=item w->set<function> (void *data = 0)
2091		4116
2092	Also sets a callback, but uses a static method or plain function as	4117	Also sets a callback, but uses a static method or plain function as
2093	callback. The optional C<data> argument will be stored in the watcher's	4118	callback. The optional C<data> argument will be stored in the watcher's
…		…
2095		4120
2096	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.	4121	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
2097		4122
2098	See the method-C<set> above for more details.	4123	See the method-C<set> above for more details.
2099		4124
2100	Example:	4125	Example: Use a plain function as callback.
2101		4126
2102	static void io_cb (ev::io &w, int revents) { }	4127	static void io_cb (ev::io &w, int revents) { }
2103	iow.set <io_cb> ();	4128	iow.set <io_cb> ();
2104		4129
2105	=item w->set (struct ev_loop *)	4130	=item w->set (loop)
2106		4131
2107	Associates a different C<struct ev_loop> with this watcher. You can only	4132	Associates a different C<struct ev_loop> with this watcher. You can only
2108	do this when the watcher is inactive (and not pending either).	4133	do this when the watcher is inactive (and not pending either).
2109		4134
2110	=item w->set ([args])	4135	=item w->set ([arguments])
2111		4136
2112	Basically the same as C<ev_TYPE_set>, with the same args. Must be	4137	Basically the same as C<ev_TYPE_set> (except for C<ev::embed> watchers>),
		4138	with the same arguments. Either this method or a suitable start method
2113	called at least once. Unlike the C counterpart, an active watcher gets	4139	must be called at least once. Unlike the C counterpart, an active watcher
2114	automatically stopped and restarted when reconfiguring it with this	4140	gets automatically stopped and restarted when reconfiguring it with this
2115	method.	4141	method.
		4142
		4143	For C<ev::embed> watchers this method is called C<set_embed>, to avoid
		4144	clashing with the C<set (loop)> method.
2116		4145
2117	=item w->start ()	4146	=item w->start ()
2118		4147
2119	Starts the watcher. Note that there is no C<loop> argument, as the	4148	Starts the watcher. Note that there is no C<loop> argument, as the
2120	constructor already stores the event loop.	4149	constructor already stores the event loop.
2121		4150
		4151	=item w->start ([arguments])
		4152
		4153	Instead of calling C<set> and C<start> methods separately, it is often
		4154	convenient to wrap them in one call. Uses the same type of arguments as
		4155	the configure C<set> method of the watcher.
		4156
2122	=item w->stop ()	4157	=item w->stop ()
2123		4158
2124	Stops the watcher if it is active. Again, no C<loop> argument.	4159	Stops the watcher if it is active. Again, no C<loop> argument.
2125		4160
2126	=item w->again () (C<ev::timer>, C<ev::periodic> only)	4161	=item w->again () (C<ev::timer>, C<ev::periodic> only)
…		…
2138		4173
2139	=back	4174	=back
2140		4175
2141	=back	4176	=back
2142		4177
2143	Example: Define a class with an IO and idle watcher, start one of them in	4178	Example: Define a class with two I/O and idle watchers, start the I/O
2144	the constructor.	4179	watchers in the constructor.
2145		4180
2146	class myclass	4181	class myclass
2147	{	4182	{
2148	ev_io io; void io_cb (ev::io &w, int revents);	4183	ev::io io ; void io_cb (ev::io &w, int revents);
		4184	ev::io io2 ; void io2_cb (ev::io &w, int revents);
2149	ev_idle idle void idle_cb (ev::idle &w, int revents);	4185	ev::idle idle; void idle_cb (ev::idle &w, int revents);
2150		4186
2151	myclass ();	4187	myclass (int fd)
2152	}	4188	{
2153
2154	myclass::myclass (int fd)
2155	{
2156	io .set <myclass, &myclass::io_cb > (this);	4189	io .set <myclass, &myclass::io_cb > (this);
		4190	io2 .set <myclass, &myclass::io2_cb > (this);
2157	idle.set <myclass, &myclass::idle_cb> (this);	4191	idle.set <myclass, &myclass::idle_cb> (this);
2158		4192
2159	io.start (fd, ev::READ);	4193	io.set (fd, ev::WRITE); // configure the watcher
		4194	io.start (); // start it whenever convenient
		4195
		4196	io2.start (fd, ev::READ); // set + start in one call
		4197	}
2160	}	4198	};
		4199
		4200
		4201	=head1 OTHER LANGUAGE BINDINGS
		4202
		4203	Libev does not offer other language bindings itself, but bindings for a
		4204	number of languages exist in the form of third-party packages. If you know
		4205	any interesting language binding in addition to the ones listed here, drop
		4206	me a note.
		4207
		4208	=over 4
		4209
		4210	=item Perl
		4211
		4212	The EV module implements the full libev API and is actually used to test
		4213	libev. EV is developed together with libev. Apart from the EV core module,
		4214	there are additional modules that implement libev-compatible interfaces
		4215	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		4216	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		4217	and C<EV::Glib>).
		4218
		4219	It can be found and installed via CPAN, its homepage is at
		4220	L<http://software.schmorp.de/pkg/EV>.
		4221
		4222	=item Python
		4223
		4224	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		4225	seems to be quite complete and well-documented.
		4226
		4227	=item Ruby
		4228
		4229	Tony Arcieri has written a ruby extension that offers access to a subset
		4230	of the libev API and adds file handle abstractions, asynchronous DNS and
		4231	more on top of it. It can be found via gem servers. Its homepage is at
		4232	L<http://rev.rubyforge.org/>.
		4233
		4234	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		4235	makes rev work even on mingw.
		4236
		4237	=item Haskell
		4238
		4239	A haskell binding to libev is available at
		4240	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		4241
		4242	=item D
		4243
		4244	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		4245	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
		4246
		4247	=item Ocaml
		4248
		4249	Erkki Seppala has written Ocaml bindings for libev, to be found at
		4250	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		4251
		4252	=item Lua
		4253
		4254	Brian Maher has written a partial interface to libev for lua (at the
		4255	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		4256	L<http://github.com/brimworks/lua-ev>.
		4257
		4258	=item Javascript
		4259
		4260	Node.js (L<http://nodejs.org>) uses libev as the underlying event library.
		4261
		4262	=item Others
		4263
		4264	There are others, and I stopped counting.
		4265
		4266	=back
2161		4267
2162		4268
2163	=head1 MACRO MAGIC	4269	=head1 MACRO MAGIC
2164		4270
2165	Libev can be compiled with a variety of options, the most fundamantal	4271	Libev can be compiled with a variety of options, the most fundamental
2166	of which is C<EV_MULTIPLICITY>. This option determines whether (most)	4272	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
2167	functions and callbacks have an initial C<struct ev_loop *> argument.	4273	functions and callbacks have an initial C<struct ev_loop *> argument.
2168		4274
2169	To make it easier to write programs that cope with either variant, the	4275	To make it easier to write programs that cope with either variant, the
2170	following macros are defined:	4276	following macros are defined:
…		…
2175		4281
2176	This provides the loop I<argument> for functions, if one is required ("ev	4282	This provides the loop I<argument> for functions, if one is required ("ev
2177	loop argument"). The C<EV_A> form is used when this is the sole argument,	4283	loop argument"). The C<EV_A> form is used when this is the sole argument,
2178	C<EV_A_> is used when other arguments are following. Example:	4284	C<EV_A_> is used when other arguments are following. Example:
2179		4285
2180	ev_unref (EV_A);	4286	ev_unref (EV_A);
2181	ev_timer_add (EV_A_ watcher);	4287	ev_timer_add (EV_A_ watcher);
2182	ev_loop (EV_A_ 0);	4288	ev_run (EV_A_ 0);
2183		4289
2184	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	4290	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
2185	which is often provided by the following macro.	4291	which is often provided by the following macro.
2186		4292
2187	=item C<EV_P>, C<EV_P_>	4293	=item C<EV_P>, C<EV_P_>
2188		4294
2189	This provides the loop I<parameter> for functions, if one is required ("ev	4295	This provides the loop I<parameter> for functions, if one is required ("ev
2190	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	4296	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
2191	C<EV_P_> is used when other parameters are following. Example:	4297	C<EV_P_> is used when other parameters are following. Example:
2192		4298
2193	// this is how ev_unref is being declared	4299	// this is how ev_unref is being declared
2194	static void ev_unref (EV_P);	4300	static void ev_unref (EV_P);
2195		4301
2196	// this is how you can declare your typical callback	4302	// this is how you can declare your typical callback
2197	static void cb (EV_P_ ev_timer *w, int revents)	4303	static void cb (EV_P_ ev_timer *w, int revents)
2198		4304
2199	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	4305	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
2200	suitable for use with C<EV_A>.	4306	suitable for use with C<EV_A>.
2201		4307
2202	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	4308	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
2203		4309
2204	Similar to the other two macros, this gives you the value of the default	4310	Similar to the other two macros, this gives you the value of the default
2205	loop, if multiple loops are supported ("ev loop default").	4311	loop, if multiple loops are supported ("ev loop default"). The default loop
		4312	will be initialised if it isn't already initialised.
		4313
		4314	For non-multiplicity builds, these macros do nothing, so you always have
		4315	to initialise the loop somewhere.
		4316
		4317	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		4318
		4319	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		4320	default loop has been initialised (C<UC> == unchecked). Their behaviour
		4321	is undefined when the default loop has not been initialised by a previous
		4322	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		4323
		4324	It is often prudent to use C<EV_DEFAULT> when initialising the first
		4325	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
2206		4326
2207	=back	4327	=back
2208		4328
2209	Example: Declare and initialise a check watcher, utilising the above	4329	Example: Declare and initialise a check watcher, utilising the above
2210	macros so it will work regardless of whether multiple loops are supported	4330	macros so it will work regardless of whether multiple loops are supported
2211	or not.	4331	or not.
2212		4332
2213	static void	4333	static void
2214	check_cb (EV_P_ ev_timer *w, int revents)	4334	check_cb (EV_P_ ev_timer *w, int revents)
2215	{	4335	{
2216	ev_check_stop (EV_A_ w);	4336	ev_check_stop (EV_A_ w);
2217	}	4337	}
2218		4338
2219	ev_check check;	4339	ev_check check;
2220	ev_check_init (&check, check_cb);	4340	ev_check_init (&check, check_cb);
2221	ev_check_start (EV_DEFAULT_ &check);	4341	ev_check_start (EV_DEFAULT_ &check);
2222	ev_loop (EV_DEFAULT_ 0);	4342	ev_run (EV_DEFAULT_ 0);
2223		4343
2224	=head1 EMBEDDING	4344	=head1 EMBEDDING
2225		4345
2226	Libev can (and often is) directly embedded into host	4346	Libev can (and often is) directly embedded into host
2227	applications. Examples of applications that embed it include the Deliantra	4347	applications. Examples of applications that embed it include the Deliantra
…		…
2234	libev somewhere in your source tree).	4354	libev somewhere in your source tree).
2235		4355
2236	=head2 FILESETS	4356	=head2 FILESETS
2237		4357
2238	Depending on what features you need you need to include one or more sets of files	4358	Depending on what features you need you need to include one or more sets of files
2239	in your app.	4359	in your application.
2240		4360
2241	=head3 CORE EVENT LOOP	4361	=head3 CORE EVENT LOOP
2242		4362
2243	To include only the libev core (all the C<ev_*> functions), with manual	4363	To include only the libev core (all the C<ev_*> functions), with manual
2244	configuration (no autoconf):	4364	configuration (no autoconf):
2245		4365
2246	#define EV_STANDALONE 1	4366	#define EV_STANDALONE 1
2247	#include "ev.c"	4367	#include "ev.c"
2248		4368
2249	This will automatically include F<ev.h>, too, and should be done in a	4369	This will automatically include F<ev.h>, too, and should be done in a
2250	single C source file only to provide the function implementations. To use	4370	single C source file only to provide the function implementations. To use
2251	it, do the same for F<ev.h> in all files wishing to use this API (best	4371	it, do the same for F<ev.h> in all files wishing to use this API (best
2252	done by writing a wrapper around F<ev.h> that you can include instead and	4372	done by writing a wrapper around F<ev.h> that you can include instead and
2253	where you can put other configuration options):	4373	where you can put other configuration options):
2254		4374
2255	#define EV_STANDALONE 1	4375	#define EV_STANDALONE 1
2256	#include "ev.h"	4376	#include "ev.h"
2257		4377
2258	Both header files and implementation files can be compiled with a C++	4378	Both header files and implementation files can be compiled with a C++
2259	compiler (at least, thats a stated goal, and breakage will be treated	4379	compiler (at least, that's a stated goal, and breakage will be treated
2260	as a bug).	4380	as a bug).
2261		4381
2262	You need the following files in your source tree, or in a directory	4382	You need the following files in your source tree, or in a directory
2263	in your include path (e.g. in libev/ when using -Ilibev):	4383	in your include path (e.g. in libev/ when using -Ilibev):
2264		4384
2265	ev.h	4385	ev.h
2266	ev.c	4386	ev.c
2267	ev_vars.h	4387	ev_vars.h
2268	ev_wrap.h	4388	ev_wrap.h
2269		4389
2270	ev_win32.c required on win32 platforms only	4390	ev_win32.c required on win32 platforms only
2271		4391
2272	ev_select.c only when select backend is enabled (which is enabled by default)	4392	ev_select.c only when select backend is enabled
2273	ev_poll.c only when poll backend is enabled (disabled by default)	4393	ev_poll.c only when poll backend is enabled
2274	ev_epoll.c only when the epoll backend is enabled (disabled by default)	4394	ev_epoll.c only when the epoll backend is enabled
2275	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	4395	ev_kqueue.c only when the kqueue backend is enabled
2276	ev_port.c only when the solaris port backend is enabled (disabled by default)	4396	ev_port.c only when the solaris port backend is enabled
2277		4397
2278	F<ev.c> includes the backend files directly when enabled, so you only need	4398	F<ev.c> includes the backend files directly when enabled, so you only need
2279	to compile this single file.	4399	to compile this single file.
2280		4400
2281	=head3 LIBEVENT COMPATIBILITY API	4401	=head3 LIBEVENT COMPATIBILITY API
2282		4402
2283	To include the libevent compatibility API, also include:	4403	To include the libevent compatibility API, also include:
2284		4404
2285	#include "event.c"	4405	#include "event.c"
2286		4406
2287	in the file including F<ev.c>, and:	4407	in the file including F<ev.c>, and:
2288		4408
2289	#include "event.h"	4409	#include "event.h"
2290		4410
2291	in the files that want to use the libevent API. This also includes F<ev.h>.	4411	in the files that want to use the libevent API. This also includes F<ev.h>.
2292		4412
2293	You need the following additional files for this:	4413	You need the following additional files for this:
2294		4414
2295	event.h	4415	event.h
2296	event.c	4416	event.c
2297		4417
2298	=head3 AUTOCONF SUPPORT	4418	=head3 AUTOCONF SUPPORT
2299		4419
2300	Instead of using C<EV_STANDALONE=1> and providing your config in	4420	Instead of using C<EV_STANDALONE=1> and providing your configuration in
2301	whatever way you want, you can also C<m4_include([libev.m4])> in your	4421	whatever way you want, you can also C<m4_include([libev.m4])> in your
2302	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	4422	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
2303	include F<config.h> and configure itself accordingly.	4423	include F<config.h> and configure itself accordingly.
2304		4424
2305	For this of course you need the m4 file:	4425	For this of course you need the m4 file:
2306		4426
2307	libev.m4	4427	libev.m4
2308		4428
2309	=head2 PREPROCESSOR SYMBOLS/MACROS	4429	=head2 PREPROCESSOR SYMBOLS/MACROS
2310		4430
2311	Libev can be configured via a variety of preprocessor symbols you have to define	4431	Libev can be configured via a variety of preprocessor symbols you have to
2312	before including any of its files. The default is not to build for multiplicity	4432	define before including (or compiling) any of its files. The default in
2313	and only include the select backend.	4433	the absence of autoconf is documented for every option.
		4434
		4435	Symbols marked with "(h)" do not change the ABI, and can have different
		4436	values when compiling libev vs. including F<ev.h>, so it is permissible
		4437	to redefine them before including F<ev.h> without breaking compatibility
		4438	to a compiled library. All other symbols change the ABI, which means all
		4439	users of libev and the libev code itself must be compiled with compatible
		4440	settings.
2314		4441
2315	=over 4	4442	=over 4
2316		4443
		4444	=item EV_COMPAT3 (h)
		4445
		4446	Backwards compatibility is a major concern for libev. This is why this
		4447	release of libev comes with wrappers for the functions and symbols that
		4448	have been renamed between libev version 3 and 4.
		4449
		4450	You can disable these wrappers (to test compatibility with future
		4451	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4452	sources. This has the additional advantage that you can drop the C<struct>
		4453	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4454	typedef in that case.
		4455
		4456	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4457	and in some even more future version the compatibility code will be
		4458	removed completely.
		4459
2317	=item EV_STANDALONE	4460	=item EV_STANDALONE (h)
2318		4461
2319	Must always be C<1> if you do not use autoconf configuration, which	4462	Must always be C<1> if you do not use autoconf configuration, which
2320	keeps libev from including F<config.h>, and it also defines dummy	4463	keeps libev from including F<config.h>, and it also defines dummy
2321	implementations for some libevent functions (such as logging, which is not	4464	implementations for some libevent functions (such as logging, which is not
2322	supported). It will also not define any of the structs usually found in	4465	supported). It will also not define any of the structs usually found in
2323	F<event.h> that are not directly supported by the libev core alone.	4466	F<event.h> that are not directly supported by the libev core alone.
2324		4467
		4468	In standalone mode, libev will still try to automatically deduce the
		4469	configuration, but has to be more conservative.
		4470
		4471	=item EV_USE_FLOOR
		4472
		4473	If defined to be C<1>, libev will use the C<floor ()> function for its
		4474	periodic reschedule calculations, otherwise libev will fall back on a
		4475	portable (slower) implementation. If you enable this, you usually have to
		4476	link against libm or something equivalent. Enabling this when the C<floor>
		4477	function is not available will fail, so the safe default is to not enable
		4478	this.
		4479
2325	=item EV_USE_MONOTONIC	4480	=item EV_USE_MONOTONIC
2326		4481
2327	If defined to be C<1>, libev will try to detect the availability of the	4482	If defined to be C<1>, libev will try to detect the availability of the
2328	monotonic clock option at both compiletime and runtime. Otherwise no use	4483	monotonic clock option at both compile time and runtime. Otherwise no
2329	of the monotonic clock option will be attempted. If you enable this, you	4484	use of the monotonic clock option will be attempted. If you enable this,
2330	usually have to link against librt or something similar. Enabling it when	4485	you usually have to link against librt or something similar. Enabling it
2331	the functionality isn't available is safe, though, although you have	4486	when the functionality isn't available is safe, though, although you have
2332	to make sure you link against any libraries where the C<clock_gettime>	4487	to make sure you link against any libraries where the C<clock_gettime>
2333	function is hiding in (often F<-lrt>).	4488	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
2334		4489
2335	=item EV_USE_REALTIME	4490	=item EV_USE_REALTIME
2336		4491
2337	If defined to be C<1>, libev will try to detect the availability of the	4492	If defined to be C<1>, libev will try to detect the availability of the
2338	realtime clock option at compiletime (and assume its availability at	4493	real-time clock option at compile time (and assume its availability
2339	runtime if successful). Otherwise no use of the realtime clock option will	4494	at runtime if successful). Otherwise no use of the real-time clock
2340	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	4495	option will be attempted. This effectively replaces C<gettimeofday>
2341	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the	4496	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
2342	note about libraries in the description of C<EV_USE_MONOTONIC>, though.	4497	correctness. See the note about libraries in the description of
		4498	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4499	C<EV_USE_CLOCK_SYSCALL>.
		4500
		4501	=item EV_USE_CLOCK_SYSCALL
		4502
		4503	If defined to be C<1>, libev will try to use a direct syscall instead
		4504	of calling the system-provided C<clock_gettime> function. This option
		4505	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4506	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4507	programs needlessly. Using a direct syscall is slightly slower (in
		4508	theory), because no optimised vdso implementation can be used, but avoids
		4509	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4510	higher, as it simplifies linking (no need for C<-lrt>).
2343		4511
2344	=item EV_USE_NANOSLEEP	4512	=item EV_USE_NANOSLEEP
2345		4513
2346	If defined to be C<1>, libev will assume that C<nanosleep ()> is available	4514	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
2347	and will use it for delays. Otherwise it will use C<select ()>.	4515	and will use it for delays. Otherwise it will use C<select ()>.
2348		4516
		4517	=item EV_USE_EVENTFD
		4518
		4519	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4520	available and will probe for kernel support at runtime. This will improve
		4521	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4522	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4523	2.7 or newer, otherwise disabled.
		4524
2349	=item EV_USE_SELECT	4525	=item EV_USE_SELECT
2350		4526
2351	If undefined or defined to be C<1>, libev will compile in support for the	4527	If undefined or defined to be C<1>, libev will compile in support for the
2352	C<select>(2) backend. No attempt at autodetection will be done: if no	4528	C<select>(2) backend. No attempt at auto-detection will be done: if no
2353	other method takes over, select will be it. Otherwise the select backend	4529	other method takes over, select will be it. Otherwise the select backend
2354	will not be compiled in.	4530	will not be compiled in.
2355		4531
2356	=item EV_SELECT_USE_FD_SET	4532	=item EV_SELECT_USE_FD_SET
2357		4533
2358	If defined to C<1>, then the select backend will use the system C<fd_set>	4534	If defined to C<1>, then the select backend will use the system C<fd_set>
2359	structure. This is useful if libev doesn't compile due to a missing	4535	structure. This is useful if libev doesn't compile due to a missing
2360	C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on	4536	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
2361	exotic systems. This usually limits the range of file descriptors to some	4537	on exotic systems. This usually limits the range of file descriptors to
2362	low limit such as 1024 or might have other limitations (winsocket only	4538	some low limit such as 1024 or might have other limitations (winsocket
2363	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	4539	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
2364	influence the size of the C<fd_set> used.	4540	configures the maximum size of the C<fd_set>.
2365		4541
2366	=item EV_SELECT_IS_WINSOCKET	4542	=item EV_SELECT_IS_WINSOCKET
2367		4543
2368	When defined to C<1>, the select backend will assume that	4544	When defined to C<1>, the select backend will assume that
2369	select/socket/connect etc. don't understand file descriptors but	4545	select/socket/connect etc. don't understand file descriptors but
…		…
2371	be used is the winsock select). This means that it will call	4547	be used is the winsock select). This means that it will call
2372	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	4548	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2373	it is assumed that all these functions actually work on fds, even	4549	it is assumed that all these functions actually work on fds, even
2374	on win32. Should not be defined on non-win32 platforms.	4550	on win32. Should not be defined on non-win32 platforms.
2375		4551
		4552	=item EV_FD_TO_WIN32_HANDLE(fd)
		4553
		4554	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		4555	file descriptors to socket handles. When not defining this symbol (the
		4556	default), then libev will call C<_get_osfhandle>, which is usually
		4557	correct. In some cases, programs use their own file descriptor management,
		4558	in which case they can provide this function to map fds to socket handles.
		4559
		4560	=item EV_WIN32_HANDLE_TO_FD(handle)
		4561
		4562	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4563	using the standard C<_open_osfhandle> function. For programs implementing
		4564	their own fd to handle mapping, overwriting this function makes it easier
		4565	to do so. This can be done by defining this macro to an appropriate value.
		4566
		4567	=item EV_WIN32_CLOSE_FD(fd)
		4568
		4569	If programs implement their own fd to handle mapping on win32, then this
		4570	macro can be used to override the C<close> function, useful to unregister
		4571	file descriptors again. Note that the replacement function has to close
		4572	the underlying OS handle.
		4573
		4574	=item EV_USE_WSASOCKET
		4575
		4576	If defined to be C<1>, libev will use C<WSASocket> to create its internal
		4577	communication socket, which works better in some environments. Otherwise,
		4578	the normal C<socket> function will be used, which works better in other
		4579	environments.
		4580
2376	=item EV_USE_POLL	4581	=item EV_USE_POLL
2377		4582
2378	If defined to be C<1>, libev will compile in support for the C<poll>(2)	4583	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2379	backend. Otherwise it will be enabled on non-win32 platforms. It	4584	backend. Otherwise it will be enabled on non-win32 platforms. It
2380	takes precedence over select.	4585	takes precedence over select.
2381		4586
2382	=item EV_USE_EPOLL	4587	=item EV_USE_EPOLL
2383		4588
2384	If defined to be C<1>, libev will compile in support for the Linux	4589	If defined to be C<1>, libev will compile in support for the Linux
2385	C<epoll>(7) backend. Its availability will be detected at runtime,	4590	C<epoll>(7) backend. Its availability will be detected at runtime,
2386	otherwise another method will be used as fallback. This is the	4591	otherwise another method will be used as fallback. This is the preferred
2387	preferred backend for GNU/Linux systems.	4592	backend for GNU/Linux systems. If undefined, it will be enabled if the
		4593	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
2388		4594
2389	=item EV_USE_KQUEUE	4595	=item EV_USE_KQUEUE
2390		4596
2391	If defined to be C<1>, libev will compile in support for the BSD style	4597	If defined to be C<1>, libev will compile in support for the BSD style
2392	C<kqueue>(2) backend. Its actual availability will be detected at runtime,	4598	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
…		…
2405	otherwise another method will be used as fallback. This is the preferred	4611	otherwise another method will be used as fallback. This is the preferred
2406	backend for Solaris 10 systems.	4612	backend for Solaris 10 systems.
2407		4613
2408	=item EV_USE_DEVPOLL	4614	=item EV_USE_DEVPOLL
2409		4615
2410	reserved for future expansion, works like the USE symbols above.	4616	Reserved for future expansion, works like the USE symbols above.
2411		4617
2412	=item EV_USE_INOTIFY	4618	=item EV_USE_INOTIFY
2413		4619
2414	If defined to be C<1>, libev will compile in support for the Linux inotify	4620	If defined to be C<1>, libev will compile in support for the Linux inotify
2415	interface to speed up C<ev_stat> watchers. Its actual availability will	4621	interface to speed up C<ev_stat> watchers. Its actual availability will
2416	be detected at runtime.	4622	be detected at runtime. If undefined, it will be enabled if the headers
		4623	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
2417		4624
		4625	=item EV_NO_SMP
		4626
		4627	If defined to be C<1>, libev will assume that memory is always coherent
		4628	between threads, that is, threads can be used, but threads never run on
		4629	different cpus (or different cpu cores). This reduces dependencies
		4630	and makes libev faster.
		4631
		4632	=item EV_NO_THREADS
		4633
		4634	If defined to be C<1>, libev will assume that it will never be called from
		4635	different threads (that includes signal handlers), which is a stronger
		4636	assumption than C<EV_NO_SMP>, above. This reduces dependencies and makes
		4637	libev faster.
		4638
		4639	=item EV_ATOMIC_T
		4640
		4641	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		4642	access is atomic with respect to other threads or signal contexts. No
		4643	such type is easily found in the C language, so you can provide your own
		4644	type that you know is safe for your purposes. It is used both for signal
		4645	handler "locking" as well as for signal and thread safety in C<ev_async>
		4646	watchers.
		4647
		4648	In the absence of this define, libev will use C<sig_atomic_t volatile>
		4649	(from F<signal.h>), which is usually good enough on most platforms.
		4650
2418	=item EV_H	4651	=item EV_H (h)
2419		4652
2420	The name of the F<ev.h> header file used to include it. The default if	4653	The name of the F<ev.h> header file used to include it. The default if
2421	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	4654	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2422	can be used to virtually rename the F<ev.h> header file in case of conflicts.	4655	used to virtually rename the F<ev.h> header file in case of conflicts.
2423		4656
2424	=item EV_CONFIG_H	4657	=item EV_CONFIG_H (h)
2425		4658
2426	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	4659	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2427	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	4660	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2428	C<EV_H>, above.	4661	C<EV_H>, above.
2429		4662
2430	=item EV_EVENT_H	4663	=item EV_EVENT_H (h)
2431		4664
2432	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	4665	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2433	of how the F<event.h> header can be found.	4666	of how the F<event.h> header can be found, the default is C<"event.h">.
2434		4667
2435	=item EV_PROTOTYPES	4668	=item EV_PROTOTYPES (h)
2436		4669
2437	If defined to be C<0>, then F<ev.h> will not define any function	4670	If defined to be C<0>, then F<ev.h> will not define any function
2438	prototypes, but still define all the structs and other symbols. This is	4671	prototypes, but still define all the structs and other symbols. This is
2439	occasionally useful if you want to provide your own wrapper functions	4672	occasionally useful if you want to provide your own wrapper functions
2440	around libev functions.	4673	around libev functions.
…		…
2445	will have the C<struct ev_loop *> as first argument, and you can create	4678	will have the C<struct ev_loop *> as first argument, and you can create
2446	additional independent event loops. Otherwise there will be no support	4679	additional independent event loops. Otherwise there will be no support
2447	for multiple event loops and there is no first event loop pointer	4680	for multiple event loops and there is no first event loop pointer
2448	argument. Instead, all functions act on the single default loop.	4681	argument. Instead, all functions act on the single default loop.
2449		4682
		4683	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4684	default loop when multiplicity is switched off - you always have to
		4685	initialise the loop manually in this case.
		4686
2450	=item EV_MINPRI	4687	=item EV_MINPRI
2451		4688
2452	=item EV_MAXPRI	4689	=item EV_MAXPRI
2453		4690
2454	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to	4691	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
…		…
2459	When doing priority-based operations, libev usually has to linearly search	4696	When doing priority-based operations, libev usually has to linearly search
2460	all the priorities, so having many of them (hundreds) uses a lot of space	4697	all the priorities, so having many of them (hundreds) uses a lot of space
2461	and time, so using the defaults of five priorities (-2 .. +2) is usually	4698	and time, so using the defaults of five priorities (-2 .. +2) is usually
2462	fine.	4699	fine.
2463		4700
2464	If your embedding app does not need any priorities, defining these both to	4701	If your embedding application does not need any priorities, defining these
2465	C<0> will save some memory and cpu.	4702	both to C<0> will save some memory and CPU.
2466		4703
2467	=item EV_PERIODIC_ENABLE	4704	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4705	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4706	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
2468		4707
2469	If undefined or defined to be C<1>, then periodic timers are supported. If	4708	If undefined or defined to be C<1> (and the platform supports it), then
2470	defined to be C<0>, then they are not. Disabling them saves a few kB of	4709	the respective watcher type is supported. If defined to be C<0>, then it
2471	code.	4710	is not. Disabling watcher types mainly saves code size.
2472		4711
2473	=item EV_IDLE_ENABLE	4712	=item EV_FEATURES
2474
2475	If undefined or defined to be C<1>, then idle watchers are supported. If
2476	defined to be C<0>, then they are not. Disabling them saves a few kB of
2477	code.
2478
2479	=item EV_EMBED_ENABLE
2480
2481	If undefined or defined to be C<1>, then embed watchers are supported. If
2482	defined to be C<0>, then they are not.
2483
2484	=item EV_STAT_ENABLE
2485
2486	If undefined or defined to be C<1>, then stat watchers are supported. If
2487	defined to be C<0>, then they are not.
2488
2489	=item EV_FORK_ENABLE
2490
2491	If undefined or defined to be C<1>, then fork watchers are supported. If
2492	defined to be C<0>, then they are not.
2493
2494	=item EV_MINIMAL
2495		4713
2496	If you need to shave off some kilobytes of code at the expense of some	4714	If you need to shave off some kilobytes of code at the expense of some
2497	speed, define this symbol to C<1>. Currently only used for gcc to override	4715	speed (but with the full API), you can define this symbol to request
2498	some inlining decisions, saves roughly 30% codesize of amd64.	4716	certain subsets of functionality. The default is to enable all features
		4717	that can be enabled on the platform.
		4718
		4719	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4720	with some broad features you want) and then selectively re-enable
		4721	additional parts you want, for example if you want everything minimal,
		4722	but multiple event loop support, async and child watchers and the poll
		4723	backend, use this:
		4724
		4725	#define EV_FEATURES 0
		4726	#define EV_MULTIPLICITY 1
		4727	#define EV_USE_POLL 1
		4728	#define EV_CHILD_ENABLE 1
		4729	#define EV_ASYNC_ENABLE 1
		4730
		4731	The actual value is a bitset, it can be a combination of the following
		4732	values (by default, all of these are enabled):
		4733
		4734	=over 4
		4735
		4736	=item C<1> - faster/larger code
		4737
		4738	Use larger code to speed up some operations.
		4739
		4740	Currently this is used to override some inlining decisions (enlarging the
		4741	code size by roughly 30% on amd64).
		4742
		4743	When optimising for size, use of compiler flags such as C<-Os> with
		4744	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4745	assertions.
		4746
		4747	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4748	(e.g. gcc with C<-Os>).
		4749
		4750	=item C<2> - faster/larger data structures
		4751
		4752	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4753	hash table sizes and so on. This will usually further increase code size
		4754	and can additionally have an effect on the size of data structures at
		4755	runtime.
		4756
		4757	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4758	(e.g. gcc with C<-Os>).
		4759
		4760	=item C<4> - full API configuration
		4761
		4762	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4763	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4764
		4765	=item C<8> - full API
		4766
		4767	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4768	details on which parts of the API are still available without this
		4769	feature, and do not complain if this subset changes over time.
		4770
		4771	=item C<16> - enable all optional watcher types
		4772
		4773	Enables all optional watcher types. If you want to selectively enable
		4774	only some watcher types other than I/O and timers (e.g. prepare,
		4775	embed, async, child...) you can enable them manually by defining
		4776	C<EV_watchertype_ENABLE> to C<1> instead.
		4777
		4778	=item C<32> - enable all backends
		4779
		4780	This enables all backends - without this feature, you need to enable at
		4781	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4782
		4783	=item C<64> - enable OS-specific "helper" APIs
		4784
		4785	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4786	default.
		4787
		4788	=back
		4789
		4790	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4791	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4792	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4793	watchers, timers and monotonic clock support.
		4794
		4795	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4796	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4797	your program might be left out as well - a binary starting a timer and an
		4798	I/O watcher then might come out at only 5Kb.
		4799
		4800	=item EV_API_STATIC
		4801
		4802	If this symbol is defined (by default it is not), then all identifiers
		4803	will have static linkage. This means that libev will not export any
		4804	identifiers, and you cannot link against libev anymore. This can be useful
		4805	when you embed libev, only want to use libev functions in a single file,
		4806	and do not want its identifiers to be visible.
		4807
		4808	To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
		4809	wants to use libev.
		4810
		4811	This option only works when libev is compiled with a C compiler, as C++
		4812	doesn't support the required declaration syntax.
		4813
		4814	=item EV_AVOID_STDIO
		4815
		4816	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4817	functions (printf, scanf, perror etc.). This will increase the code size
		4818	somewhat, but if your program doesn't otherwise depend on stdio and your
		4819	libc allows it, this avoids linking in the stdio library which is quite
		4820	big.
		4821
		4822	Note that error messages might become less precise when this option is
		4823	enabled.
		4824
		4825	=item EV_NSIG
		4826
		4827	The highest supported signal number, +1 (or, the number of
		4828	signals): Normally, libev tries to deduce the maximum number of signals
		4829	automatically, but sometimes this fails, in which case it can be
		4830	specified. Also, using a lower number than detected (C<32> should be
		4831	good for about any system in existence) can save some memory, as libev
		4832	statically allocates some 12-24 bytes per signal number.
2499		4833
2500	=item EV_PID_HASHSIZE	4834	=item EV_PID_HASHSIZE
2501		4835
2502	C<ev_child> watchers use a small hash table to distribute workload by	4836	C<ev_child> watchers use a small hash table to distribute workload by
2503	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	4837	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
2504	than enough. If you need to manage thousands of children you might want to	4838	usually more than enough. If you need to manage thousands of children you
2505	increase this value (I<must> be a power of two).	4839	might want to increase this value (I<must> be a power of two).
2506		4840
2507	=item EV_INOTIFY_HASHSIZE	4841	=item EV_INOTIFY_HASHSIZE
2508		4842
2509	C<ev_staz> watchers use a small hash table to distribute workload by	4843	C<ev_stat> watchers use a small hash table to distribute workload by
2510	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	4844	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
2511	usually more than enough. If you need to manage thousands of C<ev_stat>	4845	disabled), usually more than enough. If you need to manage thousands of
2512	watchers you might want to increase this value (I<must> be a power of	4846	C<ev_stat> watchers you might want to increase this value (I<must> be a
2513	two).	4847	power of two).
		4848
		4849	=item EV_USE_4HEAP
		4850
		4851	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4852	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4853	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4854	faster performance with many (thousands) of watchers.
		4855
		4856	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4857	will be C<0>.
		4858
		4859	=item EV_HEAP_CACHE_AT
		4860
		4861	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4862	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4863	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4864	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4865	but avoids random read accesses on heap changes. This improves performance
		4866	noticeably with many (hundreds) of watchers.
		4867
		4868	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4869	will be C<0>.
		4870
		4871	=item EV_VERIFY
		4872
		4873	Controls how much internal verification (see C<ev_verify ()>) will
		4874	be done: If set to C<0>, no internal verification code will be compiled
		4875	in. If set to C<1>, then verification code will be compiled in, but not
		4876	called. If set to C<2>, then the internal verification code will be
		4877	called once per loop, which can slow down libev. If set to C<3>, then the
		4878	verification code will be called very frequently, which will slow down
		4879	libev considerably.
		4880
		4881	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4882	will be C<0>.
2514		4883
2515	=item EV_COMMON	4884	=item EV_COMMON
2516		4885
2517	By default, all watchers have a C<void *data> member. By redefining	4886	By default, all watchers have a C<void *data> member. By redefining
2518	this macro to a something else you can include more and other types of	4887	this macro to something else you can include more and other types of
2519	members. You have to define it each time you include one of the files,	4888	members. You have to define it each time you include one of the files,
2520	though, and it must be identical each time.	4889	though, and it must be identical each time.
2521		4890
2522	For example, the perl EV module uses something like this:	4891	For example, the perl EV module uses something like this:
2523		4892
2524	#define EV_COMMON \	4893	#define EV_COMMON \
2525	SV *self; /* contains this struct */ \	4894	SV *self; /* contains this struct */ \
2526	SV *cb_sv, *fh /* note no trailing ";" */	4895	SV *cb_sv, *fh /* note no trailing ";" */
2527		4896
2528	=item EV_CB_DECLARE (type)	4897	=item EV_CB_DECLARE (type)
2529		4898
2530	=item EV_CB_INVOKE (watcher, revents)	4899	=item EV_CB_INVOKE (watcher, revents)
2531		4900
…		…
2536	definition and a statement, respectively. See the F<ev.h> header file for	4905	definition and a statement, respectively. See the F<ev.h> header file for
2537	their default definitions. One possible use for overriding these is to	4906	their default definitions. One possible use for overriding these is to
2538	avoid the C<struct ev_loop *> as first argument in all cases, or to use	4907	avoid the C<struct ev_loop *> as first argument in all cases, or to use
2539	method calls instead of plain function calls in C++.	4908	method calls instead of plain function calls in C++.
2540		4909
		4910	=back
		4911
2541	=head2 EXPORTED API SYMBOLS	4912	=head2 EXPORTED API SYMBOLS
2542		4913
2543	If you need to re-export the API (e.g. via a dll) and you need a list of	4914	If you need to re-export the API (e.g. via a DLL) and you need a list of
2544	exported symbols, you can use the provided F<Symbol.*> files which list	4915	exported symbols, you can use the provided F<Symbol.*> files which list
2545	all public symbols, one per line:	4916	all public symbols, one per line:
2546		4917
2547	Symbols.ev for libev proper	4918	Symbols.ev for libev proper
2548	Symbols.event for the libevent emulation	4919	Symbols.event for the libevent emulation
2549		4920
2550	This can also be used to rename all public symbols to avoid clashes with	4921	This can also be used to rename all public symbols to avoid clashes with
2551	multiple versions of libev linked together (which is obviously bad in	4922	multiple versions of libev linked together (which is obviously bad in
2552	itself, but sometimes it is inconvinient to avoid this).	4923	itself, but sometimes it is inconvenient to avoid this).
2553		4924
2554	A sed command like this will create wrapper C<#define>'s that you need to	4925	A sed command like this will create wrapper C<#define>'s that you need to
2555	include before including F<ev.h>:	4926	include before including F<ev.h>:
2556		4927
2557	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h	4928	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
…		…
2574	file.	4945	file.
2575		4946
2576	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	4947	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
2577	that everybody includes and which overrides some configure choices:	4948	that everybody includes and which overrides some configure choices:
2578		4949
2579	#define EV_MINIMAL 1	4950	#define EV_FEATURES 8
2580	#define EV_USE_POLL 0	4951	#define EV_USE_SELECT 1
2581	#define EV_MULTIPLICITY 0
2582	#define EV_PERIODIC_ENABLE 0	4952	#define EV_PREPARE_ENABLE 1
		4953	#define EV_IDLE_ENABLE 1
2583	#define EV_STAT_ENABLE 0	4954	#define EV_SIGNAL_ENABLE 1
2584	#define EV_FORK_ENABLE 0	4955	#define EV_CHILD_ENABLE 1
		4956	#define EV_USE_STDEXCEPT 0
2585	#define EV_CONFIG_H <config.h>	4957	#define EV_CONFIG_H <config.h>
2586	#define EV_MINPRI 0
2587	#define EV_MAXPRI 0
2588		4958
2589	#include "ev++.h"	4959	#include "ev++.h"
2590		4960
2591	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	4961	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
2592		4962
2593	#include "ev_cpp.h"	4963	#include "ev_cpp.h"
2594	#include "ev.c"	4964	#include "ev.c"
2595		4965
		4966	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
2596		4967
		4968	=head2 THREADS AND COROUTINES
		4969
		4970	=head3 THREADS
		4971
		4972	All libev functions are reentrant and thread-safe unless explicitly
		4973	documented otherwise, but libev implements no locking itself. This means
		4974	that you can use as many loops as you want in parallel, as long as there
		4975	are no concurrent calls into any libev function with the same loop
		4976	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4977	of course): libev guarantees that different event loops share no data
		4978	structures that need any locking.
		4979
		4980	Or to put it differently: calls with different loop parameters can be done
		4981	concurrently from multiple threads, calls with the same loop parameter
		4982	must be done serially (but can be done from different threads, as long as
		4983	only one thread ever is inside a call at any point in time, e.g. by using
		4984	a mutex per loop).
		4985
		4986	Specifically to support threads (and signal handlers), libev implements
		4987	so-called C<ev_async> watchers, which allow some limited form of
		4988	concurrency on the same event loop, namely waking it up "from the
		4989	outside".
		4990
		4991	If you want to know which design (one loop, locking, or multiple loops
		4992	without or something else still) is best for your problem, then I cannot
		4993	help you, but here is some generic advice:
		4994
		4995	=over 4
		4996
		4997	=item * most applications have a main thread: use the default libev loop
		4998	in that thread, or create a separate thread running only the default loop.
		4999
		5000	This helps integrating other libraries or software modules that use libev
		5001	themselves and don't care/know about threading.
		5002
		5003	=item * one loop per thread is usually a good model.
		5004
		5005	Doing this is almost never wrong, sometimes a better-performance model
		5006	exists, but it is always a good start.
		5007
		5008	=item * other models exist, such as the leader/follower pattern, where one
		5009	loop is handed through multiple threads in a kind of round-robin fashion.
		5010
		5011	Choosing a model is hard - look around, learn, know that usually you can do
		5012	better than you currently do :-)
		5013
		5014	=item * often you need to talk to some other thread which blocks in the
		5015	event loop.
		5016
		5017	C<ev_async> watchers can be used to wake them up from other threads safely
		5018	(or from signal contexts...).
		5019
		5020	An example use would be to communicate signals or other events that only
		5021	work in the default loop by registering the signal watcher with the
		5022	default loop and triggering an C<ev_async> watcher from the default loop
		5023	watcher callback into the event loop interested in the signal.
		5024
		5025	=back
		5026
		5027	See also L</THREAD LOCKING EXAMPLE>.
		5028
		5029	=head3 COROUTINES
		5030
		5031	Libev is very accommodating to coroutines ("cooperative threads"):
		5032	libev fully supports nesting calls to its functions from different
		5033	coroutines (e.g. you can call C<ev_run> on the same loop from two
		5034	different coroutines, and switch freely between both coroutines running
		5035	the loop, as long as you don't confuse yourself). The only exception is
		5036	that you must not do this from C<ev_periodic> reschedule callbacks.
		5037
		5038	Care has been taken to ensure that libev does not keep local state inside
		5039	C<ev_run>, and other calls do not usually allow for coroutine switches as
		5040	they do not call any callbacks.
		5041
		5042	=head2 COMPILER WARNINGS
		5043
		5044	Depending on your compiler and compiler settings, you might get no or a
		5045	lot of warnings when compiling libev code. Some people are apparently
		5046	scared by this.
		5047
		5048	However, these are unavoidable for many reasons. For one, each compiler
		5049	has different warnings, and each user has different tastes regarding
		5050	warning options. "Warn-free" code therefore cannot be a goal except when
		5051	targeting a specific compiler and compiler-version.
		5052
		5053	Another reason is that some compiler warnings require elaborate
		5054	workarounds, or other changes to the code that make it less clear and less
		5055	maintainable.
		5056
		5057	And of course, some compiler warnings are just plain stupid, or simply
		5058	wrong (because they don't actually warn about the condition their message
		5059	seems to warn about). For example, certain older gcc versions had some
		5060	warnings that resulted in an extreme number of false positives. These have
		5061	been fixed, but some people still insist on making code warn-free with
		5062	such buggy versions.
		5063
		5064	While libev is written to generate as few warnings as possible,
		5065	"warn-free" code is not a goal, and it is recommended not to build libev
		5066	with any compiler warnings enabled unless you are prepared to cope with
		5067	them (e.g. by ignoring them). Remember that warnings are just that:
		5068	warnings, not errors, or proof of bugs.
		5069
		5070
		5071	=head2 VALGRIND
		5072
		5073	Valgrind has a special section here because it is a popular tool that is
		5074	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		5075
		5076	If you think you found a bug (memory leak, uninitialised data access etc.)
		5077	in libev, then check twice: If valgrind reports something like:
		5078
		5079	==2274== definitely lost: 0 bytes in 0 blocks.
		5080	==2274== possibly lost: 0 bytes in 0 blocks.
		5081	==2274== still reachable: 256 bytes in 1 blocks.
		5082
		5083	Then there is no memory leak, just as memory accounted to global variables
		5084	is not a memleak - the memory is still being referenced, and didn't leak.
		5085
		5086	Similarly, under some circumstances, valgrind might report kernel bugs
		5087	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		5088	although an acceptable workaround has been found here), or it might be
		5089	confused.
		5090
		5091	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		5092	make it into some kind of religion.
		5093
		5094	If you are unsure about something, feel free to contact the mailing list
		5095	with the full valgrind report and an explanation on why you think this
		5096	is a bug in libev (best check the archives, too :). However, don't be
		5097	annoyed when you get a brisk "this is no bug" answer and take the chance
		5098	of learning how to interpret valgrind properly.
		5099
		5100	If you need, for some reason, empty reports from valgrind for your project
		5101	I suggest using suppression lists.
		5102
		5103
		5104	=head1 PORTABILITY NOTES
		5105
		5106	=head2 GNU/LINUX 32 BIT LIMITATIONS
		5107
		5108	GNU/Linux is the only common platform that supports 64 bit file/large file
		5109	interfaces but I<disables> them by default.
		5110
		5111	That means that libev compiled in the default environment doesn't support
		5112	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		5113
		5114	Unfortunately, many programs try to work around this GNU/Linux issue
		5115	by enabling the large file API, which makes them incompatible with the
		5116	standard libev compiled for their system.
		5117
		5118	Likewise, libev cannot enable the large file API itself as this would
		5119	suddenly make it incompatible to the default compile time environment,
		5120	i.e. all programs not using special compile switches.
		5121
		5122	=head2 OS/X AND DARWIN BUGS
		5123
		5124	The whole thing is a bug if you ask me - basically any system interface
		5125	you touch is broken, whether it is locales, poll, kqueue or even the
		5126	OpenGL drivers.
		5127
		5128	=head3 C<kqueue> is buggy
		5129
		5130	The kqueue syscall is broken in all known versions - most versions support
		5131	only sockets, many support pipes.
		5132
		5133	Libev tries to work around this by not using C<kqueue> by default on this
		5134	rotten platform, but of course you can still ask for it when creating a
		5135	loop - embedding a socket-only kqueue loop into a select-based one is
		5136	probably going to work well.
		5137
		5138	=head3 C<poll> is buggy
		5139
		5140	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		5141	implementation by something calling C<kqueue> internally around the 10.5.6
		5142	release, so now C<kqueue> I<and> C<poll> are broken.
		5143
		5144	Libev tries to work around this by not using C<poll> by default on
		5145	this rotten platform, but of course you can still ask for it when creating
		5146	a loop.
		5147
		5148	=head3 C<select> is buggy
		5149
		5150	All that's left is C<select>, and of course Apple found a way to fuck this
		5151	one up as well: On OS/X, C<select> actively limits the number of file
		5152	descriptors you can pass in to 1024 - your program suddenly crashes when
		5153	you use more.
		5154
		5155	There is an undocumented "workaround" for this - defining
		5156	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		5157	work on OS/X.
		5158
		5159	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		5160
		5161	=head3 C<errno> reentrancy
		5162
		5163	The default compile environment on Solaris is unfortunately so
		5164	thread-unsafe that you can't even use components/libraries compiled
		5165	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		5166	defined by default. A valid, if stupid, implementation choice.
		5167
		5168	If you want to use libev in threaded environments you have to make sure
		5169	it's compiled with C<_REENTRANT> defined.
		5170
		5171	=head3 Event port backend
		5172
		5173	The scalable event interface for Solaris is called "event
		5174	ports". Unfortunately, this mechanism is very buggy in all major
		5175	releases. If you run into high CPU usage, your program freezes or you get
		5176	a large number of spurious wakeups, make sure you have all the relevant
		5177	and latest kernel patches applied. No, I don't know which ones, but there
		5178	are multiple ones to apply, and afterwards, event ports actually work
		5179	great.
		5180
		5181	If you can't get it to work, you can try running the program by setting
		5182	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		5183	C<select> backends.
		5184
		5185	=head2 AIX POLL BUG
		5186
		5187	AIX unfortunately has a broken C<poll.h> header. Libev works around
		5188	this by trying to avoid the poll backend altogether (i.e. it's not even
		5189	compiled in), which normally isn't a big problem as C<select> works fine
		5190	with large bitsets on AIX, and AIX is dead anyway.
		5191
		5192	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		5193
		5194	=head3 General issues
		5195
		5196	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		5197	requires, and its I/O model is fundamentally incompatible with the POSIX
		5198	model. Libev still offers limited functionality on this platform in
		5199	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		5200	descriptors. This only applies when using Win32 natively, not when using
		5201	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		5202	as every compiler comes with a slightly differently broken/incompatible
		5203	environment.
		5204
		5205	Lifting these limitations would basically require the full
		5206	re-implementation of the I/O system. If you are into this kind of thing,
		5207	then note that glib does exactly that for you in a very portable way (note
		5208	also that glib is the slowest event library known to man).
		5209
		5210	There is no supported compilation method available on windows except
		5211	embedding it into other applications.
		5212
		5213	Sensible signal handling is officially unsupported by Microsoft - libev
		5214	tries its best, but under most conditions, signals will simply not work.
		5215
		5216	Not a libev limitation but worth mentioning: windows apparently doesn't
		5217	accept large writes: instead of resulting in a partial write, windows will
		5218	either accept everything or return C<ENOBUFS> if the buffer is too large,
		5219	so make sure you only write small amounts into your sockets (less than a
		5220	megabyte seems safe, but this apparently depends on the amount of memory
		5221	available).
		5222
		5223	Due to the many, low, and arbitrary limits on the win32 platform and
		5224	the abysmal performance of winsockets, using a large number of sockets
		5225	is not recommended (and not reasonable). If your program needs to use
		5226	more than a hundred or so sockets, then likely it needs to use a totally
		5227	different implementation for windows, as libev offers the POSIX readiness
		5228	notification model, which cannot be implemented efficiently on windows
		5229	(due to Microsoft monopoly games).
		5230
		5231	A typical way to use libev under windows is to embed it (see the embedding
		5232	section for details) and use the following F<evwrap.h> header file instead
		5233	of F<ev.h>:
		5234
		5235	#define EV_STANDALONE /* keeps ev from requiring config.h */
		5236	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		5237
		5238	#include "ev.h"
		5239
		5240	And compile the following F<evwrap.c> file into your project (make sure
		5241	you do I<not> compile the F<ev.c> or any other embedded source files!):
		5242
		5243	#include "evwrap.h"
		5244	#include "ev.c"
		5245
		5246	=head3 The winsocket C<select> function
		5247
		5248	The winsocket C<select> function doesn't follow POSIX in that it
		5249	requires socket I<handles> and not socket I<file descriptors> (it is
		5250	also extremely buggy). This makes select very inefficient, and also
		5251	requires a mapping from file descriptors to socket handles (the Microsoft
		5252	C runtime provides the function C<_open_osfhandle> for this). See the
		5253	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		5254	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		5255
		5256	The configuration for a "naked" win32 using the Microsoft runtime
		5257	libraries and raw winsocket select is:
		5258
		5259	#define EV_USE_SELECT 1
		5260	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		5261
		5262	Note that winsockets handling of fd sets is O(n), so you can easily get a
		5263	complexity in the O(n²) range when using win32.
		5264
		5265	=head3 Limited number of file descriptors
		5266
		5267	Windows has numerous arbitrary (and low) limits on things.
		5268
		5269	Early versions of winsocket's select only supported waiting for a maximum
		5270	of C<64> handles (probably owning to the fact that all windows kernels
		5271	can only wait for C<64> things at the same time internally; Microsoft
		5272	recommends spawning a chain of threads and wait for 63 handles and the
		5273	previous thread in each. Sounds great!).
		5274
		5275	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		5276	to some high number (e.g. C<2048>) before compiling the winsocket select
		5277	call (which might be in libev or elsewhere, for example, perl and many
		5278	other interpreters do their own select emulation on windows).
		5279
		5280	Another limit is the number of file descriptors in the Microsoft runtime
		5281	libraries, which by default is C<64> (there must be a hidden I<64>
		5282	fetish or something like this inside Microsoft). You can increase this
		5283	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		5284	(another arbitrary limit), but is broken in many versions of the Microsoft
		5285	runtime libraries. This might get you to about C<512> or C<2048> sockets
		5286	(depending on windows version and/or the phase of the moon). To get more,
		5287	you need to wrap all I/O functions and provide your own fd management, but
		5288	the cost of calling select (O(n²)) will likely make this unworkable.
		5289
		5290	=head2 PORTABILITY REQUIREMENTS
		5291
		5292	In addition to a working ISO-C implementation and of course the
		5293	backend-specific APIs, libev relies on a few additional extensions:
		5294
		5295	=over 4
		5296
		5297	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		5298	calling conventions regardless of C<ev_watcher_type *>.
		5299
		5300	Libev assumes not only that all watcher pointers have the same internal
		5301	structure (guaranteed by POSIX but not by ISO C for example), but it also
		5302	assumes that the same (machine) code can be used to call any watcher
		5303	callback: The watcher callbacks have different type signatures, but libev
		5304	calls them using an C<ev_watcher *> internally.
		5305
		5306	=item null pointers and integer zero are represented by 0 bytes
		5307
		5308	Libev uses C<memset> to initialise structs and arrays to C<0> bytes, and
		5309	relies on this setting pointers and integers to null.
		5310
		5311	=item pointer accesses must be thread-atomic
		5312
		5313	Accessing a pointer value must be atomic, it must both be readable and
		5314	writable in one piece - this is the case on all current architectures.
		5315
		5316	=item C<sig_atomic_t volatile> must be thread-atomic as well
		5317
		5318	The type C<sig_atomic_t volatile> (or whatever is defined as
		5319	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		5320	threads. This is not part of the specification for C<sig_atomic_t>, but is
		5321	believed to be sufficiently portable.
		5322
		5323	=item C<sigprocmask> must work in a threaded environment
		5324
		5325	Libev uses C<sigprocmask> to temporarily block signals. This is not
		5326	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		5327	pthread implementations will either allow C<sigprocmask> in the "main
		5328	thread" or will block signals process-wide, both behaviours would
		5329	be compatible with libev. Interaction between C<sigprocmask> and
		5330	C<pthread_sigmask> could complicate things, however.
		5331
		5332	The most portable way to handle signals is to block signals in all threads
		5333	except the initial one, and run the signal handling loop in the initial
		5334	thread as well.
		5335
		5336	=item C<long> must be large enough for common memory allocation sizes
		5337
		5338	To improve portability and simplify its API, libev uses C<long> internally
		5339	instead of C<size_t> when allocating its data structures. On non-POSIX
		5340	systems (Microsoft...) this might be unexpectedly low, but is still at
		5341	least 31 bits everywhere, which is enough for hundreds of millions of
		5342	watchers.
		5343
		5344	=item C<double> must hold a time value in seconds with enough accuracy
		5345
		5346	The type C<double> is used to represent timestamps. It is required to
		5347	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		5348	good enough for at least into the year 4000 with millisecond accuracy
		5349	(the design goal for libev). This requirement is overfulfilled by
		5350	implementations using IEEE 754, which is basically all existing ones.
		5351
		5352	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5353	year 2255 (and millisecond accuracy till the year 287396 - by then, libev
		5354	is either obsolete or somebody patched it to use C<long double> or
		5355	something like that, just kidding).
		5356
		5357	=back
		5358
		5359	If you know of other additional requirements drop me a note.
		5360
		5361
2597	=head1 COMPLEXITIES	5362	=head1 ALGORITHMIC COMPLEXITIES
2598		5363
2599	In this section the complexities of (many of) the algorithms used inside	5364	In this section the complexities of (many of) the algorithms used inside
2600	libev will be explained. For complexity discussions about backends see the	5365	libev will be documented. For complexity discussions about backends see
2601	documentation for C<ev_default_init>.	5366	the documentation for C<ev_default_init>.
2602		5367
2603	All of the following are about amortised time: If an array needs to be	5368	All of the following are about amortised time: If an array needs to be
2604	extended, libev needs to realloc and move the whole array, but this	5369	extended, libev needs to realloc and move the whole array, but this
2605	happens asymptotically never with higher number of elements, so O(1) might	5370	happens asymptotically rarer with higher number of elements, so O(1) might
2606	mean it might do a lengthy realloc operation in rare cases, but on average	5371	mean that libev does a lengthy realloc operation in rare cases, but on
2607	it is much faster and asymptotically approaches constant time.	5372	average it is much faster and asymptotically approaches constant time.
2608		5373
2609	=over 4	5374	=over 4
2610		5375
2611	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	5376	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2612		5377
2613	This means that, when you have a watcher that triggers in one hour and	5378	This means that, when you have a watcher that triggers in one hour and
2614	there are 100 watchers that would trigger before that then inserting will	5379	there are 100 watchers that would trigger before that, then inserting will
2615	have to skip those 100 watchers.	5380	have to skip roughly seven (C<ld 100>) of these watchers.
2616		5381
2617	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	5382	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2618		5383
2619	That means that for changing a timer costs less than removing/adding them	5384	That means that changing a timer costs less than removing/adding them,
2620	as only the relative motion in the event queue has to be paid for.	5385	as only the relative motion in the event queue has to be paid for.
2621		5386
2622	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	5387	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2623		5388
2624	These just add the watcher into an array or at the head of a list.	5389	These just add the watcher into an array or at the head of a list.
		5390
2625	=item Stopping check/prepare/idle watchers: O(1)	5391	=item Stopping check/prepare/idle/fork/async watchers: O(1)
2626		5392
2627	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	5393	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2628		5394
2629	These watchers are stored in lists then need to be walked to find the	5395	These watchers are stored in lists, so they need to be walked to find the
2630	correct watcher to remove. The lists are usually short (you don't usually	5396	correct watcher to remove. The lists are usually short (you don't usually
2631	have many watchers waiting for the same fd or signal).	5397	have many watchers waiting for the same fd or signal: one is typical, two
		5398	is rare).
2632		5399
2633	=item Finding the next timer per loop iteration: O(1)	5400	=item Finding the next timer in each loop iteration: O(1)
		5401
		5402	By virtue of using a binary or 4-heap, the next timer is always found at a
		5403	fixed position in the storage array.
2634		5404
2635	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	5405	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2636		5406
2637	A change means an I/O watcher gets started or stopped, which requires	5407	A change means an I/O watcher gets started or stopped, which requires
2638	libev to recalculate its status (and possibly tell the kernel).	5408	libev to recalculate its status (and possibly tell the kernel, depending
		5409	on backend and whether C<ev_io_set> was used).
2639		5410
2640	=item Activating one watcher: O(1)	5411	=item Activating one watcher (putting it into the pending state): O(1)
2641		5412
2642	=item Priority handling: O(number_of_priorities)	5413	=item Priority handling: O(number_of_priorities)
2643		5414
2644	Priorities are implemented by allocating some space for each	5415	Priorities are implemented by allocating some space for each
2645	priority. When doing priority-based operations, libev usually has to	5416	priority. When doing priority-based operations, libev usually has to
2646	linearly search all the priorities.	5417	linearly search all the priorities, but starting/stopping and activating
		5418	watchers becomes O(1) with respect to priority handling.
		5419
		5420	=item Sending an ev_async: O(1)
		5421
		5422	=item Processing ev_async_send: O(number_of_async_watchers)
		5423
		5424	=item Processing signals: O(max_signal_number)
		5425
		5426	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5427	calls in the current loop iteration and the loop is currently
		5428	blocked. Checking for async and signal events involves iterating over all
		5429	running async watchers or all signal numbers.
2647		5430
2648	=back	5431	=back
2649		5432
2650		5433
		5434	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5435
		5436	The major version 4 introduced some incompatible changes to the API.
		5437
		5438	At the moment, the C<ev.h> header file provides compatibility definitions
		5439	for all changes, so most programs should still compile. The compatibility
		5440	layer might be removed in later versions of libev, so better update to the
		5441	new API early than late.
		5442
		5443	=over 4
		5444
		5445	=item C<EV_COMPAT3> backwards compatibility mechanism
		5446
		5447	The backward compatibility mechanism can be controlled by
		5448	C<EV_COMPAT3>. See L</"PREPROCESSOR SYMBOLS/MACROS"> in the L</EMBEDDING>
		5449	section.
		5450
		5451	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5452
		5453	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5454
		5455	ev_loop_destroy (EV_DEFAULT_UC);
		5456	ev_loop_fork (EV_DEFAULT);
		5457
		5458	=item function/symbol renames
		5459
		5460	A number of functions and symbols have been renamed:
		5461
		5462	ev_loop => ev_run
		5463	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5464	EVLOOP_ONESHOT => EVRUN_ONCE
		5465
		5466	ev_unloop => ev_break
		5467	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5468	EVUNLOOP_ONE => EVBREAK_ONE
		5469	EVUNLOOP_ALL => EVBREAK_ALL
		5470
		5471	EV_TIMEOUT => EV_TIMER
		5472
		5473	ev_loop_count => ev_iteration
		5474	ev_loop_depth => ev_depth
		5475	ev_loop_verify => ev_verify
		5476
		5477	Most functions working on C<struct ev_loop> objects don't have an
		5478	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5479	associated constants have been renamed to not collide with the C<struct
		5480	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5481	as all other watcher types. Note that C<ev_loop_fork> is still called
		5482	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5483	typedef.
		5484
		5485	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5486
		5487	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5488	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5489	and work, but the library code will of course be larger.
		5490
		5491	=back
		5492
		5493
		5494	=head1 GLOSSARY
		5495
		5496	=over 4
		5497
		5498	=item active
		5499
		5500	A watcher is active as long as it has been started and not yet stopped.
		5501	See L</WATCHER STATES> for details.
		5502
		5503	=item application
		5504
		5505	In this document, an application is whatever is using libev.
		5506
		5507	=item backend
		5508
		5509	The part of the code dealing with the operating system interfaces.
		5510
		5511	=item callback
		5512
		5513	The address of a function that is called when some event has been
		5514	detected. Callbacks are being passed the event loop, the watcher that
		5515	received the event, and the actual event bitset.
		5516
		5517	=item callback/watcher invocation
		5518
		5519	The act of calling the callback associated with a watcher.
		5520
		5521	=item event
		5522
		5523	A change of state of some external event, such as data now being available
		5524	for reading on a file descriptor, time having passed or simply not having
		5525	any other events happening anymore.
		5526
		5527	In libev, events are represented as single bits (such as C<EV_READ> or
		5528	C<EV_TIMER>).
		5529
		5530	=item event library
		5531
		5532	A software package implementing an event model and loop.
		5533
		5534	=item event loop
		5535
		5536	An entity that handles and processes external events and converts them
		5537	into callback invocations.
		5538
		5539	=item event model
		5540
		5541	The model used to describe how an event loop handles and processes
		5542	watchers and events.
		5543
		5544	=item pending
		5545
		5546	A watcher is pending as soon as the corresponding event has been
		5547	detected. See L</WATCHER STATES> for details.
		5548
		5549	=item real time
		5550
		5551	The physical time that is observed. It is apparently strictly monotonic :)
		5552
		5553	=item wall-clock time
		5554
		5555	The time and date as shown on clocks. Unlike real time, it can actually
		5556	be wrong and jump forwards and backwards, e.g. when you adjust your
		5557	clock.
		5558
		5559	=item watcher
		5560
		5561	A data structure that describes interest in certain events. Watchers need
		5562	to be started (attached to an event loop) before they can receive events.
		5563
		5564	=back
		5565
2651	=head1 AUTHOR	5566	=head1 AUTHOR
2652		5567
2653	Marc Lehmann <libev@schmorp.de>.	5568	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5569	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
2654		5570

Diff Legend

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