ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.37 by root, Sat Nov 24 07:20:43 2007 UTC vs.
Revision 1.360 by root, Mon Jan 17 12:11:12 2011 UTC

…		…
2		2
3	libev - a high performance full-featured event loop written in C	3	libev - a high performance full-featured event loop written in C
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	#include <ev.h>	7	#include <ev.h>
8		8
9	=head1 DESCRIPTION	9	=head2 EXAMPLE PROGRAM
		10
		11	// a single header file is required
		12	#include <ev.h>
		13
		14	#include <stdio.h> // for puts
		15
		16	// every watcher type has its own typedef'd struct
		17	// with the name ev_TYPE
		18	ev_io stdin_watcher;
		19	ev_timer timeout_watcher;
		20
		21	// all watcher callbacks have a similar signature
		22	// this callback is called when data is readable on stdin
		23	static void
		24	stdin_cb (EV_P_ ev_io *w, int revents)
		25	{
		26	puts ("stdin ready");
		27	// for one-shot events, one must manually stop the watcher
		28	// with its corresponding stop function.
		29	ev_io_stop (EV_A_ w);
		30
		31	// this causes all nested ev_run's to stop iterating
		32	ev_break (EV_A_ EVBREAK_ALL);
		33	}
		34
		35	// another callback, this time for a time-out
		36	static void
		37	timeout_cb (EV_P_ ev_timer *w, int revents)
		38	{
		39	puts ("timeout");
		40	// this causes the innermost ev_run to stop iterating
		41	ev_break (EV_A_ EVBREAK_ONE);
		42	}
		43
		44	int
		45	main (void)
		46	{
		47	// use the default event loop unless you have special needs
		48	struct ev_loop *loop = EV_DEFAULT;
		49
		50	// initialise an io watcher, then start it
		51	// this one will watch for stdin to become readable
		52	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
		53	ev_io_start (loop, &stdin_watcher);
		54
		55	// initialise a timer watcher, then start it
		56	// simple non-repeating 5.5 second timeout
		57	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		58	ev_timer_start (loop, &timeout_watcher);
		59
		60	// now wait for events to arrive
		61	ev_run (loop, 0);
		62
		63	// unloop was called, so exit
		64	return 0;
		65	}
		66
		67	=head1 ABOUT THIS DOCUMENT
		68
		69	This document documents the libev software package.
		70
		71	The newest version of this document is also available as an html-formatted
		72	web page you might find easier to navigate when reading it for the first
		73	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		74
		75	While this document tries to be as complete as possible in documenting
		76	libev, its usage and the rationale behind its design, it is not a tutorial
		77	on event-based programming, nor will it introduce event-based programming
		78	with libev.
		79
		80	Familiarity with event based programming techniques in general is assumed
		81	throughout this document.
		82
		83	=head1 WHAT TO READ WHEN IN A HURRY
		84
		85	This manual tries to be very detailed, but unfortunately, this also makes
		86	it very long. If you just want to know the basics of libev, I suggest
		87	reading L<ANATOMY OF A WATCHER>, then the L<EXAMPLE PROGRAM> above and
		88	look up the missing functions in L<GLOBAL FUNCTIONS> and the C<ev_io> and
		89	C<ev_timer> sections in L<WATCHER TYPES>.
		90
		91	=head1 ABOUT LIBEV
10		92
11	Libev is an event loop: you register interest in certain events (such as a	93	Libev is an event loop: you register interest in certain events (such as a
12	file descriptor being readable or a timeout occuring), and it will manage	94	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	95	these event sources and provide your program with events.
14		96
15	To do this, it must take more or less complete control over your process	97	To do this, it must take more or less complete control over your process
16	(or thread) by executing the I<event loop> handler, and will then	98	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	99	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	101	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	102	watchers>, which are relatively small C structures you initialise with the
21	details of the event, and then hand it over to libev by I<starting> the	103	details of the event, and then hand it over to libev by I<starting> the
22	watcher.	104	watcher.
23		105
24	=head1 FEATURES	106	=head2 FEATURES
25		107
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	108	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27	kqueue mechanisms for file descriptor events, relative timers, absolute	109	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28	timers with customised rescheduling, signal events, process status change	110	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
		111	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
		112	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
		113	timers (C<ev_timer>), absolute timers with customised rescheduling
		114	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
29	events (related to SIGCHLD), and event watchers dealing with the event	115	change events (C<ev_child>), and event watchers dealing with the event
30	loop mechanism itself (idle, prepare and check watchers). It also is quite	116	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
		117	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		118	limited support for fork events (C<ev_fork>).
		119
		120	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	121	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	122	for example).
33		123
34	=head1 CONVENTIONS	124	=head2 CONVENTIONS
35		125
36	Libev is very configurable. In this manual the default configuration	126	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	127	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	128	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	129	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	130	for multiple event loops, then all functions taking an initial argument of
41	argument of name C<loop> (which is always of type C<struct ev_loop *>)	131	name C<loop> (which is always of type C<struct ev_loop *>) will not have
42	will not have this argument.	132	this argument.
43		133
44	=head1 TIME REPRESENTATION	134	=head2 TIME REPRESENTATION
45		135
46	Libev represents time as a single floating point number, representing the	136	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	137	the (fractional) number of seconds since the (POSIX) epoch (in practice
48	the beginning of 1970, details are complicated, don't ask). This type is	138	somewhere near the beginning of 1970, details are complicated, don't
49	called C<ev_tstamp>, which is what you should use too. It usually aliases	139	ask). This type is called C<ev_tstamp>, which is what you should use
50	to the C<double> type in C, and when you need to do any calculations on	140	too. It usually aliases to the C<double> type in C. When you need to do
51	it, you should treat it as such.	141	any calculations on it, you should treat it as some floating point value.
		142
		143	Unlike the name component C<stamp> might indicate, it is also used for
		144	time differences (e.g. delays) throughout libev.
		145
		146	=head1 ERROR HANDLING
		147
		148	Libev knows three classes of errors: operating system errors, usage errors
		149	and internal errors (bugs).
		150
		151	When libev catches an operating system error it cannot handle (for example
		152	a system call indicating a condition libev cannot fix), it calls the callback
		153	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		154	abort. The default is to print a diagnostic message and to call C<abort
		155	()>.
		156
		157	When libev detects a usage error such as a negative timer interval, then
		158	it will print a diagnostic message and abort (via the C<assert> mechanism,
		159	so C<NDEBUG> will disable this checking): these are programming errors in
		160	the libev caller and need to be fixed there.
		161
		162	Libev also has a few internal error-checking C<assert>ions, and also has
		163	extensive consistency checking code. These do not trigger under normal
		164	circumstances, as they indicate either a bug in libev or worse.
52		165
53		166
54	=head1 GLOBAL FUNCTIONS	167	=head1 GLOBAL FUNCTIONS
55		168
56	These functions can be called anytime, even before initialising the	169	These functions can be called anytime, even before initialising the
…		…
60		173
61	=item ev_tstamp ev_time ()	174	=item ev_tstamp ev_time ()
62		175
63	Returns the current time as libev would use it. Please note that the	176	Returns the current time as libev would use it. Please note that the
64	C<ev_now> function is usually faster and also often returns the timestamp	177	C<ev_now> function is usually faster and also often returns the timestamp
65	you actually want to know.	178	you actually want to know. Also interesting is the combination of
		179	C<ev_update_now> and C<ev_now>.
		180
		181	=item ev_sleep (ev_tstamp interval)
		182
		183	Sleep for the given interval: The current thread will be blocked until
		184	either it is interrupted or the given time interval has passed. Basically
		185	this is a sub-second-resolution C<sleep ()>.
66		186
67	=item int ev_version_major ()	187	=item int ev_version_major ()
68		188
69	=item int ev_version_minor ()	189	=item int ev_version_minor ()
70		190
71	You can find out the major and minor version numbers of the library	191	You can find out the major and minor ABI version numbers of the library
72	you linked against by calling the functions C<ev_version_major> and	192	you linked against by calling the functions C<ev_version_major> and
73	C<ev_version_minor>. If you want, you can compare against the global	193	C<ev_version_minor>. If you want, you can compare against the global
74	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	194	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
75	version of the library your program was compiled against.	195	version of the library your program was compiled against.
76		196
		197	These version numbers refer to the ABI version of the library, not the
		198	release version.
		199
77	Usually, it's a good idea to terminate if the major versions mismatch,	200	Usually, it's a good idea to terminate if the major versions mismatch,
78	as this indicates an incompatible change. Minor versions are usually	201	as this indicates an incompatible change. Minor versions are usually
79	compatible to older versions, so a larger minor version alone is usually	202	compatible to older versions, so a larger minor version alone is usually
80	not a problem.	203	not a problem.
81		204
82	Example: make sure we haven't accidentally been linked against the wrong	205	Example: Make sure we haven't accidentally been linked against the wrong
83	version:	206	version (note, however, that this will not detect other ABI mismatches,
		207	such as LFS or reentrancy).
84		208
85	assert (("libev version mismatch",	209	assert (("libev version mismatch",
86	ev_version_major () == EV_VERSION_MAJOR	210	ev_version_major () == EV_VERSION_MAJOR
87	&& ev_version_minor () >= EV_VERSION_MINOR));	211	&& ev_version_minor () >= EV_VERSION_MINOR));
88		212
89	=item unsigned int ev_supported_backends ()	213	=item unsigned int ev_supported_backends ()
90		214
91	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	215	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
92	value) compiled into this binary of libev (independent of their	216	value) compiled into this binary of libev (independent of their
…		…
94	a description of the set values.	218	a description of the set values.
95		219
96	Example: make sure we have the epoll method, because yeah this is cool and	220	Example: make sure we have the epoll method, because yeah this is cool and
97	a must have and can we have a torrent of it please!!!11	221	a must have and can we have a torrent of it please!!!11
98		222
99	assert (("sorry, no epoll, no sex",	223	assert (("sorry, no epoll, no sex",
100	ev_supported_backends () & EVBACKEND_EPOLL));	224	ev_supported_backends () & EVBACKEND_EPOLL));
101		225
102	=item unsigned int ev_recommended_backends ()	226	=item unsigned int ev_recommended_backends ()
103		227
104	Return the set of all backends compiled into this binary of libev and also	228	Return the set of all backends compiled into this binary of libev and
105	recommended for this platform. This set is often smaller than the one	229	also recommended for this platform, meaning it will work for most file
		230	descriptor types. This set is often smaller than the one returned by
106	returned by C<ev_supported_backends>, as for example kqueue is broken on	231	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
107	most BSDs and will not be autodetected unless you explicitly request it	232	and will not be auto-detected unless you explicitly request it (assuming
108	(assuming you know what you are doing). This is the set of backends that	233	you know what you are doing). This is the set of backends that libev will
109	libev will probe for if you specify no backends explicitly.	234	probe for if you specify no backends explicitly.
110		235
111	=item unsigned int ev_embeddable_backends ()	236	=item unsigned int ev_embeddable_backends ()
112		237
113	Returns the set of backends that are embeddable in other event loops. This	238	Returns the set of backends that are embeddable in other event loops. This
114	is the theoretical, all-platform, value. To find which backends	239	value is platform-specific but can include backends not available on the
115	might be supported on the current system, you would need to look at	240	current system. To find which embeddable backends might be supported on
116	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	241	the current system, you would need to look at C<ev_embeddable_backends ()
117	recommended ones.	242	& ev_supported_backends ()>, likewise for recommended ones.
118		243
119	See the description of C<ev_embed> watchers for more info.	244	See the description of C<ev_embed> watchers for more info.
120		245
121	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	246	=item ev_set_allocator (void *(*cb)(void *ptr, long size))
122		247
123	Sets the allocation function to use (the prototype is similar to the	248	Sets the allocation function to use (the prototype is similar - the
124	realloc C function, the semantics are identical). It is used to allocate	249	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
125	and free memory (no surprises here). If it returns zero when memory	250	used to allocate and free memory (no surprises here). If it returns zero
126	needs to be allocated, the library might abort or take some potentially	251	when memory needs to be allocated (C<size != 0>), the library might abort
127	destructive action. The default is your system realloc function.	252	or take some potentially destructive action.
		253
		254	Since some systems (at least OpenBSD and Darwin) fail to implement
		255	correct C<realloc> semantics, libev will use a wrapper around the system
		256	C<realloc> and C<free> functions by default.
128		257
129	You could override this function in high-availability programs to, say,	258	You could override this function in high-availability programs to, say,
130	free some memory if it cannot allocate memory, to use a special allocator,	259	free some memory if it cannot allocate memory, to use a special allocator,
131	or even to sleep a while and retry until some memory is available.	260	or even to sleep a while and retry until some memory is available.
132		261
133	Example: replace the libev allocator with one that waits a bit and then	262	Example: Replace the libev allocator with one that waits a bit and then
134	retries: better than mine).	263	retries (example requires a standards-compliant C<realloc>).
135		264
136	static void *	265	static void *
137	persistent_realloc (void *ptr, long size)	266	persistent_realloc (void *ptr, size_t size)
138	{	267	{
139	for (;;)	268	for (;;)
140	{	269	{
141	void *newptr = realloc (ptr, size);	270	void *newptr = realloc (ptr, size);
142		271
…		…
148	}	277	}
149		278
150	...	279	...
151	ev_set_allocator (persistent_realloc);	280	ev_set_allocator (persistent_realloc);
152		281
153	=item ev_set_syserr_cb (void (*cb)(const char *msg));	282	=item ev_set_syserr_cb (void (*cb)(const char *msg))
154		283
155	Set the callback function to call on a retryable syscall error (such	284	Set the callback function to call on a retryable system call error (such
156	as failed select, poll, epoll_wait). The message is a printable string	285	as failed select, poll, epoll_wait). The message is a printable string
157	indicating the system call or subsystem causing the problem. If this	286	indicating the system call or subsystem causing the problem. If this
158	callback is set, then libev will expect it to remedy the sitution, no	287	callback is set, then libev will expect it to remedy the situation, no
159	matter what, when it returns. That is, libev will generally retry the	288	matter what, when it returns. That is, libev will generally retry the
160	requested operation, or, if the condition doesn't go away, do bad stuff	289	requested operation, or, if the condition doesn't go away, do bad stuff
161	(such as abort).	290	(such as abort).
162		291
163	Example: do the same thing as libev does internally:	292	Example: This is basically the same thing that libev does internally, too.
164		293
165	static void	294	static void
166	fatal_error (const char *msg)	295	fatal_error (const char *msg)
167	{	296	{
168	perror (msg);	297	perror (msg);
…		…
170	}	299	}
171		300
172	...	301	...
173	ev_set_syserr_cb (fatal_error);	302	ev_set_syserr_cb (fatal_error);
174		303
		304	=item ev_feed_signal (int signum)
		305
		306	This function can be used to "simulate" a signal receive. It is completely
		307	safe to call this function at any time, from any context, including signal
		308	handlers or random threads.
		309
		310	Its main use is to customise signal handling in your process, especially
		311	in the presence of threads. For example, you could block signals
		312	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		313	creating any loops), and in one thread, use C<sigwait> or any other
		314	mechanism to wait for signals, then "deliver" them to libev by calling
		315	C<ev_feed_signal>.
		316
175	=back	317	=back
176		318
177	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	319	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
178		320
179	An event loop is described by a C<struct ev_loop *>. The library knows two	321	An event loop is described by a C<struct ev_loop *> (the C<struct> is
180	types of such loops, the I<default> loop, which supports signals and child	322	I<not> optional in this case unless libev 3 compatibility is disabled, as
181	events, and dynamically created loops which do not.	323	libev 3 had an C<ev_loop> function colliding with the struct name).
182		324
183	If you use threads, a common model is to run the default event loop	325	The library knows two types of such loops, the I<default> loop, which
184	in your main thread (or in a separate thread) and for each thread you	326	supports child process events, and dynamically created event loops which
185	create, you also create another event loop. Libev itself does no locking	327	do not.
186	whatsoever, so if you mix calls to the same event loop in different
187	threads, make sure you lock (this is usually a bad idea, though, even if
188	done correctly, because it's hideous and inefficient).
189		328
190	=over 4	329	=over 4
191		330
192	=item struct ev_loop *ev_default_loop (unsigned int flags)	331	=item struct ev_loop *ev_default_loop (unsigned int flags)
193		332
194	This will initialise the default event loop if it hasn't been initialised	333	This returns the "default" event loop object, which is what you should
195	yet and return it. If the default loop could not be initialised, returns	334	normally use when you just need "the event loop". Event loop objects and
196	false. If it already was initialised it simply returns it (and ignores the	335	the C<flags> parameter are described in more detail in the entry for
197	flags. If that is troubling you, check C<ev_backend ()> afterwards).	336	C<ev_loop_new>.
		337
		338	If the default loop is already initialised then this function simply
		339	returns it (and ignores the flags. If that is troubling you, check
		340	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		341	flags, which should almost always be C<0>, unless the caller is also the
		342	one calling C<ev_run> or otherwise qualifies as "the main program".
198		343
199	If you don't know what event loop to use, use the one returned from this	344	If you don't know what event loop to use, use the one returned from this
200	function.	345	function (or via the C<EV_DEFAULT> macro).
		346
		347	Note that this function is I<not> thread-safe, so if you want to use it
		348	from multiple threads, you have to employ some kind of mutex (note also
		349	that this case is unlikely, as loops cannot be shared easily between
		350	threads anyway).
		351
		352	The default loop is the only loop that can handle C<ev_child> watchers,
		353	and to do this, it always registers a handler for C<SIGCHLD>. If this is
		354	a problem for your application you can either create a dynamic loop with
		355	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
		356	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
		357
		358	Example: This is the most typical usage.
		359
		360	if (!ev_default_loop (0))
		361	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		362
		363	Example: Restrict libev to the select and poll backends, and do not allow
		364	environment settings to be taken into account:
		365
		366	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		367
		368	=item struct ev_loop *ev_loop_new (unsigned int flags)
		369
		370	This will create and initialise a new event loop object. If the loop
		371	could not be initialised, returns false.
		372
		373	This function is thread-safe, and one common way to use libev with
		374	threads is indeed to create one loop per thread, and using the default
		375	loop in the "main" or "initial" thread.
201		376
202	The flags argument can be used to specify special behaviour or specific	377	The flags argument can be used to specify special behaviour or specific
203	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	378	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
204		379
205	The following flags are supported:	380	The following flags are supported:
…		…
211	The default flags value. Use this if you have no clue (it's the right	386	The default flags value. Use this if you have no clue (it's the right
212	thing, believe me).	387	thing, believe me).
213		388
214	=item C<EVFLAG_NOENV>	389	=item C<EVFLAG_NOENV>
215		390
216	If this flag bit is ored into the flag value (or the program runs setuid	391	If this flag bit is or'ed into the flag value (or the program runs setuid
217	or setgid) then libev will I<not> look at the environment variable	392	or setgid) then libev will I<not> look at the environment variable
218	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	393	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
219	override the flags completely if it is found in the environment. This is	394	override the flags completely if it is found in the environment. This is
220	useful to try out specific backends to test their performance, or to work	395	useful to try out specific backends to test their performance, or to work
221	around bugs.	396	around bugs.
222		397
		398	=item C<EVFLAG_FORKCHECK>
		399
		400	Instead of calling C<ev_loop_fork> manually after a fork, you can also
		401	make libev check for a fork in each iteration by enabling this flag.
		402
		403	This works by calling C<getpid ()> on every iteration of the loop,
		404	and thus this might slow down your event loop if you do a lot of loop
		405	iterations and little real work, but is usually not noticeable (on my
		406	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		407	without a system call and thus I<very> fast, but my GNU/Linux system also has
		408	C<pthread_atfork> which is even faster).
		409
		410	The big advantage of this flag is that you can forget about fork (and
		411	forget about forgetting to tell libev about forking) when you use this
		412	flag.
		413
		414	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		415	environment variable.
		416
		417	=item C<EVFLAG_NOINOTIFY>
		418
		419	When this flag is specified, then libev will not attempt to use the
		420	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
		421	testing, this flag can be useful to conserve inotify file descriptors, as
		422	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		423
		424	=item C<EVFLAG_SIGNALFD>
		425
		426	When this flag is specified, then libev will attempt to use the
		427	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
		428	delivers signals synchronously, which makes it both faster and might make
		429	it possible to get the queued signal data. It can also simplify signal
		430	handling with threads, as long as you properly block signals in your
		431	threads that are not interested in handling them.
		432
		433	Signalfd will not be used by default as this changes your signal mask, and
		434	there are a lot of shoddy libraries and programs (glib's threadpool for
		435	example) that can't properly initialise their signal masks.
		436
		437	=item C<EVFLAG_NOSIGMASK>
		438
		439	When this flag is specified, then libev will avoid to modify the signal
		440	mask. Specifically, this means you ahve to make sure signals are unblocked
		441	when you want to receive them.
		442
		443	This behaviour is useful when you want to do your own signal handling, or
		444	want to handle signals only in specific threads and want to avoid libev
		445	unblocking the signals.
		446
		447	It's also required by POSIX in a threaded program, as libev calls
		448	C<sigprocmask>, whose behaviour is officially unspecified.
		449
		450	This flag's behaviour will become the default in future versions of libev.
		451
223	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	452	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
224		453
225	This is your standard select(2) backend. Not I<completely> standard, as	454	This is your standard select(2) backend. Not I<completely> standard, as
226	libev tries to roll its own fd_set with no limits on the number of fds,	455	libev tries to roll its own fd_set with no limits on the number of fds,
227	but if that fails, expect a fairly low limit on the number of fds when	456	but if that fails, expect a fairly low limit on the number of fds when
228	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	457	using this backend. It doesn't scale too well (O(highest_fd)), but its
229	the fastest backend for a low number of fds.	458	usually the fastest backend for a low number of (low-numbered :) fds.
		459
		460	To get good performance out of this backend you need a high amount of
		461	parallelism (most of the file descriptors should be busy). If you are
		462	writing a server, you should C<accept ()> in a loop to accept as many
		463	connections as possible during one iteration. You might also want to have
		464	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		465	readiness notifications you get per iteration.
		466
		467	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		468	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		469	C<exceptfds> set on that platform).
230		470
231	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	471	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
232		472
233	And this is your standard poll(2) backend. It's more complicated than	473	And this is your standard poll(2) backend. It's more complicated
234	select, but handles sparse fds better and has no artificial limit on the	474	than select, but handles sparse fds better and has no artificial
235	number of fds you can use (except it will slow down considerably with a	475	limit on the number of fds you can use (except it will slow down
236	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	476	considerably with a lot of inactive fds). It scales similarly to select,
		477	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		478	performance tips.
		479
		480	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		481	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
237		482
238	=item C<EVBACKEND_EPOLL> (value 4, Linux)	483	=item C<EVBACKEND_EPOLL> (value 4, Linux)
239		484
		485	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		486	kernels).
		487
240	For few fds, this backend is a bit little slower than poll and select,	488	For few fds, this backend is a bit little slower than poll and select,
241	but it scales phenomenally better. While poll and select usually scale like	489	but it scales phenomenally better. While poll and select usually scale
242	O(total_fds) where n is the total number of fds (or the highest fd), epoll scales	490	like O(total_fds) where n is the total number of fds (or the highest fd),
243	either O(1) or O(active_fds).	491	epoll scales either O(1) or O(active_fds).
244		492
		493	The epoll mechanism deserves honorable mention as the most misdesigned
		494	of the more advanced event mechanisms: mere annoyances include silently
		495	dropping file descriptors, requiring a system call per change per file
		496	descriptor (and unnecessary guessing of parameters), problems with dup,
		497	returning before the timeout value, resulting in additional iterations
		498	(and only giving 5ms accuracy while select on the same platform gives
		499	0.1ms) and so on. The biggest issue is fork races, however - if a program
		500	forks then I<both> parent and child process have to recreate the epoll
		501	set, which can take considerable time (one syscall per file descriptor)
		502	and is of course hard to detect.
		503
		504	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		505	of course I<doesn't>, and epoll just loves to report events for totally
		506	I<different> file descriptors (even already closed ones, so one cannot
		507	even remove them from the set) than registered in the set (especially
		508	on SMP systems). Libev tries to counter these spurious notifications by
		509	employing an additional generation counter and comparing that against the
		510	events to filter out spurious ones, recreating the set when required. Last
		511	not least, it also refuses to work with some file descriptors which work
		512	perfectly fine with C<select> (files, many character devices...).
		513
		514	Epoll is truly the train wreck analog among event poll mechanisms,
		515	a frankenpoll, cobbled together in a hurry, no thought to design or
		516	interaction with others.
		517
245	While stopping and starting an I/O watcher in the same iteration will	518	While stopping, setting and starting an I/O watcher in the same iteration
246	result in some caching, there is still a syscall per such incident	519	will result in some caching, there is still a system call per such
247	(because the fd could point to a different file description now), so its	520	incident (because the same I<file descriptor> could point to a different
248	best to avoid that. Also, dup()ed file descriptors might not work very	521	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
249	well if you register events for both fds.	522	file descriptors might not work very well if you register events for both
		523	file descriptors.
250		524
251	Please note that epoll sometimes generates spurious notifications, so you	525	Best performance from this backend is achieved by not unregistering all
252	need to use non-blocking I/O or other means to avoid blocking when no data	526	watchers for a file descriptor until it has been closed, if possible,
253	(or space) is available.	527	i.e. keep at least one watcher active per fd at all times. Stopping and
		528	starting a watcher (without re-setting it) also usually doesn't cause
		529	extra overhead. A fork can both result in spurious notifications as well
		530	as in libev having to destroy and recreate the epoll object, which can
		531	take considerable time and thus should be avoided.
		532
		533	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		534	faster than epoll for maybe up to a hundred file descriptors, depending on
		535	the usage. So sad.
		536
		537	While nominally embeddable in other event loops, this feature is broken in
		538	all kernel versions tested so far.
		539
		540	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		541	C<EVBACKEND_POLL>.
254		542
255	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	543	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
256		544
257	Kqueue deserves special mention, as at the time of this writing, it	545	Kqueue deserves special mention, as at the time of this writing, it
258	was broken on all BSDs except NetBSD (usually it doesn't work with	546	was broken on all BSDs except NetBSD (usually it doesn't work reliably
259	anything but sockets and pipes, except on Darwin, where of course its	547	with anything but sockets and pipes, except on Darwin, where of course
260	completely useless). For this reason its not being "autodetected"	548	it's completely useless). Unlike epoll, however, whose brokenness
		549	is by design, these kqueue bugs can (and eventually will) be fixed
		550	without API changes to existing programs. For this reason it's not being
261	unless you explicitly specify it explicitly in the flags (i.e. using	551	"auto-detected" unless you explicitly specify it in the flags (i.e. using
262	C<EVBACKEND_KQUEUE>).	552	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		553	system like NetBSD.
		554
		555	You still can embed kqueue into a normal poll or select backend and use it
		556	only for sockets (after having made sure that sockets work with kqueue on
		557	the target platform). See C<ev_embed> watchers for more info.
263		558
264	It scales in the same way as the epoll backend, but the interface to the	559	It scales in the same way as the epoll backend, but the interface to the
265	kernel is more efficient (which says nothing about its actual speed, of	560	kernel is more efficient (which says nothing about its actual speed, of
266	course). While starting and stopping an I/O watcher does not cause an	561	course). While stopping, setting and starting an I/O watcher does never
267	extra syscall as with epoll, it still adds up to four event changes per	562	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
268	incident, so its best to avoid that.	563	two event changes per incident. Support for C<fork ()> is very bad (but
		564	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		565	cases
		566
		567	This backend usually performs well under most conditions.
		568
		569	While nominally embeddable in other event loops, this doesn't work
		570	everywhere, so you might need to test for this. And since it is broken
		571	almost everywhere, you should only use it when you have a lot of sockets
		572	(for which it usually works), by embedding it into another event loop
		573	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		574	also broken on OS X)) and, did I mention it, using it only for sockets.
		575
		576	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		577	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		578	C<NOTE_EOF>.
269		579
270	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	580	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
271		581
272	This is not implemented yet (and might never be).	582	This is not implemented yet (and might never be, unless you send me an
		583	implementation). According to reports, C</dev/poll> only supports sockets
		584	and is not embeddable, which would limit the usefulness of this backend
		585	immensely.
273		586
274	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	587	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
275		588
276	This uses the Solaris 10 port mechanism. As with everything on Solaris,	589	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
277	it's really slow, but it still scales very well (O(active_fds)).	590	it's really slow, but it still scales very well (O(active_fds)).
278		591
279	Please note that solaris ports can result in a lot of spurious	592	While this backend scales well, it requires one system call per active
280	notifications, so you need to use non-blocking I/O or other means to avoid	593	file descriptor per loop iteration. For small and medium numbers of file
281	blocking when no data (or space) is available.	594	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		595	might perform better.
		596
		597	On the positive side, this backend actually performed fully to
		598	specification in all tests and is fully embeddable, which is a rare feat
		599	among the OS-specific backends (I vastly prefer correctness over speed
		600	hacks).
		601
		602	On the negative side, the interface is I<bizarre> - so bizarre that
		603	even sun itself gets it wrong in their code examples: The event polling
		604	function sometimes returning events to the caller even though an error
		605	occurred, but with no indication whether it has done so or not (yes, it's
		606	even documented that way) - deadly for edge-triggered interfaces where
		607	you absolutely have to know whether an event occurred or not because you
		608	have to re-arm the watcher.
		609
		610	Fortunately libev seems to be able to work around these idiocies.
		611
		612	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		613	C<EVBACKEND_POLL>.
282		614
283	=item C<EVBACKEND_ALL>	615	=item C<EVBACKEND_ALL>
284		616
285	Try all backends (even potentially broken ones that wouldn't be tried	617	Try all backends (even potentially broken ones that wouldn't be tried
286	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	618	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
287	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	619	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
288		620
		621	It is definitely not recommended to use this flag, use whatever
		622	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		623	at all.
		624
		625	=item C<EVBACKEND_MASK>
		626
		627	Not a backend at all, but a mask to select all backend bits from a
		628	C<flags> value, in case you want to mask out any backends from a flags
		629	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
		630
289	=back	631	=back
290		632
291	If one or more of these are ored into the flags value, then only these	633	If one or more of the backend flags are or'ed into the flags value,
292	backends will be tried (in the reverse order as given here). If none are	634	then only these backends will be tried (in the reverse order as listed
293	specified, most compiled-in backend will be tried, usually in reverse	635	here). If none are specified, all backends in C<ev_recommended_backends
294	order of their flag values :)	636	()> will be tried.
295		637
296	The most typical usage is like this:
297
298	if (!ev_default_loop (0))
299	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
300
301	Restrict libev to the select and poll backends, and do not allow
302	environment settings to be taken into account:
303
304	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
305
306	Use whatever libev has to offer, but make sure that kqueue is used if
307	available (warning, breaks stuff, best use only with your own private
308	event loop and only if you know the OS supports your types of fds):
309
310	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
311
312	=item struct ev_loop *ev_loop_new (unsigned int flags)
313
314	Similar to C<ev_default_loop>, but always creates a new event loop that is
315	always distinct from the default loop. Unlike the default loop, it cannot
316	handle signal and child watchers, and attempts to do so will be greeted by
317	undefined behaviour (or a failed assertion if assertions are enabled).
318
319	Example: try to create a event loop that uses epoll and nothing else.	638	Example: Try to create a event loop that uses epoll and nothing else.
320		639
321	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	640	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
322	if (!epoller)	641	if (!epoller)
323	fatal ("no epoll found here, maybe it hides under your chair");	642	fatal ("no epoll found here, maybe it hides under your chair");
324		643
		644	Example: Use whatever libev has to offer, but make sure that kqueue is
		645	used if available.
		646
		647	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
		648
325	=item ev_default_destroy ()	649	=item ev_loop_destroy (loop)
326		650
327	Destroys the default loop again (frees all memory and kernel state	651	Destroys an event loop object (frees all memory and kernel state
328	etc.). None of the active event watchers will be stopped in the normal	652	etc.). None of the active event watchers will be stopped in the normal
329	sense, so e.g. C<ev_is_active> might still return true. It is your	653	sense, so e.g. C<ev_is_active> might still return true. It is your
330	responsibility to either stop all watchers cleanly yoursef I<before>	654	responsibility to either stop all watchers cleanly yourself I<before>
331	calling this function, or cope with the fact afterwards (which is usually	655	calling this function, or cope with the fact afterwards (which is usually
332	the easiest thing, youc na just ignore the watchers and/or C<free ()> them	656	the easiest thing, you can just ignore the watchers and/or C<free ()> them
333	for example).	657	for example).
334		658
		659	Note that certain global state, such as signal state (and installed signal
		660	handlers), will not be freed by this function, and related watchers (such
		661	as signal and child watchers) would need to be stopped manually.
		662
		663	This function is normally used on loop objects allocated by
		664	C<ev_loop_new>, but it can also be used on the default loop returned by
		665	C<ev_default_loop>, in which case it is not thread-safe.
		666
		667	Note that it is not advisable to call this function on the default loop
		668	except in the rare occasion where you really need to free its resources.
		669	If you need dynamically allocated loops it is better to use C<ev_loop_new>
		670	and C<ev_loop_destroy>.
		671
335	=item ev_loop_destroy (loop)	672	=item ev_loop_fork (loop)
336		673
337	Like C<ev_default_destroy>, but destroys an event loop created by an	674	This function sets a flag that causes subsequent C<ev_run> iterations to
338	earlier call to C<ev_loop_new>.
339
340	=item ev_default_fork ()
341
342	This function reinitialises the kernel state for backends that have	675	reinitialise the kernel state for backends that have one. Despite the
343	one. Despite the name, you can call it anytime, but it makes most sense	676	name, you can call it anytime, but it makes most sense after forking, in
344	after forking, in either the parent or child process (or both, but that	677	the child process. You I<must> call it (or use C<EVFLAG_FORKCHECK>) in the
345	again makes little sense).	678	child before resuming or calling C<ev_run>.
346		679
347	You I<must> call this function in the child process after forking if and	680	Again, you I<have> to call it on I<any> loop that you want to re-use after
348	only if you want to use the event library in both processes. If you just	681	a fork, I<even if you do not plan to use the loop in the parent>. This is
349	fork+exec, you don't have to call it.	682	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		683	during fork.
		684
		685	On the other hand, you only need to call this function in the child
		686	process if and only if you want to use the event loop in the child. If
		687	you just fork+exec or create a new loop in the child, you don't have to
		688	call it at all (in fact, C<epoll> is so badly broken that it makes a
		689	difference, but libev will usually detect this case on its own and do a
		690	costly reset of the backend).
350		691
351	The function itself is quite fast and it's usually not a problem to call	692	The function itself is quite fast and it's usually not a problem to call
352	it just in case after a fork. To make this easy, the function will fit in	693	it just in case after a fork.
353	quite nicely into a call to C<pthread_atfork>:
354		694
		695	Example: Automate calling C<ev_loop_fork> on the default loop when
		696	using pthreads.
		697
		698	static void
		699	post_fork_child (void)
		700	{
		701	ev_loop_fork (EV_DEFAULT);
		702	}
		703
		704	...
355	pthread_atfork (0, 0, ev_default_fork);	705	pthread_atfork (0, 0, post_fork_child);
356		706
357	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use	707	=item int ev_is_default_loop (loop)
358	without calling this function, so if you force one of those backends you
359	do not need to care.
360		708
361	=item ev_loop_fork (loop)	709	Returns true when the given loop is, in fact, the default loop, and false
		710	otherwise.
362		711
363	Like C<ev_default_fork>, but acts on an event loop created by	712	=item unsigned int ev_iteration (loop)
364	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	713
365	after fork, and how you do this is entirely your own problem.	714	Returns the current iteration count for the event loop, which is identical
		715	to the number of times libev did poll for new events. It starts at C<0>
		716	and happily wraps around with enough iterations.
		717
		718	This value can sometimes be useful as a generation counter of sorts (it
		719	"ticks" the number of loop iterations), as it roughly corresponds with
		720	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		721	prepare and check phases.
		722
		723	=item unsigned int ev_depth (loop)
		724
		725	Returns the number of times C<ev_run> was entered minus the number of
		726	times C<ev_run> was exited normally, in other words, the recursion depth.
		727
		728	Outside C<ev_run>, this number is zero. In a callback, this number is
		729	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		730	in which case it is higher.
		731
		732	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
		733	throwing an exception etc.), doesn't count as "exit" - consider this
		734	as a hint to avoid such ungentleman-like behaviour unless it's really
		735	convenient, in which case it is fully supported.
366		736
367	=item unsigned int ev_backend (loop)	737	=item unsigned int ev_backend (loop)
368		738
369	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	739	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
370	use.	740	use.
…		…
373		743
374	Returns the current "event loop time", which is the time the event loop	744	Returns the current "event loop time", which is the time the event loop
375	received events and started processing them. This timestamp does not	745	received events and started processing them. This timestamp does not
376	change as long as callbacks are being processed, and this is also the base	746	change as long as callbacks are being processed, and this is also the base
377	time used for relative timers. You can treat it as the timestamp of the	747	time used for relative timers. You can treat it as the timestamp of the
378	event occuring (or more correctly, libev finding out about it).	748	event occurring (or more correctly, libev finding out about it).
379		749
		750	=item ev_now_update (loop)
		751
		752	Establishes the current time by querying the kernel, updating the time
		753	returned by C<ev_now ()> in the progress. This is a costly operation and
		754	is usually done automatically within C<ev_run ()>.
		755
		756	This function is rarely useful, but when some event callback runs for a
		757	very long time without entering the event loop, updating libev's idea of
		758	the current time is a good idea.
		759
		760	See also L<The special problem of time updates> in the C<ev_timer> section.
		761
		762	=item ev_suspend (loop)
		763
		764	=item ev_resume (loop)
		765
		766	These two functions suspend and resume an event loop, for use when the
		767	loop is not used for a while and timeouts should not be processed.
		768
		769	A typical use case would be an interactive program such as a game: When
		770	the user presses C<^Z> to suspend the game and resumes it an hour later it
		771	would be best to handle timeouts as if no time had actually passed while
		772	the program was suspended. This can be achieved by calling C<ev_suspend>
		773	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		774	C<ev_resume> directly afterwards to resume timer processing.
		775
		776	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		777	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		778	will be rescheduled (that is, they will lose any events that would have
		779	occurred while suspended).
		780
		781	After calling C<ev_suspend> you B<must not> call I<any> function on the
		782	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		783	without a previous call to C<ev_suspend>.
		784
		785	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		786	event loop time (see C<ev_now_update>).
		787
380	=item ev_loop (loop, int flags)	788	=item ev_run (loop, int flags)
381		789
382	Finally, this is it, the event handler. This function usually is called	790	Finally, this is it, the event handler. This function usually is called
383	after you initialised all your watchers and you want to start handling	791	after you have initialised all your watchers and you want to start
384	events.	792	handling events. It will ask the operating system for any new events, call
		793	the watcher callbacks, an then repeat the whole process indefinitely: This
		794	is why event loops are called I<loops>.
385		795
386	If the flags argument is specified as C<0>, it will not return until	796	If the flags argument is specified as C<0>, it will keep handling events
387	either no event watchers are active anymore or C<ev_unloop> was called.	797	until either no event watchers are active anymore or C<ev_break> was
		798	called.
388		799
389	Please note that an explicit C<ev_unloop> is usually better than	800	Please note that an explicit C<ev_break> is usually better than
390	relying on all watchers to be stopped when deciding when a program has	801	relying on all watchers to be stopped when deciding when a program has
391	finished (especially in interactive programs), but having a program that	802	finished (especially in interactive programs), but having a program
392	automatically loops as long as it has to and no longer by virtue of	803	that automatically loops as long as it has to and no longer by virtue
393	relying on its watchers stopping correctly is a thing of beauty.	804	of relying on its watchers stopping correctly, that is truly a thing of
		805	beauty.
394		806
		807	This function is also I<mostly> exception-safe - you can break out of
		808	a C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		809	exception and so on. This does not decrement the C<ev_depth> value, nor
		810	will it clear any outstanding C<EVBREAK_ONE> breaks.
		811
395	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	812	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
396	those events and any outstanding ones, but will not block your process in	813	those events and any already outstanding ones, but will not wait and
397	case there are no events and will return after one iteration of the loop.	814	block your process in case there are no events and will return after one
		815	iteration of the loop. This is sometimes useful to poll and handle new
		816	events while doing lengthy calculations, to keep the program responsive.
398		817
399	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	818	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
400	neccessary) and will handle those and any outstanding ones. It will block	819	necessary) and will handle those and any already outstanding ones. It
401	your process until at least one new event arrives, and will return after	820	will block your process until at least one new event arrives (which could
402	one iteration of the loop. This is useful if you are waiting for some	821	be an event internal to libev itself, so there is no guarantee that a
403	external event in conjunction with something not expressible using other	822	user-registered callback will be called), and will return after one
		823	iteration of the loop.
		824
		825	This is useful if you are waiting for some external event in conjunction
		826	with something not expressible using other libev watchers (i.e. "roll your
404	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	827	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
405	usually a better approach for this kind of thing.	828	usually a better approach for this kind of thing.
406		829
407	Here are the gory details of what C<ev_loop> does:	830	Here are the gory details of what C<ev_run> does:
408		831
409	* If there are no active watchers (reference count is zero), return.	832	- Increment loop depth.
410	- Queue prepare watchers and then call all outstanding watchers.	833	- Reset the ev_break status.
		834	- Before the first iteration, call any pending watchers.
		835	LOOP:
		836	- If EVFLAG_FORKCHECK was used, check for a fork.
		837	- If a fork was detected (by any means), queue and call all fork watchers.
		838	- Queue and call all prepare watchers.
		839	- If ev_break was called, goto FINISH.
411	- If we have been forked, recreate the kernel state.	840	- If we have been forked, detach and recreate the kernel state
		841	as to not disturb the other process.
412	- Update the kernel state with all outstanding changes.	842	- Update the kernel state with all outstanding changes.
413	- Update the "event loop time".	843	- Update the "event loop time" (ev_now ()).
414	- Calculate for how long to block.	844	- Calculate for how long to sleep or block, if at all
		845	(active idle watchers, EVRUN_NOWAIT or not having
		846	any active watchers at all will result in not sleeping).
		847	- Sleep if the I/O and timer collect interval say so.
		848	- Increment loop iteration counter.
415	- Block the process, waiting for any events.	849	- Block the process, waiting for any events.
416	- Queue all outstanding I/O (fd) events.	850	- Queue all outstanding I/O (fd) events.
417	- Update the "event loop time" and do time jump handling.	851	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
418	- Queue all outstanding timers.	852	- Queue all expired timers.
419	- Queue all outstanding periodics.	853	- Queue all expired periodics.
420	- If no events are pending now, queue all idle watchers.	854	- Queue all idle watchers with priority higher than that of pending events.
421	- Queue all check watchers.	855	- Queue all check watchers.
422	- Call all queued watchers in reverse order (i.e. check watchers first).	856	- Call all queued watchers in reverse order (i.e. check watchers first).
423	Signals and child watchers are implemented as I/O watchers, and will	857	Signals and child watchers are implemented as I/O watchers, and will
424	be handled here by queueing them when their watcher gets executed.	858	be handled here by queueing them when their watcher gets executed.
425	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	859	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
426	were used, return, otherwise continue with step *.	860	were used, or there are no active watchers, goto FINISH, otherwise
		861	continue with step LOOP.
		862	FINISH:
		863	- Reset the ev_break status iff it was EVBREAK_ONE.
		864	- Decrement the loop depth.
		865	- Return.
427		866
428	Example: queue some jobs and then loop until no events are outsanding	867	Example: Queue some jobs and then loop until no events are outstanding
429	anymore.	868	anymore.
430		869
431	... queue jobs here, make sure they register event watchers as long	870	... queue jobs here, make sure they register event watchers as long
432	... as they still have work to do (even an idle watcher will do..)	871	... as they still have work to do (even an idle watcher will do..)
433	ev_loop (my_loop, 0);	872	ev_run (my_loop, 0);
434	... jobs done. yeah!	873	... jobs done or somebody called unloop. yeah!
435		874
436	=item ev_unloop (loop, how)	875	=item ev_break (loop, how)
437		876
438	Can be used to make a call to C<ev_loop> return early (but only after it	877	Can be used to make a call to C<ev_run> return early (but only after it
439	has processed all outstanding events). The C<how> argument must be either	878	has processed all outstanding events). The C<how> argument must be either
440	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	879	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
441	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	880	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
		881
		882	This "break state" will be cleared on the next call to C<ev_run>.
		883
		884	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		885	which case it will have no effect.
442		886
443	=item ev_ref (loop)	887	=item ev_ref (loop)
444		888
445	=item ev_unref (loop)	889	=item ev_unref (loop)
446		890
447	Ref/unref can be used to add or remove a reference count on the event	891	Ref/unref can be used to add or remove a reference count on the event
448	loop: Every watcher keeps one reference, and as long as the reference	892	loop: Every watcher keeps one reference, and as long as the reference
449	count is nonzero, C<ev_loop> will not return on its own. If you have	893	count is nonzero, C<ev_run> will not return on its own.
450	a watcher you never unregister that should not keep C<ev_loop> from	894
451	returning, ev_unref() after starting, and ev_ref() before stopping it. For	895	This is useful when you have a watcher that you never intend to
		896	unregister, but that nevertheless should not keep C<ev_run> from
		897	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		898	before stopping it.
		899
452	example, libev itself uses this for its internal signal pipe: It is not	900	As an example, libev itself uses this for its internal signal pipe: It
453	visible to the libev user and should not keep C<ev_loop> from exiting if	901	is not visible to the libev user and should not keep C<ev_run> from
454	no event watchers registered by it are active. It is also an excellent	902	exiting if no event watchers registered by it are active. It is also an
455	way to do this for generic recurring timers or from within third-party	903	excellent way to do this for generic recurring timers or from within
456	libraries. Just remember to I<unref after start> and I<ref before stop>.	904	third-party libraries. Just remember to I<unref after start> and I<ref
		905	before stop> (but only if the watcher wasn't active before, or was active
		906	before, respectively. Note also that libev might stop watchers itself
		907	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		908	in the callback).
457		909
458	Example: create a signal watcher, but keep it from keeping C<ev_loop>	910	Example: Create a signal watcher, but keep it from keeping C<ev_run>
459	running when nothing else is active.	911	running when nothing else is active.
460		912
461	struct dv_signal exitsig;	913	ev_signal exitsig;
462	ev_signal_init (&exitsig, sig_cb, SIGINT);	914	ev_signal_init (&exitsig, sig_cb, SIGINT);
463	ev_signal_start (myloop, &exitsig);	915	ev_signal_start (loop, &exitsig);
464	evf_unref (myloop);	916	ev_unref (loop);
465		917
466	Example: for some weird reason, unregister the above signal handler again.	918	Example: For some weird reason, unregister the above signal handler again.
467		919
468	ev_ref (myloop);	920	ev_ref (loop);
469	ev_signal_stop (myloop, &exitsig);	921	ev_signal_stop (loop, &exitsig);
		922
		923	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		924
		925	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		926
		927	These advanced functions influence the time that libev will spend waiting
		928	for events. Both time intervals are by default C<0>, meaning that libev
		929	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		930	latency.
		931
		932	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		933	allows libev to delay invocation of I/O and timer/periodic callbacks
		934	to increase efficiency of loop iterations (or to increase power-saving
		935	opportunities).
		936
		937	The idea is that sometimes your program runs just fast enough to handle
		938	one (or very few) event(s) per loop iteration. While this makes the
		939	program responsive, it also wastes a lot of CPU time to poll for new
		940	events, especially with backends like C<select ()> which have a high
		941	overhead for the actual polling but can deliver many events at once.
		942
		943	By setting a higher I<io collect interval> you allow libev to spend more
		944	time collecting I/O events, so you can handle more events per iteration,
		945	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		946	C<ev_timer>) will be not affected. Setting this to a non-null value will
		947	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		948	sleep time ensures that libev will not poll for I/O events more often then
		949	once per this interval, on average.
		950
		951	Likewise, by setting a higher I<timeout collect interval> you allow libev
		952	to spend more time collecting timeouts, at the expense of increased
		953	latency/jitter/inexactness (the watcher callback will be called
		954	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		955	value will not introduce any overhead in libev.
		956
		957	Many (busy) programs can usually benefit by setting the I/O collect
		958	interval to a value near C<0.1> or so, which is often enough for
		959	interactive servers (of course not for games), likewise for timeouts. It
		960	usually doesn't make much sense to set it to a lower value than C<0.01>,
		961	as this approaches the timing granularity of most systems. Note that if
		962	you do transactions with the outside world and you can't increase the
		963	parallelity, then this setting will limit your transaction rate (if you
		964	need to poll once per transaction and the I/O collect interval is 0.01,
		965	then you can't do more than 100 transactions per second).
		966
		967	Setting the I<timeout collect interval> can improve the opportunity for
		968	saving power, as the program will "bundle" timer callback invocations that
		969	are "near" in time together, by delaying some, thus reducing the number of
		970	times the process sleeps and wakes up again. Another useful technique to
		971	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		972	they fire on, say, one-second boundaries only.
		973
		974	Example: we only need 0.1s timeout granularity, and we wish not to poll
		975	more often than 100 times per second:
		976
		977	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		978	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		979
		980	=item ev_invoke_pending (loop)
		981
		982	This call will simply invoke all pending watchers while resetting their
		983	pending state. Normally, C<ev_run> does this automatically when required,
		984	but when overriding the invoke callback this call comes handy. This
		985	function can be invoked from a watcher - this can be useful for example
		986	when you want to do some lengthy calculation and want to pass further
		987	event handling to another thread (you still have to make sure only one
		988	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		989
		990	=item int ev_pending_count (loop)
		991
		992	Returns the number of pending watchers - zero indicates that no watchers
		993	are pending.
		994
		995	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		996
		997	This overrides the invoke pending functionality of the loop: Instead of
		998	invoking all pending watchers when there are any, C<ev_run> will call
		999	this callback instead. This is useful, for example, when you want to
		1000	invoke the actual watchers inside another context (another thread etc.).
		1001
		1002	If you want to reset the callback, use C<ev_invoke_pending> as new
		1003	callback.
		1004
		1005	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		1006
		1007	Sometimes you want to share the same loop between multiple threads. This
		1008	can be done relatively simply by putting mutex_lock/unlock calls around
		1009	each call to a libev function.
		1010
		1011	However, C<ev_run> can run an indefinite time, so it is not feasible
		1012	to wait for it to return. One way around this is to wake up the event
		1013	loop via C<ev_break> and C<av_async_send>, another way is to set these
		1014	I<release> and I<acquire> callbacks on the loop.
		1015
		1016	When set, then C<release> will be called just before the thread is
		1017	suspended waiting for new events, and C<acquire> is called just
		1018	afterwards.
		1019
		1020	Ideally, C<release> will just call your mutex_unlock function, and
		1021	C<acquire> will just call the mutex_lock function again.
		1022
		1023	While event loop modifications are allowed between invocations of
		1024	C<release> and C<acquire> (that's their only purpose after all), no
		1025	modifications done will affect the event loop, i.e. adding watchers will
		1026	have no effect on the set of file descriptors being watched, or the time
		1027	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		1028	to take note of any changes you made.
		1029
		1030	In theory, threads executing C<ev_run> will be async-cancel safe between
		1031	invocations of C<release> and C<acquire>.
		1032
		1033	See also the locking example in the C<THREADS> section later in this
		1034	document.
		1035
		1036	=item ev_set_userdata (loop, void *data)
		1037
		1038	=item void *ev_userdata (loop)
		1039
		1040	Set and retrieve a single C<void *> associated with a loop. When
		1041	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		1042	C<0>.
		1043
		1044	These two functions can be used to associate arbitrary data with a loop,
		1045	and are intended solely for the C<invoke_pending_cb>, C<release> and
		1046	C<acquire> callbacks described above, but of course can be (ab-)used for
		1047	any other purpose as well.
		1048
		1049	=item ev_verify (loop)
		1050
		1051	This function only does something when C<EV_VERIFY> support has been
		1052	compiled in, which is the default for non-minimal builds. It tries to go
		1053	through all internal structures and checks them for validity. If anything
		1054	is found to be inconsistent, it will print an error message to standard
		1055	error and call C<abort ()>.
		1056
		1057	This can be used to catch bugs inside libev itself: under normal
		1058	circumstances, this function will never abort as of course libev keeps its
		1059	data structures consistent.
470		1060
471	=back	1061	=back
472		1062
		1063
473	=head1 ANATOMY OF A WATCHER	1064	=head1 ANATOMY OF A WATCHER
474		1065
		1066	In the following description, uppercase C<TYPE> in names stands for the
		1067	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		1068	watchers and C<ev_io_start> for I/O watchers.
		1069
475	A watcher is a structure that you create and register to record your	1070	A watcher is an opaque structure that you allocate and register to record
476	interest in some event. For instance, if you want to wait for STDIN to	1071	your interest in some event. To make a concrete example, imagine you want
477	become readable, you would create an C<ev_io> watcher for that:	1072	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1073	for that:
478		1074
479	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1075	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
480	{	1076	{
481	ev_io_stop (w);	1077	ev_io_stop (w);
482	ev_unloop (loop, EVUNLOOP_ALL);	1078	ev_break (loop, EVBREAK_ALL);
483	}	1079	}
484		1080
485	struct ev_loop *loop = ev_default_loop (0);	1081	struct ev_loop *loop = ev_default_loop (0);
		1082
486	struct ev_io stdin_watcher;	1083	ev_io stdin_watcher;
		1084
487	ev_init (&stdin_watcher, my_cb);	1085	ev_init (&stdin_watcher, my_cb);
488	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1086	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
489	ev_io_start (loop, &stdin_watcher);	1087	ev_io_start (loop, &stdin_watcher);
		1088
490	ev_loop (loop, 0);	1089	ev_run (loop, 0);
491		1090
492	As you can see, you are responsible for allocating the memory for your	1091	As you can see, you are responsible for allocating the memory for your
493	watcher structures (and it is usually a bad idea to do this on the stack,	1092	watcher structures (and it is I<usually> a bad idea to do this on the
494	although this can sometimes be quite valid).	1093	stack).
495		1094
		1095	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1096	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1097
496	Each watcher structure must be initialised by a call to C<ev_init	1098	Each watcher structure must be initialised by a call to C<ev_init (watcher
497	(watcher *, callback)>, which expects a callback to be provided. This	1099	*, callback)>, which expects a callback to be provided. This callback is
498	callback gets invoked each time the event occurs (or, in the case of io	1100	invoked each time the event occurs (or, in the case of I/O watchers, each
499	watchers, each time the event loop detects that the file descriptor given	1101	time the event loop detects that the file descriptor given is readable
500	is readable and/or writable).	1102	and/or writable).
501		1103
502	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1104	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
503	with arguments specific to this watcher type. There is also a macro	1105	macro to configure it, with arguments specific to the watcher type. There
504	to combine initialisation and setting in one call: C<< ev_<type>_init	1106	is also a macro to combine initialisation and setting in one call: C<<
505	(watcher *, callback, ...) >>.	1107	ev_TYPE_init (watcher *, callback, ...) >>.
506		1108
507	To make the watcher actually watch out for events, you have to start it	1109	To make the watcher actually watch out for events, you have to start it
508	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1110	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
509	*) >>), and you can stop watching for events at any time by calling the	1111	*) >>), and you can stop watching for events at any time by calling the
510	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1112	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
511		1113
512	As long as your watcher is active (has been started but not stopped) you	1114	As long as your watcher is active (has been started but not stopped) you
513	must not touch the values stored in it. Most specifically you must never	1115	must not touch the values stored in it. Most specifically you must never
514	reinitialise it or call its C<set> macro.	1116	reinitialise it or call its C<ev_TYPE_set> macro.
515		1117
516	Each and every callback receives the event loop pointer as first, the	1118	Each and every callback receives the event loop pointer as first, the
517	registered watcher structure as second, and a bitset of received events as	1119	registered watcher structure as second, and a bitset of received events as
518	third argument.	1120	third argument.
519		1121
…		…
528	=item C<EV_WRITE>	1130	=item C<EV_WRITE>
529		1131
530	The file descriptor in the C<ev_io> watcher has become readable and/or	1132	The file descriptor in the C<ev_io> watcher has become readable and/or
531	writable.	1133	writable.
532		1134
533	=item C<EV_TIMEOUT>	1135	=item C<EV_TIMER>
534		1136
535	The C<ev_timer> watcher has timed out.	1137	The C<ev_timer> watcher has timed out.
536		1138
537	=item C<EV_PERIODIC>	1139	=item C<EV_PERIODIC>
538		1140
…		…
544		1146
545	=item C<EV_CHILD>	1147	=item C<EV_CHILD>
546		1148
547	The pid specified in the C<ev_child> watcher has received a status change.	1149	The pid specified in the C<ev_child> watcher has received a status change.
548		1150
		1151	=item C<EV_STAT>
		1152
		1153	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		1154
549	=item C<EV_IDLE>	1155	=item C<EV_IDLE>
550		1156
551	The C<ev_idle> watcher has determined that you have nothing better to do.	1157	The C<ev_idle> watcher has determined that you have nothing better to do.
552		1158
553	=item C<EV_PREPARE>	1159	=item C<EV_PREPARE>
554		1160
555	=item C<EV_CHECK>	1161	=item C<EV_CHECK>
556		1162
557	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1163	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts
558	to gather new events, and all C<ev_check> watchers are invoked just after	1164	to gather new events, and all C<ev_check> watchers are invoked just after
559	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1165	C<ev_run> has gathered them, but before it invokes any callbacks for any
560	received events. Callbacks of both watcher types can start and stop as	1166	received events. Callbacks of both watcher types can start and stop as
561	many watchers as they want, and all of them will be taken into account	1167	many watchers as they want, and all of them will be taken into account
562	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1168	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
563	C<ev_loop> from blocking).	1169	C<ev_run> from blocking).
		1170
		1171	=item C<EV_EMBED>
		1172
		1173	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1174
		1175	=item C<EV_FORK>
		1176
		1177	The event loop has been resumed in the child process after fork (see
		1178	C<ev_fork>).
		1179
		1180	=item C<EV_CLEANUP>
		1181
		1182	The event loop is about to be destroyed (see C<ev_cleanup>).
		1183
		1184	=item C<EV_ASYNC>
		1185
		1186	The given async watcher has been asynchronously notified (see C<ev_async>).
		1187
		1188	=item C<EV_CUSTOM>
		1189
		1190	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1191	by libev users to signal watchers (e.g. via C<ev_feed_event>).
564		1192
565	=item C<EV_ERROR>	1193	=item C<EV_ERROR>
566		1194
567	An unspecified error has occured, the watcher has been stopped. This might	1195	An unspecified error has occurred, the watcher has been stopped. This might
568	happen because the watcher could not be properly started because libev	1196	happen because the watcher could not be properly started because libev
569	ran out of memory, a file descriptor was found to be closed or any other	1197	ran out of memory, a file descriptor was found to be closed or any other
		1198	problem. Libev considers these application bugs.
		1199
570	problem. You best act on it by reporting the problem and somehow coping	1200	You best act on it by reporting the problem and somehow coping with the
571	with the watcher being stopped.	1201	watcher being stopped. Note that well-written programs should not receive
		1202	an error ever, so when your watcher receives it, this usually indicates a
		1203	bug in your program.
572		1204
573	Libev will usually signal a few "dummy" events together with an error,	1205	Libev will usually signal a few "dummy" events together with an error, for
574	for example it might indicate that a fd is readable or writable, and if	1206	example it might indicate that a fd is readable or writable, and if your
575	your callbacks is well-written it can just attempt the operation and cope	1207	callbacks is well-written it can just attempt the operation and cope with
576	with the error from read() or write(). This will not work in multithreaded	1208	the error from read() or write(). This will not work in multi-threaded
577	programs, though, so beware.	1209	programs, though, as the fd could already be closed and reused for another
		1210	thing, so beware.
578		1211
579	=back	1212	=back
580		1213
581	=head2 SUMMARY OF GENERIC WATCHER FUNCTIONS	1214	=head2 GENERIC WATCHER FUNCTIONS
582
583	In the following description, C<TYPE> stands for the watcher type,
584	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
585		1215
586	=over 4	1216	=over 4
587		1217
588	=item C<ev_init> (ev_TYPE *watcher, callback)	1218	=item C<ev_init> (ev_TYPE *watcher, callback)
589		1219
…		…
595	which rolls both calls into one.	1225	which rolls both calls into one.
596		1226
597	You can reinitialise a watcher at any time as long as it has been stopped	1227	You can reinitialise a watcher at any time as long as it has been stopped
598	(or never started) and there are no pending events outstanding.	1228	(or never started) and there are no pending events outstanding.
599		1229
600	The callbakc is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	1230	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
601	int revents)>.	1231	int revents)>.
602		1232
		1233	Example: Initialise an C<ev_io> watcher in two steps.
		1234
		1235	ev_io w;
		1236	ev_init (&w, my_cb);
		1237	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1238
603	=item C<ev_TYPE_set> (ev_TYPE *, [args])	1239	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
604		1240
605	This macro initialises the type-specific parts of a watcher. You need to	1241	This macro initialises the type-specific parts of a watcher. You need to
606	call C<ev_init> at least once before you call this macro, but you can	1242	call C<ev_init> at least once before you call this macro, but you can
607	call C<ev_TYPE_set> any number of times. You must not, however, call this	1243	call C<ev_TYPE_set> any number of times. You must not, however, call this
608	macro on a watcher that is active (it can be pending, however, which is a	1244	macro on a watcher that is active (it can be pending, however, which is a
609	difference to the C<ev_init> macro).	1245	difference to the C<ev_init> macro).
610		1246
611	Although some watcher types do not have type-specific arguments	1247	Although some watcher types do not have type-specific arguments
612	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	1248	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
613		1249
		1250	See C<ev_init>, above, for an example.
		1251
614	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	1252	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
615		1253
616	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	1254	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
617	calls into a single call. This is the most convinient method to initialise	1255	calls into a single call. This is the most convenient method to initialise
618	a watcher. The same limitations apply, of course.	1256	a watcher. The same limitations apply, of course.
619		1257
		1258	Example: Initialise and set an C<ev_io> watcher in one step.
		1259
		1260	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1261
620	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	1262	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
621		1263
622	Starts (activates) the given watcher. Only active watchers will receive	1264	Starts (activates) the given watcher. Only active watchers will receive
623	events. If the watcher is already active nothing will happen.	1265	events. If the watcher is already active nothing will happen.
624		1266
		1267	Example: Start the C<ev_io> watcher that is being abused as example in this
		1268	whole section.
		1269
		1270	ev_io_start (EV_DEFAULT_UC, &w);
		1271
625	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	1272	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
626		1273
627	Stops the given watcher again (if active) and clears the pending	1274	Stops the given watcher if active, and clears the pending status (whether
		1275	the watcher was active or not).
		1276
628	status. It is possible that stopped watchers are pending (for example,	1277	It is possible that stopped watchers are pending - for example,
629	non-repeating timers are being stopped when they become pending), but	1278	non-repeating timers are being stopped when they become pending - but
630	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1279	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
631	you want to free or reuse the memory used by the watcher it is therefore a	1280	pending. If you want to free or reuse the memory used by the watcher it is
632	good idea to always call its C<ev_TYPE_stop> function.	1281	therefore a good idea to always call its C<ev_TYPE_stop> function.
633		1282
634	=item bool ev_is_active (ev_TYPE *watcher)	1283	=item bool ev_is_active (ev_TYPE *watcher)
635		1284
636	Returns a true value iff the watcher is active (i.e. it has been started	1285	Returns a true value iff the watcher is active (i.e. it has been started
637	and not yet been stopped). As long as a watcher is active you must not modify	1286	and not yet been stopped). As long as a watcher is active you must not modify
…		…
640	=item bool ev_is_pending (ev_TYPE *watcher)	1289	=item bool ev_is_pending (ev_TYPE *watcher)
641		1290
642	Returns a true value iff the watcher is pending, (i.e. it has outstanding	1291	Returns a true value iff the watcher is pending, (i.e. it has outstanding
643	events but its callback has not yet been invoked). As long as a watcher	1292	events but its callback has not yet been invoked). As long as a watcher
644	is pending (but not active) you must not call an init function on it (but	1293	is pending (but not active) you must not call an init function on it (but
645	C<ev_TYPE_set> is safe) and you must make sure the watcher is available to	1294	C<ev_TYPE_set> is safe), you must not change its priority, and you must
646	libev (e.g. you cnanot C<free ()> it).	1295	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1296	it).
647		1297
648	=item callback = ev_cb (ev_TYPE *watcher)	1298	=item callback ev_cb (ev_TYPE *watcher)
649		1299
650	Returns the callback currently set on the watcher.	1300	Returns the callback currently set on the watcher.
651		1301
652	=item ev_cb_set (ev_TYPE *watcher, callback)	1302	=item ev_cb_set (ev_TYPE *watcher, callback)
653		1303
654	Change the callback. You can change the callback at virtually any time	1304	Change the callback. You can change the callback at virtually any time
655	(modulo threads).	1305	(modulo threads).
656		1306
		1307	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1308
		1309	=item int ev_priority (ev_TYPE *watcher)
		1310
		1311	Set and query the priority of the watcher. The priority is a small
		1312	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1313	(default: C<-2>). Pending watchers with higher priority will be invoked
		1314	before watchers with lower priority, but priority will not keep watchers
		1315	from being executed (except for C<ev_idle> watchers).
		1316
		1317	If you need to suppress invocation when higher priority events are pending
		1318	you need to look at C<ev_idle> watchers, which provide this functionality.
		1319
		1320	You I<must not> change the priority of a watcher as long as it is active or
		1321	pending.
		1322
		1323	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1324	fine, as long as you do not mind that the priority value you query might
		1325	or might not have been clamped to the valid range.
		1326
		1327	The default priority used by watchers when no priority has been set is
		1328	always C<0>, which is supposed to not be too high and not be too low :).
		1329
		1330	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1331	priorities.
		1332
		1333	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1334
		1335	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1336	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1337	can deal with that fact, as both are simply passed through to the
		1338	callback.
		1339
		1340	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1341
		1342	If the watcher is pending, this function clears its pending status and
		1343	returns its C<revents> bitset (as if its callback was invoked). If the
		1344	watcher isn't pending it does nothing and returns C<0>.
		1345
		1346	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1347	callback to be invoked, which can be accomplished with this function.
		1348
		1349	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1350
		1351	Feeds the given event set into the event loop, as if the specified event
		1352	had happened for the specified watcher (which must be a pointer to an
		1353	initialised but not necessarily started event watcher). Obviously you must
		1354	not free the watcher as long as it has pending events.
		1355
		1356	Stopping the watcher, letting libev invoke it, or calling
		1357	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1358	not started in the first place.
		1359
		1360	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1361	functions that do not need a watcher.
		1362
657	=back	1363	=back
658		1364
		1365	See also the L<ASSOCIATING CUSTOM DATA WITH A WATCHER> and L<BUILDING YOUR
		1366	OWN COMPOSITE WATCHERS> idioms.
659		1367
660	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1368	=head2 WATCHER STATES
661		1369
662	Each watcher has, by default, a member C<void *data> that you can change	1370	There are various watcher states mentioned throughout this manual -
663	and read at any time, libev will completely ignore it. This can be used	1371	active, pending and so on. In this section these states and the rules to
664	to associate arbitrary data with your watcher. If you need more data and	1372	transition between them will be described in more detail - and while these
665	don't want to allocate memory and store a pointer to it in that data	1373	rules might look complicated, they usually do "the right thing".
666	member, you can also "subclass" the watcher type and provide your own
667	data:
668		1374
669	struct my_io	1375	=over 4
		1376
		1377	=item initialiased
		1378
		1379	Before a watcher can be registered with the event looop it has to be
		1380	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1381	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1382
		1383	In this state it is simply some block of memory that is suitable for use
		1384	in an event loop. It can be moved around, freed, reused etc. at will.
		1385
		1386	=item started/running/active
		1387
		1388	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1389	property of the event loop, and is actively waiting for events. While in
		1390	this state it cannot be accessed (except in a few documented ways), moved,
		1391	freed or anything else - the only legal thing is to keep a pointer to it,
		1392	and call libev functions on it that are documented to work on active watchers.
		1393
		1394	=item pending
		1395
		1396	If a watcher is active and libev determines that an event it is interested
		1397	in has occurred (such as a timer expiring), it will become pending. It will
		1398	stay in this pending state until either it is stopped or its callback is
		1399	about to be invoked, so it is not normally pending inside the watcher
		1400	callback.
		1401
		1402	The watcher might or might not be active while it is pending (for example,
		1403	an expired non-repeating timer can be pending but no longer active). If it
		1404	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1405	but it is still property of the event loop at this time, so cannot be
		1406	moved, freed or reused. And if it is active the rules described in the
		1407	previous item still apply.
		1408
		1409	It is also possible to feed an event on a watcher that is not active (e.g.
		1410	via C<ev_feed_event>), in which case it becomes pending without being
		1411	active.
		1412
		1413	=item stopped
		1414
		1415	A watcher can be stopped implicitly by libev (in which case it might still
		1416	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1417	latter will clear any pending state the watcher might be in, regardless
		1418	of whether it was active or not, so stopping a watcher explicitly before
		1419	freeing it is often a good idea.
		1420
		1421	While stopped (and not pending) the watcher is essentially in the
		1422	initialised state, that is it can be reused, moved, modified in any way
		1423	you wish.
		1424
		1425	=back
		1426
		1427	=head2 WATCHER PRIORITY MODELS
		1428
		1429	Many event loops support I<watcher priorities>, which are usually small
		1430	integers that influence the ordering of event callback invocation
		1431	between watchers in some way, all else being equal.
		1432
		1433	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1434	description for the more technical details such as the actual priority
		1435	range.
		1436
		1437	There are two common ways how these these priorities are being interpreted
		1438	by event loops:
		1439
		1440	In the more common lock-out model, higher priorities "lock out" invocation
		1441	of lower priority watchers, which means as long as higher priority
		1442	watchers receive events, lower priority watchers are not being invoked.
		1443
		1444	The less common only-for-ordering model uses priorities solely to order
		1445	callback invocation within a single event loop iteration: Higher priority
		1446	watchers are invoked before lower priority ones, but they all get invoked
		1447	before polling for new events.
		1448
		1449	Libev uses the second (only-for-ordering) model for all its watchers
		1450	except for idle watchers (which use the lock-out model).
		1451
		1452	The rationale behind this is that implementing the lock-out model for
		1453	watchers is not well supported by most kernel interfaces, and most event
		1454	libraries will just poll for the same events again and again as long as
		1455	their callbacks have not been executed, which is very inefficient in the
		1456	common case of one high-priority watcher locking out a mass of lower
		1457	priority ones.
		1458
		1459	Static (ordering) priorities are most useful when you have two or more
		1460	watchers handling the same resource: a typical usage example is having an
		1461	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1462	timeouts. Under load, data might be received while the program handles
		1463	other jobs, but since timers normally get invoked first, the timeout
		1464	handler will be executed before checking for data. In that case, giving
		1465	the timer a lower priority than the I/O watcher ensures that I/O will be
		1466	handled first even under adverse conditions (which is usually, but not
		1467	always, what you want).
		1468
		1469	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1470	will only be executed when no same or higher priority watchers have
		1471	received events, they can be used to implement the "lock-out" model when
		1472	required.
		1473
		1474	For example, to emulate how many other event libraries handle priorities,
		1475	you can associate an C<ev_idle> watcher to each such watcher, and in
		1476	the normal watcher callback, you just start the idle watcher. The real
		1477	processing is done in the idle watcher callback. This causes libev to
		1478	continuously poll and process kernel event data for the watcher, but when
		1479	the lock-out case is known to be rare (which in turn is rare :), this is
		1480	workable.
		1481
		1482	Usually, however, the lock-out model implemented that way will perform
		1483	miserably under the type of load it was designed to handle. In that case,
		1484	it might be preferable to stop the real watcher before starting the
		1485	idle watcher, so the kernel will not have to process the event in case
		1486	the actual processing will be delayed for considerable time.
		1487
		1488	Here is an example of an I/O watcher that should run at a strictly lower
		1489	priority than the default, and which should only process data when no
		1490	other events are pending:
		1491
		1492	ev_idle idle; // actual processing watcher
		1493	ev_io io; // actual event watcher
		1494
		1495	static void
		1496	io_cb (EV_P_ ev_io *w, int revents)
670	{	1497	{
671	struct ev_io io;	1498	// stop the I/O watcher, we received the event, but
672	int otherfd;	1499	// are not yet ready to handle it.
673	void *somedata;	1500	ev_io_stop (EV_A_ w);
674	struct whatever *mostinteresting;	1501
		1502	// start the idle watcher to handle the actual event.
		1503	// it will not be executed as long as other watchers
		1504	// with the default priority are receiving events.
		1505	ev_idle_start (EV_A_ &idle);
675	}	1506	}
676		1507
677	And since your callback will be called with a pointer to the watcher, you	1508	static void
678	can cast it back to your own type:	1509	idle_cb (EV_P_ ev_idle *w, int revents)
679
680	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
681	{	1510	{
682	struct my_io *w = (struct my_io *)w_;	1511	// actual processing
683	...	1512	read (STDIN_FILENO, ...);
		1513
		1514	// have to start the I/O watcher again, as
		1515	// we have handled the event
		1516	ev_io_start (EV_P_ &io);
684	}	1517	}
685		1518
686	More interesting and less C-conformant ways of catsing your callback type	1519	// initialisation
687	have been omitted....	1520	ev_idle_init (&idle, idle_cb);
		1521	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1522	ev_io_start (EV_DEFAULT_ &io);
		1523
		1524	In the "real" world, it might also be beneficial to start a timer, so that
		1525	low-priority connections can not be locked out forever under load. This
		1526	enables your program to keep a lower latency for important connections
		1527	during short periods of high load, while not completely locking out less
		1528	important ones.
688		1529
689		1530
690	=head1 WATCHER TYPES	1531	=head1 WATCHER TYPES
691		1532
692	This section describes each watcher in detail, but will not repeat	1533	This section describes each watcher in detail, but will not repeat
693	information given in the last section.	1534	information given in the last section. Any initialisation/set macros,
		1535	functions and members specific to the watcher type are explained.
694		1536
		1537	Members are additionally marked with either I<[read-only]>, meaning that,
		1538	while the watcher is active, you can look at the member and expect some
		1539	sensible content, but you must not modify it (you can modify it while the
		1540	watcher is stopped to your hearts content), or I<[read-write]>, which
		1541	means you can expect it to have some sensible content while the watcher
		1542	is active, but you can also modify it. Modifying it may not do something
		1543	sensible or take immediate effect (or do anything at all), but libev will
		1544	not crash or malfunction in any way.
695		1545
		1546
696	=head2 C<ev_io> - is this file descriptor readable or writable	1547	=head2 C<ev_io> - is this file descriptor readable or writable?
697		1548
698	I/O watchers check whether a file descriptor is readable or writable	1549	I/O watchers check whether a file descriptor is readable or writable
699	in each iteration of the event loop (This behaviour is called	1550	in each iteration of the event loop, or, more precisely, when reading
700	level-triggering because you keep receiving events as long as the	1551	would not block the process and writing would at least be able to write
701	condition persists. Remember you can stop the watcher if you don't want to	1552	some data. This behaviour is called level-triggering because you keep
702	act on the event and neither want to receive future events).	1553	receiving events as long as the condition persists. Remember you can stop
		1554	the watcher if you don't want to act on the event and neither want to
		1555	receive future events.
703		1556
704	In general you can register as many read and/or write event watchers per	1557	In general you can register as many read and/or write event watchers per
705	fd as you want (as long as you don't confuse yourself). Setting all file	1558	fd as you want (as long as you don't confuse yourself). Setting all file
706	descriptors to non-blocking mode is also usually a good idea (but not	1559	descriptors to non-blocking mode is also usually a good idea (but not
707	required if you know what you are doing).	1560	required if you know what you are doing).
708		1561
709	You have to be careful with dup'ed file descriptors, though. Some backends	1562	Another thing you have to watch out for is that it is quite easy to
710	(the linux epoll backend is a notable example) cannot handle dup'ed file	1563	receive "spurious" readiness notifications, that is, your callback might
711	descriptors correctly if you register interest in two or more fds pointing	1564	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
712	to the same underlying file/socket etc. description (that is, they share	1565	because there is no data. It is very easy to get into this situation even
713	the same underlying "file open").	1566	with a relatively standard program structure. Thus it is best to always
		1567	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
		1568	preferable to a program hanging until some data arrives.
714		1569
715	If you must do this, then force the use of a known-to-be-good backend	1570	If you cannot run the fd in non-blocking mode (for example you should
716	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	1571	not play around with an Xlib connection), then you have to separately
717	C<EVBACKEND_POLL>).	1572	re-test whether a file descriptor is really ready with a known-to-be good
		1573	interface such as poll (fortunately in the case of Xlib, it already does
		1574	this on its own, so its quite safe to use). Some people additionally
		1575	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1576	indefinitely.
		1577
		1578	But really, best use non-blocking mode.
		1579
		1580	=head3 The special problem of disappearing file descriptors
		1581
		1582	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1583	descriptor (either due to calling C<close> explicitly or any other means,
		1584	such as C<dup2>). The reason is that you register interest in some file
		1585	descriptor, but when it goes away, the operating system will silently drop
		1586	this interest. If another file descriptor with the same number then is
		1587	registered with libev, there is no efficient way to see that this is, in
		1588	fact, a different file descriptor.
		1589
		1590	To avoid having to explicitly tell libev about such cases, libev follows
		1591	the following policy: Each time C<ev_io_set> is being called, libev
		1592	will assume that this is potentially a new file descriptor, otherwise
		1593	it is assumed that the file descriptor stays the same. That means that
		1594	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1595	descriptor even if the file descriptor number itself did not change.
		1596
		1597	This is how one would do it normally anyway, the important point is that
		1598	the libev application should not optimise around libev but should leave
		1599	optimisations to libev.
		1600
		1601	=head3 The special problem of dup'ed file descriptors
		1602
		1603	Some backends (e.g. epoll), cannot register events for file descriptors,
		1604	but only events for the underlying file descriptions. That means when you
		1605	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1606	events for them, only one file descriptor might actually receive events.
		1607
		1608	There is no workaround possible except not registering events
		1609	for potentially C<dup ()>'ed file descriptors, or to resort to
		1610	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1611
		1612	=head3 The special problem of files
		1613
		1614	Many people try to use C<select> (or libev) on file descriptors
		1615	representing files, and expect it to become ready when their program
		1616	doesn't block on disk accesses (which can take a long time on their own).
		1617
		1618	However, this cannot ever work in the "expected" way - you get a readiness
		1619	notification as soon as the kernel knows whether and how much data is
		1620	there, and in the case of open files, that's always the case, so you
		1621	always get a readiness notification instantly, and your read (or possibly
		1622	write) will still block on the disk I/O.
		1623
		1624	Another way to view it is that in the case of sockets, pipes, character
		1625	devices and so on, there is another party (the sender) that delivers data
		1626	on its own, but in the case of files, there is no such thing: the disk
		1627	will not send data on its own, simply because it doesn't know what you
		1628	wish to read - you would first have to request some data.
		1629
		1630	Since files are typically not-so-well supported by advanced notification
		1631	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1632	to files, even though you should not use it. The reason for this is
		1633	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1634	usually a tty, often a pipe, but also sometimes files or special devices
		1635	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1636	F</dev/urandom>), and even though the file might better be served with
		1637	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1638	it "just works" instead of freezing.
		1639
		1640	So avoid file descriptors pointing to files when you know it (e.g. use
		1641	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1642	when you rarely read from a file instead of from a socket, and want to
		1643	reuse the same code path.
		1644
		1645	=head3 The special problem of fork
		1646
		1647	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1648	useless behaviour. Libev fully supports fork, but needs to be told about
		1649	it in the child if you want to continue to use it in the child.
		1650
		1651	To support fork in your child processes, you have to call C<ev_loop_fork
		1652	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
		1653	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1654
		1655	=head3 The special problem of SIGPIPE
		1656
		1657	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1658	when writing to a pipe whose other end has been closed, your program gets
		1659	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1660	this is sensible behaviour, for daemons, this is usually undesirable.
		1661
		1662	So when you encounter spurious, unexplained daemon exits, make sure you
		1663	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1664	somewhere, as that would have given you a big clue).
		1665
		1666	=head3 The special problem of accept()ing when you can't
		1667
		1668	Many implementations of the POSIX C<accept> function (for example,
		1669	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1670	connection from the pending queue in all error cases.
		1671
		1672	For example, larger servers often run out of file descriptors (because
		1673	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1674	rejecting the connection, leading to libev signalling readiness on
		1675	the next iteration again (the connection still exists after all), and
		1676	typically causing the program to loop at 100% CPU usage.
		1677
		1678	Unfortunately, the set of errors that cause this issue differs between
		1679	operating systems, there is usually little the app can do to remedy the
		1680	situation, and no known thread-safe method of removing the connection to
		1681	cope with overload is known (to me).
		1682
		1683	One of the easiest ways to handle this situation is to just ignore it
		1684	- when the program encounters an overload, it will just loop until the
		1685	situation is over. While this is a form of busy waiting, no OS offers an
		1686	event-based way to handle this situation, so it's the best one can do.
		1687
		1688	A better way to handle the situation is to log any errors other than
		1689	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1690	messages, and continue as usual, which at least gives the user an idea of
		1691	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1692	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1693	usage.
		1694
		1695	If your program is single-threaded, then you could also keep a dummy file
		1696	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1697	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1698	close that fd, and create a new dummy fd. This will gracefully refuse
		1699	clients under typical overload conditions.
		1700
		1701	The last way to handle it is to simply log the error and C<exit>, as
		1702	is often done with C<malloc> failures, but this results in an easy
		1703	opportunity for a DoS attack.
		1704
		1705	=head3 Watcher-Specific Functions
718		1706
719	=over 4	1707	=over 4
720		1708
721	=item ev_io_init (ev_io *, callback, int fd, int events)	1709	=item ev_io_init (ev_io *, callback, int fd, int events)
722		1710
723	=item ev_io_set (ev_io *, int fd, int events)	1711	=item ev_io_set (ev_io *, int fd, int events)
724		1712
725	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1713	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
726	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1714	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
727	EV_WRITE> to receive the given events.	1715	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
728		1716
729	Please note that most of the more scalable backend mechanisms (for example	1717	=item int fd [read-only]
730	epoll and solaris ports) can result in spurious readyness notifications	1718
731	for file descriptors, so you practically need to use non-blocking I/O (and	1719	The file descriptor being watched.
732	treat callback invocation as hint only), or retest separately with a safe	1720
733	interface before doing I/O (XLib can do this), or force the use of either	1721	=item int events [read-only]
734	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>, which don't suffer from this	1722
735	problem. Also note that it is quite easy to have your callback invoked	1723	The events being watched.
736	when the readyness condition is no longer valid even when employing
737	typical ways of handling events, so its a good idea to use non-blocking
738	I/O unconditionally.
739		1724
740	=back	1725	=back
741		1726
		1727	=head3 Examples
		1728
742	Example: call C<stdin_readable_cb> when STDIN_FILENO has become, well	1729	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
743	readable, but only once. Since it is likely line-buffered, you could	1730	readable, but only once. Since it is likely line-buffered, you could
744	attempt to read a whole line in the callback:	1731	attempt to read a whole line in the callback.
745		1732
746	static void	1733	static void
747	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1734	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
748	{	1735	{
749	ev_io_stop (loop, w);	1736	ev_io_stop (loop, w);
750	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1737	.. read from stdin here (or from w->fd) and handle any I/O errors
751	}	1738	}
752		1739
753	...	1740	...
754	struct ev_loop *loop = ev_default_init (0);	1741	struct ev_loop *loop = ev_default_init (0);
755	struct ev_io stdin_readable;	1742	ev_io stdin_readable;
756	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1743	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
757	ev_io_start (loop, &stdin_readable);	1744	ev_io_start (loop, &stdin_readable);
758	ev_loop (loop, 0);	1745	ev_run (loop, 0);
759		1746
760		1747
761	=head2 C<ev_timer> - relative and optionally recurring timeouts	1748	=head2 C<ev_timer> - relative and optionally repeating timeouts
762		1749
763	Timer watchers are simple relative timers that generate an event after a	1750	Timer watchers are simple relative timers that generate an event after a
764	given time, and optionally repeating in regular intervals after that.	1751	given time, and optionally repeating in regular intervals after that.
765		1752
766	The timers are based on real time, that is, if you register an event that	1753	The timers are based on real time, that is, if you register an event that
767	times out after an hour and you reset your system clock to last years	1754	times out after an hour and you reset your system clock to January last
768	time, it will still time out after (roughly) and hour. "Roughly" because	1755	year, it will still time out after (roughly) one hour. "Roughly" because
769	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1756	detecting time jumps is hard, and some inaccuracies are unavoidable (the
770	monotonic clock option helps a lot here).	1757	monotonic clock option helps a lot here).
		1758
		1759	The callback is guaranteed to be invoked only I<after> its timeout has
		1760	passed (not I<at>, so on systems with very low-resolution clocks this
		1761	might introduce a small delay). If multiple timers become ready during the
		1762	same loop iteration then the ones with earlier time-out values are invoked
		1763	before ones of the same priority with later time-out values (but this is
		1764	no longer true when a callback calls C<ev_run> recursively).
		1765
		1766	=head3 Be smart about timeouts
		1767
		1768	Many real-world problems involve some kind of timeout, usually for error
		1769	recovery. A typical example is an HTTP request - if the other side hangs,
		1770	you want to raise some error after a while.
		1771
		1772	What follows are some ways to handle this problem, from obvious and
		1773	inefficient to smart and efficient.
		1774
		1775	In the following, a 60 second activity timeout is assumed - a timeout that
		1776	gets reset to 60 seconds each time there is activity (e.g. each time some
		1777	data or other life sign was received).
		1778
		1779	=over 4
		1780
		1781	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1782
		1783	This is the most obvious, but not the most simple way: In the beginning,
		1784	start the watcher:
		1785
		1786	ev_timer_init (timer, callback, 60., 0.);
		1787	ev_timer_start (loop, timer);
		1788
		1789	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1790	and start it again:
		1791
		1792	ev_timer_stop (loop, timer);
		1793	ev_timer_set (timer, 60., 0.);
		1794	ev_timer_start (loop, timer);
		1795
		1796	This is relatively simple to implement, but means that each time there is
		1797	some activity, libev will first have to remove the timer from its internal
		1798	data structure and then add it again. Libev tries to be fast, but it's
		1799	still not a constant-time operation.
		1800
		1801	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1802
		1803	This is the easiest way, and involves using C<ev_timer_again> instead of
		1804	C<ev_timer_start>.
		1805
		1806	To implement this, configure an C<ev_timer> with a C<repeat> value
		1807	of C<60> and then call C<ev_timer_again> at start and each time you
		1808	successfully read or write some data. If you go into an idle state where
		1809	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1810	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1811
		1812	That means you can ignore both the C<ev_timer_start> function and the
		1813	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1814	member and C<ev_timer_again>.
		1815
		1816	At start:
		1817
		1818	ev_init (timer, callback);
		1819	timer->repeat = 60.;
		1820	ev_timer_again (loop, timer);
		1821
		1822	Each time there is some activity:
		1823
		1824	ev_timer_again (loop, timer);
		1825
		1826	It is even possible to change the time-out on the fly, regardless of
		1827	whether the watcher is active or not:
		1828
		1829	timer->repeat = 30.;
		1830	ev_timer_again (loop, timer);
		1831
		1832	This is slightly more efficient then stopping/starting the timer each time
		1833	you want to modify its timeout value, as libev does not have to completely
		1834	remove and re-insert the timer from/into its internal data structure.
		1835
		1836	It is, however, even simpler than the "obvious" way to do it.
		1837
		1838	=item 3. Let the timer time out, but then re-arm it as required.
		1839
		1840	This method is more tricky, but usually most efficient: Most timeouts are
		1841	relatively long compared to the intervals between other activity - in
		1842	our example, within 60 seconds, there are usually many I/O events with
		1843	associated activity resets.
		1844
		1845	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1846	but remember the time of last activity, and check for a real timeout only
		1847	within the callback:
		1848
		1849	ev_tstamp last_activity; // time of last activity
		1850
		1851	static void
		1852	callback (EV_P_ ev_timer *w, int revents)
		1853	{
		1854	ev_tstamp now = ev_now (EV_A);
		1855	ev_tstamp timeout = last_activity + 60.;
		1856
		1857	// if last_activity + 60. is older than now, we did time out
		1858	if (timeout < now)
		1859	{
		1860	// timeout occurred, take action
		1861	}
		1862	else
		1863	{
		1864	// callback was invoked, but there was some activity, re-arm
		1865	// the watcher to fire in last_activity + 60, which is
		1866	// guaranteed to be in the future, so "again" is positive:
		1867	w->repeat = timeout - now;
		1868	ev_timer_again (EV_A_ w);
		1869	}
		1870	}
		1871
		1872	To summarise the callback: first calculate the real timeout (defined
		1873	as "60 seconds after the last activity"), then check if that time has
		1874	been reached, which means something I<did>, in fact, time out. Otherwise
		1875	the callback was invoked too early (C<timeout> is in the future), so
		1876	re-schedule the timer to fire at that future time, to see if maybe we have
		1877	a timeout then.
		1878
		1879	Note how C<ev_timer_again> is used, taking advantage of the
		1880	C<ev_timer_again> optimisation when the timer is already running.
		1881
		1882	This scheme causes more callback invocations (about one every 60 seconds
		1883	minus half the average time between activity), but virtually no calls to
		1884	libev to change the timeout.
		1885
		1886	To start the timer, simply initialise the watcher and set C<last_activity>
		1887	to the current time (meaning we just have some activity :), then call the
		1888	callback, which will "do the right thing" and start the timer:
		1889
		1890	ev_init (timer, callback);
		1891	last_activity = ev_now (loop);
		1892	callback (loop, timer, EV_TIMER);
		1893
		1894	And when there is some activity, simply store the current time in
		1895	C<last_activity>, no libev calls at all:
		1896
		1897	last_activity = ev_now (loop);
		1898
		1899	This technique is slightly more complex, but in most cases where the
		1900	time-out is unlikely to be triggered, much more efficient.
		1901
		1902	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1903	callback :) - just change the timeout and invoke the callback, which will
		1904	fix things for you.
		1905
		1906	=item 4. Wee, just use a double-linked list for your timeouts.
		1907
		1908	If there is not one request, but many thousands (millions...), all
		1909	employing some kind of timeout with the same timeout value, then one can
		1910	do even better:
		1911
		1912	When starting the timeout, calculate the timeout value and put the timeout
		1913	at the I<end> of the list.
		1914
		1915	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1916	the list is expected to fire (for example, using the technique #3).
		1917
		1918	When there is some activity, remove the timer from the list, recalculate
		1919	the timeout, append it to the end of the list again, and make sure to
		1920	update the C<ev_timer> if it was taken from the beginning of the list.
		1921
		1922	This way, one can manage an unlimited number of timeouts in O(1) time for
		1923	starting, stopping and updating the timers, at the expense of a major
		1924	complication, and having to use a constant timeout. The constant timeout
		1925	ensures that the list stays sorted.
		1926
		1927	=back
		1928
		1929	So which method the best?
		1930
		1931	Method #2 is a simple no-brain-required solution that is adequate in most
		1932	situations. Method #3 requires a bit more thinking, but handles many cases
		1933	better, and isn't very complicated either. In most case, choosing either
		1934	one is fine, with #3 being better in typical situations.
		1935
		1936	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1937	rather complicated, but extremely efficient, something that really pays
		1938	off after the first million or so of active timers, i.e. it's usually
		1939	overkill :)
		1940
		1941	=head3 The special problem of time updates
		1942
		1943	Establishing the current time is a costly operation (it usually takes at
		1944	least two system calls): EV therefore updates its idea of the current
		1945	time only before and after C<ev_run> collects new events, which causes a
		1946	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1947	lots of events in one iteration.
771		1948
772	The relative timeouts are calculated relative to the C<ev_now ()>	1949	The relative timeouts are calculated relative to the C<ev_now ()>
773	time. This is usually the right thing as this timestamp refers to the time	1950	time. This is usually the right thing as this timestamp refers to the time
774	of the event triggering whatever timeout you are modifying/starting. If	1951	of the event triggering whatever timeout you are modifying/starting. If
775	you suspect event processing to be delayed and you I<need> to base the timeout	1952	you suspect event processing to be delayed and you I<need> to base the
776	on the current time, use something like this to adjust for this:	1953	timeout on the current time, use something like this to adjust for this:
777		1954
778	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1955	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
779		1956
780	The callback is guarenteed to be invoked only when its timeout has passed,	1957	If the event loop is suspended for a long time, you can also force an
781	but if multiple timers become ready during the same loop iteration then	1958	update of the time returned by C<ev_now ()> by calling C<ev_now_update
782	order of execution is undefined.	1959	()>.
		1960
		1961	=head3 The special problems of suspended animation
		1962
		1963	When you leave the server world it is quite customary to hit machines that
		1964	can suspend/hibernate - what happens to the clocks during such a suspend?
		1965
		1966	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		1967	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		1968	to run until the system is suspended, but they will not advance while the
		1969	system is suspended. That means, on resume, it will be as if the program
		1970	was frozen for a few seconds, but the suspend time will not be counted
		1971	towards C<ev_timer> when a monotonic clock source is used. The real time
		1972	clock advanced as expected, but if it is used as sole clocksource, then a
		1973	long suspend would be detected as a time jump by libev, and timers would
		1974	be adjusted accordingly.
		1975
		1976	I would not be surprised to see different behaviour in different between
		1977	operating systems, OS versions or even different hardware.
		1978
		1979	The other form of suspend (job control, or sending a SIGSTOP) will see a
		1980	time jump in the monotonic clocks and the realtime clock. If the program
		1981	is suspended for a very long time, and monotonic clock sources are in use,
		1982	then you can expect C<ev_timer>s to expire as the full suspension time
		1983	will be counted towards the timers. When no monotonic clock source is in
		1984	use, then libev will again assume a timejump and adjust accordingly.
		1985
		1986	It might be beneficial for this latter case to call C<ev_suspend>
		1987	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		1988	deterministic behaviour in this case (you can do nothing against
		1989	C<SIGSTOP>).
		1990
		1991	=head3 Watcher-Specific Functions and Data Members
783		1992
784	=over 4	1993	=over 4
785		1994
786	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1995	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
787		1996
788	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1997	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
789		1998
790	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1999	Configure the timer to trigger after C<after> seconds. If C<repeat>
791	C<0.>, then it will automatically be stopped. If it is positive, then the	2000	is C<0.>, then it will automatically be stopped once the timeout is
792	timer will automatically be configured to trigger again C<repeat> seconds	2001	reached. If it is positive, then the timer will automatically be
793	later, again, and again, until stopped manually.	2002	configured to trigger again C<repeat> seconds later, again, and again,
		2003	until stopped manually.
794		2004
795	The timer itself will do a best-effort at avoiding drift, that is, if you	2005	The timer itself will do a best-effort at avoiding drift, that is, if
796	configure a timer to trigger every 10 seconds, then it will trigger at	2006	you configure a timer to trigger every 10 seconds, then it will normally
797	exactly 10 second intervals. If, however, your program cannot keep up with	2007	trigger at exactly 10 second intervals. If, however, your program cannot
798	the timer (because it takes longer than those 10 seconds to do stuff) the	2008	keep up with the timer (because it takes longer than those 10 seconds to
799	timer will not fire more than once per event loop iteration.	2009	do stuff) the timer will not fire more than once per event loop iteration.
800		2010
801	=item ev_timer_again (loop)	2011	=item ev_timer_again (loop, ev_timer *)
802		2012
803	This will act as if the timer timed out and restart it again if it is	2013	This will act as if the timer timed out and restart it again if it is
804	repeating. The exact semantics are:	2014	repeating. The exact semantics are:
805		2015
		2016	If the timer is pending, its pending status is cleared.
		2017
806	If the timer is started but nonrepeating, stop it.	2018	If the timer is started but non-repeating, stop it (as if it timed out).
807		2019
808	If the timer is repeating, either start it if necessary (with the repeat	2020	If the timer is repeating, either start it if necessary (with the
809	value), or reset the running timer to the repeat value.	2021	C<repeat> value), or reset the running timer to the C<repeat> value.
810		2022
811	This sounds a bit complicated, but here is a useful and typical	2023	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
812	example: Imagine you have a tcp connection and you want a so-called idle	2024	usage example.
813	timeout, that is, you want to be called when there have been, say, 60	2025
814	seconds of inactivity on the socket. The easiest way to do this is to	2026	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
815	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each	2027
816	time you successfully read or write some data. If you go into an idle	2028	Returns the remaining time until a timer fires. If the timer is active,
817	state where you do not expect data to travel on the socket, you can stop	2029	then this time is relative to the current event loop time, otherwise it's
818	the timer, and again will automatically restart it if need be.	2030	the timeout value currently configured.
		2031
		2032	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		2033	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		2034	will return C<4>. When the timer expires and is restarted, it will return
		2035	roughly C<7> (likely slightly less as callback invocation takes some time,
		2036	too), and so on.
		2037
		2038	=item ev_tstamp repeat [read-write]
		2039
		2040	The current C<repeat> value. Will be used each time the watcher times out
		2041	or C<ev_timer_again> is called, and determines the next timeout (if any),
		2042	which is also when any modifications are taken into account.
819		2043
820	=back	2044	=back
821		2045
		2046	=head3 Examples
		2047
822	Example: create a timer that fires after 60 seconds.	2048	Example: Create a timer that fires after 60 seconds.
823		2049
824	static void	2050	static void
825	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2051	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
826	{	2052	{
827	.. one minute over, w is actually stopped right here	2053	.. one minute over, w is actually stopped right here
828	}	2054	}
829		2055
830	struct ev_timer mytimer;	2056	ev_timer mytimer;
831	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	2057	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
832	ev_timer_start (loop, &mytimer);	2058	ev_timer_start (loop, &mytimer);
833		2059
834	Example: create a timeout timer that times out after 10 seconds of	2060	Example: Create a timeout timer that times out after 10 seconds of
835	inactivity.	2061	inactivity.
836		2062
837	static void	2063	static void
838	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	2064	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
839	{	2065	{
840	.. ten seconds without any activity	2066	.. ten seconds without any activity
841	}	2067	}
842		2068
843	struct ev_timer mytimer;	2069	ev_timer mytimer;
844	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	2070	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
845	ev_timer_again (&mytimer); /* start timer */	2071	ev_timer_again (&mytimer); /* start timer */
846	ev_loop (loop, 0);	2072	ev_run (loop, 0);
847		2073
848	// and in some piece of code that gets executed on any "activity":	2074	// and in some piece of code that gets executed on any "activity":
849	// reset the timeout to start ticking again at 10 seconds	2075	// reset the timeout to start ticking again at 10 seconds
850	ev_timer_again (&mytimer);	2076	ev_timer_again (&mytimer);
851		2077
852		2078
853	=head2 C<ev_periodic> - to cron or not to cron	2079	=head2 C<ev_periodic> - to cron or not to cron?
854		2080
855	Periodic watchers are also timers of a kind, but they are very versatile	2081	Periodic watchers are also timers of a kind, but they are very versatile
856	(and unfortunately a bit complex).	2082	(and unfortunately a bit complex).
857		2083
858	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2084	Unlike C<ev_timer>, periodic watchers are not based on real time (or
859	but on wallclock time (absolute time). You can tell a periodic watcher	2085	relative time, the physical time that passes) but on wall clock time
860	to trigger "at" some specific point in time. For example, if you tell a	2086	(absolute time, the thing you can read on your calender or clock). The
861	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	2087	difference is that wall clock time can run faster or slower than real
862	+ 10.>) and then reset your system clock to the last year, then it will	2088	time, and time jumps are not uncommon (e.g. when you adjust your
863	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2089	wrist-watch).
864	roughly 10 seconds later and of course not if you reset your system time
865	again).
866		2090
867	They can also be used to implement vastly more complex timers, such as	2091	You can tell a periodic watcher to trigger after some specific point
		2092	in time: for example, if you tell a periodic watcher to trigger "in 10
		2093	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2094	not a delay) and then reset your system clock to January of the previous
		2095	year, then it will take a year or more to trigger the event (unlike an
		2096	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2097	it, as it uses a relative timeout).
		2098
		2099	C<ev_periodic> watchers can also be used to implement vastly more complex
868	triggering an event on eahc midnight, local time.	2100	timers, such as triggering an event on each "midnight, local time", or
		2101	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2102	those cannot react to time jumps.
869		2103
870	As with timers, the callback is guarenteed to be invoked only when the	2104	As with timers, the callback is guaranteed to be invoked only when the
871	time (C<at>) has been passed, but if multiple periodic timers become ready	2105	point in time where it is supposed to trigger has passed. If multiple
872	during the same loop iteration then order of execution is undefined.	2106	timers become ready during the same loop iteration then the ones with
		2107	earlier time-out values are invoked before ones with later time-out values
		2108	(but this is no longer true when a callback calls C<ev_run> recursively).
		2109
		2110	=head3 Watcher-Specific Functions and Data Members
873		2111
874	=over 4	2112	=over 4
875		2113
876	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2114	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
877		2115
878	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2116	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
879		2117
880	Lots of arguments, lets sort it out... There are basically three modes of	2118	Lots of arguments, let's sort it out... There are basically three modes of
881	operation, and we will explain them from simplest to complex:	2119	operation, and we will explain them from simplest to most complex:
882		2120
883	=over 4	2121	=over 4
884		2122
885	=item * absolute timer (interval = reschedule_cb = 0)	2123	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
886		2124
887	In this configuration the watcher triggers an event at the wallclock time	2125	In this configuration the watcher triggers an event after the wall clock
888	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2126	time C<offset> has passed. It will not repeat and will not adjust when a
889	that is, if it is to be run at January 1st 2011 then it will run when the	2127	time jump occurs, that is, if it is to be run at January 1st 2011 then it
890	system time reaches or surpasses this time.	2128	will be stopped and invoked when the system clock reaches or surpasses
		2129	this point in time.
891		2130
892	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	2131	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
893		2132
894	In this mode the watcher will always be scheduled to time out at the next	2133	In this mode the watcher will always be scheduled to time out at the next
895	C<at + N * interval> time (for some integer N) and then repeat, regardless	2134	C<offset + N * interval> time (for some integer N, which can also be
896	of any time jumps.	2135	negative) and then repeat, regardless of any time jumps. The C<offset>
		2136	argument is merely an offset into the C<interval> periods.
897		2137
898	This can be used to create timers that do not drift with respect to system	2138	This can be used to create timers that do not drift with respect to the
899	time:	2139	system clock, for example, here is an C<ev_periodic> that triggers each
		2140	hour, on the hour (with respect to UTC):
900		2141
901	ev_periodic_set (&periodic, 0., 3600., 0);	2142	ev_periodic_set (&periodic, 0., 3600., 0);
902		2143
903	This doesn't mean there will always be 3600 seconds in between triggers,	2144	This doesn't mean there will always be 3600 seconds in between triggers,
904	but only that the the callback will be called when the system time shows a	2145	but only that the callback will be called when the system time shows a
905	full hour (UTC), or more correctly, when the system time is evenly divisible	2146	full hour (UTC), or more correctly, when the system time is evenly divisible
906	by 3600.	2147	by 3600.
907		2148
908	Another way to think about it (for the mathematically inclined) is that	2149	Another way to think about it (for the mathematically inclined) is that
909	C<ev_periodic> will try to run the callback in this mode at the next possible	2150	C<ev_periodic> will try to run the callback in this mode at the next possible
910	time where C<time = at (mod interval)>, regardless of any time jumps.	2151	time where C<time = offset (mod interval)>, regardless of any time jumps.
911		2152
		2153	For numerical stability it is preferable that the C<offset> value is near
		2154	C<ev_now ()> (the current time), but there is no range requirement for
		2155	this value, and in fact is often specified as zero.
		2156
		2157	Note also that there is an upper limit to how often a timer can fire (CPU
		2158	speed for example), so if C<interval> is very small then timing stability
		2159	will of course deteriorate. Libev itself tries to be exact to be about one
		2160	millisecond (if the OS supports it and the machine is fast enough).
		2161
912	=item * manual reschedule mode (reschedule_cb = callback)	2162	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
913		2163
914	In this mode the values for C<interval> and C<at> are both being	2164	In this mode the values for C<interval> and C<offset> are both being
915	ignored. Instead, each time the periodic watcher gets scheduled, the	2165	ignored. Instead, each time the periodic watcher gets scheduled, the
916	reschedule callback will be called with the watcher as first, and the	2166	reschedule callback will be called with the watcher as first, and the
917	current time as second argument.	2167	current time as second argument.
918		2168
919	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2169	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
920	ever, or make any event loop modifications>. If you need to stop it,	2170	or make ANY other event loop modifications whatsoever, unless explicitly
921	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2171	allowed by documentation here>.
922	starting a prepare watcher).
923		2172
		2173	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2174	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2175	only event loop modification you are allowed to do).
		2176
924	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2177	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
925	ev_tstamp now)>, e.g.:	2178	*w, ev_tstamp now)>, e.g.:
926		2179
		2180	static ev_tstamp
927	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2181	my_rescheduler (ev_periodic *w, ev_tstamp now)
928	{	2182	{
929	return now + 60.;	2183	return now + 60.;
930	}	2184	}
931		2185
932	It must return the next time to trigger, based on the passed time value	2186	It must return the next time to trigger, based on the passed time value
933	(that is, the lowest time value larger than to the second argument). It	2187	(that is, the lowest time value larger than to the second argument). It
934	will usually be called just before the callback will be triggered, but	2188	will usually be called just before the callback will be triggered, but
935	might be called at other times, too.	2189	might be called at other times, too.
936		2190
937	NOTE: I<< This callback must always return a time that is later than the	2191	NOTE: I<< This callback must always return a time that is higher than or
938	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2192	equal to the passed C<now> value >>.
939		2193
940	This can be used to create very complex timers, such as a timer that	2194	This can be used to create very complex timers, such as a timer that
941	triggers on each midnight, local time. To do this, you would calculate the	2195	triggers on "next midnight, local time". To do this, you would calculate the
942	next midnight after C<now> and return the timestamp value for this. How	2196	next midnight after C<now> and return the timestamp value for this. How
943	you do this is, again, up to you (but it is not trivial, which is the main	2197	you do this is, again, up to you (but it is not trivial, which is the main
944	reason I omitted it as an example).	2198	reason I omitted it as an example).
945		2199
946	=back	2200	=back
…		…
950	Simply stops and restarts the periodic watcher again. This is only useful	2204	Simply stops and restarts the periodic watcher again. This is only useful
951	when you changed some parameters or the reschedule callback would return	2205	when you changed some parameters or the reschedule callback would return
952	a different time than the last time it was called (e.g. in a crond like	2206	a different time than the last time it was called (e.g. in a crond like
953	program when the crontabs have changed).	2207	program when the crontabs have changed).
954		2208
		2209	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2210
		2211	When active, returns the absolute time that the watcher is supposed
		2212	to trigger next. This is not the same as the C<offset> argument to
		2213	C<ev_periodic_set>, but indeed works even in interval and manual
		2214	rescheduling modes.
		2215
		2216	=item ev_tstamp offset [read-write]
		2217
		2218	When repeating, this contains the offset value, otherwise this is the
		2219	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2220	although libev might modify this value for better numerical stability).
		2221
		2222	Can be modified any time, but changes only take effect when the periodic
		2223	timer fires or C<ev_periodic_again> is being called.
		2224
		2225	=item ev_tstamp interval [read-write]
		2226
		2227	The current interval value. Can be modified any time, but changes only
		2228	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2229	called.
		2230
		2231	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2232
		2233	The current reschedule callback, or C<0>, if this functionality is
		2234	switched off. Can be changed any time, but changes only take effect when
		2235	the periodic timer fires or C<ev_periodic_again> is being called.
		2236
955	=back	2237	=back
956		2238
		2239	=head3 Examples
		2240
957	Example: call a callback every hour, or, more precisely, whenever the	2241	Example: Call a callback every hour, or, more precisely, whenever the
958	system clock is divisible by 3600. The callback invocation times have	2242	system time is divisible by 3600. The callback invocation times have
959	potentially a lot of jittering, but good long-term stability.	2243	potentially a lot of jitter, but good long-term stability.
960		2244
961	static void	2245	static void
962	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	2246	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
963	{	2247	{
964	... its now a full hour (UTC, or TAI or whatever your clock follows)	2248	... its now a full hour (UTC, or TAI or whatever your clock follows)
965	}	2249	}
966		2250
967	struct ev_periodic hourly_tick;	2251	ev_periodic hourly_tick;
968	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	2252	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
969	ev_periodic_start (loop, &hourly_tick);	2253	ev_periodic_start (loop, &hourly_tick);
970		2254
971	Example: the same as above, but use a reschedule callback to do it:	2255	Example: The same as above, but use a reschedule callback to do it:
972		2256
973	#include <math.h>	2257	#include <math.h>
974		2258
975	static ev_tstamp	2259	static ev_tstamp
976	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	2260	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
977	{	2261	{
978	return fmod (now, 3600.) + 3600.;	2262	return now + (3600. - fmod (now, 3600.));
979	}	2263	}
980		2264
981	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	2265	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
982		2266
983	Example: call a callback every hour, starting now:	2267	Example: Call a callback every hour, starting now:
984		2268
985	struct ev_periodic hourly_tick;	2269	ev_periodic hourly_tick;
986	ev_periodic_init (&hourly_tick, clock_cb,	2270	ev_periodic_init (&hourly_tick, clock_cb,
987	fmod (ev_now (loop), 3600.), 3600., 0);	2271	fmod (ev_now (loop), 3600.), 3600., 0);
988	ev_periodic_start (loop, &hourly_tick);	2272	ev_periodic_start (loop, &hourly_tick);
989		2273
990		2274
991	=head2 C<ev_signal> - signal me when a signal gets signalled	2275	=head2 C<ev_signal> - signal me when a signal gets signalled!
992		2276
993	Signal watchers will trigger an event when the process receives a specific	2277	Signal watchers will trigger an event when the process receives a specific
994	signal one or more times. Even though signals are very asynchronous, libev	2278	signal one or more times. Even though signals are very asynchronous, libev
995	will try it's best to deliver signals synchronously, i.e. as part of the	2279	will try its best to deliver signals synchronously, i.e. as part of the
996	normal event processing, like any other event.	2280	normal event processing, like any other event.
997		2281
		2282	If you want signals to be delivered truly asynchronously, just use
		2283	C<sigaction> as you would do without libev and forget about sharing
		2284	the signal. You can even use C<ev_async> from a signal handler to
		2285	synchronously wake up an event loop.
		2286
998	You can configure as many watchers as you like per signal. Only when the	2287	You can configure as many watchers as you like for the same signal, but
		2288	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2289	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2290	C<SIGINT> in both the default loop and another loop at the same time. At
		2291	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2292
999	first watcher gets started will libev actually register a signal watcher	2293	When the first watcher gets started will libev actually register something
1000	with the kernel (thus it coexists with your own signal handlers as long	2294	with the kernel (thus it coexists with your own signal handlers as long as
1001	as you don't register any with libev). Similarly, when the last signal	2295	you don't register any with libev for the same signal).
1002	watcher for a signal is stopped libev will reset the signal handler to	2296
1003	SIG_DFL (regardless of what it was set to before).	2297	If possible and supported, libev will install its handlers with
		2298	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2299	not be unduly interrupted. If you have a problem with system calls getting
		2300	interrupted by signals you can block all signals in an C<ev_check> watcher
		2301	and unblock them in an C<ev_prepare> watcher.
		2302
		2303	=head3 The special problem of inheritance over fork/execve/pthread_create
		2304
		2305	Both the signal mask (C<sigprocmask>) and the signal disposition
		2306	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2307	stopping it again), that is, libev might or might not block the signal,
		2308	and might or might not set or restore the installed signal handler (but
		2309	see C<EVFLAG_NOSIGMASK>).
		2310
		2311	While this does not matter for the signal disposition (libev never
		2312	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2313	C<execve>), this matters for the signal mask: many programs do not expect
		2314	certain signals to be blocked.
		2315
		2316	This means that before calling C<exec> (from the child) you should reset
		2317	the signal mask to whatever "default" you expect (all clear is a good
		2318	choice usually).
		2319
		2320	The simplest way to ensure that the signal mask is reset in the child is
		2321	to install a fork handler with C<pthread_atfork> that resets it. That will
		2322	catch fork calls done by libraries (such as the libc) as well.
		2323
		2324	In current versions of libev, the signal will not be blocked indefinitely
		2325	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2326	the window of opportunity for problems, it will not go away, as libev
		2327	I<has> to modify the signal mask, at least temporarily.
		2328
		2329	So I can't stress this enough: I<If you do not reset your signal mask when
		2330	you expect it to be empty, you have a race condition in your code>. This
		2331	is not a libev-specific thing, this is true for most event libraries.
		2332
		2333	=head3 The special problem of threads signal handling
		2334
		2335	POSIX threads has problematic signal handling semantics, specifically,
		2336	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2337	threads in a process block signals, which is hard to achieve.
		2338
		2339	When you want to use sigwait (or mix libev signal handling with your own
		2340	for the same signals), you can tackle this problem by globally blocking
		2341	all signals before creating any threads (or creating them with a fully set
		2342	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2343	loops. Then designate one thread as "signal receiver thread" which handles
		2344	these signals. You can pass on any signals that libev might be interested
		2345	in by calling C<ev_feed_signal>.
		2346
		2347	=head3 Watcher-Specific Functions and Data Members
1004		2348
1005	=over 4	2349	=over 4
1006		2350
1007	=item ev_signal_init (ev_signal *, callback, int signum)	2351	=item ev_signal_init (ev_signal *, callback, int signum)
1008		2352
1009	=item ev_signal_set (ev_signal *, int signum)	2353	=item ev_signal_set (ev_signal *, int signum)
1010		2354
1011	Configures the watcher to trigger on the given signal number (usually one	2355	Configures the watcher to trigger on the given signal number (usually one
1012	of the C<SIGxxx> constants).	2356	of the C<SIGxxx> constants).
1013		2357
		2358	=item int signum [read-only]
		2359
		2360	The signal the watcher watches out for.
		2361
1014	=back	2362	=back
1015		2363
		2364	=head3 Examples
1016		2365
		2366	Example: Try to exit cleanly on SIGINT.
		2367
		2368	static void
		2369	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2370	{
		2371	ev_break (loop, EVBREAK_ALL);
		2372	}
		2373
		2374	ev_signal signal_watcher;
		2375	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2376	ev_signal_start (loop, &signal_watcher);
		2377
		2378
1017	=head2 C<ev_child> - wait for pid status changes	2379	=head2 C<ev_child> - watch out for process status changes
1018		2380
1019	Child watchers trigger when your process receives a SIGCHLD in response to	2381	Child watchers trigger when your process receives a SIGCHLD in response to
1020	some child status changes (most typically when a child of yours dies).	2382	some child status changes (most typically when a child of yours dies or
		2383	exits). It is permissible to install a child watcher I<after> the child
		2384	has been forked (which implies it might have already exited), as long
		2385	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2386	forking and then immediately registering a watcher for the child is fine,
		2387	but forking and registering a watcher a few event loop iterations later or
		2388	in the next callback invocation is not.
		2389
		2390	Only the default event loop is capable of handling signals, and therefore
		2391	you can only register child watchers in the default event loop.
		2392
		2393	Due to some design glitches inside libev, child watchers will always be
		2394	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2395	libev)
		2396
		2397	=head3 Process Interaction
		2398
		2399	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2400	initialised. This is necessary to guarantee proper behaviour even if the
		2401	first child watcher is started after the child exits. The occurrence
		2402	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2403	synchronously as part of the event loop processing. Libev always reaps all
		2404	children, even ones not watched.
		2405
		2406	=head3 Overriding the Built-In Processing
		2407
		2408	Libev offers no special support for overriding the built-in child
		2409	processing, but if your application collides with libev's default child
		2410	handler, you can override it easily by installing your own handler for
		2411	C<SIGCHLD> after initialising the default loop, and making sure the
		2412	default loop never gets destroyed. You are encouraged, however, to use an
		2413	event-based approach to child reaping and thus use libev's support for
		2414	that, so other libev users can use C<ev_child> watchers freely.
		2415
		2416	=head3 Stopping the Child Watcher
		2417
		2418	Currently, the child watcher never gets stopped, even when the
		2419	child terminates, so normally one needs to stop the watcher in the
		2420	callback. Future versions of libev might stop the watcher automatically
		2421	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2422	problem).
		2423
		2424	=head3 Watcher-Specific Functions and Data Members
1021		2425
1022	=over 4	2426	=over 4
1023		2427
1024	=item ev_child_init (ev_child *, callback, int pid)	2428	=item ev_child_init (ev_child *, callback, int pid, int trace)
1025		2429
1026	=item ev_child_set (ev_child *, int pid)	2430	=item ev_child_set (ev_child *, int pid, int trace)
1027		2431
1028	Configures the watcher to wait for status changes of process C<pid> (or	2432	Configures the watcher to wait for status changes of process C<pid> (or
1029	I<any> process if C<pid> is specified as C<0>). The callback can look	2433	I<any> process if C<pid> is specified as C<0>). The callback can look
1030	at the C<rstatus> member of the C<ev_child> watcher structure to see	2434	at the C<rstatus> member of the C<ev_child> watcher structure to see
1031	the status word (use the macros from C<sys/wait.h> and see your systems	2435	the status word (use the macros from C<sys/wait.h> and see your systems
1032	C<waitpid> documentation). The C<rpid> member contains the pid of the	2436	C<waitpid> documentation). The C<rpid> member contains the pid of the
1033	process causing the status change.	2437	process causing the status change. C<trace> must be either C<0> (only
		2438	activate the watcher when the process terminates) or C<1> (additionally
		2439	activate the watcher when the process is stopped or continued).
		2440
		2441	=item int pid [read-only]
		2442
		2443	The process id this watcher watches out for, or C<0>, meaning any process id.
		2444
		2445	=item int rpid [read-write]
		2446
		2447	The process id that detected a status change.
		2448
		2449	=item int rstatus [read-write]
		2450
		2451	The process exit/trace status caused by C<rpid> (see your systems
		2452	C<waitpid> and C<sys/wait.h> documentation for details).
1034		2453
1035	=back	2454	=back
1036		2455
1037	Example: try to exit cleanly on SIGINT and SIGTERM.	2456	=head3 Examples
1038		2457
		2458	Example: C<fork()> a new process and install a child handler to wait for
		2459	its completion.
		2460
		2461	ev_child cw;
		2462
1039	static void	2463	static void
1040	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	2464	child_cb (EV_P_ ev_child *w, int revents)
1041	{	2465	{
1042	ev_unloop (loop, EVUNLOOP_ALL);	2466	ev_child_stop (EV_A_ w);
		2467	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1043	}	2468	}
1044		2469
1045	struct ev_signal signal_watcher;	2470	pid_t pid = fork ();
1046	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1047	ev_signal_start (loop, &sigint_cb);
1048		2471
		2472	if (pid < 0)
		2473	// error
		2474	else if (pid == 0)
		2475	{
		2476	// the forked child executes here
		2477	exit (1);
		2478	}
		2479	else
		2480	{
		2481	ev_child_init (&cw, child_cb, pid, 0);
		2482	ev_child_start (EV_DEFAULT_ &cw);
		2483	}
1049		2484
		2485
		2486	=head2 C<ev_stat> - did the file attributes just change?
		2487
		2488	This watches a file system path for attribute changes. That is, it calls
		2489	C<stat> on that path in regular intervals (or when the OS says it changed)
		2490	and sees if it changed compared to the last time, invoking the callback if
		2491	it did.
		2492
		2493	The path does not need to exist: changing from "path exists" to "path does
		2494	not exist" is a status change like any other. The condition "path does not
		2495	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2496	C<st_nlink> field being zero (which is otherwise always forced to be at
		2497	least one) and all the other fields of the stat buffer having unspecified
		2498	contents.
		2499
		2500	The path I<must not> end in a slash or contain special components such as
		2501	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2502	your working directory changes, then the behaviour is undefined.
		2503
		2504	Since there is no portable change notification interface available, the
		2505	portable implementation simply calls C<stat(2)> regularly on the path
		2506	to see if it changed somehow. You can specify a recommended polling
		2507	interval for this case. If you specify a polling interval of C<0> (highly
		2508	recommended!) then a I<suitable, unspecified default> value will be used
		2509	(which you can expect to be around five seconds, although this might
		2510	change dynamically). Libev will also impose a minimum interval which is
		2511	currently around C<0.1>, but that's usually overkill.
		2512
		2513	This watcher type is not meant for massive numbers of stat watchers,
		2514	as even with OS-supported change notifications, this can be
		2515	resource-intensive.
		2516
		2517	At the time of this writing, the only OS-specific interface implemented
		2518	is the Linux inotify interface (implementing kqueue support is left as an
		2519	exercise for the reader. Note, however, that the author sees no way of
		2520	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2521
		2522	=head3 ABI Issues (Largefile Support)
		2523
		2524	Libev by default (unless the user overrides this) uses the default
		2525	compilation environment, which means that on systems with large file
		2526	support disabled by default, you get the 32 bit version of the stat
		2527	structure. When using the library from programs that change the ABI to
		2528	use 64 bit file offsets the programs will fail. In that case you have to
		2529	compile libev with the same flags to get binary compatibility. This is
		2530	obviously the case with any flags that change the ABI, but the problem is
		2531	most noticeably displayed with ev_stat and large file support.
		2532
		2533	The solution for this is to lobby your distribution maker to make large
		2534	file interfaces available by default (as e.g. FreeBSD does) and not
		2535	optional. Libev cannot simply switch on large file support because it has
		2536	to exchange stat structures with application programs compiled using the
		2537	default compilation environment.
		2538
		2539	=head3 Inotify and Kqueue
		2540
		2541	When C<inotify (7)> support has been compiled into libev and present at
		2542	runtime, it will be used to speed up change detection where possible. The
		2543	inotify descriptor will be created lazily when the first C<ev_stat>
		2544	watcher is being started.
		2545
		2546	Inotify presence does not change the semantics of C<ev_stat> watchers
		2547	except that changes might be detected earlier, and in some cases, to avoid
		2548	making regular C<stat> calls. Even in the presence of inotify support
		2549	there are many cases where libev has to resort to regular C<stat> polling,
		2550	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2551	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2552	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2553	xfs are fully working) libev usually gets away without polling.
		2554
		2555	There is no support for kqueue, as apparently it cannot be used to
		2556	implement this functionality, due to the requirement of having a file
		2557	descriptor open on the object at all times, and detecting renames, unlinks
		2558	etc. is difficult.
		2559
		2560	=head3 C<stat ()> is a synchronous operation
		2561
		2562	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2563	the process. The exception are C<ev_stat> watchers - those call C<stat
		2564	()>, which is a synchronous operation.
		2565
		2566	For local paths, this usually doesn't matter: unless the system is very
		2567	busy or the intervals between stat's are large, a stat call will be fast,
		2568	as the path data is usually in memory already (except when starting the
		2569	watcher).
		2570
		2571	For networked file systems, calling C<stat ()> can block an indefinite
		2572	time due to network issues, and even under good conditions, a stat call
		2573	often takes multiple milliseconds.
		2574
		2575	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2576	paths, although this is fully supported by libev.
		2577
		2578	=head3 The special problem of stat time resolution
		2579
		2580	The C<stat ()> system call only supports full-second resolution portably,
		2581	and even on systems where the resolution is higher, most file systems
		2582	still only support whole seconds.
		2583
		2584	That means that, if the time is the only thing that changes, you can
		2585	easily miss updates: on the first update, C<ev_stat> detects a change and
		2586	calls your callback, which does something. When there is another update
		2587	within the same second, C<ev_stat> will be unable to detect unless the
		2588	stat data does change in other ways (e.g. file size).
		2589
		2590	The solution to this is to delay acting on a change for slightly more
		2591	than a second (or till slightly after the next full second boundary), using
		2592	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2593	ev_timer_again (loop, w)>).
		2594
		2595	The C<.02> offset is added to work around small timing inconsistencies
		2596	of some operating systems (where the second counter of the current time
		2597	might be be delayed. One such system is the Linux kernel, where a call to
		2598	C<gettimeofday> might return a timestamp with a full second later than
		2599	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2600	update file times then there will be a small window where the kernel uses
		2601	the previous second to update file times but libev might already execute
		2602	the timer callback).
		2603
		2604	=head3 Watcher-Specific Functions and Data Members
		2605
		2606	=over 4
		2607
		2608	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2609
		2610	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2611
		2612	Configures the watcher to wait for status changes of the given
		2613	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2614	be detected and should normally be specified as C<0> to let libev choose
		2615	a suitable value. The memory pointed to by C<path> must point to the same
		2616	path for as long as the watcher is active.
		2617
		2618	The callback will receive an C<EV_STAT> event when a change was detected,
		2619	relative to the attributes at the time the watcher was started (or the
		2620	last change was detected).
		2621
		2622	=item ev_stat_stat (loop, ev_stat *)
		2623
		2624	Updates the stat buffer immediately with new values. If you change the
		2625	watched path in your callback, you could call this function to avoid
		2626	detecting this change (while introducing a race condition if you are not
		2627	the only one changing the path). Can also be useful simply to find out the
		2628	new values.
		2629
		2630	=item ev_statdata attr [read-only]
		2631
		2632	The most-recently detected attributes of the file. Although the type is
		2633	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2634	suitable for your system, but you can only rely on the POSIX-standardised
		2635	members to be present. If the C<st_nlink> member is C<0>, then there was
		2636	some error while C<stat>ing the file.
		2637
		2638	=item ev_statdata prev [read-only]
		2639
		2640	The previous attributes of the file. The callback gets invoked whenever
		2641	C<prev> != C<attr>, or, more precisely, one or more of these members
		2642	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2643	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2644
		2645	=item ev_tstamp interval [read-only]
		2646
		2647	The specified interval.
		2648
		2649	=item const char *path [read-only]
		2650
		2651	The file system path that is being watched.
		2652
		2653	=back
		2654
		2655	=head3 Examples
		2656
		2657	Example: Watch C</etc/passwd> for attribute changes.
		2658
		2659	static void
		2660	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2661	{
		2662	/* /etc/passwd changed in some way */
		2663	if (w->attr.st_nlink)
		2664	{
		2665	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2666	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2667	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2668	}
		2669	else
		2670	/* you shalt not abuse printf for puts */
		2671	puts ("wow, /etc/passwd is not there, expect problems. "
		2672	"if this is windows, they already arrived\n");
		2673	}
		2674
		2675	...
		2676	ev_stat passwd;
		2677
		2678	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2679	ev_stat_start (loop, &passwd);
		2680
		2681	Example: Like above, but additionally use a one-second delay so we do not
		2682	miss updates (however, frequent updates will delay processing, too, so
		2683	one might do the work both on C<ev_stat> callback invocation I<and> on
		2684	C<ev_timer> callback invocation).
		2685
		2686	static ev_stat passwd;
		2687	static ev_timer timer;
		2688
		2689	static void
		2690	timer_cb (EV_P_ ev_timer *w, int revents)
		2691	{
		2692	ev_timer_stop (EV_A_ w);
		2693
		2694	/* now it's one second after the most recent passwd change */
		2695	}
		2696
		2697	static void
		2698	stat_cb (EV_P_ ev_stat *w, int revents)
		2699	{
		2700	/* reset the one-second timer */
		2701	ev_timer_again (EV_A_ &timer);
		2702	}
		2703
		2704	...
		2705	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2706	ev_stat_start (loop, &passwd);
		2707	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2708
		2709
1050	=head2 C<ev_idle> - when you've got nothing better to do	2710	=head2 C<ev_idle> - when you've got nothing better to do...
1051		2711
1052	Idle watchers trigger events when there are no other events are pending	2712	Idle watchers trigger events when no other events of the same or higher
1053	(prepare, check and other idle watchers do not count). That is, as long	2713	priority are pending (prepare, check and other idle watchers do not count
1054	as your process is busy handling sockets or timeouts (or even signals,	2714	as receiving "events").
1055	imagine) it will not be triggered. But when your process is idle all idle	2715
1056	watchers are being called again and again, once per event loop iteration -	2716	That is, as long as your process is busy handling sockets or timeouts
		2717	(or even signals, imagine) of the same or higher priority it will not be
		2718	triggered. But when your process is idle (or only lower-priority watchers
		2719	are pending), the idle watchers are being called once per event loop
1057	until stopped, that is, or your process receives more events and becomes	2720	iteration - until stopped, that is, or your process receives more events
1058	busy.	2721	and becomes busy again with higher priority stuff.
1059		2722
1060	The most noteworthy effect is that as long as any idle watchers are	2723	The most noteworthy effect is that as long as any idle watchers are
1061	active, the process will not block when waiting for new events.	2724	active, the process will not block when waiting for new events.
1062		2725
1063	Apart from keeping your process non-blocking (which is a useful	2726	Apart from keeping your process non-blocking (which is a useful
1064	effect on its own sometimes), idle watchers are a good place to do	2727	effect on its own sometimes), idle watchers are a good place to do
1065	"pseudo-background processing", or delay processing stuff to after the	2728	"pseudo-background processing", or delay processing stuff to after the
1066	event loop has handled all outstanding events.	2729	event loop has handled all outstanding events.
1067		2730
		2731	=head3 Watcher-Specific Functions and Data Members
		2732
1068	=over 4	2733	=over 4
1069		2734
1070	=item ev_idle_init (ev_signal *, callback)	2735	=item ev_idle_init (ev_idle *, callback)
1071		2736
1072	Initialises and configures the idle watcher - it has no parameters of any	2737	Initialises and configures the idle watcher - it has no parameters of any
1073	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2738	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1074	believe me.	2739	believe me.
1075		2740
1076	=back	2741	=back
1077		2742
		2743	=head3 Examples
		2744
1078	Example: dynamically allocate an C<ev_idle>, start it, and in the	2745	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1079	callback, free it. Alos, use no error checking, as usual.	2746	callback, free it. Also, use no error checking, as usual.
1080		2747
1081	static void	2748	static void
1082	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2749	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1083	{	2750	{
1084	free (w);	2751	free (w);
1085	// now do something you wanted to do when the program has	2752	// now do something you wanted to do when the program has
1086	// no longer asnything immediate to do.	2753	// no longer anything immediate to do.
1087	}	2754	}
1088		2755
1089	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2756	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1090	ev_idle_init (idle_watcher, idle_cb);	2757	ev_idle_init (idle_watcher, idle_cb);
1091	ev_idle_start (loop, idle_cb);	2758	ev_idle_start (loop, idle_watcher);
1092		2759
1093		2760
1094	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2761	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1095		2762
1096	Prepare and check watchers are usually (but not always) used in tandem:	2763	Prepare and check watchers are usually (but not always) used in pairs:
1097	prepare watchers get invoked before the process blocks and check watchers	2764	prepare watchers get invoked before the process blocks and check watchers
1098	afterwards.	2765	afterwards.
1099		2766
		2767	You I<must not> call C<ev_run> or similar functions that enter
		2768	the current event loop from either C<ev_prepare> or C<ev_check>
		2769	watchers. Other loops than the current one are fine, however. The
		2770	rationale behind this is that you do not need to check for recursion in
		2771	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2772	C<ev_check> so if you have one watcher of each kind they will always be
		2773	called in pairs bracketing the blocking call.
		2774
1100	Their main purpose is to integrate other event mechanisms into libev and	2775	Their main purpose is to integrate other event mechanisms into libev and
1101	their use is somewhat advanced. This could be used, for example, to track	2776	their use is somewhat advanced. They could be used, for example, to track
1102	variable changes, implement your own watchers, integrate net-snmp or a	2777	variable changes, implement your own watchers, integrate net-snmp or a
1103	coroutine library and lots more.	2778	coroutine library and lots more. They are also occasionally useful if
		2779	you cache some data and want to flush it before blocking (for example,
		2780	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2781	watcher).
1104		2782
1105	This is done by examining in each prepare call which file descriptors need	2783	This is done by examining in each prepare call which file descriptors
1106	to be watched by the other library, registering C<ev_io> watchers for	2784	need to be watched by the other library, registering C<ev_io> watchers
1107	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2785	for them and starting an C<ev_timer> watcher for any timeouts (many
1108	provide just this functionality). Then, in the check watcher you check for	2786	libraries provide exactly this functionality). Then, in the check watcher,
1109	any events that occured (by checking the pending status of all watchers	2787	you check for any events that occurred (by checking the pending status
1110	and stopping them) and call back into the library. The I/O and timer	2788	of all watchers and stopping them) and call back into the library. The
1111	callbacks will never actually be called (but must be valid nevertheless,	2789	I/O and timer callbacks will never actually be called (but must be valid
1112	because you never know, you know?).	2790	nevertheless, because you never know, you know?).
1113		2791
1114	As another example, the Perl Coro module uses these hooks to integrate	2792	As another example, the Perl Coro module uses these hooks to integrate
1115	coroutines into libev programs, by yielding to other active coroutines	2793	coroutines into libev programs, by yielding to other active coroutines
1116	during each prepare and only letting the process block if no coroutines	2794	during each prepare and only letting the process block if no coroutines
1117	are ready to run (it's actually more complicated: it only runs coroutines	2795	are ready to run (it's actually more complicated: it only runs coroutines
1118	with priority higher than or equal to the event loop and one coroutine	2796	with priority higher than or equal to the event loop and one coroutine
1119	of lower priority, but only once, using idle watchers to keep the event	2797	of lower priority, but only once, using idle watchers to keep the event
1120	loop from blocking if lower-priority coroutines are active, thus mapping	2798	loop from blocking if lower-priority coroutines are active, thus mapping
1121	low-priority coroutines to idle/background tasks).	2799	low-priority coroutines to idle/background tasks).
1122		2800
		2801	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2802	priority, to ensure that they are being run before any other watchers
		2803	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2804
		2805	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2806	activate ("feed") events into libev. While libev fully supports this, they
		2807	might get executed before other C<ev_check> watchers did their job. As
		2808	C<ev_check> watchers are often used to embed other (non-libev) event
		2809	loops those other event loops might be in an unusable state until their
		2810	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2811	others).
		2812
		2813	=head3 Watcher-Specific Functions and Data Members
		2814
1123	=over 4	2815	=over 4
1124		2816
1125	=item ev_prepare_init (ev_prepare *, callback)	2817	=item ev_prepare_init (ev_prepare *, callback)
1126		2818
1127	=item ev_check_init (ev_check *, callback)	2819	=item ev_check_init (ev_check *, callback)
1128		2820
1129	Initialises and configures the prepare or check watcher - they have no	2821	Initialises and configures the prepare or check watcher - they have no
1130	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2822	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1131	macros, but using them is utterly, utterly and completely pointless.	2823	macros, but using them is utterly, utterly, utterly and completely
		2824	pointless.
1132		2825
1133	=back	2826	=back
1134		2827
1135	Example: *TODO*.	2828	=head3 Examples
1136		2829
		2830	There are a number of principal ways to embed other event loops or modules
		2831	into libev. Here are some ideas on how to include libadns into libev
		2832	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2833	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2834	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2835	Glib event loop).
1137		2836
		2837	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2838	and in a check watcher, destroy them and call into libadns. What follows
		2839	is pseudo-code only of course. This requires you to either use a low
		2840	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2841	the callbacks for the IO/timeout watchers might not have been called yet.
		2842
		2843	static ev_io iow [nfd];
		2844	static ev_timer tw;
		2845
		2846	static void
		2847	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2848	{
		2849	}
		2850
		2851	// create io watchers for each fd and a timer before blocking
		2852	static void
		2853	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2854	{
		2855	int timeout = 3600000;
		2856	struct pollfd fds [nfd];
		2857	// actual code will need to loop here and realloc etc.
		2858	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2859
		2860	/* the callback is illegal, but won't be called as we stop during check */
		2861	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2862	ev_timer_start (loop, &tw);
		2863
		2864	// create one ev_io per pollfd
		2865	for (int i = 0; i < nfd; ++i)
		2866	{
		2867	ev_io_init (iow + i, io_cb, fds [i].fd,
		2868	((fds [i].events & POLLIN ? EV_READ : 0)
		2869	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2870
		2871	fds [i].revents = 0;
		2872	ev_io_start (loop, iow + i);
		2873	}
		2874	}
		2875
		2876	// stop all watchers after blocking
		2877	static void
		2878	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2879	{
		2880	ev_timer_stop (loop, &tw);
		2881
		2882	for (int i = 0; i < nfd; ++i)
		2883	{
		2884	// set the relevant poll flags
		2885	// could also call adns_processreadable etc. here
		2886	struct pollfd *fd = fds + i;
		2887	int revents = ev_clear_pending (iow + i);
		2888	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2889	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2890
		2891	// now stop the watcher
		2892	ev_io_stop (loop, iow + i);
		2893	}
		2894
		2895	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2896	}
		2897
		2898	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2899	in the prepare watcher and would dispose of the check watcher.
		2900
		2901	Method 3: If the module to be embedded supports explicit event
		2902	notification (libadns does), you can also make use of the actual watcher
		2903	callbacks, and only destroy/create the watchers in the prepare watcher.
		2904
		2905	static void
		2906	timer_cb (EV_P_ ev_timer *w, int revents)
		2907	{
		2908	adns_state ads = (adns_state)w->data;
		2909	update_now (EV_A);
		2910
		2911	adns_processtimeouts (ads, &tv_now);
		2912	}
		2913
		2914	static void
		2915	io_cb (EV_P_ ev_io *w, int revents)
		2916	{
		2917	adns_state ads = (adns_state)w->data;
		2918	update_now (EV_A);
		2919
		2920	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2921	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2922	}
		2923
		2924	// do not ever call adns_afterpoll
		2925
		2926	Method 4: Do not use a prepare or check watcher because the module you
		2927	want to embed is not flexible enough to support it. Instead, you can
		2928	override their poll function. The drawback with this solution is that the
		2929	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2930	this approach, effectively embedding EV as a client into the horrible
		2931	libglib event loop.
		2932
		2933	static gint
		2934	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2935	{
		2936	int got_events = 0;
		2937
		2938	for (n = 0; n < nfds; ++n)
		2939	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2940
		2941	if (timeout >= 0)
		2942	// create/start timer
		2943
		2944	// poll
		2945	ev_run (EV_A_ 0);
		2946
		2947	// stop timer again
		2948	if (timeout >= 0)
		2949	ev_timer_stop (EV_A_ &to);
		2950
		2951	// stop io watchers again - their callbacks should have set
		2952	for (n = 0; n < nfds; ++n)
		2953	ev_io_stop (EV_A_ iow [n]);
		2954
		2955	return got_events;
		2956	}
		2957
		2958
1138	=head2 C<ev_embed> - when one backend isn't enough	2959	=head2 C<ev_embed> - when one backend isn't enough...
1139		2960
1140	This is a rather advanced watcher type that lets you embed one event loop	2961	This is a rather advanced watcher type that lets you embed one event loop
1141	into another (currently only C<ev_io> events are supported in the embedded	2962	into another (currently only C<ev_io> events are supported in the embedded
1142	loop, other types of watchers might be handled in a delayed or incorrect	2963	loop, other types of watchers might be handled in a delayed or incorrect
1143	fashion and must not be used).	2964	fashion and must not be used).
…		…
1146	prioritise I/O.	2967	prioritise I/O.
1147		2968
1148	As an example for a bug workaround, the kqueue backend might only support	2969	As an example for a bug workaround, the kqueue backend might only support
1149	sockets on some platform, so it is unusable as generic backend, but you	2970	sockets on some platform, so it is unusable as generic backend, but you
1150	still want to make use of it because you have many sockets and it scales	2971	still want to make use of it because you have many sockets and it scales
1151	so nicely. In this case, you would create a kqueue-based loop and embed it	2972	so nicely. In this case, you would create a kqueue-based loop and embed
1152	into your default loop (which might use e.g. poll). Overall operation will	2973	it into your default loop (which might use e.g. poll). Overall operation
1153	be a bit slower because first libev has to poll and then call kevent, but	2974	will be a bit slower because first libev has to call C<poll> and then
1154	at least you can use both at what they are best.	2975	C<kevent>, but at least you can use both mechanisms for what they are
		2976	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
1155		2977
1156	As for prioritising I/O: rarely you have the case where some fds have	2978	As for prioritising I/O: under rare circumstances you have the case where
1157	to be watched and handled very quickly (with low latency), and even	2979	some fds have to be watched and handled very quickly (with low latency),
1158	priorities and idle watchers might have too much overhead. In this case	2980	and even priorities and idle watchers might have too much overhead. In
1159	you would put all the high priority stuff in one loop and all the rest in	2981	this case you would put all the high priority stuff in one loop and all
1160	a second one, and embed the second one in the first.	2982	the rest in a second one, and embed the second one in the first.
1161		2983
1162	As long as the watcher is active, the callback will be invoked every time	2984	As long as the watcher is active, the callback will be invoked every
1163	there might be events pending in the embedded loop. The callback must then	2985	time there might be events pending in the embedded loop. The callback
1164	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	2986	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
1165	their callbacks (you could also start an idle watcher to give the embedded	2987	sweep and invoke their callbacks (the callback doesn't need to invoke the
1166	loop strictly lower priority for example). You can also set the callback	2988	C<ev_embed_sweep> function directly, it could also start an idle watcher
1167	to C<0>, in which case the embed watcher will automatically execute the	2989	to give the embedded loop strictly lower priority for example).
1168	embedded loop sweep.
1169		2990
1170	As long as the watcher is started it will automatically handle events. The	2991	You can also set the callback to C<0>, in which case the embed watcher
1171	callback will be invoked whenever some events have been handled. You can	2992	will automatically execute the embedded loop sweep whenever necessary.
1172	set the callback to C<0> to avoid having to specify one if you are not
1173	interested in that.
1174		2993
1175	Also, there have not currently been made special provisions for forking:	2994	Fork detection will be handled transparently while the C<ev_embed> watcher
1176	when you fork, you not only have to call C<ev_loop_fork> on both loops,	2995	is active, i.e., the embedded loop will automatically be forked when the
1177	but you will also have to stop and restart any C<ev_embed> watchers	2996	embedding loop forks. In other cases, the user is responsible for calling
1178	yourself.	2997	C<ev_loop_fork> on the embedded loop.
1179		2998
1180	Unfortunately, not all backends are embeddable, only the ones returned by	2999	Unfortunately, not all backends are embeddable: only the ones returned by
1181	C<ev_embeddable_backends> are, which, unfortunately, does not include any	3000	C<ev_embeddable_backends> are, which, unfortunately, does not include any
1182	portable one.	3001	portable one.
1183		3002
1184	So when you want to use this feature you will always have to be prepared	3003	So when you want to use this feature you will always have to be prepared
1185	that you cannot get an embeddable loop. The recommended way to get around	3004	that you cannot get an embeddable loop. The recommended way to get around
1186	this is to have a separate variables for your embeddable loop, try to	3005	this is to have a separate variables for your embeddable loop, try to
1187	create it, and if that fails, use the normal loop for everything:	3006	create it, and if that fails, use the normal loop for everything.
1188		3007
1189	struct ev_loop *loop_hi = ev_default_init (0);	3008	=head3 C<ev_embed> and fork
1190	struct ev_loop *loop_lo = 0;
1191	struct ev_embed embed;
1192
1193	// see if there is a chance of getting one that works
1194	// (remember that a flags value of 0 means autodetection)
1195	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1196	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1197	: 0;
1198		3009
1199	// if we got one, then embed it, otherwise default to loop_hi	3010	While the C<ev_embed> watcher is running, forks in the embedding loop will
1200	if (loop_lo)	3011	automatically be applied to the embedded loop as well, so no special
1201	{	3012	fork handling is required in that case. When the watcher is not running,
1202	ev_embed_init (&embed, 0, loop_lo);	3013	however, it is still the task of the libev user to call C<ev_loop_fork ()>
1203	ev_embed_start (loop_hi, &embed);	3014	as applicable.
1204	}	3015
1205	else	3016	=head3 Watcher-Specific Functions and Data Members
1206	loop_lo = loop_hi;
1207		3017
1208	=over 4	3018	=over 4
1209		3019
1210	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	3020	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1211		3021
…		…
1213		3023
1214	Configures the watcher to embed the given loop, which must be	3024	Configures the watcher to embed the given loop, which must be
1215	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	3025	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1216	invoked automatically, otherwise it is the responsibility of the callback	3026	invoked automatically, otherwise it is the responsibility of the callback
1217	to invoke it (it will continue to be called until the sweep has been done,	3027	to invoke it (it will continue to be called until the sweep has been done,
1218	if you do not want thta, you need to temporarily stop the embed watcher).	3028	if you do not want that, you need to temporarily stop the embed watcher).
1219		3029
1220	=item ev_embed_sweep (loop, ev_embed *)	3030	=item ev_embed_sweep (loop, ev_embed *)
1221		3031
1222	Make a single, non-blocking sweep over the embedded loop. This works	3032	Make a single, non-blocking sweep over the embedded loop. This works
1223	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	3033	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
1224	apropriate way for embedded loops.	3034	appropriate way for embedded loops.
		3035
		3036	=item struct ev_loop *other [read-only]
		3037
		3038	The embedded event loop.
1225		3039
1226	=back	3040	=back
1227		3041
		3042	=head3 Examples
		3043
		3044	Example: Try to get an embeddable event loop and embed it into the default
		3045	event loop. If that is not possible, use the default loop. The default
		3046	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		3047	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		3048	used).
		3049
		3050	struct ev_loop *loop_hi = ev_default_init (0);
		3051	struct ev_loop *loop_lo = 0;
		3052	ev_embed embed;
		3053
		3054	// see if there is a chance of getting one that works
		3055	// (remember that a flags value of 0 means autodetection)
		3056	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3057	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3058	: 0;
		3059
		3060	// if we got one, then embed it, otherwise default to loop_hi
		3061	if (loop_lo)
		3062	{
		3063	ev_embed_init (&embed, 0, loop_lo);
		3064	ev_embed_start (loop_hi, &embed);
		3065	}
		3066	else
		3067	loop_lo = loop_hi;
		3068
		3069	Example: Check if kqueue is available but not recommended and create
		3070	a kqueue backend for use with sockets (which usually work with any
		3071	kqueue implementation). Store the kqueue/socket-only event loop in
		3072	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3073
		3074	struct ev_loop *loop = ev_default_init (0);
		3075	struct ev_loop *loop_socket = 0;
		3076	ev_embed embed;
		3077
		3078	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3079	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3080	{
		3081	ev_embed_init (&embed, 0, loop_socket);
		3082	ev_embed_start (loop, &embed);
		3083	}
		3084
		3085	if (!loop_socket)
		3086	loop_socket = loop;
		3087
		3088	// now use loop_socket for all sockets, and loop for everything else
		3089
		3090
		3091	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		3092
		3093	Fork watchers are called when a C<fork ()> was detected (usually because
		3094	whoever is a good citizen cared to tell libev about it by calling
		3095	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		3096	event loop blocks next and before C<ev_check> watchers are being called,
		3097	and only in the child after the fork. If whoever good citizen calling
		3098	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		3099	handlers will be invoked, too, of course.
		3100
		3101	=head3 The special problem of life after fork - how is it possible?
		3102
		3103	Most uses of C<fork()> consist of forking, then some simple calls to set
		3104	up/change the process environment, followed by a call to C<exec()>. This
		3105	sequence should be handled by libev without any problems.
		3106
		3107	This changes when the application actually wants to do event handling
		3108	in the child, or both parent in child, in effect "continuing" after the
		3109	fork.
		3110
		3111	The default mode of operation (for libev, with application help to detect
		3112	forks) is to duplicate all the state in the child, as would be expected
		3113	when I<either> the parent I<or> the child process continues.
		3114
		3115	When both processes want to continue using libev, then this is usually the
		3116	wrong result. In that case, usually one process (typically the parent) is
		3117	supposed to continue with all watchers in place as before, while the other
		3118	process typically wants to start fresh, i.e. without any active watchers.
		3119
		3120	The cleanest and most efficient way to achieve that with libev is to
		3121	simply create a new event loop, which of course will be "empty", and
		3122	use that for new watchers. This has the advantage of not touching more
		3123	memory than necessary, and thus avoiding the copy-on-write, and the
		3124	disadvantage of having to use multiple event loops (which do not support
		3125	signal watchers).
		3126
		3127	When this is not possible, or you want to use the default loop for
		3128	other reasons, then in the process that wants to start "fresh", call
		3129	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
		3130	Destroying the default loop will "orphan" (not stop) all registered
		3131	watchers, so you have to be careful not to execute code that modifies
		3132	those watchers. Note also that in that case, you have to re-register any
		3133	signal watchers.
		3134
		3135	=head3 Watcher-Specific Functions and Data Members
		3136
		3137	=over 4
		3138
		3139	=item ev_fork_init (ev_fork *, callback)
		3140
		3141	Initialises and configures the fork watcher - it has no parameters of any
		3142	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		3143	really.
		3144
		3145	=back
		3146
		3147
		3148	=head2 C<ev_cleanup> - even the best things end
		3149
		3150	Cleanup watchers are called just before the event loop is being destroyed
		3151	by a call to C<ev_loop_destroy>.
		3152
		3153	While there is no guarantee that the event loop gets destroyed, cleanup
		3154	watchers provide a convenient method to install cleanup hooks for your
		3155	program, worker threads and so on - you just to make sure to destroy the
		3156	loop when you want them to be invoked.
		3157
		3158	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3159	all other watchers, they do not keep a reference to the event loop (which
		3160	makes a lot of sense if you think about it). Like all other watchers, you
		3161	can call libev functions in the callback, except C<ev_cleanup_start>.
		3162
		3163	=head3 Watcher-Specific Functions and Data Members
		3164
		3165	=over 4
		3166
		3167	=item ev_cleanup_init (ev_cleanup *, callback)
		3168
		3169	Initialises and configures the cleanup watcher - it has no parameters of
		3170	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3171	pointless, I assure you.
		3172
		3173	=back
		3174
		3175	Example: Register an atexit handler to destroy the default loop, so any
		3176	cleanup functions are called.
		3177
		3178	static void
		3179	program_exits (void)
		3180	{
		3181	ev_loop_destroy (EV_DEFAULT_UC);
		3182	}
		3183
		3184	...
		3185	atexit (program_exits);
		3186
		3187
		3188	=head2 C<ev_async> - how to wake up an event loop
		3189
		3190	In general, you cannot use an C<ev_run> from multiple threads or other
		3191	asynchronous sources such as signal handlers (as opposed to multiple event
		3192	loops - those are of course safe to use in different threads).
		3193
		3194	Sometimes, however, you need to wake up an event loop you do not control,
		3195	for example because it belongs to another thread. This is what C<ev_async>
		3196	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3197	it by calling C<ev_async_send>, which is thread- and signal safe.
		3198
		3199	This functionality is very similar to C<ev_signal> watchers, as signals,
		3200	too, are asynchronous in nature, and signals, too, will be compressed
		3201	(i.e. the number of callback invocations may be less than the number of
		3202	C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
		3203	of "global async watchers" by using a watcher on an otherwise unused
		3204	signal, and C<ev_feed_signal> to signal this watcher from another thread,
		3205	even without knowing which loop owns the signal.
		3206
		3207	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		3208	just the default loop.
		3209
		3210	=head3 Queueing
		3211
		3212	C<ev_async> does not support queueing of data in any way. The reason
		3213	is that the author does not know of a simple (or any) algorithm for a
		3214	multiple-writer-single-reader queue that works in all cases and doesn't
		3215	need elaborate support such as pthreads or unportable memory access
		3216	semantics.
		3217
		3218	That means that if you want to queue data, you have to provide your own
		3219	queue. But at least I can tell you how to implement locking around your
		3220	queue:
		3221
		3222	=over 4
		3223
		3224	=item queueing from a signal handler context
		3225
		3226	To implement race-free queueing, you simply add to the queue in the signal
		3227	handler but you block the signal handler in the watcher callback. Here is
		3228	an example that does that for some fictitious SIGUSR1 handler:
		3229
		3230	static ev_async mysig;
		3231
		3232	static void
		3233	sigusr1_handler (void)
		3234	{
		3235	sometype data;
		3236
		3237	// no locking etc.
		3238	queue_put (data);
		3239	ev_async_send (EV_DEFAULT_ &mysig);
		3240	}
		3241
		3242	static void
		3243	mysig_cb (EV_P_ ev_async *w, int revents)
		3244	{
		3245	sometype data;
		3246	sigset_t block, prev;
		3247
		3248	sigemptyset (&block);
		3249	sigaddset (&block, SIGUSR1);
		3250	sigprocmask (SIG_BLOCK, &block, &prev);
		3251
		3252	while (queue_get (&data))
		3253	process (data);
		3254
		3255	if (sigismember (&prev, SIGUSR1)
		3256	sigprocmask (SIG_UNBLOCK, &block, 0);
		3257	}
		3258
		3259	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3260	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3261	either...).
		3262
		3263	=item queueing from a thread context
		3264
		3265	The strategy for threads is different, as you cannot (easily) block
		3266	threads but you can easily preempt them, so to queue safely you need to
		3267	employ a traditional mutex lock, such as in this pthread example:
		3268
		3269	static ev_async mysig;
		3270	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3271
		3272	static void
		3273	otherthread (void)
		3274	{
		3275	// only need to lock the actual queueing operation
		3276	pthread_mutex_lock (&mymutex);
		3277	queue_put (data);
		3278	pthread_mutex_unlock (&mymutex);
		3279
		3280	ev_async_send (EV_DEFAULT_ &mysig);
		3281	}
		3282
		3283	static void
		3284	mysig_cb (EV_P_ ev_async *w, int revents)
		3285	{
		3286	pthread_mutex_lock (&mymutex);
		3287
		3288	while (queue_get (&data))
		3289	process (data);
		3290
		3291	pthread_mutex_unlock (&mymutex);
		3292	}
		3293
		3294	=back
		3295
		3296
		3297	=head3 Watcher-Specific Functions and Data Members
		3298
		3299	=over 4
		3300
		3301	=item ev_async_init (ev_async *, callback)
		3302
		3303	Initialises and configures the async watcher - it has no parameters of any
		3304	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3305	trust me.
		3306
		3307	=item ev_async_send (loop, ev_async *)
		3308
		3309	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3310	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		3311	C<ev_feed_event>, this call is safe to do from other threads, signal or
		3312	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		3313	section below on what exactly this means).
		3314
		3315	Note that, as with other watchers in libev, multiple events might get
		3316	compressed into a single callback invocation (another way to look at this
		3317	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3318	reset when the event loop detects that).
		3319
		3320	This call incurs the overhead of a system call only once per event loop
		3321	iteration, so while the overhead might be noticeable, it doesn't apply to
		3322	repeated calls to C<ev_async_send> for the same event loop.
		3323
		3324	=item bool = ev_async_pending (ev_async *)
		3325
		3326	Returns a non-zero value when C<ev_async_send> has been called on the
		3327	watcher but the event has not yet been processed (or even noted) by the
		3328	event loop.
		3329
		3330	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3331	the loop iterates next and checks for the watcher to have become active,
		3332	it will reset the flag again. C<ev_async_pending> can be used to very
		3333	quickly check whether invoking the loop might be a good idea.
		3334
		3335	Not that this does I<not> check whether the watcher itself is pending,
		3336	only whether it has been requested to make this watcher pending: there
		3337	is a time window between the event loop checking and resetting the async
		3338	notification, and the callback being invoked.
		3339
		3340	=back
		3341
1228		3342
1229	=head1 OTHER FUNCTIONS	3343	=head1 OTHER FUNCTIONS
1230		3344
1231	There are some other functions of possible interest. Described. Here. Now.	3345	There are some other functions of possible interest. Described. Here. Now.
1232		3346
1233	=over 4	3347	=over 4
1234		3348
1235	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3349	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
1236		3350
1237	This function combines a simple timer and an I/O watcher, calls your	3351	This function combines a simple timer and an I/O watcher, calls your
1238	callback on whichever event happens first and automatically stop both	3352	callback on whichever event happens first and automatically stops both
1239	watchers. This is useful if you want to wait for a single event on an fd	3353	watchers. This is useful if you want to wait for a single event on an fd
1240	or timeout without having to allocate/configure/start/stop/free one or	3354	or timeout without having to allocate/configure/start/stop/free one or
1241	more watchers yourself.	3355	more watchers yourself.
1242		3356
1243	If C<fd> is less than 0, then no I/O watcher will be started and events	3357	If C<fd> is less than 0, then no I/O watcher will be started and the
1244	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3358	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
1245	C<events> set will be craeted and started.	3359	the given C<fd> and C<events> set will be created and started.
1246		3360
1247	If C<timeout> is less than 0, then no timeout watcher will be	3361	If C<timeout> is less than 0, then no timeout watcher will be
1248	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3362	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
1249	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3363	repeat = 0) will be started. C<0> is a valid timeout.
1250	dubious value.
1251		3364
1252	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3365	The callback has the type C<void (*cb)(int revents, void *arg)> and is
1253	passed an C<revents> set like normal event callbacks (a combination of	3366	passed an C<revents> set like normal event callbacks (a combination of
1254	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3367	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
1255	value passed to C<ev_once>:	3368	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3369	a timeout and an io event at the same time - you probably should give io
		3370	events precedence.
1256		3371
		3372	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3373
1257	static void stdin_ready (int revents, void *arg)	3374	static void stdin_ready (int revents, void *arg)
1258	{	3375	{
1259	if (revents & EV_TIMEOUT)
1260	/* doh, nothing entered */;
1261	else if (revents & EV_READ)	3376	if (revents & EV_READ)
1262	/* stdin might have data for us, joy! */;	3377	/* stdin might have data for us, joy! */;
		3378	else if (revents & EV_TIMER)
		3379	/* doh, nothing entered */;
1263	}	3380	}
1264		3381
1265	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3382	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1266		3383
1267	=item ev_feed_event (ev_loop *, watcher *, int revents)
1268
1269	Feeds the given event set into the event loop, as if the specified event
1270	had happened for the specified watcher (which must be a pointer to an
1271	initialised but not necessarily started event watcher).
1272
1273	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	3384	=item ev_feed_fd_event (loop, int fd, int revents)
1274		3385
1275	Feed an event on the given fd, as if a file descriptor backend detected	3386	Feed an event on the given fd, as if a file descriptor backend detected
1276	the given events it.	3387	the given events it.
1277		3388
1278	=item ev_feed_signal_event (ev_loop *loop, int signum)	3389	=item ev_feed_signal_event (loop, int signum)
1279		3390
1280	Feed an event as if the given signal occured (C<loop> must be the default	3391	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
1281	loop!).	3392	which is async-safe.
1282		3393
1283	=back	3394	=back
		3395
		3396
		3397	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3398
		3399	This section explains some common idioms that are not immediately
		3400	obvious. Note that examples are sprinkled over the whole manual, and this
		3401	section only contains stuff that wouldn't fit anywhere else.
		3402
		3403	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3404
		3405	Each watcher has, by default, a C<void *data> member that you can read
		3406	or modify at any time: libev will completely ignore it. This can be used
		3407	to associate arbitrary data with your watcher. If you need more data and
		3408	don't want to allocate memory separately and store a pointer to it in that
		3409	data member, you can also "subclass" the watcher type and provide your own
		3410	data:
		3411
		3412	struct my_io
		3413	{
		3414	ev_io io;
		3415	int otherfd;
		3416	void *somedata;
		3417	struct whatever *mostinteresting;
		3418	};
		3419
		3420	...
		3421	struct my_io w;
		3422	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3423
		3424	And since your callback will be called with a pointer to the watcher, you
		3425	can cast it back to your own type:
		3426
		3427	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3428	{
		3429	struct my_io *w = (struct my_io *)w_;
		3430	...
		3431	}
		3432
		3433	More interesting and less C-conformant ways of casting your callback
		3434	function type instead have been omitted.
		3435
		3436	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3437
		3438	Another common scenario is to use some data structure with multiple
		3439	embedded watchers, in effect creating your own watcher that combines
		3440	multiple libev event sources into one "super-watcher":
		3441
		3442	struct my_biggy
		3443	{
		3444	int some_data;
		3445	ev_timer t1;
		3446	ev_timer t2;
		3447	}
		3448
		3449	In this case getting the pointer to C<my_biggy> is a bit more
		3450	complicated: Either you store the address of your C<my_biggy> struct in
		3451	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3452	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3453	real programmers):
		3454
		3455	#include <stddef.h>
		3456
		3457	static void
		3458	t1_cb (EV_P_ ev_timer *w, int revents)
		3459	{
		3460	struct my_biggy big = (struct my_biggy *)
		3461	(((char *)w) - offsetof (struct my_biggy, t1));
		3462	}
		3463
		3464	static void
		3465	t2_cb (EV_P_ ev_timer *w, int revents)
		3466	{
		3467	struct my_biggy big = (struct my_biggy *)
		3468	(((char *)w) - offsetof (struct my_biggy, t2));
		3469	}
		3470
		3471	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3472
		3473	Often (especially in GUI toolkits) there are places where you have
		3474	I<modal> interaction, which is most easily implemented by recursively
		3475	invoking C<ev_run>.
		3476
		3477	This brings the problem of exiting - a callback might want to finish the
		3478	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3479	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3480	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3481	other combination: In these cases, C<ev_break> will not work alone.
		3482
		3483	The solution is to maintain "break this loop" variable for each C<ev_run>
		3484	invocation, and use a loop around C<ev_run> until the condition is
		3485	triggered, using C<EVRUN_ONCE>:
		3486
		3487	// main loop
		3488	int exit_main_loop = 0;
		3489
		3490	while (!exit_main_loop)
		3491	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3492
		3493	// in a model watcher
		3494	int exit_nested_loop = 0;
		3495
		3496	while (!exit_nested_loop)
		3497	ev_run (EV_A_ EVRUN_ONCE);
		3498
		3499	To exit from any of these loops, just set the corresponding exit variable:
		3500
		3501	// exit modal loop
		3502	exit_nested_loop = 1;
		3503
		3504	// exit main program, after modal loop is finished
		3505	exit_main_loop = 1;
		3506
		3507	// exit both
		3508	exit_main_loop = exit_nested_loop = 1;
		3509
		3510	=head2 THREAD LOCKING EXAMPLE
		3511
		3512	Here is a fictitious example of how to run an event loop in a different
		3513	thread from where callbacks are being invoked and watchers are
		3514	created/added/removed.
		3515
		3516	For a real-world example, see the C<EV::Loop::Async> perl module,
		3517	which uses exactly this technique (which is suited for many high-level
		3518	languages).
		3519
		3520	The example uses a pthread mutex to protect the loop data, a condition
		3521	variable to wait for callback invocations, an async watcher to notify the
		3522	event loop thread and an unspecified mechanism to wake up the main thread.
		3523
		3524	First, you need to associate some data with the event loop:
		3525
		3526	typedef struct {
		3527	mutex_t lock; /* global loop lock */
		3528	ev_async async_w;
		3529	thread_t tid;
		3530	cond_t invoke_cv;
		3531	} userdata;
		3532
		3533	void prepare_loop (EV_P)
		3534	{
		3535	// for simplicity, we use a static userdata struct.
		3536	static userdata u;
		3537
		3538	ev_async_init (&u->async_w, async_cb);
		3539	ev_async_start (EV_A_ &u->async_w);
		3540
		3541	pthread_mutex_init (&u->lock, 0);
		3542	pthread_cond_init (&u->invoke_cv, 0);
		3543
		3544	// now associate this with the loop
		3545	ev_set_userdata (EV_A_ u);
		3546	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3547	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3548
		3549	// then create the thread running ev_loop
		3550	pthread_create (&u->tid, 0, l_run, EV_A);
		3551	}
		3552
		3553	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3554	solely to wake up the event loop so it takes notice of any new watchers
		3555	that might have been added:
		3556
		3557	static void
		3558	async_cb (EV_P_ ev_async *w, int revents)
		3559	{
		3560	// just used for the side effects
		3561	}
		3562
		3563	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3564	protecting the loop data, respectively.
		3565
		3566	static void
		3567	l_release (EV_P)
		3568	{
		3569	userdata *u = ev_userdata (EV_A);
		3570	pthread_mutex_unlock (&u->lock);
		3571	}
		3572
		3573	static void
		3574	l_acquire (EV_P)
		3575	{
		3576	userdata *u = ev_userdata (EV_A);
		3577	pthread_mutex_lock (&u->lock);
		3578	}
		3579
		3580	The event loop thread first acquires the mutex, and then jumps straight
		3581	into C<ev_run>:
		3582
		3583	void *
		3584	l_run (void *thr_arg)
		3585	{
		3586	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3587
		3588	l_acquire (EV_A);
		3589	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3590	ev_run (EV_A_ 0);
		3591	l_release (EV_A);
		3592
		3593	return 0;
		3594	}
		3595
		3596	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3597	signal the main thread via some unspecified mechanism (signals? pipe
		3598	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3599	have been called (in a while loop because a) spurious wakeups are possible
		3600	and b) skipping inter-thread-communication when there are no pending
		3601	watchers is very beneficial):
		3602
		3603	static void
		3604	l_invoke (EV_P)
		3605	{
		3606	userdata *u = ev_userdata (EV_A);
		3607
		3608	while (ev_pending_count (EV_A))
		3609	{
		3610	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3611	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3612	}
		3613	}
		3614
		3615	Now, whenever the main thread gets told to invoke pending watchers, it
		3616	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3617	thread to continue:
		3618
		3619	static void
		3620	real_invoke_pending (EV_P)
		3621	{
		3622	userdata *u = ev_userdata (EV_A);
		3623
		3624	pthread_mutex_lock (&u->lock);
		3625	ev_invoke_pending (EV_A);
		3626	pthread_cond_signal (&u->invoke_cv);
		3627	pthread_mutex_unlock (&u->lock);
		3628	}
		3629
		3630	Whenever you want to start/stop a watcher or do other modifications to an
		3631	event loop, you will now have to lock:
		3632
		3633	ev_timer timeout_watcher;
		3634	userdata *u = ev_userdata (EV_A);
		3635
		3636	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3637
		3638	pthread_mutex_lock (&u->lock);
		3639	ev_timer_start (EV_A_ &timeout_watcher);
		3640	ev_async_send (EV_A_ &u->async_w);
		3641	pthread_mutex_unlock (&u->lock);
		3642
		3643	Note that sending the C<ev_async> watcher is required because otherwise
		3644	an event loop currently blocking in the kernel will have no knowledge
		3645	about the newly added timer. By waking up the loop it will pick up any new
		3646	watchers in the next event loop iteration.
		3647
		3648	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3649
		3650	While the overhead of a callback that e.g. schedules a thread is small, it
		3651	is still an overhead. If you embed libev, and your main usage is with some
		3652	kind of threads or coroutines, you might want to customise libev so that
		3653	doesn't need callbacks anymore.
		3654
		3655	Imagine you have coroutines that you can switch to using a function
		3656	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3657	and that due to some magic, the currently active coroutine is stored in a
		3658	global called C<current_coro>. Then you can build your own "wait for libev
		3659	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3660	the differing C<;> conventions):
		3661
		3662	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3663	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3664
		3665	That means instead of having a C callback function, you store the
		3666	coroutine to switch to in each watcher, and instead of having libev call
		3667	your callback, you instead have it switch to that coroutine.
		3668
		3669	A coroutine might now wait for an event with a function called
		3670	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		3671	matter when, or whether the watcher is active or not when this function is
		3672	called):
		3673
		3674	void
		3675	wait_for_event (ev_watcher *w)
		3676	{
		3677	ev_cb_set (w) = current_coro;
		3678	switch_to (libev_coro);
		3679	}
		3680
		3681	That basically suspends the coroutine inside C<wait_for_event> and
		3682	continues the libev coroutine, which, when appropriate, switches back to
		3683	this or any other coroutine. I am sure if you sue this your own :)
		3684
		3685	You can do similar tricks if you have, say, threads with an event queue -
		3686	instead of storing a coroutine, you store the queue object and instead of
		3687	switching to a coroutine, you push the watcher onto the queue and notify
		3688	any waiters.
		3689
		3690	To embed libev, see L<EMBEDDING>, but in short, it's easiest to create two
		3691	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		3692
		3693	// my_ev.h
		3694	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3695	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb);
		3696	#include "../libev/ev.h"
		3697
		3698	// my_ev.c
		3699	#define EV_H "my_ev.h"
		3700	#include "../libev/ev.c"
		3701
		3702	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		3703	F<my_ev.c> into your project. When properly specifying include paths, you
		3704	can even use F<ev.h> as header file name directly.
1284		3705
1285		3706
1286	=head1 LIBEVENT EMULATION	3707	=head1 LIBEVENT EMULATION
1287		3708
1288	Libev offers a compatibility emulation layer for libevent. It cannot	3709	Libev offers a compatibility emulation layer for libevent. It cannot
1289	emulate the internals of libevent, so here are some usage hints:	3710	emulate the internals of libevent, so here are some usage hints:
1290		3711
1291	=over 4	3712	=over 4
		3713
		3714	=item * Only the libevent-1.4.1-beta API is being emulated.
		3715
		3716	This was the newest libevent version available when libev was implemented,
		3717	and is still mostly unchanged in 2010.
1292		3718
1293	=item * Use it by including <event.h>, as usual.	3719	=item * Use it by including <event.h>, as usual.
1294		3720
1295	=item * The following members are fully supported: ev_base, ev_callback,	3721	=item * The following members are fully supported: ev_base, ev_callback,
1296	ev_arg, ev_fd, ev_res, ev_events.	3722	ev_arg, ev_fd, ev_res, ev_events.
…		…
1301		3727
1302	=item * Priorities are not currently supported. Initialising priorities	3728	=item * Priorities are not currently supported. Initialising priorities
1303	will fail and all watchers will have the same priority, even though there	3729	will fail and all watchers will have the same priority, even though there
1304	is an ev_pri field.	3730	is an ev_pri field.
1305		3731
		3732	=item * In libevent, the last base created gets the signals, in libev, the
		3733	base that registered the signal gets the signals.
		3734
1306	=item * Other members are not supported.	3735	=item * Other members are not supported.
1307		3736
1308	=item * The libev emulation is I<not> ABI compatible to libevent, you need	3737	=item * The libev emulation is I<not> ABI compatible to libevent, you need
1309	to use the libev header file and library.	3738	to use the libev header file and library.
1310		3739
1311	=back	3740	=back
1312		3741
1313	=head1 C++ SUPPORT	3742	=head1 C++ SUPPORT
1314		3743
1315	TBD.	3744	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3745	you to use some convenience methods to start/stop watchers and also change
		3746	the callback model to a model using method callbacks on objects.
		3747
		3748	To use it,
		3749
		3750	#include <ev++.h>
		3751
		3752	This automatically includes F<ev.h> and puts all of its definitions (many
		3753	of them macros) into the global namespace. All C++ specific things are
		3754	put into the C<ev> namespace. It should support all the same embedding
		3755	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3756
		3757	Care has been taken to keep the overhead low. The only data member the C++
		3758	classes add (compared to plain C-style watchers) is the event loop pointer
		3759	that the watcher is associated with (or no additional members at all if
		3760	you disable C<EV_MULTIPLICITY> when embedding libev).
		3761
		3762	Currently, functions, static and non-static member functions and classes
		3763	with C<operator ()> can be used as callbacks. Other types should be easy
		3764	to add as long as they only need one additional pointer for context. If
		3765	you need support for other types of functors please contact the author
		3766	(preferably after implementing it).
		3767
		3768	Here is a list of things available in the C<ev> namespace:
		3769
		3770	=over 4
		3771
		3772	=item C<ev::READ>, C<ev::WRITE> etc.
		3773
		3774	These are just enum values with the same values as the C<EV_READ> etc.
		3775	macros from F<ev.h>.
		3776
		3777	=item C<ev::tstamp>, C<ev::now>
		3778
		3779	Aliases to the same types/functions as with the C<ev_> prefix.
		3780
		3781	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3782
		3783	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3784	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3785	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3786	defines by many implementations.
		3787
		3788	All of those classes have these methods:
		3789
		3790	=over 4
		3791
		3792	=item ev::TYPE::TYPE ()
		3793
		3794	=item ev::TYPE::TYPE (loop)
		3795
		3796	=item ev::TYPE::~TYPE
		3797
		3798	The constructor (optionally) takes an event loop to associate the watcher
		3799	with. If it is omitted, it will use C<EV_DEFAULT>.
		3800
		3801	The constructor calls C<ev_init> for you, which means you have to call the
		3802	C<set> method before starting it.
		3803
		3804	It will not set a callback, however: You have to call the templated C<set>
		3805	method to set a callback before you can start the watcher.
		3806
		3807	(The reason why you have to use a method is a limitation in C++ which does
		3808	not allow explicit template arguments for constructors).
		3809
		3810	The destructor automatically stops the watcher if it is active.
		3811
		3812	=item w->set<class, &class::method> (object *)
		3813
		3814	This method sets the callback method to call. The method has to have a
		3815	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3816	first argument and the C<revents> as second. The object must be given as
		3817	parameter and is stored in the C<data> member of the watcher.
		3818
		3819	This method synthesizes efficient thunking code to call your method from
		3820	the C callback that libev requires. If your compiler can inline your
		3821	callback (i.e. it is visible to it at the place of the C<set> call and
		3822	your compiler is good :), then the method will be fully inlined into the
		3823	thunking function, making it as fast as a direct C callback.
		3824
		3825	Example: simple class declaration and watcher initialisation
		3826
		3827	struct myclass
		3828	{
		3829	void io_cb (ev::io &w, int revents) { }
		3830	}
		3831
		3832	myclass obj;
		3833	ev::io iow;
		3834	iow.set <myclass, &myclass::io_cb> (&obj);
		3835
		3836	=item w->set (object *)
		3837
		3838	This is a variation of a method callback - leaving out the method to call
		3839	will default the method to C<operator ()>, which makes it possible to use
		3840	functor objects without having to manually specify the C<operator ()> all
		3841	the time. Incidentally, you can then also leave out the template argument
		3842	list.
		3843
		3844	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3845	int revents)>.
		3846
		3847	See the method-C<set> above for more details.
		3848
		3849	Example: use a functor object as callback.
		3850
		3851	struct myfunctor
		3852	{
		3853	void operator() (ev::io &w, int revents)
		3854	{
		3855	...
		3856	}
		3857	}
		3858
		3859	myfunctor f;
		3860
		3861	ev::io w;
		3862	w.set (&f);
		3863
		3864	=item w->set<function> (void *data = 0)
		3865
		3866	Also sets a callback, but uses a static method or plain function as
		3867	callback. The optional C<data> argument will be stored in the watcher's
		3868	C<data> member and is free for you to use.
		3869
		3870	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3871
		3872	See the method-C<set> above for more details.
		3873
		3874	Example: Use a plain function as callback.
		3875
		3876	static void io_cb (ev::io &w, int revents) { }
		3877	iow.set <io_cb> ();
		3878
		3879	=item w->set (loop)
		3880
		3881	Associates a different C<struct ev_loop> with this watcher. You can only
		3882	do this when the watcher is inactive (and not pending either).
		3883
		3884	=item w->set ([arguments])
		3885
		3886	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this
		3887	method or a suitable start method must be called at least once. Unlike the
		3888	C counterpart, an active watcher gets automatically stopped and restarted
		3889	when reconfiguring it with this method.
		3890
		3891	=item w->start ()
		3892
		3893	Starts the watcher. Note that there is no C<loop> argument, as the
		3894	constructor already stores the event loop.
		3895
		3896	=item w->start ([arguments])
		3897
		3898	Instead of calling C<set> and C<start> methods separately, it is often
		3899	convenient to wrap them in one call. Uses the same type of arguments as
		3900	the configure C<set> method of the watcher.
		3901
		3902	=item w->stop ()
		3903
		3904	Stops the watcher if it is active. Again, no C<loop> argument.
		3905
		3906	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3907
		3908	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3909	C<ev_TYPE_again> function.
		3910
		3911	=item w->sweep () (C<ev::embed> only)
		3912
		3913	Invokes C<ev_embed_sweep>.
		3914
		3915	=item w->update () (C<ev::stat> only)
		3916
		3917	Invokes C<ev_stat_stat>.
		3918
		3919	=back
		3920
		3921	=back
		3922
		3923	Example: Define a class with two I/O and idle watchers, start the I/O
		3924	watchers in the constructor.
		3925
		3926	class myclass
		3927	{
		3928	ev::io io ; void io_cb (ev::io &w, int revents);
		3929	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);
		3930	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3931
		3932	myclass (int fd)
		3933	{
		3934	io .set <myclass, &myclass::io_cb > (this);
		3935	io2 .set <myclass, &myclass::io2_cb > (this);
		3936	idle.set <myclass, &myclass::idle_cb> (this);
		3937
		3938	io.set (fd, ev::WRITE); // configure the watcher
		3939	io.start (); // start it whenever convenient
		3940
		3941	io2.start (fd, ev::READ); // set + start in one call
		3942	}
		3943	};
		3944
		3945
		3946	=head1 OTHER LANGUAGE BINDINGS
		3947
		3948	Libev does not offer other language bindings itself, but bindings for a
		3949	number of languages exist in the form of third-party packages. If you know
		3950	any interesting language binding in addition to the ones listed here, drop
		3951	me a note.
		3952
		3953	=over 4
		3954
		3955	=item Perl
		3956
		3957	The EV module implements the full libev API and is actually used to test
		3958	libev. EV is developed together with libev. Apart from the EV core module,
		3959	there are additional modules that implement libev-compatible interfaces
		3960	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3961	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3962	and C<EV::Glib>).
		3963
		3964	It can be found and installed via CPAN, its homepage is at
		3965	L<http://software.schmorp.de/pkg/EV>.
		3966
		3967	=item Python
		3968
		3969	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3970	seems to be quite complete and well-documented.
		3971
		3972	=item Ruby
		3973
		3974	Tony Arcieri has written a ruby extension that offers access to a subset
		3975	of the libev API and adds file handle abstractions, asynchronous DNS and
		3976	more on top of it. It can be found via gem servers. Its homepage is at
		3977	L<http://rev.rubyforge.org/>.
		3978
		3979	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3980	makes rev work even on mingw.
		3981
		3982	=item Haskell
		3983
		3984	A haskell binding to libev is available at
		3985	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3986
		3987	=item D
		3988
		3989	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3990	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3991
		3992	=item Ocaml
		3993
		3994	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3995	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3996
		3997	=item Lua
		3998
		3999	Brian Maher has written a partial interface to libev for lua (at the
		4000	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		4001	L<http://github.com/brimworks/lua-ev>.
		4002
		4003	=back
		4004
		4005
		4006	=head1 MACRO MAGIC
		4007
		4008	Libev can be compiled with a variety of options, the most fundamental
		4009	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		4010	functions and callbacks have an initial C<struct ev_loop *> argument.
		4011
		4012	To make it easier to write programs that cope with either variant, the
		4013	following macros are defined:
		4014
		4015	=over 4
		4016
		4017	=item C<EV_A>, C<EV_A_>
		4018
		4019	This provides the loop I<argument> for functions, if one is required ("ev
		4020	loop argument"). The C<EV_A> form is used when this is the sole argument,
		4021	C<EV_A_> is used when other arguments are following. Example:
		4022
		4023	ev_unref (EV_A);
		4024	ev_timer_add (EV_A_ watcher);
		4025	ev_run (EV_A_ 0);
		4026
		4027	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		4028	which is often provided by the following macro.
		4029
		4030	=item C<EV_P>, C<EV_P_>
		4031
		4032	This provides the loop I<parameter> for functions, if one is required ("ev
		4033	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		4034	C<EV_P_> is used when other parameters are following. Example:
		4035
		4036	// this is how ev_unref is being declared
		4037	static void ev_unref (EV_P);
		4038
		4039	// this is how you can declare your typical callback
		4040	static void cb (EV_P_ ev_timer *w, int revents)
		4041
		4042	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		4043	suitable for use with C<EV_A>.
		4044
		4045	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		4046
		4047	Similar to the other two macros, this gives you the value of the default
		4048	loop, if multiple loops are supported ("ev loop default").
		4049
		4050	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		4051
		4052	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		4053	default loop has been initialised (C<UC> == unchecked). Their behaviour
		4054	is undefined when the default loop has not been initialised by a previous
		4055	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		4056
		4057	It is often prudent to use C<EV_DEFAULT> when initialising the first
		4058	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		4059
		4060	=back
		4061
		4062	Example: Declare and initialise a check watcher, utilising the above
		4063	macros so it will work regardless of whether multiple loops are supported
		4064	or not.
		4065
		4066	static void
		4067	check_cb (EV_P_ ev_timer *w, int revents)
		4068	{
		4069	ev_check_stop (EV_A_ w);
		4070	}
		4071
		4072	ev_check check;
		4073	ev_check_init (&check, check_cb);
		4074	ev_check_start (EV_DEFAULT_ &check);
		4075	ev_run (EV_DEFAULT_ 0);
		4076
		4077	=head1 EMBEDDING
		4078
		4079	Libev can (and often is) directly embedded into host
		4080	applications. Examples of applications that embed it include the Deliantra
		4081	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		4082	and rxvt-unicode.
		4083
		4084	The goal is to enable you to just copy the necessary files into your
		4085	source directory without having to change even a single line in them, so
		4086	you can easily upgrade by simply copying (or having a checked-out copy of
		4087	libev somewhere in your source tree).
		4088
		4089	=head2 FILESETS
		4090
		4091	Depending on what features you need you need to include one or more sets of files
		4092	in your application.
		4093
		4094	=head3 CORE EVENT LOOP
		4095
		4096	To include only the libev core (all the C<ev_*> functions), with manual
		4097	configuration (no autoconf):
		4098
		4099	#define EV_STANDALONE 1
		4100	#include "ev.c"
		4101
		4102	This will automatically include F<ev.h>, too, and should be done in a
		4103	single C source file only to provide the function implementations. To use
		4104	it, do the same for F<ev.h> in all files wishing to use this API (best
		4105	done by writing a wrapper around F<ev.h> that you can include instead and
		4106	where you can put other configuration options):
		4107
		4108	#define EV_STANDALONE 1
		4109	#include "ev.h"
		4110
		4111	Both header files and implementation files can be compiled with a C++
		4112	compiler (at least, that's a stated goal, and breakage will be treated
		4113	as a bug).
		4114
		4115	You need the following files in your source tree, or in a directory
		4116	in your include path (e.g. in libev/ when using -Ilibev):
		4117
		4118	ev.h
		4119	ev.c
		4120	ev_vars.h
		4121	ev_wrap.h
		4122
		4123	ev_win32.c required on win32 platforms only
		4124
		4125	ev_select.c only when select backend is enabled (which is enabled by default)
		4126	ev_poll.c only when poll backend is enabled (disabled by default)
		4127	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		4128	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		4129	ev_port.c only when the solaris port backend is enabled (disabled by default)
		4130
		4131	F<ev.c> includes the backend files directly when enabled, so you only need
		4132	to compile this single file.
		4133
		4134	=head3 LIBEVENT COMPATIBILITY API
		4135
		4136	To include the libevent compatibility API, also include:
		4137
		4138	#include "event.c"
		4139
		4140	in the file including F<ev.c>, and:
		4141
		4142	#include "event.h"
		4143
		4144	in the files that want to use the libevent API. This also includes F<ev.h>.
		4145
		4146	You need the following additional files for this:
		4147
		4148	event.h
		4149	event.c
		4150
		4151	=head3 AUTOCONF SUPPORT
		4152
		4153	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		4154	whatever way you want, you can also C<m4_include([libev.m4])> in your
		4155	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		4156	include F<config.h> and configure itself accordingly.
		4157
		4158	For this of course you need the m4 file:
		4159
		4160	libev.m4
		4161
		4162	=head2 PREPROCESSOR SYMBOLS/MACROS
		4163
		4164	Libev can be configured via a variety of preprocessor symbols you have to
		4165	define before including (or compiling) any of its files. The default in
		4166	the absence of autoconf is documented for every option.
		4167
		4168	Symbols marked with "(h)" do not change the ABI, and can have different
		4169	values when compiling libev vs. including F<ev.h>, so it is permissible
		4170	to redefine them before including F<ev.h> without breaking compatibility
		4171	to a compiled library. All other symbols change the ABI, which means all
		4172	users of libev and the libev code itself must be compiled with compatible
		4173	settings.
		4174
		4175	=over 4
		4176
		4177	=item EV_COMPAT3 (h)
		4178
		4179	Backwards compatibility is a major concern for libev. This is why this
		4180	release of libev comes with wrappers for the functions and symbols that
		4181	have been renamed between libev version 3 and 4.
		4182
		4183	You can disable these wrappers (to test compatibility with future
		4184	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		4185	sources. This has the additional advantage that you can drop the C<struct>
		4186	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		4187	typedef in that case.
		4188
		4189	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		4190	and in some even more future version the compatibility code will be
		4191	removed completely.
		4192
		4193	=item EV_STANDALONE (h)
		4194
		4195	Must always be C<1> if you do not use autoconf configuration, which
		4196	keeps libev from including F<config.h>, and it also defines dummy
		4197	implementations for some libevent functions (such as logging, which is not
		4198	supported). It will also not define any of the structs usually found in
		4199	F<event.h> that are not directly supported by the libev core alone.
		4200
		4201	In standalone mode, libev will still try to automatically deduce the
		4202	configuration, but has to be more conservative.
		4203
		4204	=item EV_USE_MONOTONIC
		4205
		4206	If defined to be C<1>, libev will try to detect the availability of the
		4207	monotonic clock option at both compile time and runtime. Otherwise no
		4208	use of the monotonic clock option will be attempted. If you enable this,
		4209	you usually have to link against librt or something similar. Enabling it
		4210	when the functionality isn't available is safe, though, although you have
		4211	to make sure you link against any libraries where the C<clock_gettime>
		4212	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		4213
		4214	=item EV_USE_REALTIME
		4215
		4216	If defined to be C<1>, libev will try to detect the availability of the
		4217	real-time clock option at compile time (and assume its availability
		4218	at runtime if successful). Otherwise no use of the real-time clock
		4219	option will be attempted. This effectively replaces C<gettimeofday>
		4220	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		4221	correctness. See the note about libraries in the description of
		4222	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		4223	C<EV_USE_CLOCK_SYSCALL>.
		4224
		4225	=item EV_USE_CLOCK_SYSCALL
		4226
		4227	If defined to be C<1>, libev will try to use a direct syscall instead
		4228	of calling the system-provided C<clock_gettime> function. This option
		4229	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		4230	unconditionally pulls in C<libpthread>, slowing down single-threaded
		4231	programs needlessly. Using a direct syscall is slightly slower (in
		4232	theory), because no optimised vdso implementation can be used, but avoids
		4233	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		4234	higher, as it simplifies linking (no need for C<-lrt>).
		4235
		4236	=item EV_USE_NANOSLEEP
		4237
		4238	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		4239	and will use it for delays. Otherwise it will use C<select ()>.
		4240
		4241	=item EV_USE_EVENTFD
		4242
		4243	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		4244	available and will probe for kernel support at runtime. This will improve
		4245	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		4246	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		4247	2.7 or newer, otherwise disabled.
		4248
		4249	=item EV_USE_SELECT
		4250
		4251	If undefined or defined to be C<1>, libev will compile in support for the
		4252	C<select>(2) backend. No attempt at auto-detection will be done: if no
		4253	other method takes over, select will be it. Otherwise the select backend
		4254	will not be compiled in.
		4255
		4256	=item EV_SELECT_USE_FD_SET
		4257
		4258	If defined to C<1>, then the select backend will use the system C<fd_set>
		4259	structure. This is useful if libev doesn't compile due to a missing
		4260	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		4261	on exotic systems. This usually limits the range of file descriptors to
		4262	some low limit such as 1024 or might have other limitations (winsocket
		4263	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		4264	configures the maximum size of the C<fd_set>.
		4265
		4266	=item EV_SELECT_IS_WINSOCKET
		4267
		4268	When defined to C<1>, the select backend will assume that
		4269	select/socket/connect etc. don't understand file descriptors but
		4270	wants osf handles on win32 (this is the case when the select to
		4271	be used is the winsock select). This means that it will call
		4272	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		4273	it is assumed that all these functions actually work on fds, even
		4274	on win32. Should not be defined on non-win32 platforms.
		4275
		4276	=item EV_FD_TO_WIN32_HANDLE(fd)
		4277
		4278	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		4279	file descriptors to socket handles. When not defining this symbol (the
		4280	default), then libev will call C<_get_osfhandle>, which is usually
		4281	correct. In some cases, programs use their own file descriptor management,
		4282	in which case they can provide this function to map fds to socket handles.
		4283
		4284	=item EV_WIN32_HANDLE_TO_FD(handle)
		4285
		4286	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		4287	using the standard C<_open_osfhandle> function. For programs implementing
		4288	their own fd to handle mapping, overwriting this function makes it easier
		4289	to do so. This can be done by defining this macro to an appropriate value.
		4290
		4291	=item EV_WIN32_CLOSE_FD(fd)
		4292
		4293	If programs implement their own fd to handle mapping on win32, then this
		4294	macro can be used to override the C<close> function, useful to unregister
		4295	file descriptors again. Note that the replacement function has to close
		4296	the underlying OS handle.
		4297
		4298	=item EV_USE_POLL
		4299
		4300	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		4301	backend. Otherwise it will be enabled on non-win32 platforms. It
		4302	takes precedence over select.
		4303
		4304	=item EV_USE_EPOLL
		4305
		4306	If defined to be C<1>, libev will compile in support for the Linux
		4307	C<epoll>(7) backend. Its availability will be detected at runtime,
		4308	otherwise another method will be used as fallback. This is the preferred
		4309	backend for GNU/Linux systems. If undefined, it will be enabled if the
		4310	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4311
		4312	=item EV_USE_KQUEUE
		4313
		4314	If defined to be C<1>, libev will compile in support for the BSD style
		4315	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		4316	otherwise another method will be used as fallback. This is the preferred
		4317	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		4318	supports some types of fds correctly (the only platform we found that
		4319	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		4320	not be used unless explicitly requested. The best way to use it is to find
		4321	out whether kqueue supports your type of fd properly and use an embedded
		4322	kqueue loop.
		4323
		4324	=item EV_USE_PORT
		4325
		4326	If defined to be C<1>, libev will compile in support for the Solaris
		4327	10 port style backend. Its availability will be detected at runtime,
		4328	otherwise another method will be used as fallback. This is the preferred
		4329	backend for Solaris 10 systems.
		4330
		4331	=item EV_USE_DEVPOLL
		4332
		4333	Reserved for future expansion, works like the USE symbols above.
		4334
		4335	=item EV_USE_INOTIFY
		4336
		4337	If defined to be C<1>, libev will compile in support for the Linux inotify
		4338	interface to speed up C<ev_stat> watchers. Its actual availability will
		4339	be detected at runtime. If undefined, it will be enabled if the headers
		4340	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4341
		4342	=item EV_ATOMIC_T
		4343
		4344	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		4345	access is atomic with respect to other threads or signal contexts. No such
		4346	type is easily found in the C language, so you can provide your own type
		4347	that you know is safe for your purposes. It is used both for signal handler "locking"
		4348	as well as for signal and thread safety in C<ev_async> watchers.
		4349
		4350	In the absence of this define, libev will use C<sig_atomic_t volatile>
		4351	(from F<signal.h>), which is usually good enough on most platforms.
		4352
		4353	=item EV_H (h)
		4354
		4355	The name of the F<ev.h> header file used to include it. The default if
		4356	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		4357	used to virtually rename the F<ev.h> header file in case of conflicts.
		4358
		4359	=item EV_CONFIG_H (h)
		4360
		4361	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		4362	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		4363	C<EV_H>, above.
		4364
		4365	=item EV_EVENT_H (h)
		4366
		4367	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		4368	of how the F<event.h> header can be found, the default is C<"event.h">.
		4369
		4370	=item EV_PROTOTYPES (h)
		4371
		4372	If defined to be C<0>, then F<ev.h> will not define any function
		4373	prototypes, but still define all the structs and other symbols. This is
		4374	occasionally useful if you want to provide your own wrapper functions
		4375	around libev functions.
		4376
		4377	=item EV_MULTIPLICITY
		4378
		4379	If undefined or defined to C<1>, then all event-loop-specific functions
		4380	will have the C<struct ev_loop *> as first argument, and you can create
		4381	additional independent event loops. Otherwise there will be no support
		4382	for multiple event loops and there is no first event loop pointer
		4383	argument. Instead, all functions act on the single default loop.
		4384
		4385	=item EV_MINPRI
		4386
		4387	=item EV_MAXPRI
		4388
		4389	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		4390	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		4391	provide for more priorities by overriding those symbols (usually defined
		4392	to be C<-2> and C<2>, respectively).
		4393
		4394	When doing priority-based operations, libev usually has to linearly search
		4395	all the priorities, so having many of them (hundreds) uses a lot of space
		4396	and time, so using the defaults of five priorities (-2 .. +2) is usually
		4397	fine.
		4398
		4399	If your embedding application does not need any priorities, defining these
		4400	both to C<0> will save some memory and CPU.
		4401
		4402	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		4403	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		4404	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
		4405
		4406	If undefined or defined to be C<1> (and the platform supports it), then
		4407	the respective watcher type is supported. If defined to be C<0>, then it
		4408	is not. Disabling watcher types mainly saves code size.
		4409
		4410	=item EV_FEATURES
		4411
		4412	If you need to shave off some kilobytes of code at the expense of some
		4413	speed (but with the full API), you can define this symbol to request
		4414	certain subsets of functionality. The default is to enable all features
		4415	that can be enabled on the platform.
		4416
		4417	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4418	with some broad features you want) and then selectively re-enable
		4419	additional parts you want, for example if you want everything minimal,
		4420	but multiple event loop support, async and child watchers and the poll
		4421	backend, use this:
		4422
		4423	#define EV_FEATURES 0
		4424	#define EV_MULTIPLICITY 1
		4425	#define EV_USE_POLL 1
		4426	#define EV_CHILD_ENABLE 1
		4427	#define EV_ASYNC_ENABLE 1
		4428
		4429	The actual value is a bitset, it can be a combination of the following
		4430	values:
		4431
		4432	=over 4
		4433
		4434	=item C<1> - faster/larger code
		4435
		4436	Use larger code to speed up some operations.
		4437
		4438	Currently this is used to override some inlining decisions (enlarging the
		4439	code size by roughly 30% on amd64).
		4440
		4441	When optimising for size, use of compiler flags such as C<-Os> with
		4442	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4443	assertions.
		4444
		4445	=item C<2> - faster/larger data structures
		4446
		4447	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4448	hash table sizes and so on. This will usually further increase code size
		4449	and can additionally have an effect on the size of data structures at
		4450	runtime.
		4451
		4452	=item C<4> - full API configuration
		4453
		4454	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4455	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4456
		4457	=item C<8> - full API
		4458
		4459	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4460	details on which parts of the API are still available without this
		4461	feature, and do not complain if this subset changes over time.
		4462
		4463	=item C<16> - enable all optional watcher types
		4464
		4465	Enables all optional watcher types. If you want to selectively enable
		4466	only some watcher types other than I/O and timers (e.g. prepare,
		4467	embed, async, child...) you can enable them manually by defining
		4468	C<EV_watchertype_ENABLE> to C<1> instead.
		4469
		4470	=item C<32> - enable all backends
		4471
		4472	This enables all backends - without this feature, you need to enable at
		4473	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4474
		4475	=item C<64> - enable OS-specific "helper" APIs
		4476
		4477	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4478	default.
		4479
		4480	=back
		4481
		4482	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4483	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4484	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4485	watchers, timers and monotonic clock support.
		4486
		4487	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4488	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4489	your program might be left out as well - a binary starting a timer and an
		4490	I/O watcher then might come out at only 5Kb.
		4491
		4492	=item EV_AVOID_STDIO
		4493
		4494	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4495	functions (printf, scanf, perror etc.). This will increase the code size
		4496	somewhat, but if your program doesn't otherwise depend on stdio and your
		4497	libc allows it, this avoids linking in the stdio library which is quite
		4498	big.
		4499
		4500	Note that error messages might become less precise when this option is
		4501	enabled.
		4502
		4503	=item EV_NSIG
		4504
		4505	The highest supported signal number, +1 (or, the number of
		4506	signals): Normally, libev tries to deduce the maximum number of signals
		4507	automatically, but sometimes this fails, in which case it can be
		4508	specified. Also, using a lower number than detected (C<32> should be
		4509	good for about any system in existence) can save some memory, as libev
		4510	statically allocates some 12-24 bytes per signal number.
		4511
		4512	=item EV_PID_HASHSIZE
		4513
		4514	C<ev_child> watchers use a small hash table to distribute workload by
		4515	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
		4516	usually more than enough. If you need to manage thousands of children you
		4517	might want to increase this value (I<must> be a power of two).
		4518
		4519	=item EV_INOTIFY_HASHSIZE
		4520
		4521	C<ev_stat> watchers use a small hash table to distribute workload by
		4522	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
		4523	disabled), usually more than enough. If you need to manage thousands of
		4524	C<ev_stat> watchers you might want to increase this value (I<must> be a
		4525	power of two).
		4526
		4527	=item EV_USE_4HEAP
		4528
		4529	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4530	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4531	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4532	faster performance with many (thousands) of watchers.
		4533
		4534	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4535	will be C<0>.
		4536
		4537	=item EV_HEAP_CACHE_AT
		4538
		4539	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4540	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4541	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4542	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4543	but avoids random read accesses on heap changes. This improves performance
		4544	noticeably with many (hundreds) of watchers.
		4545
		4546	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4547	will be C<0>.
		4548
		4549	=item EV_VERIFY
		4550
		4551	Controls how much internal verification (see C<ev_verify ()>) will
		4552	be done: If set to C<0>, no internal verification code will be compiled
		4553	in. If set to C<1>, then verification code will be compiled in, but not
		4554	called. If set to C<2>, then the internal verification code will be
		4555	called once per loop, which can slow down libev. If set to C<3>, then the
		4556	verification code will be called very frequently, which will slow down
		4557	libev considerably.
		4558
		4559	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4560	will be C<0>.
		4561
		4562	=item EV_COMMON
		4563
		4564	By default, all watchers have a C<void *data> member. By redefining
		4565	this macro to something else you can include more and other types of
		4566	members. You have to define it each time you include one of the files,
		4567	though, and it must be identical each time.
		4568
		4569	For example, the perl EV module uses something like this:
		4570
		4571	#define EV_COMMON \
		4572	SV *self; /* contains this struct */ \
		4573	SV *cb_sv, *fh /* note no trailing ";" */
		4574
		4575	=item EV_CB_DECLARE (type)
		4576
		4577	=item EV_CB_INVOKE (watcher, revents)
		4578
		4579	=item ev_set_cb (ev, cb)
		4580
		4581	Can be used to change the callback member declaration in each watcher,
		4582	and the way callbacks are invoked and set. Must expand to a struct member
		4583	definition and a statement, respectively. See the F<ev.h> header file for
		4584	their default definitions. One possible use for overriding these is to
		4585	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		4586	method calls instead of plain function calls in C++.
		4587
		4588	=back
		4589
		4590	=head2 EXPORTED API SYMBOLS
		4591
		4592	If you need to re-export the API (e.g. via a DLL) and you need a list of
		4593	exported symbols, you can use the provided F<Symbol.*> files which list
		4594	all public symbols, one per line:
		4595
		4596	Symbols.ev for libev proper
		4597	Symbols.event for the libevent emulation
		4598
		4599	This can also be used to rename all public symbols to avoid clashes with
		4600	multiple versions of libev linked together (which is obviously bad in
		4601	itself, but sometimes it is inconvenient to avoid this).
		4602
		4603	A sed command like this will create wrapper C<#define>'s that you need to
		4604	include before including F<ev.h>:
		4605
		4606	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		4607
		4608	This would create a file F<wrap.h> which essentially looks like this:
		4609
		4610	#define ev_backend myprefix_ev_backend
		4611	#define ev_check_start myprefix_ev_check_start
		4612	#define ev_check_stop myprefix_ev_check_stop
		4613	...
		4614
		4615	=head2 EXAMPLES
		4616
		4617	For a real-world example of a program the includes libev
		4618	verbatim, you can have a look at the EV perl module
		4619	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		4620	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		4621	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		4622	will be compiled. It is pretty complex because it provides its own header
		4623	file.
		4624
		4625	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		4626	that everybody includes and which overrides some configure choices:
		4627
		4628	#define EV_FEATURES 8
		4629	#define EV_USE_SELECT 1
		4630	#define EV_PREPARE_ENABLE 1
		4631	#define EV_IDLE_ENABLE 1
		4632	#define EV_SIGNAL_ENABLE 1
		4633	#define EV_CHILD_ENABLE 1
		4634	#define EV_USE_STDEXCEPT 0
		4635	#define EV_CONFIG_H <config.h>
		4636
		4637	#include "ev++.h"
		4638
		4639	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4640
		4641	#include "ev_cpp.h"
		4642	#include "ev.c"
		4643
		4644	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
		4645
		4646	=head2 THREADS AND COROUTINES
		4647
		4648	=head3 THREADS
		4649
		4650	All libev functions are reentrant and thread-safe unless explicitly
		4651	documented otherwise, but libev implements no locking itself. This means
		4652	that you can use as many loops as you want in parallel, as long as there
		4653	are no concurrent calls into any libev function with the same loop
		4654	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4655	of course): libev guarantees that different event loops share no data
		4656	structures that need any locking.
		4657
		4658	Or to put it differently: calls with different loop parameters can be done
		4659	concurrently from multiple threads, calls with the same loop parameter
		4660	must be done serially (but can be done from different threads, as long as
		4661	only one thread ever is inside a call at any point in time, e.g. by using
		4662	a mutex per loop).
		4663
		4664	Specifically to support threads (and signal handlers), libev implements
		4665	so-called C<ev_async> watchers, which allow some limited form of
		4666	concurrency on the same event loop, namely waking it up "from the
		4667	outside".
		4668
		4669	If you want to know which design (one loop, locking, or multiple loops
		4670	without or something else still) is best for your problem, then I cannot
		4671	help you, but here is some generic advice:
		4672
		4673	=over 4
		4674
		4675	=item * most applications have a main thread: use the default libev loop
		4676	in that thread, or create a separate thread running only the default loop.
		4677
		4678	This helps integrating other libraries or software modules that use libev
		4679	themselves and don't care/know about threading.
		4680
		4681	=item * one loop per thread is usually a good model.
		4682
		4683	Doing this is almost never wrong, sometimes a better-performance model
		4684	exists, but it is always a good start.
		4685
		4686	=item * other models exist, such as the leader/follower pattern, where one
		4687	loop is handed through multiple threads in a kind of round-robin fashion.
		4688
		4689	Choosing a model is hard - look around, learn, know that usually you can do
		4690	better than you currently do :-)
		4691
		4692	=item * often you need to talk to some other thread which blocks in the
		4693	event loop.
		4694
		4695	C<ev_async> watchers can be used to wake them up from other threads safely
		4696	(or from signal contexts...).
		4697
		4698	An example use would be to communicate signals or other events that only
		4699	work in the default loop by registering the signal watcher with the
		4700	default loop and triggering an C<ev_async> watcher from the default loop
		4701	watcher callback into the event loop interested in the signal.
		4702
		4703	=back
		4704
		4705	See also L<THREAD LOCKING EXAMPLE>.
		4706
		4707	=head3 COROUTINES
		4708
		4709	Libev is very accommodating to coroutines ("cooperative threads"):
		4710	libev fully supports nesting calls to its functions from different
		4711	coroutines (e.g. you can call C<ev_run> on the same loop from two
		4712	different coroutines, and switch freely between both coroutines running
		4713	the loop, as long as you don't confuse yourself). The only exception is
		4714	that you must not do this from C<ev_periodic> reschedule callbacks.
		4715
		4716	Care has been taken to ensure that libev does not keep local state inside
		4717	C<ev_run>, and other calls do not usually allow for coroutine switches as
		4718	they do not call any callbacks.
		4719
		4720	=head2 COMPILER WARNINGS
		4721
		4722	Depending on your compiler and compiler settings, you might get no or a
		4723	lot of warnings when compiling libev code. Some people are apparently
		4724	scared by this.
		4725
		4726	However, these are unavoidable for many reasons. For one, each compiler
		4727	has different warnings, and each user has different tastes regarding
		4728	warning options. "Warn-free" code therefore cannot be a goal except when
		4729	targeting a specific compiler and compiler-version.
		4730
		4731	Another reason is that some compiler warnings require elaborate
		4732	workarounds, or other changes to the code that make it less clear and less
		4733	maintainable.
		4734
		4735	And of course, some compiler warnings are just plain stupid, or simply
		4736	wrong (because they don't actually warn about the condition their message
		4737	seems to warn about). For example, certain older gcc versions had some
		4738	warnings that resulted in an extreme number of false positives. These have
		4739	been fixed, but some people still insist on making code warn-free with
		4740	such buggy versions.
		4741
		4742	While libev is written to generate as few warnings as possible,
		4743	"warn-free" code is not a goal, and it is recommended not to build libev
		4744	with any compiler warnings enabled unless you are prepared to cope with
		4745	them (e.g. by ignoring them). Remember that warnings are just that:
		4746	warnings, not errors, or proof of bugs.
		4747
		4748
		4749	=head2 VALGRIND
		4750
		4751	Valgrind has a special section here because it is a popular tool that is
		4752	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4753
		4754	If you think you found a bug (memory leak, uninitialised data access etc.)
		4755	in libev, then check twice: If valgrind reports something like:
		4756
		4757	==2274== definitely lost: 0 bytes in 0 blocks.
		4758	==2274== possibly lost: 0 bytes in 0 blocks.
		4759	==2274== still reachable: 256 bytes in 1 blocks.
		4760
		4761	Then there is no memory leak, just as memory accounted to global variables
		4762	is not a memleak - the memory is still being referenced, and didn't leak.
		4763
		4764	Similarly, under some circumstances, valgrind might report kernel bugs
		4765	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4766	although an acceptable workaround has been found here), or it might be
		4767	confused.
		4768
		4769	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4770	make it into some kind of religion.
		4771
		4772	If you are unsure about something, feel free to contact the mailing list
		4773	with the full valgrind report and an explanation on why you think this
		4774	is a bug in libev (best check the archives, too :). However, don't be
		4775	annoyed when you get a brisk "this is no bug" answer and take the chance
		4776	of learning how to interpret valgrind properly.
		4777
		4778	If you need, for some reason, empty reports from valgrind for your project
		4779	I suggest using suppression lists.
		4780
		4781
		4782	=head1 PORTABILITY NOTES
		4783
		4784	=head2 GNU/LINUX 32 BIT LIMITATIONS
		4785
		4786	GNU/Linux is the only common platform that supports 64 bit file/large file
		4787	interfaces but I<disables> them by default.
		4788
		4789	That means that libev compiled in the default environment doesn't support
		4790	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		4791
		4792	Unfortunately, many programs try to work around this GNU/Linux issue
		4793	by enabling the large file API, which makes them incompatible with the
		4794	standard libev compiled for their system.
		4795
		4796	Likewise, libev cannot enable the large file API itself as this would
		4797	suddenly make it incompatible to the default compile time environment,
		4798	i.e. all programs not using special compile switches.
		4799
		4800	=head2 OS/X AND DARWIN BUGS
		4801
		4802	The whole thing is a bug if you ask me - basically any system interface
		4803	you touch is broken, whether it is locales, poll, kqueue or even the
		4804	OpenGL drivers.
		4805
		4806	=head3 C<kqueue> is buggy
		4807
		4808	The kqueue syscall is broken in all known versions - most versions support
		4809	only sockets, many support pipes.
		4810
		4811	Libev tries to work around this by not using C<kqueue> by default on this
		4812	rotten platform, but of course you can still ask for it when creating a
		4813	loop - embedding a socket-only kqueue loop into a select-based one is
		4814	probably going to work well.
		4815
		4816	=head3 C<poll> is buggy
		4817
		4818	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		4819	implementation by something calling C<kqueue> internally around the 10.5.6
		4820	release, so now C<kqueue> I<and> C<poll> are broken.
		4821
		4822	Libev tries to work around this by not using C<poll> by default on
		4823	this rotten platform, but of course you can still ask for it when creating
		4824	a loop.
		4825
		4826	=head3 C<select> is buggy
		4827
		4828	All that's left is C<select>, and of course Apple found a way to fuck this
		4829	one up as well: On OS/X, C<select> actively limits the number of file
		4830	descriptors you can pass in to 1024 - your program suddenly crashes when
		4831	you use more.
		4832
		4833	There is an undocumented "workaround" for this - defining
		4834	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		4835	work on OS/X.
		4836
		4837	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		4838
		4839	=head3 C<errno> reentrancy
		4840
		4841	The default compile environment on Solaris is unfortunately so
		4842	thread-unsafe that you can't even use components/libraries compiled
		4843	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		4844	defined by default. A valid, if stupid, implementation choice.
		4845
		4846	If you want to use libev in threaded environments you have to make sure
		4847	it's compiled with C<_REENTRANT> defined.
		4848
		4849	=head3 Event port backend
		4850
		4851	The scalable event interface for Solaris is called "event
		4852	ports". Unfortunately, this mechanism is very buggy in all major
		4853	releases. If you run into high CPU usage, your program freezes or you get
		4854	a large number of spurious wakeups, make sure you have all the relevant
		4855	and latest kernel patches applied. No, I don't know which ones, but there
		4856	are multiple ones to apply, and afterwards, event ports actually work
		4857	great.
		4858
		4859	If you can't get it to work, you can try running the program by setting
		4860	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		4861	C<select> backends.
		4862
		4863	=head2 AIX POLL BUG
		4864
		4865	AIX unfortunately has a broken C<poll.h> header. Libev works around
		4866	this by trying to avoid the poll backend altogether (i.e. it's not even
		4867	compiled in), which normally isn't a big problem as C<select> works fine
		4868	with large bitsets on AIX, and AIX is dead anyway.
		4869
		4870	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4871
		4872	=head3 General issues
		4873
		4874	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		4875	requires, and its I/O model is fundamentally incompatible with the POSIX
		4876	model. Libev still offers limited functionality on this platform in
		4877	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		4878	descriptors. This only applies when using Win32 natively, not when using
		4879	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		4880	as every compielr comes with a slightly differently broken/incompatible
		4881	environment.
		4882
		4883	Lifting these limitations would basically require the full
		4884	re-implementation of the I/O system. If you are into this kind of thing,
		4885	then note that glib does exactly that for you in a very portable way (note
		4886	also that glib is the slowest event library known to man).
		4887
		4888	There is no supported compilation method available on windows except
		4889	embedding it into other applications.
		4890
		4891	Sensible signal handling is officially unsupported by Microsoft - libev
		4892	tries its best, but under most conditions, signals will simply not work.
		4893
		4894	Not a libev limitation but worth mentioning: windows apparently doesn't
		4895	accept large writes: instead of resulting in a partial write, windows will
		4896	either accept everything or return C<ENOBUFS> if the buffer is too large,
		4897	so make sure you only write small amounts into your sockets (less than a
		4898	megabyte seems safe, but this apparently depends on the amount of memory
		4899	available).
		4900
		4901	Due to the many, low, and arbitrary limits on the win32 platform and
		4902	the abysmal performance of winsockets, using a large number of sockets
		4903	is not recommended (and not reasonable). If your program needs to use
		4904	more than a hundred or so sockets, then likely it needs to use a totally
		4905	different implementation for windows, as libev offers the POSIX readiness
		4906	notification model, which cannot be implemented efficiently on windows
		4907	(due to Microsoft monopoly games).
		4908
		4909	A typical way to use libev under windows is to embed it (see the embedding
		4910	section for details) and use the following F<evwrap.h> header file instead
		4911	of F<ev.h>:
		4912
		4913	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4914	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4915
		4916	#include "ev.h"
		4917
		4918	And compile the following F<evwrap.c> file into your project (make sure
		4919	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4920
		4921	#include "evwrap.h"
		4922	#include "ev.c"
		4923
		4924	=head3 The winsocket C<select> function
		4925
		4926	The winsocket C<select> function doesn't follow POSIX in that it
		4927	requires socket I<handles> and not socket I<file descriptors> (it is
		4928	also extremely buggy). This makes select very inefficient, and also
		4929	requires a mapping from file descriptors to socket handles (the Microsoft
		4930	C runtime provides the function C<_open_osfhandle> for this). See the
		4931	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		4932	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		4933
		4934	The configuration for a "naked" win32 using the Microsoft runtime
		4935	libraries and raw winsocket select is:
		4936
		4937	#define EV_USE_SELECT 1
		4938	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		4939
		4940	Note that winsockets handling of fd sets is O(n), so you can easily get a
		4941	complexity in the O(n²) range when using win32.
		4942
		4943	=head3 Limited number of file descriptors
		4944
		4945	Windows has numerous arbitrary (and low) limits on things.
		4946
		4947	Early versions of winsocket's select only supported waiting for a maximum
		4948	of C<64> handles (probably owning to the fact that all windows kernels
		4949	can only wait for C<64> things at the same time internally; Microsoft
		4950	recommends spawning a chain of threads and wait for 63 handles and the
		4951	previous thread in each. Sounds great!).
		4952
		4953	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		4954	to some high number (e.g. C<2048>) before compiling the winsocket select
		4955	call (which might be in libev or elsewhere, for example, perl and many
		4956	other interpreters do their own select emulation on windows).
		4957
		4958	Another limit is the number of file descriptors in the Microsoft runtime
		4959	libraries, which by default is C<64> (there must be a hidden I<64>
		4960	fetish or something like this inside Microsoft). You can increase this
		4961	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		4962	(another arbitrary limit), but is broken in many versions of the Microsoft
		4963	runtime libraries. This might get you to about C<512> or C<2048> sockets
		4964	(depending on windows version and/or the phase of the moon). To get more,
		4965	you need to wrap all I/O functions and provide your own fd management, but
		4966	the cost of calling select (O(n²)) will likely make this unworkable.
		4967
		4968	=head2 PORTABILITY REQUIREMENTS
		4969
		4970	In addition to a working ISO-C implementation and of course the
		4971	backend-specific APIs, libev relies on a few additional extensions:
		4972
		4973	=over 4
		4974
		4975	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4976	calling conventions regardless of C<ev_watcher_type *>.
		4977
		4978	Libev assumes not only that all watcher pointers have the same internal
		4979	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4980	assumes that the same (machine) code can be used to call any watcher
		4981	callback: The watcher callbacks have different type signatures, but libev
		4982	calls them using an C<ev_watcher *> internally.
		4983
		4984	=item pointer accesses must be thread-atomic
		4985
		4986	Accessing a pointer value must be atomic, it must both be readable and
		4987	writable in one piece - this is the case on all current architectures.
		4988
		4989	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4990
		4991	The type C<sig_atomic_t volatile> (or whatever is defined as
		4992	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4993	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4994	believed to be sufficiently portable.
		4995
		4996	=item C<sigprocmask> must work in a threaded environment
		4997
		4998	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4999	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		5000	pthread implementations will either allow C<sigprocmask> in the "main
		5001	thread" or will block signals process-wide, both behaviours would
		5002	be compatible with libev. Interaction between C<sigprocmask> and
		5003	C<pthread_sigmask> could complicate things, however.
		5004
		5005	The most portable way to handle signals is to block signals in all threads
		5006	except the initial one, and run the default loop in the initial thread as
		5007	well.
		5008
		5009	=item C<long> must be large enough for common memory allocation sizes
		5010
		5011	To improve portability and simplify its API, libev uses C<long> internally
		5012	instead of C<size_t> when allocating its data structures. On non-POSIX
		5013	systems (Microsoft...) this might be unexpectedly low, but is still at
		5014	least 31 bits everywhere, which is enough for hundreds of millions of
		5015	watchers.
		5016
		5017	=item C<double> must hold a time value in seconds with enough accuracy
		5018
		5019	The type C<double> is used to represent timestamps. It is required to
		5020	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		5021	good enough for at least into the year 4000 with millisecond accuracy
		5022	(the design goal for libev). This requirement is overfulfilled by
		5023	implementations using IEEE 754, which is basically all existing ones. With
		5024	IEEE 754 doubles, you get microsecond accuracy until at least 2200.
		5025
		5026	=back
		5027
		5028	If you know of other additional requirements drop me a note.
		5029
		5030
		5031	=head1 ALGORITHMIC COMPLEXITIES
		5032
		5033	In this section the complexities of (many of) the algorithms used inside
		5034	libev will be documented. For complexity discussions about backends see
		5035	the documentation for C<ev_default_init>.
		5036
		5037	All of the following are about amortised time: If an array needs to be
		5038	extended, libev needs to realloc and move the whole array, but this
		5039	happens asymptotically rarer with higher number of elements, so O(1) might
		5040	mean that libev does a lengthy realloc operation in rare cases, but on
		5041	average it is much faster and asymptotically approaches constant time.
		5042
		5043	=over 4
		5044
		5045	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		5046
		5047	This means that, when you have a watcher that triggers in one hour and
		5048	there are 100 watchers that would trigger before that, then inserting will
		5049	have to skip roughly seven (C<ld 100>) of these watchers.
		5050
		5051	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		5052
		5053	That means that changing a timer costs less than removing/adding them,
		5054	as only the relative motion in the event queue has to be paid for.
		5055
		5056	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		5057
		5058	These just add the watcher into an array or at the head of a list.
		5059
		5060	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		5061
		5062	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		5063
		5064	These watchers are stored in lists, so they need to be walked to find the
		5065	correct watcher to remove. The lists are usually short (you don't usually
		5066	have many watchers waiting for the same fd or signal: one is typical, two
		5067	is rare).
		5068
		5069	=item Finding the next timer in each loop iteration: O(1)
		5070
		5071	By virtue of using a binary or 4-heap, the next timer is always found at a
		5072	fixed position in the storage array.
		5073
		5074	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		5075
		5076	A change means an I/O watcher gets started or stopped, which requires
		5077	libev to recalculate its status (and possibly tell the kernel, depending
		5078	on backend and whether C<ev_io_set> was used).
		5079
		5080	=item Activating one watcher (putting it into the pending state): O(1)
		5081
		5082	=item Priority handling: O(number_of_priorities)
		5083
		5084	Priorities are implemented by allocating some space for each
		5085	priority. When doing priority-based operations, libev usually has to
		5086	linearly search all the priorities, but starting/stopping and activating
		5087	watchers becomes O(1) with respect to priority handling.
		5088
		5089	=item Sending an ev_async: O(1)
		5090
		5091	=item Processing ev_async_send: O(number_of_async_watchers)
		5092
		5093	=item Processing signals: O(max_signal_number)
		5094
		5095	Sending involves a system call I<iff> there were no other C<ev_async_send>
		5096	calls in the current loop iteration. Checking for async and signal events
		5097	involves iterating over all running async watchers or all signal numbers.
		5098
		5099	=back
		5100
		5101
		5102	=head1 PORTING FROM LIBEV 3.X TO 4.X
		5103
		5104	The major version 4 introduced some incompatible changes to the API.
		5105
		5106	At the moment, the C<ev.h> header file provides compatibility definitions
		5107	for all changes, so most programs should still compile. The compatibility
		5108	layer might be removed in later versions of libev, so better update to the
		5109	new API early than late.
		5110
		5111	=over 4
		5112
		5113	=item C<EV_COMPAT3> backwards compatibility mechanism
		5114
		5115	The backward compatibility mechanism can be controlled by
		5116	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
		5117	section.
		5118
		5119	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5120
		5121	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5122
		5123	ev_loop_destroy (EV_DEFAULT_UC);
		5124	ev_loop_fork (EV_DEFAULT);
		5125
		5126	=item function/symbol renames
		5127
		5128	A number of functions and symbols have been renamed:
		5129
		5130	ev_loop => ev_run
		5131	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		5132	EVLOOP_ONESHOT => EVRUN_ONCE
		5133
		5134	ev_unloop => ev_break
		5135	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		5136	EVUNLOOP_ONE => EVBREAK_ONE
		5137	EVUNLOOP_ALL => EVBREAK_ALL
		5138
		5139	EV_TIMEOUT => EV_TIMER
		5140
		5141	ev_loop_count => ev_iteration
		5142	ev_loop_depth => ev_depth
		5143	ev_loop_verify => ev_verify
		5144
		5145	Most functions working on C<struct ev_loop> objects don't have an
		5146	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		5147	associated constants have been renamed to not collide with the C<struct
		5148	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		5149	as all other watcher types. Note that C<ev_loop_fork> is still called
		5150	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		5151	typedef.
		5152
		5153	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		5154
		5155	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		5156	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		5157	and work, but the library code will of course be larger.
		5158
		5159	=back
		5160
		5161
		5162	=head1 GLOSSARY
		5163
		5164	=over 4
		5165
		5166	=item active
		5167
		5168	A watcher is active as long as it has been started and not yet stopped.
		5169	See L<WATCHER STATES> for details.
		5170
		5171	=item application
		5172
		5173	In this document, an application is whatever is using libev.
		5174
		5175	=item backend
		5176
		5177	The part of the code dealing with the operating system interfaces.
		5178
		5179	=item callback
		5180
		5181	The address of a function that is called when some event has been
		5182	detected. Callbacks are being passed the event loop, the watcher that
		5183	received the event, and the actual event bitset.
		5184
		5185	=item callback/watcher invocation
		5186
		5187	The act of calling the callback associated with a watcher.
		5188
		5189	=item event
		5190
		5191	A change of state of some external event, such as data now being available
		5192	for reading on a file descriptor, time having passed or simply not having
		5193	any other events happening anymore.
		5194
		5195	In libev, events are represented as single bits (such as C<EV_READ> or
		5196	C<EV_TIMER>).
		5197
		5198	=item event library
		5199
		5200	A software package implementing an event model and loop.
		5201
		5202	=item event loop
		5203
		5204	An entity that handles and processes external events and converts them
		5205	into callback invocations.
		5206
		5207	=item event model
		5208
		5209	The model used to describe how an event loop handles and processes
		5210	watchers and events.
		5211
		5212	=item pending
		5213
		5214	A watcher is pending as soon as the corresponding event has been
		5215	detected. See L<WATCHER STATES> for details.
		5216
		5217	=item real time
		5218
		5219	The physical time that is observed. It is apparently strictly monotonic :)
		5220
		5221	=item wall-clock time
		5222
		5223	The time and date as shown on clocks. Unlike real time, it can actually
		5224	be wrong and jump forwards and backwards, e.g. when the you adjust your
		5225	clock.
		5226
		5227	=item watcher
		5228
		5229	A data structure that describes interest in certain events. Watchers need
		5230	to be started (attached to an event loop) before they can receive events.
		5231
		5232	=back
1316		5233
1317	=head1 AUTHOR	5234	=head1 AUTHOR
1318		5235
1319	Marc Lehmann <libev@schmorp.de>.	5236	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5237	Magnusson and Emanuele Giaquinta.
1320		5238

Diff Legend

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