ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.8 by root, Mon Nov 12 08:20:02 2007 UTC vs.
Revision 1.315 by root, Fri Oct 22 09:33:14 2010 UTC

…		…
2		2
3	libev - a high performance full-featured event loop written in C	3	libev - a high performance full-featured event loop written in C
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	#include <ev.h>	7	#include <ev.h>
8		8
9	=head1 DESCRIPTION	9	=head2 EXAMPLE PROGRAM
		10
		11	// a single header file is required
		12	#include <ev.h>
		13
		14	#include <stdio.h> // for puts
		15
		16	// every watcher type has its own typedef'd struct
		17	// with the name ev_TYPE
		18	ev_io stdin_watcher;
		19	ev_timer timeout_watcher;
		20
		21	// all watcher callbacks have a similar signature
		22	// this callback is called when data is readable on stdin
		23	static void
		24	stdin_cb (EV_P_ ev_io *w, int revents)
		25	{
		26	puts ("stdin ready");
		27	// for one-shot events, one must manually stop the watcher
		28	// with its corresponding stop function.
		29	ev_io_stop (EV_A_ w);
		30
		31	// this causes all nested ev_run's to stop iterating
		32	ev_break (EV_A_ EVBREAK_ALL);
		33	}
		34
		35	// another callback, this time for a time-out
		36	static void
		37	timeout_cb (EV_P_ ev_timer *w, int revents)
		38	{
		39	puts ("timeout");
		40	// this causes the innermost ev_run to stop iterating
		41	ev_break (EV_A_ EVBREAK_ONE);
		42	}
		43
		44	int
		45	main (void)
		46	{
		47	// use the default event loop unless you have special needs
		48	struct ev_loop *loop = ev_default_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_run (loop, 0);
		62
		63	// unloop was called, so exit
		64	return 0;
		65	}
		66
		67	=head1 ABOUT THIS DOCUMENT
		68
		69	This document documents the libev software package.
		70
		71	The newest version of this document is also available as an html-formatted
		72	web page you might find easier to navigate when reading it for the first
		73	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		74
		75	While this document tries to be as complete as possible in documenting
		76	libev, its usage and the rationale behind its design, it is not a tutorial
		77	on event-based programming, nor will it introduce event-based programming
		78	with libev.
		79
		80	Familiarity with event based programming techniques in general is assumed
		81	throughout this document.
		82
		83	=head1 ABOUT LIBEV
10		84
11	Libev is an event loop: you register interest in certain events (such as a	85	Libev is an event loop: you register interest in certain events (such as a
12	file descriptor being readable or a timeout occuring), and it will manage	86	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	87	these event sources and provide your program with events.
14		88
15	To do this, it must take more or less complete control over your process	89	To do this, it must take more or less complete control over your process
16	(or thread) by executing the I<event loop> handler, and will then	90	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	91	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	93	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	94	watchers>, which are relatively small C structures you initialise with the
21	details of the event, and then hand it over to libev by I<starting> the	95	details of the event, and then hand it over to libev by I<starting> the
22	watcher.	96	watcher.
23		97
24	=head1 FEATURES	98	=head2 FEATURES
25		99
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	100	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27	kqueue mechanisms for file descriptor events, relative timers, absolute	101	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28	timers with customised rescheduling, signal events, process status change	102	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
		103	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
		104	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
		105	timers (C<ev_timer>), absolute timers with customised rescheduling
		106	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
29	events (related to SIGCHLD), and event watchers dealing with the event	107	change events (C<ev_child>), and event watchers dealing with the event
30	loop mechanism itself (idle, prepare and check watchers). It also is quite	108	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
		109	C<ev_check> watchers) as well as file watchers (C<ev_stat>) and even
		110	limited support for fork events (C<ev_fork>).
		111
		112	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	113	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	114	for example).
33		115
34	=head1 CONVENTIONS	116	=head2 CONVENTIONS
35		117
36	Libev is very configurable. In this manual the default configuration	118	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	119	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	120	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	121	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	122	for multiple event loops, then all functions taking an initial argument of
41	argument of name C<loop> (which is always of type C<struct ev_loop *>)	123	name C<loop> (which is always of type C<struct ev_loop *>) will not have
42	will not have this argument.	124	this argument.
43		125
44	=head1 TIME AND OTHER GLOBAL FUNCTIONS	126	=head2 TIME REPRESENTATION
45		127
46	Libev represents time as a single floating point number, representing the	128	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	129	the (fractional) number of seconds since the (POSIX) epoch (in practise
48	the beginning of 1970, details are complicated, don't ask). This type is	130	somewhere near the beginning of 1970, details are complicated, don't
49	called C<ev_tstamp>, which is what you should use too. It usually aliases	131	ask). This type is called C<ev_tstamp>, which is what you should use
50	to the double type in C.	132	too. It usually aliases to the C<double> type in C. When you need to do
		133	any calculations on it, you should treat it as some floating point value.
		134
		135	Unlike the name component C<stamp> might indicate, it is also used for
		136	time differences (e.g. delays) throughout libev.
		137
		138	=head1 ERROR HANDLING
		139
		140	Libev knows three classes of errors: operating system errors, usage errors
		141	and internal errors (bugs).
		142
		143	When libev catches an operating system error it cannot handle (for example
		144	a system call indicating a condition libev cannot fix), it calls the callback
		145	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		146	abort. The default is to print a diagnostic message and to call C<abort
		147	()>.
		148
		149	When libev detects a usage error such as a negative timer interval, then
		150	it will print a diagnostic message and abort (via the C<assert> mechanism,
		151	so C<NDEBUG> will disable this checking): these are programming errors in
		152	the libev caller and need to be fixed there.
		153
		154	Libev also has a few internal error-checking C<assert>ions, and also has
		155	extensive consistency checking code. These do not trigger under normal
		156	circumstances, as they indicate either a bug in libev or worse.
		157
		158
		159	=head1 GLOBAL FUNCTIONS
		160
		161	These functions can be called anytime, even before initialising the
		162	library in any way.
51		163
52	=over 4	164	=over 4
53		165
54	=item ev_tstamp ev_time ()	166	=item ev_tstamp ev_time ()
55		167
56	Returns the current time as libev would use it.	168	Returns the current time as libev would use it. Please note that the
		169	C<ev_now> function is usually faster and also often returns the timestamp
		170	you actually want to know.
		171
		172	=item ev_sleep (ev_tstamp interval)
		173
		174	Sleep for the given interval: The current thread will be blocked until
		175	either it is interrupted or the given time interval has passed. Basically
		176	this is a sub-second-resolution C<sleep ()>.
57		177
58	=item int ev_version_major ()	178	=item int ev_version_major ()
59		179
60	=item int ev_version_minor ()	180	=item int ev_version_minor ()
61		181
62	You can find out the major and minor version numbers of the library	182	You can find out the major and minor ABI version numbers of the library
63	you linked against by calling the functions C<ev_version_major> and	183	you linked against by calling the functions C<ev_version_major> and
64	C<ev_version_minor>. If you want, you can compare against the global	184	C<ev_version_minor>. If you want, you can compare against the global
65	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	185	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
66	version of the library your program was compiled against.	186	version of the library your program was compiled against.
67		187
		188	These version numbers refer to the ABI version of the library, not the
		189	release version.
		190
68	Usually, its a good idea to terminate if the major versions mismatch,	191	Usually, it's a good idea to terminate if the major versions mismatch,
69	as this indicates an incompatible change. Minor versions are usually	192	as this indicates an incompatible change. Minor versions are usually
70	compatible to older versions, so a larger minor version alone is usually	193	compatible to older versions, so a larger minor version alone is usually
71	not a problem.	194	not a problem.
72		195
		196	Example: Make sure we haven't accidentally been linked against the wrong
		197	version (note, however, that this will not detect ABI mismatches :).
		198
		199	assert (("libev version mismatch",
		200	ev_version_major () == EV_VERSION_MAJOR
		201	&& ev_version_minor () >= EV_VERSION_MINOR));
		202
		203	=item unsigned int ev_supported_backends ()
		204
		205	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		206	value) compiled into this binary of libev (independent of their
		207	availability on the system you are running on). See C<ev_default_loop> for
		208	a description of the set values.
		209
		210	Example: make sure we have the epoll method, because yeah this is cool and
		211	a must have and can we have a torrent of it please!!!11
		212
		213	assert (("sorry, no epoll, no sex",
		214	ev_supported_backends () & EVBACKEND_EPOLL));
		215
		216	=item unsigned int ev_recommended_backends ()
		217
		218	Return the set of all backends compiled into this binary of libev and also
		219	recommended for this platform. This set is often smaller than the one
		220	returned by C<ev_supported_backends>, as for example kqueue is broken on
		221	most BSDs and will not be auto-detected unless you explicitly request it
		222	(assuming you know what you are doing). This is the set of backends that
		223	libev will probe for if you specify no backends explicitly.
		224
		225	=item unsigned int ev_embeddable_backends ()
		226
		227	Returns the set of backends that are embeddable in other event loops. This
		228	is the theoretical, all-platform, value. To find which backends
		229	might be supported on the current system, you would need to look at
		230	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
		231	recommended ones.
		232
		233	See the description of C<ev_embed> watchers for more info.
		234
73	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	235	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
74		236
75	Sets the allocation function to use (the prototype is similar to the	237	Sets the allocation function to use (the prototype is similar - the
76	realloc C function, the semantics are identical). It is used to allocate	238	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
77	and free memory (no surprises here). If it returns zero when memory	239	used to allocate and free memory (no surprises here). If it returns zero
78	needs to be allocated, the library might abort or take some potentially	240	when memory needs to be allocated (C<size != 0>), the library might abort
79	destructive action. The default is your system realloc function.	241	or take some potentially destructive action.
		242
		243	Since some systems (at least OpenBSD and Darwin) fail to implement
		244	correct C<realloc> semantics, libev will use a wrapper around the system
		245	C<realloc> and C<free> functions by default.
80		246
81	You could override this function in high-availability programs to, say,	247	You could override this function in high-availability programs to, say,
82	free some memory if it cannot allocate memory, to use a special allocator,	248	free some memory if it cannot allocate memory, to use a special allocator,
83	or even to sleep a while and retry until some memory is available.	249	or even to sleep a while and retry until some memory is available.
84		250
		251	Example: Replace the libev allocator with one that waits a bit and then
		252	retries (example requires a standards-compliant C<realloc>).
		253
		254	static void *
		255	persistent_realloc (void *ptr, size_t size)
		256	{
		257	for (;;)
		258	{
		259	void *newptr = realloc (ptr, size);
		260
		261	if (newptr)
		262	return newptr;
		263
		264	sleep (60);
		265	}
		266	}
		267
		268	...
		269	ev_set_allocator (persistent_realloc);
		270
85	=item ev_set_syserr_cb (void (*cb)(const char *msg));	271	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
86		272
87	Set the callback function to call on a retryable syscall error (such	273	Set the callback function to call on a retryable system call error (such
88	as failed select, poll, epoll_wait). The message is a printable string	274	as failed select, poll, epoll_wait). The message is a printable string
89	indicating the system call or subsystem causing the problem. If this	275	indicating the system call or subsystem causing the problem. If this
90	callback is set, then libev will expect it to remedy the sitution, no	276	callback is set, then libev will expect it to remedy the situation, no
91	matter what, when it returns. That is, libev will generally retry the	277	matter what, when it returns. That is, libev will generally retry the
92	requested operation, or, if the condition doesn't go away, do bad stuff	278	requested operation, or, if the condition doesn't go away, do bad stuff
93	(such as abort).	279	(such as abort).
94		280
		281	Example: This is basically the same thing that libev does internally, too.
		282
		283	static void
		284	fatal_error (const char *msg)
		285	{
		286	perror (msg);
		287	abort ();
		288	}
		289
		290	...
		291	ev_set_syserr_cb (fatal_error);
		292
95	=back	293	=back
96		294
97	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	295	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
98		296
99	An event loop is described by a C<struct ev_loop *>. The library knows two	297	An event loop is described by a C<struct ev_loop *> (the C<struct> is
100	types of such loops, the I<default> loop, which supports signals and child	298	I<not> optional in this case unless libev 3 compatibility is disabled, as
101	events, and dynamically created loops which do not.	299	libev 3 had an C<ev_loop> function colliding with the struct name).
102		300
103	If you use threads, a common model is to run the default event loop	301	The library knows two types of such loops, the I<default> loop, which
104	in your main thread (or in a separate thrad) and for each thread you	302	supports signals and child events, and dynamically created event loops
105	create, you also create another event loop. Libev itself does no locking	303	which do not.
106	whatsoever, so if you mix calls to the same event loop in different
107	threads, make sure you lock (this is usually a bad idea, though, even if
108	done correctly, because its hideous and inefficient).
109		304
110	=over 4	305	=over 4
111		306
112	=item struct ev_loop *ev_default_loop (unsigned int flags)	307	=item struct ev_loop *ev_default_loop (unsigned int flags)
113		308
114	This will initialise the default event loop if it hasn't been initialised	309	This will initialise the default event loop if it hasn't been initialised
115	yet and return it. If the default loop could not be initialised, returns	310	yet and return it. If the default loop could not be initialised, returns
116	false. If it already was initialised it simply returns it (and ignores the	311	false. If it already was initialised it simply returns it (and ignores the
117	flags).	312	flags. If that is troubling you, check C<ev_backend ()> afterwards).
118		313
119	If you don't know what event loop to use, use the one returned from this	314	If you don't know what event loop to use, use the one returned from this
120	function.	315	function.
121		316
		317	Note that this function is I<not> thread-safe, so if you want to use it
		318	from multiple threads, you have to lock (note also that this is unlikely,
		319	as loops cannot be shared easily between threads anyway).
		320
		321	The default loop is the only loop that can handle C<ev_signal> and
		322	C<ev_child> watchers, and to do this, it always registers a handler
		323	for C<SIGCHLD>. If this is a problem for your application you can either
		324	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
		325	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
		326	C<ev_default_init>.
		327
122	The flags argument can be used to specify special behaviour or specific	328	The flags argument can be used to specify special behaviour or specific
123	backends to use, and is usually specified as 0 (or EVFLAG_AUTO).	329	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
124		330
125	It supports the following flags:	331	The following flags are supported:
126		332
127	=over 4	333	=over 4
128		334
129	=item EVFLAG_AUTO	335	=item C<EVFLAG_AUTO>
130		336
131	The default flags value. Use this if you have no clue (its the right	337	The default flags value. Use this if you have no clue (it's the right
132	thing, believe me).	338	thing, believe me).
133		339
134	=item EVFLAG_NOENV	340	=item C<EVFLAG_NOENV>
135		341
136	If this flag bit is ored into the flag value (or the program runs setuid	342	If this flag bit is or'ed into the flag value (or the program runs setuid
137	or setgid) then libev will I<not> look at the environment variable	343	or setgid) then libev will I<not> look at the environment variable
138	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	344	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
139	override the flags completely if it is found in the environment. This is	345	override the flags completely if it is found in the environment. This is
140	useful to try out specific backends to test their performance, or to work	346	useful to try out specific backends to test their performance, or to work
141	around bugs.	347	around bugs.
142		348
		349	=item C<EVFLAG_FORKCHECK>
		350
		351	Instead of calling C<ev_loop_fork> manually after a fork, you can also
		352	make libev check for a fork in each iteration by 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_SIGNALFD>
		376
		377	When this flag is specified, then libev will attempt to use the
		378	I<signalfd> API for it's C<ev_signal> (and C<ev_child>) watchers. This API
		379	delivers signals synchronously, which makes it both faster and might make
		380	it possible to get the queued signal data. It can also simplify signal
		381	handling with threads, as long as you properly block signals in your
		382	threads that are not interested in handling them.
		383
		384	Signalfd will not be used by default as this changes your signal mask, and
		385	there are a lot of shoddy libraries and programs (glib's threadpool for
		386	example) that can't properly initialise their signal masks.
		387
143	=item EVMETHOD_SELECT portable select backend	388	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
144		389
145	=item EVMETHOD_POLL poll backend (everywhere except windows)	390	This is your standard select(2) backend. Not I<completely> standard, as
		391	libev tries to roll its own fd_set with no limits on the number of fds,
		392	but if that fails, expect a fairly low limit on the number of fds when
		393	using this backend. It doesn't scale too well (O(highest_fd)), but its
		394	usually the fastest backend for a low number of (low-numbered :) fds.
146		395
147	=item EVMETHOD_EPOLL linux only	396	To get good performance out of this backend you need a high amount of
		397	parallelism (most of the file descriptors should be busy). If you are
		398	writing a server, you should C<accept ()> in a loop to accept as many
		399	connections as possible during one iteration. You might also want to have
		400	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		401	readiness notifications you get per iteration.
148		402
149	=item EVMETHOD_KQUEUE some bsds only	403	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		404	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		405	C<exceptfds> set on that platform).
150		406
151	=item EVMETHOD_DEVPOLL solaris 8 only	407	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
152		408
153	=item EVMETHOD_PORT solaris 10 only	409	And this is your standard poll(2) backend. It's more complicated
		410	than select, but handles sparse fds better and has no artificial
		411	limit on the number of fds you can use (except it will slow down
		412	considerably with a lot of inactive fds). It scales similarly to select,
		413	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		414	performance tips.
154		415
155	If one or more of these are ored into the flags value, then only these	416	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
156	backends will be tried (in the reverse order as given here). If one are	417	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
157	specified, any backend will do.	418
		419	=item C<EVBACKEND_EPOLL> (value 4, Linux)
		420
		421	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		422	kernels).
		423
		424	For few fds, this backend is a bit little slower than poll and select,
		425	but it scales phenomenally better. While poll and select usually scale
		426	like O(total_fds) where n is the total number of fds (or the highest fd),
		427	epoll scales either O(1) or O(active_fds).
		428
		429	The epoll mechanism deserves honorable mention as the most misdesigned
		430	of the more advanced event mechanisms: mere annoyances include silently
		431	dropping file descriptors, requiring a system call per change per file
		432	descriptor (and unnecessary guessing of parameters), problems with dup and
		433	so on. The biggest issue is fork races, however - if a program forks then
		434	I<both> parent and child process have to recreate the epoll set, which can
		435	take considerable time (one syscall per file descriptor) and is of course
		436	hard to detect.
		437
		438	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		439	of course I<doesn't>, and epoll just loves to report events for totally
		440	I<different> file descriptors (even already closed ones, so one cannot
		441	even remove them from the set) than registered in the set (especially
		442	on SMP systems). Libev tries to counter these spurious notifications by
		443	employing an additional generation counter and comparing that against the
		444	events to filter out spurious ones, recreating the set when required. Last
		445	not least, it also refuses to work with some file descriptors which work
		446	perfectly fine with C<select> (files, many character devices...).
		447
		448	While stopping, setting and starting an I/O watcher in the same iteration
		449	will result in some caching, there is still a system call per such
		450	incident (because the same I<file descriptor> could point to a different
		451	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
		452	file descriptors might not work very well if you register events for both
		453	file descriptors.
		454
		455	Best performance from this backend is achieved by not unregistering all
		456	watchers for a file descriptor until it has been closed, if possible,
		457	i.e. keep at least one watcher active per fd at all times. Stopping and
		458	starting a watcher (without re-setting it) also usually doesn't cause
		459	extra overhead. A fork can both result in spurious notifications as well
		460	as in libev having to destroy and recreate the epoll object, which can
		461	take considerable time and thus should be avoided.
		462
		463	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		464	faster than epoll for maybe up to a hundred file descriptors, depending on
		465	the usage. So sad.
		466
		467	While nominally embeddable in other event loops, this feature is broken in
		468	all kernel versions tested so far.
		469
		470	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		471	C<EVBACKEND_POLL>.
		472
		473	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
		474
		475	Kqueue deserves special mention, as at the time of this writing, it
		476	was broken on all BSDs except NetBSD (usually it doesn't work reliably
		477	with anything but sockets and pipes, except on Darwin, where of course
		478	it's completely useless). Unlike epoll, however, whose brokenness
		479	is by design, these kqueue bugs can (and eventually will) be fixed
		480	without API changes to existing programs. For this reason it's not being
		481	"auto-detected" unless you explicitly specify it in the flags (i.e. using
		482	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		483	system like NetBSD.
		484
		485	You still can embed kqueue into a normal poll or select backend and use it
		486	only for sockets (after having made sure that sockets work with kqueue on
		487	the target platform). See C<ev_embed> watchers for more info.
		488
		489	It scales in the same way as the epoll backend, but the interface to the
		490	kernel is more efficient (which says nothing about its actual speed, of
		491	course). While stopping, setting and starting an I/O watcher does never
		492	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
		493	two event changes per incident. Support for C<fork ()> is very bad (but
		494	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		495	cases
		496
		497	This backend usually performs well under most conditions.
		498
		499	While nominally embeddable in other event loops, this doesn't work
		500	everywhere, so you might need to test for this. And since it is broken
		501	almost everywhere, you should only use it when you have a lot of sockets
		502	(for which it usually works), by embedding it into another event loop
		503	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		504	also broken on OS X)) and, did I mention it, using it only for sockets.
		505
		506	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		507	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		508	C<NOTE_EOF>.
		509
		510	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
		511
		512	This is not implemented yet (and might never be, unless you send me an
		513	implementation). According to reports, C</dev/poll> only supports sockets
		514	and is not embeddable, which would limit the usefulness of this backend
		515	immensely.
		516
		517	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
		518
		519	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
		520	it's really slow, but it still scales very well (O(active_fds)).
		521
		522	Please note that Solaris event ports can deliver a lot of spurious
		523	notifications, so you need to use non-blocking I/O or other means to avoid
		524	blocking when no data (or space) is available.
		525
		526	While this backend scales well, it requires one system call per active
		527	file descriptor per loop iteration. For small and medium numbers of file
		528	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		529	might perform better.
		530
		531	On the positive side, with the exception of the spurious readiness
		532	notifications, this backend actually performed fully to specification
		533	in all tests and is fully embeddable, which is a rare feat among the
		534	OS-specific backends (I vastly prefer correctness over speed hacks).
		535
		536	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		537	C<EVBACKEND_POLL>.
		538
		539	=item C<EVBACKEND_ALL>
		540
		541	Try all backends (even potentially broken ones that wouldn't be tried
		542	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
		543	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		544
		545	It is definitely not recommended to use this flag.
158		546
159	=back	547	=back
160		548
		549	If one or more of the backend flags are or'ed into the flags value,
		550	then only these backends will be tried (in the reverse order as listed
		551	here). If none are specified, all backends in C<ev_recommended_backends
		552	()> will be tried.
		553
		554	Example: This is the most typical usage.
		555
		556	if (!ev_default_loop (0))
		557	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		558
		559	Example: Restrict libev to the select and poll backends, and do not allow
		560	environment settings to be taken into account:
		561
		562	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		563
		564	Example: Use whatever libev has to offer, but make sure that kqueue is
		565	used if available (warning, breaks stuff, best use only with your own
		566	private event loop and only if you know the OS supports your types of
		567	fds):
		568
		569	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
		570
161	=item struct ev_loop *ev_loop_new (unsigned int flags)	571	=item struct ev_loop *ev_loop_new (unsigned int flags)
162		572
163	Similar to C<ev_default_loop>, but always creates a new event loop that is	573	Similar to C<ev_default_loop>, but always creates a new event loop that is
164	always distinct from the default loop. Unlike the default loop, it cannot	574	always distinct from the default loop.
165	handle signal and child watchers, and attempts to do so will be greeted by	575
166	undefined behaviour (or a failed assertion if assertions are enabled).	576	Note that this function I<is> thread-safe, and one common way to use
		577	libev with threads is indeed to create one loop per thread, and using the
		578	default loop in the "main" or "initial" thread.
		579
		580	Example: Try to create a event loop that uses epoll and nothing else.
		581
		582	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
		583	if (!epoller)
		584	fatal ("no epoll found here, maybe it hides under your chair");
167		585
168	=item ev_default_destroy ()	586	=item ev_default_destroy ()
169		587
170	Destroys the default loop again (frees all memory and kernel state	588	Destroys the default loop (frees all memory and kernel state etc.). None
171	etc.). This stops all registered event watchers (by not touching them in	589	of the active event watchers will be stopped in the normal sense, so
172	any way whatsoever, although you cnanot rely on this :).	590	e.g. C<ev_is_active> might still return true. It is your responsibility to
		591	either stop all watchers cleanly yourself I<before> calling this function,
		592	or cope with the fact afterwards (which is usually the easiest thing, you
		593	can just ignore the watchers and/or C<free ()> them for example).
		594
		595	Note that certain global state, such as signal state (and installed signal
		596	handlers), will not be freed by this function, and related watchers (such
		597	as signal and child watchers) would need to be stopped manually.
		598
		599	In general it is not advisable to call this function except in the
		600	rare occasion where you really need to free e.g. the signal handling
		601	pipe fds. If you need dynamically allocated loops it is better to use
		602	C<ev_loop_new> and C<ev_loop_destroy>.
173		603
174	=item ev_loop_destroy (loop)	604	=item ev_loop_destroy (loop)
175		605
176	Like C<ev_default_destroy>, but destroys an event loop created by an	606	Like C<ev_default_destroy>, but destroys an event loop created by an
177	earlier call to C<ev_loop_new>.	607	earlier call to C<ev_loop_new>.
178		608
179	=item ev_default_fork ()	609	=item ev_default_fork ()
180		610
		611	This function sets a flag that causes subsequent C<ev_run> iterations
181	This function reinitialises the kernel state for backends that have	612	to reinitialise the kernel state for backends that have one. Despite the
182	one. Despite the name, you can call it anytime, but it makes most sense	613	name, you can call it anytime, but it makes most sense after forking, in
183	after forking, in either the parent or child process (or both, but that	614	the child process (or both child and parent, but that again makes little
184	again makes little sense).	615	sense). You I<must> call it in the child before using any of the libev
		616	functions, and it will only take effect at the next C<ev_run> iteration.
185		617
186	You I<must> call this function after forking if and only if you want to	618	Again, you I<have> to call it on I<any> loop that you want to re-use after
187	use the event library in both processes. If you just fork+exec, you don't	619	a fork, I<even if you do not plan to use the loop in the parent>. This is
188	have to call it.	620	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		621	during fork.
189		622
		623	On the other hand, you only need to call this function in the child
		624	process if and only if you want to use the event loop in the child. If
		625	you just fork+exec or create a new loop in the child, you don't have to
		626	call it at all (in fact, C<epoll> is so badly broken that it makes a
		627	difference, but libev will usually detect this case on its own and do a
		628	costly reset of the backend).
		629
190	The function itself is quite fast and its usually not a problem to call	630	The function itself is quite fast and it's usually not a problem to call
191	it just in case after a fork. To make this easy, the function will fit in	631	it just in case after a fork. To make this easy, the function will fit in
192	quite nicely into a call to C<pthread_atfork>:	632	quite nicely into a call to C<pthread_atfork>:
193		633
194	pthread_atfork (0, 0, ev_default_fork);	634	pthread_atfork (0, 0, ev_default_fork);
195		635
196	=item ev_loop_fork (loop)	636	=item ev_loop_fork (loop)
197		637
198	Like C<ev_default_fork>, but acts on an event loop created by	638	Like C<ev_default_fork>, but acts on an event loop created by
199	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	639	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
200	after fork, and how you do this is entirely your own problem.	640	after fork that you want to re-use in the child, and how you keep track of
		641	them is entirely your own problem.
201		642
		643	=item int ev_is_default_loop (loop)
		644
		645	Returns true when the given loop is, in fact, the default loop, and false
		646	otherwise.
		647
		648	=item unsigned int ev_iteration (loop)
		649
		650	Returns the current iteration count for the event loop, which is identical
		651	to the number of times libev did poll for new events. It starts at C<0>
		652	and happily wraps around with enough iterations.
		653
		654	This value can sometimes be useful as a generation counter of sorts (it
		655	"ticks" the number of loop iterations), as it roughly corresponds with
		656	C<ev_prepare> and C<ev_check> calls - and is incremented between the
		657	prepare and check phases.
		658
202	=item unsigned int ev_method (loop)	659	=item unsigned int ev_depth (loop)
203		660
		661	Returns the number of times C<ev_run> was entered minus the number of
		662	times C<ev_run> was exited, in other words, the recursion depth.
		663
		664	Outside C<ev_run>, this number is zero. In a callback, this number is
		665	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
		666	in which case it is higher.
		667
		668	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread
		669	etc.), doesn't count as "exit" - consider this as a hint to avoid such
		670	ungentleman-like behaviour unless it's really convenient.
		671
		672	=item unsigned int ev_backend (loop)
		673
204	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	674	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
205	use.	675	use.
206		676
207	=item ev_tstamp = ev_now (loop)	677	=item ev_tstamp ev_now (loop)
208		678
209	Returns the current "event loop time", which is the time the event loop	679	Returns the current "event loop time", which is the time the event loop
210	got events and started processing them. This timestamp does not change	680	received events and started processing them. This timestamp does not
211	as long as callbacks are being processed, and this is also the base time	681	change as long as callbacks are being processed, and this is also the base
212	used for relative timers. You can treat it as the timestamp of the event	682	time used for relative timers. You can treat it as the timestamp of the
213	occuring (or more correctly, the mainloop finding out about it).	683	event occurring (or more correctly, libev finding out about it).
214		684
		685	=item ev_now_update (loop)
		686
		687	Establishes the current time by querying the kernel, updating the time
		688	returned by C<ev_now ()> in the progress. This is a costly operation and
		689	is usually done automatically within C<ev_run ()>.
		690
		691	This function is rarely useful, but when some event callback runs for a
		692	very long time without entering the event loop, updating libev's idea of
		693	the current time is a good idea.
		694
		695	See also L<The special problem of time updates> in the C<ev_timer> section.
		696
		697	=item ev_suspend (loop)
		698
		699	=item ev_resume (loop)
		700
		701	These two functions suspend and resume an event loop, for use when the
		702	loop is not used for a while and timeouts should not be processed.
		703
		704	A typical use case would be an interactive program such as a game: When
		705	the user presses C<^Z> to suspend the game and resumes it an hour later it
		706	would be best to handle timeouts as if no time had actually passed while
		707	the program was suspended. This can be achieved by calling C<ev_suspend>
		708	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		709	C<ev_resume> directly afterwards to resume timer processing.
		710
		711	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		712	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		713	will be rescheduled (that is, they will lose any events that would have
		714	occurred while suspended).
		715
		716	After calling C<ev_suspend> you B<must not> call I<any> function on the
		717	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		718	without a previous call to C<ev_suspend>.
		719
		720	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		721	event loop time (see C<ev_now_update>).
		722
215	=item ev_loop (loop, int flags)	723	=item ev_run (loop, int flags)
216		724
217	Finally, this is it, the event handler. This function usually is called	725	Finally, this is it, the event handler. This function usually is called
218	after you initialised all your watchers and you want to start handling	726	after you have initialised all your watchers and you want to start
219	events.	727	handling events. It will ask the operating system for any new events, call
		728	the watcher callbacks, an then repeat the whole process indefinitely: This
		729	is why event loops are called I<loops>.
220		730
221	If the flags argument is specified as 0, it will not return until either	731	If the flags argument is specified as C<0>, it will keep handling events
222	no event watchers are active anymore or C<ev_unloop> was called.	732	until either no event watchers are active anymore or C<ev_break> was
		733	called.
223		734
		735	Please note that an explicit C<ev_break> is usually better than
		736	relying on all watchers to be stopped when deciding when a program has
		737	finished (especially in interactive programs), but having a program
		738	that automatically loops as long as it has to and no longer by virtue
		739	of relying on its watchers stopping correctly, that is truly a thing of
		740	beauty.
		741
224	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	742	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
225	those events and any outstanding ones, but will not block your process in	743	those events and any already outstanding ones, but will not wait and
226	case there are no events.	744	block your process in case there are no events and will return after one
		745	iteration of the loop. This is sometimes useful to poll and handle new
		746	events while doing lengthy calculations, to keep the program responsive.
227		747
228	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	748	A flags value of C<EVRUN_ONCE> will look for new events (waiting if
229	neccessary) and will handle those and any outstanding ones. It will block	749	necessary) and will handle those and any already outstanding ones. It
230	your process until at least one new event arrives.	750	will block your process until at least one new event arrives (which could
		751	be an event internal to libev itself, so there is no guarantee that a
		752	user-registered callback will be called), and will return after one
		753	iteration of the loop.
231		754
232	This flags value could be used to implement alternative looping	755	This is useful if you are waiting for some external event in conjunction
233	constructs, but the C<prepare> and C<check> watchers provide a better and	756	with something not expressible using other libev watchers (i.e. "roll your
234	more generic mechanism.	757	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		758	usually a better approach for this kind of thing.
235		759
		760	Here are the gory details of what C<ev_run> does:
		761
		762	- Increment loop depth.
		763	- Reset the ev_break status.
		764	- Before the first iteration, call any pending watchers.
		765	LOOP:
		766	- If EVFLAG_FORKCHECK was used, check for a fork.
		767	- If a fork was detected (by any means), queue and call all fork watchers.
		768	- Queue and call all prepare watchers.
		769	- If ev_break was called, goto FINISH.
		770	- If we have been forked, detach and recreate the kernel state
		771	as to not disturb the other process.
		772	- Update the kernel state with all outstanding changes.
		773	- Update the "event loop time" (ev_now ()).
		774	- Calculate for how long to sleep or block, if at all
		775	(active idle watchers, EVRUN_NOWAIT or not having
		776	any active watchers at all will result in not sleeping).
		777	- Sleep if the I/O and timer collect interval say so.
		778	- Increment loop iteration counter.
		779	- Block the process, waiting for any events.
		780	- Queue all outstanding I/O (fd) events.
		781	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
		782	- Queue all expired timers.
		783	- Queue all expired periodics.
		784	- Queue all idle watchers with priority higher than that of pending events.
		785	- Queue all check watchers.
		786	- Call all queued watchers in reverse order (i.e. check watchers first).
		787	Signals and child watchers are implemented as I/O watchers, and will
		788	be handled here by queueing them when their watcher gets executed.
		789	- If ev_break has been called, or EVRUN_ONCE or EVRUN_NOWAIT
		790	were used, or there are no active watchers, goto FINISH, otherwise
		791	continue with step LOOP.
		792	FINISH:
		793	- Reset the ev_break status iff it was EVBREAK_ONE.
		794	- Decrement the loop depth.
		795	- Return.
		796
		797	Example: Queue some jobs and then loop until no events are outstanding
		798	anymore.
		799
		800	... queue jobs here, make sure they register event watchers as long
		801	... as they still have work to do (even an idle watcher will do..)
		802	ev_run (my_loop, 0);
		803	... jobs done or somebody called unloop. yeah!
		804
236	=item ev_unloop (loop, how)	805	=item ev_break (loop, how)
237		806
238	Can be used to make a call to C<ev_loop> return early. The C<how> argument	807	Can be used to make a call to C<ev_run> return early (but only after it
239	must be either C<EVUNLOOP_ONCE>, which will make the innermost C<ev_loop>	808	has processed all outstanding events). The C<how> argument must be either
240	call return, or C<EVUNLOOP_ALL>, which will make all nested C<ev_loop>	809	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
241	calls return.	810	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
		811
		812	This "unloop state" will be cleared when entering C<ev_run> again.
		813
		814	It is safe to call C<ev_break> from outside any C<ev_run> calls. ##TODO##
242		815
243	=item ev_ref (loop)	816	=item ev_ref (loop)
244		817
245	=item ev_unref (loop)	818	=item ev_unref (loop)
246		819
247	Ref/unref can be used to add or remove a refcount on the event loop: Every	820	Ref/unref can be used to add or remove a reference count on the event
248	watcher keeps one reference. If you have a long-runing watcher you never	821	loop: Every watcher keeps one reference, and as long as the reference
249	unregister that should not keep ev_loop from running, ev_unref() after	822	count is nonzero, C<ev_run> will not return on its own.
250	starting, and ev_ref() before stopping it. Libev itself uses this for	823
251	example for its internal signal pipe: It is not visible to you as a user	824	This is useful when you have a watcher that you never intend to
252	and should not keep C<ev_loop> from exiting if the work is done. It is	825	unregister, but that nevertheless should not keep C<ev_run> from
		826	returning. In such a case, call C<ev_unref> after starting, and C<ev_ref>
		827	before stopping it.
		828
		829	As an example, libev itself uses this for its internal signal pipe: It
		830	is not visible to the libev user and should not keep C<ev_run> from
		831	exiting if no event watchers registered by it are active. It is also an
253	also an excellent way to do this for generic recurring timers or from	832	excellent way to do this for generic recurring timers or from within
254	within third-party libraries. Just remember to unref after start and ref	833	third-party libraries. Just remember to I<unref after start> and I<ref
255	before stop.	834	before stop> (but only if the watcher wasn't active before, or was active
		835	before, respectively. Note also that libev might stop watchers itself
		836	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		837	in the callback).
		838
		839	Example: Create a signal watcher, but keep it from keeping C<ev_run>
		840	running when nothing else is active.
		841
		842	ev_signal exitsig;
		843	ev_signal_init (&exitsig, sig_cb, SIGINT);
		844	ev_signal_start (loop, &exitsig);
		845	evf_unref (loop);
		846
		847	Example: For some weird reason, unregister the above signal handler again.
		848
		849	ev_ref (loop);
		850	ev_signal_stop (loop, &exitsig);
		851
		852	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		853
		854	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		855
		856	These advanced functions influence the time that libev will spend waiting
		857	for events. Both time intervals are by default C<0>, meaning that libev
		858	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		859	latency.
		860
		861	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		862	allows libev to delay invocation of I/O and timer/periodic callbacks
		863	to increase efficiency of loop iterations (or to increase power-saving
		864	opportunities).
		865
		866	The idea is that sometimes your program runs just fast enough to handle
		867	one (or very few) event(s) per loop iteration. While this makes the
		868	program responsive, it also wastes a lot of CPU time to poll for new
		869	events, especially with backends like C<select ()> which have a high
		870	overhead for the actual polling but can deliver many events at once.
		871
		872	By setting a higher I<io collect interval> you allow libev to spend more
		873	time collecting I/O events, so you can handle more events per iteration,
		874	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		875	C<ev_timer>) will be not affected. Setting this to a non-null value will
		876	introduce an additional C<ev_sleep ()> call into most loop iterations. The
		877	sleep time ensures that libev will not poll for I/O events more often then
		878	once per this interval, on average.
		879
		880	Likewise, by setting a higher I<timeout collect interval> you allow libev
		881	to spend more time collecting timeouts, at the expense of increased
		882	latency/jitter/inexactness (the watcher callback will be called
		883	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		884	value will not introduce any overhead in libev.
		885
		886	Many (busy) programs can usually benefit by setting the I/O collect
		887	interval to a value near C<0.1> or so, which is often enough for
		888	interactive servers (of course not for games), likewise for timeouts. It
		889	usually doesn't make much sense to set it to a lower value than C<0.01>,
		890	as this approaches the timing granularity of most systems. Note that if
		891	you do transactions with the outside world and you can't increase the
		892	parallelity, then this setting will limit your transaction rate (if you
		893	need to poll once per transaction and the I/O collect interval is 0.01,
		894	then you can't do more than 100 transactions per second).
		895
		896	Setting the I<timeout collect interval> can improve the opportunity for
		897	saving power, as the program will "bundle" timer callback invocations that
		898	are "near" in time together, by delaying some, thus reducing the number of
		899	times the process sleeps and wakes up again. Another useful technique to
		900	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		901	they fire on, say, one-second boundaries only.
		902
		903	Example: we only need 0.1s timeout granularity, and we wish not to poll
		904	more often than 100 times per second:
		905
		906	ev_set_timeout_collect_interval (EV_DEFAULT_UC_ 0.1);
		907	ev_set_io_collect_interval (EV_DEFAULT_UC_ 0.01);
		908
		909	=item ev_invoke_pending (loop)
		910
		911	This call will simply invoke all pending watchers while resetting their
		912	pending state. Normally, C<ev_run> does this automatically when required,
		913	but when overriding the invoke callback this call comes handy. This
		914	function can be invoked from a watcher - this can be useful for example
		915	when you want to do some lengthy calculation and want to pass further
		916	event handling to another thread (you still have to make sure only one
		917	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
		918
		919	=item int ev_pending_count (loop)
		920
		921	Returns the number of pending watchers - zero indicates that no watchers
		922	are pending.
		923
		924	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		925
		926	This overrides the invoke pending functionality of the loop: Instead of
		927	invoking all pending watchers when there are any, C<ev_run> will call
		928	this callback instead. This is useful, for example, when you want to
		929	invoke the actual watchers inside another context (another thread etc.).
		930
		931	If you want to reset the callback, use C<ev_invoke_pending> as new
		932	callback.
		933
		934	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		935
		936	Sometimes you want to share the same loop between multiple threads. This
		937	can be done relatively simply by putting mutex_lock/unlock calls around
		938	each call to a libev function.
		939
		940	However, C<ev_run> can run an indefinite time, so it is not feasible
		941	to wait for it to return. One way around this is to wake up the event
		942	loop via C<ev_break> and C<av_async_send>, another way is to set these
		943	I<release> and I<acquire> callbacks on the loop.
		944
		945	When set, then C<release> will be called just before the thread is
		946	suspended waiting for new events, and C<acquire> is called just
		947	afterwards.
		948
		949	Ideally, C<release> will just call your mutex_unlock function, and
		950	C<acquire> will just call the mutex_lock function again.
		951
		952	While event loop modifications are allowed between invocations of
		953	C<release> and C<acquire> (that's their only purpose after all), no
		954	modifications done will affect the event loop, i.e. adding watchers will
		955	have no effect on the set of file descriptors being watched, or the time
		956	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		957	to take note of any changes you made.
		958
		959	In theory, threads executing C<ev_run> will be async-cancel safe between
		960	invocations of C<release> and C<acquire>.
		961
		962	See also the locking example in the C<THREADS> section later in this
		963	document.
		964
		965	=item ev_set_userdata (loop, void *data)
		966
		967	=item ev_userdata (loop)
		968
		969	Set and retrieve a single C<void *> associated with a loop. When
		970	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		971	C<0.>
		972
		973	These two functions can be used to associate arbitrary data with a loop,
		974	and are intended solely for the C<invoke_pending_cb>, C<release> and
		975	C<acquire> callbacks described above, but of course can be (ab-)used for
		976	any other purpose as well.
		977
		978	=item ev_verify (loop)
		979
		980	This function only does something when C<EV_VERIFY> support has been
		981	compiled in, which is the default for non-minimal builds. It tries to go
		982	through all internal structures and checks them for validity. If anything
		983	is found to be inconsistent, it will print an error message to standard
		984	error and call C<abort ()>.
		985
		986	This can be used to catch bugs inside libev itself: under normal
		987	circumstances, this function will never abort as of course libev keeps its
		988	data structures consistent.
256		989
257	=back	990	=back
258		991
		992
259	=head1 ANATOMY OF A WATCHER	993	=head1 ANATOMY OF A WATCHER
260		994
		995	In the following description, uppercase C<TYPE> in names stands for the
		996	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		997	watchers and C<ev_io_start> for I/O watchers.
		998
261	A watcher is a structure that you create and register to record your	999	A watcher is an opaque structure that you allocate and register to record
262	interest in some event. For instance, if you want to wait for STDIN to	1000	your interest in some event. To make a concrete example, imagine you want
263	become readable, you would create an ev_io watcher for that:	1001	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1002	for that:
264		1003
265	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1004	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
266	{	1005	{
267	ev_io_stop (w);	1006	ev_io_stop (w);
268	ev_unloop (loop, EVUNLOOP_ALL);	1007	ev_break (loop, EVBREAK_ALL);
269	}	1008	}
270		1009
271	struct ev_loop *loop = ev_default_loop (0);	1010	struct ev_loop *loop = ev_default_loop (0);
		1011
272	struct ev_io stdin_watcher;	1012	ev_io stdin_watcher;
		1013
273	ev_init (&stdin_watcher, my_cb);	1014	ev_init (&stdin_watcher, my_cb);
274	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1015	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
275	ev_io_start (loop, &stdin_watcher);	1016	ev_io_start (loop, &stdin_watcher);
		1017
276	ev_loop (loop, 0);	1018	ev_run (loop, 0);
277		1019
278	As you can see, you are responsible for allocating the memory for your	1020	As you can see, you are responsible for allocating the memory for your
279	watcher structures (and it is usually a bad idea to do this on the stack,	1021	watcher structures (and it is I<usually> a bad idea to do this on the
280	although this can sometimes be quite valid).	1022	stack).
281		1023
		1024	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1025	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1026
282	Each watcher structure must be initialised by a call to C<ev_init	1027	Each watcher structure must be initialised by a call to C<ev_init (watcher
283	(watcher *, callback)>, which expects a callback to be provided. This	1028	*, callback)>, which expects a callback to be provided. This callback is
284	callback gets invoked each time the event occurs (or, in the case of io	1029	invoked each time the event occurs (or, in the case of I/O watchers, each
285	watchers, each time the event loop detects that the file descriptor given	1030	time the event loop detects that the file descriptor given is readable
286	is readable and/or writable).	1031	and/or writable).
287		1032
288	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1033	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
289	with arguments specific to this watcher type. There is also a macro	1034	macro to configure it, with arguments specific to the watcher type. There
290	to combine initialisation and setting in one call: C<< ev_<type>_init	1035	is also a macro to combine initialisation and setting in one call: C<<
291	(watcher *, callback, ...) >>.	1036	ev_TYPE_init (watcher *, callback, ...) >>.
292		1037
293	To make the watcher actually watch out for events, you have to start it	1038	To make the watcher actually watch out for events, you have to start it
294	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1039	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
295	*) >>), and you can stop watching for events at any time by calling the	1040	*) >>), and you can stop watching for events at any time by calling the
296	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1041	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
297		1042
298	As long as your watcher is active (has been started but not stopped) you	1043	As long as your watcher is active (has been started but not stopped) you
299	must not touch the values stored in it. Most specifically you must never	1044	must not touch the values stored in it. Most specifically you must never
300	reinitialise it or call its set method.	1045	reinitialise it or call its C<ev_TYPE_set> macro.
301
302	You cna check whether an event is active by calling the C<ev_is_active
303	(watcher *)> macro. To see whether an event is outstanding (but the
304	callback for it has not been called yet) you cna use the C<ev_is_pending
305	(watcher *)> macro.
306		1046
307	Each and every callback receives the event loop pointer as first, the	1047	Each and every callback receives the event loop pointer as first, the
308	registered watcher structure as second, and a bitset of received events as	1048	registered watcher structure as second, and a bitset of received events as
309	third argument.	1049	third argument.
310		1050
311	The rceeived events usually include a single bit per event type received	1051	The received events usually include a single bit per event type received
312	(you can receive multiple events at the same time). The possible bit masks	1052	(you can receive multiple events at the same time). The possible bit masks
313	are:	1053	are:
314		1054
315	=over 4	1055	=over 4
316		1056
317	=item EV_READ	1057	=item C<EV_READ>
318		1058
319	=item EV_WRITE	1059	=item C<EV_WRITE>
320		1060
321	The file descriptor in the ev_io watcher has become readable and/or	1061	The file descriptor in the C<ev_io> watcher has become readable and/or
322	writable.	1062	writable.
323		1063
324	=item EV_TIMEOUT	1064	=item C<EV_TIMER>
325		1065
326	The ev_timer watcher has timed out.	1066	The C<ev_timer> watcher has timed out.
327		1067
328	=item EV_PERIODIC	1068	=item C<EV_PERIODIC>
329		1069
330	The ev_periodic watcher has timed out.	1070	The C<ev_periodic> watcher has timed out.
331		1071
332	=item EV_SIGNAL	1072	=item C<EV_SIGNAL>
333		1073
334	The signal specified in the ev_signal watcher has been received by a thread.	1074	The signal specified in the C<ev_signal> watcher has been received by a thread.
335		1075
336	=item EV_CHILD	1076	=item C<EV_CHILD>
337		1077
338	The pid specified in the ev_child watcher has received a status change.	1078	The pid specified in the C<ev_child> watcher has received a status change.
339		1079
		1080	=item C<EV_STAT>
		1081
		1082	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		1083
340	=item EV_IDLE	1084	=item C<EV_IDLE>
341		1085
342	The ev_idle watcher has determined that you have nothing better to do.	1086	The C<ev_idle> watcher has determined that you have nothing better to do.
343		1087
344	=item EV_PREPARE	1088	=item C<EV_PREPARE>
345		1089
346	=item EV_CHECK	1090	=item C<EV_CHECK>
347		1091
348	All ev_prepare watchers are invoked just I<before> C<ev_loop> starts	1092	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts
349	to gather new events, and all ev_check watchers are invoked just after	1093	to gather new events, and all C<ev_check> watchers are invoked just after
350	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1094	C<ev_run> has gathered them, but before it invokes any callbacks for any
351	received events. Callbacks of both watcher types can start and stop as	1095	received events. Callbacks of both watcher types can start and stop as
352	many watchers as they want, and all of them will be taken into account	1096	many watchers as they want, and all of them will be taken into account
353	(for example, a ev_prepare watcher might start an idle watcher to keep	1097	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
354	C<ev_loop> from blocking).	1098	C<ev_run> from blocking).
355		1099
		1100	=item C<EV_EMBED>
		1101
		1102	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1103
		1104	=item C<EV_FORK>
		1105
		1106	The event loop has been resumed in the child process after fork (see
		1107	C<ev_fork>).
		1108
		1109	=item C<EV_ASYNC>
		1110
		1111	The given async watcher has been asynchronously notified (see C<ev_async>).
		1112
		1113	=item C<EV_CUSTOM>
		1114
		1115	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1116	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		1117
356	=item EV_ERROR	1118	=item C<EV_ERROR>
357		1119
358	An unspecified error has occured, the watcher has been stopped. This might	1120	An unspecified error has occurred, the watcher has been stopped. This might
359	happen because the watcher could not be properly started because libev	1121	happen because the watcher could not be properly started because libev
360	ran out of memory, a file descriptor was found to be closed or any other	1122	ran out of memory, a file descriptor was found to be closed or any other
		1123	problem. Libev considers these application bugs.
		1124
361	problem. You best act on it by reporting the problem and somehow coping	1125	You best act on it by reporting the problem and somehow coping with the
362	with the watcher being stopped.	1126	watcher being stopped. Note that well-written programs should not receive
		1127	an error ever, so when your watcher receives it, this usually indicates a
		1128	bug in your program.
363		1129
364	Libev will usually signal a few "dummy" events together with an error,	1130	Libev will usually signal a few "dummy" events together with an error, for
365	for example it might indicate that a fd is readable or writable, and if	1131	example it might indicate that a fd is readable or writable, and if your
366	your callbacks is well-written it can just attempt the operation and cope	1132	callbacks is well-written it can just attempt the operation and cope with
367	with the error from read() or write(). This will not work in multithreaded	1133	the error from read() or write(). This will not work in multi-threaded
368	programs, though, so beware.	1134	programs, though, as the fd could already be closed and reused for another
		1135	thing, so beware.
369		1136
370	=back	1137	=back
371		1138
		1139	=head2 WATCHER STATES
		1140
		1141	There are various watcher states mentioned throughout this manual -
		1142	active, pending and so on. In this section these states and the rules to
		1143	transition between them will be described in more detail - and while these
		1144	rules might look complicated, they usually do "the right thing".
		1145
		1146	=over 4
		1147
		1148	=item initialiased
		1149
		1150	Before a watcher can be registered with the event looop it has to be
		1151	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1152	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1153
		1154	In this state it is simply some block of memory that is suitable for use
		1155	in an event loop. It can be moved around, freed, reused etc. at will.
		1156
		1157	=item started/running/active
		1158
		1159	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1160	property of the event loop, and is actively waiting for events. While in
		1161	this state it cannot be accessed (except in a few documented ways), moved,
		1162	freed or anything else - the only legal thing is to keep a pointer to it,
		1163	and call libev functions on it that are documented to work on active watchers.
		1164
		1165	=item pending
		1166
		1167	If a watcher is active and libev determines that an event it is interested
		1168	in has occurred (such as a timer expiring), it will become pending. It will
		1169	stay in this pending state until either it is stopped or its callback is
		1170	about to be invoked, so it is not normally pending inside the watcher
		1171	callback.
		1172
		1173	The watcher might or might not be active while it is pending (for example,
		1174	an expired non-repeating timer can be pending but no longer active). If it
		1175	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1176	but it is still property of the event loop at this time, so cannot be
		1177	moved, freed or reused. And if it is active the rules described in the
		1178	previous item still apply.
		1179
		1180	It is also possible to feed an event on a watcher that is not active (e.g.
		1181	via C<ev_feed_event>), in which case it becomes pending without being
		1182	active.
		1183
		1184	=item stopped
		1185
		1186	A watcher can be stopped implicitly by libev (in which case it might still
		1187	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1188	latter will clear any pending state the watcher might be in, regardless
		1189	of whether it was active or not, so stopping a watcher explicitly before
		1190	freeing it is often a good idea.
		1191
		1192	While stopped (and not pending) the watcher is essentially in the
		1193	initialised state, that is it can be reused, moved, modified in any way
		1194	you wish.
		1195
		1196	=back
		1197
		1198	=head2 GENERIC WATCHER FUNCTIONS
		1199
		1200	=over 4
		1201
		1202	=item C<ev_init> (ev_TYPE *watcher, callback)
		1203
		1204	This macro initialises the generic portion of a watcher. The contents
		1205	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1206	the generic parts of the watcher are initialised, you I<need> to call
		1207	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1208	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1209	which rolls both calls into one.
		1210
		1211	You can reinitialise a watcher at any time as long as it has been stopped
		1212	(or never started) and there are no pending events outstanding.
		1213
		1214	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1215	int revents)>.
		1216
		1217	Example: Initialise an C<ev_io> watcher in two steps.
		1218
		1219	ev_io w;
		1220	ev_init (&w, my_cb);
		1221	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1222
		1223	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1224
		1225	This macro initialises the type-specific parts of a watcher. You need to
		1226	call C<ev_init> at least once before you call this macro, but you can
		1227	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1228	macro on a watcher that is active (it can be pending, however, which is a
		1229	difference to the C<ev_init> macro).
		1230
		1231	Although some watcher types do not have type-specific arguments
		1232	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1233
		1234	See C<ev_init>, above, for an example.
		1235
		1236	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1237
		1238	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1239	calls into a single call. This is the most convenient method to initialise
		1240	a watcher. The same limitations apply, of course.
		1241
		1242	Example: Initialise and set an C<ev_io> watcher in one step.
		1243
		1244	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1245
		1246	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1247
		1248	Starts (activates) the given watcher. Only active watchers will receive
		1249	events. If the watcher is already active nothing will happen.
		1250
		1251	Example: Start the C<ev_io> watcher that is being abused as example in this
		1252	whole section.
		1253
		1254	ev_io_start (EV_DEFAULT_UC, &w);
		1255
		1256	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1257
		1258	Stops the given watcher if active, and clears the pending status (whether
		1259	the watcher was active or not).
		1260
		1261	It is possible that stopped watchers are pending - for example,
		1262	non-repeating timers are being stopped when they become pending - but
		1263	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1264	pending. If you want to free or reuse the memory used by the watcher it is
		1265	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1266
		1267	=item bool ev_is_active (ev_TYPE *watcher)
		1268
		1269	Returns a true value iff the watcher is active (i.e. it has been started
		1270	and not yet been stopped). As long as a watcher is active you must not modify
		1271	it.
		1272
		1273	=item bool ev_is_pending (ev_TYPE *watcher)
		1274
		1275	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1276	events but its callback has not yet been invoked). As long as a watcher
		1277	is pending (but not active) you must not call an init function on it (but
		1278	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1279	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1280	it).
		1281
		1282	=item callback ev_cb (ev_TYPE *watcher)
		1283
		1284	Returns the callback currently set on the watcher.
		1285
		1286	=item ev_cb_set (ev_TYPE *watcher, callback)
		1287
		1288	Change the callback. You can change the callback at virtually any time
		1289	(modulo threads).
		1290
		1291	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1292
		1293	=item int ev_priority (ev_TYPE *watcher)
		1294
		1295	Set and query the priority of the watcher. The priority is a small
		1296	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1297	(default: C<-2>). Pending watchers with higher priority will be invoked
		1298	before watchers with lower priority, but priority will not keep watchers
		1299	from being executed (except for C<ev_idle> watchers).
		1300
		1301	If you need to suppress invocation when higher priority events are pending
		1302	you need to look at C<ev_idle> watchers, which provide this functionality.
		1303
		1304	You I<must not> change the priority of a watcher as long as it is active or
		1305	pending.
		1306
		1307	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1308	fine, as long as you do not mind that the priority value you query might
		1309	or might not have been clamped to the valid range.
		1310
		1311	The default priority used by watchers when no priority has been set is
		1312	always C<0>, which is supposed to not be too high and not be too low :).
		1313
		1314	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1315	priorities.
		1316
		1317	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1318
		1319	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1320	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1321	can deal with that fact, as both are simply passed through to the
		1322	callback.
		1323
		1324	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1325
		1326	If the watcher is pending, this function clears its pending status and
		1327	returns its C<revents> bitset (as if its callback was invoked). If the
		1328	watcher isn't pending it does nothing and returns C<0>.
		1329
		1330	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1331	callback to be invoked, which can be accomplished with this function.
		1332
		1333	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1334
		1335	Feeds the given event set into the event loop, as if the specified event
		1336	had happened for the specified watcher (which must be a pointer to an
		1337	initialised but not necessarily started event watcher). Obviously you must
		1338	not free the watcher as long as it has pending events.
		1339
		1340	Stopping the watcher, letting libev invoke it, or calling
		1341	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1342	not started in the first place.
		1343
		1344	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1345	functions that do not need a watcher.
		1346
		1347	=back
		1348
		1349
372	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1350	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
373		1351
374	Each watcher has, by default, a member C<void *data> that you can change	1352	Each watcher has, by default, a member C<void *data> that you can change
375	and read at any time, libev will completely ignore it. This cna be used	1353	and read at any time: libev will completely ignore it. This can be used
376	to associate arbitrary data with your watcher. If you need more data and	1354	to associate arbitrary data with your watcher. If you need more data and
377	don't want to allocate memory and store a pointer to it in that data	1355	don't want to allocate memory and store a pointer to it in that data
378	member, you can also "subclass" the watcher type and provide your own	1356	member, you can also "subclass" the watcher type and provide your own
379	data:	1357	data:
380		1358
381	struct my_io	1359	struct my_io
382	{	1360	{
383	struct ev_io io;	1361	ev_io io;
384	int otherfd;	1362	int otherfd;
385	void *somedata;	1363	void *somedata;
386	struct whatever *mostinteresting;	1364	struct whatever *mostinteresting;
387	}	1365	};
		1366
		1367	...
		1368	struct my_io w;
		1369	ev_io_init (&w.io, my_cb, fd, EV_READ);
388		1370
389	And since your callback will be called with a pointer to the watcher, you	1371	And since your callback will be called with a pointer to the watcher, you
390	can cast it back to your own type:	1372	can cast it back to your own type:
391		1373
392	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1374	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
393	{	1375	{
394	struct my_io *w = (struct my_io *)w_;	1376	struct my_io *w = (struct my_io *)w_;
395	...	1377	...
396	}	1378	}
397		1379
398	More interesting and less C-conformant ways of catsing your callback type	1380	More interesting and less C-conformant ways of casting your callback type
399	have been omitted....	1381	instead have been omitted.
		1382
		1383	Another common scenario is to use some data structure with multiple
		1384	embedded watchers:
		1385
		1386	struct my_biggy
		1387	{
		1388	int some_data;
		1389	ev_timer t1;
		1390	ev_timer t2;
		1391	}
		1392
		1393	In this case getting the pointer to C<my_biggy> is a bit more
		1394	complicated: Either you store the address of your C<my_biggy> struct
		1395	in the C<data> member of the watcher (for woozies), or you need to use
		1396	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1397	programmers):
		1398
		1399	#include <stddef.h>
		1400
		1401	static void
		1402	t1_cb (EV_P_ ev_timer *w, int revents)
		1403	{
		1404	struct my_biggy big = (struct my_biggy *)
		1405	(((char *)w) - offsetof (struct my_biggy, t1));
		1406	}
		1407
		1408	static void
		1409	t2_cb (EV_P_ ev_timer *w, int revents)
		1410	{
		1411	struct my_biggy big = (struct my_biggy *)
		1412	(((char *)w) - offsetof (struct my_biggy, t2));
		1413	}
		1414
		1415	=head2 WATCHER PRIORITY MODELS
		1416
		1417	Many event loops support I<watcher priorities>, which are usually small
		1418	integers that influence the ordering of event callback invocation
		1419	between watchers in some way, all else being equal.
		1420
		1421	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1422	description for the more technical details such as the actual priority
		1423	range.
		1424
		1425	There are two common ways how these these priorities are being interpreted
		1426	by event loops:
		1427
		1428	In the more common lock-out model, higher priorities "lock out" invocation
		1429	of lower priority watchers, which means as long as higher priority
		1430	watchers receive events, lower priority watchers are not being invoked.
		1431
		1432	The less common only-for-ordering model uses priorities solely to order
		1433	callback invocation within a single event loop iteration: Higher priority
		1434	watchers are invoked before lower priority ones, but they all get invoked
		1435	before polling for new events.
		1436
		1437	Libev uses the second (only-for-ordering) model for all its watchers
		1438	except for idle watchers (which use the lock-out model).
		1439
		1440	The rationale behind this is that implementing the lock-out model for
		1441	watchers is not well supported by most kernel interfaces, and most event
		1442	libraries will just poll for the same events again and again as long as
		1443	their callbacks have not been executed, which is very inefficient in the
		1444	common case of one high-priority watcher locking out a mass of lower
		1445	priority ones.
		1446
		1447	Static (ordering) priorities are most useful when you have two or more
		1448	watchers handling the same resource: a typical usage example is having an
		1449	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1450	timeouts. Under load, data might be received while the program handles
		1451	other jobs, but since timers normally get invoked first, the timeout
		1452	handler will be executed before checking for data. In that case, giving
		1453	the timer a lower priority than the I/O watcher ensures that I/O will be
		1454	handled first even under adverse conditions (which is usually, but not
		1455	always, what you want).
		1456
		1457	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1458	will only be executed when no same or higher priority watchers have
		1459	received events, they can be used to implement the "lock-out" model when
		1460	required.
		1461
		1462	For example, to emulate how many other event libraries handle priorities,
		1463	you can associate an C<ev_idle> watcher to each such watcher, and in
		1464	the normal watcher callback, you just start the idle watcher. The real
		1465	processing is done in the idle watcher callback. This causes libev to
		1466	continuously poll and process kernel event data for the watcher, but when
		1467	the lock-out case is known to be rare (which in turn is rare :), this is
		1468	workable.
		1469
		1470	Usually, however, the lock-out model implemented that way will perform
		1471	miserably under the type of load it was designed to handle. In that case,
		1472	it might be preferable to stop the real watcher before starting the
		1473	idle watcher, so the kernel will not have to process the event in case
		1474	the actual processing will be delayed for considerable time.
		1475
		1476	Here is an example of an I/O watcher that should run at a strictly lower
		1477	priority than the default, and which should only process data when no
		1478	other events are pending:
		1479
		1480	ev_idle idle; // actual processing watcher
		1481	ev_io io; // actual event watcher
		1482
		1483	static void
		1484	io_cb (EV_P_ ev_io *w, int revents)
		1485	{
		1486	// stop the I/O watcher, we received the event, but
		1487	// are not yet ready to handle it.
		1488	ev_io_stop (EV_A_ w);
		1489
		1490	// start the idle watcher to handle the actual event.
		1491	// it will not be executed as long as other watchers
		1492	// with the default priority are receiving events.
		1493	ev_idle_start (EV_A_ &idle);
		1494	}
		1495
		1496	static void
		1497	idle_cb (EV_P_ ev_idle *w, int revents)
		1498	{
		1499	// actual processing
		1500	read (STDIN_FILENO, ...);
		1501
		1502	// have to start the I/O watcher again, as
		1503	// we have handled the event
		1504	ev_io_start (EV_P_ &io);
		1505	}
		1506
		1507	// initialisation
		1508	ev_idle_init (&idle, idle_cb);
		1509	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1510	ev_io_start (EV_DEFAULT_ &io);
		1511
		1512	In the "real" world, it might also be beneficial to start a timer, so that
		1513	low-priority connections can not be locked out forever under load. This
		1514	enables your program to keep a lower latency for important connections
		1515	during short periods of high load, while not completely locking out less
		1516	important ones.
400		1517
401		1518
402	=head1 WATCHER TYPES	1519	=head1 WATCHER TYPES
403		1520
404	This section describes each watcher in detail, but will not repeat	1521	This section describes each watcher in detail, but will not repeat
405	information given in the last section.	1522	information given in the last section. Any initialisation/set macros,
		1523	functions and members specific to the watcher type are explained.
406		1524
		1525	Members are additionally marked with either I<[read-only]>, meaning that,
		1526	while the watcher is active, you can look at the member and expect some
		1527	sensible content, but you must not modify it (you can modify it while the
		1528	watcher is stopped to your hearts content), or I<[read-write]>, which
		1529	means you can expect it to have some sensible content while the watcher
		1530	is active, but you can also modify it. Modifying it may not do something
		1531	sensible or take immediate effect (or do anything at all), but libev will
		1532	not crash or malfunction in any way.
		1533
		1534
407	=head2 struct ev_io - is my file descriptor readable or writable	1535	=head2 C<ev_io> - is this file descriptor readable or writable?
408		1536
409	I/O watchers check whether a file descriptor is readable or writable	1537	I/O watchers check whether a file descriptor is readable or writable
410	in each iteration of the event loop (This behaviour is called	1538	in each iteration of the event loop, or, more precisely, when reading
411	level-triggering because you keep receiving events as long as the	1539	would not block the process and writing would at least be able to write
412	condition persists. Remember you cna stop the watcher if you don't want to	1540	some data. This behaviour is called level-triggering because you keep
413	act on the event and neither want to receive future events).	1541	receiving events as long as the condition persists. Remember you can stop
		1542	the watcher if you don't want to act on the event and neither want to
		1543	receive future events.
414		1544
415	In general you can register as many read and/or write event watchers oer	1545	In general you can register as many read and/or write event watchers per
416	fd as you want (as long as you don't confuse yourself). Setting all file	1546	fd as you want (as long as you don't confuse yourself). Setting all file
417	descriptors to non-blocking mode is also usually a good idea (but not	1547	descriptors to non-blocking mode is also usually a good idea (but not
418	required if you know what you are doing).	1548	required if you know what you are doing).
419		1549
420	You have to be careful with dup'ed file descriptors, though. Some backends	1550	If you cannot use non-blocking mode, then force the use of a
421	(the linux epoll backend is a notable example) cannot handle dup'ed file	1551	known-to-be-good backend (at the time of this writing, this includes only
422	descriptors correctly if you register interest in two or more fds pointing	1552	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
423	to the same file/socket etc. description.	1553	descriptors for which non-blocking operation makes no sense (such as
		1554	files) - libev doesn't guarantee any specific behaviour in that case.
424		1555
425	If you must do this, then force the use of a known-to-be-good backend	1556	Another thing you have to watch out for is that it is quite easy to
426	(at the time of this writing, this includes only EVMETHOD_SELECT and	1557	receive "spurious" readiness notifications, that is your callback might
427	EVMETHOD_POLL).	1558	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1559	because there is no data. Not only are some backends known to create a
		1560	lot of those (for example Solaris ports), it is very easy to get into
		1561	this situation even with a relatively standard program structure. Thus
		1562	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1563	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1564
		1565	If you cannot run the fd in non-blocking mode (for example you should
		1566	not play around with an Xlib connection), then you have to separately
		1567	re-test whether a file descriptor is really ready with a known-to-be good
		1568	interface such as poll (fortunately in our Xlib example, Xlib already
		1569	does this on its own, so its quite safe to use). Some people additionally
		1570	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1571	indefinitely.
		1572
		1573	But really, best use non-blocking mode.
		1574
		1575	=head3 The special problem of disappearing file descriptors
		1576
		1577	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1578	descriptor (either due to calling C<close> explicitly or any other means,
		1579	such as C<dup2>). The reason is that you register interest in some file
		1580	descriptor, but when it goes away, the operating system will silently drop
		1581	this interest. If another file descriptor with the same number then is
		1582	registered with libev, there is no efficient way to see that this is, in
		1583	fact, a different file descriptor.
		1584
		1585	To avoid having to explicitly tell libev about such cases, libev follows
		1586	the following policy: Each time C<ev_io_set> is being called, libev
		1587	will assume that this is potentially a new file descriptor, otherwise
		1588	it is assumed that the file descriptor stays the same. That means that
		1589	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1590	descriptor even if the file descriptor number itself did not change.
		1591
		1592	This is how one would do it normally anyway, the important point is that
		1593	the libev application should not optimise around libev but should leave
		1594	optimisations to libev.
		1595
		1596	=head3 The special problem of dup'ed file descriptors
		1597
		1598	Some backends (e.g. epoll), cannot register events for file descriptors,
		1599	but only events for the underlying file descriptions. That means when you
		1600	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1601	events for them, only one file descriptor might actually receive events.
		1602
		1603	There is no workaround possible except not registering events
		1604	for potentially C<dup ()>'ed file descriptors, or to resort to
		1605	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1606
		1607	=head3 The special problem of fork
		1608
		1609	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1610	useless behaviour. Libev fully supports fork, but needs to be told about
		1611	it in the child.
		1612
		1613	To support fork in your programs, you either have to call
		1614	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1615	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1616	C<EVBACKEND_POLL>.
		1617
		1618	=head3 The special problem of SIGPIPE
		1619
		1620	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1621	when writing to a pipe whose other end has been closed, your program gets
		1622	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1623	this is sensible behaviour, for daemons, this is usually undesirable.
		1624
		1625	So when you encounter spurious, unexplained daemon exits, make sure you
		1626	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1627	somewhere, as that would have given you a big clue).
		1628
		1629	=head3 The special problem of accept()ing when you can't
		1630
		1631	Many implementations of the POSIX C<accept> function (for example,
		1632	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1633	connection from the pending queue in all error cases.
		1634
		1635	For example, larger servers often run out of file descriptors (because
		1636	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1637	rejecting the connection, leading to libev signalling readiness on
		1638	the next iteration again (the connection still exists after all), and
		1639	typically causing the program to loop at 100% CPU usage.
		1640
		1641	Unfortunately, the set of errors that cause this issue differs between
		1642	operating systems, there is usually little the app can do to remedy the
		1643	situation, and no known thread-safe method of removing the connection to
		1644	cope with overload is known (to me).
		1645
		1646	One of the easiest ways to handle this situation is to just ignore it
		1647	- when the program encounters an overload, it will just loop until the
		1648	situation is over. While this is a form of busy waiting, no OS offers an
		1649	event-based way to handle this situation, so it's the best one can do.
		1650
		1651	A better way to handle the situation is to log any errors other than
		1652	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1653	messages, and continue as usual, which at least gives the user an idea of
		1654	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1655	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1656	usage.
		1657
		1658	If your program is single-threaded, then you could also keep a dummy file
		1659	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1660	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1661	close that fd, and create a new dummy fd. This will gracefully refuse
		1662	clients under typical overload conditions.
		1663
		1664	The last way to handle it is to simply log the error and C<exit>, as
		1665	is often done with C<malloc> failures, but this results in an easy
		1666	opportunity for a DoS attack.
		1667
		1668	=head3 Watcher-Specific Functions
428		1669
429	=over 4	1670	=over 4
430		1671
431	=item ev_io_init (ev_io *, callback, int fd, int events)	1672	=item ev_io_init (ev_io *, callback, int fd, int events)
432		1673
433	=item ev_io_set (ev_io *, int fd, int events)	1674	=item ev_io_set (ev_io *, int fd, int events)
434		1675
435	Configures an ev_io watcher. The fd is the file descriptor to rceeive	1676	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
436	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1677	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
437	EV_WRITE> to receive the given events.	1678	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1679
		1680	=item int fd [read-only]
		1681
		1682	The file descriptor being watched.
		1683
		1684	=item int events [read-only]
		1685
		1686	The events being watched.
438		1687
439	=back	1688	=back
440		1689
		1690	=head3 Examples
		1691
		1692	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1693	readable, but only once. Since it is likely line-buffered, you could
		1694	attempt to read a whole line in the callback.
		1695
		1696	static void
		1697	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1698	{
		1699	ev_io_stop (loop, w);
		1700	.. read from stdin here (or from w->fd) and handle any I/O errors
		1701	}
		1702
		1703	...
		1704	struct ev_loop *loop = ev_default_init (0);
		1705	ev_io stdin_readable;
		1706	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1707	ev_io_start (loop, &stdin_readable);
		1708	ev_run (loop, 0);
		1709
		1710
441	=head2 struct ev_timer - relative and optionally recurring timeouts	1711	=head2 C<ev_timer> - relative and optionally repeating timeouts
442		1712
443	Timer watchers are simple relative timers that generate an event after a	1713	Timer watchers are simple relative timers that generate an event after a
444	given time, and optionally repeating in regular intervals after that.	1714	given time, and optionally repeating in regular intervals after that.
445		1715
446	The timers are based on real time, that is, if you register an event that	1716	The timers are based on real time, that is, if you register an event that
447	times out after an hour and youreset your system clock to last years	1717	times out after an hour and you reset your system clock to January last
448	time, it will still time out after (roughly) and hour. "Roughly" because	1718	year, it will still time out after (roughly) one hour. "Roughly" because
449	detecting time jumps is hard, and soem inaccuracies are unavoidable (the	1719	detecting time jumps is hard, and some inaccuracies are unavoidable (the
450	monotonic clock option helps a lot here).	1720	monotonic clock option helps a lot here).
451		1721
		1722	The callback is guaranteed to be invoked only I<after> its timeout has
		1723	passed (not I<at>, so on systems with very low-resolution clocks this
		1724	might introduce a small delay). If multiple timers become ready during the
		1725	same loop iteration then the ones with earlier time-out values are invoked
		1726	before ones of the same priority with later time-out values (but this is
		1727	no longer true when a callback calls C<ev_run> recursively).
		1728
		1729	=head3 Be smart about timeouts
		1730
		1731	Many real-world problems involve some kind of timeout, usually for error
		1732	recovery. A typical example is an HTTP request - if the other side hangs,
		1733	you want to raise some error after a while.
		1734
		1735	What follows are some ways to handle this problem, from obvious and
		1736	inefficient to smart and efficient.
		1737
		1738	In the following, a 60 second activity timeout is assumed - a timeout that
		1739	gets reset to 60 seconds each time there is activity (e.g. each time some
		1740	data or other life sign was received).
		1741
452	=over 4	1742	=over 4
453		1743
		1744	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1745
		1746	This is the most obvious, but not the most simple way: In the beginning,
		1747	start the watcher:
		1748
		1749	ev_timer_init (timer, callback, 60., 0.);
		1750	ev_timer_start (loop, timer);
		1751
		1752	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1753	and start it again:
		1754
		1755	ev_timer_stop (loop, timer);
		1756	ev_timer_set (timer, 60., 0.);
		1757	ev_timer_start (loop, timer);
		1758
		1759	This is relatively simple to implement, but means that each time there is
		1760	some activity, libev will first have to remove the timer from its internal
		1761	data structure and then add it again. Libev tries to be fast, but it's
		1762	still not a constant-time operation.
		1763
		1764	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1765
		1766	This is the easiest way, and involves using C<ev_timer_again> instead of
		1767	C<ev_timer_start>.
		1768
		1769	To implement this, configure an C<ev_timer> with a C<repeat> value
		1770	of C<60> and then call C<ev_timer_again> at start and each time you
		1771	successfully read or write some data. If you go into an idle state where
		1772	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1773	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1774
		1775	That means you can ignore both the C<ev_timer_start> function and the
		1776	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1777	member and C<ev_timer_again>.
		1778
		1779	At start:
		1780
		1781	ev_init (timer, callback);
		1782	timer->repeat = 60.;
		1783	ev_timer_again (loop, timer);
		1784
		1785	Each time there is some activity:
		1786
		1787	ev_timer_again (loop, timer);
		1788
		1789	It is even possible to change the time-out on the fly, regardless of
		1790	whether the watcher is active or not:
		1791
		1792	timer->repeat = 30.;
		1793	ev_timer_again (loop, timer);
		1794
		1795	This is slightly more efficient then stopping/starting the timer each time
		1796	you want to modify its timeout value, as libev does not have to completely
		1797	remove and re-insert the timer from/into its internal data structure.
		1798
		1799	It is, however, even simpler than the "obvious" way to do it.
		1800
		1801	=item 3. Let the timer time out, but then re-arm it as required.
		1802
		1803	This method is more tricky, but usually most efficient: Most timeouts are
		1804	relatively long compared to the intervals between other activity - in
		1805	our example, within 60 seconds, there are usually many I/O events with
		1806	associated activity resets.
		1807
		1808	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1809	but remember the time of last activity, and check for a real timeout only
		1810	within the callback:
		1811
		1812	ev_tstamp last_activity; // time of last activity
		1813
		1814	static void
		1815	callback (EV_P_ ev_timer *w, int revents)
		1816	{
		1817	ev_tstamp now = ev_now (EV_A);
		1818	ev_tstamp timeout = last_activity + 60.;
		1819
		1820	// if last_activity + 60. is older than now, we did time out
		1821	if (timeout < now)
		1822	{
		1823	// timeout occurred, take action
		1824	}
		1825	else
		1826	{
		1827	// callback was invoked, but there was some activity, re-arm
		1828	// the watcher to fire in last_activity + 60, which is
		1829	// guaranteed to be in the future, so "again" is positive:
		1830	w->repeat = timeout - now;
		1831	ev_timer_again (EV_A_ w);
		1832	}
		1833	}
		1834
		1835	To summarise the callback: first calculate the real timeout (defined
		1836	as "60 seconds after the last activity"), then check if that time has
		1837	been reached, which means something I<did>, in fact, time out. Otherwise
		1838	the callback was invoked too early (C<timeout> is in the future), so
		1839	re-schedule the timer to fire at that future time, to see if maybe we have
		1840	a timeout then.
		1841
		1842	Note how C<ev_timer_again> is used, taking advantage of the
		1843	C<ev_timer_again> optimisation when the timer is already running.
		1844
		1845	This scheme causes more callback invocations (about one every 60 seconds
		1846	minus half the average time between activity), but virtually no calls to
		1847	libev to change the timeout.
		1848
		1849	To start the timer, simply initialise the watcher and set C<last_activity>
		1850	to the current time (meaning we just have some activity :), then call the
		1851	callback, which will "do the right thing" and start the timer:
		1852
		1853	ev_init (timer, callback);
		1854	last_activity = ev_now (loop);
		1855	callback (loop, timer, EV_TIMER);
		1856
		1857	And when there is some activity, simply store the current time in
		1858	C<last_activity>, no libev calls at all:
		1859
		1860	last_activity = ev_now (loop);
		1861
		1862	This technique is slightly more complex, but in most cases where the
		1863	time-out is unlikely to be triggered, much more efficient.
		1864
		1865	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1866	callback :) - just change the timeout and invoke the callback, which will
		1867	fix things for you.
		1868
		1869	=item 4. Wee, just use a double-linked list for your timeouts.
		1870
		1871	If there is not one request, but many thousands (millions...), all
		1872	employing some kind of timeout with the same timeout value, then one can
		1873	do even better:
		1874
		1875	When starting the timeout, calculate the timeout value and put the timeout
		1876	at the I<end> of the list.
		1877
		1878	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1879	the list is expected to fire (for example, using the technique #3).
		1880
		1881	When there is some activity, remove the timer from the list, recalculate
		1882	the timeout, append it to the end of the list again, and make sure to
		1883	update the C<ev_timer> if it was taken from the beginning of the list.
		1884
		1885	This way, one can manage an unlimited number of timeouts in O(1) time for
		1886	starting, stopping and updating the timers, at the expense of a major
		1887	complication, and having to use a constant timeout. The constant timeout
		1888	ensures that the list stays sorted.
		1889
		1890	=back
		1891
		1892	So which method the best?
		1893
		1894	Method #2 is a simple no-brain-required solution that is adequate in most
		1895	situations. Method #3 requires a bit more thinking, but handles many cases
		1896	better, and isn't very complicated either. In most case, choosing either
		1897	one is fine, with #3 being better in typical situations.
		1898
		1899	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1900	rather complicated, but extremely efficient, something that really pays
		1901	off after the first million or so of active timers, i.e. it's usually
		1902	overkill :)
		1903
		1904	=head3 The special problem of time updates
		1905
		1906	Establishing the current time is a costly operation (it usually takes at
		1907	least two system calls): EV therefore updates its idea of the current
		1908	time only before and after C<ev_run> collects new events, which causes a
		1909	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1910	lots of events in one iteration.
		1911
		1912	The relative timeouts are calculated relative to the C<ev_now ()>
		1913	time. This is usually the right thing as this timestamp refers to the time
		1914	of the event triggering whatever timeout you are modifying/starting. If
		1915	you suspect event processing to be delayed and you I<need> to base the
		1916	timeout on the current time, use something like this to adjust for this:
		1917
		1918	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
		1919
		1920	If the event loop is suspended for a long time, you can also force an
		1921	update of the time returned by C<ev_now ()> by calling C<ev_now_update
		1922	()>.
		1923
		1924	=head3 The special problems of suspended animation
		1925
		1926	When you leave the server world it is quite customary to hit machines that
		1927	can suspend/hibernate - what happens to the clocks during such a suspend?
		1928
		1929	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		1930	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		1931	to run until the system is suspended, but they will not advance while the
		1932	system is suspended. That means, on resume, it will be as if the program
		1933	was frozen for a few seconds, but the suspend time will not be counted
		1934	towards C<ev_timer> when a monotonic clock source is used. The real time
		1935	clock advanced as expected, but if it is used as sole clocksource, then a
		1936	long suspend would be detected as a time jump by libev, and timers would
		1937	be adjusted accordingly.
		1938
		1939	I would not be surprised to see different behaviour in different between
		1940	operating systems, OS versions or even different hardware.
		1941
		1942	The other form of suspend (job control, or sending a SIGSTOP) will see a
		1943	time jump in the monotonic clocks and the realtime clock. If the program
		1944	is suspended for a very long time, and monotonic clock sources are in use,
		1945	then you can expect C<ev_timer>s to expire as the full suspension time
		1946	will be counted towards the timers. When no monotonic clock source is in
		1947	use, then libev will again assume a timejump and adjust accordingly.
		1948
		1949	It might be beneficial for this latter case to call C<ev_suspend>
		1950	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		1951	deterministic behaviour in this case (you can do nothing against
		1952	C<SIGSTOP>).
		1953
		1954	=head3 Watcher-Specific Functions and Data Members
		1955
		1956	=over 4
		1957
454	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1958	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
455		1959
456	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1960	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
457		1961
458	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1962	Configure the timer to trigger after C<after> seconds. If C<repeat>
459	C<0.>, then it will automatically be stopped. If it is positive, then the	1963	is C<0.>, then it will automatically be stopped once the timeout is
460	timer will automatically be configured to trigger again C<repeat> seconds	1964	reached. If it is positive, then the timer will automatically be
461	later, again, and again, until stopped manually.	1965	configured to trigger again C<repeat> seconds later, again, and again,
		1966	until stopped manually.
462		1967
463	The timer itself will do a best-effort at avoiding drift, that is, if you	1968	The timer itself will do a best-effort at avoiding drift, that is, if
464	configure a timer to trigger every 10 seconds, then it will trigger at	1969	you configure a timer to trigger every 10 seconds, then it will normally
465	exactly 10 second intervals. If, however, your program cannot keep up with	1970	trigger at exactly 10 second intervals. If, however, your program cannot
466	the timer (ecause it takes longer than those 10 seconds to do stuff) the	1971	keep up with the timer (because it takes longer than those 10 seconds to
467	timer will not fire more than once per event loop iteration.	1972	do stuff) the timer will not fire more than once per event loop iteration.
468		1973
469	=item ev_timer_again (loop)	1974	=item ev_timer_again (loop, ev_timer *)
470		1975
471	This will act as if the timer timed out and restart it again if it is	1976	This will act as if the timer timed out and restart it again if it is
472	repeating. The exact semantics are:	1977	repeating. The exact semantics are:
473		1978
		1979	If the timer is pending, its pending status is cleared.
		1980
474	If the timer is started but nonrepeating, stop it.	1981	If the timer is started but non-repeating, stop it (as if it timed out).
475		1982
476	If the timer is repeating, either start it if necessary (with the repeat	1983	If the timer is repeating, either start it if necessary (with the
477	value), or reset the running timer to the repeat value.	1984	C<repeat> value), or reset the running timer to the C<repeat> value.
478		1985
479	This sounds a bit complicated, but here is a useful and typical	1986	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
480	example: Imagine you have a tcp connection and you want a so-called idle	1987	usage example.
481	timeout, that is, you want to be called when there have been, say, 60	1988
482	seconds of inactivity on the socket. The easiest way to do this is to	1989	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
483	configure an ev_timer with after=repeat=60 and calling ev_timer_again each	1990
484	time you successfully read or write some data. If you go into an idle	1991	Returns the remaining time until a timer fires. If the timer is active,
485	state where you do not expect data to travel on the socket, you can stop	1992	then this time is relative to the current event loop time, otherwise it's
486	the timer, and again will automatically restart it if need be.	1993	the timeout value currently configured.
		1994
		1995	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		1996	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		1997	will return C<4>. When the timer expires and is restarted, it will return
		1998	roughly C<7> (likely slightly less as callback invocation takes some time,
		1999	too), and so on.
		2000
		2001	=item ev_tstamp repeat [read-write]
		2002
		2003	The current C<repeat> value. Will be used each time the watcher times out
		2004	or C<ev_timer_again> is called, and determines the next timeout (if any),
		2005	which is also when any modifications are taken into account.
487		2006
488	=back	2007	=back
489		2008
		2009	=head3 Examples
		2010
		2011	Example: Create a timer that fires after 60 seconds.
		2012
		2013	static void
		2014	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2015	{
		2016	.. one minute over, w is actually stopped right here
		2017	}
		2018
		2019	ev_timer mytimer;
		2020	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		2021	ev_timer_start (loop, &mytimer);
		2022
		2023	Example: Create a timeout timer that times out after 10 seconds of
		2024	inactivity.
		2025
		2026	static void
		2027	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2028	{
		2029	.. ten seconds without any activity
		2030	}
		2031
		2032	ev_timer mytimer;
		2033	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		2034	ev_timer_again (&mytimer); /* start timer */
		2035	ev_run (loop, 0);
		2036
		2037	// and in some piece of code that gets executed on any "activity":
		2038	// reset the timeout to start ticking again at 10 seconds
		2039	ev_timer_again (&mytimer);
		2040
		2041
490	=head2 ev_periodic - to cron or not to cron it	2042	=head2 C<ev_periodic> - to cron or not to cron?
491		2043
492	Periodic watchers are also timers of a kind, but they are very versatile	2044	Periodic watchers are also timers of a kind, but they are very versatile
493	(and unfortunately a bit complex).	2045	(and unfortunately a bit complex).
494		2046
495	Unlike ev_timer's, they are not based on real time (or relative time)	2047	Unlike C<ev_timer>, periodic watchers are not based on real time (or
496	but on wallclock time (absolute time). You can tell a periodic watcher	2048	relative time, the physical time that passes) but on wall clock time
497	to trigger "at" some specific point in time. For example, if you tell a	2049	(absolute time, the thing you can read on your calender or clock). The
498	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	2050	difference is that wall clock time can run faster or slower than real
499	+ 10.>) and then reset your system clock to the last year, then it will	2051	time, and time jumps are not uncommon (e.g. when you adjust your
500	take a year to trigger the event (unlike an ev_timer, which would trigger	2052	wrist-watch).
501	roughly 10 seconds later and of course not if you reset your system time
502	again).
503		2053
504	They can also be used to implement vastly more complex timers, such as	2054	You can tell a periodic watcher to trigger after some specific point
		2055	in time: for example, if you tell a periodic watcher to trigger "in 10
		2056	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2057	not a delay) and then reset your system clock to January of the previous
		2058	year, then it will take a year or more to trigger the event (unlike an
		2059	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2060	it, as it uses a relative timeout).
		2061
		2062	C<ev_periodic> watchers can also be used to implement vastly more complex
505	triggering an event on eahc midnight, local time.	2063	timers, such as triggering an event on each "midnight, local time", or
		2064	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2065	those cannot react to time jumps.
		2066
		2067	As with timers, the callback is guaranteed to be invoked only when the
		2068	point in time where it is supposed to trigger has passed. If multiple
		2069	timers become ready during the same loop iteration then the ones with
		2070	earlier time-out values are invoked before ones with later time-out values
		2071	(but this is no longer true when a callback calls C<ev_run> recursively).
		2072
		2073	=head3 Watcher-Specific Functions and Data Members
506		2074
507	=over 4	2075	=over 4
508		2076
509	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2077	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
510		2078
511	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2079	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
512		2080
513	Lots of arguments, lets sort it out... There are basically three modes of	2081	Lots of arguments, let's sort it out... There are basically three modes of
514	operation, and we will explain them from simplest to complex:	2082	operation, and we will explain them from simplest to most complex:
515
516		2083
517	=over 4	2084	=over 4
518		2085
519	=item * absolute timer (interval = reschedule_cb = 0)	2086	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
520		2087
521	In this configuration the watcher triggers an event at the wallclock time	2088	In this configuration the watcher triggers an event after the wall clock
522	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2089	time C<offset> has passed. It will not repeat and will not adjust when a
523	that is, if it is to be run at January 1st 2011 then it will run when the	2090	time jump occurs, that is, if it is to be run at January 1st 2011 then it
524	system time reaches or surpasses this time.	2091	will be stopped and invoked when the system clock reaches or surpasses
		2092	this point in time.
525		2093
526	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	2094	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
527		2095
528	In this mode the watcher will always be scheduled to time out at the next	2096	In this mode the watcher will always be scheduled to time out at the next
529	C<at + N * interval> time (for some integer N) and then repeat, regardless	2097	C<offset + N * interval> time (for some integer N, which can also be
530	of any time jumps.	2098	negative) and then repeat, regardless of any time jumps. The C<offset>
		2099	argument is merely an offset into the C<interval> periods.
531		2100
532	This can be used to create timers that do not drift with respect to system	2101	This can be used to create timers that do not drift with respect to the
533	time:	2102	system clock, for example, here is an C<ev_periodic> that triggers each
		2103	hour, on the hour (with respect to UTC):
534		2104
535	ev_periodic_set (&periodic, 0., 3600., 0);	2105	ev_periodic_set (&periodic, 0., 3600., 0);
536		2106
537	This doesn't mean there will always be 3600 seconds in between triggers,	2107	This doesn't mean there will always be 3600 seconds in between triggers,
538	but only that the the callback will be called when the system time shows a	2108	but only that the callback will be called when the system time shows a
539	full hour (UTC), or more correct, when the system time is evenly divisible	2109	full hour (UTC), or more correctly, when the system time is evenly divisible
540	by 3600.	2110	by 3600.
541		2111
542	Another way to think about it (for the mathematically inclined) is that	2112	Another way to think about it (for the mathematically inclined) is that
543	ev_periodic will try to run the callback in this mode at the next possible	2113	C<ev_periodic> will try to run the callback in this mode at the next possible
544	time where C<time = at (mod interval)>, regardless of any time jumps.	2114	time where C<time = offset (mod interval)>, regardless of any time jumps.
545		2115
		2116	For numerical stability it is preferable that the C<offset> value is near
		2117	C<ev_now ()> (the current time), but there is no range requirement for
		2118	this value, and in fact is often specified as zero.
		2119
		2120	Note also that there is an upper limit to how often a timer can fire (CPU
		2121	speed for example), so if C<interval> is very small then timing stability
		2122	will of course deteriorate. Libev itself tries to be exact to be about one
		2123	millisecond (if the OS supports it and the machine is fast enough).
		2124
546	=item * manual reschedule mode (reschedule_cb = callback)	2125	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
547		2126
548	In this mode the values for C<interval> and C<at> are both being	2127	In this mode the values for C<interval> and C<offset> are both being
549	ignored. Instead, each time the periodic watcher gets scheduled, the	2128	ignored. Instead, each time the periodic watcher gets scheduled, the
550	reschedule callback will be called with the watcher as first, and the	2129	reschedule callback will be called with the watcher as first, and the
551	current time as second argument.	2130	current time as second argument.
552		2131
553	NOTE: I<This callback MUST NOT stop or destroy the periodic or any other	2132	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
554	periodic watcher, ever, or make any event loop modificstions>. If you need	2133	or make ANY other event loop modifications whatsoever, unless explicitly
555	to stop it, return 1e30 (or so, fudge fudge) and stop it afterwards.	2134	allowed by documentation here>.
556		2135
		2136	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2137	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2138	only event loop modification you are allowed to do).
		2139
557	Its prototype is c<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2140	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
558	ev_tstamp now)>, e.g.:	2141	*w, ev_tstamp now)>, e.g.:
559		2142
		2143	static ev_tstamp
560	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2144	my_rescheduler (ev_periodic *w, ev_tstamp now)
561	{	2145	{
562	return now + 60.;	2146	return now + 60.;
563	}	2147	}
564		2148
565	It must return the next time to trigger, based on the passed time value	2149	It must return the next time to trigger, based on the passed time value
566	(that is, the lowest time value larger than to the second argument). It	2150	(that is, the lowest time value larger than to the second argument). It
567	will usually be called just before the callback will be triggered, but	2151	will usually be called just before the callback will be triggered, but
568	might be called at other times, too.	2152	might be called at other times, too.
569		2153
		2154	NOTE: I<< This callback must always return a time that is higher than or
		2155	equal to the passed C<now> value >>.
		2156
570	This can be used to create very complex timers, such as a timer that	2157	This can be used to create very complex timers, such as a timer that
571	triggers on each midnight, local time. To do this, you would calculate the	2158	triggers on "next midnight, local time". To do this, you would calculate the
572	next midnight after C<now> and return the timestamp value for this. How you do this	2159	next midnight after C<now> and return the timestamp value for this. How
573	is, again, up to you (but it is not trivial).	2160	you do this is, again, up to you (but it is not trivial, which is the main
		2161	reason I omitted it as an example).
574		2162
575	=back	2163	=back
576		2164
577	=item ev_periodic_again (loop, ev_periodic *)	2165	=item ev_periodic_again (loop, ev_periodic *)
578		2166
579	Simply stops and restarts the periodic watcher again. This is only useful	2167	Simply stops and restarts the periodic watcher again. This is only useful
580	when you changed some parameters or the reschedule callback would return	2168	when you changed some parameters or the reschedule callback would return
581	a different time than the last time it was called (e.g. in a crond like	2169	a different time than the last time it was called (e.g. in a crond like
582	program when the crontabs have changed).	2170	program when the crontabs have changed).
583		2171
		2172	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2173
		2174	When active, returns the absolute time that the watcher is supposed
		2175	to trigger next. This is not the same as the C<offset> argument to
		2176	C<ev_periodic_set>, but indeed works even in interval and manual
		2177	rescheduling modes.
		2178
		2179	=item ev_tstamp offset [read-write]
		2180
		2181	When repeating, this contains the offset value, otherwise this is the
		2182	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2183	although libev might modify this value for better numerical stability).
		2184
		2185	Can be modified any time, but changes only take effect when the periodic
		2186	timer fires or C<ev_periodic_again> is being called.
		2187
		2188	=item ev_tstamp interval [read-write]
		2189
		2190	The current interval value. Can be modified any time, but changes only
		2191	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2192	called.
		2193
		2194	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2195
		2196	The current reschedule callback, or C<0>, if this functionality is
		2197	switched off. Can be changed any time, but changes only take effect when
		2198	the periodic timer fires or C<ev_periodic_again> is being called.
		2199
584	=back	2200	=back
585		2201
		2202	=head3 Examples
		2203
		2204	Example: Call a callback every hour, or, more precisely, whenever the
		2205	system time is divisible by 3600. The callback invocation times have
		2206	potentially a lot of jitter, but good long-term stability.
		2207
		2208	static void
		2209	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
		2210	{
		2211	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2212	}
		2213
		2214	ev_periodic hourly_tick;
		2215	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2216	ev_periodic_start (loop, &hourly_tick);
		2217
		2218	Example: The same as above, but use a reschedule callback to do it:
		2219
		2220	#include <math.h>
		2221
		2222	static ev_tstamp
		2223	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2224	{
		2225	return now + (3600. - fmod (now, 3600.));
		2226	}
		2227
		2228	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2229
		2230	Example: Call a callback every hour, starting now:
		2231
		2232	ev_periodic hourly_tick;
		2233	ev_periodic_init (&hourly_tick, clock_cb,
		2234	fmod (ev_now (loop), 3600.), 3600., 0);
		2235	ev_periodic_start (loop, &hourly_tick);
		2236
		2237
586	=head2 ev_signal - signal me when a signal gets signalled	2238	=head2 C<ev_signal> - signal me when a signal gets signalled!
587		2239
588	Signal watchers will trigger an event when the process receives a specific	2240	Signal watchers will trigger an event when the process receives a specific
589	signal one or more times. Even though signals are very asynchronous, libev	2241	signal one or more times. Even though signals are very asynchronous, libev
590	will try its best to deliver signals synchronously, i.e. as part of the	2242	will try it's best to deliver signals synchronously, i.e. as part of the
591	normal event processing, like any other event.	2243	normal event processing, like any other event.
592		2244
		2245	If you want signals to be delivered truly asynchronously, just use
		2246	C<sigaction> as you would do without libev and forget about sharing
		2247	the signal. You can even use C<ev_async> from a signal handler to
		2248	synchronously wake up an event loop.
		2249
593	You cna configure as many watchers as you like per signal. Only when the	2250	You can configure as many watchers as you like for the same signal, but
		2251	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2252	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2253	C<SIGINT> in both the default loop and another loop at the same time. At
		2254	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2255
594	first watcher gets started will libev actually register a signal watcher	2256	When the first watcher gets started will libev actually register something
595	with the kernel (thus it coexists with your own signal handlers as long	2257	with the kernel (thus it coexists with your own signal handlers as long as
596	as you don't register any with libev). Similarly, when the last signal	2258	you don't register any with libev for the same signal).
597	watcher for a signal is stopped libev will reset the signal handler to	2259
598	SIG_DFL (regardless of what it was set to before).	2260	If possible and supported, libev will install its handlers with
		2261	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2262	not be unduly interrupted. If you have a problem with system calls getting
		2263	interrupted by signals you can block all signals in an C<ev_check> watcher
		2264	and unblock them in an C<ev_prepare> watcher.
		2265
		2266	=head3 The special problem of inheritance over fork/execve/pthread_create
		2267
		2268	Both the signal mask (C<sigprocmask>) and the signal disposition
		2269	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2270	stopping it again), that is, libev might or might not block the signal,
		2271	and might or might not set or restore the installed signal handler.
		2272
		2273	While this does not matter for the signal disposition (libev never
		2274	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2275	C<execve>), this matters for the signal mask: many programs do not expect
		2276	certain signals to be blocked.
		2277
		2278	This means that before calling C<exec> (from the child) you should reset
		2279	the signal mask to whatever "default" you expect (all clear is a good
		2280	choice usually).
		2281
		2282	The simplest way to ensure that the signal mask is reset in the child is
		2283	to install a fork handler with C<pthread_atfork> that resets it. That will
		2284	catch fork calls done by libraries (such as the libc) as well.
		2285
		2286	In current versions of libev, the signal will not be blocked indefinitely
		2287	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2288	the window of opportunity for problems, it will not go away, as libev
		2289	I<has> to modify the signal mask, at least temporarily.
		2290
		2291	So I can't stress this enough: I<If you do not reset your signal mask when
		2292	you expect it to be empty, you have a race condition in your code>. This
		2293	is not a libev-specific thing, this is true for most event libraries.
		2294
		2295	=head3 Watcher-Specific Functions and Data Members
599		2296
600	=over 4	2297	=over 4
601		2298
602	=item ev_signal_init (ev_signal *, callback, int signum)	2299	=item ev_signal_init (ev_signal *, callback, int signum)
603		2300
604	=item ev_signal_set (ev_signal *, int signum)	2301	=item ev_signal_set (ev_signal *, int signum)
605		2302
606	Configures the watcher to trigger on the given signal number (usually one	2303	Configures the watcher to trigger on the given signal number (usually one
607	of the C<SIGxxx> constants).	2304	of the C<SIGxxx> constants).
608		2305
		2306	=item int signum [read-only]
		2307
		2308	The signal the watcher watches out for.
		2309
609	=back	2310	=back
610		2311
		2312	=head3 Examples
		2313
		2314	Example: Try to exit cleanly on SIGINT.
		2315
		2316	static void
		2317	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2318	{
		2319	ev_break (loop, EVBREAK_ALL);
		2320	}
		2321
		2322	ev_signal signal_watcher;
		2323	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2324	ev_signal_start (loop, &signal_watcher);
		2325
		2326
611	=head2 ev_child - wait for pid status changes	2327	=head2 C<ev_child> - watch out for process status changes
612		2328
613	Child watchers trigger when your process receives a SIGCHLD in response to	2329	Child watchers trigger when your process receives a SIGCHLD in response to
614	some child status changes (most typically when a child of yours dies).	2330	some child status changes (most typically when a child of yours dies or
		2331	exits). It is permissible to install a child watcher I<after> the child
		2332	has been forked (which implies it might have already exited), as long
		2333	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2334	forking and then immediately registering a watcher for the child is fine,
		2335	but forking and registering a watcher a few event loop iterations later or
		2336	in the next callback invocation is not.
		2337
		2338	Only the default event loop is capable of handling signals, and therefore
		2339	you can only register child watchers in the default event loop.
		2340
		2341	Due to some design glitches inside libev, child watchers will always be
		2342	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2343	libev)
		2344
		2345	=head3 Process Interaction
		2346
		2347	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2348	initialised. This is necessary to guarantee proper behaviour even if the
		2349	first child watcher is started after the child exits. The occurrence
		2350	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2351	synchronously as part of the event loop processing. Libev always reaps all
		2352	children, even ones not watched.
		2353
		2354	=head3 Overriding the Built-In Processing
		2355
		2356	Libev offers no special support for overriding the built-in child
		2357	processing, but if your application collides with libev's default child
		2358	handler, you can override it easily by installing your own handler for
		2359	C<SIGCHLD> after initialising the default loop, and making sure the
		2360	default loop never gets destroyed. You are encouraged, however, to use an
		2361	event-based approach to child reaping and thus use libev's support for
		2362	that, so other libev users can use C<ev_child> watchers freely.
		2363
		2364	=head3 Stopping the Child Watcher
		2365
		2366	Currently, the child watcher never gets stopped, even when the
		2367	child terminates, so normally one needs to stop the watcher in the
		2368	callback. Future versions of libev might stop the watcher automatically
		2369	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2370	problem).
		2371
		2372	=head3 Watcher-Specific Functions and Data Members
615		2373
616	=over 4	2374	=over 4
617		2375
618	=item ev_child_init (ev_child *, callback, int pid)	2376	=item ev_child_init (ev_child *, callback, int pid, int trace)
619		2377
620	=item ev_child_set (ev_child *, int pid)	2378	=item ev_child_set (ev_child *, int pid, int trace)
621		2379
622	Configures the watcher to wait for status changes of process C<pid> (or	2380	Configures the watcher to wait for status changes of process C<pid> (or
623	I<any> process if C<pid> is specified as C<0>). The callback can look	2381	I<any> process if C<pid> is specified as C<0>). The callback can look
624	at the C<rstatus> member of the C<ev_child> watcher structure to see	2382	at the C<rstatus> member of the C<ev_child> watcher structure to see
625	the status word (use the macros from C<sys/wait.h>). The C<rpid> member	2383	the status word (use the macros from C<sys/wait.h> and see your systems
626	contains the pid of the process causing the status change.	2384	C<waitpid> documentation). The C<rpid> member contains the pid of the
		2385	process causing the status change. C<trace> must be either C<0> (only
		2386	activate the watcher when the process terminates) or C<1> (additionally
		2387	activate the watcher when the process is stopped or continued).
		2388
		2389	=item int pid [read-only]
		2390
		2391	The process id this watcher watches out for, or C<0>, meaning any process id.
		2392
		2393	=item int rpid [read-write]
		2394
		2395	The process id that detected a status change.
		2396
		2397	=item int rstatus [read-write]
		2398
		2399	The process exit/trace status caused by C<rpid> (see your systems
		2400	C<waitpid> and C<sys/wait.h> documentation for details).
627		2401
628	=back	2402	=back
629		2403
		2404	=head3 Examples
		2405
		2406	Example: C<fork()> a new process and install a child handler to wait for
		2407	its completion.
		2408
		2409	ev_child cw;
		2410
		2411	static void
		2412	child_cb (EV_P_ ev_child *w, int revents)
		2413	{
		2414	ev_child_stop (EV_A_ w);
		2415	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2416	}
		2417
		2418	pid_t pid = fork ();
		2419
		2420	if (pid < 0)
		2421	// error
		2422	else if (pid == 0)
		2423	{
		2424	// the forked child executes here
		2425	exit (1);
		2426	}
		2427	else
		2428	{
		2429	ev_child_init (&cw, child_cb, pid, 0);
		2430	ev_child_start (EV_DEFAULT_ &cw);
		2431	}
		2432
		2433
		2434	=head2 C<ev_stat> - did the file attributes just change?
		2435
		2436	This watches a file system path for attribute changes. That is, it calls
		2437	C<stat> on that path in regular intervals (or when the OS says it changed)
		2438	and sees if it changed compared to the last time, invoking the callback if
		2439	it did.
		2440
		2441	The path does not need to exist: changing from "path exists" to "path does
		2442	not exist" is a status change like any other. The condition "path does not
		2443	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2444	C<st_nlink> field being zero (which is otherwise always forced to be at
		2445	least one) and all the other fields of the stat buffer having unspecified
		2446	contents.
		2447
		2448	The path I<must not> end in a slash or contain special components such as
		2449	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2450	your working directory changes, then the behaviour is undefined.
		2451
		2452	Since there is no portable change notification interface available, the
		2453	portable implementation simply calls C<stat(2)> regularly on the path
		2454	to see if it changed somehow. You can specify a recommended polling
		2455	interval for this case. If you specify a polling interval of C<0> (highly
		2456	recommended!) then a I<suitable, unspecified default> value will be used
		2457	(which you can expect to be around five seconds, although this might
		2458	change dynamically). Libev will also impose a minimum interval which is
		2459	currently around C<0.1>, but that's usually overkill.
		2460
		2461	This watcher type is not meant for massive numbers of stat watchers,
		2462	as even with OS-supported change notifications, this can be
		2463	resource-intensive.
		2464
		2465	At the time of this writing, the only OS-specific interface implemented
		2466	is the Linux inotify interface (implementing kqueue support is left as an
		2467	exercise for the reader. Note, however, that the author sees no way of
		2468	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2469
		2470	=head3 ABI Issues (Largefile Support)
		2471
		2472	Libev by default (unless the user overrides this) uses the default
		2473	compilation environment, which means that on systems with large file
		2474	support disabled by default, you get the 32 bit version of the stat
		2475	structure. When using the library from programs that change the ABI to
		2476	use 64 bit file offsets the programs will fail. In that case you have to
		2477	compile libev with the same flags to get binary compatibility. This is
		2478	obviously the case with any flags that change the ABI, but the problem is
		2479	most noticeably displayed with ev_stat and large file support.
		2480
		2481	The solution for this is to lobby your distribution maker to make large
		2482	file interfaces available by default (as e.g. FreeBSD does) and not
		2483	optional. Libev cannot simply switch on large file support because it has
		2484	to exchange stat structures with application programs compiled using the
		2485	default compilation environment.
		2486
		2487	=head3 Inotify and Kqueue
		2488
		2489	When C<inotify (7)> support has been compiled into libev and present at
		2490	runtime, it will be used to speed up change detection where possible. The
		2491	inotify descriptor will be created lazily when the first C<ev_stat>
		2492	watcher is being started.
		2493
		2494	Inotify presence does not change the semantics of C<ev_stat> watchers
		2495	except that changes might be detected earlier, and in some cases, to avoid
		2496	making regular C<stat> calls. Even in the presence of inotify support
		2497	there are many cases where libev has to resort to regular C<stat> polling,
		2498	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2499	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2500	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2501	xfs are fully working) libev usually gets away without polling.
		2502
		2503	There is no support for kqueue, as apparently it cannot be used to
		2504	implement this functionality, due to the requirement of having a file
		2505	descriptor open on the object at all times, and detecting renames, unlinks
		2506	etc. is difficult.
		2507
		2508	=head3 C<stat ()> is a synchronous operation
		2509
		2510	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2511	the process. The exception are C<ev_stat> watchers - those call C<stat
		2512	()>, which is a synchronous operation.
		2513
		2514	For local paths, this usually doesn't matter: unless the system is very
		2515	busy or the intervals between stat's are large, a stat call will be fast,
		2516	as the path data is usually in memory already (except when starting the
		2517	watcher).
		2518
		2519	For networked file systems, calling C<stat ()> can block an indefinite
		2520	time due to network issues, and even under good conditions, a stat call
		2521	often takes multiple milliseconds.
		2522
		2523	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2524	paths, although this is fully supported by libev.
		2525
		2526	=head3 The special problem of stat time resolution
		2527
		2528	The C<stat ()> system call only supports full-second resolution portably,
		2529	and even on systems where the resolution is higher, most file systems
		2530	still only support whole seconds.
		2531
		2532	That means that, if the time is the only thing that changes, you can
		2533	easily miss updates: on the first update, C<ev_stat> detects a change and
		2534	calls your callback, which does something. When there is another update
		2535	within the same second, C<ev_stat> will be unable to detect unless the
		2536	stat data does change in other ways (e.g. file size).
		2537
		2538	The solution to this is to delay acting on a change for slightly more
		2539	than a second (or till slightly after the next full second boundary), using
		2540	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2541	ev_timer_again (loop, w)>).
		2542
		2543	The C<.02> offset is added to work around small timing inconsistencies
		2544	of some operating systems (where the second counter of the current time
		2545	might be be delayed. One such system is the Linux kernel, where a call to
		2546	C<gettimeofday> might return a timestamp with a full second later than
		2547	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2548	update file times then there will be a small window where the kernel uses
		2549	the previous second to update file times but libev might already execute
		2550	the timer callback).
		2551
		2552	=head3 Watcher-Specific Functions and Data Members
		2553
		2554	=over 4
		2555
		2556	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2557
		2558	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2559
		2560	Configures the watcher to wait for status changes of the given
		2561	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2562	be detected and should normally be specified as C<0> to let libev choose
		2563	a suitable value. The memory pointed to by C<path> must point to the same
		2564	path for as long as the watcher is active.
		2565
		2566	The callback will receive an C<EV_STAT> event when a change was detected,
		2567	relative to the attributes at the time the watcher was started (or the
		2568	last change was detected).
		2569
		2570	=item ev_stat_stat (loop, ev_stat *)
		2571
		2572	Updates the stat buffer immediately with new values. If you change the
		2573	watched path in your callback, you could call this function to avoid
		2574	detecting this change (while introducing a race condition if you are not
		2575	the only one changing the path). Can also be useful simply to find out the
		2576	new values.
		2577
		2578	=item ev_statdata attr [read-only]
		2579
		2580	The most-recently detected attributes of the file. Although the type is
		2581	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2582	suitable for your system, but you can only rely on the POSIX-standardised
		2583	members to be present. If the C<st_nlink> member is C<0>, then there was
		2584	some error while C<stat>ing the file.
		2585
		2586	=item ev_statdata prev [read-only]
		2587
		2588	The previous attributes of the file. The callback gets invoked whenever
		2589	C<prev> != C<attr>, or, more precisely, one or more of these members
		2590	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2591	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2592
		2593	=item ev_tstamp interval [read-only]
		2594
		2595	The specified interval.
		2596
		2597	=item const char *path [read-only]
		2598
		2599	The file system path that is being watched.
		2600
		2601	=back
		2602
		2603	=head3 Examples
		2604
		2605	Example: Watch C</etc/passwd> for attribute changes.
		2606
		2607	static void
		2608	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2609	{
		2610	/* /etc/passwd changed in some way */
		2611	if (w->attr.st_nlink)
		2612	{
		2613	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2614	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2615	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2616	}
		2617	else
		2618	/* you shalt not abuse printf for puts */
		2619	puts ("wow, /etc/passwd is not there, expect problems. "
		2620	"if this is windows, they already arrived\n");
		2621	}
		2622
		2623	...
		2624	ev_stat passwd;
		2625
		2626	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2627	ev_stat_start (loop, &passwd);
		2628
		2629	Example: Like above, but additionally use a one-second delay so we do not
		2630	miss updates (however, frequent updates will delay processing, too, so
		2631	one might do the work both on C<ev_stat> callback invocation I<and> on
		2632	C<ev_timer> callback invocation).
		2633
		2634	static ev_stat passwd;
		2635	static ev_timer timer;
		2636
		2637	static void
		2638	timer_cb (EV_P_ ev_timer *w, int revents)
		2639	{
		2640	ev_timer_stop (EV_A_ w);
		2641
		2642	/* now it's one second after the most recent passwd change */
		2643	}
		2644
		2645	static void
		2646	stat_cb (EV_P_ ev_stat *w, int revents)
		2647	{
		2648	/* reset the one-second timer */
		2649	ev_timer_again (EV_A_ &timer);
		2650	}
		2651
		2652	...
		2653	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2654	ev_stat_start (loop, &passwd);
		2655	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2656
		2657
630	=head2 ev_idle - when you've got nothing better to do	2658	=head2 C<ev_idle> - when you've got nothing better to do...
631		2659
632	Idle watchers trigger events when there are no other I/O or timer (or	2660	Idle watchers trigger events when no other events of the same or higher
633	periodic) events pending. That is, as long as your process is busy	2661	priority are pending (prepare, check and other idle watchers do not count
634	handling sockets or timeouts it will not be called. But when your process	2662	as receiving "events").
635	is idle all idle watchers are being called again and again - until	2663
		2664	That is, as long as your process is busy handling sockets or timeouts
		2665	(or even signals, imagine) of the same or higher priority it will not be
		2666	triggered. But when your process is idle (or only lower-priority watchers
		2667	are pending), the idle watchers are being called once per event loop
636	stopped, that is, or your process receives more events.	2668	iteration - until stopped, that is, or your process receives more events
		2669	and becomes busy again with higher priority stuff.
637		2670
638	The most noteworthy effect is that as long as any idle watchers are	2671	The most noteworthy effect is that as long as any idle watchers are
639	active, the process will not block when waiting for new events.	2672	active, the process will not block when waiting for new events.
640		2673
641	Apart from keeping your process non-blocking (which is a useful	2674	Apart from keeping your process non-blocking (which is a useful
642	effect on its own sometimes), idle watchers are a good place to do	2675	effect on its own sometimes), idle watchers are a good place to do
643	"pseudo-background processing", or delay processing stuff to after the	2676	"pseudo-background processing", or delay processing stuff to after the
644	event loop has handled all outstanding events.	2677	event loop has handled all outstanding events.
645		2678
		2679	=head3 Watcher-Specific Functions and Data Members
		2680
646	=over 4	2681	=over 4
647		2682
648	=item ev_idle_init (ev_signal *, callback)	2683	=item ev_idle_init (ev_idle *, callback)
649		2684
650	Initialises and configures the idle watcher - it has no parameters of any	2685	Initialises and configures the idle watcher - it has no parameters of any
651	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2686	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
652	believe me.	2687	believe me.
653		2688
654	=back	2689	=back
655		2690
656	=head2 prepare and check - your hooks into the event loop	2691	=head3 Examples
657		2692
		2693	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2694	callback, free it. Also, use no error checking, as usual.
		2695
		2696	static void
		2697	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2698	{
		2699	free (w);
		2700	// now do something you wanted to do when the program has
		2701	// no longer anything immediate to do.
		2702	}
		2703
		2704	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2705	ev_idle_init (idle_watcher, idle_cb);
		2706	ev_idle_start (loop, idle_watcher);
		2707
		2708
		2709	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
		2710
658	Prepare and check watchers usually (but not always) are used in	2711	Prepare and check watchers are usually (but not always) used in pairs:
659	tandom. Prepare watchers get invoked before the process blocks and check	2712	prepare watchers get invoked before the process blocks and check watchers
660	watchers afterwards.	2713	afterwards.
661		2714
		2715	You I<must not> call C<ev_run> or similar functions that enter
		2716	the current event loop from either C<ev_prepare> or C<ev_check>
		2717	watchers. Other loops than the current one are fine, however. The
		2718	rationale behind this is that you do not need to check for recursion in
		2719	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2720	C<ev_check> so if you have one watcher of each kind they will always be
		2721	called in pairs bracketing the blocking call.
		2722
662	Their main purpose is to integrate other event mechanisms into libev. This	2723	Their main purpose is to integrate other event mechanisms into libev and
663	could be used, for example, to track variable changes, implement your own	2724	their use is somewhat advanced. They could be used, for example, to track
664	watchers, integrate net-snmp or a coroutine library and lots more.	2725	variable changes, implement your own watchers, integrate net-snmp or a
		2726	coroutine library and lots more. They are also occasionally useful if
		2727	you cache some data and want to flush it before blocking (for example,
		2728	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2729	watcher).
665		2730
666	This is done by examining in each prepare call which file descriptors need	2731	This is done by examining in each prepare call which file descriptors
667	to be watched by the other library, registering ev_io watchers for them	2732	need to be watched by the other library, registering C<ev_io> watchers
668	and starting an ev_timer watcher for any timeouts (many libraries provide	2733	for them and starting an C<ev_timer> watcher for any timeouts (many
669	just this functionality). Then, in the check watcher you check for any	2734	libraries provide exactly this functionality). Then, in the check watcher,
670	events that occured (by making your callbacks set soem flags for example)	2735	you check for any events that occurred (by checking the pending status
671	and call back into the library.	2736	of all watchers and stopping them) and call back into the library. The
		2737	I/O and timer callbacks will never actually be called (but must be valid
		2738	nevertheless, because you never know, you know?).
672		2739
673	As another example, the perl Coro module uses these hooks to integrate	2740	As another example, the Perl Coro module uses these hooks to integrate
674	coroutines into libev programs, by yielding to other active coroutines	2741	coroutines into libev programs, by yielding to other active coroutines
675	during each prepare and only letting the process block if no coroutines	2742	during each prepare and only letting the process block if no coroutines
676	are ready to run.	2743	are ready to run (it's actually more complicated: it only runs coroutines
		2744	with priority higher than or equal to the event loop and one coroutine
		2745	of lower priority, but only once, using idle watchers to keep the event
		2746	loop from blocking if lower-priority coroutines are active, thus mapping
		2747	low-priority coroutines to idle/background tasks).
		2748
		2749	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2750	priority, to ensure that they are being run before any other watchers
		2751	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2752
		2753	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2754	activate ("feed") events into libev. While libev fully supports this, they
		2755	might get executed before other C<ev_check> watchers did their job. As
		2756	C<ev_check> watchers are often used to embed other (non-libev) event
		2757	loops those other event loops might be in an unusable state until their
		2758	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2759	others).
		2760
		2761	=head3 Watcher-Specific Functions and Data Members
677		2762
678	=over 4	2763	=over 4
679		2764
680	=item ev_prepare_init (ev_prepare *, callback)	2765	=item ev_prepare_init (ev_prepare *, callback)
681		2766
682	=item ev_check_init (ev_check *, callback)	2767	=item ev_check_init (ev_check *, callback)
683		2768
684	Initialises and configures the prepare or check watcher - they have no	2769	Initialises and configures the prepare or check watcher - they have no
685	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2770	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
686	macros, but using them is utterly, utterly pointless.	2771	macros, but using them is utterly, utterly, utterly and completely
		2772	pointless.
687		2773
688	=back	2774	=back
689		2775
		2776	=head3 Examples
		2777
		2778	There are a number of principal ways to embed other event loops or modules
		2779	into libev. Here are some ideas on how to include libadns into libev
		2780	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2781	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2782	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2783	Glib event loop).
		2784
		2785	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2786	and in a check watcher, destroy them and call into libadns. What follows
		2787	is pseudo-code only of course. This requires you to either use a low
		2788	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2789	the callbacks for the IO/timeout watchers might not have been called yet.
		2790
		2791	static ev_io iow [nfd];
		2792	static ev_timer tw;
		2793
		2794	static void
		2795	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2796	{
		2797	}
		2798
		2799	// create io watchers for each fd and a timer before blocking
		2800	static void
		2801	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2802	{
		2803	int timeout = 3600000;
		2804	struct pollfd fds [nfd];
		2805	// actual code will need to loop here and realloc etc.
		2806	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2807
		2808	/* the callback is illegal, but won't be called as we stop during check */
		2809	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2810	ev_timer_start (loop, &tw);
		2811
		2812	// create one ev_io per pollfd
		2813	for (int i = 0; i < nfd; ++i)
		2814	{
		2815	ev_io_init (iow + i, io_cb, fds [i].fd,
		2816	((fds [i].events & POLLIN ? EV_READ : 0)
		2817	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2818
		2819	fds [i].revents = 0;
		2820	ev_io_start (loop, iow + i);
		2821	}
		2822	}
		2823
		2824	// stop all watchers after blocking
		2825	static void
		2826	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2827	{
		2828	ev_timer_stop (loop, &tw);
		2829
		2830	for (int i = 0; i < nfd; ++i)
		2831	{
		2832	// set the relevant poll flags
		2833	// could also call adns_processreadable etc. here
		2834	struct pollfd *fd = fds + i;
		2835	int revents = ev_clear_pending (iow + i);
		2836	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2837	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2838
		2839	// now stop the watcher
		2840	ev_io_stop (loop, iow + i);
		2841	}
		2842
		2843	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2844	}
		2845
		2846	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2847	in the prepare watcher and would dispose of the check watcher.
		2848
		2849	Method 3: If the module to be embedded supports explicit event
		2850	notification (libadns does), you can also make use of the actual watcher
		2851	callbacks, and only destroy/create the watchers in the prepare watcher.
		2852
		2853	static void
		2854	timer_cb (EV_P_ ev_timer *w, int revents)
		2855	{
		2856	adns_state ads = (adns_state)w->data;
		2857	update_now (EV_A);
		2858
		2859	adns_processtimeouts (ads, &tv_now);
		2860	}
		2861
		2862	static void
		2863	io_cb (EV_P_ ev_io *w, int revents)
		2864	{
		2865	adns_state ads = (adns_state)w->data;
		2866	update_now (EV_A);
		2867
		2868	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2869	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2870	}
		2871
		2872	// do not ever call adns_afterpoll
		2873
		2874	Method 4: Do not use a prepare or check watcher because the module you
		2875	want to embed is not flexible enough to support it. Instead, you can
		2876	override their poll function. The drawback with this solution is that the
		2877	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2878	this approach, effectively embedding EV as a client into the horrible
		2879	libglib event loop.
		2880
		2881	static gint
		2882	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2883	{
		2884	int got_events = 0;
		2885
		2886	for (n = 0; n < nfds; ++n)
		2887	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2888
		2889	if (timeout >= 0)
		2890	// create/start timer
		2891
		2892	// poll
		2893	ev_run (EV_A_ 0);
		2894
		2895	// stop timer again
		2896	if (timeout >= 0)
		2897	ev_timer_stop (EV_A_ &to);
		2898
		2899	// stop io watchers again - their callbacks should have set
		2900	for (n = 0; n < nfds; ++n)
		2901	ev_io_stop (EV_A_ iow [n]);
		2902
		2903	return got_events;
		2904	}
		2905
		2906
		2907	=head2 C<ev_embed> - when one backend isn't enough...
		2908
		2909	This is a rather advanced watcher type that lets you embed one event loop
		2910	into another (currently only C<ev_io> events are supported in the embedded
		2911	loop, other types of watchers might be handled in a delayed or incorrect
		2912	fashion and must not be used).
		2913
		2914	There are primarily two reasons you would want that: work around bugs and
		2915	prioritise I/O.
		2916
		2917	As an example for a bug workaround, the kqueue backend might only support
		2918	sockets on some platform, so it is unusable as generic backend, but you
		2919	still want to make use of it because you have many sockets and it scales
		2920	so nicely. In this case, you would create a kqueue-based loop and embed
		2921	it into your default loop (which might use e.g. poll). Overall operation
		2922	will be a bit slower because first libev has to call C<poll> and then
		2923	C<kevent>, but at least you can use both mechanisms for what they are
		2924	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2925
		2926	As for prioritising I/O: under rare circumstances you have the case where
		2927	some fds have to be watched and handled very quickly (with low latency),
		2928	and even priorities and idle watchers might have too much overhead. In
		2929	this case you would put all the high priority stuff in one loop and all
		2930	the rest in a second one, and embed the second one in the first.
		2931
		2932	As long as the watcher is active, the callback will be invoked every
		2933	time there might be events pending in the embedded loop. The callback
		2934	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		2935	sweep and invoke their callbacks (the callback doesn't need to invoke the
		2936	C<ev_embed_sweep> function directly, it could also start an idle watcher
		2937	to give the embedded loop strictly lower priority for example).
		2938
		2939	You can also set the callback to C<0>, in which case the embed watcher
		2940	will automatically execute the embedded loop sweep whenever necessary.
		2941
		2942	Fork detection will be handled transparently while the C<ev_embed> watcher
		2943	is active, i.e., the embedded loop will automatically be forked when the
		2944	embedding loop forks. In other cases, the user is responsible for calling
		2945	C<ev_loop_fork> on the embedded loop.
		2946
		2947	Unfortunately, not all backends are embeddable: only the ones returned by
		2948	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2949	portable one.
		2950
		2951	So when you want to use this feature you will always have to be prepared
		2952	that you cannot get an embeddable loop. The recommended way to get around
		2953	this is to have a separate variables for your embeddable loop, try to
		2954	create it, and if that fails, use the normal loop for everything.
		2955
		2956	=head3 C<ev_embed> and fork
		2957
		2958	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2959	automatically be applied to the embedded loop as well, so no special
		2960	fork handling is required in that case. When the watcher is not running,
		2961	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2962	as applicable.
		2963
		2964	=head3 Watcher-Specific Functions and Data Members
		2965
		2966	=over 4
		2967
		2968	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2969
		2970	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2971
		2972	Configures the watcher to embed the given loop, which must be
		2973	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2974	invoked automatically, otherwise it is the responsibility of the callback
		2975	to invoke it (it will continue to be called until the sweep has been done,
		2976	if you do not want that, you need to temporarily stop the embed watcher).
		2977
		2978	=item ev_embed_sweep (loop, ev_embed *)
		2979
		2980	Make a single, non-blocking sweep over the embedded loop. This works
		2981	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
		2982	appropriate way for embedded loops.
		2983
		2984	=item struct ev_loop *other [read-only]
		2985
		2986	The embedded event loop.
		2987
		2988	=back
		2989
		2990	=head3 Examples
		2991
		2992	Example: Try to get an embeddable event loop and embed it into the default
		2993	event loop. If that is not possible, use the default loop. The default
		2994	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2995	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2996	used).
		2997
		2998	struct ev_loop *loop_hi = ev_default_init (0);
		2999	struct ev_loop *loop_lo = 0;
		3000	ev_embed embed;
		3001
		3002	// see if there is a chance of getting one that works
		3003	// (remember that a flags value of 0 means autodetection)
		3004	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3005	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3006	: 0;
		3007
		3008	// if we got one, then embed it, otherwise default to loop_hi
		3009	if (loop_lo)
		3010	{
		3011	ev_embed_init (&embed, 0, loop_lo);
		3012	ev_embed_start (loop_hi, &embed);
		3013	}
		3014	else
		3015	loop_lo = loop_hi;
		3016
		3017	Example: Check if kqueue is available but not recommended and create
		3018	a kqueue backend for use with sockets (which usually work with any
		3019	kqueue implementation). Store the kqueue/socket-only event loop in
		3020	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3021
		3022	struct ev_loop *loop = ev_default_init (0);
		3023	struct ev_loop *loop_socket = 0;
		3024	ev_embed embed;
		3025
		3026	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3027	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3028	{
		3029	ev_embed_init (&embed, 0, loop_socket);
		3030	ev_embed_start (loop, &embed);
		3031	}
		3032
		3033	if (!loop_socket)
		3034	loop_socket = loop;
		3035
		3036	// now use loop_socket for all sockets, and loop for everything else
		3037
		3038
		3039	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		3040
		3041	Fork watchers are called when a C<fork ()> was detected (usually because
		3042	whoever is a good citizen cared to tell libev about it by calling
		3043	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		3044	event loop blocks next and before C<ev_check> watchers are being called,
		3045	and only in the child after the fork. If whoever good citizen calling
		3046	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		3047	handlers will be invoked, too, of course.
		3048
		3049	=head3 The special problem of life after fork - how is it possible?
		3050
		3051	Most uses of C<fork()> consist of forking, then some simple calls to set
		3052	up/change the process environment, followed by a call to C<exec()>. This
		3053	sequence should be handled by libev without any problems.
		3054
		3055	This changes when the application actually wants to do event handling
		3056	in the child, or both parent in child, in effect "continuing" after the
		3057	fork.
		3058
		3059	The default mode of operation (for libev, with application help to detect
		3060	forks) is to duplicate all the state in the child, as would be expected
		3061	when I<either> the parent I<or> the child process continues.
		3062
		3063	When both processes want to continue using libev, then this is usually the
		3064	wrong result. In that case, usually one process (typically the parent) is
		3065	supposed to continue with all watchers in place as before, while the other
		3066	process typically wants to start fresh, i.e. without any active watchers.
		3067
		3068	The cleanest and most efficient way to achieve that with libev is to
		3069	simply create a new event loop, which of course will be "empty", and
		3070	use that for new watchers. This has the advantage of not touching more
		3071	memory than necessary, and thus avoiding the copy-on-write, and the
		3072	disadvantage of having to use multiple event loops (which do not support
		3073	signal watchers).
		3074
		3075	When this is not possible, or you want to use the default loop for
		3076	other reasons, then in the process that wants to start "fresh", call
		3077	C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying
		3078	the default loop will "orphan" (not stop) all registered watchers, so you
		3079	have to be careful not to execute code that modifies those watchers. Note
		3080	also that in that case, you have to re-register any signal watchers.
		3081
		3082	=head3 Watcher-Specific Functions and Data Members
		3083
		3084	=over 4
		3085
		3086	=item ev_fork_init (ev_signal *, callback)
		3087
		3088	Initialises and configures the fork watcher - it has no parameters of any
		3089	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		3090	believe me.
		3091
		3092	=back
		3093
		3094
		3095	=head2 C<ev_async> - how to wake up an event loop
		3096
		3097	In general, you cannot use an C<ev_run> from multiple threads or other
		3098	asynchronous sources such as signal handlers (as opposed to multiple event
		3099	loops - those are of course safe to use in different threads).
		3100
		3101	Sometimes, however, you need to wake up an event loop you do not control,
		3102	for example because it belongs to another thread. This is what C<ev_async>
		3103	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3104	it by calling C<ev_async_send>, which is thread- and signal safe.
		3105
		3106	This functionality is very similar to C<ev_signal> watchers, as signals,
		3107	too, are asynchronous in nature, and signals, too, will be compressed
		3108	(i.e. the number of callback invocations may be less than the number of
		3109	C<ev_async_sent> calls).
		3110
		3111	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		3112	just the default loop.
		3113
		3114	=head3 Queueing
		3115
		3116	C<ev_async> does not support queueing of data in any way. The reason
		3117	is that the author does not know of a simple (or any) algorithm for a
		3118	multiple-writer-single-reader queue that works in all cases and doesn't
		3119	need elaborate support such as pthreads or unportable memory access
		3120	semantics.
		3121
		3122	That means that if you want to queue data, you have to provide your own
		3123	queue. But at least I can tell you how to implement locking around your
		3124	queue:
		3125
		3126	=over 4
		3127
		3128	=item queueing from a signal handler context
		3129
		3130	To implement race-free queueing, you simply add to the queue in the signal
		3131	handler but you block the signal handler in the watcher callback. Here is
		3132	an example that does that for some fictitious SIGUSR1 handler:
		3133
		3134	static ev_async mysig;
		3135
		3136	static void
		3137	sigusr1_handler (void)
		3138	{
		3139	sometype data;
		3140
		3141	// no locking etc.
		3142	queue_put (data);
		3143	ev_async_send (EV_DEFAULT_ &mysig);
		3144	}
		3145
		3146	static void
		3147	mysig_cb (EV_P_ ev_async *w, int revents)
		3148	{
		3149	sometype data;
		3150	sigset_t block, prev;
		3151
		3152	sigemptyset (&block);
		3153	sigaddset (&block, SIGUSR1);
		3154	sigprocmask (SIG_BLOCK, &block, &prev);
		3155
		3156	while (queue_get (&data))
		3157	process (data);
		3158
		3159	if (sigismember (&prev, SIGUSR1)
		3160	sigprocmask (SIG_UNBLOCK, &block, 0);
		3161	}
		3162
		3163	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3164	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3165	either...).
		3166
		3167	=item queueing from a thread context
		3168
		3169	The strategy for threads is different, as you cannot (easily) block
		3170	threads but you can easily preempt them, so to queue safely you need to
		3171	employ a traditional mutex lock, such as in this pthread example:
		3172
		3173	static ev_async mysig;
		3174	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3175
		3176	static void
		3177	otherthread (void)
		3178	{
		3179	// only need to lock the actual queueing operation
		3180	pthread_mutex_lock (&mymutex);
		3181	queue_put (data);
		3182	pthread_mutex_unlock (&mymutex);
		3183
		3184	ev_async_send (EV_DEFAULT_ &mysig);
		3185	}
		3186
		3187	static void
		3188	mysig_cb (EV_P_ ev_async *w, int revents)
		3189	{
		3190	pthread_mutex_lock (&mymutex);
		3191
		3192	while (queue_get (&data))
		3193	process (data);
		3194
		3195	pthread_mutex_unlock (&mymutex);
		3196	}
		3197
		3198	=back
		3199
		3200
		3201	=head3 Watcher-Specific Functions and Data Members
		3202
		3203	=over 4
		3204
		3205	=item ev_async_init (ev_async *, callback)
		3206
		3207	Initialises and configures the async watcher - it has no parameters of any
		3208	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3209	trust me.
		3210
		3211	=item ev_async_send (loop, ev_async *)
		3212
		3213	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3214	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		3215	C<ev_feed_event>, this call is safe to do from other threads, signal or
		3216	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		3217	section below on what exactly this means).
		3218
		3219	Note that, as with other watchers in libev, multiple events might get
		3220	compressed into a single callback invocation (another way to look at this
		3221	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3222	reset when the event loop detects that).
		3223
		3224	This call incurs the overhead of a system call only once per event loop
		3225	iteration, so while the overhead might be noticeable, it doesn't apply to
		3226	repeated calls to C<ev_async_send> for the same event loop.
		3227
		3228	=item bool = ev_async_pending (ev_async *)
		3229
		3230	Returns a non-zero value when C<ev_async_send> has been called on the
		3231	watcher but the event has not yet been processed (or even noted) by the
		3232	event loop.
		3233
		3234	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3235	the loop iterates next and checks for the watcher to have become active,
		3236	it will reset the flag again. C<ev_async_pending> can be used to very
		3237	quickly check whether invoking the loop might be a good idea.
		3238
		3239	Not that this does I<not> check whether the watcher itself is pending,
		3240	only whether it has been requested to make this watcher pending: there
		3241	is a time window between the event loop checking and resetting the async
		3242	notification, and the callback being invoked.
		3243
		3244	=back
		3245
		3246
690	=head1 OTHER FUNCTIONS	3247	=head1 OTHER FUNCTIONS
691		3248
692	There are some other fucntions of possible interest. Described. Here. Now.	3249	There are some other functions of possible interest. Described. Here. Now.
693		3250
694	=over 4	3251	=over 4
695		3252
696	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3253	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
697		3254
698	This function combines a simple timer and an I/O watcher, calls your	3255	This function combines a simple timer and an I/O watcher, calls your
699	callback on whichever event happens first and automatically stop both	3256	callback on whichever event happens first and automatically stops both
700	watchers. This is useful if you want to wait for a single event on an fd	3257	watchers. This is useful if you want to wait for a single event on an fd
701	or timeout without havign to allocate/configure/start/stop/free one or	3258	or timeout without having to allocate/configure/start/stop/free one or
702	more watchers yourself.	3259	more watchers yourself.
703		3260
704	If C<fd> is less than 0, then no I/O watcher will be started and events is	3261	If C<fd> is less than 0, then no I/O watcher will be started and the
705	ignored. Otherwise, an ev_io watcher for the given C<fd> and C<events> set	3262	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
706	will be craeted and started.	3263	the given C<fd> and C<events> set will be created and started.
707		3264
708	If C<timeout> is less than 0, then no timeout watcher will be	3265	If C<timeout> is less than 0, then no timeout watcher will be
709	started. Otherwise an ev_timer watcher with after = C<timeout> (and repeat	3266	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
710	= 0) will be started.	3267	repeat = 0) will be started. C<0> is a valid timeout.
711		3268
712	The callback has the type C<void (*cb)(int revents, void *arg)> and	3269	The callback has the type C<void (*cb)(int revents, void *arg)> and is
713	gets passed an events set (normally a combination of EV_ERROR, EV_READ,	3270	passed an C<revents> set like normal event callbacks (a combination of
714	EV_WRITE or EV_TIMEOUT) and the C<arg> value passed to C<ev_once>:	3271	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
		3272	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3273	a timeout and an io event at the same time - you probably should give io
		3274	events precedence.
715		3275
		3276	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3277
716	static void stdin_ready (int revents, void *arg)	3278	static void stdin_ready (int revents, void *arg)
717	{	3279	{
718	if (revents & EV_TIMEOUT)
719	/* doh, nothing entered */
720	else if (revents & EV_READ)	3280	if (revents & EV_READ)
721	/* stdin might have data for us, joy! */	3281	/* stdin might have data for us, joy! */;
		3282	else if (revents & EV_TIMER)
		3283	/* doh, nothing entered */;
722	}	3284	}
723		3285
724	ev_once (STDIN_FILENO, EV_READm 10., stdin_ready, 0);	3286	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
725
726	=item ev_feed_event (loop, watcher, int events)
727
728	Feeds the given event set into the event loop, as if the specified event
729	has happened for the specified watcher (which must be a pointer to an
730	initialised but not necessarily active event watcher).
731		3287
732	=item ev_feed_fd_event (loop, int fd, int revents)	3288	=item ev_feed_fd_event (loop, int fd, int revents)
733		3289
734	Feed an event on the given fd, as if a file descriptor backend detected it.	3290	Feed an event on the given fd, as if a file descriptor backend detected
		3291	the given events it.
735		3292
736	=item ev_feed_signal_event (loop, int signum)	3293	=item ev_feed_signal_event (loop, int signum)
737		3294
738	Feed an event as if the given signal occured (loop must be the default loop!).	3295	Feed an event as if the given signal occurred (C<loop> must be the default
		3296	loop!).
739		3297
740	=back	3298	=back
741		3299
		3300
		3301	=head1 LIBEVENT EMULATION
		3302
		3303	Libev offers a compatibility emulation layer for libevent. It cannot
		3304	emulate the internals of libevent, so here are some usage hints:
		3305
		3306	=over 4
		3307
		3308	=item * Use it by including <event.h>, as usual.
		3309
		3310	=item * The following members are fully supported: ev_base, ev_callback,
		3311	ev_arg, ev_fd, ev_res, ev_events.
		3312
		3313	=item * Avoid using ev_flags and the EVLIST_*-macros, while it is
		3314	maintained by libev, it does not work exactly the same way as in libevent (consider
		3315	it a private API).
		3316
		3317	=item * Priorities are not currently supported. Initialising priorities
		3318	will fail and all watchers will have the same priority, even though there
		3319	is an ev_pri field.
		3320
		3321	=item * In libevent, the last base created gets the signals, in libev, the
		3322	first base created (== the default loop) gets the signals.
		3323
		3324	=item * Other members are not supported.
		3325
		3326	=item * The libev emulation is I<not> ABI compatible to libevent, you need
		3327	to use the libev header file and library.
		3328
		3329	=back
		3330
		3331	=head1 C++ SUPPORT
		3332
		3333	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3334	you to use some convenience methods to start/stop watchers and also change
		3335	the callback model to a model using method callbacks on objects.
		3336
		3337	To use it,
		3338
		3339	#include <ev++.h>
		3340
		3341	This automatically includes F<ev.h> and puts all of its definitions (many
		3342	of them macros) into the global namespace. All C++ specific things are
		3343	put into the C<ev> namespace. It should support all the same embedding
		3344	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3345
		3346	Care has been taken to keep the overhead low. The only data member the C++
		3347	classes add (compared to plain C-style watchers) is the event loop pointer
		3348	that the watcher is associated with (or no additional members at all if
		3349	you disable C<EV_MULTIPLICITY> when embedding libev).
		3350
		3351	Currently, functions, and static and non-static member functions can be
		3352	used as callbacks. Other types should be easy to add as long as they only
		3353	need one additional pointer for context. If you need support for other
		3354	types of functors please contact the author (preferably after implementing
		3355	it).
		3356
		3357	Here is a list of things available in the C<ev> namespace:
		3358
		3359	=over 4
		3360
		3361	=item C<ev::READ>, C<ev::WRITE> etc.
		3362
		3363	These are just enum values with the same values as the C<EV_READ> etc.
		3364	macros from F<ev.h>.
		3365
		3366	=item C<ev::tstamp>, C<ev::now>
		3367
		3368	Aliases to the same types/functions as with the C<ev_> prefix.
		3369
		3370	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3371
		3372	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3373	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3374	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3375	defines by many implementations.
		3376
		3377	All of those classes have these methods:
		3378
		3379	=over 4
		3380
		3381	=item ev::TYPE::TYPE ()
		3382
		3383	=item ev::TYPE::TYPE (loop)
		3384
		3385	=item ev::TYPE::~TYPE
		3386
		3387	The constructor (optionally) takes an event loop to associate the watcher
		3388	with. If it is omitted, it will use C<EV_DEFAULT>.
		3389
		3390	The constructor calls C<ev_init> for you, which means you have to call the
		3391	C<set> method before starting it.
		3392
		3393	It will not set a callback, however: You have to call the templated C<set>
		3394	method to set a callback before you can start the watcher.
		3395
		3396	(The reason why you have to use a method is a limitation in C++ which does
		3397	not allow explicit template arguments for constructors).
		3398
		3399	The destructor automatically stops the watcher if it is active.
		3400
		3401	=item w->set<class, &class::method> (object *)
		3402
		3403	This method sets the callback method to call. The method has to have a
		3404	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3405	first argument and the C<revents> as second. The object must be given as
		3406	parameter and is stored in the C<data> member of the watcher.
		3407
		3408	This method synthesizes efficient thunking code to call your method from
		3409	the C callback that libev requires. If your compiler can inline your
		3410	callback (i.e. it is visible to it at the place of the C<set> call and
		3411	your compiler is good :), then the method will be fully inlined into the
		3412	thunking function, making it as fast as a direct C callback.
		3413
		3414	Example: simple class declaration and watcher initialisation
		3415
		3416	struct myclass
		3417	{
		3418	void io_cb (ev::io &w, int revents) { }
		3419	}
		3420
		3421	myclass obj;
		3422	ev::io iow;
		3423	iow.set <myclass, &myclass::io_cb> (&obj);
		3424
		3425	=item w->set (object *)
		3426
		3427	This is a variation of a method callback - leaving out the method to call
		3428	will default the method to C<operator ()>, which makes it possible to use
		3429	functor objects without having to manually specify the C<operator ()> all
		3430	the time. Incidentally, you can then also leave out the template argument
		3431	list.
		3432
		3433	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3434	int revents)>.
		3435
		3436	See the method-C<set> above for more details.
		3437
		3438	Example: use a functor object as callback.
		3439
		3440	struct myfunctor
		3441	{
		3442	void operator() (ev::io &w, int revents)
		3443	{
		3444	...
		3445	}
		3446	}
		3447
		3448	myfunctor f;
		3449
		3450	ev::io w;
		3451	w.set (&f);
		3452
		3453	=item w->set<function> (void *data = 0)
		3454
		3455	Also sets a callback, but uses a static method or plain function as
		3456	callback. The optional C<data> argument will be stored in the watcher's
		3457	C<data> member and is free for you to use.
		3458
		3459	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3460
		3461	See the method-C<set> above for more details.
		3462
		3463	Example: Use a plain function as callback.
		3464
		3465	static void io_cb (ev::io &w, int revents) { }
		3466	iow.set <io_cb> ();
		3467
		3468	=item w->set (loop)
		3469
		3470	Associates a different C<struct ev_loop> with this watcher. You can only
		3471	do this when the watcher is inactive (and not pending either).
		3472
		3473	=item w->set ([arguments])
		3474
		3475	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this
		3476	method or a suitable start method must be called at least once. Unlike the
		3477	C counterpart, an active watcher gets automatically stopped and restarted
		3478	when reconfiguring it with this method.
		3479
		3480	=item w->start ()
		3481
		3482	Starts the watcher. Note that there is no C<loop> argument, as the
		3483	constructor already stores the event loop.
		3484
		3485	=item w->start ([arguments])
		3486
		3487	Instead of calling C<set> and C<start> methods separately, it is often
		3488	convenient to wrap them in one call. Uses the same type of arguments as
		3489	the configure C<set> method of the watcher.
		3490
		3491	=item w->stop ()
		3492
		3493	Stops the watcher if it is active. Again, no C<loop> argument.
		3494
		3495	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3496
		3497	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3498	C<ev_TYPE_again> function.
		3499
		3500	=item w->sweep () (C<ev::embed> only)
		3501
		3502	Invokes C<ev_embed_sweep>.
		3503
		3504	=item w->update () (C<ev::stat> only)
		3505
		3506	Invokes C<ev_stat_stat>.
		3507
		3508	=back
		3509
		3510	=back
		3511
		3512	Example: Define a class with two I/O and idle watchers, start the I/O
		3513	watchers in the constructor.
		3514
		3515	class myclass
		3516	{
		3517	ev::io io ; void io_cb (ev::io &w, int revents);
		3518	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);
		3519	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3520
		3521	myclass (int fd)
		3522	{
		3523	io .set <myclass, &myclass::io_cb > (this);
		3524	io2 .set <myclass, &myclass::io2_cb > (this);
		3525	idle.set <myclass, &myclass::idle_cb> (this);
		3526
		3527	io.set (fd, ev::WRITE); // configure the watcher
		3528	io.start (); // start it whenever convenient
		3529
		3530	io2.start (fd, ev::READ); // set + start in one call
		3531	}
		3532	};
		3533
		3534
		3535	=head1 OTHER LANGUAGE BINDINGS
		3536
		3537	Libev does not offer other language bindings itself, but bindings for a
		3538	number of languages exist in the form of third-party packages. If you know
		3539	any interesting language binding in addition to the ones listed here, drop
		3540	me a note.
		3541
		3542	=over 4
		3543
		3544	=item Perl
		3545
		3546	The EV module implements the full libev API and is actually used to test
		3547	libev. EV is developed together with libev. Apart from the EV core module,
		3548	there are additional modules that implement libev-compatible interfaces
		3549	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3550	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3551	and C<EV::Glib>).
		3552
		3553	It can be found and installed via CPAN, its homepage is at
		3554	L<http://software.schmorp.de/pkg/EV>.
		3555
		3556	=item Python
		3557
		3558	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3559	seems to be quite complete and well-documented.
		3560
		3561	=item Ruby
		3562
		3563	Tony Arcieri has written a ruby extension that offers access to a subset
		3564	of the libev API and adds file handle abstractions, asynchronous DNS and
		3565	more on top of it. It can be found via gem servers. Its homepage is at
		3566	L<http://rev.rubyforge.org/>.
		3567
		3568	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3569	makes rev work even on mingw.
		3570
		3571	=item Haskell
		3572
		3573	A haskell binding to libev is available at
		3574	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3575
		3576	=item D
		3577
		3578	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3579	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3580
		3581	=item Ocaml
		3582
		3583	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3584	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3585
		3586	=item Lua
		3587
		3588	Brian Maher has written a partial interface to libev for lua (at the
		3589	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		3590	L<http://github.com/brimworks/lua-ev>.
		3591
		3592	=back
		3593
		3594
		3595	=head1 MACRO MAGIC
		3596
		3597	Libev can be compiled with a variety of options, the most fundamental
		3598	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3599	functions and callbacks have an initial C<struct ev_loop *> argument.
		3600
		3601	To make it easier to write programs that cope with either variant, the
		3602	following macros are defined:
		3603
		3604	=over 4
		3605
		3606	=item C<EV_A>, C<EV_A_>
		3607
		3608	This provides the loop I<argument> for functions, if one is required ("ev
		3609	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3610	C<EV_A_> is used when other arguments are following. Example:
		3611
		3612	ev_unref (EV_A);
		3613	ev_timer_add (EV_A_ watcher);
		3614	ev_run (EV_A_ 0);
		3615
		3616	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3617	which is often provided by the following macro.
		3618
		3619	=item C<EV_P>, C<EV_P_>
		3620
		3621	This provides the loop I<parameter> for functions, if one is required ("ev
		3622	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3623	C<EV_P_> is used when other parameters are following. Example:
		3624
		3625	// this is how ev_unref is being declared
		3626	static void ev_unref (EV_P);
		3627
		3628	// this is how you can declare your typical callback
		3629	static void cb (EV_P_ ev_timer *w, int revents)
		3630
		3631	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3632	suitable for use with C<EV_A>.
		3633
		3634	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3635
		3636	Similar to the other two macros, this gives you the value of the default
		3637	loop, if multiple loops are supported ("ev loop default").
		3638
		3639	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3640
		3641	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3642	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3643	is undefined when the default loop has not been initialised by a previous
		3644	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3645
		3646	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3647	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3648
		3649	=back
		3650
		3651	Example: Declare and initialise a check watcher, utilising the above
		3652	macros so it will work regardless of whether multiple loops are supported
		3653	or not.
		3654
		3655	static void
		3656	check_cb (EV_P_ ev_timer *w, int revents)
		3657	{
		3658	ev_check_stop (EV_A_ w);
		3659	}
		3660
		3661	ev_check check;
		3662	ev_check_init (&check, check_cb);
		3663	ev_check_start (EV_DEFAULT_ &check);
		3664	ev_run (EV_DEFAULT_ 0);
		3665
		3666	=head1 EMBEDDING
		3667
		3668	Libev can (and often is) directly embedded into host
		3669	applications. Examples of applications that embed it include the Deliantra
		3670	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3671	and rxvt-unicode.
		3672
		3673	The goal is to enable you to just copy the necessary files into your
		3674	source directory without having to change even a single line in them, so
		3675	you can easily upgrade by simply copying (or having a checked-out copy of
		3676	libev somewhere in your source tree).
		3677
		3678	=head2 FILESETS
		3679
		3680	Depending on what features you need you need to include one or more sets of files
		3681	in your application.
		3682
		3683	=head3 CORE EVENT LOOP
		3684
		3685	To include only the libev core (all the C<ev_*> functions), with manual
		3686	configuration (no autoconf):
		3687
		3688	#define EV_STANDALONE 1
		3689	#include "ev.c"
		3690
		3691	This will automatically include F<ev.h>, too, and should be done in a
		3692	single C source file only to provide the function implementations. To use
		3693	it, do the same for F<ev.h> in all files wishing to use this API (best
		3694	done by writing a wrapper around F<ev.h> that you can include instead and
		3695	where you can put other configuration options):
		3696
		3697	#define EV_STANDALONE 1
		3698	#include "ev.h"
		3699
		3700	Both header files and implementation files can be compiled with a C++
		3701	compiler (at least, that's a stated goal, and breakage will be treated
		3702	as a bug).
		3703
		3704	You need the following files in your source tree, or in a directory
		3705	in your include path (e.g. in libev/ when using -Ilibev):
		3706
		3707	ev.h
		3708	ev.c
		3709	ev_vars.h
		3710	ev_wrap.h
		3711
		3712	ev_win32.c required on win32 platforms only
		3713
		3714	ev_select.c only when select backend is enabled (which is enabled by default)
		3715	ev_poll.c only when poll backend is enabled (disabled by default)
		3716	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3717	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3718	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3719
		3720	F<ev.c> includes the backend files directly when enabled, so you only need
		3721	to compile this single file.
		3722
		3723	=head3 LIBEVENT COMPATIBILITY API
		3724
		3725	To include the libevent compatibility API, also include:
		3726
		3727	#include "event.c"
		3728
		3729	in the file including F<ev.c>, and:
		3730
		3731	#include "event.h"
		3732
		3733	in the files that want to use the libevent API. This also includes F<ev.h>.
		3734
		3735	You need the following additional files for this:
		3736
		3737	event.h
		3738	event.c
		3739
		3740	=head3 AUTOCONF SUPPORT
		3741
		3742	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3743	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3744	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3745	include F<config.h> and configure itself accordingly.
		3746
		3747	For this of course you need the m4 file:
		3748
		3749	libev.m4
		3750
		3751	=head2 PREPROCESSOR SYMBOLS/MACROS
		3752
		3753	Libev can be configured via a variety of preprocessor symbols you have to
		3754	define before including (or compiling) any of its files. The default in
		3755	the absence of autoconf is documented for every option.
		3756
		3757	Symbols marked with "(h)" do not change the ABI, and can have different
		3758	values when compiling libev vs. including F<ev.h>, so it is permissible
		3759	to redefine them before including F<ev.h> without breaking compatibility
		3760	to a compiled library. All other symbols change the ABI, which means all
		3761	users of libev and the libev code itself must be compiled with compatible
		3762	settings.
		3763
		3764	=over 4
		3765
		3766	=item EV_COMPAT3 (h)
		3767
		3768	Backwards compatibility is a major concern for libev. This is why this
		3769	release of libev comes with wrappers for the functions and symbols that
		3770	have been renamed between libev version 3 and 4.
		3771
		3772	You can disable these wrappers (to test compatibility with future
		3773	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		3774	sources. This has the additional advantage that you can drop the C<struct>
		3775	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		3776	typedef in that case.
		3777
		3778	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		3779	and in some even more future version the compatibility code will be
		3780	removed completely.
		3781
		3782	=item EV_STANDALONE (h)
		3783
		3784	Must always be C<1> if you do not use autoconf configuration, which
		3785	keeps libev from including F<config.h>, and it also defines dummy
		3786	implementations for some libevent functions (such as logging, which is not
		3787	supported). It will also not define any of the structs usually found in
		3788	F<event.h> that are not directly supported by the libev core alone.
		3789
		3790	In standalone mode, libev will still try to automatically deduce the
		3791	configuration, but has to be more conservative.
		3792
		3793	=item EV_USE_MONOTONIC
		3794
		3795	If defined to be C<1>, libev will try to detect the availability of the
		3796	monotonic clock option at both compile time and runtime. Otherwise no
		3797	use of the monotonic clock option will be attempted. If you enable this,
		3798	you usually have to link against librt or something similar. Enabling it
		3799	when the functionality isn't available is safe, though, although you have
		3800	to make sure you link against any libraries where the C<clock_gettime>
		3801	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		3802
		3803	=item EV_USE_REALTIME
		3804
		3805	If defined to be C<1>, libev will try to detect the availability of the
		3806	real-time clock option at compile time (and assume its availability
		3807	at runtime if successful). Otherwise no use of the real-time clock
		3808	option will be attempted. This effectively replaces C<gettimeofday>
		3809	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		3810	correctness. See the note about libraries in the description of
		3811	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3812	C<EV_USE_CLOCK_SYSCALL>.
		3813
		3814	=item EV_USE_CLOCK_SYSCALL
		3815
		3816	If defined to be C<1>, libev will try to use a direct syscall instead
		3817	of calling the system-provided C<clock_gettime> function. This option
		3818	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3819	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3820	programs needlessly. Using a direct syscall is slightly slower (in
		3821	theory), because no optimised vdso implementation can be used, but avoids
		3822	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		3823	higher, as it simplifies linking (no need for C<-lrt>).
		3824
		3825	=item EV_USE_NANOSLEEP
		3826
		3827	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3828	and will use it for delays. Otherwise it will use C<select ()>.
		3829
		3830	=item EV_USE_EVENTFD
		3831
		3832	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3833	available and will probe for kernel support at runtime. This will improve
		3834	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3835	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3836	2.7 or newer, otherwise disabled.
		3837
		3838	=item EV_USE_SELECT
		3839
		3840	If undefined or defined to be C<1>, libev will compile in support for the
		3841	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3842	other method takes over, select will be it. Otherwise the select backend
		3843	will not be compiled in.
		3844
		3845	=item EV_SELECT_USE_FD_SET
		3846
		3847	If defined to C<1>, then the select backend will use the system C<fd_set>
		3848	structure. This is useful if libev doesn't compile due to a missing
		3849	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		3850	on exotic systems. This usually limits the range of file descriptors to
		3851	some low limit such as 1024 or might have other limitations (winsocket
		3852	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		3853	configures the maximum size of the C<fd_set>.
		3854
		3855	=item EV_SELECT_IS_WINSOCKET
		3856
		3857	When defined to C<1>, the select backend will assume that
		3858	select/socket/connect etc. don't understand file descriptors but
		3859	wants osf handles on win32 (this is the case when the select to
		3860	be used is the winsock select). This means that it will call
		3861	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3862	it is assumed that all these functions actually work on fds, even
		3863	on win32. Should not be defined on non-win32 platforms.
		3864
		3865	=item EV_FD_TO_WIN32_HANDLE(fd)
		3866
		3867	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3868	file descriptors to socket handles. When not defining this symbol (the
		3869	default), then libev will call C<_get_osfhandle>, which is usually
		3870	correct. In some cases, programs use their own file descriptor management,
		3871	in which case they can provide this function to map fds to socket handles.
		3872
		3873	=item EV_WIN32_HANDLE_TO_FD(handle)
		3874
		3875	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		3876	using the standard C<_open_osfhandle> function. For programs implementing
		3877	their own fd to handle mapping, overwriting this function makes it easier
		3878	to do so. This can be done by defining this macro to an appropriate value.
		3879
		3880	=item EV_WIN32_CLOSE_FD(fd)
		3881
		3882	If programs implement their own fd to handle mapping on win32, then this
		3883	macro can be used to override the C<close> function, useful to unregister
		3884	file descriptors again. Note that the replacement function has to close
		3885	the underlying OS handle.
		3886
		3887	=item EV_USE_POLL
		3888
		3889	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3890	backend. Otherwise it will be enabled on non-win32 platforms. It
		3891	takes precedence over select.
		3892
		3893	=item EV_USE_EPOLL
		3894
		3895	If defined to be C<1>, libev will compile in support for the Linux
		3896	C<epoll>(7) backend. Its availability will be detected at runtime,
		3897	otherwise another method will be used as fallback. This is the preferred
		3898	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3899	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3900
		3901	=item EV_USE_KQUEUE
		3902
		3903	If defined to be C<1>, libev will compile in support for the BSD style
		3904	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3905	otherwise another method will be used as fallback. This is the preferred
		3906	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3907	supports some types of fds correctly (the only platform we found that
		3908	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3909	not be used unless explicitly requested. The best way to use it is to find
		3910	out whether kqueue supports your type of fd properly and use an embedded
		3911	kqueue loop.
		3912
		3913	=item EV_USE_PORT
		3914
		3915	If defined to be C<1>, libev will compile in support for the Solaris
		3916	10 port style backend. Its availability will be detected at runtime,
		3917	otherwise another method will be used as fallback. This is the preferred
		3918	backend for Solaris 10 systems.
		3919
		3920	=item EV_USE_DEVPOLL
		3921
		3922	Reserved for future expansion, works like the USE symbols above.
		3923
		3924	=item EV_USE_INOTIFY
		3925
		3926	If defined to be C<1>, libev will compile in support for the Linux inotify
		3927	interface to speed up C<ev_stat> watchers. Its actual availability will
		3928	be detected at runtime. If undefined, it will be enabled if the headers
		3929	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3930
		3931	=item EV_ATOMIC_T
		3932
		3933	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3934	access is atomic with respect to other threads or signal contexts. No such
		3935	type is easily found in the C language, so you can provide your own type
		3936	that you know is safe for your purposes. It is used both for signal handler "locking"
		3937	as well as for signal and thread safety in C<ev_async> watchers.
		3938
		3939	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3940	(from F<signal.h>), which is usually good enough on most platforms.
		3941
		3942	=item EV_H (h)
		3943
		3944	The name of the F<ev.h> header file used to include it. The default if
		3945	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3946	used to virtually rename the F<ev.h> header file in case of conflicts.
		3947
		3948	=item EV_CONFIG_H (h)
		3949
		3950	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3951	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3952	C<EV_H>, above.
		3953
		3954	=item EV_EVENT_H (h)
		3955
		3956	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3957	of how the F<event.h> header can be found, the default is C<"event.h">.
		3958
		3959	=item EV_PROTOTYPES (h)
		3960
		3961	If defined to be C<0>, then F<ev.h> will not define any function
		3962	prototypes, but still define all the structs and other symbols. This is
		3963	occasionally useful if you want to provide your own wrapper functions
		3964	around libev functions.
		3965
		3966	=item EV_MULTIPLICITY
		3967
		3968	If undefined or defined to C<1>, then all event-loop-specific functions
		3969	will have the C<struct ev_loop *> as first argument, and you can create
		3970	additional independent event loops. Otherwise there will be no support
		3971	for multiple event loops and there is no first event loop pointer
		3972	argument. Instead, all functions act on the single default loop.
		3973
		3974	=item EV_MINPRI
		3975
		3976	=item EV_MAXPRI
		3977
		3978	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3979	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3980	provide for more priorities by overriding those symbols (usually defined
		3981	to be C<-2> and C<2>, respectively).
		3982
		3983	When doing priority-based operations, libev usually has to linearly search
		3984	all the priorities, so having many of them (hundreds) uses a lot of space
		3985	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3986	fine.
		3987
		3988	If your embedding application does not need any priorities, defining these
		3989	both to C<0> will save some memory and CPU.
		3990
		3991	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		3992	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		3993	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
		3994
		3995	If undefined or defined to be C<1> (and the platform supports it), then
		3996	the respective watcher type is supported. If defined to be C<0>, then it
		3997	is not. Disabling watcher types mainly saves code size.
		3998
		3999	=item EV_FEATURES
		4000
		4001	If you need to shave off some kilobytes of code at the expense of some
		4002	speed (but with the full API), you can define this symbol to request
		4003	certain subsets of functionality. The default is to enable all features
		4004	that can be enabled on the platform.
		4005
		4006	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4007	with some broad features you want) and then selectively re-enable
		4008	additional parts you want, for example if you want everything minimal,
		4009	but multiple event loop support, async and child watchers and the poll
		4010	backend, use this:
		4011
		4012	#define EV_FEATURES 0
		4013	#define EV_MULTIPLICITY 1
		4014	#define EV_USE_POLL 1
		4015	#define EV_CHILD_ENABLE 1
		4016	#define EV_ASYNC_ENABLE 1
		4017
		4018	The actual value is a bitset, it can be a combination of the following
		4019	values:
		4020
		4021	=over 4
		4022
		4023	=item C<1> - faster/larger code
		4024
		4025	Use larger code to speed up some operations.
		4026
		4027	Currently this is used to override some inlining decisions (enlarging the
		4028	code size by roughly 30% on amd64).
		4029
		4030	When optimising for size, use of compiler flags such as C<-Os> with
		4031	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4032	assertions.
		4033
		4034	=item C<2> - faster/larger data structures
		4035
		4036	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4037	hash table sizes and so on. This will usually further increase code size
		4038	and can additionally have an effect on the size of data structures at
		4039	runtime.
		4040
		4041	=item C<4> - full API configuration
		4042
		4043	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4044	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4045
		4046	=item C<8> - full API
		4047
		4048	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4049	details on which parts of the API are still available without this
		4050	feature, and do not complain if this subset changes over time.
		4051
		4052	=item C<16> - enable all optional watcher types
		4053
		4054	Enables all optional watcher types. If you want to selectively enable
		4055	only some watcher types other than I/O and timers (e.g. prepare,
		4056	embed, async, child...) you can enable them manually by defining
		4057	C<EV_watchertype_ENABLE> to C<1> instead.
		4058
		4059	=item C<32> - enable all backends
		4060
		4061	This enables all backends - without this feature, you need to enable at
		4062	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4063
		4064	=item C<64> - enable OS-specific "helper" APIs
		4065
		4066	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4067	default.
		4068
		4069	=back
		4070
		4071	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4072	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4073	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4074	watchers, timers and monotonic clock support.
		4075
		4076	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4077	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4078	your program might be left out as well - a binary starting a timer and an
		4079	I/O watcher then might come out at only 5Kb.
		4080
		4081	=item EV_AVOID_STDIO
		4082
		4083	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4084	functions (printf, scanf, perror etc.). This will increase the code size
		4085	somewhat, but if your program doesn't otherwise depend on stdio and your
		4086	libc allows it, this avoids linking in the stdio library which is quite
		4087	big.
		4088
		4089	Note that error messages might become less precise when this option is
		4090	enabled.
		4091
		4092	=item EV_NSIG
		4093
		4094	The highest supported signal number, +1 (or, the number of
		4095	signals): Normally, libev tries to deduce the maximum number of signals
		4096	automatically, but sometimes this fails, in which case it can be
		4097	specified. Also, using a lower number than detected (C<32> should be
		4098	good for about any system in existence) can save some memory, as libev
		4099	statically allocates some 12-24 bytes per signal number.
		4100
		4101	=item EV_PID_HASHSIZE
		4102
		4103	C<ev_child> watchers use a small hash table to distribute workload by
		4104	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
		4105	usually more than enough. If you need to manage thousands of children you
		4106	might want to increase this value (I<must> be a power of two).
		4107
		4108	=item EV_INOTIFY_HASHSIZE
		4109
		4110	C<ev_stat> watchers use a small hash table to distribute workload by
		4111	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
		4112	disabled), usually more than enough. If you need to manage thousands of
		4113	C<ev_stat> watchers you might want to increase this value (I<must> be a
		4114	power of two).
		4115
		4116	=item EV_USE_4HEAP
		4117
		4118	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4119	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4120	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4121	faster performance with many (thousands) of watchers.
		4122
		4123	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4124	will be C<0>.
		4125
		4126	=item EV_HEAP_CACHE_AT
		4127
		4128	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4129	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4130	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4131	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4132	but avoids random read accesses on heap changes. This improves performance
		4133	noticeably with many (hundreds) of watchers.
		4134
		4135	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4136	will be C<0>.
		4137
		4138	=item EV_VERIFY
		4139
		4140	Controls how much internal verification (see C<ev_verify ()>) will
		4141	be done: If set to C<0>, no internal verification code will be compiled
		4142	in. If set to C<1>, then verification code will be compiled in, but not
		4143	called. If set to C<2>, then the internal verification code will be
		4144	called once per loop, which can slow down libev. If set to C<3>, then the
		4145	verification code will be called very frequently, which will slow down
		4146	libev considerably.
		4147
		4148	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4149	will be C<0>.
		4150
		4151	=item EV_COMMON
		4152
		4153	By default, all watchers have a C<void *data> member. By redefining
		4154	this macro to something else you can include more and other types of
		4155	members. You have to define it each time you include one of the files,
		4156	though, and it must be identical each time.
		4157
		4158	For example, the perl EV module uses something like this:
		4159
		4160	#define EV_COMMON \
		4161	SV *self; /* contains this struct */ \
		4162	SV *cb_sv, *fh /* note no trailing ";" */
		4163
		4164	=item EV_CB_DECLARE (type)
		4165
		4166	=item EV_CB_INVOKE (watcher, revents)
		4167
		4168	=item ev_set_cb (ev, cb)
		4169
		4170	Can be used to change the callback member declaration in each watcher,
		4171	and the way callbacks are invoked and set. Must expand to a struct member
		4172	definition and a statement, respectively. See the F<ev.h> header file for
		4173	their default definitions. One possible use for overriding these is to
		4174	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		4175	method calls instead of plain function calls in C++.
		4176
		4177	=back
		4178
		4179	=head2 EXPORTED API SYMBOLS
		4180
		4181	If you need to re-export the API (e.g. via a DLL) and you need a list of
		4182	exported symbols, you can use the provided F<Symbol.*> files which list
		4183	all public symbols, one per line:
		4184
		4185	Symbols.ev for libev proper
		4186	Symbols.event for the libevent emulation
		4187
		4188	This can also be used to rename all public symbols to avoid clashes with
		4189	multiple versions of libev linked together (which is obviously bad in
		4190	itself, but sometimes it is inconvenient to avoid this).
		4191
		4192	A sed command like this will create wrapper C<#define>'s that you need to
		4193	include before including F<ev.h>:
		4194
		4195	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		4196
		4197	This would create a file F<wrap.h> which essentially looks like this:
		4198
		4199	#define ev_backend myprefix_ev_backend
		4200	#define ev_check_start myprefix_ev_check_start
		4201	#define ev_check_stop myprefix_ev_check_stop
		4202	...
		4203
		4204	=head2 EXAMPLES
		4205
		4206	For a real-world example of a program the includes libev
		4207	verbatim, you can have a look at the EV perl module
		4208	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		4209	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		4210	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		4211	will be compiled. It is pretty complex because it provides its own header
		4212	file.
		4213
		4214	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		4215	that everybody includes and which overrides some configure choices:
		4216
		4217	#define EV_FEATURES 8
		4218	#define EV_USE_SELECT 1
		4219	#define EV_PREPARE_ENABLE 1
		4220	#define EV_IDLE_ENABLE 1
		4221	#define EV_SIGNAL_ENABLE 1
		4222	#define EV_CHILD_ENABLE 1
		4223	#define EV_USE_STDEXCEPT 0
		4224	#define EV_CONFIG_H <config.h>
		4225
		4226	#include "ev++.h"
		4227
		4228	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4229
		4230	#include "ev_cpp.h"
		4231	#include "ev.c"
		4232
		4233	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		4234
		4235	=head2 THREADS AND COROUTINES
		4236
		4237	=head3 THREADS
		4238
		4239	All libev functions are reentrant and thread-safe unless explicitly
		4240	documented otherwise, but libev implements no locking itself. This means
		4241	that you can use as many loops as you want in parallel, as long as there
		4242	are no concurrent calls into any libev function with the same loop
		4243	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4244	of course): libev guarantees that different event loops share no data
		4245	structures that need any locking.
		4246
		4247	Or to put it differently: calls with different loop parameters can be done
		4248	concurrently from multiple threads, calls with the same loop parameter
		4249	must be done serially (but can be done from different threads, as long as
		4250	only one thread ever is inside a call at any point in time, e.g. by using
		4251	a mutex per loop).
		4252
		4253	Specifically to support threads (and signal handlers), libev implements
		4254	so-called C<ev_async> watchers, which allow some limited form of
		4255	concurrency on the same event loop, namely waking it up "from the
		4256	outside".
		4257
		4258	If you want to know which design (one loop, locking, or multiple loops
		4259	without or something else still) is best for your problem, then I cannot
		4260	help you, but here is some generic advice:
		4261
		4262	=over 4
		4263
		4264	=item * most applications have a main thread: use the default libev loop
		4265	in that thread, or create a separate thread running only the default loop.
		4266
		4267	This helps integrating other libraries or software modules that use libev
		4268	themselves and don't care/know about threading.
		4269
		4270	=item * one loop per thread is usually a good model.
		4271
		4272	Doing this is almost never wrong, sometimes a better-performance model
		4273	exists, but it is always a good start.
		4274
		4275	=item * other models exist, such as the leader/follower pattern, where one
		4276	loop is handed through multiple threads in a kind of round-robin fashion.
		4277
		4278	Choosing a model is hard - look around, learn, know that usually you can do
		4279	better than you currently do :-)
		4280
		4281	=item * often you need to talk to some other thread which blocks in the
		4282	event loop.
		4283
		4284	C<ev_async> watchers can be used to wake them up from other threads safely
		4285	(or from signal contexts...).
		4286
		4287	An example use would be to communicate signals or other events that only
		4288	work in the default loop by registering the signal watcher with the
		4289	default loop and triggering an C<ev_async> watcher from the default loop
		4290	watcher callback into the event loop interested in the signal.
		4291
		4292	=back
		4293
		4294	=head4 THREAD LOCKING EXAMPLE
		4295
		4296	Here is a fictitious example of how to run an event loop in a different
		4297	thread than where callbacks are being invoked and watchers are
		4298	created/added/removed.
		4299
		4300	For a real-world example, see the C<EV::Loop::Async> perl module,
		4301	which uses exactly this technique (which is suited for many high-level
		4302	languages).
		4303
		4304	The example uses a pthread mutex to protect the loop data, a condition
		4305	variable to wait for callback invocations, an async watcher to notify the
		4306	event loop thread and an unspecified mechanism to wake up the main thread.
		4307
		4308	First, you need to associate some data with the event loop:
		4309
		4310	typedef struct {
		4311	mutex_t lock; /* global loop lock */
		4312	ev_async async_w;
		4313	thread_t tid;
		4314	cond_t invoke_cv;
		4315	} userdata;
		4316
		4317	void prepare_loop (EV_P)
		4318	{
		4319	// for simplicity, we use a static userdata struct.
		4320	static userdata u;
		4321
		4322	ev_async_init (&u->async_w, async_cb);
		4323	ev_async_start (EV_A_ &u->async_w);
		4324
		4325	pthread_mutex_init (&u->lock, 0);
		4326	pthread_cond_init (&u->invoke_cv, 0);
		4327
		4328	// now associate this with the loop
		4329	ev_set_userdata (EV_A_ u);
		4330	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		4331	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		4332
		4333	// then create the thread running ev_loop
		4334	pthread_create (&u->tid, 0, l_run, EV_A);
		4335	}
		4336
		4337	The callback for the C<ev_async> watcher does nothing: the watcher is used
		4338	solely to wake up the event loop so it takes notice of any new watchers
		4339	that might have been added:
		4340
		4341	static void
		4342	async_cb (EV_P_ ev_async *w, int revents)
		4343	{
		4344	// just used for the side effects
		4345	}
		4346
		4347	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		4348	protecting the loop data, respectively.
		4349
		4350	static void
		4351	l_release (EV_P)
		4352	{
		4353	userdata *u = ev_userdata (EV_A);
		4354	pthread_mutex_unlock (&u->lock);
		4355	}
		4356
		4357	static void
		4358	l_acquire (EV_P)
		4359	{
		4360	userdata *u = ev_userdata (EV_A);
		4361	pthread_mutex_lock (&u->lock);
		4362	}
		4363
		4364	The event loop thread first acquires the mutex, and then jumps straight
		4365	into C<ev_run>:
		4366
		4367	void *
		4368	l_run (void *thr_arg)
		4369	{
		4370	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		4371
		4372	l_acquire (EV_A);
		4373	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		4374	ev_run (EV_A_ 0);
		4375	l_release (EV_A);
		4376
		4377	return 0;
		4378	}
		4379
		4380	Instead of invoking all pending watchers, the C<l_invoke> callback will
		4381	signal the main thread via some unspecified mechanism (signals? pipe
		4382	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		4383	have been called (in a while loop because a) spurious wakeups are possible
		4384	and b) skipping inter-thread-communication when there are no pending
		4385	watchers is very beneficial):
		4386
		4387	static void
		4388	l_invoke (EV_P)
		4389	{
		4390	userdata *u = ev_userdata (EV_A);
		4391
		4392	while (ev_pending_count (EV_A))
		4393	{
		4394	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		4395	pthread_cond_wait (&u->invoke_cv, &u->lock);
		4396	}
		4397	}
		4398
		4399	Now, whenever the main thread gets told to invoke pending watchers, it
		4400	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		4401	thread to continue:
		4402
		4403	static void
		4404	real_invoke_pending (EV_P)
		4405	{
		4406	userdata *u = ev_userdata (EV_A);
		4407
		4408	pthread_mutex_lock (&u->lock);
		4409	ev_invoke_pending (EV_A);
		4410	pthread_cond_signal (&u->invoke_cv);
		4411	pthread_mutex_unlock (&u->lock);
		4412	}
		4413
		4414	Whenever you want to start/stop a watcher or do other modifications to an
		4415	event loop, you will now have to lock:
		4416
		4417	ev_timer timeout_watcher;
		4418	userdata *u = ev_userdata (EV_A);
		4419
		4420	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		4421
		4422	pthread_mutex_lock (&u->lock);
		4423	ev_timer_start (EV_A_ &timeout_watcher);
		4424	ev_async_send (EV_A_ &u->async_w);
		4425	pthread_mutex_unlock (&u->lock);
		4426
		4427	Note that sending the C<ev_async> watcher is required because otherwise
		4428	an event loop currently blocking in the kernel will have no knowledge
		4429	about the newly added timer. By waking up the loop it will pick up any new
		4430	watchers in the next event loop iteration.
		4431
		4432	=head3 COROUTINES
		4433
		4434	Libev is very accommodating to coroutines ("cooperative threads"):
		4435	libev fully supports nesting calls to its functions from different
		4436	coroutines (e.g. you can call C<ev_run> on the same loop from two
		4437	different coroutines, and switch freely between both coroutines running
		4438	the loop, as long as you don't confuse yourself). The only exception is
		4439	that you must not do this from C<ev_periodic> reschedule callbacks.
		4440
		4441	Care has been taken to ensure that libev does not keep local state inside
		4442	C<ev_run>, and other calls do not usually allow for coroutine switches as
		4443	they do not call any callbacks.
		4444
		4445	=head2 COMPILER WARNINGS
		4446
		4447	Depending on your compiler and compiler settings, you might get no or a
		4448	lot of warnings when compiling libev code. Some people are apparently
		4449	scared by this.
		4450
		4451	However, these are unavoidable for many reasons. For one, each compiler
		4452	has different warnings, and each user has different tastes regarding
		4453	warning options. "Warn-free" code therefore cannot be a goal except when
		4454	targeting a specific compiler and compiler-version.
		4455
		4456	Another reason is that some compiler warnings require elaborate
		4457	workarounds, or other changes to the code that make it less clear and less
		4458	maintainable.
		4459
		4460	And of course, some compiler warnings are just plain stupid, or simply
		4461	wrong (because they don't actually warn about the condition their message
		4462	seems to warn about). For example, certain older gcc versions had some
		4463	warnings that resulted in an extreme number of false positives. These have
		4464	been fixed, but some people still insist on making code warn-free with
		4465	such buggy versions.
		4466
		4467	While libev is written to generate as few warnings as possible,
		4468	"warn-free" code is not a goal, and it is recommended not to build libev
		4469	with any compiler warnings enabled unless you are prepared to cope with
		4470	them (e.g. by ignoring them). Remember that warnings are just that:
		4471	warnings, not errors, or proof of bugs.
		4472
		4473
		4474	=head2 VALGRIND
		4475
		4476	Valgrind has a special section here because it is a popular tool that is
		4477	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4478
		4479	If you think you found a bug (memory leak, uninitialised data access etc.)
		4480	in libev, then check twice: If valgrind reports something like:
		4481
		4482	==2274== definitely lost: 0 bytes in 0 blocks.
		4483	==2274== possibly lost: 0 bytes in 0 blocks.
		4484	==2274== still reachable: 256 bytes in 1 blocks.
		4485
		4486	Then there is no memory leak, just as memory accounted to global variables
		4487	is not a memleak - the memory is still being referenced, and didn't leak.
		4488
		4489	Similarly, under some circumstances, valgrind might report kernel bugs
		4490	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4491	although an acceptable workaround has been found here), or it might be
		4492	confused.
		4493
		4494	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4495	make it into some kind of religion.
		4496
		4497	If you are unsure about something, feel free to contact the mailing list
		4498	with the full valgrind report and an explanation on why you think this
		4499	is a bug in libev (best check the archives, too :). However, don't be
		4500	annoyed when you get a brisk "this is no bug" answer and take the chance
		4501	of learning how to interpret valgrind properly.
		4502
		4503	If you need, for some reason, empty reports from valgrind for your project
		4504	I suggest using suppression lists.
		4505
		4506
		4507	=head1 PORTABILITY NOTES
		4508
		4509	=head2 GNU/LINUX 32 BIT LIMITATIONS
		4510
		4511	GNU/Linux is the only common platform that supports 64 bit file/large file
		4512	interfaces but I<disables> them by default.
		4513
		4514	That means that libev compiled in the default environment doesn't support
		4515	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		4516
		4517	Unfortunately, many programs try to work around this GNU/Linux issue
		4518	by enabling the large file API, which makes them incompatible with the
		4519	standard libev compiled for their system.
		4520
		4521	Likewise, libev cannot enable the large file API itself as this would
		4522	suddenly make it incompatible to the default compile time environment,
		4523	i.e. all programs not using special compile switches.
		4524
		4525	=head2 OS/X AND DARWIN BUGS
		4526
		4527	The whole thing is a bug if you ask me - basically any system interface
		4528	you touch is broken, whether it is locales, poll, kqueue or even the
		4529	OpenGL drivers.
		4530
		4531	=head3 C<kqueue> is buggy
		4532
		4533	The kqueue syscall is broken in all known versions - most versions support
		4534	only sockets, many support pipes.
		4535
		4536	Libev tries to work around this by not using C<kqueue> by default on this
		4537	rotten platform, but of course you can still ask for it when creating a
		4538	loop - embedding a socket-only kqueue loop into a select-based one is
		4539	probably going to work well.
		4540
		4541	=head3 C<poll> is buggy
		4542
		4543	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		4544	implementation by something calling C<kqueue> internally around the 10.5.6
		4545	release, so now C<kqueue> I<and> C<poll> are broken.
		4546
		4547	Libev tries to work around this by not using C<poll> by default on
		4548	this rotten platform, but of course you can still ask for it when creating
		4549	a loop.
		4550
		4551	=head3 C<select> is buggy
		4552
		4553	All that's left is C<select>, and of course Apple found a way to fuck this
		4554	one up as well: On OS/X, C<select> actively limits the number of file
		4555	descriptors you can pass in to 1024 - your program suddenly crashes when
		4556	you use more.
		4557
		4558	There is an undocumented "workaround" for this - defining
		4559	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		4560	work on OS/X.
		4561
		4562	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		4563
		4564	=head3 C<errno> reentrancy
		4565
		4566	The default compile environment on Solaris is unfortunately so
		4567	thread-unsafe that you can't even use components/libraries compiled
		4568	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
		4569	defined by default. A valid, if stupid, implementation choice.
		4570
		4571	If you want to use libev in threaded environments you have to make sure
		4572	it's compiled with C<_REENTRANT> defined.
		4573
		4574	=head3 Event port backend
		4575
		4576	The scalable event interface for Solaris is called "event
		4577	ports". Unfortunately, this mechanism is very buggy in all major
		4578	releases. If you run into high CPU usage, your program freezes or you get
		4579	a large number of spurious wakeups, make sure you have all the relevant
		4580	and latest kernel patches applied. No, I don't know which ones, but there
		4581	are multiple ones to apply, and afterwards, event ports actually work
		4582	great.
		4583
		4584	If you can't get it to work, you can try running the program by setting
		4585	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		4586	C<select> backends.
		4587
		4588	=head2 AIX POLL BUG
		4589
		4590	AIX unfortunately has a broken C<poll.h> header. Libev works around
		4591	this by trying to avoid the poll backend altogether (i.e. it's not even
		4592	compiled in), which normally isn't a big problem as C<select> works fine
		4593	with large bitsets on AIX, and AIX is dead anyway.
		4594
		4595	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4596
		4597	=head3 General issues
		4598
		4599	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		4600	requires, and its I/O model is fundamentally incompatible with the POSIX
		4601	model. Libev still offers limited functionality on this platform in
		4602	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		4603	descriptors. This only applies when using Win32 natively, not when using
		4604	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		4605	as every compielr comes with a slightly differently broken/incompatible
		4606	environment.
		4607
		4608	Lifting these limitations would basically require the full
		4609	re-implementation of the I/O system. If you are into this kind of thing,
		4610	then note that glib does exactly that for you in a very portable way (note
		4611	also that glib is the slowest event library known to man).
		4612
		4613	There is no supported compilation method available on windows except
		4614	embedding it into other applications.
		4615
		4616	Sensible signal handling is officially unsupported by Microsoft - libev
		4617	tries its best, but under most conditions, signals will simply not work.
		4618
		4619	Not a libev limitation but worth mentioning: windows apparently doesn't
		4620	accept large writes: instead of resulting in a partial write, windows will
		4621	either accept everything or return C<ENOBUFS> if the buffer is too large,
		4622	so make sure you only write small amounts into your sockets (less than a
		4623	megabyte seems safe, but this apparently depends on the amount of memory
		4624	available).
		4625
		4626	Due to the many, low, and arbitrary limits on the win32 platform and
		4627	the abysmal performance of winsockets, using a large number of sockets
		4628	is not recommended (and not reasonable). If your program needs to use
		4629	more than a hundred or so sockets, then likely it needs to use a totally
		4630	different implementation for windows, as libev offers the POSIX readiness
		4631	notification model, which cannot be implemented efficiently on windows
		4632	(due to Microsoft monopoly games).
		4633
		4634	A typical way to use libev under windows is to embed it (see the embedding
		4635	section for details) and use the following F<evwrap.h> header file instead
		4636	of F<ev.h>:
		4637
		4638	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4639	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4640
		4641	#include "ev.h"
		4642
		4643	And compile the following F<evwrap.c> file into your project (make sure
		4644	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4645
		4646	#include "evwrap.h"
		4647	#include "ev.c"
		4648
		4649	=head3 The winsocket C<select> function
		4650
		4651	The winsocket C<select> function doesn't follow POSIX in that it
		4652	requires socket I<handles> and not socket I<file descriptors> (it is
		4653	also extremely buggy). This makes select very inefficient, and also
		4654	requires a mapping from file descriptors to socket handles (the Microsoft
		4655	C runtime provides the function C<_open_osfhandle> for this). See the
		4656	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		4657	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		4658
		4659	The configuration for a "naked" win32 using the Microsoft runtime
		4660	libraries and raw winsocket select is:
		4661
		4662	#define EV_USE_SELECT 1
		4663	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		4664
		4665	Note that winsockets handling of fd sets is O(n), so you can easily get a
		4666	complexity in the O(n²) range when using win32.
		4667
		4668	=head3 Limited number of file descriptors
		4669
		4670	Windows has numerous arbitrary (and low) limits on things.
		4671
		4672	Early versions of winsocket's select only supported waiting for a maximum
		4673	of C<64> handles (probably owning to the fact that all windows kernels
		4674	can only wait for C<64> things at the same time internally; Microsoft
		4675	recommends spawning a chain of threads and wait for 63 handles and the
		4676	previous thread in each. Sounds great!).
		4677
		4678	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		4679	to some high number (e.g. C<2048>) before compiling the winsocket select
		4680	call (which might be in libev or elsewhere, for example, perl and many
		4681	other interpreters do their own select emulation on windows).
		4682
		4683	Another limit is the number of file descriptors in the Microsoft runtime
		4684	libraries, which by default is C<64> (there must be a hidden I<64>
		4685	fetish or something like this inside Microsoft). You can increase this
		4686	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		4687	(another arbitrary limit), but is broken in many versions of the Microsoft
		4688	runtime libraries. This might get you to about C<512> or C<2048> sockets
		4689	(depending on windows version and/or the phase of the moon). To get more,
		4690	you need to wrap all I/O functions and provide your own fd management, but
		4691	the cost of calling select (O(n²)) will likely make this unworkable.
		4692
		4693	=head2 PORTABILITY REQUIREMENTS
		4694
		4695	In addition to a working ISO-C implementation and of course the
		4696	backend-specific APIs, libev relies on a few additional extensions:
		4697
		4698	=over 4
		4699
		4700	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4701	calling conventions regardless of C<ev_watcher_type *>.
		4702
		4703	Libev assumes not only that all watcher pointers have the same internal
		4704	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4705	assumes that the same (machine) code can be used to call any watcher
		4706	callback: The watcher callbacks have different type signatures, but libev
		4707	calls them using an C<ev_watcher *> internally.
		4708
		4709	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4710
		4711	The type C<sig_atomic_t volatile> (or whatever is defined as
		4712	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4713	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4714	believed to be sufficiently portable.
		4715
		4716	=item C<sigprocmask> must work in a threaded environment
		4717
		4718	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4719	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		4720	pthread implementations will either allow C<sigprocmask> in the "main
		4721	thread" or will block signals process-wide, both behaviours would
		4722	be compatible with libev. Interaction between C<sigprocmask> and
		4723	C<pthread_sigmask> could complicate things, however.
		4724
		4725	The most portable way to handle signals is to block signals in all threads
		4726	except the initial one, and run the default loop in the initial thread as
		4727	well.
		4728
		4729	=item C<long> must be large enough for common memory allocation sizes
		4730
		4731	To improve portability and simplify its API, libev uses C<long> internally
		4732	instead of C<size_t> when allocating its data structures. On non-POSIX
		4733	systems (Microsoft...) this might be unexpectedly low, but is still at
		4734	least 31 bits everywhere, which is enough for hundreds of millions of
		4735	watchers.
		4736
		4737	=item C<double> must hold a time value in seconds with enough accuracy
		4738
		4739	The type C<double> is used to represent timestamps. It is required to
		4740	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		4741	good enough for at least into the year 4000 with millisecond accuracy
		4742	(the design goal for libev). This requirement is overfulfilled by
		4743	implementations using IEEE 754, which is basically all existing ones. With
		4744	IEEE 754 doubles, you get microsecond accuracy until at least 2200.
		4745
		4746	=back
		4747
		4748	If you know of other additional requirements drop me a note.
		4749
		4750
		4751	=head1 ALGORITHMIC COMPLEXITIES
		4752
		4753	In this section the complexities of (many of) the algorithms used inside
		4754	libev will be documented. For complexity discussions about backends see
		4755	the documentation for C<ev_default_init>.
		4756
		4757	All of the following are about amortised time: If an array needs to be
		4758	extended, libev needs to realloc and move the whole array, but this
		4759	happens asymptotically rarer with higher number of elements, so O(1) might
		4760	mean that libev does a lengthy realloc operation in rare cases, but on
		4761	average it is much faster and asymptotically approaches constant time.
		4762
		4763	=over 4
		4764
		4765	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		4766
		4767	This means that, when you have a watcher that triggers in one hour and
		4768	there are 100 watchers that would trigger before that, then inserting will
		4769	have to skip roughly seven (C<ld 100>) of these watchers.
		4770
		4771	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		4772
		4773	That means that changing a timer costs less than removing/adding them,
		4774	as only the relative motion in the event queue has to be paid for.
		4775
		4776	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		4777
		4778	These just add the watcher into an array or at the head of a list.
		4779
		4780	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		4781
		4782	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		4783
		4784	These watchers are stored in lists, so they need to be walked to find the
		4785	correct watcher to remove. The lists are usually short (you don't usually
		4786	have many watchers waiting for the same fd or signal: one is typical, two
		4787	is rare).
		4788
		4789	=item Finding the next timer in each loop iteration: O(1)
		4790
		4791	By virtue of using a binary or 4-heap, the next timer is always found at a
		4792	fixed position in the storage array.
		4793
		4794	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		4795
		4796	A change means an I/O watcher gets started or stopped, which requires
		4797	libev to recalculate its status (and possibly tell the kernel, depending
		4798	on backend and whether C<ev_io_set> was used).
		4799
		4800	=item Activating one watcher (putting it into the pending state): O(1)
		4801
		4802	=item Priority handling: O(number_of_priorities)
		4803
		4804	Priorities are implemented by allocating some space for each
		4805	priority. When doing priority-based operations, libev usually has to
		4806	linearly search all the priorities, but starting/stopping and activating
		4807	watchers becomes O(1) with respect to priority handling.
		4808
		4809	=item Sending an ev_async: O(1)
		4810
		4811	=item Processing ev_async_send: O(number_of_async_watchers)
		4812
		4813	=item Processing signals: O(max_signal_number)
		4814
		4815	Sending involves a system call I<iff> there were no other C<ev_async_send>
		4816	calls in the current loop iteration. Checking for async and signal events
		4817	involves iterating over all running async watchers or all signal numbers.
		4818
		4819	=back
		4820
		4821
		4822	=head1 PORTING FROM LIBEV 3.X TO 4.X
		4823
		4824	The major version 4 introduced some minor incompatible changes to the API.
		4825
		4826	At the moment, the C<ev.h> header file tries to implement superficial
		4827	compatibility, so most programs should still compile. Those might be
		4828	removed in later versions of libev, so better update early than late.
		4829
		4830	=over 4
		4831
		4832	=item function/symbol renames
		4833
		4834	A number of functions and symbols have been renamed:
		4835
		4836	ev_loop => ev_run
		4837	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		4838	EVLOOP_ONESHOT => EVRUN_ONCE
		4839
		4840	ev_unloop => ev_break
		4841	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		4842	EVUNLOOP_ONE => EVBREAK_ONE
		4843	EVUNLOOP_ALL => EVBREAK_ALL
		4844
		4845	EV_TIMEOUT => EV_TIMER
		4846
		4847	ev_loop_count => ev_iteration
		4848	ev_loop_depth => ev_depth
		4849	ev_loop_verify => ev_verify
		4850
		4851	Most functions working on C<struct ev_loop> objects don't have an
		4852	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		4853	associated constants have been renamed to not collide with the C<struct
		4854	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		4855	as all other watcher types. Note that C<ev_loop_fork> is still called
		4856	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		4857	typedef.
		4858
		4859	=item C<EV_COMPAT3> backwards compatibility mechanism
		4860
		4861	The backward compatibility mechanism can be controlled by
		4862	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
		4863	section.
		4864
		4865	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		4866
		4867	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		4868	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		4869	and work, but the library code will of course be larger.
		4870
		4871	=back
		4872
		4873
		4874	=head1 GLOSSARY
		4875
		4876	=over 4
		4877
		4878	=item active
		4879
		4880	A watcher is active as long as it has been started and not yet stopped.
		4881	See L<WATCHER STATES> for details.
		4882
		4883	=item application
		4884
		4885	In this document, an application is whatever is using libev.
		4886
		4887	=item callback
		4888
		4889	The address of a function that is called when some event has been
		4890	detected. Callbacks are being passed the event loop, the watcher that
		4891	received the event, and the actual event bitset.
		4892
		4893	=item callback/watcher invocation
		4894
		4895	The act of calling the callback associated with a watcher.
		4896
		4897	=item event
		4898
		4899	A change of state of some external event, such as data now being available
		4900	for reading on a file descriptor, time having passed or simply not having
		4901	any other events happening anymore.
		4902
		4903	In libev, events are represented as single bits (such as C<EV_READ> or
		4904	C<EV_TIMER>).
		4905
		4906	=item event library
		4907
		4908	A software package implementing an event model and loop.
		4909
		4910	=item event loop
		4911
		4912	An entity that handles and processes external events and converts them
		4913	into callback invocations.
		4914
		4915	=item event model
		4916
		4917	The model used to describe how an event loop handles and processes
		4918	watchers and events.
		4919
		4920	=item pending
		4921
		4922	A watcher is pending as soon as the corresponding event has been
		4923	detected. See L<WATCHER STATES> for details.
		4924
		4925	=item real time
		4926
		4927	The physical time that is observed. It is apparently strictly monotonic :)
		4928
		4929	=item wall-clock time
		4930
		4931	The time and date as shown on clocks. Unlike real time, it can actually
		4932	be wrong and jump forwards and backwards, e.g. when the you adjust your
		4933	clock.
		4934
		4935	=item watcher
		4936
		4937	A data structure that describes interest in certain events. Watchers need
		4938	to be started (attached to an event loop) before they can receive events.
		4939
		4940	=back
		4941
742	=head1 AUTHOR	4942	=head1 AUTHOR
743		4943
744	Marc Lehmann <libev@schmorp.de>.	4944	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
745		4945

Diff Legend

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