ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.16 by root, Mon Nov 12 08:47:14 2007 UTC vs.
Revision 1.276 by root, Tue Dec 29 13:11:00 2009 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_loop's to stop iterating
		32	ev_unloop (EV_A_ EVUNLOOP_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_loop to stop iterating
		41	ev_unloop (EV_A_ EVUNLOOP_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_loop (0);
		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_loop (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	Familarity with event based programming techniques in general is assumed
		81	throughout this document.
		82
		83	=head1 ABOUT LIBEV
10		84
11	Libev is an event loop: you register interest in certain events (such as a	85	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	86	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	87	these event sources and provide your program with events.
14		88
15	To do this, it must take more or less complete control over your process	89	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	90	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	91	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	93	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	94	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	95	details of the event, and then hand it over to libev by I<starting> the
22	watcher.	96	watcher.
23		97
24	=head1 FEATURES	98	=head2 FEATURES
25		99
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	100	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27	kqueue mechanisms for file descriptor events, relative timers, absolute	101	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28	timers with customised rescheduling, signal events, process status change	102	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
		103	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
		104	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
		105	timers (C<ev_timer>), absolute timers with customised rescheduling
		106	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
29	events (related to SIGCHLD), and event watchers dealing with the event	107	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	108	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
		109	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		110	limited support for fork events (C<ev_fork>).
		111
		112	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	113	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	114	for example).
33		115
34	=head1 CONVENTIONS	116	=head2 CONVENTIONS
35		117
36	Libev is very configurable. In this manual the default configuration	118	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	119	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	120	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	121	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	122	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 *>)	123	name C<loop> (which is always of type C<struct ev_loop *>) will not have
42	will not have this argument.	124	this argument.
43		125
44	=head1 TIME AND OTHER GLOBAL FUNCTIONS	126	=head2 TIME REPRESENTATION
45		127
46	Libev represents time as a single floating point number, representing the	128	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	129	the (fractional) number of seconds since the (POSIX) epoch (somewhere
48	the beginning of 1970, details are complicated, don't ask). This type is	130	near the beginning of 1970, details are complicated, don't ask). This
49	called C<ev_tstamp>, which is what you should use too. It usually aliases	131	type is called C<ev_tstamp>, which is what you should use too. It usually
50	to the double type in C.	132	aliases to the C<double> type in C. When you need to do any calculations
		133	on it, you should treat it as some floating point value. Unlike the name
		134	component C<stamp> might indicate, it is also used for time differences
		135	throughout libev.
		136
		137	=head1 ERROR HANDLING
		138
		139	Libev knows three classes of errors: operating system errors, usage errors
		140	and internal errors (bugs).
		141
		142	When libev catches an operating system error it cannot handle (for example
		143	a system call indicating a condition libev cannot fix), it calls the callback
		144	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		145	abort. The default is to print a diagnostic message and to call C<abort
		146	()>.
		147
		148	When libev detects a usage error such as a negative timer interval, then
		149	it will print a diagnostic message and abort (via the C<assert> mechanism,
		150	so C<NDEBUG> will disable this checking): these are programming errors in
		151	the libev caller and need to be fixed there.
		152
		153	Libev also has a few internal error-checking C<assert>ions, and also has
		154	extensive consistency checking code. These do not trigger under normal
		155	circumstances, as they indicate either a bug in libev or worse.
		156
		157
		158	=head1 GLOBAL FUNCTIONS
		159
		160	These functions can be called anytime, even before initialising the
		161	library in any way.
51		162
52	=over 4	163	=over 4
53		164
54	=item ev_tstamp ev_time ()	165	=item ev_tstamp ev_time ()
55		166
56	Returns the current time as libev would use it.	167	Returns the current time as libev would use it. Please note that the
		168	C<ev_now> function is usually faster and also often returns the timestamp
		169	you actually want to know.
		170
		171	=item ev_sleep (ev_tstamp interval)
		172
		173	Sleep for the given interval: The current thread will be blocked until
		174	either it is interrupted or the given time interval has passed. Basically
		175	this is a sub-second-resolution C<sleep ()>.
57		176
58	=item int ev_version_major ()	177	=item int ev_version_major ()
59		178
60	=item int ev_version_minor ()	179	=item int ev_version_minor ()
61		180
62	You can find out the major and minor version numbers of the library	181	You can find out the major and minor ABI version numbers of the library
63	you linked against by calling the functions C<ev_version_major> and	182	you linked against by calling the functions C<ev_version_major> and
64	C<ev_version_minor>. If you want, you can compare against the global	183	C<ev_version_minor>. If you want, you can compare against the global
65	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	184	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
66	version of the library your program was compiled against.	185	version of the library your program was compiled against.
67		186
		187	These version numbers refer to the ABI version of the library, not the
		188	release version.
		189
68	Usually, it's a good idea to terminate if the major versions mismatch,	190	Usually, it's a good idea to terminate if the major versions mismatch,
69	as this indicates an incompatible change. Minor versions are usually	191	as this indicates an incompatible change. Minor versions are usually
70	compatible to older versions, so a larger minor version alone is usually	192	compatible to older versions, so a larger minor version alone is usually
71	not a problem.	193	not a problem.
72		194
		195	Example: Make sure we haven't accidentally been linked against the wrong
		196	version.
		197
		198	assert (("libev version mismatch",
		199	ev_version_major () == EV_VERSION_MAJOR
		200	&& ev_version_minor () >= EV_VERSION_MINOR));
		201
		202	=item unsigned int ev_supported_backends ()
		203
		204	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		205	value) compiled into this binary of libev (independent of their
		206	availability on the system you are running on). See C<ev_default_loop> for
		207	a description of the set values.
		208
		209	Example: make sure we have the epoll method, because yeah this is cool and
		210	a must have and can we have a torrent of it please!!!11
		211
		212	assert (("sorry, no epoll, no sex",
		213	ev_supported_backends () & EVBACKEND_EPOLL));
		214
		215	=item unsigned int ev_recommended_backends ()
		216
		217	Return the set of all backends compiled into this binary of libev and also
		218	recommended for this platform. This set is often smaller than the one
		219	returned by C<ev_supported_backends>, as for example kqueue is broken on
		220	most BSDs and will not be auto-detected unless you explicitly request it
		221	(assuming you know what you are doing). This is the set of backends that
		222	libev will probe for if you specify no backends explicitly.
		223
		224	=item unsigned int ev_embeddable_backends ()
		225
		226	Returns the set of backends that are embeddable in other event loops. This
		227	is the theoretical, all-platform, value. To find which backends
		228	might be supported on the current system, you would need to look at
		229	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
		230	recommended ones.
		231
		232	See the description of C<ev_embed> watchers for more info.
		233
73	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	234	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
74		235
75	Sets the allocation function to use (the prototype is similar to the	236	Sets the allocation function to use (the prototype is similar - the
76	realloc C function, the semantics are identical). It is used to allocate	237	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
77	and free memory (no surprises here). If it returns zero when memory	238	used to allocate and free memory (no surprises here). If it returns zero
78	needs to be allocated, the library might abort or take some potentially	239	when memory needs to be allocated (C<size != 0>), the library might abort
79	destructive action. The default is your system realloc function.	240	or take some potentially destructive action.
		241
		242	Since some systems (at least OpenBSD and Darwin) fail to implement
		243	correct C<realloc> semantics, libev will use a wrapper around the system
		244	C<realloc> and C<free> functions by default.
80		245
81	You could override this function in high-availability programs to, say,	246	You could override this function in high-availability programs to, say,
82	free some memory if it cannot allocate memory, to use a special allocator,	247	free some memory if it cannot allocate memory, to use a special allocator,
83	or even to sleep a while and retry until some memory is available.	248	or even to sleep a while and retry until some memory is available.
84		249
		250	Example: Replace the libev allocator with one that waits a bit and then
		251	retries (example requires a standards-compliant C<realloc>).
		252
		253	static void *
		254	persistent_realloc (void *ptr, size_t size)
		255	{
		256	for (;;)
		257	{
		258	void *newptr = realloc (ptr, size);
		259
		260	if (newptr)
		261	return newptr;
		262
		263	sleep (60);
		264	}
		265	}
		266
		267	...
		268	ev_set_allocator (persistent_realloc);
		269
85	=item ev_set_syserr_cb (void (*cb)(const char *msg));	270	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
86		271
87	Set the callback function to call on a retryable syscall error (such	272	Set the callback function to call on a retryable system call error (such
88	as failed select, poll, epoll_wait). The message is a printable string	273	as failed select, poll, epoll_wait). The message is a printable string
89	indicating the system call or subsystem causing the problem. If this	274	indicating the system call or subsystem causing the problem. If this
90	callback is set, then libev will expect it to remedy the sitution, no	275	callback is set, then libev will expect it to remedy the situation, no
91	matter what, when it returns. That is, libev will generally retry the	276	matter what, when it returns. That is, libev will generally retry the
92	requested operation, or, if the condition doesn't go away, do bad stuff	277	requested operation, or, if the condition doesn't go away, do bad stuff
93	(such as abort).	278	(such as abort).
94		279
		280	Example: This is basically the same thing that libev does internally, too.
		281
		282	static void
		283	fatal_error (const char *msg)
		284	{
		285	perror (msg);
		286	abort ();
		287	}
		288
		289	...
		290	ev_set_syserr_cb (fatal_error);
		291
95	=back	292	=back
96		293
97	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	294	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
98		295
99	An event loop is described by a C<struct ev_loop *>. The library knows two	296	An event loop is described by a C<struct ev_loop *> (the C<struct>
100	types of such loops, the I<default> loop, which supports signals and child	297	is I<not> optional in this case, as there is also an C<ev_loop>
101	events, and dynamically created loops which do not.	298	I<function>).
102		299
103	If you use threads, a common model is to run the default event loop	300	The library knows two types of such loops, the I<default> loop, which
104	in your main thread (or in a separate thrad) and for each thread you	301	supports signals and child events, and dynamically created loops which do
105	create, you also create another event loop. Libev itself does no locking	302	not.
106	whatsoever, so if you mix calls to the same event loop in different
107	threads, make sure you lock (this is usually a bad idea, though, even if
108	done correctly, because it's hideous and inefficient).
109		303
110	=over 4	304	=over 4
111		305
112	=item struct ev_loop *ev_default_loop (unsigned int flags)	306	=item struct ev_loop *ev_default_loop (unsigned int flags)
113		307
114	This will initialise the default event loop if it hasn't been initialised	308	This will initialise the default event loop if it hasn't been initialised
115	yet and return it. If the default loop could not be initialised, returns	309	yet and return it. If the default loop could not be initialised, returns
116	false. If it already was initialised it simply returns it (and ignores the	310	false. If it already was initialised it simply returns it (and ignores the
117	flags).	311	flags. If that is troubling you, check C<ev_backend ()> afterwards).
118		312
119	If you don't know what event loop to use, use the one returned from this	313	If you don't know what event loop to use, use the one returned from this
120	function.	314	function.
121		315
		316	Note that this function is I<not> thread-safe, so if you want to use it
		317	from multiple threads, you have to lock (note also that this is unlikely,
		318	as loops cannot be shared easily between threads anyway).
		319
		320	The default loop is the only loop that can handle C<ev_signal> and
		321	C<ev_child> watchers, and to do this, it always registers a handler
		322	for C<SIGCHLD>. If this is a problem for your application you can either
		323	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
		324	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
		325	C<ev_default_init>.
		326
122	The flags argument can be used to specify special behaviour or specific	327	The flags argument can be used to specify special behaviour or specific
123	backends to use, and is usually specified as 0 (or EVFLAG_AUTO).	328	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
124		329
125	It supports the following flags:	330	The following flags are supported:
126		331
127	=over 4	332	=over 4
128		333
129	=item C<EVFLAG_AUTO>	334	=item C<EVFLAG_AUTO>
130		335
131	The default flags value. Use this if you have no clue (it's the right	336	The default flags value. Use this if you have no clue (it's the right
132	thing, believe me).	337	thing, believe me).
133		338
134	=item C<EVFLAG_NOENV>	339	=item C<EVFLAG_NOENV>
135		340
136	If this flag bit is ored into the flag value (or the program runs setuid	341	If this flag bit is or'ed into the flag value (or the program runs setuid
137	or setgid) then libev will I<not> look at the environment variable	342	or setgid) then libev will I<not> look at the environment variable
138	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	343	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
139	override the flags completely if it is found in the environment. This is	344	override the flags completely if it is found in the environment. This is
140	useful to try out specific backends to test their performance, or to work	345	useful to try out specific backends to test their performance, or to work
141	around bugs.	346	around bugs.
142		347
		348	=item C<EVFLAG_FORKCHECK>
		349
		350	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after
		351	a fork, you can also make libev check for a fork in each iteration by
		352	enabling this flag.
		353
		354	This works by calling C<getpid ()> on every iteration of the loop,
		355	and thus this might slow down your event loop if you do a lot of loop
		356	iterations and little real work, but is usually not noticeable (on my
		357	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		358	without a system call and thus I<very> fast, but my GNU/Linux system also has
		359	C<pthread_atfork> which is even faster).
		360
		361	The big advantage of this flag is that you can forget about fork (and
		362	forget about forgetting to tell libev about forking) when you use this
		363	flag.
		364
		365	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		366	environment variable.
		367
		368	=item C<EVFLAG_NOINOTIFY>
		369
		370	When this flag is specified, then libev will not attempt to use the
		371	I<inotify> API for it's C<ev_stat> watchers. Apart from debugging and
		372	testing, this flag can be useful to conserve inotify file descriptors, as
		373	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
		374
		375	=item C<EVFLAG_NOSIGFD>
		376
		377	When this flag is specified, then libev will not attempt to use the
		378	I<signalfd> API for it's C<ev_signal> (and C<ev_child>) watchers. This is
		379	probably only useful to work around any bugs in libev. Consequently, this
		380	flag might go away once the signalfd functionality is considered stable,
		381	so it's useful mostly in environment variables and not in program code.
		382
143	=item C<EVMETHOD_SELECT> (portable select backend)	383	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
144		384
		385	This is your standard select(2) backend. Not I<completely> standard, as
		386	libev tries to roll its own fd_set with no limits on the number of fds,
		387	but if that fails, expect a fairly low limit on the number of fds when
		388	using this backend. It doesn't scale too well (O(highest_fd)), but its
		389	usually the fastest backend for a low number of (low-numbered :) fds.
		390
		391	To get good performance out of this backend you need a high amount of
		392	parallelism (most of the file descriptors should be busy). If you are
		393	writing a server, you should C<accept ()> in a loop to accept as many
		394	connections as possible during one iteration. You might also want to have
		395	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		396	readiness notifications you get per iteration.
		397
		398	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		399	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		400	C<exceptfds> set on that platform).
		401
145	=item C<EVMETHOD_POLL> (poll backend, available everywhere except on windows)	402	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
146		403
147	=item C<EVMETHOD_EPOLL> (linux only)	404	And this is your standard poll(2) backend. It's more complicated
		405	than select, but handles sparse fds better and has no artificial
		406	limit on the number of fds you can use (except it will slow down
		407	considerably with a lot of inactive fds). It scales similarly to select,
		408	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		409	performance tips.
148		410
149	=item C<EVMETHOD_KQUEUE> (some bsds only)	411	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		412	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
150		413
151	=item C<EVMETHOD_DEVPOLL> (solaris 8 only)	414	=item C<EVBACKEND_EPOLL> (value 4, Linux)
152		415
153	=item C<EVMETHOD_PORT> (solaris 10 only)	416	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		417	kernels).
154		418
155	If one or more of these are ored into the flags value, then only these	419	For few fds, this backend is a bit little slower than poll and select,
156	backends will be tried (in the reverse order as given here). If one are	420	but it scales phenomenally better. While poll and select usually scale
157	specified, any backend will do.	421	like O(total_fds) where n is the total number of fds (or the highest fd),
		422	epoll scales either O(1) or O(active_fds).
		423
		424	The epoll mechanism deserves honorable mention as the most misdesigned
		425	of the more advanced event mechanisms: mere annoyances include silently
		426	dropping file descriptors, requiring a system call per change per file
		427	descriptor (and unnecessary guessing of parameters), problems with dup and
		428	so on. The biggest issue is fork races, however - if a program forks then
		429	I<both> parent and child process have to recreate the epoll set, which can
		430	take considerable time (one syscall per file descriptor) and is of course
		431	hard to detect.
		432
		433	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		434	of course I<doesn't>, and epoll just loves to report events for totally
		435	I<different> file descriptors (even already closed ones, so one cannot
		436	even remove them from the set) than registered in the set (especially
		437	on SMP systems). Libev tries to counter these spurious notifications by
		438	employing an additional generation counter and comparing that against the
		439	events to filter out spurious ones, recreating the set when required.
		440
		441	While stopping, setting and starting an I/O watcher in the same iteration
		442	will result in some caching, there is still a system call per such
		443	incident (because the same I<file descriptor> could point to a different
		444	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
		445	file descriptors might not work very well if you register events for both
		446	file descriptors.
		447
		448	Best performance from this backend is achieved by not unregistering all
		449	watchers for a file descriptor until it has been closed, if possible,
		450	i.e. keep at least one watcher active per fd at all times. Stopping and
		451	starting a watcher (without re-setting it) also usually doesn't cause
		452	extra overhead. A fork can both result in spurious notifications as well
		453	as in libev having to destroy and recreate the epoll object, which can
		454	take considerable time and thus should be avoided.
		455
		456	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		457	faster than epoll for maybe up to a hundred file descriptors, depending on
		458	the usage. So sad.
		459
		460	While nominally embeddable in other event loops, this feature is broken in
		461	all kernel versions tested so far.
		462
		463	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		464	C<EVBACKEND_POLL>.
		465
		466	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
		467
		468	Kqueue deserves special mention, as at the time of this writing, it
		469	was broken on all BSDs except NetBSD (usually it doesn't work reliably
		470	with anything but sockets and pipes, except on Darwin, where of course
		471	it's completely useless). Unlike epoll, however, whose brokenness
		472	is by design, these kqueue bugs can (and eventually will) be fixed
		473	without API changes to existing programs. For this reason it's not being
		474	"auto-detected" unless you explicitly specify it in the flags (i.e. using
		475	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		476	system like NetBSD.
		477
		478	You still can embed kqueue into a normal poll or select backend and use it
		479	only for sockets (after having made sure that sockets work with kqueue on
		480	the target platform). See C<ev_embed> watchers for more info.
		481
		482	It scales in the same way as the epoll backend, but the interface to the
		483	kernel is more efficient (which says nothing about its actual speed, of
		484	course). While stopping, setting and starting an I/O watcher does never
		485	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
		486	two event changes per incident. Support for C<fork ()> is very bad (but
		487	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		488	cases
		489
		490	This backend usually performs well under most conditions.
		491
		492	While nominally embeddable in other event loops, this doesn't work
		493	everywhere, so you might need to test for this. And since it is broken
		494	almost everywhere, you should only use it when you have a lot of sockets
		495	(for which it usually works), by embedding it into another event loop
		496	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		497	also broken on OS X)) and, did I mention it, using it only for sockets.
		498
		499	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		500	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		501	C<NOTE_EOF>.
		502
		503	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
		504
		505	This is not implemented yet (and might never be, unless you send me an
		506	implementation). According to reports, C</dev/poll> only supports sockets
		507	and is not embeddable, which would limit the usefulness of this backend
		508	immensely.
		509
		510	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
		511
		512	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
		513	it's really slow, but it still scales very well (O(active_fds)).
		514
		515	Please note that Solaris event ports can deliver a lot of spurious
		516	notifications, so you need to use non-blocking I/O or other means to avoid
		517	blocking when no data (or space) is available.
		518
		519	While this backend scales well, it requires one system call per active
		520	file descriptor per loop iteration. For small and medium numbers of file
		521	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		522	might perform better.
		523
		524	On the positive side, with the exception of the spurious readiness
		525	notifications, this backend actually performed fully to specification
		526	in all tests and is fully embeddable, which is a rare feat among the
		527	OS-specific backends (I vastly prefer correctness over speed hacks).
		528
		529	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		530	C<EVBACKEND_POLL>.
		531
		532	=item C<EVBACKEND_ALL>
		533
		534	Try all backends (even potentially broken ones that wouldn't be tried
		535	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
		536	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		537
		538	It is definitely not recommended to use this flag.
158		539
159	=back	540	=back
		541
		542	If one or more of the backend flags are or'ed into the flags value,
		543	then only these backends will be tried (in the reverse order as listed
		544	here). If none are specified, all backends in C<ev_recommended_backends
		545	()> will be tried.
		546
		547	Example: This is the most typical usage.
		548
		549	if (!ev_default_loop (0))
		550	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		551
		552	Example: Restrict libev to the select and poll backends, and do not allow
		553	environment settings to be taken into account:
		554
		555	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		556
		557	Example: Use whatever libev has to offer, but make sure that kqueue is
		558	used if available (warning, breaks stuff, best use only with your own
		559	private event loop and only if you know the OS supports your types of
		560	fds):
		561
		562	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
160		563
161	=item struct ev_loop *ev_loop_new (unsigned int flags)	564	=item struct ev_loop *ev_loop_new (unsigned int flags)
162		565
163	Similar to C<ev_default_loop>, but always creates a new event loop that is	566	Similar to C<ev_default_loop>, but always creates a new event loop that is
164	always distinct from the default loop. Unlike the default loop, it cannot	567	always distinct from the default loop. Unlike the default loop, it cannot
165	handle signal and child watchers, and attempts to do so will be greeted by	568	handle signal and child watchers, and attempts to do so will be greeted by
166	undefined behaviour (or a failed assertion if assertions are enabled).	569	undefined behaviour (or a failed assertion if assertions are enabled).
167		570
		571	Note that this function I<is> thread-safe, and the recommended way to use
		572	libev with threads is indeed to create one loop per thread, and using the
		573	default loop in the "main" or "initial" thread.
		574
		575	Example: Try to create a event loop that uses epoll and nothing else.
		576
		577	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
		578	if (!epoller)
		579	fatal ("no epoll found here, maybe it hides under your chair");
		580
168	=item ev_default_destroy ()	581	=item ev_default_destroy ()
169		582
170	Destroys the default loop again (frees all memory and kernel state	583	Destroys the default loop again (frees all memory and kernel state
171	etc.). This stops all registered event watchers (by not touching them in	584	etc.). None of the active event watchers will be stopped in the normal
172	any way whatsoever, although you cannot rely on this :).	585	sense, so e.g. C<ev_is_active> might still return true. It is your
		586	responsibility to either stop all watchers cleanly yourself I<before>
		587	calling this function, or cope with the fact afterwards (which is usually
		588	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		589	for example).
		590
		591	Note that certain global state, such as signal state (and installed signal
		592	handlers), will not be freed by this function, and related watchers (such
		593	as signal and child watchers) would need to be stopped manually.
		594
		595	In general it is not advisable to call this function except in the
		596	rare occasion where you really need to free e.g. the signal handling
		597	pipe fds. If you need dynamically allocated loops it is better to use
		598	C<ev_loop_new> and C<ev_loop_destroy>.
173		599
174	=item ev_loop_destroy (loop)	600	=item ev_loop_destroy (loop)
175		601
176	Like C<ev_default_destroy>, but destroys an event loop created by an	602	Like C<ev_default_destroy>, but destroys an event loop created by an
177	earlier call to C<ev_loop_new>.	603	earlier call to C<ev_loop_new>.
178		604
179	=item ev_default_fork ()	605	=item ev_default_fork ()
180		606
		607	This function sets a flag that causes subsequent C<ev_loop> iterations
181	This function reinitialises the kernel state for backends that have	608	to reinitialise the kernel state for backends that have one. Despite the
182	one. Despite the name, you can call it anytime, but it makes most sense	609	name, you can call it anytime, but it makes most sense after forking, in
183	after forking, in either the parent or child process (or both, but that	610	the child process (or both child and parent, but that again makes little
184	again makes little sense).	611	sense). You I<must> call it in the child before using any of the libev
		612	functions, and it will only take effect at the next C<ev_loop> iteration.
185		613
186	You I<must> call this function after forking if and only if you want to	614	On the other hand, you only need to call this function in the child
187	use the event library in both processes. If you just fork+exec, you don't	615	process if and only if you want to use the event library in the child. If
188	have to call it.	616	you just fork+exec, you don't have to call it at all.
189		617
190	The function itself is quite fast and it's usually not a problem to call	618	The function itself is quite fast and it's usually not a problem to call
191	it just in case after a fork. To make this easy, the function will fit in	619	it just in case after a fork. To make this easy, the function will fit in
192	quite nicely into a call to C<pthread_atfork>:	620	quite nicely into a call to C<pthread_atfork>:
193		621
…		…
195		623
196	=item ev_loop_fork (loop)	624	=item ev_loop_fork (loop)
197		625
198	Like C<ev_default_fork>, but acts on an event loop created by	626	Like C<ev_default_fork>, but acts on an event loop created by
199	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	627	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
200	after fork, and how you do this is entirely your own problem.	628	after fork that you want to re-use in the child, and how you do this is
		629	entirely your own problem.
201		630
		631	=item int ev_is_default_loop (loop)
		632
		633	Returns true when the given loop is, in fact, the default loop, and false
		634	otherwise.
		635
		636	=item unsigned int ev_loop_count (loop)
		637
		638	Returns the count of loop iterations for the loop, which is identical to
		639	the number of times libev did poll for new events. It starts at C<0> and
		640	happily wraps around with enough iterations.
		641
		642	This value can sometimes be useful as a generation counter of sorts (it
		643	"ticks" the number of loop iterations), as it roughly corresponds with
		644	C<ev_prepare> and C<ev_check> calls.
		645
		646	=item unsigned int ev_loop_depth (loop)
		647
		648	Returns the number of times C<ev_loop> was entered minus the number of
		649	times C<ev_loop> was exited, in other words, the recursion depth.
		650
		651	Outside C<ev_loop>, this number is zero. In a callback, this number is
		652	C<1>, unless C<ev_loop> was invoked recursively (or from another thread),
		653	in which case it is higher.
		654
		655	Leaving C<ev_loop> abnormally (setjmp/longjmp, cancelling the thread
		656	etc.), doesn't count as exit.
		657
202	=item unsigned int ev_method (loop)	658	=item unsigned int ev_backend (loop)
203		659
204	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	660	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
205	use.	661	use.
206		662
207	=item ev_tstamp ev_now (loop)	663	=item ev_tstamp ev_now (loop)
208		664
209	Returns the current "event loop time", which is the time the event loop	665	Returns the current "event loop time", which is the time the event loop
210	got events and started processing them. This timestamp does not change	666	received events and started processing them. This timestamp does not
211	as long as callbacks are being processed, and this is also the base time	667	change as long as callbacks are being processed, and this is also the base
212	used for relative timers. You can treat it as the timestamp of the event	668	time used for relative timers. You can treat it as the timestamp of the
213	occuring (or more correctly, the mainloop finding out about it).	669	event occurring (or more correctly, libev finding out about it).
		670
		671	=item ev_now_update (loop)
		672
		673	Establishes the current time by querying the kernel, updating the time
		674	returned by C<ev_now ()> in the progress. This is a costly operation and
		675	is usually done automatically within C<ev_loop ()>.
		676
		677	This function is rarely useful, but when some event callback runs for a
		678	very long time without entering the event loop, updating libev's idea of
		679	the current time is a good idea.
		680
		681	See also L<The special problem of time updates> in the C<ev_timer> section.
		682
		683	=item ev_suspend (loop)
		684
		685	=item ev_resume (loop)
		686
		687	These two functions suspend and resume a loop, for use when the loop is
		688	not used for a while and timeouts should not be processed.
		689
		690	A typical use case would be an interactive program such as a game: When
		691	the user presses C<^Z> to suspend the game and resumes it an hour later it
		692	would be best to handle timeouts as if no time had actually passed while
		693	the program was suspended. This can be achieved by calling C<ev_suspend>
		694	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		695	C<ev_resume> directly afterwards to resume timer processing.
		696
		697	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		698	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		699	will be rescheduled (that is, they will lose any events that would have
		700	occured while suspended).
		701
		702	After calling C<ev_suspend> you B<must not> call I<any> function on the
		703	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		704	without a previous call to C<ev_suspend>.
		705
		706	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		707	event loop time (see C<ev_now_update>).
214		708
215	=item ev_loop (loop, int flags)	709	=item ev_loop (loop, int flags)
216		710
217	Finally, this is it, the event handler. This function usually is called	711	Finally, this is it, the event handler. This function usually is called
218	after you initialised all your watchers and you want to start handling	712	after you have initialised all your watchers and you want to start
219	events.	713	handling events.
220		714
221	If the flags argument is specified as 0, it will not return until either	715	If the flags argument is specified as C<0>, it will not return until
222	no event watchers are active anymore or C<ev_unloop> was called.	716	either no event watchers are active anymore or C<ev_unloop> was called.
		717
		718	Please note that an explicit C<ev_unloop> is usually better than
		719	relying on all watchers to be stopped when deciding when a program has
		720	finished (especially in interactive programs), but having a program
		721	that automatically loops as long as it has to and no longer by virtue
		722	of relying on its watchers stopping correctly, that is truly a thing of
		723	beauty.
223		724
224	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	725	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
225	those events and any outstanding ones, but will not block your process in	726	those events and any already outstanding ones, but will not block your
226	case there are no events and will return after one iteration of the loop.	727	process in case there are no events and will return after one iteration of
		728	the loop.
227		729
228	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	730	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
229	neccessary) and will handle those and any outstanding ones. It will block	731	necessary) and will handle those and any already outstanding ones. It
230	your process until at least one new event arrives, and will return after	732	will block your process until at least one new event arrives (which could
		733	be an event internal to libev itself, so there is no guarantee that a
		734	user-registered callback will be called), and will return after one
231	one iteration of the loop.	735	iteration of the loop.
232		736
233	This flags value could be used to implement alternative looping	737	This is useful if you are waiting for some external event in conjunction
234	constructs, but the C<prepare> and C<check> watchers provide a better and	738	with something not expressible using other libev watchers (i.e. "roll your
235	more generic mechanism.	739	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		740	usually a better approach for this kind of thing.
		741
		742	Here are the gory details of what C<ev_loop> does:
		743
		744	- Before the first iteration, call any pending watchers.
		745	* If EVFLAG_FORKCHECK was used, check for a fork.
		746	- If a fork was detected (by any means), queue and call all fork watchers.
		747	- Queue and call all prepare watchers.
		748	- If we have been forked, detach and recreate the kernel state
		749	as to not disturb the other process.
		750	- Update the kernel state with all outstanding changes.
		751	- Update the "event loop time" (ev_now ()).
		752	- Calculate for how long to sleep or block, if at all
		753	(active idle watchers, EVLOOP_NONBLOCK or not having
		754	any active watchers at all will result in not sleeping).
		755	- Sleep if the I/O and timer collect interval say so.
		756	- Block the process, waiting for any events.
		757	- Queue all outstanding I/O (fd) events.
		758	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
		759	- Queue all expired timers.
		760	- Queue all expired periodics.
		761	- Unless any events are pending now, queue all idle watchers.
		762	- Queue all check watchers.
		763	- Call all queued watchers in reverse order (i.e. check watchers first).
		764	Signals and child watchers are implemented as I/O watchers, and will
		765	be handled here by queueing them when their watcher gets executed.
		766	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
		767	were used, or there are no active watchers, return, otherwise
		768	continue with step *.
		769
		770	Example: Queue some jobs and then loop until no events are outstanding
		771	anymore.
		772
		773	... queue jobs here, make sure they register event watchers as long
		774	... as they still have work to do (even an idle watcher will do..)
		775	ev_loop (my_loop, 0);
		776	... jobs done or somebody called unloop. yeah!
236		777
237	=item ev_unloop (loop, how)	778	=item ev_unloop (loop, how)
238		779
239	Can be used to make a call to C<ev_loop> return early (but only after it	780	Can be used to make a call to C<ev_loop> return early (but only after it
240	has processed all outstanding events). The C<how> argument must be either	781	has processed all outstanding events). The C<how> argument must be either
241	C<EVUNLOOP_ONCE>, which will make the innermost C<ev_loop> call return, or	782	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
242	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	783	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
		784
		785	This "unloop state" will be cleared when entering C<ev_loop> again.
		786
		787	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
243		788
244	=item ev_ref (loop)	789	=item ev_ref (loop)
245		790
246	=item ev_unref (loop)	791	=item ev_unref (loop)
247		792
248	Ref/unref can be used to add or remove a reference count on the event	793	Ref/unref can be used to add or remove a reference count on the event
249	loop: Every watcher keeps one reference, and as long as the reference	794	loop: Every watcher keeps one reference, and as long as the reference
250	count is nonzero, C<ev_loop> will not return on its own. If you have	795	count is nonzero, C<ev_loop> will not return on its own.
251	a watcher you never unregister that should not keep C<ev_loop> from	796
252	returning, ev_unref() after starting, and ev_ref() before stopping it. For	797	This is useful when you have a watcher that you never intend to
		798	unregister, but that nevertheless should not keep C<ev_loop> from
		799	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		800	before stopping it.
		801
253	example, libev itself uses this for its internal signal pipe: It is not	802	As an example, libev itself uses this for its internal signal pipe: It
254	visible to the libev user and should not keep C<ev_loop> from exiting if	803	is not visible to the libev user and should not keep C<ev_loop> from
255	no event watchers registered by it are active. It is also an excellent	804	exiting if no event watchers registered by it are active. It is also an
256	way to do this for generic recurring timers or from within third-party	805	excellent way to do this for generic recurring timers or from within
257	libraries. Just remember to I<unref after start> and I<ref before stop>.	806	third-party libraries. Just remember to I<unref after start> and I<ref
		807	before stop> (but only if the watcher wasn't active before, or was active
		808	before, respectively. Note also that libev might stop watchers itself
		809	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		810	in the callback).
		811
		812	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
		813	running when nothing else is active.
		814
		815	ev_signal exitsig;
		816	ev_signal_init (&exitsig, sig_cb, SIGINT);
		817	ev_signal_start (loop, &exitsig);
		818	evf_unref (loop);
		819
		820	Example: For some weird reason, unregister the above signal handler again.
		821
		822	ev_ref (loop);
		823	ev_signal_stop (loop, &exitsig);
		824
		825	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		826
		827	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		828
		829	These advanced functions influence the time that libev will spend waiting
		830	for events. Both time intervals are by default C<0>, meaning that libev
		831	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		832	latency.
		833
		834	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		835	allows libev to delay invocation of I/O and timer/periodic callbacks
		836	to increase efficiency of loop iterations (or to increase power-saving
		837	opportunities).
		838
		839	The idea is that sometimes your program runs just fast enough to handle
		840	one (or very few) event(s) per loop iteration. While this makes the
		841	program responsive, it also wastes a lot of CPU time to poll for new
		842	events, especially with backends like C<select ()> which have a high
		843	overhead for the actual polling but can deliver many events at once.
		844
		845	By setting a higher I<io collect interval> you allow libev to spend more
		846	time collecting I/O events, so you can handle more events per iteration,
		847	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		848	C<ev_timer>) will be not affected. Setting this to a non-null value will
		849	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		850	sleep time ensures that libev will not poll for I/O events more often then
		851	once per this interval, on average.
		852
		853	Likewise, by setting a higher I<timeout collect interval> you allow libev
		854	to spend more time collecting timeouts, at the expense of increased
		855	latency/jitter/inexactness (the watcher callback will be called
		856	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		857	value will not introduce any overhead in libev.
		858
		859	Many (busy) programs can usually benefit by setting the I/O collect
		860	interval to a value near C<0.1> or so, which is often enough for
		861	interactive servers (of course not for games), likewise for timeouts. It
		862	usually doesn't make much sense to set it to a lower value than C<0.01>,
		863	as this approaches the timing granularity of most systems. Note that if
		864	you do transactions with the outside world and you can't increase the
		865	parallelity, then this setting will limit your transaction rate (if you
		866	need to poll once per transaction and the I/O collect interval is 0.01,
		867	then you can't do more than 100 transations per second).
		868
		869	Setting the I<timeout collect interval> can improve the opportunity for
		870	saving power, as the program will "bundle" timer callback invocations that
		871	are "near" in time together, by delaying some, thus reducing the number of
		872	times the process sleeps and wakes up again. Another useful technique to
		873	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		874	they fire on, say, one-second boundaries only.
		875
		876	Example: we only need 0.1s timeout granularity, and we wish not to poll
		877	more often than 100 times per second:
		878
		879	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		880	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		881
		882	=item ev_invoke_pending (loop)
		883
		884	This call will simply invoke all pending watchers while resetting their
		885	pending state. Normally, C<ev_loop> does this automatically when required,
		886	but when overriding the invoke callback this call comes handy.
		887
		888	=item int ev_pending_count (loop)
		889
		890	Returns the number of pending watchers - zero indicates that no watchers
		891	are pending.
		892
		893	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		894
		895	This overrides the invoke pending functionality of the loop: Instead of
		896	invoking all pending watchers when there are any, C<ev_loop> will call
		897	this callback instead. This is useful, for example, when you want to
		898	invoke the actual watchers inside another context (another thread etc.).
		899
		900	If you want to reset the callback, use C<ev_invoke_pending> as new
		901	callback.
		902
		903	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		904
		905	Sometimes you want to share the same loop between multiple threads. This
		906	can be done relatively simply by putting mutex_lock/unlock calls around
		907	each call to a libev function.
		908
		909	However, C<ev_loop> can run an indefinite time, so it is not feasible to
		910	wait for it to return. One way around this is to wake up the loop via
		911	C<ev_unloop> and C<av_async_send>, another way is to set these I<release>
		912	and I<acquire> callbacks on the loop.
		913
		914	When set, then C<release> will be called just before the thread is
		915	suspended waiting for new events, and C<acquire> is called just
		916	afterwards.
		917
		918	Ideally, C<release> will just call your mutex_unlock function, and
		919	C<acquire> will just call the mutex_lock function again.
		920
		921	While event loop modifications are allowed between invocations of
		922	C<release> and C<acquire> (that's their only purpose after all), no
		923	modifications done will affect the event loop, i.e. adding watchers will
		924	have no effect on the set of file descriptors being watched, or the time
		925	waited. Use an C<ev_async> watcher to wake up C<ev_loop> when you want it
		926	to take note of any changes you made.
		927
		928	In theory, threads executing C<ev_loop> will be async-cancel safe between
		929	invocations of C<release> and C<acquire>.
		930
		931	See also the locking example in the C<THREADS> section later in this
		932	document.
		933
		934	=item ev_set_userdata (loop, void *data)
		935
		936	=item ev_userdata (loop)
		937
		938	Set and retrieve a single C<void *> associated with a loop. When
		939	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		940	C<0.>
		941
		942	These two functions can be used to associate arbitrary data with a loop,
		943	and are intended solely for the C<invoke_pending_cb>, C<release> and
		944	C<acquire> callbacks described above, but of course can be (ab-)used for
		945	any other purpose as well.
		946
		947	=item ev_loop_verify (loop)
		948
		949	This function only does something when C<EV_VERIFY> support has been
		950	compiled in, which is the default for non-minimal builds. It tries to go
		951	through all internal structures and checks them for validity. If anything
		952	is found to be inconsistent, it will print an error message to standard
		953	error and call C<abort ()>.
		954
		955	This can be used to catch bugs inside libev itself: under normal
		956	circumstances, this function will never abort as of course libev keeps its
		957	data structures consistent.
258		958
259	=back	959	=back
260		960
		961
261	=head1 ANATOMY OF A WATCHER	962	=head1 ANATOMY OF A WATCHER
		963
		964	In the following description, uppercase C<TYPE> in names stands for the
		965	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		966	watchers and C<ev_io_start> for I/O watchers.
262		967
263	A watcher is a structure that you create and register to record your	968	A watcher is a structure that you create and register to record your
264	interest in some event. For instance, if you want to wait for STDIN to	969	interest in some event. For instance, if you want to wait for STDIN to
265	become readable, you would create an C<ev_io> watcher for that:	970	become readable, you would create an C<ev_io> watcher for that:
266		971
267	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	972	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
268	{	973	{
269	ev_io_stop (w);	974	ev_io_stop (w);
270	ev_unloop (loop, EVUNLOOP_ALL);	975	ev_unloop (loop, EVUNLOOP_ALL);
271	}	976	}
272		977
273	struct ev_loop *loop = ev_default_loop (0);	978	struct ev_loop *loop = ev_default_loop (0);
		979
274	struct ev_io stdin_watcher;	980	ev_io stdin_watcher;
		981
275	ev_init (&stdin_watcher, my_cb);	982	ev_init (&stdin_watcher, my_cb);
276	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	983	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
277	ev_io_start (loop, &stdin_watcher);	984	ev_io_start (loop, &stdin_watcher);
		985
278	ev_loop (loop, 0);	986	ev_loop (loop, 0);
279		987
280	As you can see, you are responsible for allocating the memory for your	988	As you can see, you are responsible for allocating the memory for your
281	watcher structures (and it is usually a bad idea to do this on the stack,	989	watcher structures (and it is I<usually> a bad idea to do this on the
282	although this can sometimes be quite valid).	990	stack).
		991
		992	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		993	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
283		994
284	Each watcher structure must be initialised by a call to C<ev_init	995	Each watcher structure must be initialised by a call to C<ev_init
285	(watcher *, callback)>, which expects a callback to be provided. This	996	(watcher *, callback)>, which expects a callback to be provided. This
286	callback gets invoked each time the event occurs (or, in the case of io	997	callback gets invoked each time the event occurs (or, in the case of I/O
287	watchers, each time the event loop detects that the file descriptor given	998	watchers, each time the event loop detects that the file descriptor given
288	is readable and/or writable).	999	is readable and/or writable).
289		1000
290	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1001	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
291	with arguments specific to this watcher type. There is also a macro	1002	macro to configure it, with arguments specific to the watcher type. There
292	to combine initialisation and setting in one call: C<< ev_<type>_init	1003	is also a macro to combine initialisation and setting in one call: C<<
293	(watcher *, callback, ...) >>.	1004	ev_TYPE_init (watcher *, callback, ...) >>.
294		1005
295	To make the watcher actually watch out for events, you have to start it	1006	To make the watcher actually watch out for events, you have to start it
296	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1007	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
297	*) >>), and you can stop watching for events at any time by calling the	1008	*) >>), and you can stop watching for events at any time by calling the
298	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1009	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
299		1010
300	As long as your watcher is active (has been started but not stopped) you	1011	As long as your watcher is active (has been started but not stopped) you
301	must not touch the values stored in it. Most specifically you must never	1012	must not touch the values stored in it. Most specifically you must never
302	reinitialise it or call its set method.	1013	reinitialise it or call its C<ev_TYPE_set> macro.
303
304	You can check whether an event is active by calling the C<ev_is_active
305	(watcher *)> macro. To see whether an event is outstanding (but the
306	callback for it has not been called yet) you can use the C<ev_is_pending
307	(watcher *)> macro.
308		1014
309	Each and every callback receives the event loop pointer as first, the	1015	Each and every callback receives the event loop pointer as first, the
310	registered watcher structure as second, and a bitset of received events as	1016	registered watcher structure as second, and a bitset of received events as
311	third argument.	1017	third argument.
312		1018
…		…
336	The signal specified in the C<ev_signal> watcher has been received by a thread.	1042	The signal specified in the C<ev_signal> watcher has been received by a thread.
337		1043
338	=item C<EV_CHILD>	1044	=item C<EV_CHILD>
339		1045
340	The pid specified in the C<ev_child> watcher has received a status change.	1046	The pid specified in the C<ev_child> watcher has received a status change.
		1047
		1048	=item C<EV_STAT>
		1049
		1050	The path specified in the C<ev_stat> watcher changed its attributes somehow.
341		1051
342	=item C<EV_IDLE>	1052	=item C<EV_IDLE>
343		1053
344	The C<ev_idle> watcher has determined that you have nothing better to do.	1054	The C<ev_idle> watcher has determined that you have nothing better to do.
345		1055
…		…
353	received events. Callbacks of both watcher types can start and stop as	1063	received events. Callbacks of both watcher types can start and stop as
354	many watchers as they want, and all of them will be taken into account	1064	many watchers as they want, and all of them will be taken into account
355	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1065	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
356	C<ev_loop> from blocking).	1066	C<ev_loop> from blocking).
357		1067
		1068	=item C<EV_EMBED>
		1069
		1070	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1071
		1072	=item C<EV_FORK>
		1073
		1074	The event loop has been resumed in the child process after fork (see
		1075	C<ev_fork>).
		1076
		1077	=item C<EV_ASYNC>
		1078
		1079	The given async watcher has been asynchronously notified (see C<ev_async>).
		1080
		1081	=item C<EV_CUSTOM>
		1082
		1083	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1084	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1085
358	=item C<EV_ERROR>	1086	=item C<EV_ERROR>
359		1087
360	An unspecified error has occured, the watcher has been stopped. This might	1088	An unspecified error has occurred, the watcher has been stopped. This might
361	happen because the watcher could not be properly started because libev	1089	happen because the watcher could not be properly started because libev
362	ran out of memory, a file descriptor was found to be closed or any other	1090	ran out of memory, a file descriptor was found to be closed or any other
		1091	problem. Libev considers these application bugs.
		1092
363	problem. You best act on it by reporting the problem and somehow coping	1093	You best act on it by reporting the problem and somehow coping with the
364	with the watcher being stopped.	1094	watcher being stopped. Note that well-written programs should not receive
		1095	an error ever, so when your watcher receives it, this usually indicates a
		1096	bug in your program.
365		1097
366	Libev will usually signal a few "dummy" events together with an error,	1098	Libev will usually signal a few "dummy" events together with an error, for
367	for example it might indicate that a fd is readable or writable, and if	1099	example it might indicate that a fd is readable or writable, and if your
368	your callbacks is well-written it can just attempt the operation and cope	1100	callbacks is well-written it can just attempt the operation and cope with
369	with the error from read() or write(). This will not work in multithreaded	1101	the error from read() or write(). This will not work in multi-threaded
370	programs, though, so beware.	1102	programs, though, as the fd could already be closed and reused for another
		1103	thing, so beware.
371		1104
372	=back	1105	=back
373		1106
		1107	=head2 GENERIC WATCHER FUNCTIONS
		1108
		1109	=over 4
		1110
		1111	=item C<ev_init> (ev_TYPE *watcher, callback)
		1112
		1113	This macro initialises the generic portion of a watcher. The contents
		1114	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1115	the generic parts of the watcher are initialised, you I<need> to call
		1116	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1117	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1118	which rolls both calls into one.
		1119
		1120	You can reinitialise a watcher at any time as long as it has been stopped
		1121	(or never started) and there are no pending events outstanding.
		1122
		1123	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1124	int revents)>.
		1125
		1126	Example: Initialise an C<ev_io> watcher in two steps.
		1127
		1128	ev_io w;
		1129	ev_init (&w, my_cb);
		1130	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1131
		1132	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1133
		1134	This macro initialises the type-specific parts of a watcher. You need to
		1135	call C<ev_init> at least once before you call this macro, but you can
		1136	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1137	macro on a watcher that is active (it can be pending, however, which is a
		1138	difference to the C<ev_init> macro).
		1139
		1140	Although some watcher types do not have type-specific arguments
		1141	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1142
		1143	See C<ev_init>, above, for an example.
		1144
		1145	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1146
		1147	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1148	calls into a single call. This is the most convenient method to initialise
		1149	a watcher. The same limitations apply, of course.
		1150
		1151	Example: Initialise and set an C<ev_io> watcher in one step.
		1152
		1153	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1154
		1155	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1156
		1157	Starts (activates) the given watcher. Only active watchers will receive
		1158	events. If the watcher is already active nothing will happen.
		1159
		1160	Example: Start the C<ev_io> watcher that is being abused as example in this
		1161	whole section.
		1162
		1163	ev_io_start (EV_DEFAULT_UC, &w);
		1164
		1165	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1166
		1167	Stops the given watcher if active, and clears the pending status (whether
		1168	the watcher was active or not).
		1169
		1170	It is possible that stopped watchers are pending - for example,
		1171	non-repeating timers are being stopped when they become pending - but
		1172	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1173	pending. If you want to free or reuse the memory used by the watcher it is
		1174	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1175
		1176	=item bool ev_is_active (ev_TYPE *watcher)
		1177
		1178	Returns a true value iff the watcher is active (i.e. it has been started
		1179	and not yet been stopped). As long as a watcher is active you must not modify
		1180	it.
		1181
		1182	=item bool ev_is_pending (ev_TYPE *watcher)
		1183
		1184	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1185	events but its callback has not yet been invoked). As long as a watcher
		1186	is pending (but not active) you must not call an init function on it (but
		1187	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1188	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1189	it).
		1190
		1191	=item callback ev_cb (ev_TYPE *watcher)
		1192
		1193	Returns the callback currently set on the watcher.
		1194
		1195	=item ev_cb_set (ev_TYPE *watcher, callback)
		1196
		1197	Change the callback. You can change the callback at virtually any time
		1198	(modulo threads).
		1199
		1200	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1201
		1202	=item int ev_priority (ev_TYPE *watcher)
		1203
		1204	Set and query the priority of the watcher. The priority is a small
		1205	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1206	(default: C<-2>). Pending watchers with higher priority will be invoked
		1207	before watchers with lower priority, but priority will not keep watchers
		1208	from being executed (except for C<ev_idle> watchers).
		1209
		1210	If you need to suppress invocation when higher priority events are pending
		1211	you need to look at C<ev_idle> watchers, which provide this functionality.
		1212
		1213	You I<must not> change the priority of a watcher as long as it is active or
		1214	pending.
		1215
		1216	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1217	fine, as long as you do not mind that the priority value you query might
		1218	or might not have been clamped to the valid range.
		1219
		1220	The default priority used by watchers when no priority has been set is
		1221	always C<0>, which is supposed to not be too high and not be too low :).
		1222
		1223	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1224	priorities.
		1225
		1226	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1227
		1228	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1229	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1230	can deal with that fact, as both are simply passed through to the
		1231	callback.
		1232
		1233	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1234
		1235	If the watcher is pending, this function clears its pending status and
		1236	returns its C<revents> bitset (as if its callback was invoked). If the
		1237	watcher isn't pending it does nothing and returns C<0>.
		1238
		1239	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1240	callback to be invoked, which can be accomplished with this function.
		1241
		1242	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1243
		1244	Feeds the given event set into the event loop, as if the specified event
		1245	had happened for the specified watcher (which must be a pointer to an
		1246	initialised but not necessarily started event watcher). Obviously you must
		1247	not free the watcher as long as it has pending events.
		1248
		1249	Stopping the watcher, letting libev invoke it, or calling
		1250	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1251	not started in the first place.
		1252
		1253	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1254	functions that do not need a watcher.
		1255
		1256	=back
		1257
		1258
374	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1259	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
375		1260
376	Each watcher has, by default, a member C<void *data> that you can change	1261	Each watcher has, by default, a member C<void *data> that you can change
377	and read at any time, libev will completely ignore it. This can be used	1262	and read at any time: libev will completely ignore it. This can be used
378	to associate arbitrary data with your watcher. If you need more data and	1263	to associate arbitrary data with your watcher. If you need more data and
379	don't want to allocate memory and store a pointer to it in that data	1264	don't want to allocate memory and store a pointer to it in that data
380	member, you can also "subclass" the watcher type and provide your own	1265	member, you can also "subclass" the watcher type and provide your own
381	data:	1266	data:
382		1267
383	struct my_io	1268	struct my_io
384	{	1269	{
385	struct ev_io io;	1270	ev_io io;
386	int otherfd;	1271	int otherfd;
387	void *somedata;	1272	void *somedata;
388	struct whatever *mostinteresting;	1273	struct whatever *mostinteresting;
389	}	1274	};
		1275
		1276	...
		1277	struct my_io w;
		1278	ev_io_init (&w.io, my_cb, fd, EV_READ);
390		1279
391	And since your callback will be called with a pointer to the watcher, you	1280	And since your callback will be called with a pointer to the watcher, you
392	can cast it back to your own type:	1281	can cast it back to your own type:
393		1282
394	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1283	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
395	{	1284	{
396	struct my_io *w = (struct my_io *)w_;	1285	struct my_io *w = (struct my_io *)w_;
397	...	1286	...
398	}	1287	}
399		1288
400	More interesting and less C-conformant ways of catsing your callback type	1289	More interesting and less C-conformant ways of casting your callback type
401	have been omitted....	1290	instead have been omitted.
		1291
		1292	Another common scenario is to use some data structure with multiple
		1293	embedded watchers:
		1294
		1295	struct my_biggy
		1296	{
		1297	int some_data;
		1298	ev_timer t1;
		1299	ev_timer t2;
		1300	}
		1301
		1302	In this case getting the pointer to C<my_biggy> is a bit more
		1303	complicated: Either you store the address of your C<my_biggy> struct
		1304	in the C<data> member of the watcher (for woozies), or you need to use
		1305	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1306	programmers):
		1307
		1308	#include <stddef.h>
		1309
		1310	static void
		1311	t1_cb (EV_P_ ev_timer *w, int revents)
		1312	{
		1313	struct my_biggy big = (struct my_biggy *)
		1314	(((char *)w) - offsetof (struct my_biggy, t1));
		1315	}
		1316
		1317	static void
		1318	t2_cb (EV_P_ ev_timer *w, int revents)
		1319	{
		1320	struct my_biggy big = (struct my_biggy *)
		1321	(((char *)w) - offsetof (struct my_biggy, t2));
		1322	}
		1323
		1324	=head2 WATCHER PRIORITY MODELS
		1325
		1326	Many event loops support I<watcher priorities>, which are usually small
		1327	integers that influence the ordering of event callback invocation
		1328	between watchers in some way, all else being equal.
		1329
		1330	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1331	description for the more technical details such as the actual priority
		1332	range.
		1333
		1334	There are two common ways how these these priorities are being interpreted
		1335	by event loops:
		1336
		1337	In the more common lock-out model, higher priorities "lock out" invocation
		1338	of lower priority watchers, which means as long as higher priority
		1339	watchers receive events, lower priority watchers are not being invoked.
		1340
		1341	The less common only-for-ordering model uses priorities solely to order
		1342	callback invocation within a single event loop iteration: Higher priority
		1343	watchers are invoked before lower priority ones, but they all get invoked
		1344	before polling for new events.
		1345
		1346	Libev uses the second (only-for-ordering) model for all its watchers
		1347	except for idle watchers (which use the lock-out model).
		1348
		1349	The rationale behind this is that implementing the lock-out model for
		1350	watchers is not well supported by most kernel interfaces, and most event
		1351	libraries will just poll for the same events again and again as long as
		1352	their callbacks have not been executed, which is very inefficient in the
		1353	common case of one high-priority watcher locking out a mass of lower
		1354	priority ones.
		1355
		1356	Static (ordering) priorities are most useful when you have two or more
		1357	watchers handling the same resource: a typical usage example is having an
		1358	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1359	timeouts. Under load, data might be received while the program handles
		1360	other jobs, but since timers normally get invoked first, the timeout
		1361	handler will be executed before checking for data. In that case, giving
		1362	the timer a lower priority than the I/O watcher ensures that I/O will be
		1363	handled first even under adverse conditions (which is usually, but not
		1364	always, what you want).
		1365
		1366	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1367	will only be executed when no same or higher priority watchers have
		1368	received events, they can be used to implement the "lock-out" model when
		1369	required.
		1370
		1371	For example, to emulate how many other event libraries handle priorities,
		1372	you can associate an C<ev_idle> watcher to each such watcher, and in
		1373	the normal watcher callback, you just start the idle watcher. The real
		1374	processing is done in the idle watcher callback. This causes libev to
		1375	continously poll and process kernel event data for the watcher, but when
		1376	the lock-out case is known to be rare (which in turn is rare :), this is
		1377	workable.
		1378
		1379	Usually, however, the lock-out model implemented that way will perform
		1380	miserably under the type of load it was designed to handle. In that case,
		1381	it might be preferable to stop the real watcher before starting the
		1382	idle watcher, so the kernel will not have to process the event in case
		1383	the actual processing will be delayed for considerable time.
		1384
		1385	Here is an example of an I/O watcher that should run at a strictly lower
		1386	priority than the default, and which should only process data when no
		1387	other events are pending:
		1388
		1389	ev_idle idle; // actual processing watcher
		1390	ev_io io; // actual event watcher
		1391
		1392	static void
		1393	io_cb (EV_P_ ev_io *w, int revents)
		1394	{
		1395	// stop the I/O watcher, we received the event, but
		1396	// are not yet ready to handle it.
		1397	ev_io_stop (EV_A_ w);
		1398
		1399	// start the idle watcher to ahndle the actual event.
		1400	// it will not be executed as long as other watchers
		1401	// with the default priority are receiving events.
		1402	ev_idle_start (EV_A_ &idle);
		1403	}
		1404
		1405	static void
		1406	idle_cb (EV_P_ ev_idle *w, int revents)
		1407	{
		1408	// actual processing
		1409	read (STDIN_FILENO, ...);
		1410
		1411	// have to start the I/O watcher again, as
		1412	// we have handled the event
		1413	ev_io_start (EV_P_ &io);
		1414	}
		1415
		1416	// initialisation
		1417	ev_idle_init (&idle, idle_cb);
		1418	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1419	ev_io_start (EV_DEFAULT_ &io);
		1420
		1421	In the "real" world, it might also be beneficial to start a timer, so that
		1422	low-priority connections can not be locked out forever under load. This
		1423	enables your program to keep a lower latency for important connections
		1424	during short periods of high load, while not completely locking out less
		1425	important ones.
402		1426
403		1427
404	=head1 WATCHER TYPES	1428	=head1 WATCHER TYPES
405		1429
406	This section describes each watcher in detail, but will not repeat	1430	This section describes each watcher in detail, but will not repeat
407	information given in the last section.	1431	information given in the last section. Any initialisation/set macros,
		1432	functions and members specific to the watcher type are explained.
408		1433
		1434	Members are additionally marked with either I<[read-only]>, meaning that,
		1435	while the watcher is active, you can look at the member and expect some
		1436	sensible content, but you must not modify it (you can modify it while the
		1437	watcher is stopped to your hearts content), or I<[read-write]>, which
		1438	means you can expect it to have some sensible content while the watcher
		1439	is active, but you can also modify it. Modifying it may not do something
		1440	sensible or take immediate effect (or do anything at all), but libev will
		1441	not crash or malfunction in any way.
		1442
		1443
409	=head2 C<ev_io> - is this file descriptor readable or writable	1444	=head2 C<ev_io> - is this file descriptor readable or writable?
410		1445
411	I/O watchers check whether a file descriptor is readable or writable	1446	I/O watchers check whether a file descriptor is readable or writable
412	in each iteration of the event loop (This behaviour is called	1447	in each iteration of the event loop, or, more precisely, when reading
413	level-triggering because you keep receiving events as long as the	1448	would not block the process and writing would at least be able to write
414	condition persists. Remember you can stop the watcher if you don't want to	1449	some data. This behaviour is called level-triggering because you keep
415	act on the event and neither want to receive future events).	1450	receiving events as long as the condition persists. Remember you can stop
		1451	the watcher if you don't want to act on the event and neither want to
		1452	receive future events.
416		1453
417	In general you can register as many read and/or write event watchers oer	1454	In general you can register as many read and/or write event watchers per
418	fd as you want (as long as you don't confuse yourself). Setting all file	1455	fd as you want (as long as you don't confuse yourself). Setting all file
419	descriptors to non-blocking mode is also usually a good idea (but not	1456	descriptors to non-blocking mode is also usually a good idea (but not
420	required if you know what you are doing).	1457	required if you know what you are doing).
421		1458
422	You have to be careful with dup'ed file descriptors, though. Some backends	1459	If you cannot use non-blocking mode, then force the use of a
423	(the linux epoll backend is a notable example) cannot handle dup'ed file	1460	known-to-be-good backend (at the time of this writing, this includes only
424	descriptors correctly if you register interest in two or more fds pointing	1461	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
425	to the same file/socket etc. description.	1462	descriptors for which non-blocking operation makes no sense (such as
		1463	files) - libev doesn't guarentee any specific behaviour in that case.
426		1464
427	If you must do this, then force the use of a known-to-be-good backend	1465	Another thing you have to watch out for is that it is quite easy to
428	(at the time of this writing, this includes only EVMETHOD_SELECT and	1466	receive "spurious" readiness notifications, that is your callback might
429	EVMETHOD_POLL).	1467	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1468	because there is no data. Not only are some backends known to create a
		1469	lot of those (for example Solaris ports), it is very easy to get into
		1470	this situation even with a relatively standard program structure. Thus
		1471	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1472	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1473
		1474	If you cannot run the fd in non-blocking mode (for example you should
		1475	not play around with an Xlib connection), then you have to separately
		1476	re-test whether a file descriptor is really ready with a known-to-be good
		1477	interface such as poll (fortunately in our Xlib example, Xlib already
		1478	does this on its own, so its quite safe to use). Some people additionally
		1479	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1480	indefinitely.
		1481
		1482	But really, best use non-blocking mode.
		1483
		1484	=head3 The special problem of disappearing file descriptors
		1485
		1486	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1487	descriptor (either due to calling C<close> explicitly or any other means,
		1488	such as C<dup2>). The reason is that you register interest in some file
		1489	descriptor, but when it goes away, the operating system will silently drop
		1490	this interest. If another file descriptor with the same number then is
		1491	registered with libev, there is no efficient way to see that this is, in
		1492	fact, a different file descriptor.
		1493
		1494	To avoid having to explicitly tell libev about such cases, libev follows
		1495	the following policy: Each time C<ev_io_set> is being called, libev
		1496	will assume that this is potentially a new file descriptor, otherwise
		1497	it is assumed that the file descriptor stays the same. That means that
		1498	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1499	descriptor even if the file descriptor number itself did not change.
		1500
		1501	This is how one would do it normally anyway, the important point is that
		1502	the libev application should not optimise around libev but should leave
		1503	optimisations to libev.
		1504
		1505	=head3 The special problem of dup'ed file descriptors
		1506
		1507	Some backends (e.g. epoll), cannot register events for file descriptors,
		1508	but only events for the underlying file descriptions. That means when you
		1509	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1510	events for them, only one file descriptor might actually receive events.
		1511
		1512	There is no workaround possible except not registering events
		1513	for potentially C<dup ()>'ed file descriptors, or to resort to
		1514	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1515
		1516	=head3 The special problem of fork
		1517
		1518	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1519	useless behaviour. Libev fully supports fork, but needs to be told about
		1520	it in the child.
		1521
		1522	To support fork in your programs, you either have to call
		1523	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1524	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1525	C<EVBACKEND_POLL>.
		1526
		1527	=head3 The special problem of SIGPIPE
		1528
		1529	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1530	when writing to a pipe whose other end has been closed, your program gets
		1531	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1532	this is sensible behaviour, for daemons, this is usually undesirable.
		1533
		1534	So when you encounter spurious, unexplained daemon exits, make sure you
		1535	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1536	somewhere, as that would have given you a big clue).
		1537
		1538
		1539	=head3 Watcher-Specific Functions
430		1540
431	=over 4	1541	=over 4
432		1542
433	=item ev_io_init (ev_io *, callback, int fd, int events)	1543	=item ev_io_init (ev_io *, callback, int fd, int events)
434		1544
435	=item ev_io_set (ev_io *, int fd, int events)	1545	=item ev_io_set (ev_io *, int fd, int events)
436		1546
437	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1547	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
438	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1548	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
439	EV_WRITE> to receive the given events.	1549	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1550
		1551	=item int fd [read-only]
		1552
		1553	The file descriptor being watched.
		1554
		1555	=item int events [read-only]
		1556
		1557	The events being watched.
440		1558
441	=back	1559	=back
442		1560
		1561	=head3 Examples
		1562
		1563	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1564	readable, but only once. Since it is likely line-buffered, you could
		1565	attempt to read a whole line in the callback.
		1566
		1567	static void
		1568	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1569	{
		1570	ev_io_stop (loop, w);
		1571	.. read from stdin here (or from w->fd) and handle any I/O errors
		1572	}
		1573
		1574	...
		1575	struct ev_loop *loop = ev_default_init (0);
		1576	ev_io stdin_readable;
		1577	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1578	ev_io_start (loop, &stdin_readable);
		1579	ev_loop (loop, 0);
		1580
		1581
443	=head2 C<ev_timer> - relative and optionally recurring timeouts	1582	=head2 C<ev_timer> - relative and optionally repeating timeouts
444		1583
445	Timer watchers are simple relative timers that generate an event after a	1584	Timer watchers are simple relative timers that generate an event after a
446	given time, and optionally repeating in regular intervals after that.	1585	given time, and optionally repeating in regular intervals after that.
447		1586
448	The timers are based on real time, that is, if you register an event that	1587	The timers are based on real time, that is, if you register an event that
449	times out after an hour and youreset your system clock to last years	1588	times out after an hour and you reset your system clock to January last
450	time, it will still time out after (roughly) and hour. "Roughly" because	1589	year, it will still time out after (roughly) one hour. "Roughly" because
451	detecting time jumps is hard, and soem inaccuracies are unavoidable (the	1590	detecting time jumps is hard, and some inaccuracies are unavoidable (the
452	monotonic clock option helps a lot here).	1591	monotonic clock option helps a lot here).
		1592
		1593	The callback is guaranteed to be invoked only I<after> its timeout has
		1594	passed (not I<at>, so on systems with very low-resolution clocks this
		1595	might introduce a small delay). If multiple timers become ready during the
		1596	same loop iteration then the ones with earlier time-out values are invoked
		1597	before ones of the same priority with later time-out values (but this is
		1598	no longer true when a callback calls C<ev_loop> recursively).
		1599
		1600	=head3 Be smart about timeouts
		1601
		1602	Many real-world problems involve some kind of timeout, usually for error
		1603	recovery. A typical example is an HTTP request - if the other side hangs,
		1604	you want to raise some error after a while.
		1605
		1606	What follows are some ways to handle this problem, from obvious and
		1607	inefficient to smart and efficient.
		1608
		1609	In the following, a 60 second activity timeout is assumed - a timeout that
		1610	gets reset to 60 seconds each time there is activity (e.g. each time some
		1611	data or other life sign was received).
		1612
		1613	=over 4
		1614
		1615	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1616
		1617	This is the most obvious, but not the most simple way: In the beginning,
		1618	start the watcher:
		1619
		1620	ev_timer_init (timer, callback, 60., 0.);
		1621	ev_timer_start (loop, timer);
		1622
		1623	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1624	and start it again:
		1625
		1626	ev_timer_stop (loop, timer);
		1627	ev_timer_set (timer, 60., 0.);
		1628	ev_timer_start (loop, timer);
		1629
		1630	This is relatively simple to implement, but means that each time there is
		1631	some activity, libev will first have to remove the timer from its internal
		1632	data structure and then add it again. Libev tries to be fast, but it's
		1633	still not a constant-time operation.
		1634
		1635	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1636
		1637	This is the easiest way, and involves using C<ev_timer_again> instead of
		1638	C<ev_timer_start>.
		1639
		1640	To implement this, configure an C<ev_timer> with a C<repeat> value
		1641	of C<60> and then call C<ev_timer_again> at start and each time you
		1642	successfully read or write some data. If you go into an idle state where
		1643	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1644	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1645
		1646	That means you can ignore both the C<ev_timer_start> function and the
		1647	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1648	member and C<ev_timer_again>.
		1649
		1650	At start:
		1651
		1652	ev_init (timer, callback);
		1653	timer->repeat = 60.;
		1654	ev_timer_again (loop, timer);
		1655
		1656	Each time there is some activity:
		1657
		1658	ev_timer_again (loop, timer);
		1659
		1660	It is even possible to change the time-out on the fly, regardless of
		1661	whether the watcher is active or not:
		1662
		1663	timer->repeat = 30.;
		1664	ev_timer_again (loop, timer);
		1665
		1666	This is slightly more efficient then stopping/starting the timer each time
		1667	you want to modify its timeout value, as libev does not have to completely
		1668	remove and re-insert the timer from/into its internal data structure.
		1669
		1670	It is, however, even simpler than the "obvious" way to do it.
		1671
		1672	=item 3. Let the timer time out, but then re-arm it as required.
		1673
		1674	This method is more tricky, but usually most efficient: Most timeouts are
		1675	relatively long compared to the intervals between other activity - in
		1676	our example, within 60 seconds, there are usually many I/O events with
		1677	associated activity resets.
		1678
		1679	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1680	but remember the time of last activity, and check for a real timeout only
		1681	within the callback:
		1682
		1683	ev_tstamp last_activity; // time of last activity
		1684
		1685	static void
		1686	callback (EV_P_ ev_timer *w, int revents)
		1687	{
		1688	ev_tstamp now = ev_now (EV_A);
		1689	ev_tstamp timeout = last_activity + 60.;
		1690
		1691	// if last_activity + 60. is older than now, we did time out
		1692	if (timeout < now)
		1693	{
		1694	// timeout occured, take action
		1695	}
		1696	else
		1697	{
		1698	// callback was invoked, but there was some activity, re-arm
		1699	// the watcher to fire in last_activity + 60, which is
		1700	// guaranteed to be in the future, so "again" is positive:
		1701	w->repeat = timeout - now;
		1702	ev_timer_again (EV_A_ w);
		1703	}
		1704	}
		1705
		1706	To summarise the callback: first calculate the real timeout (defined
		1707	as "60 seconds after the last activity"), then check if that time has
		1708	been reached, which means something I<did>, in fact, time out. Otherwise
		1709	the callback was invoked too early (C<timeout> is in the future), so
		1710	re-schedule the timer to fire at that future time, to see if maybe we have
		1711	a timeout then.
		1712
		1713	Note how C<ev_timer_again> is used, taking advantage of the
		1714	C<ev_timer_again> optimisation when the timer is already running.
		1715
		1716	This scheme causes more callback invocations (about one every 60 seconds
		1717	minus half the average time between activity), but virtually no calls to
		1718	libev to change the timeout.
		1719
		1720	To start the timer, simply initialise the watcher and set C<last_activity>
		1721	to the current time (meaning we just have some activity :), then call the
		1722	callback, which will "do the right thing" and start the timer:
		1723
		1724	ev_init (timer, callback);
		1725	last_activity = ev_now (loop);
		1726	callback (loop, timer, EV_TIMEOUT);
		1727
		1728	And when there is some activity, simply store the current time in
		1729	C<last_activity>, no libev calls at all:
		1730
		1731	last_actiivty = ev_now (loop);
		1732
		1733	This technique is slightly more complex, but in most cases where the
		1734	time-out is unlikely to be triggered, much more efficient.
		1735
		1736	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1737	callback :) - just change the timeout and invoke the callback, which will
		1738	fix things for you.
		1739
		1740	=item 4. Wee, just use a double-linked list for your timeouts.
		1741
		1742	If there is not one request, but many thousands (millions...), all
		1743	employing some kind of timeout with the same timeout value, then one can
		1744	do even better:
		1745
		1746	When starting the timeout, calculate the timeout value and put the timeout
		1747	at the I<end> of the list.
		1748
		1749	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1750	the list is expected to fire (for example, using the technique #3).
		1751
		1752	When there is some activity, remove the timer from the list, recalculate
		1753	the timeout, append it to the end of the list again, and make sure to
		1754	update the C<ev_timer> if it was taken from the beginning of the list.
		1755
		1756	This way, one can manage an unlimited number of timeouts in O(1) time for
		1757	starting, stopping and updating the timers, at the expense of a major
		1758	complication, and having to use a constant timeout. The constant timeout
		1759	ensures that the list stays sorted.
		1760
		1761	=back
		1762
		1763	So which method the best?
		1764
		1765	Method #2 is a simple no-brain-required solution that is adequate in most
		1766	situations. Method #3 requires a bit more thinking, but handles many cases
		1767	better, and isn't very complicated either. In most case, choosing either
		1768	one is fine, with #3 being better in typical situations.
		1769
		1770	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1771	rather complicated, but extremely efficient, something that really pays
		1772	off after the first million or so of active timers, i.e. it's usually
		1773	overkill :)
		1774
		1775	=head3 The special problem of time updates
		1776
		1777	Establishing the current time is a costly operation (it usually takes at
		1778	least two system calls): EV therefore updates its idea of the current
		1779	time only before and after C<ev_loop> collects new events, which causes a
		1780	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1781	lots of events in one iteration.
453		1782
454	The relative timeouts are calculated relative to the C<ev_now ()>	1783	The relative timeouts are calculated relative to the C<ev_now ()>
455	time. This is usually the right thing as this timestamp refers to the time	1784	time. This is usually the right thing as this timestamp refers to the time
456	of the event triggering whatever timeout you are modifying/starting. If	1785	of the event triggering whatever timeout you are modifying/starting. If
457	you suspect event processing to be delayed and you *need* to base the timeout	1786	you suspect event processing to be delayed and you I<need> to base the
458	ion the current time, use something like this to adjust for this:	1787	timeout on the current time, use something like this to adjust for this:
459		1788
460	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1789	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
461		1790
		1791	If the event loop is suspended for a long time, you can also force an
		1792	update of the time returned by C<ev_now ()> by calling C<ev_now_update
		1793	()>.
		1794
		1795	=head3 The special problems of suspended animation
		1796
		1797	When you leave the server world it is quite customary to hit machines that
		1798	can suspend/hibernate - what happens to the clocks during such a suspend?
		1799
		1800	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		1801	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		1802	to run until the system is suspended, but they will not advance while the
		1803	system is suspended. That means, on resume, it will be as if the program
		1804	was frozen for a few seconds, but the suspend time will not be counted
		1805	towards C<ev_timer> when a monotonic clock source is used. The real time
		1806	clock advanced as expected, but if it is used as sole clocksource, then a
		1807	long suspend would be detected as a time jump by libev, and timers would
		1808	be adjusted accordingly.
		1809
		1810	I would not be surprised to see different behaviour in different between
		1811	operating systems, OS versions or even different hardware.
		1812
		1813	The other form of suspend (job control, or sending a SIGSTOP) will see a
		1814	time jump in the monotonic clocks and the realtime clock. If the program
		1815	is suspended for a very long time, and monotonic clock sources are in use,
		1816	then you can expect C<ev_timer>s to expire as the full suspension time
		1817	will be counted towards the timers. When no monotonic clock source is in
		1818	use, then libev will again assume a timejump and adjust accordingly.
		1819
		1820	It might be beneficial for this latter case to call C<ev_suspend>
		1821	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		1822	deterministic behaviour in this case (you can do nothing against
		1823	C<SIGSTOP>).
		1824
		1825	=head3 Watcher-Specific Functions and Data Members
		1826
462	=over 4	1827	=over 4
463		1828
464	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1829	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
465		1830
466	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1831	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
467		1832
468	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1833	Configure the timer to trigger after C<after> seconds. If C<repeat>
469	C<0.>, then it will automatically be stopped. If it is positive, then the	1834	is C<0.>, then it will automatically be stopped once the timeout is
470	timer will automatically be configured to trigger again C<repeat> seconds	1835	reached. If it is positive, then the timer will automatically be
471	later, again, and again, until stopped manually.	1836	configured to trigger again C<repeat> seconds later, again, and again,
		1837	until stopped manually.
472		1838
473	The timer itself will do a best-effort at avoiding drift, that is, if you	1839	The timer itself will do a best-effort at avoiding drift, that is, if
474	configure a timer to trigger every 10 seconds, then it will trigger at	1840	you configure a timer to trigger every 10 seconds, then it will normally
475	exactly 10 second intervals. If, however, your program cannot keep up with	1841	trigger at exactly 10 second intervals. If, however, your program cannot
476	the timer (ecause it takes longer than those 10 seconds to do stuff) the	1842	keep up with the timer (because it takes longer than those 10 seconds to
477	timer will not fire more than once per event loop iteration.	1843	do stuff) the timer will not fire more than once per event loop iteration.
478		1844
479	=item ev_timer_again (loop)	1845	=item ev_timer_again (loop, ev_timer *)
480		1846
481	This will act as if the timer timed out and restart it again if it is	1847	This will act as if the timer timed out and restart it again if it is
482	repeating. The exact semantics are:	1848	repeating. The exact semantics are:
483		1849
		1850	If the timer is pending, its pending status is cleared.
		1851
484	If the timer is started but nonrepeating, stop it.	1852	If the timer is started but non-repeating, stop it (as if it timed out).
485		1853
486	If the timer is repeating, either start it if necessary (with the repeat	1854	If the timer is repeating, either start it if necessary (with the
487	value), or reset the running timer to the repeat value.	1855	C<repeat> value), or reset the running timer to the C<repeat> value.
488		1856
489	This sounds a bit complicated, but here is a useful and typical	1857	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
490	example: Imagine you have a tcp connection and you want a so-called idle	1858	usage example.
491	timeout, that is, you want to be called when there have been, say, 60	1859
492	seconds of inactivity on the socket. The easiest way to do this is to	1860	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
493	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each	1861
494	time you successfully read or write some data. If you go into an idle	1862	Returns the remaining time until a timer fires. If the timer is active,
495	state where you do not expect data to travel on the socket, you can stop	1863	then this time is relative to the current event loop time, otherwise it's
496	the timer, and again will automatically restart it if need be.	1864	the timeout value currently configured.
		1865
		1866	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		1867	C<5>. When the timer is started and one second passes, C<ev_timer_remain>
		1868	will return C<4>. When the timer expires and is restarted, it will return
		1869	roughly C<7> (likely slightly less as callback invocation takes some time,
		1870	too), and so on.
		1871
		1872	=item ev_tstamp repeat [read-write]
		1873
		1874	The current C<repeat> value. Will be used each time the watcher times out
		1875	or C<ev_timer_again> is called, and determines the next timeout (if any),
		1876	which is also when any modifications are taken into account.
497		1877
498	=back	1878	=back
499		1879
		1880	=head3 Examples
		1881
		1882	Example: Create a timer that fires after 60 seconds.
		1883
		1884	static void
		1885	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1886	{
		1887	.. one minute over, w is actually stopped right here
		1888	}
		1889
		1890	ev_timer mytimer;
		1891	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		1892	ev_timer_start (loop, &mytimer);
		1893
		1894	Example: Create a timeout timer that times out after 10 seconds of
		1895	inactivity.
		1896
		1897	static void
		1898	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1899	{
		1900	.. ten seconds without any activity
		1901	}
		1902
		1903	ev_timer mytimer;
		1904	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		1905	ev_timer_again (&mytimer); /* start timer */
		1906	ev_loop (loop, 0);
		1907
		1908	// and in some piece of code that gets executed on any "activity":
		1909	// reset the timeout to start ticking again at 10 seconds
		1910	ev_timer_again (&mytimer);
		1911
		1912
500	=head2 C<ev_periodic> - to cron or not to cron	1913	=head2 C<ev_periodic> - to cron or not to cron?
501		1914
502	Periodic watchers are also timers of a kind, but they are very versatile	1915	Periodic watchers are also timers of a kind, but they are very versatile
503	(and unfortunately a bit complex).	1916	(and unfortunately a bit complex).
504		1917
505	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1918	Unlike C<ev_timer>, periodic watchers are not based on real time (or
506	but on wallclock time (absolute time). You can tell a periodic watcher	1919	relative time, the physical time that passes) but on wall clock time
507	to trigger "at" some specific point in time. For example, if you tell a	1920	(absolute time, the thing you can read on your calender or clock). The
508	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	1921	difference is that wall clock time can run faster or slower than real
509	+ 10.>) and then reset your system clock to the last year, then it will	1922	time, and time jumps are not uncommon (e.g. when you adjust your
510	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1923	wrist-watch).
511	roughly 10 seconds later and of course not if you reset your system time
512	again).
513		1924
514	They can also be used to implement vastly more complex timers, such as	1925	You can tell a periodic watcher to trigger after some specific point
		1926	in time: for example, if you tell a periodic watcher to trigger "in 10
		1927	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		1928	not a delay) and then reset your system clock to January of the previous
		1929	year, then it will take a year or more to trigger the event (unlike an
		1930	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		1931	it, as it uses a relative timeout).
		1932
		1933	C<ev_periodic> watchers can also be used to implement vastly more complex
515	triggering an event on eahc midnight, local time.	1934	timers, such as triggering an event on each "midnight, local time", or
		1935	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		1936	those cannot react to time jumps.
		1937
		1938	As with timers, the callback is guaranteed to be invoked only when the
		1939	point in time where it is supposed to trigger has passed. If multiple
		1940	timers become ready during the same loop iteration then the ones with
		1941	earlier time-out values are invoked before ones with later time-out values
		1942	(but this is no longer true when a callback calls C<ev_loop> recursively).
		1943
		1944	=head3 Watcher-Specific Functions and Data Members
516		1945
517	=over 4	1946	=over 4
518		1947
519	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1948	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
520		1949
521	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1950	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
522		1951
523	Lots of arguments, lets sort it out... There are basically three modes of	1952	Lots of arguments, let's sort it out... There are basically three modes of
524	operation, and we will explain them from simplest to complex:	1953	operation, and we will explain them from simplest to most complex:
525
526		1954
527	=over 4	1955	=over 4
528		1956
529	=item * absolute timer (interval = reschedule_cb = 0)	1957	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
530		1958
531	In this configuration the watcher triggers an event at the wallclock time	1959	In this configuration the watcher triggers an event after the wall clock
532	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1960	time C<offset> has passed. It will not repeat and will not adjust when a
533	that is, if it is to be run at January 1st 2011 then it will run when the	1961	time jump occurs, that is, if it is to be run at January 1st 2011 then it
534	system time reaches or surpasses this time.	1962	will be stopped and invoked when the system clock reaches or surpasses
		1963	this point in time.
535		1964
536	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1965	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
537		1966
538	In this mode the watcher will always be scheduled to time out at the next	1967	In this mode the watcher will always be scheduled to time out at the next
539	C<at + N * interval> time (for some integer N) and then repeat, regardless	1968	C<offset + N * interval> time (for some integer N, which can also be
540	of any time jumps.	1969	negative) and then repeat, regardless of any time jumps. The C<offset>
		1970	argument is merely an offset into the C<interval> periods.
541		1971
542	This can be used to create timers that do not drift with respect to system	1972	This can be used to create timers that do not drift with respect to the
543	time:	1973	system clock, for example, here is an C<ev_periodic> that triggers each
		1974	hour, on the hour (with respect to UTC):
544		1975
545	ev_periodic_set (&periodic, 0., 3600., 0);	1976	ev_periodic_set (&periodic, 0., 3600., 0);
546		1977
547	This doesn't mean there will always be 3600 seconds in between triggers,	1978	This doesn't mean there will always be 3600 seconds in between triggers,
548	but only that the the callback will be called when the system time shows a	1979	but only that the callback will be called when the system time shows a
549	full hour (UTC), or more correctly, when the system time is evenly divisible	1980	full hour (UTC), or more correctly, when the system time is evenly divisible
550	by 3600.	1981	by 3600.
551		1982
552	Another way to think about it (for the mathematically inclined) is that	1983	Another way to think about it (for the mathematically inclined) is that
553	C<ev_periodic> will try to run the callback in this mode at the next possible	1984	C<ev_periodic> will try to run the callback in this mode at the next possible
554	time where C<time = at (mod interval)>, regardless of any time jumps.	1985	time where C<time = offset (mod interval)>, regardless of any time jumps.
555		1986
		1987	For numerical stability it is preferable that the C<offset> value is near
		1988	C<ev_now ()> (the current time), but there is no range requirement for
		1989	this value, and in fact is often specified as zero.
		1990
		1991	Note also that there is an upper limit to how often a timer can fire (CPU
		1992	speed for example), so if C<interval> is very small then timing stability
		1993	will of course deteriorate. Libev itself tries to be exact to be about one
		1994	millisecond (if the OS supports it and the machine is fast enough).
		1995
556	=item * manual reschedule mode (reschedule_cb = callback)	1996	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
557		1997
558	In this mode the values for C<interval> and C<at> are both being	1998	In this mode the values for C<interval> and C<offset> are both being
559	ignored. Instead, each time the periodic watcher gets scheduled, the	1999	ignored. Instead, each time the periodic watcher gets scheduled, the
560	reschedule callback will be called with the watcher as first, and the	2000	reschedule callback will be called with the watcher as first, and the
561	current time as second argument.	2001	current time as second argument.
562		2002
563	NOTE: I<This callback MUST NOT stop or destroy the periodic or any other	2003	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
564	periodic watcher, ever, or make any event loop modifications>. If you need	2004	or make ANY other event loop modifications whatsoever, unless explicitly
		2005	allowed by documentation here>.
		2006
565	to stop it, return C<now + 1e30> (or so, fudge fudge) and stop it afterwards.	2007	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2008	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2009	only event loop modification you are allowed to do).
566		2010
567	Also, I<< this callback must always return a time that is later than the
568	passed C<now> value >>. Not even C<now> itself will be ok.
569
570	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2011	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
571	ev_tstamp now)>, e.g.:	2012	*w, ev_tstamp now)>, e.g.:
572		2013
		2014	static ev_tstamp
573	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2015	my_rescheduler (ev_periodic *w, ev_tstamp now)
574	{	2016	{
575	return now + 60.;	2017	return now + 60.;
576	}	2018	}
577		2019
578	It must return the next time to trigger, based on the passed time value	2020	It must return the next time to trigger, based on the passed time value
579	(that is, the lowest time value larger than to the second argument). It	2021	(that is, the lowest time value larger than to the second argument). It
580	will usually be called just before the callback will be triggered, but	2022	will usually be called just before the callback will be triggered, but
581	might be called at other times, too.	2023	might be called at other times, too.
582		2024
		2025	NOTE: I<< This callback must always return a time that is higher than or
		2026	equal to the passed C<now> value >>.
		2027
583	This can be used to create very complex timers, such as a timer that	2028	This can be used to create very complex timers, such as a timer that
584	triggers on each midnight, local time. To do this, you would calculate the	2029	triggers on "next midnight, local time". To do this, you would calculate the
585	next midnight after C<now> and return the timestamp value for this. How you do this	2030	next midnight after C<now> and return the timestamp value for this. How
586	is, again, up to you (but it is not trivial).	2031	you do this is, again, up to you (but it is not trivial, which is the main
		2032	reason I omitted it as an example).
587		2033
588	=back	2034	=back
589		2035
590	=item ev_periodic_again (loop, ev_periodic *)	2036	=item ev_periodic_again (loop, ev_periodic *)
591		2037
592	Simply stops and restarts the periodic watcher again. This is only useful	2038	Simply stops and restarts the periodic watcher again. This is only useful
593	when you changed some parameters or the reschedule callback would return	2039	when you changed some parameters or the reschedule callback would return
594	a different time than the last time it was called (e.g. in a crond like	2040	a different time than the last time it was called (e.g. in a crond like
595	program when the crontabs have changed).	2041	program when the crontabs have changed).
596		2042
		2043	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2044
		2045	When active, returns the absolute time that the watcher is supposed
		2046	to trigger next. This is not the same as the C<offset> argument to
		2047	C<ev_periodic_set>, but indeed works even in interval and manual
		2048	rescheduling modes.
		2049
		2050	=item ev_tstamp offset [read-write]
		2051
		2052	When repeating, this contains the offset value, otherwise this is the
		2053	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2054	although libev might modify this value for better numerical stability).
		2055
		2056	Can be modified any time, but changes only take effect when the periodic
		2057	timer fires or C<ev_periodic_again> is being called.
		2058
		2059	=item ev_tstamp interval [read-write]
		2060
		2061	The current interval value. Can be modified any time, but changes only
		2062	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2063	called.
		2064
		2065	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2066
		2067	The current reschedule callback, or C<0>, if this functionality is
		2068	switched off. Can be changed any time, but changes only take effect when
		2069	the periodic timer fires or C<ev_periodic_again> is being called.
		2070
597	=back	2071	=back
598		2072
		2073	=head3 Examples
		2074
		2075	Example: Call a callback every hour, or, more precisely, whenever the
		2076	system time is divisible by 3600. The callback invocation times have
		2077	potentially a lot of jitter, but good long-term stability.
		2078
		2079	static void
		2080	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
		2081	{
		2082	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2083	}
		2084
		2085	ev_periodic hourly_tick;
		2086	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2087	ev_periodic_start (loop, &hourly_tick);
		2088
		2089	Example: The same as above, but use a reschedule callback to do it:
		2090
		2091	#include <math.h>
		2092
		2093	static ev_tstamp
		2094	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2095	{
		2096	return now + (3600. - fmod (now, 3600.));
		2097	}
		2098
		2099	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2100
		2101	Example: Call a callback every hour, starting now:
		2102
		2103	ev_periodic hourly_tick;
		2104	ev_periodic_init (&hourly_tick, clock_cb,
		2105	fmod (ev_now (loop), 3600.), 3600., 0);
		2106	ev_periodic_start (loop, &hourly_tick);
		2107
		2108
599	=head2 C<ev_signal> - signal me when a signal gets signalled	2109	=head2 C<ev_signal> - signal me when a signal gets signalled!
600		2110
601	Signal watchers will trigger an event when the process receives a specific	2111	Signal watchers will trigger an event when the process receives a specific
602	signal one or more times. Even though signals are very asynchronous, libev	2112	signal one or more times. Even though signals are very asynchronous, libev
603	will try it's best to deliver signals synchronously, i.e. as part of the	2113	will try it's best to deliver signals synchronously, i.e. as part of the
604	normal event processing, like any other event.	2114	normal event processing, like any other event.
605		2115
		2116	If you want signals to be delivered truly asynchronously, just use
		2117	C<sigaction> as you would do without libev and forget about sharing
		2118	the signal. You can even use C<ev_async> from a signal handler to
		2119	synchronously wake up an event loop.
		2120
606	You can configure as many watchers as you like per signal. Only when the	2121	You can configure as many watchers as you like for the same signal, but
		2122	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2123	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2124	C<SIGINT> in both the default loop and another loop at the same time. At
		2125	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2126
607	first watcher gets started will libev actually register a signal watcher	2127	When the first watcher gets started will libev actually register something
608	with the kernel (thus it coexists with your own signal handlers as long	2128	with the kernel (thus it coexists with your own signal handlers as long as
609	as you don't register any with libev). Similarly, when the last signal	2129	you don't register any with libev for the same signal).
610	watcher for a signal is stopped libev will reset the signal handler to	2130
611	SIG_DFL (regardless of what it was set to before).	2131	If possible and supported, libev will install its handlers with
		2132	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2133	not be unduly interrupted. If you have a problem with system calls getting
		2134	interrupted by signals you can block all signals in an C<ev_check> watcher
		2135	and unblock them in an C<ev_prepare> watcher.
		2136
		2137	=head3 The special problem of inheritance over execve
		2138
		2139	Both the signal mask (C<sigprocmask>) and the signal disposition
		2140	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2141	stopping it again), that is, libev might or might not block the signal,
		2142	and might or might not set or restore the installed signal handler.
		2143
		2144	While this does not matter for the signal disposition (libev never
		2145	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2146	C<execve>), this matters for the signal mask: many programs do not expect
		2147	certain signals to be blocked.
		2148
		2149	This means that before calling C<exec> (from the child) you should reset
		2150	the signal mask to whatever "default" you expect (all clear is a good
		2151	choice usually).
		2152
		2153	The simplest way to ensure that the signal mask is reset in the child is
		2154	to install a fork handler with C<pthread_atfork> that resets it. That will
		2155	catch fork calls done by libraries (such as the libc) as well.
		2156
		2157	In current versions of libev, you can also ensure that the signal mask is
		2158	not blocking any signals (except temporarily, so thread users watch out)
		2159	by specifying the C<EVFLAG_NOSIGFD> when creating the event loop. This
		2160	is not guaranteed for future versions, however.
		2161
		2162	=head3 Watcher-Specific Functions and Data Members
612		2163
613	=over 4	2164	=over 4
614		2165
615	=item ev_signal_init (ev_signal *, callback, int signum)	2166	=item ev_signal_init (ev_signal *, callback, int signum)
616		2167
617	=item ev_signal_set (ev_signal *, int signum)	2168	=item ev_signal_set (ev_signal *, int signum)
618		2169
619	Configures the watcher to trigger on the given signal number (usually one	2170	Configures the watcher to trigger on the given signal number (usually one
620	of the C<SIGxxx> constants).	2171	of the C<SIGxxx> constants).
621		2172
		2173	=item int signum [read-only]
		2174
		2175	The signal the watcher watches out for.
		2176
622	=back	2177	=back
623		2178
		2179	=head3 Examples
		2180
		2181	Example: Try to exit cleanly on SIGINT.
		2182
		2183	static void
		2184	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2185	{
		2186	ev_unloop (loop, EVUNLOOP_ALL);
		2187	}
		2188
		2189	ev_signal signal_watcher;
		2190	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2191	ev_signal_start (loop, &signal_watcher);
		2192
		2193
624	=head2 C<ev_child> - wait for pid status changes	2194	=head2 C<ev_child> - watch out for process status changes
625		2195
626	Child watchers trigger when your process receives a SIGCHLD in response to	2196	Child watchers trigger when your process receives a SIGCHLD in response to
627	some child status changes (most typically when a child of yours dies).	2197	some child status changes (most typically when a child of yours dies or
		2198	exits). It is permissible to install a child watcher I<after> the child
		2199	has been forked (which implies it might have already exited), as long
		2200	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2201	forking and then immediately registering a watcher for the child is fine,
		2202	but forking and registering a watcher a few event loop iterations later or
		2203	in the next callback invocation is not.
		2204
		2205	Only the default event loop is capable of handling signals, and therefore
		2206	you can only register child watchers in the default event loop.
		2207
		2208	Due to some design glitches inside libev, child watchers will always be
		2209	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2210	libev)
		2211
		2212	=head3 Process Interaction
		2213
		2214	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2215	initialised. This is necessary to guarantee proper behaviour even if the
		2216	first child watcher is started after the child exits. The occurrence
		2217	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2218	synchronously as part of the event loop processing. Libev always reaps all
		2219	children, even ones not watched.
		2220
		2221	=head3 Overriding the Built-In Processing
		2222
		2223	Libev offers no special support for overriding the built-in child
		2224	processing, but if your application collides with libev's default child
		2225	handler, you can override it easily by installing your own handler for
		2226	C<SIGCHLD> after initialising the default loop, and making sure the
		2227	default loop never gets destroyed. You are encouraged, however, to use an
		2228	event-based approach to child reaping and thus use libev's support for
		2229	that, so other libev users can use C<ev_child> watchers freely.
		2230
		2231	=head3 Stopping the Child Watcher
		2232
		2233	Currently, the child watcher never gets stopped, even when the
		2234	child terminates, so normally one needs to stop the watcher in the
		2235	callback. Future versions of libev might stop the watcher automatically
		2236	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2237	problem).
		2238
		2239	=head3 Watcher-Specific Functions and Data Members
628		2240
629	=over 4	2241	=over 4
630		2242
631	=item ev_child_init (ev_child *, callback, int pid)	2243	=item ev_child_init (ev_child *, callback, int pid, int trace)
632		2244
633	=item ev_child_set (ev_child *, int pid)	2245	=item ev_child_set (ev_child *, int pid, int trace)
634		2246
635	Configures the watcher to wait for status changes of process C<pid> (or	2247	Configures the watcher to wait for status changes of process C<pid> (or
636	I<any> process if C<pid> is specified as C<0>). The callback can look	2248	I<any> process if C<pid> is specified as C<0>). The callback can look
637	at the C<rstatus> member of the C<ev_child> watcher structure to see	2249	at the C<rstatus> member of the C<ev_child> watcher structure to see
638	the status word (use the macros from C<sys/wait.h> and see your systems	2250	the status word (use the macros from C<sys/wait.h> and see your systems
639	C<waitpid> documentation). The C<rpid> member contains the pid of the	2251	C<waitpid> documentation). The C<rpid> member contains the pid of the
640	process causing the status change.	2252	process causing the status change. C<trace> must be either C<0> (only
		2253	activate the watcher when the process terminates) or C<1> (additionally
		2254	activate the watcher when the process is stopped or continued).
		2255
		2256	=item int pid [read-only]
		2257
		2258	The process id this watcher watches out for, or C<0>, meaning any process id.
		2259
		2260	=item int rpid [read-write]
		2261
		2262	The process id that detected a status change.
		2263
		2264	=item int rstatus [read-write]
		2265
		2266	The process exit/trace status caused by C<rpid> (see your systems
		2267	C<waitpid> and C<sys/wait.h> documentation for details).
641		2268
642	=back	2269	=back
643		2270
		2271	=head3 Examples
		2272
		2273	Example: C<fork()> a new process and install a child handler to wait for
		2274	its completion.
		2275
		2276	ev_child cw;
		2277
		2278	static void
		2279	child_cb (EV_P_ ev_child *w, int revents)
		2280	{
		2281	ev_child_stop (EV_A_ w);
		2282	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2283	}
		2284
		2285	pid_t pid = fork ();
		2286
		2287	if (pid < 0)
		2288	// error
		2289	else if (pid == 0)
		2290	{
		2291	// the forked child executes here
		2292	exit (1);
		2293	}
		2294	else
		2295	{
		2296	ev_child_init (&cw, child_cb, pid, 0);
		2297	ev_child_start (EV_DEFAULT_ &cw);
		2298	}
		2299
		2300
		2301	=head2 C<ev_stat> - did the file attributes just change?
		2302
		2303	This watches a file system path for attribute changes. That is, it calls
		2304	C<stat> on that path in regular intervals (or when the OS says it changed)
		2305	and sees if it changed compared to the last time, invoking the callback if
		2306	it did.
		2307
		2308	The path does not need to exist: changing from "path exists" to "path does
		2309	not exist" is a status change like any other. The condition "path does not
		2310	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2311	C<st_nlink> field being zero (which is otherwise always forced to be at
		2312	least one) and all the other fields of the stat buffer having unspecified
		2313	contents.
		2314
		2315	The path I<must not> end in a slash or contain special components such as
		2316	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2317	your working directory changes, then the behaviour is undefined.
		2318
		2319	Since there is no portable change notification interface available, the
		2320	portable implementation simply calls C<stat(2)> regularly on the path
		2321	to see if it changed somehow. You can specify a recommended polling
		2322	interval for this case. If you specify a polling interval of C<0> (highly
		2323	recommended!) then a I<suitable, unspecified default> value will be used
		2324	(which you can expect to be around five seconds, although this might
		2325	change dynamically). Libev will also impose a minimum interval which is
		2326	currently around C<0.1>, but that's usually overkill.
		2327
		2328	This watcher type is not meant for massive numbers of stat watchers,
		2329	as even with OS-supported change notifications, this can be
		2330	resource-intensive.
		2331
		2332	At the time of this writing, the only OS-specific interface implemented
		2333	is the Linux inotify interface (implementing kqueue support is left as an
		2334	exercise for the reader. Note, however, that the author sees no way of
		2335	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2336
		2337	=head3 ABI Issues (Largefile Support)
		2338
		2339	Libev by default (unless the user overrides this) uses the default
		2340	compilation environment, which means that on systems with large file
		2341	support disabled by default, you get the 32 bit version of the stat
		2342	structure. When using the library from programs that change the ABI to
		2343	use 64 bit file offsets the programs will fail. In that case you have to
		2344	compile libev with the same flags to get binary compatibility. This is
		2345	obviously the case with any flags that change the ABI, but the problem is
		2346	most noticeably displayed with ev_stat and large file support.
		2347
		2348	The solution for this is to lobby your distribution maker to make large
		2349	file interfaces available by default (as e.g. FreeBSD does) and not
		2350	optional. Libev cannot simply switch on large file support because it has
		2351	to exchange stat structures with application programs compiled using the
		2352	default compilation environment.
		2353
		2354	=head3 Inotify and Kqueue
		2355
		2356	When C<inotify (7)> support has been compiled into libev and present at
		2357	runtime, it will be used to speed up change detection where possible. The
		2358	inotify descriptor will be created lazily when the first C<ev_stat>
		2359	watcher is being started.
		2360
		2361	Inotify presence does not change the semantics of C<ev_stat> watchers
		2362	except that changes might be detected earlier, and in some cases, to avoid
		2363	making regular C<stat> calls. Even in the presence of inotify support
		2364	there are many cases where libev has to resort to regular C<stat> polling,
		2365	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2366	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2367	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2368	xfs are fully working) libev usually gets away without polling.
		2369
		2370	There is no support for kqueue, as apparently it cannot be used to
		2371	implement this functionality, due to the requirement of having a file
		2372	descriptor open on the object at all times, and detecting renames, unlinks
		2373	etc. is difficult.
		2374
		2375	=head3 C<stat ()> is a synchronous operation
		2376
		2377	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2378	the process. The exception are C<ev_stat> watchers - those call C<stat
		2379	()>, which is a synchronous operation.
		2380
		2381	For local paths, this usually doesn't matter: unless the system is very
		2382	busy or the intervals between stat's are large, a stat call will be fast,
		2383	as the path data is usually in memory already (except when starting the
		2384	watcher).
		2385
		2386	For networked file systems, calling C<stat ()> can block an indefinite
		2387	time due to network issues, and even under good conditions, a stat call
		2388	often takes multiple milliseconds.
		2389
		2390	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2391	paths, although this is fully supported by libev.
		2392
		2393	=head3 The special problem of stat time resolution
		2394
		2395	The C<stat ()> system call only supports full-second resolution portably,
		2396	and even on systems where the resolution is higher, most file systems
		2397	still only support whole seconds.
		2398
		2399	That means that, if the time is the only thing that changes, you can
		2400	easily miss updates: on the first update, C<ev_stat> detects a change and
		2401	calls your callback, which does something. When there is another update
		2402	within the same second, C<ev_stat> will be unable to detect unless the
		2403	stat data does change in other ways (e.g. file size).
		2404
		2405	The solution to this is to delay acting on a change for slightly more
		2406	than a second (or till slightly after the next full second boundary), using
		2407	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2408	ev_timer_again (loop, w)>).
		2409
		2410	The C<.02> offset is added to work around small timing inconsistencies
		2411	of some operating systems (where the second counter of the current time
		2412	might be be delayed. One such system is the Linux kernel, where a call to
		2413	C<gettimeofday> might return a timestamp with a full second later than
		2414	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2415	update file times then there will be a small window where the kernel uses
		2416	the previous second to update file times but libev might already execute
		2417	the timer callback).
		2418
		2419	=head3 Watcher-Specific Functions and Data Members
		2420
		2421	=over 4
		2422
		2423	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2424
		2425	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2426
		2427	Configures the watcher to wait for status changes of the given
		2428	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2429	be detected and should normally be specified as C<0> to let libev choose
		2430	a suitable value. The memory pointed to by C<path> must point to the same
		2431	path for as long as the watcher is active.
		2432
		2433	The callback will receive an C<EV_STAT> event when a change was detected,
		2434	relative to the attributes at the time the watcher was started (or the
		2435	last change was detected).
		2436
		2437	=item ev_stat_stat (loop, ev_stat *)
		2438
		2439	Updates the stat buffer immediately with new values. If you change the
		2440	watched path in your callback, you could call this function to avoid
		2441	detecting this change (while introducing a race condition if you are not
		2442	the only one changing the path). Can also be useful simply to find out the
		2443	new values.
		2444
		2445	=item ev_statdata attr [read-only]
		2446
		2447	The most-recently detected attributes of the file. Although the type is
		2448	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2449	suitable for your system, but you can only rely on the POSIX-standardised
		2450	members to be present. If the C<st_nlink> member is C<0>, then there was
		2451	some error while C<stat>ing the file.
		2452
		2453	=item ev_statdata prev [read-only]
		2454
		2455	The previous attributes of the file. The callback gets invoked whenever
		2456	C<prev> != C<attr>, or, more precisely, one or more of these members
		2457	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2458	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2459
		2460	=item ev_tstamp interval [read-only]
		2461
		2462	The specified interval.
		2463
		2464	=item const char *path [read-only]
		2465
		2466	The file system path that is being watched.
		2467
		2468	=back
		2469
		2470	=head3 Examples
		2471
		2472	Example: Watch C</etc/passwd> for attribute changes.
		2473
		2474	static void
		2475	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2476	{
		2477	/* /etc/passwd changed in some way */
		2478	if (w->attr.st_nlink)
		2479	{
		2480	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2481	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2482	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2483	}
		2484	else
		2485	/* you shalt not abuse printf for puts */
		2486	puts ("wow, /etc/passwd is not there, expect problems. "
		2487	"if this is windows, they already arrived\n");
		2488	}
		2489
		2490	...
		2491	ev_stat passwd;
		2492
		2493	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2494	ev_stat_start (loop, &passwd);
		2495
		2496	Example: Like above, but additionally use a one-second delay so we do not
		2497	miss updates (however, frequent updates will delay processing, too, so
		2498	one might do the work both on C<ev_stat> callback invocation I<and> on
		2499	C<ev_timer> callback invocation).
		2500
		2501	static ev_stat passwd;
		2502	static ev_timer timer;
		2503
		2504	static void
		2505	timer_cb (EV_P_ ev_timer *w, int revents)
		2506	{
		2507	ev_timer_stop (EV_A_ w);
		2508
		2509	/* now it's one second after the most recent passwd change */
		2510	}
		2511
		2512	static void
		2513	stat_cb (EV_P_ ev_stat *w, int revents)
		2514	{
		2515	/* reset the one-second timer */
		2516	ev_timer_again (EV_A_ &timer);
		2517	}
		2518
		2519	...
		2520	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2521	ev_stat_start (loop, &passwd);
		2522	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2523
		2524
644	=head2 C<ev_idle> - when you've got nothing better to do	2525	=head2 C<ev_idle> - when you've got nothing better to do...
645		2526
646	Idle watchers trigger events when there are no other events are pending	2527	Idle watchers trigger events when no other events of the same or higher
647	(prepare, check and other idle watchers do not count). That is, as long	2528	priority are pending (prepare, check and other idle watchers do not count
648	as your process is busy handling sockets or timeouts (or even signals,	2529	as receiving "events").
649	imagine) it will not be triggered. But when your process is idle all idle	2530
650	watchers are being called again and again, once per event loop iteration -	2531	That is, as long as your process is busy handling sockets or timeouts
		2532	(or even signals, imagine) of the same or higher priority it will not be
		2533	triggered. But when your process is idle (or only lower-priority watchers
		2534	are pending), the idle watchers are being called once per event loop
651	until stopped, that is, or your process receives more events and becomes	2535	iteration - until stopped, that is, or your process receives more events
652	busy.	2536	and becomes busy again with higher priority stuff.
653		2537
654	The most noteworthy effect is that as long as any idle watchers are	2538	The most noteworthy effect is that as long as any idle watchers are
655	active, the process will not block when waiting for new events.	2539	active, the process will not block when waiting for new events.
656		2540
657	Apart from keeping your process non-blocking (which is a useful	2541	Apart from keeping your process non-blocking (which is a useful
658	effect on its own sometimes), idle watchers are a good place to do	2542	effect on its own sometimes), idle watchers are a good place to do
659	"pseudo-background processing", or delay processing stuff to after the	2543	"pseudo-background processing", or delay processing stuff to after the
660	event loop has handled all outstanding events.	2544	event loop has handled all outstanding events.
661		2545
		2546	=head3 Watcher-Specific Functions and Data Members
		2547
662	=over 4	2548	=over 4
663		2549
664	=item ev_idle_init (ev_signal *, callback)	2550	=item ev_idle_init (ev_idle *, callback)
665		2551
666	Initialises and configures the idle watcher - it has no parameters of any	2552	Initialises and configures the idle watcher - it has no parameters of any
667	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2553	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
668	believe me.	2554	believe me.
669		2555
670	=back	2556	=back
671		2557
		2558	=head3 Examples
		2559
		2560	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2561	callback, free it. Also, use no error checking, as usual.
		2562
		2563	static void
		2564	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2565	{
		2566	free (w);
		2567	// now do something you wanted to do when the program has
		2568	// no longer anything immediate to do.
		2569	}
		2570
		2571	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2572	ev_idle_init (idle_watcher, idle_cb);
		2573	ev_idle_start (loop, idle_watcher);
		2574
		2575
672	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2576	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
673		2577
674	Prepare and check watchers are usually (but not always) used in tandem:	2578	Prepare and check watchers are usually (but not always) used in pairs:
675	Prepare watchers get invoked before the process blocks and check watchers	2579	prepare watchers get invoked before the process blocks and check watchers
676	afterwards.	2580	afterwards.
677		2581
		2582	You I<must not> call C<ev_loop> or similar functions that enter
		2583	the current event loop from either C<ev_prepare> or C<ev_check>
		2584	watchers. Other loops than the current one are fine, however. The
		2585	rationale behind this is that you do not need to check for recursion in
		2586	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2587	C<ev_check> so if you have one watcher of each kind they will always be
		2588	called in pairs bracketing the blocking call.
		2589
678	Their main purpose is to integrate other event mechanisms into libev. This	2590	Their main purpose is to integrate other event mechanisms into libev and
679	could be used, for example, to track variable changes, implement your own	2591	their use is somewhat advanced. They could be used, for example, to track
680	watchers, integrate net-snmp or a coroutine library and lots more.	2592	variable changes, implement your own watchers, integrate net-snmp or a
		2593	coroutine library and lots more. They are also occasionally useful if
		2594	you cache some data and want to flush it before blocking (for example,
		2595	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2596	watcher).
681		2597
682	This is done by examining in each prepare call which file descriptors need	2598	This is done by examining in each prepare call which file descriptors
683	to be watched by the other library, registering C<ev_io> watchers for	2599	need to be watched by the other library, registering C<ev_io> watchers
684	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2600	for them and starting an C<ev_timer> watcher for any timeouts (many
685	provide just this functionality). Then, in the check watcher you check for	2601	libraries provide exactly this functionality). Then, in the check watcher,
686	any events that occured (by checking the pending status of all watchers	2602	you check for any events that occurred (by checking the pending status
687	and stopping them) and call back into the library. The I/O and timer	2603	of all watchers and stopping them) and call back into the library. The
688	callbacks will never actually be called (but must be valid neverthelles,	2604	I/O and timer callbacks will never actually be called (but must be valid
689	because you never know, you know?).	2605	nevertheless, because you never know, you know?).
690		2606
691	As another example, the Perl Coro module uses these hooks to integrate	2607	As another example, the Perl Coro module uses these hooks to integrate
692	coroutines into libev programs, by yielding to other active coroutines	2608	coroutines into libev programs, by yielding to other active coroutines
693	during each prepare and only letting the process block if no coroutines	2609	during each prepare and only letting the process block if no coroutines
694	are ready to run (its actually more complicated, it only runs coroutines	2610	are ready to run (it's actually more complicated: it only runs coroutines
695	with priority higher than the event loop and one lower priority once,	2611	with priority higher than or equal to the event loop and one coroutine
696	using idle watchers to keep the event loop from blocking if lower-priority	2612	of lower priority, but only once, using idle watchers to keep the event
697	coroutines exist, thus mapping low-priority coroutines to idle/background	2613	loop from blocking if lower-priority coroutines are active, thus mapping
698	tasks).	2614	low-priority coroutines to idle/background tasks).
		2615
		2616	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2617	priority, to ensure that they are being run before any other watchers
		2618	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2619
		2620	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2621	activate ("feed") events into libev. While libev fully supports this, they
		2622	might get executed before other C<ev_check> watchers did their job. As
		2623	C<ev_check> watchers are often used to embed other (non-libev) event
		2624	loops those other event loops might be in an unusable state until their
		2625	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2626	others).
		2627
		2628	=head3 Watcher-Specific Functions and Data Members
699		2629
700	=over 4	2630	=over 4
701		2631
702	=item ev_prepare_init (ev_prepare *, callback)	2632	=item ev_prepare_init (ev_prepare *, callback)
703		2633
704	=item ev_check_init (ev_check *, callback)	2634	=item ev_check_init (ev_check *, callback)
705		2635
706	Initialises and configures the prepare or check watcher - they have no	2636	Initialises and configures the prepare or check watcher - they have no
707	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2637	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
708	macros, but using them is utterly, utterly and completely pointless.	2638	macros, but using them is utterly, utterly, utterly and completely
		2639	pointless.
709		2640
710	=back	2641	=back
711		2642
		2643	=head3 Examples
		2644
		2645	There are a number of principal ways to embed other event loops or modules
		2646	into libev. Here are some ideas on how to include libadns into libev
		2647	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2648	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2649	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2650	Glib event loop).
		2651
		2652	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2653	and in a check watcher, destroy them and call into libadns. What follows
		2654	is pseudo-code only of course. This requires you to either use a low
		2655	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2656	the callbacks for the IO/timeout watchers might not have been called yet.
		2657
		2658	static ev_io iow [nfd];
		2659	static ev_timer tw;
		2660
		2661	static void
		2662	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2663	{
		2664	}
		2665
		2666	// create io watchers for each fd and a timer before blocking
		2667	static void
		2668	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2669	{
		2670	int timeout = 3600000;
		2671	struct pollfd fds [nfd];
		2672	// actual code will need to loop here and realloc etc.
		2673	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2674
		2675	/* the callback is illegal, but won't be called as we stop during check */
		2676	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2677	ev_timer_start (loop, &tw);
		2678
		2679	// create one ev_io per pollfd
		2680	for (int i = 0; i < nfd; ++i)
		2681	{
		2682	ev_io_init (iow + i, io_cb, fds [i].fd,
		2683	((fds [i].events & POLLIN ? EV_READ : 0)
		2684	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2685
		2686	fds [i].revents = 0;
		2687	ev_io_start (loop, iow + i);
		2688	}
		2689	}
		2690
		2691	// stop all watchers after blocking
		2692	static void
		2693	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2694	{
		2695	ev_timer_stop (loop, &tw);
		2696
		2697	for (int i = 0; i < nfd; ++i)
		2698	{
		2699	// set the relevant poll flags
		2700	// could also call adns_processreadable etc. here
		2701	struct pollfd *fd = fds + i;
		2702	int revents = ev_clear_pending (iow + i);
		2703	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2704	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2705
		2706	// now stop the watcher
		2707	ev_io_stop (loop, iow + i);
		2708	}
		2709
		2710	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2711	}
		2712
		2713	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2714	in the prepare watcher and would dispose of the check watcher.
		2715
		2716	Method 3: If the module to be embedded supports explicit event
		2717	notification (libadns does), you can also make use of the actual watcher
		2718	callbacks, and only destroy/create the watchers in the prepare watcher.
		2719
		2720	static void
		2721	timer_cb (EV_P_ ev_timer *w, int revents)
		2722	{
		2723	adns_state ads = (adns_state)w->data;
		2724	update_now (EV_A);
		2725
		2726	adns_processtimeouts (ads, &tv_now);
		2727	}
		2728
		2729	static void
		2730	io_cb (EV_P_ ev_io *w, int revents)
		2731	{
		2732	adns_state ads = (adns_state)w->data;
		2733	update_now (EV_A);
		2734
		2735	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2736	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2737	}
		2738
		2739	// do not ever call adns_afterpoll
		2740
		2741	Method 4: Do not use a prepare or check watcher because the module you
		2742	want to embed is not flexible enough to support it. Instead, you can
		2743	override their poll function. The drawback with this solution is that the
		2744	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2745	this approach, effectively embedding EV as a client into the horrible
		2746	libglib event loop.
		2747
		2748	static gint
		2749	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2750	{
		2751	int got_events = 0;
		2752
		2753	for (n = 0; n < nfds; ++n)
		2754	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2755
		2756	if (timeout >= 0)
		2757	// create/start timer
		2758
		2759	// poll
		2760	ev_loop (EV_A_ 0);
		2761
		2762	// stop timer again
		2763	if (timeout >= 0)
		2764	ev_timer_stop (EV_A_ &to);
		2765
		2766	// stop io watchers again - their callbacks should have set
		2767	for (n = 0; n < nfds; ++n)
		2768	ev_io_stop (EV_A_ iow [n]);
		2769
		2770	return got_events;
		2771	}
		2772
		2773
		2774	=head2 C<ev_embed> - when one backend isn't enough...
		2775
		2776	This is a rather advanced watcher type that lets you embed one event loop
		2777	into another (currently only C<ev_io> events are supported in the embedded
		2778	loop, other types of watchers might be handled in a delayed or incorrect
		2779	fashion and must not be used).
		2780
		2781	There are primarily two reasons you would want that: work around bugs and
		2782	prioritise I/O.
		2783
		2784	As an example for a bug workaround, the kqueue backend might only support
		2785	sockets on some platform, so it is unusable as generic backend, but you
		2786	still want to make use of it because you have many sockets and it scales
		2787	so nicely. In this case, you would create a kqueue-based loop and embed
		2788	it into your default loop (which might use e.g. poll). Overall operation
		2789	will be a bit slower because first libev has to call C<poll> and then
		2790	C<kevent>, but at least you can use both mechanisms for what they are
		2791	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2792
		2793	As for prioritising I/O: under rare circumstances you have the case where
		2794	some fds have to be watched and handled very quickly (with low latency),
		2795	and even priorities and idle watchers might have too much overhead. In
		2796	this case you would put all the high priority stuff in one loop and all
		2797	the rest in a second one, and embed the second one in the first.
		2798
		2799	As long as the watcher is active, the callback will be invoked every
		2800	time there might be events pending in the embedded loop. The callback
		2801	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		2802	sweep and invoke their callbacks (the callback doesn't need to invoke the
		2803	C<ev_embed_sweep> function directly, it could also start an idle watcher
		2804	to give the embedded loop strictly lower priority for example).
		2805
		2806	You can also set the callback to C<0>, in which case the embed watcher
		2807	will automatically execute the embedded loop sweep whenever necessary.
		2808
		2809	Fork detection will be handled transparently while the C<ev_embed> watcher
		2810	is active, i.e., the embedded loop will automatically be forked when the
		2811	embedding loop forks. In other cases, the user is responsible for calling
		2812	C<ev_loop_fork> on the embedded loop.
		2813
		2814	Unfortunately, not all backends are embeddable: only the ones returned by
		2815	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2816	portable one.
		2817
		2818	So when you want to use this feature you will always have to be prepared
		2819	that you cannot get an embeddable loop. The recommended way to get around
		2820	this is to have a separate variables for your embeddable loop, try to
		2821	create it, and if that fails, use the normal loop for everything.
		2822
		2823	=head3 C<ev_embed> and fork
		2824
		2825	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2826	automatically be applied to the embedded loop as well, so no special
		2827	fork handling is required in that case. When the watcher is not running,
		2828	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2829	as applicable.
		2830
		2831	=head3 Watcher-Specific Functions and Data Members
		2832
		2833	=over 4
		2834
		2835	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2836
		2837	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2838
		2839	Configures the watcher to embed the given loop, which must be
		2840	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2841	invoked automatically, otherwise it is the responsibility of the callback
		2842	to invoke it (it will continue to be called until the sweep has been done,
		2843	if you do not want that, you need to temporarily stop the embed watcher).
		2844
		2845	=item ev_embed_sweep (loop, ev_embed *)
		2846
		2847	Make a single, non-blocking sweep over the embedded loop. This works
		2848	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		2849	appropriate way for embedded loops.
		2850
		2851	=item struct ev_loop *other [read-only]
		2852
		2853	The embedded event loop.
		2854
		2855	=back
		2856
		2857	=head3 Examples
		2858
		2859	Example: Try to get an embeddable event loop and embed it into the default
		2860	event loop. If that is not possible, use the default loop. The default
		2861	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2862	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2863	used).
		2864
		2865	struct ev_loop *loop_hi = ev_default_init (0);
		2866	struct ev_loop *loop_lo = 0;
		2867	ev_embed embed;
		2868
		2869	// see if there is a chance of getting one that works
		2870	// (remember that a flags value of 0 means autodetection)
		2871	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2872	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2873	: 0;
		2874
		2875	// if we got one, then embed it, otherwise default to loop_hi
		2876	if (loop_lo)
		2877	{
		2878	ev_embed_init (&embed, 0, loop_lo);
		2879	ev_embed_start (loop_hi, &embed);
		2880	}
		2881	else
		2882	loop_lo = loop_hi;
		2883
		2884	Example: Check if kqueue is available but not recommended and create
		2885	a kqueue backend for use with sockets (which usually work with any
		2886	kqueue implementation). Store the kqueue/socket-only event loop in
		2887	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2888
		2889	struct ev_loop *loop = ev_default_init (0);
		2890	struct ev_loop *loop_socket = 0;
		2891	ev_embed embed;
		2892
		2893	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2894	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2895	{
		2896	ev_embed_init (&embed, 0, loop_socket);
		2897	ev_embed_start (loop, &embed);
		2898	}
		2899
		2900	if (!loop_socket)
		2901	loop_socket = loop;
		2902
		2903	// now use loop_socket for all sockets, and loop for everything else
		2904
		2905
		2906	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		2907
		2908	Fork watchers are called when a C<fork ()> was detected (usually because
		2909	whoever is a good citizen cared to tell libev about it by calling
		2910	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		2911	event loop blocks next and before C<ev_check> watchers are being called,
		2912	and only in the child after the fork. If whoever good citizen calling
		2913	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		2914	handlers will be invoked, too, of course.
		2915
		2916	=head3 The special problem of life after fork - how is it possible?
		2917
		2918	Most uses of C<fork()> consist of forking, then some simple calls to ste
		2919	up/change the process environment, followed by a call to C<exec()>. This
		2920	sequence should be handled by libev without any problems.
		2921
		2922	This changes when the application actually wants to do event handling
		2923	in the child, or both parent in child, in effect "continuing" after the
		2924	fork.
		2925
		2926	The default mode of operation (for libev, with application help to detect
		2927	forks) is to duplicate all the state in the child, as would be expected
		2928	when I<either> the parent I<or> the child process continues.
		2929
		2930	When both processes want to continue using libev, then this is usually the
		2931	wrong result. In that case, usually one process (typically the parent) is
		2932	supposed to continue with all watchers in place as before, while the other
		2933	process typically wants to start fresh, i.e. without any active watchers.
		2934
		2935	The cleanest and most efficient way to achieve that with libev is to
		2936	simply create a new event loop, which of course will be "empty", and
		2937	use that for new watchers. This has the advantage of not touching more
		2938	memory than necessary, and thus avoiding the copy-on-write, and the
		2939	disadvantage of having to use multiple event loops (which do not support
		2940	signal watchers).
		2941
		2942	When this is not possible, or you want to use the default loop for
		2943	other reasons, then in the process that wants to start "fresh", call
		2944	C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying
		2945	the default loop will "orphan" (not stop) all registered watchers, so you
		2946	have to be careful not to execute code that modifies those watchers. Note
		2947	also that in that case, you have to re-register any signal watchers.
		2948
		2949	=head3 Watcher-Specific Functions and Data Members
		2950
		2951	=over 4
		2952
		2953	=item ev_fork_init (ev_signal *, callback)
		2954
		2955	Initialises and configures the fork watcher - it has no parameters of any
		2956	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		2957	believe me.
		2958
		2959	=back
		2960
		2961
		2962	=head2 C<ev_async> - how to wake up another event loop
		2963
		2964	In general, you cannot use an C<ev_loop> from multiple threads or other
		2965	asynchronous sources such as signal handlers (as opposed to multiple event
		2966	loops - those are of course safe to use in different threads).
		2967
		2968	Sometimes, however, you need to wake up another event loop you do not
		2969	control, for example because it belongs to another thread. This is what
		2970	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2971	can signal it by calling C<ev_async_send>, which is thread- and signal
		2972	safe.
		2973
		2974	This functionality is very similar to C<ev_signal> watchers, as signals,
		2975	too, are asynchronous in nature, and signals, too, will be compressed
		2976	(i.e. the number of callback invocations may be less than the number of
		2977	C<ev_async_sent> calls).
		2978
		2979	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2980	just the default loop.
		2981
		2982	=head3 Queueing
		2983
		2984	C<ev_async> does not support queueing of data in any way. The reason
		2985	is that the author does not know of a simple (or any) algorithm for a
		2986	multiple-writer-single-reader queue that works in all cases and doesn't
		2987	need elaborate support such as pthreads or unportable memory access
		2988	semantics.
		2989
		2990	That means that if you want to queue data, you have to provide your own
		2991	queue. But at least I can tell you how to implement locking around your
		2992	queue:
		2993
		2994	=over 4
		2995
		2996	=item queueing from a signal handler context
		2997
		2998	To implement race-free queueing, you simply add to the queue in the signal
		2999	handler but you block the signal handler in the watcher callback. Here is
		3000	an example that does that for some fictitious SIGUSR1 handler:
		3001
		3002	static ev_async mysig;
		3003
		3004	static void
		3005	sigusr1_handler (void)
		3006	{
		3007	sometype data;
		3008
		3009	// no locking etc.
		3010	queue_put (data);
		3011	ev_async_send (EV_DEFAULT_ &mysig);
		3012	}
		3013
		3014	static void
		3015	mysig_cb (EV_P_ ev_async *w, int revents)
		3016	{
		3017	sometype data;
		3018	sigset_t block, prev;
		3019
		3020	sigemptyset (&block);
		3021	sigaddset (&block, SIGUSR1);
		3022	sigprocmask (SIG_BLOCK, &block, &prev);
		3023
		3024	while (queue_get (&data))
		3025	process (data);
		3026
		3027	if (sigismember (&prev, SIGUSR1)
		3028	sigprocmask (SIG_UNBLOCK, &block, 0);
		3029	}
		3030
		3031	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3032	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3033	either...).
		3034
		3035	=item queueing from a thread context
		3036
		3037	The strategy for threads is different, as you cannot (easily) block
		3038	threads but you can easily preempt them, so to queue safely you need to
		3039	employ a traditional mutex lock, such as in this pthread example:
		3040
		3041	static ev_async mysig;
		3042	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3043
		3044	static void
		3045	otherthread (void)
		3046	{
		3047	// only need to lock the actual queueing operation
		3048	pthread_mutex_lock (&mymutex);
		3049	queue_put (data);
		3050	pthread_mutex_unlock (&mymutex);
		3051
		3052	ev_async_send (EV_DEFAULT_ &mysig);
		3053	}
		3054
		3055	static void
		3056	mysig_cb (EV_P_ ev_async *w, int revents)
		3057	{
		3058	pthread_mutex_lock (&mymutex);
		3059
		3060	while (queue_get (&data))
		3061	process (data);
		3062
		3063	pthread_mutex_unlock (&mymutex);
		3064	}
		3065
		3066	=back
		3067
		3068
		3069	=head3 Watcher-Specific Functions and Data Members
		3070
		3071	=over 4
		3072
		3073	=item ev_async_init (ev_async *, callback)
		3074
		3075	Initialises and configures the async watcher - it has no parameters of any
		3076	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3077	trust me.
		3078
		3079	=item ev_async_send (loop, ev_async *)
		3080
		3081	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3082	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		3083	C<ev_feed_event>, this call is safe to do from other threads, signal or
		3084	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		3085	section below on what exactly this means).
		3086
		3087	Note that, as with other watchers in libev, multiple events might get
		3088	compressed into a single callback invocation (another way to look at this
		3089	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3090	reset when the event loop detects that).
		3091
		3092	This call incurs the overhead of a system call only once per event loop
		3093	iteration, so while the overhead might be noticeable, it doesn't apply to
		3094	repeated calls to C<ev_async_send> for the same event loop.
		3095
		3096	=item bool = ev_async_pending (ev_async *)
		3097
		3098	Returns a non-zero value when C<ev_async_send> has been called on the
		3099	watcher but the event has not yet been processed (or even noted) by the
		3100	event loop.
		3101
		3102	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3103	the loop iterates next and checks for the watcher to have become active,
		3104	it will reset the flag again. C<ev_async_pending> can be used to very
		3105	quickly check whether invoking the loop might be a good idea.
		3106
		3107	Not that this does I<not> check whether the watcher itself is pending,
		3108	only whether it has been requested to make this watcher pending: there
		3109	is a time window between the event loop checking and resetting the async
		3110	notification, and the callback being invoked.
		3111
		3112	=back
		3113
		3114
712	=head1 OTHER FUNCTIONS	3115	=head1 OTHER FUNCTIONS
713		3116
714	There are some other functions of possible interest. Described. Here. Now.	3117	There are some other functions of possible interest. Described. Here. Now.
715		3118
716	=over 4	3119	=over 4
717		3120
718	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3121	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
719		3122
720	This function combines a simple timer and an I/O watcher, calls your	3123	This function combines a simple timer and an I/O watcher, calls your
721	callback on whichever event happens first and automatically stop both	3124	callback on whichever event happens first and automatically stops both
722	watchers. This is useful if you want to wait for a single event on an fd	3125	watchers. This is useful if you want to wait for a single event on an fd
723	or timeout without havign to allocate/configure/start/stop/free one or	3126	or timeout without having to allocate/configure/start/stop/free one or
724	more watchers yourself.	3127	more watchers yourself.
725		3128
726	If C<fd> is less than 0, then no I/O watcher will be started and events	3129	If C<fd> is less than 0, then no I/O watcher will be started and the
727	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3130	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
728	C<events> set will be craeted and started.	3131	the given C<fd> and C<events> set will be created and started.
729		3132
730	If C<timeout> is less than 0, then no timeout watcher will be	3133	If C<timeout> is less than 0, then no timeout watcher will be
731	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3134	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
732	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3135	repeat = 0) will be started. C<0> is a valid timeout.
733	dubious value.
734		3136
735	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3137	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
736	passed an events set like normal event callbacks (with a combination of	3138	passed an C<revents> set like normal event callbacks (a combination of
737	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3139	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
738	value passed to C<ev_once>:	3140	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3141	a timeout and an io event at the same time - you probably should give io
		3142	events precedence.
739		3143
		3144	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3145
740	static void stdin_ready (int revents, void *arg)	3146	static void stdin_ready (int revents, void *arg)
741	{	3147	{
742	if (revents & EV_TIMEOUT)
743	/* doh, nothing entered */;
744	else if (revents & EV_READ)	3148	if (revents & EV_READ)
745	/* stdin might have data for us, joy! */;	3149	/* stdin might have data for us, joy! */;
		3150	else if (revents & EV_TIMEOUT)
		3151	/* doh, nothing entered */;
746	}	3152	}
747		3153
748	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3154	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
749
750	=item ev_feed_event (loop, watcher, int events)
751
752	Feeds the given event set into the event loop, as if the specified event
753	had happened for the specified watcher (which must be a pointer to an
754	initialised but not necessarily started event watcher).
755		3155
756	=item ev_feed_fd_event (loop, int fd, int revents)	3156	=item ev_feed_fd_event (loop, int fd, int revents)
757		3157
758	Feed an event on the given fd, as if a file descriptor backend detected	3158	Feed an event on the given fd, as if a file descriptor backend detected
759	the given events it.	3159	the given events it.
760		3160
761	=item ev_feed_signal_event (loop, int signum)	3161	=item ev_feed_signal_event (loop, int signum)
762		3162
763	Feed an event as if the given signal occured (loop must be the default loop!).	3163	Feed an event as if the given signal occurred (C<loop> must be the default
		3164	loop!).
764		3165
765	=back	3166	=back
766		3167
		3168
		3169	=head1 LIBEVENT EMULATION
		3170
		3171	Libev offers a compatibility emulation layer for libevent. It cannot
		3172	emulate the internals of libevent, so here are some usage hints:
		3173
		3174	=over 4
		3175
		3176	=item * Use it by including <event.h>, as usual.
		3177
		3178	=item * The following members are fully supported: ev_base, ev_callback,
		3179	ev_arg, ev_fd, ev_res, ev_events.
		3180
		3181	=item * Avoid using ev_flags and the EVLIST_*-macros, while it is
		3182	maintained by libev, it does not work exactly the same way as in libevent (consider
		3183	it a private API).
		3184
		3185	=item * Priorities are not currently supported. Initialising priorities
		3186	will fail and all watchers will have the same priority, even though there
		3187	is an ev_pri field.
		3188
		3189	=item * In libevent, the last base created gets the signals, in libev, the
		3190	first base created (== the default loop) gets the signals.
		3191
		3192	=item * Other members are not supported.
		3193
		3194	=item * The libev emulation is I<not> ABI compatible to libevent, you need
		3195	to use the libev header file and library.
		3196
		3197	=back
		3198
		3199	=head1 C++ SUPPORT
		3200
		3201	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3202	you to use some convenience methods to start/stop watchers and also change
		3203	the callback model to a model using method callbacks on objects.
		3204
		3205	To use it,
		3206
		3207	#include <ev++.h>
		3208
		3209	This automatically includes F<ev.h> and puts all of its definitions (many
		3210	of them macros) into the global namespace. All C++ specific things are
		3211	put into the C<ev> namespace. It should support all the same embedding
		3212	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3213
		3214	Care has been taken to keep the overhead low. The only data member the C++
		3215	classes add (compared to plain C-style watchers) is the event loop pointer
		3216	that the watcher is associated with (or no additional members at all if
		3217	you disable C<EV_MULTIPLICITY> when embedding libev).
		3218
		3219	Currently, functions, and static and non-static member functions can be
		3220	used as callbacks. Other types should be easy to add as long as they only
		3221	need one additional pointer for context. If you need support for other
		3222	types of functors please contact the author (preferably after implementing
		3223	it).
		3224
		3225	Here is a list of things available in the C<ev> namespace:
		3226
		3227	=over 4
		3228
		3229	=item C<ev::READ>, C<ev::WRITE> etc.
		3230
		3231	These are just enum values with the same values as the C<EV_READ> etc.
		3232	macros from F<ev.h>.
		3233
		3234	=item C<ev::tstamp>, C<ev::now>
		3235
		3236	Aliases to the same types/functions as with the C<ev_> prefix.
		3237
		3238	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3239
		3240	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3241	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3242	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3243	defines by many implementations.
		3244
		3245	All of those classes have these methods:
		3246
		3247	=over 4
		3248
		3249	=item ev::TYPE::TYPE ()
		3250
		3251	=item ev::TYPE::TYPE (loop)
		3252
		3253	=item ev::TYPE::~TYPE
		3254
		3255	The constructor (optionally) takes an event loop to associate the watcher
		3256	with. If it is omitted, it will use C<EV_DEFAULT>.
		3257
		3258	The constructor calls C<ev_init> for you, which means you have to call the
		3259	C<set> method before starting it.
		3260
		3261	It will not set a callback, however: You have to call the templated C<set>
		3262	method to set a callback before you can start the watcher.
		3263
		3264	(The reason why you have to use a method is a limitation in C++ which does
		3265	not allow explicit template arguments for constructors).
		3266
		3267	The destructor automatically stops the watcher if it is active.
		3268
		3269	=item w->set<class, &class::method> (object *)
		3270
		3271	This method sets the callback method to call. The method has to have a
		3272	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3273	first argument and the C<revents> as second. The object must be given as
		3274	parameter and is stored in the C<data> member of the watcher.
		3275
		3276	This method synthesizes efficient thunking code to call your method from
		3277	the C callback that libev requires. If your compiler can inline your
		3278	callback (i.e. it is visible to it at the place of the C<set> call and
		3279	your compiler is good :), then the method will be fully inlined into the
		3280	thunking function, making it as fast as a direct C callback.
		3281
		3282	Example: simple class declaration and watcher initialisation
		3283
		3284	struct myclass
		3285	{
		3286	void io_cb (ev::io &w, int revents) { }
		3287	}
		3288
		3289	myclass obj;
		3290	ev::io iow;
		3291	iow.set <myclass, &myclass::io_cb> (&obj);
		3292
		3293	=item w->set (object *)
		3294
		3295	This is an B<experimental> feature that might go away in a future version.
		3296
		3297	This is a variation of a method callback - leaving out the method to call
		3298	will default the method to C<operator ()>, which makes it possible to use
		3299	functor objects without having to manually specify the C<operator ()> all
		3300	the time. Incidentally, you can then also leave out the template argument
		3301	list.
		3302
		3303	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3304	int revents)>.
		3305
		3306	See the method-C<set> above for more details.
		3307
		3308	Example: use a functor object as callback.
		3309
		3310	struct myfunctor
		3311	{
		3312	void operator() (ev::io &w, int revents)
		3313	{
		3314	...
		3315	}
		3316	}
		3317
		3318	myfunctor f;
		3319
		3320	ev::io w;
		3321	w.set (&f);
		3322
		3323	=item w->set<function> (void *data = 0)
		3324
		3325	Also sets a callback, but uses a static method or plain function as
		3326	callback. The optional C<data> argument will be stored in the watcher's
		3327	C<data> member and is free for you to use.
		3328
		3329	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3330
		3331	See the method-C<set> above for more details.
		3332
		3333	Example: Use a plain function as callback.
		3334
		3335	static void io_cb (ev::io &w, int revents) { }
		3336	iow.set <io_cb> ();
		3337
		3338	=item w->set (loop)
		3339
		3340	Associates a different C<struct ev_loop> with this watcher. You can only
		3341	do this when the watcher is inactive (and not pending either).
		3342
		3343	=item w->set ([arguments])
		3344
		3345	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
		3346	called at least once. Unlike the C counterpart, an active watcher gets
		3347	automatically stopped and restarted when reconfiguring it with this
		3348	method.
		3349
		3350	=item w->start ()
		3351
		3352	Starts the watcher. Note that there is no C<loop> argument, as the
		3353	constructor already stores the event loop.
		3354
		3355	=item w->stop ()
		3356
		3357	Stops the watcher if it is active. Again, no C<loop> argument.
		3358
		3359	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3360
		3361	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3362	C<ev_TYPE_again> function.
		3363
		3364	=item w->sweep () (C<ev::embed> only)
		3365
		3366	Invokes C<ev_embed_sweep>.
		3367
		3368	=item w->update () (C<ev::stat> only)
		3369
		3370	Invokes C<ev_stat_stat>.
		3371
		3372	=back
		3373
		3374	=back
		3375
		3376	Example: Define a class with an IO and idle watcher, start one of them in
		3377	the constructor.
		3378
		3379	class myclass
		3380	{
		3381	ev::io io ; void io_cb (ev::io &w, int revents);
		3382	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3383
		3384	myclass (int fd)
		3385	{
		3386	io .set <myclass, &myclass::io_cb > (this);
		3387	idle.set <myclass, &myclass::idle_cb> (this);
		3388
		3389	io.start (fd, ev::READ);
		3390	}
		3391	};
		3392
		3393
		3394	=head1 OTHER LANGUAGE BINDINGS
		3395
		3396	Libev does not offer other language bindings itself, but bindings for a
		3397	number of languages exist in the form of third-party packages. If you know
		3398	any interesting language binding in addition to the ones listed here, drop
		3399	me a note.
		3400
		3401	=over 4
		3402
		3403	=item Perl
		3404
		3405	The EV module implements the full libev API and is actually used to test
		3406	libev. EV is developed together with libev. Apart from the EV core module,
		3407	there are additional modules that implement libev-compatible interfaces
		3408	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3409	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3410	and C<EV::Glib>).
		3411
		3412	It can be found and installed via CPAN, its homepage is at
		3413	L<http://software.schmorp.de/pkg/EV>.
		3414
		3415	=item Python
		3416
		3417	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3418	seems to be quite complete and well-documented.
		3419
		3420	=item Ruby
		3421
		3422	Tony Arcieri has written a ruby extension that offers access to a subset
		3423	of the libev API and adds file handle abstractions, asynchronous DNS and
		3424	more on top of it. It can be found via gem servers. Its homepage is at
		3425	L<http://rev.rubyforge.org/>.
		3426
		3427	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3428	makes rev work even on mingw.
		3429
		3430	=item Haskell
		3431
		3432	A haskell binding to libev is available at
		3433	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3434
		3435	=item D
		3436
		3437	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3438	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3439
		3440	=item Ocaml
		3441
		3442	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3443	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3444
		3445	=item Lua
		3446
		3447	Brian Maher has written a partial interface to libev
		3448	for lua (only C<ev_io> and C<ev_timer>), to be found at
		3449	L<http://github.com/brimworks/lua-ev>.
		3450
		3451	=back
		3452
		3453
		3454	=head1 MACRO MAGIC
		3455
		3456	Libev can be compiled with a variety of options, the most fundamental
		3457	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3458	functions and callbacks have an initial C<struct ev_loop *> argument.
		3459
		3460	To make it easier to write programs that cope with either variant, the
		3461	following macros are defined:
		3462
		3463	=over 4
		3464
		3465	=item C<EV_A>, C<EV_A_>
		3466
		3467	This provides the loop I<argument> for functions, if one is required ("ev
		3468	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3469	C<EV_A_> is used when other arguments are following. Example:
		3470
		3471	ev_unref (EV_A);
		3472	ev_timer_add (EV_A_ watcher);
		3473	ev_loop (EV_A_ 0);
		3474
		3475	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3476	which is often provided by the following macro.
		3477
		3478	=item C<EV_P>, C<EV_P_>
		3479
		3480	This provides the loop I<parameter> for functions, if one is required ("ev
		3481	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3482	C<EV_P_> is used when other parameters are following. Example:
		3483
		3484	// this is how ev_unref is being declared
		3485	static void ev_unref (EV_P);
		3486
		3487	// this is how you can declare your typical callback
		3488	static void cb (EV_P_ ev_timer *w, int revents)
		3489
		3490	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3491	suitable for use with C<EV_A>.
		3492
		3493	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3494
		3495	Similar to the other two macros, this gives you the value of the default
		3496	loop, if multiple loops are supported ("ev loop default").
		3497
		3498	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3499
		3500	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3501	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3502	is undefined when the default loop has not been initialised by a previous
		3503	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3504
		3505	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3506	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3507
		3508	=back
		3509
		3510	Example: Declare and initialise a check watcher, utilising the above
		3511	macros so it will work regardless of whether multiple loops are supported
		3512	or not.
		3513
		3514	static void
		3515	check_cb (EV_P_ ev_timer *w, int revents)
		3516	{
		3517	ev_check_stop (EV_A_ w);
		3518	}
		3519
		3520	ev_check check;
		3521	ev_check_init (&check, check_cb);
		3522	ev_check_start (EV_DEFAULT_ &check);
		3523	ev_loop (EV_DEFAULT_ 0);
		3524
		3525	=head1 EMBEDDING
		3526
		3527	Libev can (and often is) directly embedded into host
		3528	applications. Examples of applications that embed it include the Deliantra
		3529	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3530	and rxvt-unicode.
		3531
		3532	The goal is to enable you to just copy the necessary files into your
		3533	source directory without having to change even a single line in them, so
		3534	you can easily upgrade by simply copying (or having a checked-out copy of
		3535	libev somewhere in your source tree).
		3536
		3537	=head2 FILESETS
		3538
		3539	Depending on what features you need you need to include one or more sets of files
		3540	in your application.
		3541
		3542	=head3 CORE EVENT LOOP
		3543
		3544	To include only the libev core (all the C<ev_*> functions), with manual
		3545	configuration (no autoconf):
		3546
		3547	#define EV_STANDALONE 1
		3548	#include "ev.c"
		3549
		3550	This will automatically include F<ev.h>, too, and should be done in a
		3551	single C source file only to provide the function implementations. To use
		3552	it, do the same for F<ev.h> in all files wishing to use this API (best
		3553	done by writing a wrapper around F<ev.h> that you can include instead and
		3554	where you can put other configuration options):
		3555
		3556	#define EV_STANDALONE 1
		3557	#include "ev.h"
		3558
		3559	Both header files and implementation files can be compiled with a C++
		3560	compiler (at least, that's a stated goal, and breakage will be treated
		3561	as a bug).
		3562
		3563	You need the following files in your source tree, or in a directory
		3564	in your include path (e.g. in libev/ when using -Ilibev):
		3565
		3566	ev.h
		3567	ev.c
		3568	ev_vars.h
		3569	ev_wrap.h
		3570
		3571	ev_win32.c required on win32 platforms only
		3572
		3573	ev_select.c only when select backend is enabled (which is enabled by default)
		3574	ev_poll.c only when poll backend is enabled (disabled by default)
		3575	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3576	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3577	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3578
		3579	F<ev.c> includes the backend files directly when enabled, so you only need
		3580	to compile this single file.
		3581
		3582	=head3 LIBEVENT COMPATIBILITY API
		3583
		3584	To include the libevent compatibility API, also include:
		3585
		3586	#include "event.c"
		3587
		3588	in the file including F<ev.c>, and:
		3589
		3590	#include "event.h"
		3591
		3592	in the files that want to use the libevent API. This also includes F<ev.h>.
		3593
		3594	You need the following additional files for this:
		3595
		3596	event.h
		3597	event.c
		3598
		3599	=head3 AUTOCONF SUPPORT
		3600
		3601	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3602	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3603	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3604	include F<config.h> and configure itself accordingly.
		3605
		3606	For this of course you need the m4 file:
		3607
		3608	libev.m4
		3609
		3610	=head2 PREPROCESSOR SYMBOLS/MACROS
		3611
		3612	Libev can be configured via a variety of preprocessor symbols you have to
		3613	define before including any of its files. The default in the absence of
		3614	autoconf is documented for every option.
		3615
		3616	=over 4
		3617
		3618	=item EV_STANDALONE
		3619
		3620	Must always be C<1> if you do not use autoconf configuration, which
		3621	keeps libev from including F<config.h>, and it also defines dummy
		3622	implementations for some libevent functions (such as logging, which is not
		3623	supported). It will also not define any of the structs usually found in
		3624	F<event.h> that are not directly supported by the libev core alone.
		3625
		3626	In standalone mode, libev will still try to automatically deduce the
		3627	configuration, but has to be more conservative.
		3628
		3629	=item EV_USE_MONOTONIC
		3630
		3631	If defined to be C<1>, libev will try to detect the availability of the
		3632	monotonic clock option at both compile time and runtime. Otherwise no
		3633	use of the monotonic clock option will be attempted. If you enable this,
		3634	you usually have to link against librt or something similar. Enabling it
		3635	when the functionality isn't available is safe, though, although you have
		3636	to make sure you link against any libraries where the C<clock_gettime>
		3637	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		3638
		3639	=item EV_USE_REALTIME
		3640
		3641	If defined to be C<1>, libev will try to detect the availability of the
		3642	real-time clock option at compile time (and assume its availability
		3643	at runtime if successful). Otherwise no use of the real-time clock
		3644	option will be attempted. This effectively replaces C<gettimeofday>
		3645	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		3646	correctness. See the note about libraries in the description of
		3647	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3648	C<EV_USE_CLOCK_SYSCALL>.
		3649
		3650	=item EV_USE_CLOCK_SYSCALL
		3651
		3652	If defined to be C<1>, libev will try to use a direct syscall instead
		3653	of calling the system-provided C<clock_gettime> function. This option
		3654	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3655	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3656	programs needlessly. Using a direct syscall is slightly slower (in
		3657	theory), because no optimised vdso implementation can be used, but avoids
		3658	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		3659	higher, as it simplifies linking (no need for C<-lrt>).
		3660
		3661	=item EV_USE_NANOSLEEP
		3662
		3663	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3664	and will use it for delays. Otherwise it will use C<select ()>.
		3665
		3666	=item EV_USE_EVENTFD
		3667
		3668	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3669	available and will probe for kernel support at runtime. This will improve
		3670	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3671	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3672	2.7 or newer, otherwise disabled.
		3673
		3674	=item EV_USE_SELECT
		3675
		3676	If undefined or defined to be C<1>, libev will compile in support for the
		3677	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3678	other method takes over, select will be it. Otherwise the select backend
		3679	will not be compiled in.
		3680
		3681	=item EV_SELECT_USE_FD_SET
		3682
		3683	If defined to C<1>, then the select backend will use the system C<fd_set>
		3684	structure. This is useful if libev doesn't compile due to a missing
		3685	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		3686	on exotic systems. This usually limits the range of file descriptors to
		3687	some low limit such as 1024 or might have other limitations (winsocket
		3688	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		3689	configures the maximum size of the C<fd_set>.
		3690
		3691	=item EV_SELECT_IS_WINSOCKET
		3692
		3693	When defined to C<1>, the select backend will assume that
		3694	select/socket/connect etc. don't understand file descriptors but
		3695	wants osf handles on win32 (this is the case when the select to
		3696	be used is the winsock select). This means that it will call
		3697	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3698	it is assumed that all these functions actually work on fds, even
		3699	on win32. Should not be defined on non-win32 platforms.
		3700
		3701	=item EV_FD_TO_WIN32_HANDLE(fd)
		3702
		3703	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3704	file descriptors to socket handles. When not defining this symbol (the
		3705	default), then libev will call C<_get_osfhandle>, which is usually
		3706	correct. In some cases, programs use their own file descriptor management,
		3707	in which case they can provide this function to map fds to socket handles.
		3708
		3709	=item EV_WIN32_HANDLE_TO_FD(handle)
		3710
		3711	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		3712	using the standard C<_open_osfhandle> function. For programs implementing
		3713	their own fd to handle mapping, overwriting this function makes it easier
		3714	to do so. This can be done by defining this macro to an appropriate value.
		3715
		3716	=item EV_WIN32_CLOSE_FD(fd)
		3717
		3718	If programs implement their own fd to handle mapping on win32, then this
		3719	macro can be used to override the C<close> function, useful to unregister
		3720	file descriptors again. Note that the replacement function has to close
		3721	the underlying OS handle.
		3722
		3723	=item EV_USE_POLL
		3724
		3725	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3726	backend. Otherwise it will be enabled on non-win32 platforms. It
		3727	takes precedence over select.
		3728
		3729	=item EV_USE_EPOLL
		3730
		3731	If defined to be C<1>, libev will compile in support for the Linux
		3732	C<epoll>(7) backend. Its availability will be detected at runtime,
		3733	otherwise another method will be used as fallback. This is the preferred
		3734	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3735	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3736
		3737	=item EV_USE_KQUEUE
		3738
		3739	If defined to be C<1>, libev will compile in support for the BSD style
		3740	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3741	otherwise another method will be used as fallback. This is the preferred
		3742	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3743	supports some types of fds correctly (the only platform we found that
		3744	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3745	not be used unless explicitly requested. The best way to use it is to find
		3746	out whether kqueue supports your type of fd properly and use an embedded
		3747	kqueue loop.
		3748
		3749	=item EV_USE_PORT
		3750
		3751	If defined to be C<1>, libev will compile in support for the Solaris
		3752	10 port style backend. Its availability will be detected at runtime,
		3753	otherwise another method will be used as fallback. This is the preferred
		3754	backend for Solaris 10 systems.
		3755
		3756	=item EV_USE_DEVPOLL
		3757
		3758	Reserved for future expansion, works like the USE symbols above.
		3759
		3760	=item EV_USE_INOTIFY
		3761
		3762	If defined to be C<1>, libev will compile in support for the Linux inotify
		3763	interface to speed up C<ev_stat> watchers. Its actual availability will
		3764	be detected at runtime. If undefined, it will be enabled if the headers
		3765	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3766
		3767	=item EV_ATOMIC_T
		3768
		3769	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3770	access is atomic with respect to other threads or signal contexts. No such
		3771	type is easily found in the C language, so you can provide your own type
		3772	that you know is safe for your purposes. It is used both for signal handler "locking"
		3773	as well as for signal and thread safety in C<ev_async> watchers.
		3774
		3775	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3776	(from F<signal.h>), which is usually good enough on most platforms.
		3777
		3778	=item EV_H
		3779
		3780	The name of the F<ev.h> header file used to include it. The default if
		3781	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3782	used to virtually rename the F<ev.h> header file in case of conflicts.
		3783
		3784	=item EV_CONFIG_H
		3785
		3786	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3787	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3788	C<EV_H>, above.
		3789
		3790	=item EV_EVENT_H
		3791
		3792	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3793	of how the F<event.h> header can be found, the default is C<"event.h">.
		3794
		3795	=item EV_PROTOTYPES
		3796
		3797	If defined to be C<0>, then F<ev.h> will not define any function
		3798	prototypes, but still define all the structs and other symbols. This is
		3799	occasionally useful if you want to provide your own wrapper functions
		3800	around libev functions.
		3801
		3802	=item EV_MULTIPLICITY
		3803
		3804	If undefined or defined to C<1>, then all event-loop-specific functions
		3805	will have the C<struct ev_loop *> as first argument, and you can create
		3806	additional independent event loops. Otherwise there will be no support
		3807	for multiple event loops and there is no first event loop pointer
		3808	argument. Instead, all functions act on the single default loop.
		3809
		3810	=item EV_MINPRI
		3811
		3812	=item EV_MAXPRI
		3813
		3814	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3815	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3816	provide for more priorities by overriding those symbols (usually defined
		3817	to be C<-2> and C<2>, respectively).
		3818
		3819	When doing priority-based operations, libev usually has to linearly search
		3820	all the priorities, so having many of them (hundreds) uses a lot of space
		3821	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3822	fine.
		3823
		3824	If your embedding application does not need any priorities, defining these
		3825	both to C<0> will save some memory and CPU.
		3826
		3827	=item EV_PERIODIC_ENABLE
		3828
		3829	If undefined or defined to be C<1>, then periodic timers are supported. If
		3830	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3831	code.
		3832
		3833	=item EV_IDLE_ENABLE
		3834
		3835	If undefined or defined to be C<1>, then idle watchers are supported. If
		3836	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3837	code.
		3838
		3839	=item EV_EMBED_ENABLE
		3840
		3841	If undefined or defined to be C<1>, then embed watchers are supported. If
		3842	defined to be C<0>, then they are not. Embed watchers rely on most other
		3843	watcher types, which therefore must not be disabled.
		3844
		3845	=item EV_STAT_ENABLE
		3846
		3847	If undefined or defined to be C<1>, then stat watchers are supported. If
		3848	defined to be C<0>, then they are not.
		3849
		3850	=item EV_FORK_ENABLE
		3851
		3852	If undefined or defined to be C<1>, then fork watchers are supported. If
		3853	defined to be C<0>, then they are not.
		3854
		3855	=item EV_ASYNC_ENABLE
		3856
		3857	If undefined or defined to be C<1>, then async watchers are supported. If
		3858	defined to be C<0>, then they are not.
		3859
		3860	=item EV_MINIMAL
		3861
		3862	If you need to shave off some kilobytes of code at the expense of some
		3863	speed (but with the full API), define this symbol to C<1>. Currently this
		3864	is used to override some inlining decisions, saves roughly 30% code size
		3865	on amd64. It also selects a much smaller 2-heap for timer management over
		3866	the default 4-heap.
		3867
		3868	You can save even more by disabling watcher types you do not need
		3869	and setting C<EV_MAXPRI> == C<EV_MINPRI>. Also, disabling C<assert>
		3870	(C<-DNDEBUG>) will usually reduce code size a lot.
		3871
		3872	Defining C<EV_MINIMAL> to C<2> will additionally reduce the core API to
		3873	provide a bare-bones event library. See C<ev.h> for details on what parts
		3874	of the API are still available, and do not complain if this subset changes
		3875	over time.
		3876
		3877	=item EV_NSIG
		3878
		3879	The highest supported signal number, +1 (or, the number of
		3880	signals): Normally, libev tries to deduce the maximum number of signals
		3881	automatically, but sometimes this fails, in which case it can be
		3882	specified. Also, using a lower number than detected (C<32> should be
		3883	good for about any system in existance) can save some memory, as libev
		3884	statically allocates some 12-24 bytes per signal number.
		3885
		3886	=item EV_PID_HASHSIZE
		3887
		3888	C<ev_child> watchers use a small hash table to distribute workload by
		3889	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
		3890	than enough. If you need to manage thousands of children you might want to
		3891	increase this value (I<must> be a power of two).
		3892
		3893	=item EV_INOTIFY_HASHSIZE
		3894
		3895	C<ev_stat> watchers use a small hash table to distribute workload by
		3896	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		3897	usually more than enough. If you need to manage thousands of C<ev_stat>
		3898	watchers you might want to increase this value (I<must> be a power of
		3899	two).
		3900
		3901	=item EV_USE_4HEAP
		3902
		3903	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3904	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3905	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3906	faster performance with many (thousands) of watchers.
		3907
		3908	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3909	(disabled).
		3910
		3911	=item EV_HEAP_CACHE_AT
		3912
		3913	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3914	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3915	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3916	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3917	but avoids random read accesses on heap changes. This improves performance
		3918	noticeably with many (hundreds) of watchers.
		3919
		3920	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3921	(disabled).
		3922
		3923	=item EV_VERIFY
		3924
		3925	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3926	be done: If set to C<0>, no internal verification code will be compiled
		3927	in. If set to C<1>, then verification code will be compiled in, but not
		3928	called. If set to C<2>, then the internal verification code will be
		3929	called once per loop, which can slow down libev. If set to C<3>, then the
		3930	verification code will be called very frequently, which will slow down
		3931	libev considerably.
		3932
		3933	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3934	C<0>.
		3935
		3936	=item EV_COMMON
		3937
		3938	By default, all watchers have a C<void *data> member. By redefining
		3939	this macro to a something else you can include more and other types of
		3940	members. You have to define it each time you include one of the files,
		3941	though, and it must be identical each time.
		3942
		3943	For example, the perl EV module uses something like this:
		3944
		3945	#define EV_COMMON \
		3946	SV *self; /* contains this struct */ \
		3947	SV *cb_sv, *fh /* note no trailing ";" */
		3948
		3949	=item EV_CB_DECLARE (type)
		3950
		3951	=item EV_CB_INVOKE (watcher, revents)
		3952
		3953	=item ev_set_cb (ev, cb)
		3954
		3955	Can be used to change the callback member declaration in each watcher,
		3956	and the way callbacks are invoked and set. Must expand to a struct member
		3957	definition and a statement, respectively. See the F<ev.h> header file for
		3958	their default definitions. One possible use for overriding these is to
		3959	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		3960	method calls instead of plain function calls in C++.
		3961
		3962	=back
		3963
		3964	=head2 EXPORTED API SYMBOLS
		3965
		3966	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3967	exported symbols, you can use the provided F<Symbol.*> files which list
		3968	all public symbols, one per line:
		3969
		3970	Symbols.ev for libev proper
		3971	Symbols.event for the libevent emulation
		3972
		3973	This can also be used to rename all public symbols to avoid clashes with
		3974	multiple versions of libev linked together (which is obviously bad in
		3975	itself, but sometimes it is inconvenient to avoid this).
		3976
		3977	A sed command like this will create wrapper C<#define>'s that you need to
		3978	include before including F<ev.h>:
		3979
		3980	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3981
		3982	This would create a file F<wrap.h> which essentially looks like this:
		3983
		3984	#define ev_backend myprefix_ev_backend
		3985	#define ev_check_start myprefix_ev_check_start
		3986	#define ev_check_stop myprefix_ev_check_stop
		3987	...
		3988
		3989	=head2 EXAMPLES
		3990
		3991	For a real-world example of a program the includes libev
		3992	verbatim, you can have a look at the EV perl module
		3993	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		3994	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		3995	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		3996	will be compiled. It is pretty complex because it provides its own header
		3997	file.
		3998
		3999	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		4000	that everybody includes and which overrides some configure choices:
		4001
		4002	#define EV_MINIMAL 1
		4003	#define EV_USE_POLL 0
		4004	#define EV_MULTIPLICITY 0
		4005	#define EV_PERIODIC_ENABLE 0
		4006	#define EV_STAT_ENABLE 0
		4007	#define EV_FORK_ENABLE 0
		4008	#define EV_CONFIG_H <config.h>
		4009	#define EV_MINPRI 0
		4010	#define EV_MAXPRI 0
		4011
		4012	#include "ev++.h"
		4013
		4014	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4015
		4016	#include "ev_cpp.h"
		4017	#include "ev.c"
		4018
		4019	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		4020
		4021	=head2 THREADS AND COROUTINES
		4022
		4023	=head3 THREADS
		4024
		4025	All libev functions are reentrant and thread-safe unless explicitly
		4026	documented otherwise, but libev implements no locking itself. This means
		4027	that you can use as many loops as you want in parallel, as long as there
		4028	are no concurrent calls into any libev function with the same loop
		4029	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4030	of course): libev guarantees that different event loops share no data
		4031	structures that need any locking.
		4032
		4033	Or to put it differently: calls with different loop parameters can be done
		4034	concurrently from multiple threads, calls with the same loop parameter
		4035	must be done serially (but can be done from different threads, as long as
		4036	only one thread ever is inside a call at any point in time, e.g. by using
		4037	a mutex per loop).
		4038
		4039	Specifically to support threads (and signal handlers), libev implements
		4040	so-called C<ev_async> watchers, which allow some limited form of
		4041	concurrency on the same event loop, namely waking it up "from the
		4042	outside".
		4043
		4044	If you want to know which design (one loop, locking, or multiple loops
		4045	without or something else still) is best for your problem, then I cannot
		4046	help you, but here is some generic advice:
		4047
		4048	=over 4
		4049
		4050	=item * most applications have a main thread: use the default libev loop
		4051	in that thread, or create a separate thread running only the default loop.
		4052
		4053	This helps integrating other libraries or software modules that use libev
		4054	themselves and don't care/know about threading.
		4055
		4056	=item * one loop per thread is usually a good model.
		4057
		4058	Doing this is almost never wrong, sometimes a better-performance model
		4059	exists, but it is always a good start.
		4060
		4061	=item * other models exist, such as the leader/follower pattern, where one
		4062	loop is handed through multiple threads in a kind of round-robin fashion.
		4063
		4064	Choosing a model is hard - look around, learn, know that usually you can do
		4065	better than you currently do :-)
		4066
		4067	=item * often you need to talk to some other thread which blocks in the
		4068	event loop.
		4069
		4070	C<ev_async> watchers can be used to wake them up from other threads safely
		4071	(or from signal contexts...).
		4072
		4073	An example use would be to communicate signals or other events that only
		4074	work in the default loop by registering the signal watcher with the
		4075	default loop and triggering an C<ev_async> watcher from the default loop
		4076	watcher callback into the event loop interested in the signal.
		4077
		4078	=back
		4079
		4080	=head4 THREAD LOCKING EXAMPLE
		4081
		4082	Here is a fictitious example of how to run an event loop in a different
		4083	thread than where callbacks are being invoked and watchers are
		4084	created/added/removed.
		4085
		4086	For a real-world example, see the C<EV::Loop::Async> perl module,
		4087	which uses exactly this technique (which is suited for many high-level
		4088	languages).
		4089
		4090	The example uses a pthread mutex to protect the loop data, a condition
		4091	variable to wait for callback invocations, an async watcher to notify the
		4092	event loop thread and an unspecified mechanism to wake up the main thread.
		4093
		4094	First, you need to associate some data with the event loop:
		4095
		4096	typedef struct {
		4097	mutex_t lock; /* global loop lock */
		4098	ev_async async_w;
		4099	thread_t tid;
		4100	cond_t invoke_cv;
		4101	} userdata;
		4102
		4103	void prepare_loop (EV_P)
		4104	{
		4105	// for simplicity, we use a static userdata struct.
		4106	static userdata u;
		4107
		4108	ev_async_init (&u->async_w, async_cb);
		4109	ev_async_start (EV_A_ &u->async_w);
		4110
		4111	pthread_mutex_init (&u->lock, 0);
		4112	pthread_cond_init (&u->invoke_cv, 0);
		4113
		4114	// now associate this with the loop
		4115	ev_set_userdata (EV_A_ u);
		4116	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		4117	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		4118
		4119	// then create the thread running ev_loop
		4120	pthread_create (&u->tid, 0, l_run, EV_A);
		4121	}
		4122
		4123	The callback for the C<ev_async> watcher does nothing: the watcher is used
		4124	solely to wake up the event loop so it takes notice of any new watchers
		4125	that might have been added:
		4126
		4127	static void
		4128	async_cb (EV_P_ ev_async *w, int revents)
		4129	{
		4130	// just used for the side effects
		4131	}
		4132
		4133	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		4134	protecting the loop data, respectively.
		4135
		4136	static void
		4137	l_release (EV_P)
		4138	{
		4139	userdata *u = ev_userdata (EV_A);
		4140	pthread_mutex_unlock (&u->lock);
		4141	}
		4142
		4143	static void
		4144	l_acquire (EV_P)
		4145	{
		4146	userdata *u = ev_userdata (EV_A);
		4147	pthread_mutex_lock (&u->lock);
		4148	}
		4149
		4150	The event loop thread first acquires the mutex, and then jumps straight
		4151	into C<ev_loop>:
		4152
		4153	void *
		4154	l_run (void *thr_arg)
		4155	{
		4156	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		4157
		4158	l_acquire (EV_A);
		4159	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		4160	ev_loop (EV_A_ 0);
		4161	l_release (EV_A);
		4162
		4163	return 0;
		4164	}
		4165
		4166	Instead of invoking all pending watchers, the C<l_invoke> callback will
		4167	signal the main thread via some unspecified mechanism (signals? pipe
		4168	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		4169	have been called (in a while loop because a) spurious wakeups are possible
		4170	and b) skipping inter-thread-communication when there are no pending
		4171	watchers is very beneficial):
		4172
		4173	static void
		4174	l_invoke (EV_P)
		4175	{
		4176	userdata *u = ev_userdata (EV_A);
		4177
		4178	while (ev_pending_count (EV_A))
		4179	{
		4180	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		4181	pthread_cond_wait (&u->invoke_cv, &u->lock);
		4182	}
		4183	}
		4184
		4185	Now, whenever the main thread gets told to invoke pending watchers, it
		4186	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		4187	thread to continue:
		4188
		4189	static void
		4190	real_invoke_pending (EV_P)
		4191	{
		4192	userdata *u = ev_userdata (EV_A);
		4193
		4194	pthread_mutex_lock (&u->lock);
		4195	ev_invoke_pending (EV_A);
		4196	pthread_cond_signal (&u->invoke_cv);
		4197	pthread_mutex_unlock (&u->lock);
		4198	}
		4199
		4200	Whenever you want to start/stop a watcher or do other modifications to an
		4201	event loop, you will now have to lock:
		4202
		4203	ev_timer timeout_watcher;
		4204	userdata *u = ev_userdata (EV_A);
		4205
		4206	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		4207
		4208	pthread_mutex_lock (&u->lock);
		4209	ev_timer_start (EV_A_ &timeout_watcher);
		4210	ev_async_send (EV_A_ &u->async_w);
		4211	pthread_mutex_unlock (&u->lock);
		4212
		4213	Note that sending the C<ev_async> watcher is required because otherwise
		4214	an event loop currently blocking in the kernel will have no knowledge
		4215	about the newly added timer. By waking up the loop it will pick up any new
		4216	watchers in the next event loop iteration.
		4217
		4218	=head3 COROUTINES
		4219
		4220	Libev is very accommodating to coroutines ("cooperative threads"):
		4221	libev fully supports nesting calls to its functions from different
		4222	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		4223	different coroutines, and switch freely between both coroutines running
		4224	the loop, as long as you don't confuse yourself). The only exception is
		4225	that you must not do this from C<ev_periodic> reschedule callbacks.
		4226
		4227	Care has been taken to ensure that libev does not keep local state inside
		4228	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		4229	they do not call any callbacks.
		4230
		4231	=head2 COMPILER WARNINGS
		4232
		4233	Depending on your compiler and compiler settings, you might get no or a
		4234	lot of warnings when compiling libev code. Some people are apparently
		4235	scared by this.
		4236
		4237	However, these are unavoidable for many reasons. For one, each compiler
		4238	has different warnings, and each user has different tastes regarding
		4239	warning options. "Warn-free" code therefore cannot be a goal except when
		4240	targeting a specific compiler and compiler-version.
		4241
		4242	Another reason is that some compiler warnings require elaborate
		4243	workarounds, or other changes to the code that make it less clear and less
		4244	maintainable.
		4245
		4246	And of course, some compiler warnings are just plain stupid, or simply
		4247	wrong (because they don't actually warn about the condition their message
		4248	seems to warn about). For example, certain older gcc versions had some
		4249	warnings that resulted an extreme number of false positives. These have
		4250	been fixed, but some people still insist on making code warn-free with
		4251	such buggy versions.
		4252
		4253	While libev is written to generate as few warnings as possible,
		4254	"warn-free" code is not a goal, and it is recommended not to build libev
		4255	with any compiler warnings enabled unless you are prepared to cope with
		4256	them (e.g. by ignoring them). Remember that warnings are just that:
		4257	warnings, not errors, or proof of bugs.
		4258
		4259
		4260	=head2 VALGRIND
		4261
		4262	Valgrind has a special section here because it is a popular tool that is
		4263	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4264
		4265	If you think you found a bug (memory leak, uninitialised data access etc.)
		4266	in libev, then check twice: If valgrind reports something like:
		4267
		4268	==2274== definitely lost: 0 bytes in 0 blocks.
		4269	==2274== possibly lost: 0 bytes in 0 blocks.
		4270	==2274== still reachable: 256 bytes in 1 blocks.
		4271
		4272	Then there is no memory leak, just as memory accounted to global variables
		4273	is not a memleak - the memory is still being referenced, and didn't leak.
		4274
		4275	Similarly, under some circumstances, valgrind might report kernel bugs
		4276	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4277	although an acceptable workaround has been found here), or it might be
		4278	confused.
		4279
		4280	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4281	make it into some kind of religion.
		4282
		4283	If you are unsure about something, feel free to contact the mailing list
		4284	with the full valgrind report and an explanation on why you think this
		4285	is a bug in libev (best check the archives, too :). However, don't be
		4286	annoyed when you get a brisk "this is no bug" answer and take the chance
		4287	of learning how to interpret valgrind properly.
		4288
		4289	If you need, for some reason, empty reports from valgrind for your project
		4290	I suggest using suppression lists.
		4291
		4292
		4293	=head1 PORTABILITY NOTES
		4294
		4295	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4296
		4297	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		4298	requires, and its I/O model is fundamentally incompatible with the POSIX
		4299	model. Libev still offers limited functionality on this platform in
		4300	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		4301	descriptors. This only applies when using Win32 natively, not when using
		4302	e.g. cygwin.
		4303
		4304	Lifting these limitations would basically require the full
		4305	re-implementation of the I/O system. If you are into these kinds of
		4306	things, then note that glib does exactly that for you in a very portable
		4307	way (note also that glib is the slowest event library known to man).
		4308
		4309	There is no supported compilation method available on windows except
		4310	embedding it into other applications.
		4311
		4312	Sensible signal handling is officially unsupported by Microsoft - libev
		4313	tries its best, but under most conditions, signals will simply not work.
		4314
		4315	Not a libev limitation but worth mentioning: windows apparently doesn't
		4316	accept large writes: instead of resulting in a partial write, windows will
		4317	either accept everything or return C<ENOBUFS> if the buffer is too large,
		4318	so make sure you only write small amounts into your sockets (less than a
		4319	megabyte seems safe, but this apparently depends on the amount of memory
		4320	available).
		4321
		4322	Due to the many, low, and arbitrary limits on the win32 platform and
		4323	the abysmal performance of winsockets, using a large number of sockets
		4324	is not recommended (and not reasonable). If your program needs to use
		4325	more than a hundred or so sockets, then likely it needs to use a totally
		4326	different implementation for windows, as libev offers the POSIX readiness
		4327	notification model, which cannot be implemented efficiently on windows
		4328	(due to Microsoft monopoly games).
		4329
		4330	A typical way to use libev under windows is to embed it (see the embedding
		4331	section for details) and use the following F<evwrap.h> header file instead
		4332	of F<ev.h>:
		4333
		4334	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4335	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4336
		4337	#include "ev.h"
		4338
		4339	And compile the following F<evwrap.c> file into your project (make sure
		4340	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4341
		4342	#include "evwrap.h"
		4343	#include "ev.c"
		4344
		4345	=over 4
		4346
		4347	=item The winsocket select function
		4348
		4349	The winsocket C<select> function doesn't follow POSIX in that it
		4350	requires socket I<handles> and not socket I<file descriptors> (it is
		4351	also extremely buggy). This makes select very inefficient, and also
		4352	requires a mapping from file descriptors to socket handles (the Microsoft
		4353	C runtime provides the function C<_open_osfhandle> for this). See the
		4354	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		4355	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		4356
		4357	The configuration for a "naked" win32 using the Microsoft runtime
		4358	libraries and raw winsocket select is:
		4359
		4360	#define EV_USE_SELECT 1
		4361	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		4362
		4363	Note that winsockets handling of fd sets is O(n), so you can easily get a
		4364	complexity in the O(n²) range when using win32.
		4365
		4366	=item Limited number of file descriptors
		4367
		4368	Windows has numerous arbitrary (and low) limits on things.
		4369
		4370	Early versions of winsocket's select only supported waiting for a maximum
		4371	of C<64> handles (probably owning to the fact that all windows kernels
		4372	can only wait for C<64> things at the same time internally; Microsoft
		4373	recommends spawning a chain of threads and wait for 63 handles and the
		4374	previous thread in each. Sounds great!).
		4375
		4376	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		4377	to some high number (e.g. C<2048>) before compiling the winsocket select
		4378	call (which might be in libev or elsewhere, for example, perl and many
		4379	other interpreters do their own select emulation on windows).
		4380
		4381	Another limit is the number of file descriptors in the Microsoft runtime
		4382	libraries, which by default is C<64> (there must be a hidden I<64>
		4383	fetish or something like this inside Microsoft). You can increase this
		4384	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		4385	(another arbitrary limit), but is broken in many versions of the Microsoft
		4386	runtime libraries. This might get you to about C<512> or C<2048> sockets
		4387	(depending on windows version and/or the phase of the moon). To get more,
		4388	you need to wrap all I/O functions and provide your own fd management, but
		4389	the cost of calling select (O(n²)) will likely make this unworkable.
		4390
		4391	=back
		4392
		4393	=head2 PORTABILITY REQUIREMENTS
		4394
		4395	In addition to a working ISO-C implementation and of course the
		4396	backend-specific APIs, libev relies on a few additional extensions:
		4397
		4398	=over 4
		4399
		4400	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4401	calling conventions regardless of C<ev_watcher_type *>.
		4402
		4403	Libev assumes not only that all watcher pointers have the same internal
		4404	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4405	assumes that the same (machine) code can be used to call any watcher
		4406	callback: The watcher callbacks have different type signatures, but libev
		4407	calls them using an C<ev_watcher *> internally.
		4408
		4409	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4410
		4411	The type C<sig_atomic_t volatile> (or whatever is defined as
		4412	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4413	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4414	believed to be sufficiently portable.
		4415
		4416	=item C<sigprocmask> must work in a threaded environment
		4417
		4418	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4419	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		4420	pthread implementations will either allow C<sigprocmask> in the "main
		4421	thread" or will block signals process-wide, both behaviours would
		4422	be compatible with libev. Interaction between C<sigprocmask> and
		4423	C<pthread_sigmask> could complicate things, however.
		4424
		4425	The most portable way to handle signals is to block signals in all threads
		4426	except the initial one, and run the default loop in the initial thread as
		4427	well.
		4428
		4429	=item C<long> must be large enough for common memory allocation sizes
		4430
		4431	To improve portability and simplify its API, libev uses C<long> internally
		4432	instead of C<size_t> when allocating its data structures. On non-POSIX
		4433	systems (Microsoft...) this might be unexpectedly low, but is still at
		4434	least 31 bits everywhere, which is enough for hundreds of millions of
		4435	watchers.
		4436
		4437	=item C<double> must hold a time value in seconds with enough accuracy
		4438
		4439	The type C<double> is used to represent timestamps. It is required to
		4440	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		4441	enough for at least into the year 4000. This requirement is fulfilled by
		4442	implementations implementing IEEE 754, which is basically all existing
		4443	ones. With IEEE 754 doubles, you get microsecond accuracy until at least
		4444	2200.
		4445
		4446	=back
		4447
		4448	If you know of other additional requirements drop me a note.
		4449
		4450
		4451	=head1 ALGORITHMIC COMPLEXITIES
		4452
		4453	In this section the complexities of (many of) the algorithms used inside
		4454	libev will be documented. For complexity discussions about backends see
		4455	the documentation for C<ev_default_init>.
		4456
		4457	All of the following are about amortised time: If an array needs to be
		4458	extended, libev needs to realloc and move the whole array, but this
		4459	happens asymptotically rarer with higher number of elements, so O(1) might
		4460	mean that libev does a lengthy realloc operation in rare cases, but on
		4461	average it is much faster and asymptotically approaches constant time.
		4462
		4463	=over 4
		4464
		4465	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		4466
		4467	This means that, when you have a watcher that triggers in one hour and
		4468	there are 100 watchers that would trigger before that, then inserting will
		4469	have to skip roughly seven (C<ld 100>) of these watchers.
		4470
		4471	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		4472
		4473	That means that changing a timer costs less than removing/adding them,
		4474	as only the relative motion in the event queue has to be paid for.
		4475
		4476	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		4477
		4478	These just add the watcher into an array or at the head of a list.
		4479
		4480	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		4481
		4482	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		4483
		4484	These watchers are stored in lists, so they need to be walked to find the
		4485	correct watcher to remove. The lists are usually short (you don't usually
		4486	have many watchers waiting for the same fd or signal: one is typical, two
		4487	is rare).
		4488
		4489	=item Finding the next timer in each loop iteration: O(1)
		4490
		4491	By virtue of using a binary or 4-heap, the next timer is always found at a
		4492	fixed position in the storage array.
		4493
		4494	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		4495
		4496	A change means an I/O watcher gets started or stopped, which requires
		4497	libev to recalculate its status (and possibly tell the kernel, depending
		4498	on backend and whether C<ev_io_set> was used).
		4499
		4500	=item Activating one watcher (putting it into the pending state): O(1)
		4501
		4502	=item Priority handling: O(number_of_priorities)
		4503
		4504	Priorities are implemented by allocating some space for each
		4505	priority. When doing priority-based operations, libev usually has to
		4506	linearly search all the priorities, but starting/stopping and activating
		4507	watchers becomes O(1) with respect to priority handling.
		4508
		4509	=item Sending an ev_async: O(1)
		4510
		4511	=item Processing ev_async_send: O(number_of_async_watchers)
		4512
		4513	=item Processing signals: O(max_signal_number)
		4514
		4515	Sending involves a system call I<iff> there were no other C<ev_async_send>
		4516	calls in the current loop iteration. Checking for async and signal events
		4517	involves iterating over all running async watchers or all signal numbers.
		4518
		4519	=back
		4520
		4521
		4522	=head1 GLOSSARY
		4523
		4524	=over 4
		4525
		4526	=item active
		4527
		4528	A watcher is active as long as it has been started (has been attached to
		4529	an event loop) but not yet stopped (disassociated from the event loop).
		4530
		4531	=item application
		4532
		4533	In this document, an application is whatever is using libev.
		4534
		4535	=item callback
		4536
		4537	The address of a function that is called when some event has been
		4538	detected. Callbacks are being passed the event loop, the watcher that
		4539	received the event, and the actual event bitset.
		4540
		4541	=item callback invocation
		4542
		4543	The act of calling the callback associated with a watcher.
		4544
		4545	=item event
		4546
		4547	A change of state of some external event, such as data now being available
		4548	for reading on a file descriptor, time having passed or simply not having
		4549	any other events happening anymore.
		4550
		4551	In libev, events are represented as single bits (such as C<EV_READ> or
		4552	C<EV_TIMEOUT>).
		4553
		4554	=item event library
		4555
		4556	A software package implementing an event model and loop.
		4557
		4558	=item event loop
		4559
		4560	An entity that handles and processes external events and converts them
		4561	into callback invocations.
		4562
		4563	=item event model
		4564
		4565	The model used to describe how an event loop handles and processes
		4566	watchers and events.
		4567
		4568	=item pending
		4569
		4570	A watcher is pending as soon as the corresponding event has been detected,
		4571	and stops being pending as soon as the watcher will be invoked or its
		4572	pending status is explicitly cleared by the application.
		4573
		4574	A watcher can be pending, but not active. Stopping a watcher also clears
		4575	its pending status.
		4576
		4577	=item real time
		4578
		4579	The physical time that is observed. It is apparently strictly monotonic :)
		4580
		4581	=item wall-clock time
		4582
		4583	The time and date as shown on clocks. Unlike real time, it can actually
		4584	be wrong and jump forwards and backwards, e.g. when the you adjust your
		4585	clock.
		4586
		4587	=item watcher
		4588
		4589	A data structure that describes interest in certain events. Watchers need
		4590	to be started (attached to an event loop) before they can receive events.
		4591
		4592	=item watcher invocation
		4593
		4594	The act of calling the callback associated with a watcher.
		4595
		4596	=back
		4597
767	=head1 AUTHOR	4598	=head1 AUTHOR
768		4599
769	Marc Lehmann <libev@schmorp.de>.	4600	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
770		4601

Diff Legend

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