ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.20 by root, Mon Nov 12 09:06:09 2007 UTC vs.
Revision 1.275 by root, Sat Dec 26 09:21:54 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 REPRESENTATION	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
51		157
52	=head1 GLOBAL FUNCTIONS	158	=head1 GLOBAL FUNCTIONS
53		159
54	These functions can be called anytime, even before initialising the	160	These functions can be called anytime, even before initialising the
55	library in any way.	161	library in any way.
56		162
57	=over 4	163	=over 4
58		164
59	=item ev_tstamp ev_time ()	165	=item ev_tstamp ev_time ()
60		166
61	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 ()>.
62		176
63	=item int ev_version_major ()	177	=item int ev_version_major ()
64		178
65	=item int ev_version_minor ()	179	=item int ev_version_minor ()
66		180
67	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
68	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
69	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
70	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
71	version of the library your program was compiled against.	185	version of the library your program was compiled against.
72		186
		187	These version numbers refer to the ABI version of the library, not the
		188	release version.
		189
73	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,
74	as this indicates an incompatible change. Minor versions are usually	191	as this indicates an incompatible change. Minor versions are usually
75	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
76	not a problem.	193	not a problem.
77		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
78	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	234	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
79		235
80	Sets the allocation function to use (the prototype is similar to the	236	Sets the allocation function to use (the prototype is similar - the
81	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
82	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
83	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
84	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.
85		245
86	You could override this function in high-availability programs to, say,	246	You could override this function in high-availability programs to, say,
87	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,
88	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.
89		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
90	=item ev_set_syserr_cb (void (*cb)(const char *msg));	270	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
91		271
92	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
93	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
94	indicating the system call or subsystem causing the problem. If this	274	indicating the system call or subsystem causing the problem. If this
95	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
96	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
97	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
98	(such as abort).	278	(such as abort).
99		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
100	=back	292	=back
101		293
102	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	294	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
103		295
104	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>
105	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>
106	events, and dynamically created loops which do not.	298	I<function>).
107		299
108	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
109	in your main thread (or in a separate thread) and for each thread you	301	supports signals and child events, and dynamically created loops which do
110	create, you also create another event loop. Libev itself does no locking	302	not.
111	whatsoever, so if you mix calls to the same event loop in different
112	threads, make sure you lock (this is usually a bad idea, though, even if
113	done correctly, because it's hideous and inefficient).
114		303
115	=over 4	304	=over 4
116		305
117	=item struct ev_loop *ev_default_loop (unsigned int flags)	306	=item struct ev_loop *ev_default_loop (unsigned int flags)
118		307
119	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
120	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
121	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
122	flags).	311	flags. If that is troubling you, check C<ev_backend ()> afterwards).
123		312
124	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
125	function.	314	function.
126		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
127	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
128	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>).
129		329
130	It supports the following flags:	330	The following flags are supported:
131		331
132	=over 4	332	=over 4
133		333
134	=item C<EVFLAG_AUTO>	334	=item C<EVFLAG_AUTO>
135		335
136	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
137	thing, believe me).	337	thing, believe me).
138		338
139	=item C<EVFLAG_NOENV>	339	=item C<EVFLAG_NOENV>
140		340
141	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
142	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
143	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	343	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
144	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
145	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
146	around bugs.	346	around bugs.
147		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
148	=item C<EVMETHOD_SELECT> (portable select backend)	383	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
149		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
150	=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)
151		403
152	=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.
153		410
154	=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>.
155		413
156	=item C<EVMETHOD_DEVPOLL> (solaris 8 only)	414	=item C<EVBACKEND_EPOLL> (value 4, Linux)
157		415
158	=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).
159		418
160	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,
161	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
162	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.
163		539
164	=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);
165		563
166	=item struct ev_loop *ev_loop_new (unsigned int flags)	564	=item struct ev_loop *ev_loop_new (unsigned int flags)
167		565
168	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
169	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
170	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
171	undefined behaviour (or a failed assertion if assertions are enabled).	569	undefined behaviour (or a failed assertion if assertions are enabled).
172		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
173	=item ev_default_destroy ()	581	=item ev_default_destroy ()
174		582
175	Destroys the default loop again (frees all memory and kernel state	583	Destroys the default loop again (frees all memory and kernel state
176	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
177	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>.
178		599
179	=item ev_loop_destroy (loop)	600	=item ev_loop_destroy (loop)
180		601
181	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
182	earlier call to C<ev_loop_new>.	603	earlier call to C<ev_loop_new>.
183		604
184	=item ev_default_fork ()	605	=item ev_default_fork ()
185		606
		607	This function sets a flag that causes subsequent C<ev_loop> iterations
186	This function reinitialises the kernel state for backends that have	608	to reinitialise the kernel state for backends that have one. Despite the
187	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
188	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
189	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.
190		613
191	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
192	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
193	have to call it.	616	you just fork+exec, you don't have to call it at all.
194		617
195	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
196	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
197	quite nicely into a call to C<pthread_atfork>:	620	quite nicely into a call to C<pthread_atfork>:
198		621
…		…
200		623
201	=item ev_loop_fork (loop)	624	=item ev_loop_fork (loop)
202		625
203	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
204	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
205	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.
206		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
207	=item unsigned int ev_method (loop)	658	=item unsigned int ev_backend (loop)
208		659
209	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
210	use.	661	use.
211		662
212	=item ev_tstamp ev_now (loop)	663	=item ev_tstamp ev_now (loop)
213		664
214	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
215	got events and started processing them. This timestamp does not change	666	received events and started processing them. This timestamp does not
216	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
217	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
218	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>).
219		708
220	=item ev_loop (loop, int flags)	709	=item ev_loop (loop, int flags)
221		710
222	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
223	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
224	events.	713	handling events.
225		714
226	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
227	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.
228		724
229	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
230	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
231	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.
232		729
233	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
234	neccessary) and will handle those and any outstanding ones. It will block	731	necessary) and will handle those and any already outstanding ones. It
235	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
236	one iteration of the loop.	735	iteration of the loop.
237		736
238	This flags value could be used to implement alternative looping	737	This is useful if you are waiting for some external event in conjunction
239	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
240	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!
241		777
242	=item ev_unloop (loop, how)	778	=item ev_unloop (loop, how)
243		779
244	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
245	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
246	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
247	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.
248		788
249	=item ev_ref (loop)	789	=item ev_ref (loop)
250		790
251	=item ev_unref (loop)	791	=item ev_unref (loop)
252		792
253	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
254	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
255	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.
		796
256	a watcher you never unregister that should not keep C<ev_loop> from	797	If you have a watcher you never unregister that should not keep C<ev_loop>
257	returning, ev_unref() after starting, and ev_ref() before stopping it. For	798	from returning, call ev_unref() after starting, and ev_ref() before
		799	stopping it.
		800
258	example, libev itself uses this for its internal signal pipe: It is not	801	As an example, libev itself uses this for its internal signal pipe: It
259	visible to the libev user and should not keep C<ev_loop> from exiting if	802	is not visible to the libev user and should not keep C<ev_loop> from
260	no event watchers registered by it are active. It is also an excellent	803	exiting if no event watchers registered by it are active. It is also an
261	way to do this for generic recurring timers or from within third-party	804	excellent way to do this for generic recurring timers or from within
262	libraries. Just remember to I<unref after start> and I<ref before stop>.	805	third-party libraries. Just remember to I<unref after start> and I<ref
		806	before stop> (but only if the watcher wasn't active before, or was active
		807	before, respectively. Note also that libev might stop watchers itself
		808	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		809	in the callback).
		810
		811	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
		812	running when nothing else is active.
		813
		814	ev_signal exitsig;
		815	ev_signal_init (&exitsig, sig_cb, SIGINT);
		816	ev_signal_start (loop, &exitsig);
		817	evf_unref (loop);
		818
		819	Example: For some weird reason, unregister the above signal handler again.
		820
		821	ev_ref (loop);
		822	ev_signal_stop (loop, &exitsig);
		823
		824	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		825
		826	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		827
		828	These advanced functions influence the time that libev will spend waiting
		829	for events. Both time intervals are by default C<0>, meaning that libev
		830	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		831	latency.
		832
		833	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		834	allows libev to delay invocation of I/O and timer/periodic callbacks
		835	to increase efficiency of loop iterations (or to increase power-saving
		836	opportunities).
		837
		838	The idea is that sometimes your program runs just fast enough to handle
		839	one (or very few) event(s) per loop iteration. While this makes the
		840	program responsive, it also wastes a lot of CPU time to poll for new
		841	events, especially with backends like C<select ()> which have a high
		842	overhead for the actual polling but can deliver many events at once.
		843
		844	By setting a higher I<io collect interval> you allow libev to spend more
		845	time collecting I/O events, so you can handle more events per iteration,
		846	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		847	C<ev_timer>) will be not affected. Setting this to a non-null value will
		848	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		849	sleep time ensures that libev will not poll for I/O events more often then
		850	once per this interval, on average.
		851
		852	Likewise, by setting a higher I<timeout collect interval> you allow libev
		853	to spend more time collecting timeouts, at the expense of increased
		854	latency/jitter/inexactness (the watcher callback will be called
		855	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		856	value will not introduce any overhead in libev.
		857
		858	Many (busy) programs can usually benefit by setting the I/O collect
		859	interval to a value near C<0.1> or so, which is often enough for
		860	interactive servers (of course not for games), likewise for timeouts. It
		861	usually doesn't make much sense to set it to a lower value than C<0.01>,
		862	as this approaches the timing granularity of most systems. Note that if
		863	you do transactions with the outside world and you can't increase the
		864	parallelity, then this setting will limit your transaction rate (if you
		865	need to poll once per transaction and the I/O collect interval is 0.01,
		866	then you can't do more than 100 transations per second).
		867
		868	Setting the I<timeout collect interval> can improve the opportunity for
		869	saving power, as the program will "bundle" timer callback invocations that
		870	are "near" in time together, by delaying some, thus reducing the number of
		871	times the process sleeps and wakes up again. Another useful technique to
		872	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		873	they fire on, say, one-second boundaries only.
		874
		875	Example: we only need 0.1s timeout granularity, and we wish not to poll
		876	more often than 100 times per second:
		877
		878	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		879	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		880
		881	=item ev_invoke_pending (loop)
		882
		883	This call will simply invoke all pending watchers while resetting their
		884	pending state. Normally, C<ev_loop> does this automatically when required,
		885	but when overriding the invoke callback this call comes handy.
		886
		887	=item int ev_pending_count (loop)
		888
		889	Returns the number of pending watchers - zero indicates that no watchers
		890	are pending.
		891
		892	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		893
		894	This overrides the invoke pending functionality of the loop: Instead of
		895	invoking all pending watchers when there are any, C<ev_loop> will call
		896	this callback instead. This is useful, for example, when you want to
		897	invoke the actual watchers inside another context (another thread etc.).
		898
		899	If you want to reset the callback, use C<ev_invoke_pending> as new
		900	callback.
		901
		902	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		903
		904	Sometimes you want to share the same loop between multiple threads. This
		905	can be done relatively simply by putting mutex_lock/unlock calls around
		906	each call to a libev function.
		907
		908	However, C<ev_loop> can run an indefinite time, so it is not feasible to
		909	wait for it to return. One way around this is to wake up the loop via
		910	C<ev_unloop> and C<av_async_send>, another way is to set these I<release>
		911	and I<acquire> callbacks on the loop.
		912
		913	When set, then C<release> will be called just before the thread is
		914	suspended waiting for new events, and C<acquire> is called just
		915	afterwards.
		916
		917	Ideally, C<release> will just call your mutex_unlock function, and
		918	C<acquire> will just call the mutex_lock function again.
		919
		920	While event loop modifications are allowed between invocations of
		921	C<release> and C<acquire> (that's their only purpose after all), no
		922	modifications done will affect the event loop, i.e. adding watchers will
		923	have no effect on the set of file descriptors being watched, or the time
		924	waited. Use an C<ev_async> watcher to wake up C<ev_loop> when you want it
		925	to take note of any changes you made.
		926
		927	In theory, threads executing C<ev_loop> will be async-cancel safe between
		928	invocations of C<release> and C<acquire>.
		929
		930	See also the locking example in the C<THREADS> section later in this
		931	document.
		932
		933	=item ev_set_userdata (loop, void *data)
		934
		935	=item ev_userdata (loop)
		936
		937	Set and retrieve a single C<void *> associated with a loop. When
		938	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		939	C<0.>
		940
		941	These two functions can be used to associate arbitrary data with a loop,
		942	and are intended solely for the C<invoke_pending_cb>, C<release> and
		943	C<acquire> callbacks described above, but of course can be (ab-)used for
		944	any other purpose as well.
		945
		946	=item ev_loop_verify (loop)
		947
		948	This function only does something when C<EV_VERIFY> support has been
		949	compiled in, which is the default for non-minimal builds. It tries to go
		950	through all internal structures and checks them for validity. If anything
		951	is found to be inconsistent, it will print an error message to standard
		952	error and call C<abort ()>.
		953
		954	This can be used to catch bugs inside libev itself: under normal
		955	circumstances, this function will never abort as of course libev keeps its
		956	data structures consistent.
263		957
264	=back	958	=back
265		959
		960
266	=head1 ANATOMY OF A WATCHER	961	=head1 ANATOMY OF A WATCHER
		962
		963	In the following description, uppercase C<TYPE> in names stands for the
		964	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		965	watchers and C<ev_io_start> for I/O watchers.
267		966
268	A watcher is a structure that you create and register to record your	967	A watcher is a structure that you create and register to record your
269	interest in some event. For instance, if you want to wait for STDIN to	968	interest in some event. For instance, if you want to wait for STDIN to
270	become readable, you would create an C<ev_io> watcher for that:	969	become readable, you would create an C<ev_io> watcher for that:
271		970
272	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	971	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
273	{	972	{
274	ev_io_stop (w);	973	ev_io_stop (w);
275	ev_unloop (loop, EVUNLOOP_ALL);	974	ev_unloop (loop, EVUNLOOP_ALL);
276	}	975	}
277		976
278	struct ev_loop *loop = ev_default_loop (0);	977	struct ev_loop *loop = ev_default_loop (0);
		978
279	struct ev_io stdin_watcher;	979	ev_io stdin_watcher;
		980
280	ev_init (&stdin_watcher, my_cb);	981	ev_init (&stdin_watcher, my_cb);
281	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	982	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
282	ev_io_start (loop, &stdin_watcher);	983	ev_io_start (loop, &stdin_watcher);
		984
283	ev_loop (loop, 0);	985	ev_loop (loop, 0);
284		986
285	As you can see, you are responsible for allocating the memory for your	987	As you can see, you are responsible for allocating the memory for your
286	watcher structures (and it is usually a bad idea to do this on the stack,	988	watcher structures (and it is I<usually> a bad idea to do this on the
287	although this can sometimes be quite valid).	989	stack).
		990
		991	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		992	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
288		993
289	Each watcher structure must be initialised by a call to C<ev_init	994	Each watcher structure must be initialised by a call to C<ev_init
290	(watcher *, callback)>, which expects a callback to be provided. This	995	(watcher *, callback)>, which expects a callback to be provided. This
291	callback gets invoked each time the event occurs (or, in the case of io	996	callback gets invoked each time the event occurs (or, in the case of I/O
292	watchers, each time the event loop detects that the file descriptor given	997	watchers, each time the event loop detects that the file descriptor given
293	is readable and/or writable).	998	is readable and/or writable).
294		999
295	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1000	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
296	with arguments specific to this watcher type. There is also a macro	1001	macro to configure it, with arguments specific to the watcher type. There
297	to combine initialisation and setting in one call: C<< ev_<type>_init	1002	is also a macro to combine initialisation and setting in one call: C<<
298	(watcher *, callback, ...) >>.	1003	ev_TYPE_init (watcher *, callback, ...) >>.
299		1004
300	To make the watcher actually watch out for events, you have to start it	1005	To make the watcher actually watch out for events, you have to start it
301	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1006	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
302	*) >>), and you can stop watching for events at any time by calling the	1007	*) >>), and you can stop watching for events at any time by calling the
303	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1008	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
304		1009
305	As long as your watcher is active (has been started but not stopped) you	1010	As long as your watcher is active (has been started but not stopped) you
306	must not touch the values stored in it. Most specifically you must never	1011	must not touch the values stored in it. Most specifically you must never
307	reinitialise it or call its set method.	1012	reinitialise it or call its C<ev_TYPE_set> macro.
308
309	You can check whether an event is active by calling the C<ev_is_active
310	(watcher *)> macro. To see whether an event is outstanding (but the
311	callback for it has not been called yet) you can use the C<ev_is_pending
312	(watcher *)> macro.
313		1013
314	Each and every callback receives the event loop pointer as first, the	1014	Each and every callback receives the event loop pointer as first, the
315	registered watcher structure as second, and a bitset of received events as	1015	registered watcher structure as second, and a bitset of received events as
316	third argument.	1016	third argument.
317		1017
…		…
341	The signal specified in the C<ev_signal> watcher has been received by a thread.	1041	The signal specified in the C<ev_signal> watcher has been received by a thread.
342		1042
343	=item C<EV_CHILD>	1043	=item C<EV_CHILD>
344		1044
345	The pid specified in the C<ev_child> watcher has received a status change.	1045	The pid specified in the C<ev_child> watcher has received a status change.
		1046
		1047	=item C<EV_STAT>
		1048
		1049	The path specified in the C<ev_stat> watcher changed its attributes somehow.
346		1050
347	=item C<EV_IDLE>	1051	=item C<EV_IDLE>
348		1052
349	The C<ev_idle> watcher has determined that you have nothing better to do.	1053	The C<ev_idle> watcher has determined that you have nothing better to do.
350		1054
…		…
358	received events. Callbacks of both watcher types can start and stop as	1062	received events. Callbacks of both watcher types can start and stop as
359	many watchers as they want, and all of them will be taken into account	1063	many watchers as they want, and all of them will be taken into account
360	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1064	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
361	C<ev_loop> from blocking).	1065	C<ev_loop> from blocking).
362		1066
		1067	=item C<EV_EMBED>
		1068
		1069	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1070
		1071	=item C<EV_FORK>
		1072
		1073	The event loop has been resumed in the child process after fork (see
		1074	C<ev_fork>).
		1075
		1076	=item C<EV_ASYNC>
		1077
		1078	The given async watcher has been asynchronously notified (see C<ev_async>).
		1079
		1080	=item C<EV_CUSTOM>
		1081
		1082	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1083	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1084
363	=item C<EV_ERROR>	1085	=item C<EV_ERROR>
364		1086
365	An unspecified error has occured, the watcher has been stopped. This might	1087	An unspecified error has occurred, the watcher has been stopped. This might
366	happen because the watcher could not be properly started because libev	1088	happen because the watcher could not be properly started because libev
367	ran out of memory, a file descriptor was found to be closed or any other	1089	ran out of memory, a file descriptor was found to be closed or any other
		1090	problem. Libev considers these application bugs.
		1091
368	problem. You best act on it by reporting the problem and somehow coping	1092	You best act on it by reporting the problem and somehow coping with the
369	with the watcher being stopped.	1093	watcher being stopped. Note that well-written programs should not receive
		1094	an error ever, so when your watcher receives it, this usually indicates a
		1095	bug in your program.
370		1096
371	Libev will usually signal a few "dummy" events together with an error,	1097	Libev will usually signal a few "dummy" events together with an error, for
372	for example it might indicate that a fd is readable or writable, and if	1098	example it might indicate that a fd is readable or writable, and if your
373	your callbacks is well-written it can just attempt the operation and cope	1099	callbacks is well-written it can just attempt the operation and cope with
374	with the error from read() or write(). This will not work in multithreaded	1100	the error from read() or write(). This will not work in multi-threaded
375	programs, though, so beware.	1101	programs, though, as the fd could already be closed and reused for another
		1102	thing, so beware.
376		1103
377	=back	1104	=back
378		1105
		1106	=head2 GENERIC WATCHER FUNCTIONS
		1107
		1108	=over 4
		1109
		1110	=item C<ev_init> (ev_TYPE *watcher, callback)
		1111
		1112	This macro initialises the generic portion of a watcher. The contents
		1113	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1114	the generic parts of the watcher are initialised, you I<need> to call
		1115	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1116	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1117	which rolls both calls into one.
		1118
		1119	You can reinitialise a watcher at any time as long as it has been stopped
		1120	(or never started) and there are no pending events outstanding.
		1121
		1122	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1123	int revents)>.
		1124
		1125	Example: Initialise an C<ev_io> watcher in two steps.
		1126
		1127	ev_io w;
		1128	ev_init (&w, my_cb);
		1129	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1130
		1131	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1132
		1133	This macro initialises the type-specific parts of a watcher. You need to
		1134	call C<ev_init> at least once before you call this macro, but you can
		1135	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1136	macro on a watcher that is active (it can be pending, however, which is a
		1137	difference to the C<ev_init> macro).
		1138
		1139	Although some watcher types do not have type-specific arguments
		1140	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1141
		1142	See C<ev_init>, above, for an example.
		1143
		1144	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1145
		1146	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1147	calls into a single call. This is the most convenient method to initialise
		1148	a watcher. The same limitations apply, of course.
		1149
		1150	Example: Initialise and set an C<ev_io> watcher in one step.
		1151
		1152	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1153
		1154	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1155
		1156	Starts (activates) the given watcher. Only active watchers will receive
		1157	events. If the watcher is already active nothing will happen.
		1158
		1159	Example: Start the C<ev_io> watcher that is being abused as example in this
		1160	whole section.
		1161
		1162	ev_io_start (EV_DEFAULT_UC, &w);
		1163
		1164	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1165
		1166	Stops the given watcher if active, and clears the pending status (whether
		1167	the watcher was active or not).
		1168
		1169	It is possible that stopped watchers are pending - for example,
		1170	non-repeating timers are being stopped when they become pending - but
		1171	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1172	pending. If you want to free or reuse the memory used by the watcher it is
		1173	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1174
		1175	=item bool ev_is_active (ev_TYPE *watcher)
		1176
		1177	Returns a true value iff the watcher is active (i.e. it has been started
		1178	and not yet been stopped). As long as a watcher is active you must not modify
		1179	it.
		1180
		1181	=item bool ev_is_pending (ev_TYPE *watcher)
		1182
		1183	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1184	events but its callback has not yet been invoked). As long as a watcher
		1185	is pending (but not active) you must not call an init function on it (but
		1186	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1187	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1188	it).
		1189
		1190	=item callback ev_cb (ev_TYPE *watcher)
		1191
		1192	Returns the callback currently set on the watcher.
		1193
		1194	=item ev_cb_set (ev_TYPE *watcher, callback)
		1195
		1196	Change the callback. You can change the callback at virtually any time
		1197	(modulo threads).
		1198
		1199	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1200
		1201	=item int ev_priority (ev_TYPE *watcher)
		1202
		1203	Set and query the priority of the watcher. The priority is a small
		1204	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1205	(default: C<-2>). Pending watchers with higher priority will be invoked
		1206	before watchers with lower priority, but priority will not keep watchers
		1207	from being executed (except for C<ev_idle> watchers).
		1208
		1209	If you need to suppress invocation when higher priority events are pending
		1210	you need to look at C<ev_idle> watchers, which provide this functionality.
		1211
		1212	You I<must not> change the priority of a watcher as long as it is active or
		1213	pending.
		1214
		1215	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1216	fine, as long as you do not mind that the priority value you query might
		1217	or might not have been clamped to the valid range.
		1218
		1219	The default priority used by watchers when no priority has been set is
		1220	always C<0>, which is supposed to not be too high and not be too low :).
		1221
		1222	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1223	priorities.
		1224
		1225	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1226
		1227	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1228	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1229	can deal with that fact, as both are simply passed through to the
		1230	callback.
		1231
		1232	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1233
		1234	If the watcher is pending, this function clears its pending status and
		1235	returns its C<revents> bitset (as if its callback was invoked). If the
		1236	watcher isn't pending it does nothing and returns C<0>.
		1237
		1238	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1239	callback to be invoked, which can be accomplished with this function.
		1240
		1241	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1242
		1243	Feeds the given event set into the event loop, as if the specified event
		1244	had happened for the specified watcher (which must be a pointer to an
		1245	initialised but not necessarily started event watcher). Obviously you must
		1246	not free the watcher as long as it has pending events.
		1247
		1248	Stopping the watcher, letting libev invoke it, or calling
		1249	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1250	not started in the first place.
		1251
		1252	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1253	functions that do not need a watcher.
		1254
		1255	=back
		1256
		1257
379	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1258	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
380		1259
381	Each watcher has, by default, a member C<void *data> that you can change	1260	Each watcher has, by default, a member C<void *data> that you can change
382	and read at any time, libev will completely ignore it. This can be used	1261	and read at any time: libev will completely ignore it. This can be used
383	to associate arbitrary data with your watcher. If you need more data and	1262	to associate arbitrary data with your watcher. If you need more data and
384	don't want to allocate memory and store a pointer to it in that data	1263	don't want to allocate memory and store a pointer to it in that data
385	member, you can also "subclass" the watcher type and provide your own	1264	member, you can also "subclass" the watcher type and provide your own
386	data:	1265	data:
387		1266
388	struct my_io	1267	struct my_io
389	{	1268	{
390	struct ev_io io;	1269	ev_io io;
391	int otherfd;	1270	int otherfd;
392	void *somedata;	1271	void *somedata;
393	struct whatever *mostinteresting;	1272	struct whatever *mostinteresting;
394	}	1273	};
		1274
		1275	...
		1276	struct my_io w;
		1277	ev_io_init (&w.io, my_cb, fd, EV_READ);
395		1278
396	And since your callback will be called with a pointer to the watcher, you	1279	And since your callback will be called with a pointer to the watcher, you
397	can cast it back to your own type:	1280	can cast it back to your own type:
398		1281
399	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1282	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
400	{	1283	{
401	struct my_io *w = (struct my_io *)w_;	1284	struct my_io *w = (struct my_io *)w_;
402	...	1285	...
403	}	1286	}
404		1287
405	More interesting and less C-conformant ways of catsing your callback type	1288	More interesting and less C-conformant ways of casting your callback type
406	have been omitted....	1289	instead have been omitted.
		1290
		1291	Another common scenario is to use some data structure with multiple
		1292	embedded watchers:
		1293
		1294	struct my_biggy
		1295	{
		1296	int some_data;
		1297	ev_timer t1;
		1298	ev_timer t2;
		1299	}
		1300
		1301	In this case getting the pointer to C<my_biggy> is a bit more
		1302	complicated: Either you store the address of your C<my_biggy> struct
		1303	in the C<data> member of the watcher (for woozies), or you need to use
		1304	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1305	programmers):
		1306
		1307	#include <stddef.h>
		1308
		1309	static void
		1310	t1_cb (EV_P_ ev_timer *w, int revents)
		1311	{
		1312	struct my_biggy big = (struct my_biggy *)
		1313	(((char *)w) - offsetof (struct my_biggy, t1));
		1314	}
		1315
		1316	static void
		1317	t2_cb (EV_P_ ev_timer *w, int revents)
		1318	{
		1319	struct my_biggy big = (struct my_biggy *)
		1320	(((char *)w) - offsetof (struct my_biggy, t2));
		1321	}
		1322
		1323	=head2 WATCHER PRIORITY MODELS
		1324
		1325	Many event loops support I<watcher priorities>, which are usually small
		1326	integers that influence the ordering of event callback invocation
		1327	between watchers in some way, all else being equal.
		1328
		1329	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1330	description for the more technical details such as the actual priority
		1331	range.
		1332
		1333	There are two common ways how these these priorities are being interpreted
		1334	by event loops:
		1335
		1336	In the more common lock-out model, higher priorities "lock out" invocation
		1337	of lower priority watchers, which means as long as higher priority
		1338	watchers receive events, lower priority watchers are not being invoked.
		1339
		1340	The less common only-for-ordering model uses priorities solely to order
		1341	callback invocation within a single event loop iteration: Higher priority
		1342	watchers are invoked before lower priority ones, but they all get invoked
		1343	before polling for new events.
		1344
		1345	Libev uses the second (only-for-ordering) model for all its watchers
		1346	except for idle watchers (which use the lock-out model).
		1347
		1348	The rationale behind this is that implementing the lock-out model for
		1349	watchers is not well supported by most kernel interfaces, and most event
		1350	libraries will just poll for the same events again and again as long as
		1351	their callbacks have not been executed, which is very inefficient in the
		1352	common case of one high-priority watcher locking out a mass of lower
		1353	priority ones.
		1354
		1355	Static (ordering) priorities are most useful when you have two or more
		1356	watchers handling the same resource: a typical usage example is having an
		1357	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1358	timeouts. Under load, data might be received while the program handles
		1359	other jobs, but since timers normally get invoked first, the timeout
		1360	handler will be executed before checking for data. In that case, giving
		1361	the timer a lower priority than the I/O watcher ensures that I/O will be
		1362	handled first even under adverse conditions (which is usually, but not
		1363	always, what you want).
		1364
		1365	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1366	will only be executed when no same or higher priority watchers have
		1367	received events, they can be used to implement the "lock-out" model when
		1368	required.
		1369
		1370	For example, to emulate how many other event libraries handle priorities,
		1371	you can associate an C<ev_idle> watcher to each such watcher, and in
		1372	the normal watcher callback, you just start the idle watcher. The real
		1373	processing is done in the idle watcher callback. This causes libev to
		1374	continously poll and process kernel event data for the watcher, but when
		1375	the lock-out case is known to be rare (which in turn is rare :), this is
		1376	workable.
		1377
		1378	Usually, however, the lock-out model implemented that way will perform
		1379	miserably under the type of load it was designed to handle. In that case,
		1380	it might be preferable to stop the real watcher before starting the
		1381	idle watcher, so the kernel will not have to process the event in case
		1382	the actual processing will be delayed for considerable time.
		1383
		1384	Here is an example of an I/O watcher that should run at a strictly lower
		1385	priority than the default, and which should only process data when no
		1386	other events are pending:
		1387
		1388	ev_idle idle; // actual processing watcher
		1389	ev_io io; // actual event watcher
		1390
		1391	static void
		1392	io_cb (EV_P_ ev_io *w, int revents)
		1393	{
		1394	// stop the I/O watcher, we received the event, but
		1395	// are not yet ready to handle it.
		1396	ev_io_stop (EV_A_ w);
		1397
		1398	// start the idle watcher to ahndle the actual event.
		1399	// it will not be executed as long as other watchers
		1400	// with the default priority are receiving events.
		1401	ev_idle_start (EV_A_ &idle);
		1402	}
		1403
		1404	static void
		1405	idle_cb (EV_P_ ev_idle *w, int revents)
		1406	{
		1407	// actual processing
		1408	read (STDIN_FILENO, ...);
		1409
		1410	// have to start the I/O watcher again, as
		1411	// we have handled the event
		1412	ev_io_start (EV_P_ &io);
		1413	}
		1414
		1415	// initialisation
		1416	ev_idle_init (&idle, idle_cb);
		1417	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1418	ev_io_start (EV_DEFAULT_ &io);
		1419
		1420	In the "real" world, it might also be beneficial to start a timer, so that
		1421	low-priority connections can not be locked out forever under load. This
		1422	enables your program to keep a lower latency for important connections
		1423	during short periods of high load, while not completely locking out less
		1424	important ones.
407		1425
408		1426
409	=head1 WATCHER TYPES	1427	=head1 WATCHER TYPES
410		1428
411	This section describes each watcher in detail, but will not repeat	1429	This section describes each watcher in detail, but will not repeat
412	information given in the last section.	1430	information given in the last section. Any initialisation/set macros,
		1431	functions and members specific to the watcher type are explained.
413		1432
		1433	Members are additionally marked with either I<[read-only]>, meaning that,
		1434	while the watcher is active, you can look at the member and expect some
		1435	sensible content, but you must not modify it (you can modify it while the
		1436	watcher is stopped to your hearts content), or I<[read-write]>, which
		1437	means you can expect it to have some sensible content while the watcher
		1438	is active, but you can also modify it. Modifying it may not do something
		1439	sensible or take immediate effect (or do anything at all), but libev will
		1440	not crash or malfunction in any way.
		1441
		1442
414	=head2 C<ev_io> - is this file descriptor readable or writable	1443	=head2 C<ev_io> - is this file descriptor readable or writable?
415		1444
416	I/O watchers check whether a file descriptor is readable or writable	1445	I/O watchers check whether a file descriptor is readable or writable
417	in each iteration of the event loop (This behaviour is called	1446	in each iteration of the event loop, or, more precisely, when reading
418	level-triggering because you keep receiving events as long as the	1447	would not block the process and writing would at least be able to write
419	condition persists. Remember you can stop the watcher if you don't want to	1448	some data. This behaviour is called level-triggering because you keep
420	act on the event and neither want to receive future events).	1449	receiving events as long as the condition persists. Remember you can stop
		1450	the watcher if you don't want to act on the event and neither want to
		1451	receive future events.
421		1452
422	In general you can register as many read and/or write event watchers oer	1453	In general you can register as many read and/or write event watchers per
423	fd as you want (as long as you don't confuse yourself). Setting all file	1454	fd as you want (as long as you don't confuse yourself). Setting all file
424	descriptors to non-blocking mode is also usually a good idea (but not	1455	descriptors to non-blocking mode is also usually a good idea (but not
425	required if you know what you are doing).	1456	required if you know what you are doing).
426		1457
427	You have to be careful with dup'ed file descriptors, though. Some backends	1458	If you cannot use non-blocking mode, then force the use of a
428	(the linux epoll backend is a notable example) cannot handle dup'ed file	1459	known-to-be-good backend (at the time of this writing, this includes only
429	descriptors correctly if you register interest in two or more fds pointing	1460	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
430	to the same file/socket etc. description.	1461	descriptors for which non-blocking operation makes no sense (such as
		1462	files) - libev doesn't guarentee any specific behaviour in that case.
431		1463
432	If you must do this, then force the use of a known-to-be-good backend	1464	Another thing you have to watch out for is that it is quite easy to
433	(at the time of this writing, this includes only EVMETHOD_SELECT and	1465	receive "spurious" readiness notifications, that is your callback might
434	EVMETHOD_POLL).	1466	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1467	because there is no data. Not only are some backends known to create a
		1468	lot of those (for example Solaris ports), it is very easy to get into
		1469	this situation even with a relatively standard program structure. Thus
		1470	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1471	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1472
		1473	If you cannot run the fd in non-blocking mode (for example you should
		1474	not play around with an Xlib connection), then you have to separately
		1475	re-test whether a file descriptor is really ready with a known-to-be good
		1476	interface such as poll (fortunately in our Xlib example, Xlib already
		1477	does this on its own, so its quite safe to use). Some people additionally
		1478	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1479	indefinitely.
		1480
		1481	But really, best use non-blocking mode.
		1482
		1483	=head3 The special problem of disappearing file descriptors
		1484
		1485	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1486	descriptor (either due to calling C<close> explicitly or any other means,
		1487	such as C<dup2>). The reason is that you register interest in some file
		1488	descriptor, but when it goes away, the operating system will silently drop
		1489	this interest. If another file descriptor with the same number then is
		1490	registered with libev, there is no efficient way to see that this is, in
		1491	fact, a different file descriptor.
		1492
		1493	To avoid having to explicitly tell libev about such cases, libev follows
		1494	the following policy: Each time C<ev_io_set> is being called, libev
		1495	will assume that this is potentially a new file descriptor, otherwise
		1496	it is assumed that the file descriptor stays the same. That means that
		1497	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1498	descriptor even if the file descriptor number itself did not change.
		1499
		1500	This is how one would do it normally anyway, the important point is that
		1501	the libev application should not optimise around libev but should leave
		1502	optimisations to libev.
		1503
		1504	=head3 The special problem of dup'ed file descriptors
		1505
		1506	Some backends (e.g. epoll), cannot register events for file descriptors,
		1507	but only events for the underlying file descriptions. That means when you
		1508	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1509	events for them, only one file descriptor might actually receive events.
		1510
		1511	There is no workaround possible except not registering events
		1512	for potentially C<dup ()>'ed file descriptors, or to resort to
		1513	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1514
		1515	=head3 The special problem of fork
		1516
		1517	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1518	useless behaviour. Libev fully supports fork, but needs to be told about
		1519	it in the child.
		1520
		1521	To support fork in your programs, you either have to call
		1522	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1523	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1524	C<EVBACKEND_POLL>.
		1525
		1526	=head3 The special problem of SIGPIPE
		1527
		1528	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1529	when writing to a pipe whose other end has been closed, your program gets
		1530	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1531	this is sensible behaviour, for daemons, this is usually undesirable.
		1532
		1533	So when you encounter spurious, unexplained daemon exits, make sure you
		1534	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1535	somewhere, as that would have given you a big clue).
		1536
		1537
		1538	=head3 Watcher-Specific Functions
435		1539
436	=over 4	1540	=over 4
437		1541
438	=item ev_io_init (ev_io *, callback, int fd, int events)	1542	=item ev_io_init (ev_io *, callback, int fd, int events)
439		1543
440	=item ev_io_set (ev_io *, int fd, int events)	1544	=item ev_io_set (ev_io *, int fd, int events)
441		1545
442	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1546	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
443	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1547	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
444	EV_WRITE> to receive the given events.	1548	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1549
		1550	=item int fd [read-only]
		1551
		1552	The file descriptor being watched.
		1553
		1554	=item int events [read-only]
		1555
		1556	The events being watched.
445		1557
446	=back	1558	=back
447		1559
		1560	=head3 Examples
		1561
		1562	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1563	readable, but only once. Since it is likely line-buffered, you could
		1564	attempt to read a whole line in the callback.
		1565
		1566	static void
		1567	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1568	{
		1569	ev_io_stop (loop, w);
		1570	.. read from stdin here (or from w->fd) and handle any I/O errors
		1571	}
		1572
		1573	...
		1574	struct ev_loop *loop = ev_default_init (0);
		1575	ev_io stdin_readable;
		1576	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1577	ev_io_start (loop, &stdin_readable);
		1578	ev_loop (loop, 0);
		1579
		1580
448	=head2 C<ev_timer> - relative and optionally recurring timeouts	1581	=head2 C<ev_timer> - relative and optionally repeating timeouts
449		1582
450	Timer watchers are simple relative timers that generate an event after a	1583	Timer watchers are simple relative timers that generate an event after a
451	given time, and optionally repeating in regular intervals after that.	1584	given time, and optionally repeating in regular intervals after that.
452		1585
453	The timers are based on real time, that is, if you register an event that	1586	The timers are based on real time, that is, if you register an event that
454	times out after an hour and youreset your system clock to last years	1587	times out after an hour and you reset your system clock to January last
455	time, it will still time out after (roughly) and hour. "Roughly" because	1588	year, it will still time out after (roughly) one hour. "Roughly" because
456	detecting time jumps is hard, and soem inaccuracies are unavoidable (the	1589	detecting time jumps is hard, and some inaccuracies are unavoidable (the
457	monotonic clock option helps a lot here).	1590	monotonic clock option helps a lot here).
		1591
		1592	The callback is guaranteed to be invoked only I<after> its timeout has
		1593	passed (not I<at>, so on systems with very low-resolution clocks this
		1594	might introduce a small delay). If multiple timers become ready during the
		1595	same loop iteration then the ones with earlier time-out values are invoked
		1596	before ones of the same priority with later time-out values (but this is
		1597	no longer true when a callback calls C<ev_loop> recursively).
		1598
		1599	=head3 Be smart about timeouts
		1600
		1601	Many real-world problems involve some kind of timeout, usually for error
		1602	recovery. A typical example is an HTTP request - if the other side hangs,
		1603	you want to raise some error after a while.
		1604
		1605	What follows are some ways to handle this problem, from obvious and
		1606	inefficient to smart and efficient.
		1607
		1608	In the following, a 60 second activity timeout is assumed - a timeout that
		1609	gets reset to 60 seconds each time there is activity (e.g. each time some
		1610	data or other life sign was received).
		1611
		1612	=over 4
		1613
		1614	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1615
		1616	This is the most obvious, but not the most simple way: In the beginning,
		1617	start the watcher:
		1618
		1619	ev_timer_init (timer, callback, 60., 0.);
		1620	ev_timer_start (loop, timer);
		1621
		1622	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1623	and start it again:
		1624
		1625	ev_timer_stop (loop, timer);
		1626	ev_timer_set (timer, 60., 0.);
		1627	ev_timer_start (loop, timer);
		1628
		1629	This is relatively simple to implement, but means that each time there is
		1630	some activity, libev will first have to remove the timer from its internal
		1631	data structure and then add it again. Libev tries to be fast, but it's
		1632	still not a constant-time operation.
		1633
		1634	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1635
		1636	This is the easiest way, and involves using C<ev_timer_again> instead of
		1637	C<ev_timer_start>.
		1638
		1639	To implement this, configure an C<ev_timer> with a C<repeat> value
		1640	of C<60> and then call C<ev_timer_again> at start and each time you
		1641	successfully read or write some data. If you go into an idle state where
		1642	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1643	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1644
		1645	That means you can ignore both the C<ev_timer_start> function and the
		1646	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1647	member and C<ev_timer_again>.
		1648
		1649	At start:
		1650
		1651	ev_init (timer, callback);
		1652	timer->repeat = 60.;
		1653	ev_timer_again (loop, timer);
		1654
		1655	Each time there is some activity:
		1656
		1657	ev_timer_again (loop, timer);
		1658
		1659	It is even possible to change the time-out on the fly, regardless of
		1660	whether the watcher is active or not:
		1661
		1662	timer->repeat = 30.;
		1663	ev_timer_again (loop, timer);
		1664
		1665	This is slightly more efficient then stopping/starting the timer each time
		1666	you want to modify its timeout value, as libev does not have to completely
		1667	remove and re-insert the timer from/into its internal data structure.
		1668
		1669	It is, however, even simpler than the "obvious" way to do it.
		1670
		1671	=item 3. Let the timer time out, but then re-arm it as required.
		1672
		1673	This method is more tricky, but usually most efficient: Most timeouts are
		1674	relatively long compared to the intervals between other activity - in
		1675	our example, within 60 seconds, there are usually many I/O events with
		1676	associated activity resets.
		1677
		1678	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1679	but remember the time of last activity, and check for a real timeout only
		1680	within the callback:
		1681
		1682	ev_tstamp last_activity; // time of last activity
		1683
		1684	static void
		1685	callback (EV_P_ ev_timer *w, int revents)
		1686	{
		1687	ev_tstamp now = ev_now (EV_A);
		1688	ev_tstamp timeout = last_activity + 60.;
		1689
		1690	// if last_activity + 60. is older than now, we did time out
		1691	if (timeout < now)
		1692	{
		1693	// timeout occured, take action
		1694	}
		1695	else
		1696	{
		1697	// callback was invoked, but there was some activity, re-arm
		1698	// the watcher to fire in last_activity + 60, which is
		1699	// guaranteed to be in the future, so "again" is positive:
		1700	w->repeat = timeout - now;
		1701	ev_timer_again (EV_A_ w);
		1702	}
		1703	}
		1704
		1705	To summarise the callback: first calculate the real timeout (defined
		1706	as "60 seconds after the last activity"), then check if that time has
		1707	been reached, which means something I<did>, in fact, time out. Otherwise
		1708	the callback was invoked too early (C<timeout> is in the future), so
		1709	re-schedule the timer to fire at that future time, to see if maybe we have
		1710	a timeout then.
		1711
		1712	Note how C<ev_timer_again> is used, taking advantage of the
		1713	C<ev_timer_again> optimisation when the timer is already running.
		1714
		1715	This scheme causes more callback invocations (about one every 60 seconds
		1716	minus half the average time between activity), but virtually no calls to
		1717	libev to change the timeout.
		1718
		1719	To start the timer, simply initialise the watcher and set C<last_activity>
		1720	to the current time (meaning we just have some activity :), then call the
		1721	callback, which will "do the right thing" and start the timer:
		1722
		1723	ev_init (timer, callback);
		1724	last_activity = ev_now (loop);
		1725	callback (loop, timer, EV_TIMEOUT);
		1726
		1727	And when there is some activity, simply store the current time in
		1728	C<last_activity>, no libev calls at all:
		1729
		1730	last_actiivty = ev_now (loop);
		1731
		1732	This technique is slightly more complex, but in most cases where the
		1733	time-out is unlikely to be triggered, much more efficient.
		1734
		1735	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1736	callback :) - just change the timeout and invoke the callback, which will
		1737	fix things for you.
		1738
		1739	=item 4. Wee, just use a double-linked list for your timeouts.
		1740
		1741	If there is not one request, but many thousands (millions...), all
		1742	employing some kind of timeout with the same timeout value, then one can
		1743	do even better:
		1744
		1745	When starting the timeout, calculate the timeout value and put the timeout
		1746	at the I<end> of the list.
		1747
		1748	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1749	the list is expected to fire (for example, using the technique #3).
		1750
		1751	When there is some activity, remove the timer from the list, recalculate
		1752	the timeout, append it to the end of the list again, and make sure to
		1753	update the C<ev_timer> if it was taken from the beginning of the list.
		1754
		1755	This way, one can manage an unlimited number of timeouts in O(1) time for
		1756	starting, stopping and updating the timers, at the expense of a major
		1757	complication, and having to use a constant timeout. The constant timeout
		1758	ensures that the list stays sorted.
		1759
		1760	=back
		1761
		1762	So which method the best?
		1763
		1764	Method #2 is a simple no-brain-required solution that is adequate in most
		1765	situations. Method #3 requires a bit more thinking, but handles many cases
		1766	better, and isn't very complicated either. In most case, choosing either
		1767	one is fine, with #3 being better in typical situations.
		1768
		1769	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1770	rather complicated, but extremely efficient, something that really pays
		1771	off after the first million or so of active timers, i.e. it's usually
		1772	overkill :)
		1773
		1774	=head3 The special problem of time updates
		1775
		1776	Establishing the current time is a costly operation (it usually takes at
		1777	least two system calls): EV therefore updates its idea of the current
		1778	time only before and after C<ev_loop> collects new events, which causes a
		1779	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1780	lots of events in one iteration.
458		1781
459	The relative timeouts are calculated relative to the C<ev_now ()>	1782	The relative timeouts are calculated relative to the C<ev_now ()>
460	time. This is usually the right thing as this timestamp refers to the time	1783	time. This is usually the right thing as this timestamp refers to the time
461	of the event triggering whatever timeout you are modifying/starting. If	1784	of the event triggering whatever timeout you are modifying/starting. If
462	you suspect event processing to be delayed and you *need* to base the timeout	1785	you suspect event processing to be delayed and you I<need> to base the
463	ion the current time, use something like this to adjust for this:	1786	timeout on the current time, use something like this to adjust for this:
464		1787
465	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1788	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
466		1789
		1790	If the event loop is suspended for a long time, you can also force an
		1791	update of the time returned by C<ev_now ()> by calling C<ev_now_update
		1792	()>.
		1793
		1794	=head3 The special problems of suspended animation
		1795
		1796	When you leave the server world it is quite customary to hit machines that
		1797	can suspend/hibernate - what happens to the clocks during such a suspend?
		1798
		1799	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		1800	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		1801	to run until the system is suspended, but they will not advance while the
		1802	system is suspended. That means, on resume, it will be as if the program
		1803	was frozen for a few seconds, but the suspend time will not be counted
		1804	towards C<ev_timer> when a monotonic clock source is used. The real time
		1805	clock advanced as expected, but if it is used as sole clocksource, then a
		1806	long suspend would be detected as a time jump by libev, and timers would
		1807	be adjusted accordingly.
		1808
		1809	I would not be surprised to see different behaviour in different between
		1810	operating systems, OS versions or even different hardware.
		1811
		1812	The other form of suspend (job control, or sending a SIGSTOP) will see a
		1813	time jump in the monotonic clocks and the realtime clock. If the program
		1814	is suspended for a very long time, and monotonic clock sources are in use,
		1815	then you can expect C<ev_timer>s to expire as the full suspension time
		1816	will be counted towards the timers. When no monotonic clock source is in
		1817	use, then libev will again assume a timejump and adjust accordingly.
		1818
		1819	It might be beneficial for this latter case to call C<ev_suspend>
		1820	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		1821	deterministic behaviour in this case (you can do nothing against
		1822	C<SIGSTOP>).
		1823
		1824	=head3 Watcher-Specific Functions and Data Members
		1825
467	=over 4	1826	=over 4
468		1827
469	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1828	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
470		1829
471	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1830	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
472		1831
473	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1832	Configure the timer to trigger after C<after> seconds. If C<repeat>
474	C<0.>, then it will automatically be stopped. If it is positive, then the	1833	is C<0.>, then it will automatically be stopped once the timeout is
475	timer will automatically be configured to trigger again C<repeat> seconds	1834	reached. If it is positive, then the timer will automatically be
476	later, again, and again, until stopped manually.	1835	configured to trigger again C<repeat> seconds later, again, and again,
		1836	until stopped manually.
477		1837
478	The timer itself will do a best-effort at avoiding drift, that is, if you	1838	The timer itself will do a best-effort at avoiding drift, that is, if
479	configure a timer to trigger every 10 seconds, then it will trigger at	1839	you configure a timer to trigger every 10 seconds, then it will normally
480	exactly 10 second intervals. If, however, your program cannot keep up with	1840	trigger at exactly 10 second intervals. If, however, your program cannot
481	the timer (ecause it takes longer than those 10 seconds to do stuff) the	1841	keep up with the timer (because it takes longer than those 10 seconds to
482	timer will not fire more than once per event loop iteration.	1842	do stuff) the timer will not fire more than once per event loop iteration.
483		1843
484	=item ev_timer_again (loop)	1844	=item ev_timer_again (loop, ev_timer *)
485		1845
486	This will act as if the timer timed out and restart it again if it is	1846	This will act as if the timer timed out and restart it again if it is
487	repeating. The exact semantics are:	1847	repeating. The exact semantics are:
488		1848
		1849	If the timer is pending, its pending status is cleared.
		1850
489	If the timer is started but nonrepeating, stop it.	1851	If the timer is started but non-repeating, stop it (as if it timed out).
490		1852
491	If the timer is repeating, either start it if necessary (with the repeat	1853	If the timer is repeating, either start it if necessary (with the
492	value), or reset the running timer to the repeat value.	1854	C<repeat> value), or reset the running timer to the C<repeat> value.
493		1855
494	This sounds a bit complicated, but here is a useful and typical	1856	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
495	example: Imagine you have a tcp connection and you want a so-called idle	1857	usage example.
496	timeout, that is, you want to be called when there have been, say, 60	1858
497	seconds of inactivity on the socket. The easiest way to do this is to	1859	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
498	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each	1860
499	time you successfully read or write some data. If you go into an idle	1861	Returns the remaining time until a timer fires. If the timer is active,
500	state where you do not expect data to travel on the socket, you can stop	1862	then this time is relative to the current event loop time, otherwise it's
501	the timer, and again will automatically restart it if need be.	1863	the timeout value currently configured.
		1864
		1865	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		1866	C<5>. When the timer is started and one second passes, C<ev_timer_remain>
		1867	will return C<4>. When the timer expires and is restarted, it will return
		1868	roughly C<7> (likely slightly less as callback invocation takes some time,
		1869	too), and so on.
		1870
		1871	=item ev_tstamp repeat [read-write]
		1872
		1873	The current C<repeat> value. Will be used each time the watcher times out
		1874	or C<ev_timer_again> is called, and determines the next timeout (if any),
		1875	which is also when any modifications are taken into account.
502		1876
503	=back	1877	=back
504		1878
		1879	=head3 Examples
		1880
		1881	Example: Create a timer that fires after 60 seconds.
		1882
		1883	static void
		1884	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1885	{
		1886	.. one minute over, w is actually stopped right here
		1887	}
		1888
		1889	ev_timer mytimer;
		1890	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		1891	ev_timer_start (loop, &mytimer);
		1892
		1893	Example: Create a timeout timer that times out after 10 seconds of
		1894	inactivity.
		1895
		1896	static void
		1897	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1898	{
		1899	.. ten seconds without any activity
		1900	}
		1901
		1902	ev_timer mytimer;
		1903	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		1904	ev_timer_again (&mytimer); /* start timer */
		1905	ev_loop (loop, 0);
		1906
		1907	// and in some piece of code that gets executed on any "activity":
		1908	// reset the timeout to start ticking again at 10 seconds
		1909	ev_timer_again (&mytimer);
		1910
		1911
505	=head2 C<ev_periodic> - to cron or not to cron	1912	=head2 C<ev_periodic> - to cron or not to cron?
506		1913
507	Periodic watchers are also timers of a kind, but they are very versatile	1914	Periodic watchers are also timers of a kind, but they are very versatile
508	(and unfortunately a bit complex).	1915	(and unfortunately a bit complex).
509		1916
510	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1917	Unlike C<ev_timer>, periodic watchers are not based on real time (or
511	but on wallclock time (absolute time). You can tell a periodic watcher	1918	relative time, the physical time that passes) but on wall clock time
512	to trigger "at" some specific point in time. For example, if you tell a	1919	(absolute time, the thing you can read on your calender or clock). The
513	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	1920	difference is that wall clock time can run faster or slower than real
514	+ 10.>) and then reset your system clock to the last year, then it will	1921	time, and time jumps are not uncommon (e.g. when you adjust your
515	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1922	wrist-watch).
516	roughly 10 seconds later and of course not if you reset your system time
517	again).
518		1923
519	They can also be used to implement vastly more complex timers, such as	1924	You can tell a periodic watcher to trigger after some specific point
		1925	in time: for example, if you tell a periodic watcher to trigger "in 10
		1926	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		1927	not a delay) and then reset your system clock to January of the previous
		1928	year, then it will take a year or more to trigger the event (unlike an
		1929	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		1930	it, as it uses a relative timeout).
		1931
		1932	C<ev_periodic> watchers can also be used to implement vastly more complex
520	triggering an event on eahc midnight, local time.	1933	timers, such as triggering an event on each "midnight, local time", or
		1934	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		1935	those cannot react to time jumps.
		1936
		1937	As with timers, the callback is guaranteed to be invoked only when the
		1938	point in time where it is supposed to trigger has passed. If multiple
		1939	timers become ready during the same loop iteration then the ones with
		1940	earlier time-out values are invoked before ones with later time-out values
		1941	(but this is no longer true when a callback calls C<ev_loop> recursively).
		1942
		1943	=head3 Watcher-Specific Functions and Data Members
521		1944
522	=over 4	1945	=over 4
523		1946
524	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1947	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
525		1948
526	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1949	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
527		1950
528	Lots of arguments, lets sort it out... There are basically three modes of	1951	Lots of arguments, let's sort it out... There are basically three modes of
529	operation, and we will explain them from simplest to complex:	1952	operation, and we will explain them from simplest to most complex:
530
531		1953
532	=over 4	1954	=over 4
533		1955
534	=item * absolute timer (interval = reschedule_cb = 0)	1956	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
535		1957
536	In this configuration the watcher triggers an event at the wallclock time	1958	In this configuration the watcher triggers an event after the wall clock
537	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1959	time C<offset> has passed. It will not repeat and will not adjust when a
538	that is, if it is to be run at January 1st 2011 then it will run when the	1960	time jump occurs, that is, if it is to be run at January 1st 2011 then it
539	system time reaches or surpasses this time.	1961	will be stopped and invoked when the system clock reaches or surpasses
		1962	this point in time.
540		1963
541	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1964	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
542		1965
543	In this mode the watcher will always be scheduled to time out at the next	1966	In this mode the watcher will always be scheduled to time out at the next
544	C<at + N * interval> time (for some integer N) and then repeat, regardless	1967	C<offset + N * interval> time (for some integer N, which can also be
545	of any time jumps.	1968	negative) and then repeat, regardless of any time jumps. The C<offset>
		1969	argument is merely an offset into the C<interval> periods.
546		1970
547	This can be used to create timers that do not drift with respect to system	1971	This can be used to create timers that do not drift with respect to the
548	time:	1972	system clock, for example, here is an C<ev_periodic> that triggers each
		1973	hour, on the hour (with respect to UTC):
549		1974
550	ev_periodic_set (&periodic, 0., 3600., 0);	1975	ev_periodic_set (&periodic, 0., 3600., 0);
551		1976
552	This doesn't mean there will always be 3600 seconds in between triggers,	1977	This doesn't mean there will always be 3600 seconds in between triggers,
553	but only that the the callback will be called when the system time shows a	1978	but only that the callback will be called when the system time shows a
554	full hour (UTC), or more correctly, when the system time is evenly divisible	1979	full hour (UTC), or more correctly, when the system time is evenly divisible
555	by 3600.	1980	by 3600.
556		1981
557	Another way to think about it (for the mathematically inclined) is that	1982	Another way to think about it (for the mathematically inclined) is that
558	C<ev_periodic> will try to run the callback in this mode at the next possible	1983	C<ev_periodic> will try to run the callback in this mode at the next possible
559	time where C<time = at (mod interval)>, regardless of any time jumps.	1984	time where C<time = offset (mod interval)>, regardless of any time jumps.
560		1985
		1986	For numerical stability it is preferable that the C<offset> value is near
		1987	C<ev_now ()> (the current time), but there is no range requirement for
		1988	this value, and in fact is often specified as zero.
		1989
		1990	Note also that there is an upper limit to how often a timer can fire (CPU
		1991	speed for example), so if C<interval> is very small then timing stability
		1992	will of course deteriorate. Libev itself tries to be exact to be about one
		1993	millisecond (if the OS supports it and the machine is fast enough).
		1994
561	=item * manual reschedule mode (reschedule_cb = callback)	1995	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
562		1996
563	In this mode the values for C<interval> and C<at> are both being	1997	In this mode the values for C<interval> and C<offset> are both being
564	ignored. Instead, each time the periodic watcher gets scheduled, the	1998	ignored. Instead, each time the periodic watcher gets scheduled, the
565	reschedule callback will be called with the watcher as first, and the	1999	reschedule callback will be called with the watcher as first, and the
566	current time as second argument.	2000	current time as second argument.
567		2001
568	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2002	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
569	ever, or make any event loop modifications>. If you need to stop it,	2003	or make ANY other event loop modifications whatsoever, unless explicitly
570	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2004	allowed by documentation here>.
571	starting a prepare watcher).
572		2005
		2006	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2007	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2008	only event loop modification you are allowed to do).
		2009
573	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2010	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
574	ev_tstamp now)>, e.g.:	2011	*w, ev_tstamp now)>, e.g.:
575		2012
		2013	static ev_tstamp
576	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2014	my_rescheduler (ev_periodic *w, ev_tstamp now)
577	{	2015	{
578	return now + 60.;	2016	return now + 60.;
579	}	2017	}
580		2018
581	It must return the next time to trigger, based on the passed time value	2019	It must return the next time to trigger, based on the passed time value
582	(that is, the lowest time value larger than to the second argument). It	2020	(that is, the lowest time value larger than to the second argument). It
583	will usually be called just before the callback will be triggered, but	2021	will usually be called just before the callback will be triggered, but
584	might be called at other times, too.	2022	might be called at other times, too.
585		2023
586	NOTE: I<< This callback must always return a time that is later than the	2024	NOTE: I<< This callback must always return a time that is higher than or
587	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2025	equal to the passed C<now> value >>.
588		2026
589	This can be used to create very complex timers, such as a timer that	2027	This can be used to create very complex timers, such as a timer that
590	triggers on each midnight, local time. To do this, you would calculate the	2028	triggers on "next midnight, local time". To do this, you would calculate the
591	next midnight after C<now> and return the timestamp value for this. How	2029	next midnight after C<now> and return the timestamp value for this. How
592	you do this is, again, up to you (but it is not trivial, which is the main	2030	you do this is, again, up to you (but it is not trivial, which is the main
593	reason I omitted it as an example).	2031	reason I omitted it as an example).
594		2032
595	=back	2033	=back
…		…
599	Simply stops and restarts the periodic watcher again. This is only useful	2037	Simply stops and restarts the periodic watcher again. This is only useful
600	when you changed some parameters or the reschedule callback would return	2038	when you changed some parameters or the reschedule callback would return
601	a different time than the last time it was called (e.g. in a crond like	2039	a different time than the last time it was called (e.g. in a crond like
602	program when the crontabs have changed).	2040	program when the crontabs have changed).
603		2041
		2042	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2043
		2044	When active, returns the absolute time that the watcher is supposed
		2045	to trigger next. This is not the same as the C<offset> argument to
		2046	C<ev_periodic_set>, but indeed works even in interval and manual
		2047	rescheduling modes.
		2048
		2049	=item ev_tstamp offset [read-write]
		2050
		2051	When repeating, this contains the offset value, otherwise this is the
		2052	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2053	although libev might modify this value for better numerical stability).
		2054
		2055	Can be modified any time, but changes only take effect when the periodic
		2056	timer fires or C<ev_periodic_again> is being called.
		2057
		2058	=item ev_tstamp interval [read-write]
		2059
		2060	The current interval value. Can be modified any time, but changes only
		2061	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2062	called.
		2063
		2064	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2065
		2066	The current reschedule callback, or C<0>, if this functionality is
		2067	switched off. Can be changed any time, but changes only take effect when
		2068	the periodic timer fires or C<ev_periodic_again> is being called.
		2069
604	=back	2070	=back
605		2071
		2072	=head3 Examples
		2073
		2074	Example: Call a callback every hour, or, more precisely, whenever the
		2075	system time is divisible by 3600. The callback invocation times have
		2076	potentially a lot of jitter, but good long-term stability.
		2077
		2078	static void
		2079	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
		2080	{
		2081	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2082	}
		2083
		2084	ev_periodic hourly_tick;
		2085	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2086	ev_periodic_start (loop, &hourly_tick);
		2087
		2088	Example: The same as above, but use a reschedule callback to do it:
		2089
		2090	#include <math.h>
		2091
		2092	static ev_tstamp
		2093	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2094	{
		2095	return now + (3600. - fmod (now, 3600.));
		2096	}
		2097
		2098	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2099
		2100	Example: Call a callback every hour, starting now:
		2101
		2102	ev_periodic hourly_tick;
		2103	ev_periodic_init (&hourly_tick, clock_cb,
		2104	fmod (ev_now (loop), 3600.), 3600., 0);
		2105	ev_periodic_start (loop, &hourly_tick);
		2106
		2107
606	=head2 C<ev_signal> - signal me when a signal gets signalled	2108	=head2 C<ev_signal> - signal me when a signal gets signalled!
607		2109
608	Signal watchers will trigger an event when the process receives a specific	2110	Signal watchers will trigger an event when the process receives a specific
609	signal one or more times. Even though signals are very asynchronous, libev	2111	signal one or more times. Even though signals are very asynchronous, libev
610	will try it's best to deliver signals synchronously, i.e. as part of the	2112	will try it's best to deliver signals synchronously, i.e. as part of the
611	normal event processing, like any other event.	2113	normal event processing, like any other event.
612		2114
		2115	If you want signals to be delivered truly asynchronously, just use
		2116	C<sigaction> as you would do without libev and forget about sharing
		2117	the signal. You can even use C<ev_async> from a signal handler to
		2118	synchronously wake up an event loop.
		2119
613	You can configure as many watchers as you like per signal. Only when the	2120	You can configure as many watchers as you like for the same signal, but
		2121	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2122	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2123	C<SIGINT> in both the default loop and another loop at the same time. At
		2124	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2125
614	first watcher gets started will libev actually register a signal watcher	2126	When the first watcher gets started will libev actually register something
615	with the kernel (thus it coexists with your own signal handlers as long	2127	with the kernel (thus it coexists with your own signal handlers as long as
616	as you don't register any with libev). Similarly, when the last signal	2128	you don't register any with libev for the same signal).
617	watcher for a signal is stopped libev will reset the signal handler to	2129
618	SIG_DFL (regardless of what it was set to before).	2130	If possible and supported, libev will install its handlers with
		2131	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2132	not be unduly interrupted. If you have a problem with system calls getting
		2133	interrupted by signals you can block all signals in an C<ev_check> watcher
		2134	and unblock them in an C<ev_prepare> watcher.
		2135
		2136	=head3 The special problem of inheritance over execve
		2137
		2138	Both the signal mask (C<sigprocmask>) and the signal disposition
		2139	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2140	stopping it again), that is, libev might or might not block the signal,
		2141	and might or might not set or restore the installed signal handler.
		2142
		2143	While this does not matter for the signal disposition (libev never
		2144	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2145	C<execve>), this matters for the signal mask: many programs do not expect
		2146	certain signals to be blocked.
		2147
		2148	This means that before calling C<exec> (from the child) you should reset
		2149	the signal mask to whatever "default" you expect (all clear is a good
		2150	choice usually).
		2151
		2152	The simplest way to ensure that the signal mask is reset in the child is
		2153	to install a fork handler with C<pthread_atfork> that resets it. That will
		2154	catch fork calls done by libraries (such as the libc) as well.
		2155
		2156	In current versions of libev, you can also ensure that the signal mask is
		2157	not blocking any signals (except temporarily, so thread users watch out)
		2158	by specifying the C<EVFLAG_NOSIGFD> when creating the event loop. This
		2159	is not guaranteed for future versions, however.
		2160
		2161	=head3 Watcher-Specific Functions and Data Members
619		2162
620	=over 4	2163	=over 4
621		2164
622	=item ev_signal_init (ev_signal *, callback, int signum)	2165	=item ev_signal_init (ev_signal *, callback, int signum)
623		2166
624	=item ev_signal_set (ev_signal *, int signum)	2167	=item ev_signal_set (ev_signal *, int signum)
625		2168
626	Configures the watcher to trigger on the given signal number (usually one	2169	Configures the watcher to trigger on the given signal number (usually one
627	of the C<SIGxxx> constants).	2170	of the C<SIGxxx> constants).
628		2171
		2172	=item int signum [read-only]
		2173
		2174	The signal the watcher watches out for.
		2175
629	=back	2176	=back
630		2177
		2178	=head3 Examples
		2179
		2180	Example: Try to exit cleanly on SIGINT.
		2181
		2182	static void
		2183	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2184	{
		2185	ev_unloop (loop, EVUNLOOP_ALL);
		2186	}
		2187
		2188	ev_signal signal_watcher;
		2189	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2190	ev_signal_start (loop, &signal_watcher);
		2191
		2192
631	=head2 C<ev_child> - wait for pid status changes	2193	=head2 C<ev_child> - watch out for process status changes
632		2194
633	Child watchers trigger when your process receives a SIGCHLD in response to	2195	Child watchers trigger when your process receives a SIGCHLD in response to
634	some child status changes (most typically when a child of yours dies).	2196	some child status changes (most typically when a child of yours dies or
		2197	exits). It is permissible to install a child watcher I<after> the child
		2198	has been forked (which implies it might have already exited), as long
		2199	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2200	forking and then immediately registering a watcher for the child is fine,
		2201	but forking and registering a watcher a few event loop iterations later or
		2202	in the next callback invocation is not.
		2203
		2204	Only the default event loop is capable of handling signals, and therefore
		2205	you can only register child watchers in the default event loop.
		2206
		2207	Due to some design glitches inside libev, child watchers will always be
		2208	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2209	libev)
		2210
		2211	=head3 Process Interaction
		2212
		2213	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2214	initialised. This is necessary to guarantee proper behaviour even if the
		2215	first child watcher is started after the child exits. The occurrence
		2216	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2217	synchronously as part of the event loop processing. Libev always reaps all
		2218	children, even ones not watched.
		2219
		2220	=head3 Overriding the Built-In Processing
		2221
		2222	Libev offers no special support for overriding the built-in child
		2223	processing, but if your application collides with libev's default child
		2224	handler, you can override it easily by installing your own handler for
		2225	C<SIGCHLD> after initialising the default loop, and making sure the
		2226	default loop never gets destroyed. You are encouraged, however, to use an
		2227	event-based approach to child reaping and thus use libev's support for
		2228	that, so other libev users can use C<ev_child> watchers freely.
		2229
		2230	=head3 Stopping the Child Watcher
		2231
		2232	Currently, the child watcher never gets stopped, even when the
		2233	child terminates, so normally one needs to stop the watcher in the
		2234	callback. Future versions of libev might stop the watcher automatically
		2235	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2236	problem).
		2237
		2238	=head3 Watcher-Specific Functions and Data Members
635		2239
636	=over 4	2240	=over 4
637		2241
638	=item ev_child_init (ev_child *, callback, int pid)	2242	=item ev_child_init (ev_child *, callback, int pid, int trace)
639		2243
640	=item ev_child_set (ev_child *, int pid)	2244	=item ev_child_set (ev_child *, int pid, int trace)
641		2245
642	Configures the watcher to wait for status changes of process C<pid> (or	2246	Configures the watcher to wait for status changes of process C<pid> (or
643	I<any> process if C<pid> is specified as C<0>). The callback can look	2247	I<any> process if C<pid> is specified as C<0>). The callback can look
644	at the C<rstatus> member of the C<ev_child> watcher structure to see	2248	at the C<rstatus> member of the C<ev_child> watcher structure to see
645	the status word (use the macros from C<sys/wait.h> and see your systems	2249	the status word (use the macros from C<sys/wait.h> and see your systems
646	C<waitpid> documentation). The C<rpid> member contains the pid of the	2250	C<waitpid> documentation). The C<rpid> member contains the pid of the
647	process causing the status change.	2251	process causing the status change. C<trace> must be either C<0> (only
		2252	activate the watcher when the process terminates) or C<1> (additionally
		2253	activate the watcher when the process is stopped or continued).
		2254
		2255	=item int pid [read-only]
		2256
		2257	The process id this watcher watches out for, or C<0>, meaning any process id.
		2258
		2259	=item int rpid [read-write]
		2260
		2261	The process id that detected a status change.
		2262
		2263	=item int rstatus [read-write]
		2264
		2265	The process exit/trace status caused by C<rpid> (see your systems
		2266	C<waitpid> and C<sys/wait.h> documentation for details).
648		2267
649	=back	2268	=back
650		2269
		2270	=head3 Examples
		2271
		2272	Example: C<fork()> a new process and install a child handler to wait for
		2273	its completion.
		2274
		2275	ev_child cw;
		2276
		2277	static void
		2278	child_cb (EV_P_ ev_child *w, int revents)
		2279	{
		2280	ev_child_stop (EV_A_ w);
		2281	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2282	}
		2283
		2284	pid_t pid = fork ();
		2285
		2286	if (pid < 0)
		2287	// error
		2288	else if (pid == 0)
		2289	{
		2290	// the forked child executes here
		2291	exit (1);
		2292	}
		2293	else
		2294	{
		2295	ev_child_init (&cw, child_cb, pid, 0);
		2296	ev_child_start (EV_DEFAULT_ &cw);
		2297	}
		2298
		2299
		2300	=head2 C<ev_stat> - did the file attributes just change?
		2301
		2302	This watches a file system path for attribute changes. That is, it calls
		2303	C<stat> on that path in regular intervals (or when the OS says it changed)
		2304	and sees if it changed compared to the last time, invoking the callback if
		2305	it did.
		2306
		2307	The path does not need to exist: changing from "path exists" to "path does
		2308	not exist" is a status change like any other. The condition "path does not
		2309	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2310	C<st_nlink> field being zero (which is otherwise always forced to be at
		2311	least one) and all the other fields of the stat buffer having unspecified
		2312	contents.
		2313
		2314	The path I<must not> end in a slash or contain special components such as
		2315	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2316	your working directory changes, then the behaviour is undefined.
		2317
		2318	Since there is no portable change notification interface available, the
		2319	portable implementation simply calls C<stat(2)> regularly on the path
		2320	to see if it changed somehow. You can specify a recommended polling
		2321	interval for this case. If you specify a polling interval of C<0> (highly
		2322	recommended!) then a I<suitable, unspecified default> value will be used
		2323	(which you can expect to be around five seconds, although this might
		2324	change dynamically). Libev will also impose a minimum interval which is
		2325	currently around C<0.1>, but that's usually overkill.
		2326
		2327	This watcher type is not meant for massive numbers of stat watchers,
		2328	as even with OS-supported change notifications, this can be
		2329	resource-intensive.
		2330
		2331	At the time of this writing, the only OS-specific interface implemented
		2332	is the Linux inotify interface (implementing kqueue support is left as an
		2333	exercise for the reader. Note, however, that the author sees no way of
		2334	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2335
		2336	=head3 ABI Issues (Largefile Support)
		2337
		2338	Libev by default (unless the user overrides this) uses the default
		2339	compilation environment, which means that on systems with large file
		2340	support disabled by default, you get the 32 bit version of the stat
		2341	structure. When using the library from programs that change the ABI to
		2342	use 64 bit file offsets the programs will fail. In that case you have to
		2343	compile libev with the same flags to get binary compatibility. This is
		2344	obviously the case with any flags that change the ABI, but the problem is
		2345	most noticeably displayed with ev_stat and large file support.
		2346
		2347	The solution for this is to lobby your distribution maker to make large
		2348	file interfaces available by default (as e.g. FreeBSD does) and not
		2349	optional. Libev cannot simply switch on large file support because it has
		2350	to exchange stat structures with application programs compiled using the
		2351	default compilation environment.
		2352
		2353	=head3 Inotify and Kqueue
		2354
		2355	When C<inotify (7)> support has been compiled into libev and present at
		2356	runtime, it will be used to speed up change detection where possible. The
		2357	inotify descriptor will be created lazily when the first C<ev_stat>
		2358	watcher is being started.
		2359
		2360	Inotify presence does not change the semantics of C<ev_stat> watchers
		2361	except that changes might be detected earlier, and in some cases, to avoid
		2362	making regular C<stat> calls. Even in the presence of inotify support
		2363	there are many cases where libev has to resort to regular C<stat> polling,
		2364	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2365	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2366	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2367	xfs are fully working) libev usually gets away without polling.
		2368
		2369	There is no support for kqueue, as apparently it cannot be used to
		2370	implement this functionality, due to the requirement of having a file
		2371	descriptor open on the object at all times, and detecting renames, unlinks
		2372	etc. is difficult.
		2373
		2374	=head3 C<stat ()> is a synchronous operation
		2375
		2376	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2377	the process. The exception are C<ev_stat> watchers - those call C<stat
		2378	()>, which is a synchronous operation.
		2379
		2380	For local paths, this usually doesn't matter: unless the system is very
		2381	busy or the intervals between stat's are large, a stat call will be fast,
		2382	as the path data is usually in memory already (except when starting the
		2383	watcher).
		2384
		2385	For networked file systems, calling C<stat ()> can block an indefinite
		2386	time due to network issues, and even under good conditions, a stat call
		2387	often takes multiple milliseconds.
		2388
		2389	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2390	paths, although this is fully supported by libev.
		2391
		2392	=head3 The special problem of stat time resolution
		2393
		2394	The C<stat ()> system call only supports full-second resolution portably,
		2395	and even on systems where the resolution is higher, most file systems
		2396	still only support whole seconds.
		2397
		2398	That means that, if the time is the only thing that changes, you can
		2399	easily miss updates: on the first update, C<ev_stat> detects a change and
		2400	calls your callback, which does something. When there is another update
		2401	within the same second, C<ev_stat> will be unable to detect unless the
		2402	stat data does change in other ways (e.g. file size).
		2403
		2404	The solution to this is to delay acting on a change for slightly more
		2405	than a second (or till slightly after the next full second boundary), using
		2406	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2407	ev_timer_again (loop, w)>).
		2408
		2409	The C<.02> offset is added to work around small timing inconsistencies
		2410	of some operating systems (where the second counter of the current time
		2411	might be be delayed. One such system is the Linux kernel, where a call to
		2412	C<gettimeofday> might return a timestamp with a full second later than
		2413	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2414	update file times then there will be a small window where the kernel uses
		2415	the previous second to update file times but libev might already execute
		2416	the timer callback).
		2417
		2418	=head3 Watcher-Specific Functions and Data Members
		2419
		2420	=over 4
		2421
		2422	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2423
		2424	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2425
		2426	Configures the watcher to wait for status changes of the given
		2427	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2428	be detected and should normally be specified as C<0> to let libev choose
		2429	a suitable value. The memory pointed to by C<path> must point to the same
		2430	path for as long as the watcher is active.
		2431
		2432	The callback will receive an C<EV_STAT> event when a change was detected,
		2433	relative to the attributes at the time the watcher was started (or the
		2434	last change was detected).
		2435
		2436	=item ev_stat_stat (loop, ev_stat *)
		2437
		2438	Updates the stat buffer immediately with new values. If you change the
		2439	watched path in your callback, you could call this function to avoid
		2440	detecting this change (while introducing a race condition if you are not
		2441	the only one changing the path). Can also be useful simply to find out the
		2442	new values.
		2443
		2444	=item ev_statdata attr [read-only]
		2445
		2446	The most-recently detected attributes of the file. Although the type is
		2447	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2448	suitable for your system, but you can only rely on the POSIX-standardised
		2449	members to be present. If the C<st_nlink> member is C<0>, then there was
		2450	some error while C<stat>ing the file.
		2451
		2452	=item ev_statdata prev [read-only]
		2453
		2454	The previous attributes of the file. The callback gets invoked whenever
		2455	C<prev> != C<attr>, or, more precisely, one or more of these members
		2456	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2457	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2458
		2459	=item ev_tstamp interval [read-only]
		2460
		2461	The specified interval.
		2462
		2463	=item const char *path [read-only]
		2464
		2465	The file system path that is being watched.
		2466
		2467	=back
		2468
		2469	=head3 Examples
		2470
		2471	Example: Watch C</etc/passwd> for attribute changes.
		2472
		2473	static void
		2474	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2475	{
		2476	/* /etc/passwd changed in some way */
		2477	if (w->attr.st_nlink)
		2478	{
		2479	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2480	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2481	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2482	}
		2483	else
		2484	/* you shalt not abuse printf for puts */
		2485	puts ("wow, /etc/passwd is not there, expect problems. "
		2486	"if this is windows, they already arrived\n");
		2487	}
		2488
		2489	...
		2490	ev_stat passwd;
		2491
		2492	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2493	ev_stat_start (loop, &passwd);
		2494
		2495	Example: Like above, but additionally use a one-second delay so we do not
		2496	miss updates (however, frequent updates will delay processing, too, so
		2497	one might do the work both on C<ev_stat> callback invocation I<and> on
		2498	C<ev_timer> callback invocation).
		2499
		2500	static ev_stat passwd;
		2501	static ev_timer timer;
		2502
		2503	static void
		2504	timer_cb (EV_P_ ev_timer *w, int revents)
		2505	{
		2506	ev_timer_stop (EV_A_ w);
		2507
		2508	/* now it's one second after the most recent passwd change */
		2509	}
		2510
		2511	static void
		2512	stat_cb (EV_P_ ev_stat *w, int revents)
		2513	{
		2514	/* reset the one-second timer */
		2515	ev_timer_again (EV_A_ &timer);
		2516	}
		2517
		2518	...
		2519	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2520	ev_stat_start (loop, &passwd);
		2521	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2522
		2523
651	=head2 C<ev_idle> - when you've got nothing better to do	2524	=head2 C<ev_idle> - when you've got nothing better to do...
652		2525
653	Idle watchers trigger events when there are no other events are pending	2526	Idle watchers trigger events when no other events of the same or higher
654	(prepare, check and other idle watchers do not count). That is, as long	2527	priority are pending (prepare, check and other idle watchers do not count
655	as your process is busy handling sockets or timeouts (or even signals,	2528	as receiving "events").
656	imagine) it will not be triggered. But when your process is idle all idle	2529
657	watchers are being called again and again, once per event loop iteration -	2530	That is, as long as your process is busy handling sockets or timeouts
		2531	(or even signals, imagine) of the same or higher priority it will not be
		2532	triggered. But when your process is idle (or only lower-priority watchers
		2533	are pending), the idle watchers are being called once per event loop
658	until stopped, that is, or your process receives more events and becomes	2534	iteration - until stopped, that is, or your process receives more events
659	busy.	2535	and becomes busy again with higher priority stuff.
660		2536
661	The most noteworthy effect is that as long as any idle watchers are	2537	The most noteworthy effect is that as long as any idle watchers are
662	active, the process will not block when waiting for new events.	2538	active, the process will not block when waiting for new events.
663		2539
664	Apart from keeping your process non-blocking (which is a useful	2540	Apart from keeping your process non-blocking (which is a useful
665	effect on its own sometimes), idle watchers are a good place to do	2541	effect on its own sometimes), idle watchers are a good place to do
666	"pseudo-background processing", or delay processing stuff to after the	2542	"pseudo-background processing", or delay processing stuff to after the
667	event loop has handled all outstanding events.	2543	event loop has handled all outstanding events.
668		2544
		2545	=head3 Watcher-Specific Functions and Data Members
		2546
669	=over 4	2547	=over 4
670		2548
671	=item ev_idle_init (ev_signal *, callback)	2549	=item ev_idle_init (ev_idle *, callback)
672		2550
673	Initialises and configures the idle watcher - it has no parameters of any	2551	Initialises and configures the idle watcher - it has no parameters of any
674	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2552	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
675	believe me.	2553	believe me.
676		2554
677	=back	2555	=back
678		2556
		2557	=head3 Examples
		2558
		2559	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2560	callback, free it. Also, use no error checking, as usual.
		2561
		2562	static void
		2563	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2564	{
		2565	free (w);
		2566	// now do something you wanted to do when the program has
		2567	// no longer anything immediate to do.
		2568	}
		2569
		2570	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2571	ev_idle_init (idle_watcher, idle_cb);
		2572	ev_idle_start (loop, idle_watcher);
		2573
		2574
679	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2575	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
680		2576
681	Prepare and check watchers are usually (but not always) used in tandem:	2577	Prepare and check watchers are usually (but not always) used in pairs:
682	prepare watchers get invoked before the process blocks and check watchers	2578	prepare watchers get invoked before the process blocks and check watchers
683	afterwards.	2579	afterwards.
684		2580
		2581	You I<must not> call C<ev_loop> or similar functions that enter
		2582	the current event loop from either C<ev_prepare> or C<ev_check>
		2583	watchers. Other loops than the current one are fine, however. The
		2584	rationale behind this is that you do not need to check for recursion in
		2585	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2586	C<ev_check> so if you have one watcher of each kind they will always be
		2587	called in pairs bracketing the blocking call.
		2588
685	Their main purpose is to integrate other event mechanisms into libev. This	2589	Their main purpose is to integrate other event mechanisms into libev and
686	could be used, for example, to track variable changes, implement your own	2590	their use is somewhat advanced. They could be used, for example, to track
687	watchers, integrate net-snmp or a coroutine library and lots more.	2591	variable changes, implement your own watchers, integrate net-snmp or a
		2592	coroutine library and lots more. They are also occasionally useful if
		2593	you cache some data and want to flush it before blocking (for example,
		2594	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2595	watcher).
688		2596
689	This is done by examining in each prepare call which file descriptors need	2597	This is done by examining in each prepare call which file descriptors
690	to be watched by the other library, registering C<ev_io> watchers for	2598	need to be watched by the other library, registering C<ev_io> watchers
691	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2599	for them and starting an C<ev_timer> watcher for any timeouts (many
692	provide just this functionality). Then, in the check watcher you check for	2600	libraries provide exactly this functionality). Then, in the check watcher,
693	any events that occured (by checking the pending status of all watchers	2601	you check for any events that occurred (by checking the pending status
694	and stopping them) and call back into the library. The I/O and timer	2602	of all watchers and stopping them) and call back into the library. The
695	callbacks will never actually be called (but must be valid nevertheless,	2603	I/O and timer callbacks will never actually be called (but must be valid
696	because you never know, you know?).	2604	nevertheless, because you never know, you know?).
697		2605
698	As another example, the Perl Coro module uses these hooks to integrate	2606	As another example, the Perl Coro module uses these hooks to integrate
699	coroutines into libev programs, by yielding to other active coroutines	2607	coroutines into libev programs, by yielding to other active coroutines
700	during each prepare and only letting the process block if no coroutines	2608	during each prepare and only letting the process block if no coroutines
701	are ready to run (it's actually more complicated: it only runs coroutines	2609	are ready to run (it's actually more complicated: it only runs coroutines
702	with priority higher than or equal to the event loop and one coroutine	2610	with priority higher than or equal to the event loop and one coroutine
703	of lower priority, but only once, using idle watchers to keep the event	2611	of lower priority, but only once, using idle watchers to keep the event
704	loop from blocking if lower-priority coroutines are active, thus mapping	2612	loop from blocking if lower-priority coroutines are active, thus mapping
705	low-priority coroutines to idle/background tasks).	2613	low-priority coroutines to idle/background tasks).
706		2614
		2615	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2616	priority, to ensure that they are being run before any other watchers
		2617	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2618
		2619	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2620	activate ("feed") events into libev. While libev fully supports this, they
		2621	might get executed before other C<ev_check> watchers did their job. As
		2622	C<ev_check> watchers are often used to embed other (non-libev) event
		2623	loops those other event loops might be in an unusable state until their
		2624	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2625	others).
		2626
		2627	=head3 Watcher-Specific Functions and Data Members
		2628
707	=over 4	2629	=over 4
708		2630
709	=item ev_prepare_init (ev_prepare *, callback)	2631	=item ev_prepare_init (ev_prepare *, callback)
710		2632
711	=item ev_check_init (ev_check *, callback)	2633	=item ev_check_init (ev_check *, callback)
712		2634
713	Initialises and configures the prepare or check watcher - they have no	2635	Initialises and configures the prepare or check watcher - they have no
714	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2636	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
715	macros, but using them is utterly, utterly and completely pointless.	2637	macros, but using them is utterly, utterly, utterly and completely
		2638	pointless.
716		2639
717	=back	2640	=back
718		2641
		2642	=head3 Examples
		2643
		2644	There are a number of principal ways to embed other event loops or modules
		2645	into libev. Here are some ideas on how to include libadns into libev
		2646	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2647	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2648	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2649	Glib event loop).
		2650
		2651	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2652	and in a check watcher, destroy them and call into libadns. What follows
		2653	is pseudo-code only of course. This requires you to either use a low
		2654	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2655	the callbacks for the IO/timeout watchers might not have been called yet.
		2656
		2657	static ev_io iow [nfd];
		2658	static ev_timer tw;
		2659
		2660	static void
		2661	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2662	{
		2663	}
		2664
		2665	// create io watchers for each fd and a timer before blocking
		2666	static void
		2667	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2668	{
		2669	int timeout = 3600000;
		2670	struct pollfd fds [nfd];
		2671	// actual code will need to loop here and realloc etc.
		2672	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2673
		2674	/* the callback is illegal, but won't be called as we stop during check */
		2675	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2676	ev_timer_start (loop, &tw);
		2677
		2678	// create one ev_io per pollfd
		2679	for (int i = 0; i < nfd; ++i)
		2680	{
		2681	ev_io_init (iow + i, io_cb, fds [i].fd,
		2682	((fds [i].events & POLLIN ? EV_READ : 0)
		2683	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2684
		2685	fds [i].revents = 0;
		2686	ev_io_start (loop, iow + i);
		2687	}
		2688	}
		2689
		2690	// stop all watchers after blocking
		2691	static void
		2692	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2693	{
		2694	ev_timer_stop (loop, &tw);
		2695
		2696	for (int i = 0; i < nfd; ++i)
		2697	{
		2698	// set the relevant poll flags
		2699	// could also call adns_processreadable etc. here
		2700	struct pollfd *fd = fds + i;
		2701	int revents = ev_clear_pending (iow + i);
		2702	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2703	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2704
		2705	// now stop the watcher
		2706	ev_io_stop (loop, iow + i);
		2707	}
		2708
		2709	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2710	}
		2711
		2712	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2713	in the prepare watcher and would dispose of the check watcher.
		2714
		2715	Method 3: If the module to be embedded supports explicit event
		2716	notification (libadns does), you can also make use of the actual watcher
		2717	callbacks, and only destroy/create the watchers in the prepare watcher.
		2718
		2719	static void
		2720	timer_cb (EV_P_ ev_timer *w, int revents)
		2721	{
		2722	adns_state ads = (adns_state)w->data;
		2723	update_now (EV_A);
		2724
		2725	adns_processtimeouts (ads, &tv_now);
		2726	}
		2727
		2728	static void
		2729	io_cb (EV_P_ ev_io *w, int revents)
		2730	{
		2731	adns_state ads = (adns_state)w->data;
		2732	update_now (EV_A);
		2733
		2734	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2735	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2736	}
		2737
		2738	// do not ever call adns_afterpoll
		2739
		2740	Method 4: Do not use a prepare or check watcher because the module you
		2741	want to embed is not flexible enough to support it. Instead, you can
		2742	override their poll function. The drawback with this solution is that the
		2743	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2744	this approach, effectively embedding EV as a client into the horrible
		2745	libglib event loop.
		2746
		2747	static gint
		2748	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2749	{
		2750	int got_events = 0;
		2751
		2752	for (n = 0; n < nfds; ++n)
		2753	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2754
		2755	if (timeout >= 0)
		2756	// create/start timer
		2757
		2758	// poll
		2759	ev_loop (EV_A_ 0);
		2760
		2761	// stop timer again
		2762	if (timeout >= 0)
		2763	ev_timer_stop (EV_A_ &to);
		2764
		2765	// stop io watchers again - their callbacks should have set
		2766	for (n = 0; n < nfds; ++n)
		2767	ev_io_stop (EV_A_ iow [n]);
		2768
		2769	return got_events;
		2770	}
		2771
		2772
		2773	=head2 C<ev_embed> - when one backend isn't enough...
		2774
		2775	This is a rather advanced watcher type that lets you embed one event loop
		2776	into another (currently only C<ev_io> events are supported in the embedded
		2777	loop, other types of watchers might be handled in a delayed or incorrect
		2778	fashion and must not be used).
		2779
		2780	There are primarily two reasons you would want that: work around bugs and
		2781	prioritise I/O.
		2782
		2783	As an example for a bug workaround, the kqueue backend might only support
		2784	sockets on some platform, so it is unusable as generic backend, but you
		2785	still want to make use of it because you have many sockets and it scales
		2786	so nicely. In this case, you would create a kqueue-based loop and embed
		2787	it into your default loop (which might use e.g. poll). Overall operation
		2788	will be a bit slower because first libev has to call C<poll> and then
		2789	C<kevent>, but at least you can use both mechanisms for what they are
		2790	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2791
		2792	As for prioritising I/O: under rare circumstances you have the case where
		2793	some fds have to be watched and handled very quickly (with low latency),
		2794	and even priorities and idle watchers might have too much overhead. In
		2795	this case you would put all the high priority stuff in one loop and all
		2796	the rest in a second one, and embed the second one in the first.
		2797
		2798	As long as the watcher is active, the callback will be invoked every
		2799	time there might be events pending in the embedded loop. The callback
		2800	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		2801	sweep and invoke their callbacks (the callback doesn't need to invoke the
		2802	C<ev_embed_sweep> function directly, it could also start an idle watcher
		2803	to give the embedded loop strictly lower priority for example).
		2804
		2805	You can also set the callback to C<0>, in which case the embed watcher
		2806	will automatically execute the embedded loop sweep whenever necessary.
		2807
		2808	Fork detection will be handled transparently while the C<ev_embed> watcher
		2809	is active, i.e., the embedded loop will automatically be forked when the
		2810	embedding loop forks. In other cases, the user is responsible for calling
		2811	C<ev_loop_fork> on the embedded loop.
		2812
		2813	Unfortunately, not all backends are embeddable: only the ones returned by
		2814	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2815	portable one.
		2816
		2817	So when you want to use this feature you will always have to be prepared
		2818	that you cannot get an embeddable loop. The recommended way to get around
		2819	this is to have a separate variables for your embeddable loop, try to
		2820	create it, and if that fails, use the normal loop for everything.
		2821
		2822	=head3 C<ev_embed> and fork
		2823
		2824	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2825	automatically be applied to the embedded loop as well, so no special
		2826	fork handling is required in that case. When the watcher is not running,
		2827	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2828	as applicable.
		2829
		2830	=head3 Watcher-Specific Functions and Data Members
		2831
		2832	=over 4
		2833
		2834	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2835
		2836	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2837
		2838	Configures the watcher to embed the given loop, which must be
		2839	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2840	invoked automatically, otherwise it is the responsibility of the callback
		2841	to invoke it (it will continue to be called until the sweep has been done,
		2842	if you do not want that, you need to temporarily stop the embed watcher).
		2843
		2844	=item ev_embed_sweep (loop, ev_embed *)
		2845
		2846	Make a single, non-blocking sweep over the embedded loop. This works
		2847	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		2848	appropriate way for embedded loops.
		2849
		2850	=item struct ev_loop *other [read-only]
		2851
		2852	The embedded event loop.
		2853
		2854	=back
		2855
		2856	=head3 Examples
		2857
		2858	Example: Try to get an embeddable event loop and embed it into the default
		2859	event loop. If that is not possible, use the default loop. The default
		2860	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2861	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2862	used).
		2863
		2864	struct ev_loop *loop_hi = ev_default_init (0);
		2865	struct ev_loop *loop_lo = 0;
		2866	ev_embed embed;
		2867
		2868	// see if there is a chance of getting one that works
		2869	// (remember that a flags value of 0 means autodetection)
		2870	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2871	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2872	: 0;
		2873
		2874	// if we got one, then embed it, otherwise default to loop_hi
		2875	if (loop_lo)
		2876	{
		2877	ev_embed_init (&embed, 0, loop_lo);
		2878	ev_embed_start (loop_hi, &embed);
		2879	}
		2880	else
		2881	loop_lo = loop_hi;
		2882
		2883	Example: Check if kqueue is available but not recommended and create
		2884	a kqueue backend for use with sockets (which usually work with any
		2885	kqueue implementation). Store the kqueue/socket-only event loop in
		2886	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2887
		2888	struct ev_loop *loop = ev_default_init (0);
		2889	struct ev_loop *loop_socket = 0;
		2890	ev_embed embed;
		2891
		2892	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2893	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2894	{
		2895	ev_embed_init (&embed, 0, loop_socket);
		2896	ev_embed_start (loop, &embed);
		2897	}
		2898
		2899	if (!loop_socket)
		2900	loop_socket = loop;
		2901
		2902	// now use loop_socket for all sockets, and loop for everything else
		2903
		2904
		2905	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		2906
		2907	Fork watchers are called when a C<fork ()> was detected (usually because
		2908	whoever is a good citizen cared to tell libev about it by calling
		2909	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		2910	event loop blocks next and before C<ev_check> watchers are being called,
		2911	and only in the child after the fork. If whoever good citizen calling
		2912	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		2913	handlers will be invoked, too, of course.
		2914
		2915	=head3 The special problem of life after fork - how is it possible?
		2916
		2917	Most uses of C<fork()> consist of forking, then some simple calls to ste
		2918	up/change the process environment, followed by a call to C<exec()>. This
		2919	sequence should be handled by libev without any problems.
		2920
		2921	This changes when the application actually wants to do event handling
		2922	in the child, or both parent in child, in effect "continuing" after the
		2923	fork.
		2924
		2925	The default mode of operation (for libev, with application help to detect
		2926	forks) is to duplicate all the state in the child, as would be expected
		2927	when I<either> the parent I<or> the child process continues.
		2928
		2929	When both processes want to continue using libev, then this is usually the
		2930	wrong result. In that case, usually one process (typically the parent) is
		2931	supposed to continue with all watchers in place as before, while the other
		2932	process typically wants to start fresh, i.e. without any active watchers.
		2933
		2934	The cleanest and most efficient way to achieve that with libev is to
		2935	simply create a new event loop, which of course will be "empty", and
		2936	use that for new watchers. This has the advantage of not touching more
		2937	memory than necessary, and thus avoiding the copy-on-write, and the
		2938	disadvantage of having to use multiple event loops (which do not support
		2939	signal watchers).
		2940
		2941	When this is not possible, or you want to use the default loop for
		2942	other reasons, then in the process that wants to start "fresh", call
		2943	C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying
		2944	the default loop will "orphan" (not stop) all registered watchers, so you
		2945	have to be careful not to execute code that modifies those watchers. Note
		2946	also that in that case, you have to re-register any signal watchers.
		2947
		2948	=head3 Watcher-Specific Functions and Data Members
		2949
		2950	=over 4
		2951
		2952	=item ev_fork_init (ev_signal *, callback)
		2953
		2954	Initialises and configures the fork watcher - it has no parameters of any
		2955	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		2956	believe me.
		2957
		2958	=back
		2959
		2960
		2961	=head2 C<ev_async> - how to wake up another event loop
		2962
		2963	In general, you cannot use an C<ev_loop> from multiple threads or other
		2964	asynchronous sources such as signal handlers (as opposed to multiple event
		2965	loops - those are of course safe to use in different threads).
		2966
		2967	Sometimes, however, you need to wake up another event loop you do not
		2968	control, for example because it belongs to another thread. This is what
		2969	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2970	can signal it by calling C<ev_async_send>, which is thread- and signal
		2971	safe.
		2972
		2973	This functionality is very similar to C<ev_signal> watchers, as signals,
		2974	too, are asynchronous in nature, and signals, too, will be compressed
		2975	(i.e. the number of callback invocations may be less than the number of
		2976	C<ev_async_sent> calls).
		2977
		2978	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2979	just the default loop.
		2980
		2981	=head3 Queueing
		2982
		2983	C<ev_async> does not support queueing of data in any way. The reason
		2984	is that the author does not know of a simple (or any) algorithm for a
		2985	multiple-writer-single-reader queue that works in all cases and doesn't
		2986	need elaborate support such as pthreads or unportable memory access
		2987	semantics.
		2988
		2989	That means that if you want to queue data, you have to provide your own
		2990	queue. But at least I can tell you how to implement locking around your
		2991	queue:
		2992
		2993	=over 4
		2994
		2995	=item queueing from a signal handler context
		2996
		2997	To implement race-free queueing, you simply add to the queue in the signal
		2998	handler but you block the signal handler in the watcher callback. Here is
		2999	an example that does that for some fictitious SIGUSR1 handler:
		3000
		3001	static ev_async mysig;
		3002
		3003	static void
		3004	sigusr1_handler (void)
		3005	{
		3006	sometype data;
		3007
		3008	// no locking etc.
		3009	queue_put (data);
		3010	ev_async_send (EV_DEFAULT_ &mysig);
		3011	}
		3012
		3013	static void
		3014	mysig_cb (EV_P_ ev_async *w, int revents)
		3015	{
		3016	sometype data;
		3017	sigset_t block, prev;
		3018
		3019	sigemptyset (&block);
		3020	sigaddset (&block, SIGUSR1);
		3021	sigprocmask (SIG_BLOCK, &block, &prev);
		3022
		3023	while (queue_get (&data))
		3024	process (data);
		3025
		3026	if (sigismember (&prev, SIGUSR1)
		3027	sigprocmask (SIG_UNBLOCK, &block, 0);
		3028	}
		3029
		3030	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3031	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3032	either...).
		3033
		3034	=item queueing from a thread context
		3035
		3036	The strategy for threads is different, as you cannot (easily) block
		3037	threads but you can easily preempt them, so to queue safely you need to
		3038	employ a traditional mutex lock, such as in this pthread example:
		3039
		3040	static ev_async mysig;
		3041	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3042
		3043	static void
		3044	otherthread (void)
		3045	{
		3046	// only need to lock the actual queueing operation
		3047	pthread_mutex_lock (&mymutex);
		3048	queue_put (data);
		3049	pthread_mutex_unlock (&mymutex);
		3050
		3051	ev_async_send (EV_DEFAULT_ &mysig);
		3052	}
		3053
		3054	static void
		3055	mysig_cb (EV_P_ ev_async *w, int revents)
		3056	{
		3057	pthread_mutex_lock (&mymutex);
		3058
		3059	while (queue_get (&data))
		3060	process (data);
		3061
		3062	pthread_mutex_unlock (&mymutex);
		3063	}
		3064
		3065	=back
		3066
		3067
		3068	=head3 Watcher-Specific Functions and Data Members
		3069
		3070	=over 4
		3071
		3072	=item ev_async_init (ev_async *, callback)
		3073
		3074	Initialises and configures the async watcher - it has no parameters of any
		3075	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3076	trust me.
		3077
		3078	=item ev_async_send (loop, ev_async *)
		3079
		3080	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3081	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		3082	C<ev_feed_event>, this call is safe to do from other threads, signal or
		3083	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		3084	section below on what exactly this means).
		3085
		3086	Note that, as with other watchers in libev, multiple events might get
		3087	compressed into a single callback invocation (another way to look at this
		3088	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3089	reset when the event loop detects that).
		3090
		3091	This call incurs the overhead of a system call only once per event loop
		3092	iteration, so while the overhead might be noticeable, it doesn't apply to
		3093	repeated calls to C<ev_async_send> for the same event loop.
		3094
		3095	=item bool = ev_async_pending (ev_async *)
		3096
		3097	Returns a non-zero value when C<ev_async_send> has been called on the
		3098	watcher but the event has not yet been processed (or even noted) by the
		3099	event loop.
		3100
		3101	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3102	the loop iterates next and checks for the watcher to have become active,
		3103	it will reset the flag again. C<ev_async_pending> can be used to very
		3104	quickly check whether invoking the loop might be a good idea.
		3105
		3106	Not that this does I<not> check whether the watcher itself is pending,
		3107	only whether it has been requested to make this watcher pending: there
		3108	is a time window between the event loop checking and resetting the async
		3109	notification, and the callback being invoked.
		3110
		3111	=back
		3112
		3113
719	=head1 OTHER FUNCTIONS	3114	=head1 OTHER FUNCTIONS
720		3115
721	There are some other functions of possible interest. Described. Here. Now.	3116	There are some other functions of possible interest. Described. Here. Now.
722		3117
723	=over 4	3118	=over 4
724		3119
725	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3120	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
726		3121
727	This function combines a simple timer and an I/O watcher, calls your	3122	This function combines a simple timer and an I/O watcher, calls your
728	callback on whichever event happens first and automatically stop both	3123	callback on whichever event happens first and automatically stops both
729	watchers. This is useful if you want to wait for a single event on an fd	3124	watchers. This is useful if you want to wait for a single event on an fd
730	or timeout without havign to allocate/configure/start/stop/free one or	3125	or timeout without having to allocate/configure/start/stop/free one or
731	more watchers yourself.	3126	more watchers yourself.
732		3127
733	If C<fd> is less than 0, then no I/O watcher will be started and events	3128	If C<fd> is less than 0, then no I/O watcher will be started and the
734	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3129	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
735	C<events> set will be craeted and started.	3130	the given C<fd> and C<events> set will be created and started.
736		3131
737	If C<timeout> is less than 0, then no timeout watcher will be	3132	If C<timeout> is less than 0, then no timeout watcher will be
738	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3133	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
739	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3134	repeat = 0) will be started. C<0> is a valid timeout.
740	dubious value.
741		3135
742	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3136	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
743	passed an events set like normal event callbacks (with a combination of	3137	passed an C<revents> set like normal event callbacks (a combination of
744	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3138	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
745	value passed to C<ev_once>:	3139	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3140	a timeout and an io event at the same time - you probably should give io
		3141	events precedence.
746		3142
		3143	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3144
747	static void stdin_ready (int revents, void *arg)	3145	static void stdin_ready (int revents, void *arg)
748	{	3146	{
749	if (revents & EV_TIMEOUT)
750	/* doh, nothing entered */;
751	else if (revents & EV_READ)	3147	if (revents & EV_READ)
752	/* stdin might have data for us, joy! */;	3148	/* stdin might have data for us, joy! */;
		3149	else if (revents & EV_TIMEOUT)
		3150	/* doh, nothing entered */;
753	}	3151	}
754		3152
755	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3153	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
756
757	=item ev_feed_event (loop, watcher, int events)
758
759	Feeds the given event set into the event loop, as if the specified event
760	had happened for the specified watcher (which must be a pointer to an
761	initialised but not necessarily started event watcher).
762		3154
763	=item ev_feed_fd_event (loop, int fd, int revents)	3155	=item ev_feed_fd_event (loop, int fd, int revents)
764		3156
765	Feed an event on the given fd, as if a file descriptor backend detected	3157	Feed an event on the given fd, as if a file descriptor backend detected
766	the given events it.	3158	the given events it.
767		3159
768	=item ev_feed_signal_event (loop, int signum)	3160	=item ev_feed_signal_event (loop, int signum)
769		3161
770	Feed an event as if the given signal occured (loop must be the default loop!).	3162	Feed an event as if the given signal occurred (C<loop> must be the default
		3163	loop!).
771		3164
772	=back	3165	=back
773		3166
		3167
774	=head1 LIBEVENT EMULATION	3168	=head1 LIBEVENT EMULATION
775		3169
776	TBD.	3170	Libev offers a compatibility emulation layer for libevent. It cannot
		3171	emulate the internals of libevent, so here are some usage hints:
		3172
		3173	=over 4
		3174
		3175	=item * Use it by including <event.h>, as usual.
		3176
		3177	=item * The following members are fully supported: ev_base, ev_callback,
		3178	ev_arg, ev_fd, ev_res, ev_events.
		3179
		3180	=item * Avoid using ev_flags and the EVLIST_*-macros, while it is
		3181	maintained by libev, it does not work exactly the same way as in libevent (consider
		3182	it a private API).
		3183
		3184	=item * Priorities are not currently supported. Initialising priorities
		3185	will fail and all watchers will have the same priority, even though there
		3186	is an ev_pri field.
		3187
		3188	=item * In libevent, the last base created gets the signals, in libev, the
		3189	first base created (== the default loop) gets the signals.
		3190
		3191	=item * Other members are not supported.
		3192
		3193	=item * The libev emulation is I<not> ABI compatible to libevent, you need
		3194	to use the libev header file and library.
		3195
		3196	=back
777		3197
778	=head1 C++ SUPPORT	3198	=head1 C++ SUPPORT
779		3199
780	TBD.	3200	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3201	you to use some convenience methods to start/stop watchers and also change
		3202	the callback model to a model using method callbacks on objects.
		3203
		3204	To use it,
		3205
		3206	#include <ev++.h>
		3207
		3208	This automatically includes F<ev.h> and puts all of its definitions (many
		3209	of them macros) into the global namespace. All C++ specific things are
		3210	put into the C<ev> namespace. It should support all the same embedding
		3211	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3212
		3213	Care has been taken to keep the overhead low. The only data member the C++
		3214	classes add (compared to plain C-style watchers) is the event loop pointer
		3215	that the watcher is associated with (or no additional members at all if
		3216	you disable C<EV_MULTIPLICITY> when embedding libev).
		3217
		3218	Currently, functions, and static and non-static member functions can be
		3219	used as callbacks. Other types should be easy to add as long as they only
		3220	need one additional pointer for context. If you need support for other
		3221	types of functors please contact the author (preferably after implementing
		3222	it).
		3223
		3224	Here is a list of things available in the C<ev> namespace:
		3225
		3226	=over 4
		3227
		3228	=item C<ev::READ>, C<ev::WRITE> etc.
		3229
		3230	These are just enum values with the same values as the C<EV_READ> etc.
		3231	macros from F<ev.h>.
		3232
		3233	=item C<ev::tstamp>, C<ev::now>
		3234
		3235	Aliases to the same types/functions as with the C<ev_> prefix.
		3236
		3237	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3238
		3239	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3240	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3241	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3242	defines by many implementations.
		3243
		3244	All of those classes have these methods:
		3245
		3246	=over 4
		3247
		3248	=item ev::TYPE::TYPE ()
		3249
		3250	=item ev::TYPE::TYPE (loop)
		3251
		3252	=item ev::TYPE::~TYPE
		3253
		3254	The constructor (optionally) takes an event loop to associate the watcher
		3255	with. If it is omitted, it will use C<EV_DEFAULT>.
		3256
		3257	The constructor calls C<ev_init> for you, which means you have to call the
		3258	C<set> method before starting it.
		3259
		3260	It will not set a callback, however: You have to call the templated C<set>
		3261	method to set a callback before you can start the watcher.
		3262
		3263	(The reason why you have to use a method is a limitation in C++ which does
		3264	not allow explicit template arguments for constructors).
		3265
		3266	The destructor automatically stops the watcher if it is active.
		3267
		3268	=item w->set<class, &class::method> (object *)
		3269
		3270	This method sets the callback method to call. The method has to have a
		3271	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3272	first argument and the C<revents> as second. The object must be given as
		3273	parameter and is stored in the C<data> member of the watcher.
		3274
		3275	This method synthesizes efficient thunking code to call your method from
		3276	the C callback that libev requires. If your compiler can inline your
		3277	callback (i.e. it is visible to it at the place of the C<set> call and
		3278	your compiler is good :), then the method will be fully inlined into the
		3279	thunking function, making it as fast as a direct C callback.
		3280
		3281	Example: simple class declaration and watcher initialisation
		3282
		3283	struct myclass
		3284	{
		3285	void io_cb (ev::io &w, int revents) { }
		3286	}
		3287
		3288	myclass obj;
		3289	ev::io iow;
		3290	iow.set <myclass, &myclass::io_cb> (&obj);
		3291
		3292	=item w->set (object *)
		3293
		3294	This is an B<experimental> feature that might go away in a future version.
		3295
		3296	This is a variation of a method callback - leaving out the method to call
		3297	will default the method to C<operator ()>, which makes it possible to use
		3298	functor objects without having to manually specify the C<operator ()> all
		3299	the time. Incidentally, you can then also leave out the template argument
		3300	list.
		3301
		3302	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3303	int revents)>.
		3304
		3305	See the method-C<set> above for more details.
		3306
		3307	Example: use a functor object as callback.
		3308
		3309	struct myfunctor
		3310	{
		3311	void operator() (ev::io &w, int revents)
		3312	{
		3313	...
		3314	}
		3315	}
		3316
		3317	myfunctor f;
		3318
		3319	ev::io w;
		3320	w.set (&f);
		3321
		3322	=item w->set<function> (void *data = 0)
		3323
		3324	Also sets a callback, but uses a static method or plain function as
		3325	callback. The optional C<data> argument will be stored in the watcher's
		3326	C<data> member and is free for you to use.
		3327
		3328	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3329
		3330	See the method-C<set> above for more details.
		3331
		3332	Example: Use a plain function as callback.
		3333
		3334	static void io_cb (ev::io &w, int revents) { }
		3335	iow.set <io_cb> ();
		3336
		3337	=item w->set (loop)
		3338
		3339	Associates a different C<struct ev_loop> with this watcher. You can only
		3340	do this when the watcher is inactive (and not pending either).
		3341
		3342	=item w->set ([arguments])
		3343
		3344	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
		3345	called at least once. Unlike the C counterpart, an active watcher gets
		3346	automatically stopped and restarted when reconfiguring it with this
		3347	method.
		3348
		3349	=item w->start ()
		3350
		3351	Starts the watcher. Note that there is no C<loop> argument, as the
		3352	constructor already stores the event loop.
		3353
		3354	=item w->stop ()
		3355
		3356	Stops the watcher if it is active. Again, no C<loop> argument.
		3357
		3358	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3359
		3360	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3361	C<ev_TYPE_again> function.
		3362
		3363	=item w->sweep () (C<ev::embed> only)
		3364
		3365	Invokes C<ev_embed_sweep>.
		3366
		3367	=item w->update () (C<ev::stat> only)
		3368
		3369	Invokes C<ev_stat_stat>.
		3370
		3371	=back
		3372
		3373	=back
		3374
		3375	Example: Define a class with an IO and idle watcher, start one of them in
		3376	the constructor.
		3377
		3378	class myclass
		3379	{
		3380	ev::io io ; void io_cb (ev::io &w, int revents);
		3381	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3382
		3383	myclass (int fd)
		3384	{
		3385	io .set <myclass, &myclass::io_cb > (this);
		3386	idle.set <myclass, &myclass::idle_cb> (this);
		3387
		3388	io.start (fd, ev::READ);
		3389	}
		3390	};
		3391
		3392
		3393	=head1 OTHER LANGUAGE BINDINGS
		3394
		3395	Libev does not offer other language bindings itself, but bindings for a
		3396	number of languages exist in the form of third-party packages. If you know
		3397	any interesting language binding in addition to the ones listed here, drop
		3398	me a note.
		3399
		3400	=over 4
		3401
		3402	=item Perl
		3403
		3404	The EV module implements the full libev API and is actually used to test
		3405	libev. EV is developed together with libev. Apart from the EV core module,
		3406	there are additional modules that implement libev-compatible interfaces
		3407	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3408	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3409	and C<EV::Glib>).
		3410
		3411	It can be found and installed via CPAN, its homepage is at
		3412	L<http://software.schmorp.de/pkg/EV>.
		3413
		3414	=item Python
		3415
		3416	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3417	seems to be quite complete and well-documented.
		3418
		3419	=item Ruby
		3420
		3421	Tony Arcieri has written a ruby extension that offers access to a subset
		3422	of the libev API and adds file handle abstractions, asynchronous DNS and
		3423	more on top of it. It can be found via gem servers. Its homepage is at
		3424	L<http://rev.rubyforge.org/>.
		3425
		3426	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3427	makes rev work even on mingw.
		3428
		3429	=item Haskell
		3430
		3431	A haskell binding to libev is available at
		3432	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3433
		3434	=item D
		3435
		3436	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3437	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3438
		3439	=item Ocaml
		3440
		3441	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3442	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3443
		3444	=item Lua
		3445
		3446	Brian Maher has written a partial interface to libev
		3447	for lua (only C<ev_io> and C<ev_timer>), to be found at
		3448	L<http://github.com/brimworks/lua-ev>.
		3449
		3450	=back
		3451
		3452
		3453	=head1 MACRO MAGIC
		3454
		3455	Libev can be compiled with a variety of options, the most fundamental
		3456	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3457	functions and callbacks have an initial C<struct ev_loop *> argument.
		3458
		3459	To make it easier to write programs that cope with either variant, the
		3460	following macros are defined:
		3461
		3462	=over 4
		3463
		3464	=item C<EV_A>, C<EV_A_>
		3465
		3466	This provides the loop I<argument> for functions, if one is required ("ev
		3467	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3468	C<EV_A_> is used when other arguments are following. Example:
		3469
		3470	ev_unref (EV_A);
		3471	ev_timer_add (EV_A_ watcher);
		3472	ev_loop (EV_A_ 0);
		3473
		3474	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3475	which is often provided by the following macro.
		3476
		3477	=item C<EV_P>, C<EV_P_>
		3478
		3479	This provides the loop I<parameter> for functions, if one is required ("ev
		3480	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3481	C<EV_P_> is used when other parameters are following. Example:
		3482
		3483	// this is how ev_unref is being declared
		3484	static void ev_unref (EV_P);
		3485
		3486	// this is how you can declare your typical callback
		3487	static void cb (EV_P_ ev_timer *w, int revents)
		3488
		3489	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3490	suitable for use with C<EV_A>.
		3491
		3492	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3493
		3494	Similar to the other two macros, this gives you the value of the default
		3495	loop, if multiple loops are supported ("ev loop default").
		3496
		3497	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3498
		3499	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3500	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3501	is undefined when the default loop has not been initialised by a previous
		3502	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3503
		3504	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3505	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3506
		3507	=back
		3508
		3509	Example: Declare and initialise a check watcher, utilising the above
		3510	macros so it will work regardless of whether multiple loops are supported
		3511	or not.
		3512
		3513	static void
		3514	check_cb (EV_P_ ev_timer *w, int revents)
		3515	{
		3516	ev_check_stop (EV_A_ w);
		3517	}
		3518
		3519	ev_check check;
		3520	ev_check_init (&check, check_cb);
		3521	ev_check_start (EV_DEFAULT_ &check);
		3522	ev_loop (EV_DEFAULT_ 0);
		3523
		3524	=head1 EMBEDDING
		3525
		3526	Libev can (and often is) directly embedded into host
		3527	applications. Examples of applications that embed it include the Deliantra
		3528	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3529	and rxvt-unicode.
		3530
		3531	The goal is to enable you to just copy the necessary files into your
		3532	source directory without having to change even a single line in them, so
		3533	you can easily upgrade by simply copying (or having a checked-out copy of
		3534	libev somewhere in your source tree).
		3535
		3536	=head2 FILESETS
		3537
		3538	Depending on what features you need you need to include one or more sets of files
		3539	in your application.
		3540
		3541	=head3 CORE EVENT LOOP
		3542
		3543	To include only the libev core (all the C<ev_*> functions), with manual
		3544	configuration (no autoconf):
		3545
		3546	#define EV_STANDALONE 1
		3547	#include "ev.c"
		3548
		3549	This will automatically include F<ev.h>, too, and should be done in a
		3550	single C source file only to provide the function implementations. To use
		3551	it, do the same for F<ev.h> in all files wishing to use this API (best
		3552	done by writing a wrapper around F<ev.h> that you can include instead and
		3553	where you can put other configuration options):
		3554
		3555	#define EV_STANDALONE 1
		3556	#include "ev.h"
		3557
		3558	Both header files and implementation files can be compiled with a C++
		3559	compiler (at least, that's a stated goal, and breakage will be treated
		3560	as a bug).
		3561
		3562	You need the following files in your source tree, or in a directory
		3563	in your include path (e.g. in libev/ when using -Ilibev):
		3564
		3565	ev.h
		3566	ev.c
		3567	ev_vars.h
		3568	ev_wrap.h
		3569
		3570	ev_win32.c required on win32 platforms only
		3571
		3572	ev_select.c only when select backend is enabled (which is enabled by default)
		3573	ev_poll.c only when poll backend is enabled (disabled by default)
		3574	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3575	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3576	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3577
		3578	F<ev.c> includes the backend files directly when enabled, so you only need
		3579	to compile this single file.
		3580
		3581	=head3 LIBEVENT COMPATIBILITY API
		3582
		3583	To include the libevent compatibility API, also include:
		3584
		3585	#include "event.c"
		3586
		3587	in the file including F<ev.c>, and:
		3588
		3589	#include "event.h"
		3590
		3591	in the files that want to use the libevent API. This also includes F<ev.h>.
		3592
		3593	You need the following additional files for this:
		3594
		3595	event.h
		3596	event.c
		3597
		3598	=head3 AUTOCONF SUPPORT
		3599
		3600	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3601	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3602	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3603	include F<config.h> and configure itself accordingly.
		3604
		3605	For this of course you need the m4 file:
		3606
		3607	libev.m4
		3608
		3609	=head2 PREPROCESSOR SYMBOLS/MACROS
		3610
		3611	Libev can be configured via a variety of preprocessor symbols you have to
		3612	define before including any of its files. The default in the absence of
		3613	autoconf is documented for every option.
		3614
		3615	=over 4
		3616
		3617	=item EV_STANDALONE
		3618
		3619	Must always be C<1> if you do not use autoconf configuration, which
		3620	keeps libev from including F<config.h>, and it also defines dummy
		3621	implementations for some libevent functions (such as logging, which is not
		3622	supported). It will also not define any of the structs usually found in
		3623	F<event.h> that are not directly supported by the libev core alone.
		3624
		3625	In standalone mode, libev will still try to automatically deduce the
		3626	configuration, but has to be more conservative.
		3627
		3628	=item EV_USE_MONOTONIC
		3629
		3630	If defined to be C<1>, libev will try to detect the availability of the
		3631	monotonic clock option at both compile time and runtime. Otherwise no
		3632	use of the monotonic clock option will be attempted. If you enable this,
		3633	you usually have to link against librt or something similar. Enabling it
		3634	when the functionality isn't available is safe, though, although you have
		3635	to make sure you link against any libraries where the C<clock_gettime>
		3636	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		3637
		3638	=item EV_USE_REALTIME
		3639
		3640	If defined to be C<1>, libev will try to detect the availability of the
		3641	real-time clock option at compile time (and assume its availability
		3642	at runtime if successful). Otherwise no use of the real-time clock
		3643	option will be attempted. This effectively replaces C<gettimeofday>
		3644	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		3645	correctness. See the note about libraries in the description of
		3646	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3647	C<EV_USE_CLOCK_SYSCALL>.
		3648
		3649	=item EV_USE_CLOCK_SYSCALL
		3650
		3651	If defined to be C<1>, libev will try to use a direct syscall instead
		3652	of calling the system-provided C<clock_gettime> function. This option
		3653	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3654	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3655	programs needlessly. Using a direct syscall is slightly slower (in
		3656	theory), because no optimised vdso implementation can be used, but avoids
		3657	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		3658	higher, as it simplifies linking (no need for C<-lrt>).
		3659
		3660	=item EV_USE_NANOSLEEP
		3661
		3662	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3663	and will use it for delays. Otherwise it will use C<select ()>.
		3664
		3665	=item EV_USE_EVENTFD
		3666
		3667	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3668	available and will probe for kernel support at runtime. This will improve
		3669	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3670	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3671	2.7 or newer, otherwise disabled.
		3672
		3673	=item EV_USE_SELECT
		3674
		3675	If undefined or defined to be C<1>, libev will compile in support for the
		3676	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3677	other method takes over, select will be it. Otherwise the select backend
		3678	will not be compiled in.
		3679
		3680	=item EV_SELECT_USE_FD_SET
		3681
		3682	If defined to C<1>, then the select backend will use the system C<fd_set>
		3683	structure. This is useful if libev doesn't compile due to a missing
		3684	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		3685	on exotic systems. This usually limits the range of file descriptors to
		3686	some low limit such as 1024 or might have other limitations (winsocket
		3687	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		3688	configures the maximum size of the C<fd_set>.
		3689
		3690	=item EV_SELECT_IS_WINSOCKET
		3691
		3692	When defined to C<1>, the select backend will assume that
		3693	select/socket/connect etc. don't understand file descriptors but
		3694	wants osf handles on win32 (this is the case when the select to
		3695	be used is the winsock select). This means that it will call
		3696	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3697	it is assumed that all these functions actually work on fds, even
		3698	on win32. Should not be defined on non-win32 platforms.
		3699
		3700	=item EV_FD_TO_WIN32_HANDLE(fd)
		3701
		3702	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3703	file descriptors to socket handles. When not defining this symbol (the
		3704	default), then libev will call C<_get_osfhandle>, which is usually
		3705	correct. In some cases, programs use their own file descriptor management,
		3706	in which case they can provide this function to map fds to socket handles.
		3707
		3708	=item EV_WIN32_HANDLE_TO_FD(handle)
		3709
		3710	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		3711	using the standard C<_open_osfhandle> function. For programs implementing
		3712	their own fd to handle mapping, overwriting this function makes it easier
		3713	to do so. This can be done by defining this macro to an appropriate value.
		3714
		3715	=item EV_WIN32_CLOSE_FD(fd)
		3716
		3717	If programs implement their own fd to handle mapping on win32, then this
		3718	macro can be used to override the C<close> function, useful to unregister
		3719	file descriptors again. Note that the replacement function has to close
		3720	the underlying OS handle.
		3721
		3722	=item EV_USE_POLL
		3723
		3724	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3725	backend. Otherwise it will be enabled on non-win32 platforms. It
		3726	takes precedence over select.
		3727
		3728	=item EV_USE_EPOLL
		3729
		3730	If defined to be C<1>, libev will compile in support for the Linux
		3731	C<epoll>(7) backend. Its availability will be detected at runtime,
		3732	otherwise another method will be used as fallback. This is the preferred
		3733	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3734	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3735
		3736	=item EV_USE_KQUEUE
		3737
		3738	If defined to be C<1>, libev will compile in support for the BSD style
		3739	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3740	otherwise another method will be used as fallback. This is the preferred
		3741	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3742	supports some types of fds correctly (the only platform we found that
		3743	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3744	not be used unless explicitly requested. The best way to use it is to find
		3745	out whether kqueue supports your type of fd properly and use an embedded
		3746	kqueue loop.
		3747
		3748	=item EV_USE_PORT
		3749
		3750	If defined to be C<1>, libev will compile in support for the Solaris
		3751	10 port style backend. Its availability will be detected at runtime,
		3752	otherwise another method will be used as fallback. This is the preferred
		3753	backend for Solaris 10 systems.
		3754
		3755	=item EV_USE_DEVPOLL
		3756
		3757	Reserved for future expansion, works like the USE symbols above.
		3758
		3759	=item EV_USE_INOTIFY
		3760
		3761	If defined to be C<1>, libev will compile in support for the Linux inotify
		3762	interface to speed up C<ev_stat> watchers. Its actual availability will
		3763	be detected at runtime. If undefined, it will be enabled if the headers
		3764	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3765
		3766	=item EV_ATOMIC_T
		3767
		3768	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3769	access is atomic with respect to other threads or signal contexts. No such
		3770	type is easily found in the C language, so you can provide your own type
		3771	that you know is safe for your purposes. It is used both for signal handler "locking"
		3772	as well as for signal and thread safety in C<ev_async> watchers.
		3773
		3774	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3775	(from F<signal.h>), which is usually good enough on most platforms.
		3776
		3777	=item EV_H
		3778
		3779	The name of the F<ev.h> header file used to include it. The default if
		3780	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3781	used to virtually rename the F<ev.h> header file in case of conflicts.
		3782
		3783	=item EV_CONFIG_H
		3784
		3785	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3786	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3787	C<EV_H>, above.
		3788
		3789	=item EV_EVENT_H
		3790
		3791	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3792	of how the F<event.h> header can be found, the default is C<"event.h">.
		3793
		3794	=item EV_PROTOTYPES
		3795
		3796	If defined to be C<0>, then F<ev.h> will not define any function
		3797	prototypes, but still define all the structs and other symbols. This is
		3798	occasionally useful if you want to provide your own wrapper functions
		3799	around libev functions.
		3800
		3801	=item EV_MULTIPLICITY
		3802
		3803	If undefined or defined to C<1>, then all event-loop-specific functions
		3804	will have the C<struct ev_loop *> as first argument, and you can create
		3805	additional independent event loops. Otherwise there will be no support
		3806	for multiple event loops and there is no first event loop pointer
		3807	argument. Instead, all functions act on the single default loop.
		3808
		3809	=item EV_MINPRI
		3810
		3811	=item EV_MAXPRI
		3812
		3813	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3814	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3815	provide for more priorities by overriding those symbols (usually defined
		3816	to be C<-2> and C<2>, respectively).
		3817
		3818	When doing priority-based operations, libev usually has to linearly search
		3819	all the priorities, so having many of them (hundreds) uses a lot of space
		3820	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3821	fine.
		3822
		3823	If your embedding application does not need any priorities, defining these
		3824	both to C<0> will save some memory and CPU.
		3825
		3826	=item EV_PERIODIC_ENABLE
		3827
		3828	If undefined or defined to be C<1>, then periodic timers are supported. If
		3829	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3830	code.
		3831
		3832	=item EV_IDLE_ENABLE
		3833
		3834	If undefined or defined to be C<1>, then idle watchers are supported. If
		3835	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3836	code.
		3837
		3838	=item EV_EMBED_ENABLE
		3839
		3840	If undefined or defined to be C<1>, then embed watchers are supported. If
		3841	defined to be C<0>, then they are not. Embed watchers rely on most other
		3842	watcher types, which therefore must not be disabled.
		3843
		3844	=item EV_STAT_ENABLE
		3845
		3846	If undefined or defined to be C<1>, then stat watchers are supported. If
		3847	defined to be C<0>, then they are not.
		3848
		3849	=item EV_FORK_ENABLE
		3850
		3851	If undefined or defined to be C<1>, then fork watchers are supported. If
		3852	defined to be C<0>, then they are not.
		3853
		3854	=item EV_ASYNC_ENABLE
		3855
		3856	If undefined or defined to be C<1>, then async watchers are supported. If
		3857	defined to be C<0>, then they are not.
		3858
		3859	=item EV_MINIMAL
		3860
		3861	If you need to shave off some kilobytes of code at the expense of some
		3862	speed (but with the full API), define this symbol to C<1>. Currently this
		3863	is used to override some inlining decisions, saves roughly 30% code size
		3864	on amd64. It also selects a much smaller 2-heap for timer management over
		3865	the default 4-heap.
		3866
		3867	You can save even more by disabling watcher types you do not need
		3868	and setting C<EV_MAXPRI> == C<EV_MINPRI>. Also, disabling C<assert>
		3869	(C<-DNDEBUG>) will usually reduce code size a lot.
		3870
		3871	Defining C<EV_MINIMAL> to C<2> will additionally reduce the core API to
		3872	provide a bare-bones event library. See C<ev.h> for details on what parts
		3873	of the API are still available, and do not complain if this subset changes
		3874	over time.
		3875
		3876	=item EV_NSIG
		3877
		3878	The highest supported signal number, +1 (or, the number of
		3879	signals): Normally, libev tries to deduce the maximum number of signals
		3880	automatically, but sometimes this fails, in which case it can be
		3881	specified. Also, using a lower number than detected (C<32> should be
		3882	good for about any system in existance) can save some memory, as libev
		3883	statically allocates some 12-24 bytes per signal number.
		3884
		3885	=item EV_PID_HASHSIZE
		3886
		3887	C<ev_child> watchers use a small hash table to distribute workload by
		3888	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
		3889	than enough. If you need to manage thousands of children you might want to
		3890	increase this value (I<must> be a power of two).
		3891
		3892	=item EV_INOTIFY_HASHSIZE
		3893
		3894	C<ev_stat> watchers use a small hash table to distribute workload by
		3895	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		3896	usually more than enough. If you need to manage thousands of C<ev_stat>
		3897	watchers you might want to increase this value (I<must> be a power of
		3898	two).
		3899
		3900	=item EV_USE_4HEAP
		3901
		3902	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3903	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3904	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3905	faster performance with many (thousands) of watchers.
		3906
		3907	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3908	(disabled).
		3909
		3910	=item EV_HEAP_CACHE_AT
		3911
		3912	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3913	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3914	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3915	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3916	but avoids random read accesses on heap changes. This improves performance
		3917	noticeably with many (hundreds) of watchers.
		3918
		3919	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3920	(disabled).
		3921
		3922	=item EV_VERIFY
		3923
		3924	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3925	be done: If set to C<0>, no internal verification code will be compiled
		3926	in. If set to C<1>, then verification code will be compiled in, but not
		3927	called. If set to C<2>, then the internal verification code will be
		3928	called once per loop, which can slow down libev. If set to C<3>, then the
		3929	verification code will be called very frequently, which will slow down
		3930	libev considerably.
		3931
		3932	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3933	C<0>.
		3934
		3935	=item EV_COMMON
		3936
		3937	By default, all watchers have a C<void *data> member. By redefining
		3938	this macro to a something else you can include more and other types of
		3939	members. You have to define it each time you include one of the files,
		3940	though, and it must be identical each time.
		3941
		3942	For example, the perl EV module uses something like this:
		3943
		3944	#define EV_COMMON \
		3945	SV *self; /* contains this struct */ \
		3946	SV *cb_sv, *fh /* note no trailing ";" */
		3947
		3948	=item EV_CB_DECLARE (type)
		3949
		3950	=item EV_CB_INVOKE (watcher, revents)
		3951
		3952	=item ev_set_cb (ev, cb)
		3953
		3954	Can be used to change the callback member declaration in each watcher,
		3955	and the way callbacks are invoked and set. Must expand to a struct member
		3956	definition and a statement, respectively. See the F<ev.h> header file for
		3957	their default definitions. One possible use for overriding these is to
		3958	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		3959	method calls instead of plain function calls in C++.
		3960
		3961	=back
		3962
		3963	=head2 EXPORTED API SYMBOLS
		3964
		3965	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3966	exported symbols, you can use the provided F<Symbol.*> files which list
		3967	all public symbols, one per line:
		3968
		3969	Symbols.ev for libev proper
		3970	Symbols.event for the libevent emulation
		3971
		3972	This can also be used to rename all public symbols to avoid clashes with
		3973	multiple versions of libev linked together (which is obviously bad in
		3974	itself, but sometimes it is inconvenient to avoid this).
		3975
		3976	A sed command like this will create wrapper C<#define>'s that you need to
		3977	include before including F<ev.h>:
		3978
		3979	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3980
		3981	This would create a file F<wrap.h> which essentially looks like this:
		3982
		3983	#define ev_backend myprefix_ev_backend
		3984	#define ev_check_start myprefix_ev_check_start
		3985	#define ev_check_stop myprefix_ev_check_stop
		3986	...
		3987
		3988	=head2 EXAMPLES
		3989
		3990	For a real-world example of a program the includes libev
		3991	verbatim, you can have a look at the EV perl module
		3992	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		3993	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		3994	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		3995	will be compiled. It is pretty complex because it provides its own header
		3996	file.
		3997
		3998	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		3999	that everybody includes and which overrides some configure choices:
		4000
		4001	#define EV_MINIMAL 1
		4002	#define EV_USE_POLL 0
		4003	#define EV_MULTIPLICITY 0
		4004	#define EV_PERIODIC_ENABLE 0
		4005	#define EV_STAT_ENABLE 0
		4006	#define EV_FORK_ENABLE 0
		4007	#define EV_CONFIG_H <config.h>
		4008	#define EV_MINPRI 0
		4009	#define EV_MAXPRI 0
		4010
		4011	#include "ev++.h"
		4012
		4013	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4014
		4015	#include "ev_cpp.h"
		4016	#include "ev.c"
		4017
		4018	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		4019
		4020	=head2 THREADS AND COROUTINES
		4021
		4022	=head3 THREADS
		4023
		4024	All libev functions are reentrant and thread-safe unless explicitly
		4025	documented otherwise, but libev implements no locking itself. This means
		4026	that you can use as many loops as you want in parallel, as long as there
		4027	are no concurrent calls into any libev function with the same loop
		4028	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4029	of course): libev guarantees that different event loops share no data
		4030	structures that need any locking.
		4031
		4032	Or to put it differently: calls with different loop parameters can be done
		4033	concurrently from multiple threads, calls with the same loop parameter
		4034	must be done serially (but can be done from different threads, as long as
		4035	only one thread ever is inside a call at any point in time, e.g. by using
		4036	a mutex per loop).
		4037
		4038	Specifically to support threads (and signal handlers), libev implements
		4039	so-called C<ev_async> watchers, which allow some limited form of
		4040	concurrency on the same event loop, namely waking it up "from the
		4041	outside".
		4042
		4043	If you want to know which design (one loop, locking, or multiple loops
		4044	without or something else still) is best for your problem, then I cannot
		4045	help you, but here is some generic advice:
		4046
		4047	=over 4
		4048
		4049	=item * most applications have a main thread: use the default libev loop
		4050	in that thread, or create a separate thread running only the default loop.
		4051
		4052	This helps integrating other libraries or software modules that use libev
		4053	themselves and don't care/know about threading.
		4054
		4055	=item * one loop per thread is usually a good model.
		4056
		4057	Doing this is almost never wrong, sometimes a better-performance model
		4058	exists, but it is always a good start.
		4059
		4060	=item * other models exist, such as the leader/follower pattern, where one
		4061	loop is handed through multiple threads in a kind of round-robin fashion.
		4062
		4063	Choosing a model is hard - look around, learn, know that usually you can do
		4064	better than you currently do :-)
		4065
		4066	=item * often you need to talk to some other thread which blocks in the
		4067	event loop.
		4068
		4069	C<ev_async> watchers can be used to wake them up from other threads safely
		4070	(or from signal contexts...).
		4071
		4072	An example use would be to communicate signals or other events that only
		4073	work in the default loop by registering the signal watcher with the
		4074	default loop and triggering an C<ev_async> watcher from the default loop
		4075	watcher callback into the event loop interested in the signal.
		4076
		4077	=back
		4078
		4079	=head4 THREAD LOCKING EXAMPLE
		4080
		4081	Here is a fictitious example of how to run an event loop in a different
		4082	thread than where callbacks are being invoked and watchers are
		4083	created/added/removed.
		4084
		4085	For a real-world example, see the C<EV::Loop::Async> perl module,
		4086	which uses exactly this technique (which is suited for many high-level
		4087	languages).
		4088
		4089	The example uses a pthread mutex to protect the loop data, a condition
		4090	variable to wait for callback invocations, an async watcher to notify the
		4091	event loop thread and an unspecified mechanism to wake up the main thread.
		4092
		4093	First, you need to associate some data with the event loop:
		4094
		4095	typedef struct {
		4096	mutex_t lock; /* global loop lock */
		4097	ev_async async_w;
		4098	thread_t tid;
		4099	cond_t invoke_cv;
		4100	} userdata;
		4101
		4102	void prepare_loop (EV_P)
		4103	{
		4104	// for simplicity, we use a static userdata struct.
		4105	static userdata u;
		4106
		4107	ev_async_init (&u->async_w, async_cb);
		4108	ev_async_start (EV_A_ &u->async_w);
		4109
		4110	pthread_mutex_init (&u->lock, 0);
		4111	pthread_cond_init (&u->invoke_cv, 0);
		4112
		4113	// now associate this with the loop
		4114	ev_set_userdata (EV_A_ u);
		4115	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		4116	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		4117
		4118	// then create the thread running ev_loop
		4119	pthread_create (&u->tid, 0, l_run, EV_A);
		4120	}
		4121
		4122	The callback for the C<ev_async> watcher does nothing: the watcher is used
		4123	solely to wake up the event loop so it takes notice of any new watchers
		4124	that might have been added:
		4125
		4126	static void
		4127	async_cb (EV_P_ ev_async *w, int revents)
		4128	{
		4129	// just used for the side effects
		4130	}
		4131
		4132	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		4133	protecting the loop data, respectively.
		4134
		4135	static void
		4136	l_release (EV_P)
		4137	{
		4138	userdata *u = ev_userdata (EV_A);
		4139	pthread_mutex_unlock (&u->lock);
		4140	}
		4141
		4142	static void
		4143	l_acquire (EV_P)
		4144	{
		4145	userdata *u = ev_userdata (EV_A);
		4146	pthread_mutex_lock (&u->lock);
		4147	}
		4148
		4149	The event loop thread first acquires the mutex, and then jumps straight
		4150	into C<ev_loop>:
		4151
		4152	void *
		4153	l_run (void *thr_arg)
		4154	{
		4155	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		4156
		4157	l_acquire (EV_A);
		4158	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		4159	ev_loop (EV_A_ 0);
		4160	l_release (EV_A);
		4161
		4162	return 0;
		4163	}
		4164
		4165	Instead of invoking all pending watchers, the C<l_invoke> callback will
		4166	signal the main thread via some unspecified mechanism (signals? pipe
		4167	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		4168	have been called (in a while loop because a) spurious wakeups are possible
		4169	and b) skipping inter-thread-communication when there are no pending
		4170	watchers is very beneficial):
		4171
		4172	static void
		4173	l_invoke (EV_P)
		4174	{
		4175	userdata *u = ev_userdata (EV_A);
		4176
		4177	while (ev_pending_count (EV_A))
		4178	{
		4179	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		4180	pthread_cond_wait (&u->invoke_cv, &u->lock);
		4181	}
		4182	}
		4183
		4184	Now, whenever the main thread gets told to invoke pending watchers, it
		4185	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		4186	thread to continue:
		4187
		4188	static void
		4189	real_invoke_pending (EV_P)
		4190	{
		4191	userdata *u = ev_userdata (EV_A);
		4192
		4193	pthread_mutex_lock (&u->lock);
		4194	ev_invoke_pending (EV_A);
		4195	pthread_cond_signal (&u->invoke_cv);
		4196	pthread_mutex_unlock (&u->lock);
		4197	}
		4198
		4199	Whenever you want to start/stop a watcher or do other modifications to an
		4200	event loop, you will now have to lock:
		4201
		4202	ev_timer timeout_watcher;
		4203	userdata *u = ev_userdata (EV_A);
		4204
		4205	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		4206
		4207	pthread_mutex_lock (&u->lock);
		4208	ev_timer_start (EV_A_ &timeout_watcher);
		4209	ev_async_send (EV_A_ &u->async_w);
		4210	pthread_mutex_unlock (&u->lock);
		4211
		4212	Note that sending the C<ev_async> watcher is required because otherwise
		4213	an event loop currently blocking in the kernel will have no knowledge
		4214	about the newly added timer. By waking up the loop it will pick up any new
		4215	watchers in the next event loop iteration.
		4216
		4217	=head3 COROUTINES
		4218
		4219	Libev is very accommodating to coroutines ("cooperative threads"):
		4220	libev fully supports nesting calls to its functions from different
		4221	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		4222	different coroutines, and switch freely between both coroutines running
		4223	the loop, as long as you don't confuse yourself). The only exception is
		4224	that you must not do this from C<ev_periodic> reschedule callbacks.
		4225
		4226	Care has been taken to ensure that libev does not keep local state inside
		4227	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		4228	they do not call any callbacks.
		4229
		4230	=head2 COMPILER WARNINGS
		4231
		4232	Depending on your compiler and compiler settings, you might get no or a
		4233	lot of warnings when compiling libev code. Some people are apparently
		4234	scared by this.
		4235
		4236	However, these are unavoidable for many reasons. For one, each compiler
		4237	has different warnings, and each user has different tastes regarding
		4238	warning options. "Warn-free" code therefore cannot be a goal except when
		4239	targeting a specific compiler and compiler-version.
		4240
		4241	Another reason is that some compiler warnings require elaborate
		4242	workarounds, or other changes to the code that make it less clear and less
		4243	maintainable.
		4244
		4245	And of course, some compiler warnings are just plain stupid, or simply
		4246	wrong (because they don't actually warn about the condition their message
		4247	seems to warn about). For example, certain older gcc versions had some
		4248	warnings that resulted an extreme number of false positives. These have
		4249	been fixed, but some people still insist on making code warn-free with
		4250	such buggy versions.
		4251
		4252	While libev is written to generate as few warnings as possible,
		4253	"warn-free" code is not a goal, and it is recommended not to build libev
		4254	with any compiler warnings enabled unless you are prepared to cope with
		4255	them (e.g. by ignoring them). Remember that warnings are just that:
		4256	warnings, not errors, or proof of bugs.
		4257
		4258
		4259	=head2 VALGRIND
		4260
		4261	Valgrind has a special section here because it is a popular tool that is
		4262	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4263
		4264	If you think you found a bug (memory leak, uninitialised data access etc.)
		4265	in libev, then check twice: If valgrind reports something like:
		4266
		4267	==2274== definitely lost: 0 bytes in 0 blocks.
		4268	==2274== possibly lost: 0 bytes in 0 blocks.
		4269	==2274== still reachable: 256 bytes in 1 blocks.
		4270
		4271	Then there is no memory leak, just as memory accounted to global variables
		4272	is not a memleak - the memory is still being referenced, and didn't leak.
		4273
		4274	Similarly, under some circumstances, valgrind might report kernel bugs
		4275	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4276	although an acceptable workaround has been found here), or it might be
		4277	confused.
		4278
		4279	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4280	make it into some kind of religion.
		4281
		4282	If you are unsure about something, feel free to contact the mailing list
		4283	with the full valgrind report and an explanation on why you think this
		4284	is a bug in libev (best check the archives, too :). However, don't be
		4285	annoyed when you get a brisk "this is no bug" answer and take the chance
		4286	of learning how to interpret valgrind properly.
		4287
		4288	If you need, for some reason, empty reports from valgrind for your project
		4289	I suggest using suppression lists.
		4290
		4291
		4292	=head1 PORTABILITY NOTES
		4293
		4294	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4295
		4296	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		4297	requires, and its I/O model is fundamentally incompatible with the POSIX
		4298	model. Libev still offers limited functionality on this platform in
		4299	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		4300	descriptors. This only applies when using Win32 natively, not when using
		4301	e.g. cygwin.
		4302
		4303	Lifting these limitations would basically require the full
		4304	re-implementation of the I/O system. If you are into these kinds of
		4305	things, then note that glib does exactly that for you in a very portable
		4306	way (note also that glib is the slowest event library known to man).
		4307
		4308	There is no supported compilation method available on windows except
		4309	embedding it into other applications.
		4310
		4311	Sensible signal handling is officially unsupported by Microsoft - libev
		4312	tries its best, but under most conditions, signals will simply not work.
		4313
		4314	Not a libev limitation but worth mentioning: windows apparently doesn't
		4315	accept large writes: instead of resulting in a partial write, windows will
		4316	either accept everything or return C<ENOBUFS> if the buffer is too large,
		4317	so make sure you only write small amounts into your sockets (less than a
		4318	megabyte seems safe, but this apparently depends on the amount of memory
		4319	available).
		4320
		4321	Due to the many, low, and arbitrary limits on the win32 platform and
		4322	the abysmal performance of winsockets, using a large number of sockets
		4323	is not recommended (and not reasonable). If your program needs to use
		4324	more than a hundred or so sockets, then likely it needs to use a totally
		4325	different implementation for windows, as libev offers the POSIX readiness
		4326	notification model, which cannot be implemented efficiently on windows
		4327	(due to Microsoft monopoly games).
		4328
		4329	A typical way to use libev under windows is to embed it (see the embedding
		4330	section for details) and use the following F<evwrap.h> header file instead
		4331	of F<ev.h>:
		4332
		4333	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4334	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4335
		4336	#include "ev.h"
		4337
		4338	And compile the following F<evwrap.c> file into your project (make sure
		4339	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4340
		4341	#include "evwrap.h"
		4342	#include "ev.c"
		4343
		4344	=over 4
		4345
		4346	=item The winsocket select function
		4347
		4348	The winsocket C<select> function doesn't follow POSIX in that it
		4349	requires socket I<handles> and not socket I<file descriptors> (it is
		4350	also extremely buggy). This makes select very inefficient, and also
		4351	requires a mapping from file descriptors to socket handles (the Microsoft
		4352	C runtime provides the function C<_open_osfhandle> for this). See the
		4353	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		4354	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		4355
		4356	The configuration for a "naked" win32 using the Microsoft runtime
		4357	libraries and raw winsocket select is:
		4358
		4359	#define EV_USE_SELECT 1
		4360	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		4361
		4362	Note that winsockets handling of fd sets is O(n), so you can easily get a
		4363	complexity in the O(n²) range when using win32.
		4364
		4365	=item Limited number of file descriptors
		4366
		4367	Windows has numerous arbitrary (and low) limits on things.
		4368
		4369	Early versions of winsocket's select only supported waiting for a maximum
		4370	of C<64> handles (probably owning to the fact that all windows kernels
		4371	can only wait for C<64> things at the same time internally; Microsoft
		4372	recommends spawning a chain of threads and wait for 63 handles and the
		4373	previous thread in each. Sounds great!).
		4374
		4375	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		4376	to some high number (e.g. C<2048>) before compiling the winsocket select
		4377	call (which might be in libev or elsewhere, for example, perl and many
		4378	other interpreters do their own select emulation on windows).
		4379
		4380	Another limit is the number of file descriptors in the Microsoft runtime
		4381	libraries, which by default is C<64> (there must be a hidden I<64>
		4382	fetish or something like this inside Microsoft). You can increase this
		4383	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		4384	(another arbitrary limit), but is broken in many versions of the Microsoft
		4385	runtime libraries. This might get you to about C<512> or C<2048> sockets
		4386	(depending on windows version and/or the phase of the moon). To get more,
		4387	you need to wrap all I/O functions and provide your own fd management, but
		4388	the cost of calling select (O(n²)) will likely make this unworkable.
		4389
		4390	=back
		4391
		4392	=head2 PORTABILITY REQUIREMENTS
		4393
		4394	In addition to a working ISO-C implementation and of course the
		4395	backend-specific APIs, libev relies on a few additional extensions:
		4396
		4397	=over 4
		4398
		4399	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4400	calling conventions regardless of C<ev_watcher_type *>.
		4401
		4402	Libev assumes not only that all watcher pointers have the same internal
		4403	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4404	assumes that the same (machine) code can be used to call any watcher
		4405	callback: The watcher callbacks have different type signatures, but libev
		4406	calls them using an C<ev_watcher *> internally.
		4407
		4408	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4409
		4410	The type C<sig_atomic_t volatile> (or whatever is defined as
		4411	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4412	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4413	believed to be sufficiently portable.
		4414
		4415	=item C<sigprocmask> must work in a threaded environment
		4416
		4417	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4418	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		4419	pthread implementations will either allow C<sigprocmask> in the "main
		4420	thread" or will block signals process-wide, both behaviours would
		4421	be compatible with libev. Interaction between C<sigprocmask> and
		4422	C<pthread_sigmask> could complicate things, however.
		4423
		4424	The most portable way to handle signals is to block signals in all threads
		4425	except the initial one, and run the default loop in the initial thread as
		4426	well.
		4427
		4428	=item C<long> must be large enough for common memory allocation sizes
		4429
		4430	To improve portability and simplify its API, libev uses C<long> internally
		4431	instead of C<size_t> when allocating its data structures. On non-POSIX
		4432	systems (Microsoft...) this might be unexpectedly low, but is still at
		4433	least 31 bits everywhere, which is enough for hundreds of millions of
		4434	watchers.
		4435
		4436	=item C<double> must hold a time value in seconds with enough accuracy
		4437
		4438	The type C<double> is used to represent timestamps. It is required to
		4439	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		4440	enough for at least into the year 4000. This requirement is fulfilled by
		4441	implementations implementing IEEE 754, which is basically all existing
		4442	ones. With IEEE 754 doubles, you get microsecond accuracy until at least
		4443	2200.
		4444
		4445	=back
		4446
		4447	If you know of other additional requirements drop me a note.
		4448
		4449
		4450	=head1 ALGORITHMIC COMPLEXITIES
		4451
		4452	In this section the complexities of (many of) the algorithms used inside
		4453	libev will be documented. For complexity discussions about backends see
		4454	the documentation for C<ev_default_init>.
		4455
		4456	All of the following are about amortised time: If an array needs to be
		4457	extended, libev needs to realloc and move the whole array, but this
		4458	happens asymptotically rarer with higher number of elements, so O(1) might
		4459	mean that libev does a lengthy realloc operation in rare cases, but on
		4460	average it is much faster and asymptotically approaches constant time.
		4461
		4462	=over 4
		4463
		4464	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		4465
		4466	This means that, when you have a watcher that triggers in one hour and
		4467	there are 100 watchers that would trigger before that, then inserting will
		4468	have to skip roughly seven (C<ld 100>) of these watchers.
		4469
		4470	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		4471
		4472	That means that changing a timer costs less than removing/adding them,
		4473	as only the relative motion in the event queue has to be paid for.
		4474
		4475	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		4476
		4477	These just add the watcher into an array or at the head of a list.
		4478
		4479	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		4480
		4481	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		4482
		4483	These watchers are stored in lists, so they need to be walked to find the
		4484	correct watcher to remove. The lists are usually short (you don't usually
		4485	have many watchers waiting for the same fd or signal: one is typical, two
		4486	is rare).
		4487
		4488	=item Finding the next timer in each loop iteration: O(1)
		4489
		4490	By virtue of using a binary or 4-heap, the next timer is always found at a
		4491	fixed position in the storage array.
		4492
		4493	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		4494
		4495	A change means an I/O watcher gets started or stopped, which requires
		4496	libev to recalculate its status (and possibly tell the kernel, depending
		4497	on backend and whether C<ev_io_set> was used).
		4498
		4499	=item Activating one watcher (putting it into the pending state): O(1)
		4500
		4501	=item Priority handling: O(number_of_priorities)
		4502
		4503	Priorities are implemented by allocating some space for each
		4504	priority. When doing priority-based operations, libev usually has to
		4505	linearly search all the priorities, but starting/stopping and activating
		4506	watchers becomes O(1) with respect to priority handling.
		4507
		4508	=item Sending an ev_async: O(1)
		4509
		4510	=item Processing ev_async_send: O(number_of_async_watchers)
		4511
		4512	=item Processing signals: O(max_signal_number)
		4513
		4514	Sending involves a system call I<iff> there were no other C<ev_async_send>
		4515	calls in the current loop iteration. Checking for async and signal events
		4516	involves iterating over all running async watchers or all signal numbers.
		4517
		4518	=back
		4519
		4520
		4521	=head1 GLOSSARY
		4522
		4523	=over 4
		4524
		4525	=item active
		4526
		4527	A watcher is active as long as it has been started (has been attached to
		4528	an event loop) but not yet stopped (disassociated from the event loop).
		4529
		4530	=item application
		4531
		4532	In this document, an application is whatever is using libev.
		4533
		4534	=item callback
		4535
		4536	The address of a function that is called when some event has been
		4537	detected. Callbacks are being passed the event loop, the watcher that
		4538	received the event, and the actual event bitset.
		4539
		4540	=item callback invocation
		4541
		4542	The act of calling the callback associated with a watcher.
		4543
		4544	=item event
		4545
		4546	A change of state of some external event, such as data now being available
		4547	for reading on a file descriptor, time having passed or simply not having
		4548	any other events happening anymore.
		4549
		4550	In libev, events are represented as single bits (such as C<EV_READ> or
		4551	C<EV_TIMEOUT>).
		4552
		4553	=item event library
		4554
		4555	A software package implementing an event model and loop.
		4556
		4557	=item event loop
		4558
		4559	An entity that handles and processes external events and converts them
		4560	into callback invocations.
		4561
		4562	=item event model
		4563
		4564	The model used to describe how an event loop handles and processes
		4565	watchers and events.
		4566
		4567	=item pending
		4568
		4569	A watcher is pending as soon as the corresponding event has been detected,
		4570	and stops being pending as soon as the watcher will be invoked or its
		4571	pending status is explicitly cleared by the application.
		4572
		4573	A watcher can be pending, but not active. Stopping a watcher also clears
		4574	its pending status.
		4575
		4576	=item real time
		4577
		4578	The physical time that is observed. It is apparently strictly monotonic :)
		4579
		4580	=item wall-clock time
		4581
		4582	The time and date as shown on clocks. Unlike real time, it can actually
		4583	be wrong and jump forwards and backwards, e.g. when the you adjust your
		4584	clock.
		4585
		4586	=item watcher
		4587
		4588	A data structure that describes interest in certain events. Watchers need
		4589	to be started (attached to an event loop) before they can receive events.
		4590
		4591	=item watcher invocation
		4592
		4593	The act of calling the callback associated with a watcher.
		4594
		4595	=back
781		4596
782	=head1 AUTHOR	4597	=head1 AUTHOR
783		4598
784	Marc Lehmann <libev@schmorp.de>.	4599	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
785		4600

Diff Legend

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