ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.27 by root, Wed Nov 14 05:02:07 2007 UTC vs.
Revision 1.311 by root, Thu Oct 21 14:40:07 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 REPRESENTATION	126	=head2 TIME REPRESENTATION
45		127
46	Libev represents time as a single floating point number, representing the	128	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	129	the (fractional) number of seconds since the (POSIX) epoch (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
51		158
52	=head1 GLOBAL FUNCTIONS	159	=head1 GLOBAL FUNCTIONS
53		160
54	These functions can be called anytime, even before initialising the	161	These functions can be called anytime, even before initialising the
55	library in any way.	162	library in any way.
…		…
60		167
61	Returns the current time as libev would use it. Please note that the	168	Returns the current time as libev would use it. Please note that the
62	C<ev_now> function is usually faster and also often returns the timestamp	169	C<ev_now> function is usually faster and also often returns the timestamp
63	you actually want to know.	170	you actually want to know.
64		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 ()>.
		177
65	=item int ev_version_major ()	178	=item int ev_version_major ()
66		179
67	=item int ev_version_minor ()	180	=item int ev_version_minor ()
68		181
69	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
70	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
71	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
72	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
73	version of the library your program was compiled against.	186	version of the library your program was compiled against.
74		187
		188	These version numbers refer to the ABI version of the library, not the
		189	release version.
		190
75	Usually, it's a good idea to terminate if the major versions mismatch,	191	Usually, it's a good idea to terminate if the major versions mismatch,
76	as this indicates an incompatible change. Minor versions are usually	192	as this indicates an incompatible change. Minor versions are usually
77	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
78	not a problem.	194	not a problem.
79		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
80	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	235	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
81		236
82	Sets the allocation function to use (the prototype is similar to the	237	Sets the allocation function to use (the prototype is similar - the
83	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
84	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
85	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
86	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.
87		246
88	You could override this function in high-availability programs to, say,	247	You could override this function in high-availability programs to, say,
89	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,
90	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.
91		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
92	=item ev_set_syserr_cb (void (*cb)(const char *msg));	271	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
93		272
94	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
95	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
96	indicating the system call or subsystem causing the problem. If this	275	indicating the system call or subsystem causing the problem. If this
97	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
98	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
99	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
100	(such as abort).	279	(such as abort).
101		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
102	=back	293	=back
103		294
104	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	295	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
105		296
106	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
107	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
108	events, and dynamically created loops which do not.	299	libev 3 had an C<ev_loop> function colliding with the struct name).
109		300
110	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
111	in your main thread (or in a separate thread) and for each thread you	302	supports signals and child events, and dynamically created event loops
112	create, you also create another event loop. Libev itself does no locking	303	which do not.
113	whatsoever, so if you mix calls to the same event loop in different
114	threads, make sure you lock (this is usually a bad idea, though, even if
115	done correctly, because it's hideous and inefficient).
116		304
117	=over 4	305	=over 4
118		306
119	=item struct ev_loop *ev_default_loop (unsigned int flags)	307	=item struct ev_loop *ev_default_loop (unsigned int flags)
120		308
121	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
122	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
123	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
124	flags).	312	flags. If that is troubling you, check C<ev_backend ()> afterwards).
125		313
126	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
127	function.	315	function.
128		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
129	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
130	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>).
131		330
132	It supports the following flags:	331	The following flags are supported:
133		332
134	=over 4	333	=over 4
135		334
136	=item C<EVFLAG_AUTO>	335	=item C<EVFLAG_AUTO>
137		336
138	The default flags value. Use this if you have no clue (it's the right	337	The default flags value. Use this if you have no clue (it's the right
139	thing, believe me).	338	thing, believe me).
140		339
141	=item C<EVFLAG_NOENV>	340	=item C<EVFLAG_NOENV>
142		341
143	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
144	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
145	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	344	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
146	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
147	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
148	around bugs.	347	around bugs.
149		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
150	=item C<EVMETHOD_SELECT> (portable select backend)	388	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
151		389
		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.
		395
		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.
		402
		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).
		406
152	=item C<EVMETHOD_POLL> (poll backend, available everywhere except on windows)	407	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
153		408
154	=item C<EVMETHOD_EPOLL> (linux 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.
155		415
156	=item C<EVMETHOD_KQUEUE> (some bsds only)	416	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		417	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
157		418
158	=item C<EVMETHOD_DEVPOLL> (solaris 8 only)	419	=item C<EVBACKEND_EPOLL> (value 4, Linux)
159		420
160	=item C<EVMETHOD_PORT> (solaris 10 only)	421	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
		422	kernels).
161		423
162	If one or more of these are ored into the flags value, then only these	424	For few fds, this backend is a bit little slower than poll and select,
163	backends will be tried (in the reverse order as given here). If one are	425	but it scales phenomenally better. While poll and select usually scale
164	specified, any backend will do.	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.
165		546
166	=back	547	=back
167		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
168	=item struct ev_loop *ev_loop_new (unsigned int flags)	571	=item struct ev_loop *ev_loop_new (unsigned int flags)
169		572
170	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
171	always distinct from the default loop. Unlike the default loop, it cannot	574	always distinct from the default loop.
172	handle signal and child watchers, and attempts to do so will be greeted by	575
173	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");
174		585
175	=item ev_default_destroy ()	586	=item ev_default_destroy ()
176		587
177	Destroys the default loop again (frees all memory and kernel state	588	Destroys the default loop (frees all memory and kernel state etc.). None
178	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
179	any way whatsoever, although you cannot 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>.
180		603
181	=item ev_loop_destroy (loop)	604	=item ev_loop_destroy (loop)
182		605
183	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
184	earlier call to C<ev_loop_new>.	607	earlier call to C<ev_loop_new>.
185		608
186	=item ev_default_fork ()	609	=item ev_default_fork ()
187		610
		611	This function sets a flag that causes subsequent C<ev_run> iterations
188	This function reinitialises the kernel state for backends that have	612	to reinitialise the kernel state for backends that have one. Despite the
189	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
190	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
191	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.
192		617
193	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
194	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
195	have to call it.	620	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
		621	during fork.
		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).
196		629
197	The function itself is quite fast and it's usually not a problem to call	630	The function itself is quite fast and it's usually not a problem to call
198	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
199	quite nicely into a call to C<pthread_atfork>:	632	quite nicely into a call to C<pthread_atfork>:
200		633
…		…
202		635
203	=item ev_loop_fork (loop)	636	=item ev_loop_fork (loop)
204		637
205	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
206	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
207	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.
208		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
209	=item unsigned int ev_method (loop)	659	=item unsigned int ev_depth (loop)
210		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
211	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
212	use.	675	use.
213		676
214	=item ev_tstamp ev_now (loop)	677	=item ev_tstamp ev_now (loop)
215		678
216	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
217	got events and started processing them. This timestamp does not change	680	received events and started processing them. This timestamp does not
218	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
219	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
220	occuring (or more correctly, the mainloop finding out about it).	683	event occurring (or more correctly, libev finding out about it).
221		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
222	=item ev_loop (loop, int flags)	723	=item ev_run (loop, int flags)
223		724
224	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
225	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
226	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>.
227		730
228	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
229	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.
230		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
231	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
232	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
233	case there are no events and will return after one iteration of the loop.	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.
234		747
235	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
236	neccessary) and will handle those and any outstanding ones. It will block	749	necessary) and will handle those and any already outstanding ones. It
237	your process until at least one new event arrives, and will return after	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
238	one iteration of the loop.	753	iteration of the loop.
239		754
240	This flags value could be used to implement alternative looping	755	This is useful if you are waiting for some external event in conjunction
241	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
242	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.
243		759
244	Here are the gory details of what ev_loop does:	760	Here are the gory details of what C<ev_run> does:
245		761
246	1. If there are no active watchers (reference count is zero), return.	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.
247	2. Queue and immediately call all prepare watchers.	768	- Queue and call all prepare watchers.
		769	- If ev_break was called, goto FINISH.
248	3. If we have been forked, recreate the kernel state.	770	- If we have been forked, detach and recreate the kernel state
		771	as to not disturb the other process.
249	4. Update the kernel state with all outstanding changes.	772	- Update the kernel state with all outstanding changes.
250	5. Update the "event loop time".	773	- Update the "event loop time" (ev_now ()).
251	6. Calculate for how long to block.	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.
252	7. Block the process, waiting for events.	779	- Block the process, waiting for any events.
		780	- Queue all outstanding I/O (fd) events.
253	8. Update the "event loop time" and do time jump handling.	781	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
254	9. Queue all outstanding timers.	782	- Queue all expired timers.
255	10. Queue all outstanding periodics.	783	- Queue all expired periodics.
256	11. If no events are pending now, queue all idle watchers.	784	- Queue all idle watchers with priority higher than that of pending events.
257	12. Queue all check watchers.	785	- Queue all check watchers.
258	13. Call all queued watchers in reverse order (i.e. check watchers first).	786	- Call all queued watchers in reverse order (i.e. check watchers first).
259	14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	787	Signals and child watchers are implemented as I/O watchers, and will
260	was used, return, otherwise continue with step #1.	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.
261		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
262	=item ev_unloop (loop, how)	805	=item ev_break (loop, how)
263		806
264	Can be used to make a call to C<ev_loop> return early (but only after it	807	Can be used to make a call to C<ev_run> return early (but only after it
265	has processed all outstanding events). The C<how> argument must be either	808	has processed all outstanding events). The C<how> argument must be either
266	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	809	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
267	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> 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##
268		815
269	=item ev_ref (loop)	816	=item ev_ref (loop)
270		817
271	=item ev_unref (loop)	818	=item ev_unref (loop)
272		819
273	Ref/unref can be used to add or remove a reference count on the event	820	Ref/unref can be used to add or remove a reference count on the event
274	loop: Every watcher keeps one reference, and as long as the reference	821	loop: Every watcher keeps one reference, and as long as the reference
275	count is nonzero, C<ev_loop> will not return on its own. If you have	822	count is nonzero, C<ev_run> will not return on its own.
276	a watcher you never unregister that should not keep C<ev_loop> from	823
277	returning, ev_unref() after starting, and ev_ref() before stopping it. For	824	This is useful when you have a watcher that you never intend to
		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
278	example, libev itself uses this for its internal signal pipe: It is not	829	As an example, libev itself uses this for its internal signal pipe: It
279	visible to the libev user and should not keep C<ev_loop> from exiting if	830	is not visible to the libev user and should not keep C<ev_run> from
280	no event watchers registered by it are active. It is also an excellent	831	exiting if no event watchers registered by it are active. It is also an
281	way to do this for generic recurring timers or from within third-party	832	excellent way to do this for generic recurring timers or from within
282	libraries. Just remember to I<unref after start> and I<ref before stop>.	833	third-party libraries. Just remember to I<unref after start> and I<ref
		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.
		914
		915	=item int ev_pending_count (loop)
		916
		917	Returns the number of pending watchers - zero indicates that no watchers
		918	are pending.
		919
		920	=item ev_set_invoke_pending_cb (loop, void (*invoke_pending_cb)(EV_P))
		921
		922	This overrides the invoke pending functionality of the loop: Instead of
		923	invoking all pending watchers when there are any, C<ev_run> will call
		924	this callback instead. This is useful, for example, when you want to
		925	invoke the actual watchers inside another context (another thread etc.).
		926
		927	If you want to reset the callback, use C<ev_invoke_pending> as new
		928	callback.
		929
		930	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))
		931
		932	Sometimes you want to share the same loop between multiple threads. This
		933	can be done relatively simply by putting mutex_lock/unlock calls around
		934	each call to a libev function.
		935
		936	However, C<ev_run> can run an indefinite time, so it is not feasible
		937	to wait for it to return. One way around this is to wake up the event
		938	loop via C<ev_break> and C<av_async_send>, another way is to set these
		939	I<release> and I<acquire> callbacks on the loop.
		940
		941	When set, then C<release> will be called just before the thread is
		942	suspended waiting for new events, and C<acquire> is called just
		943	afterwards.
		944
		945	Ideally, C<release> will just call your mutex_unlock function, and
		946	C<acquire> will just call the mutex_lock function again.
		947
		948	While event loop modifications are allowed between invocations of
		949	C<release> and C<acquire> (that's their only purpose after all), no
		950	modifications done will affect the event loop, i.e. adding watchers will
		951	have no effect on the set of file descriptors being watched, or the time
		952	waited. Use an C<ev_async> watcher to wake up C<ev_run> when you want it
		953	to take note of any changes you made.
		954
		955	In theory, threads executing C<ev_run> will be async-cancel safe between
		956	invocations of C<release> and C<acquire>.
		957
		958	See also the locking example in the C<THREADS> section later in this
		959	document.
		960
		961	=item ev_set_userdata (loop, void *data)
		962
		963	=item ev_userdata (loop)
		964
		965	Set and retrieve a single C<void *> associated with a loop. When
		966	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
		967	C<0.>
		968
		969	These two functions can be used to associate arbitrary data with a loop,
		970	and are intended solely for the C<invoke_pending_cb>, C<release> and
		971	C<acquire> callbacks described above, but of course can be (ab-)used for
		972	any other purpose as well.
		973
		974	=item ev_verify (loop)
		975
		976	This function only does something when C<EV_VERIFY> support has been
		977	compiled in, which is the default for non-minimal builds. It tries to go
		978	through all internal structures and checks them for validity. If anything
		979	is found to be inconsistent, it will print an error message to standard
		980	error and call C<abort ()>.
		981
		982	This can be used to catch bugs inside libev itself: under normal
		983	circumstances, this function will never abort as of course libev keeps its
		984	data structures consistent.
283		985
284	=back	986	=back
285		987
		988
286	=head1 ANATOMY OF A WATCHER	989	=head1 ANATOMY OF A WATCHER
287		990
		991	In the following description, uppercase C<TYPE> in names stands for the
		992	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		993	watchers and C<ev_io_start> for I/O watchers.
		994
288	A watcher is a structure that you create and register to record your	995	A watcher is an opaque structure that you allocate and register to record
289	interest in some event. For instance, if you want to wait for STDIN to	996	your interest in some event. To make a concrete example, imagine you want
290	become readable, you would create an C<ev_io> watcher for that:	997	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		998	for that:
291		999
292	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1000	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
293	{	1001	{
294	ev_io_stop (w);	1002	ev_io_stop (w);
295	ev_unloop (loop, EVUNLOOP_ALL);	1003	ev_break (loop, EVBREAK_ALL);
296	}	1004	}
297		1005
298	struct ev_loop *loop = ev_default_loop (0);	1006	struct ev_loop *loop = ev_default_loop (0);
		1007
299	struct ev_io stdin_watcher;	1008	ev_io stdin_watcher;
		1009
300	ev_init (&stdin_watcher, my_cb);	1010	ev_init (&stdin_watcher, my_cb);
301	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	1011	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
302	ev_io_start (loop, &stdin_watcher);	1012	ev_io_start (loop, &stdin_watcher);
		1013
303	ev_loop (loop, 0);	1014	ev_run (loop, 0);
304		1015
305	As you can see, you are responsible for allocating the memory for your	1016	As you can see, you are responsible for allocating the memory for your
306	watcher structures (and it is usually a bad idea to do this on the stack,	1017	watcher structures (and it is I<usually> a bad idea to do this on the
307	although this can sometimes be quite valid).	1018	stack).
308		1019
		1020	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		1021	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
		1022
309	Each watcher structure must be initialised by a call to C<ev_init	1023	Each watcher structure must be initialised by a call to C<ev_init (watcher
310	(watcher *, callback)>, which expects a callback to be provided. This	1024	*, callback)>, which expects a callback to be provided. This callback is
311	callback gets invoked each time the event occurs (or, in the case of io	1025	invoked each time the event occurs (or, in the case of I/O watchers, each
312	watchers, each time the event loop detects that the file descriptor given	1026	time the event loop detects that the file descriptor given is readable
313	is readable and/or writable).	1027	and/or writable).
314		1028
315	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	1029	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
316	with arguments specific to this watcher type. There is also a macro	1030	macro to configure it, with arguments specific to the watcher type. There
317	to combine initialisation and setting in one call: C<< ev_<type>_init	1031	is also a macro to combine initialisation and setting in one call: C<<
318	(watcher *, callback, ...) >>.	1032	ev_TYPE_init (watcher *, callback, ...) >>.
319		1033
320	To make the watcher actually watch out for events, you have to start it	1034	To make the watcher actually watch out for events, you have to start it
321	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	1035	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
322	*) >>), and you can stop watching for events at any time by calling the	1036	*) >>), and you can stop watching for events at any time by calling the
323	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	1037	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
324		1038
325	As long as your watcher is active (has been started but not stopped) you	1039	As long as your watcher is active (has been started but not stopped) you
326	must not touch the values stored in it. Most specifically you must never	1040	must not touch the values stored in it. Most specifically you must never
327	reinitialise it or call its set method.	1041	reinitialise it or call its C<ev_TYPE_set> macro.
328
329	You can check whether an event is active by calling the C<ev_is_active
330	(watcher *)> macro. To see whether an event is outstanding (but the
331	callback for it has not been called yet) you can use the C<ev_is_pending
332	(watcher *)> macro.
333		1042
334	Each and every callback receives the event loop pointer as first, the	1043	Each and every callback receives the event loop pointer as first, the
335	registered watcher structure as second, and a bitset of received events as	1044	registered watcher structure as second, and a bitset of received events as
336	third argument.	1045	third argument.
337		1046
…		…
346	=item C<EV_WRITE>	1055	=item C<EV_WRITE>
347		1056
348	The file descriptor in the C<ev_io> watcher has become readable and/or	1057	The file descriptor in the C<ev_io> watcher has become readable and/or
349	writable.	1058	writable.
350		1059
351	=item C<EV_TIMEOUT>	1060	=item C<EV_TIMER>
352		1061
353	The C<ev_timer> watcher has timed out.	1062	The C<ev_timer> watcher has timed out.
354		1063
355	=item C<EV_PERIODIC>	1064	=item C<EV_PERIODIC>
356		1065
…		…
362		1071
363	=item C<EV_CHILD>	1072	=item C<EV_CHILD>
364		1073
365	The pid specified in the C<ev_child> watcher has received a status change.	1074	The pid specified in the C<ev_child> watcher has received a status change.
366		1075
		1076	=item C<EV_STAT>
		1077
		1078	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		1079
367	=item C<EV_IDLE>	1080	=item C<EV_IDLE>
368		1081
369	The C<ev_idle> watcher has determined that you have nothing better to do.	1082	The C<ev_idle> watcher has determined that you have nothing better to do.
370		1083
371	=item C<EV_PREPARE>	1084	=item C<EV_PREPARE>
372		1085
373	=item C<EV_CHECK>	1086	=item C<EV_CHECK>
374		1087
375	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts	1088	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts
376	to gather new events, and all C<ev_check> watchers are invoked just after	1089	to gather new events, and all C<ev_check> watchers are invoked just after
377	C<ev_loop> has gathered them, but before it invokes any callbacks for any	1090	C<ev_run> has gathered them, but before it invokes any callbacks for any
378	received events. Callbacks of both watcher types can start and stop as	1091	received events. Callbacks of both watcher types can start and stop as
379	many watchers as they want, and all of them will be taken into account	1092	many watchers as they want, and all of them will be taken into account
380	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1093	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
381	C<ev_loop> from blocking).	1094	C<ev_run> from blocking).
		1095
		1096	=item C<EV_EMBED>
		1097
		1098	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		1099
		1100	=item C<EV_FORK>
		1101
		1102	The event loop has been resumed in the child process after fork (see
		1103	C<ev_fork>).
		1104
		1105	=item C<EV_ASYNC>
		1106
		1107	The given async watcher has been asynchronously notified (see C<ev_async>).
		1108
		1109	=item C<EV_CUSTOM>
		1110
		1111	Not ever sent (or otherwise used) by libev itself, but can be freely used
		1112	by libev users to signal watchers (e.g. via C<ev_feed_event>).
382		1113
383	=item C<EV_ERROR>	1114	=item C<EV_ERROR>
384		1115
385	An unspecified error has occured, the watcher has been stopped. This might	1116	An unspecified error has occurred, the watcher has been stopped. This might
386	happen because the watcher could not be properly started because libev	1117	happen because the watcher could not be properly started because libev
387	ran out of memory, a file descriptor was found to be closed or any other	1118	ran out of memory, a file descriptor was found to be closed or any other
		1119	problem. Libev considers these application bugs.
		1120
388	problem. You best act on it by reporting the problem and somehow coping	1121	You best act on it by reporting the problem and somehow coping with the
389	with the watcher being stopped.	1122	watcher being stopped. Note that well-written programs should not receive
		1123	an error ever, so when your watcher receives it, this usually indicates a
		1124	bug in your program.
390		1125
391	Libev will usually signal a few "dummy" events together with an error,	1126	Libev will usually signal a few "dummy" events together with an error, for
392	for example it might indicate that a fd is readable or writable, and if	1127	example it might indicate that a fd is readable or writable, and if your
393	your callbacks is well-written it can just attempt the operation and cope	1128	callbacks is well-written it can just attempt the operation and cope with
394	with the error from read() or write(). This will not work in multithreaded	1129	the error from read() or write(). This will not work in multi-threaded
395	programs, though, so beware.	1130	programs, though, as the fd could already be closed and reused for another
		1131	thing, so beware.
396		1132
397	=back	1133	=back
398		1134
		1135	=head2 WATCHER STATES
		1136
		1137	There are various watcher states mentioned throughout this manual -
		1138	active, pending and so on. In this section these states and the rules to
		1139	transition between them will be described in more detail - and while these
		1140	rules might look complicated, they usually do "the right thing".
		1141
		1142	=over 4
		1143
		1144	=item initialiased
		1145
		1146	Before a watcher can be registered with the event looop it has to be
		1147	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1148	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
		1149
		1150	In this state it is simply some block of memory that is suitable for use
		1151	in an event loop. It can be moved around, freed, reused etc. at will.
		1152
		1153	=item started/running/active
		1154
		1155	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
		1156	property of the event loop, and is actively waiting for events. While in
		1157	this state it cannot be accessed (except in a few documented ways), moved,
		1158	freed or anything else - the only legal thing is to keep a pointer to it,
		1159	and call libev functions on it that are documented to work on active watchers.
		1160
		1161	=item pending
		1162
		1163	If a watcher is active and libev determines that an event it is interested
		1164	in has occured (such as a timer expiring), it will become pending. It will
		1165	stay in this pending state until either it is stopped or its callback is
		1166	about to be invoked, so it is not normally pending inside the watcher
		1167	callback.
		1168
		1169	The watcher might or might not be active while it is pending (for example,
		1170	an expired non-repeating timer can be pending but no longer active). If it
		1171	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1172	but it is still property of the event loop at this time, so cannot be
		1173	moved, freed or reused. And if it is active the rules described in the
		1174	previous item still apply.
		1175
		1176	It is also possible to feed an event on a watcher that is not active (e.g.
		1177	via C<ev_feed_event>), in which case it becomes pending without being
		1178	active.
		1179
		1180	=item stopped
		1181
		1182	A watcher can be stopped implicitly by libev (in which case it might still
		1183	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
		1184	latter will clear any pending state the watcher might be in, regardless
		1185	of whether it was active or not, so stopping a watcher explicitly before
		1186	freeing it is often a good idea.
		1187
		1188	While stopped (and not pending) the watcher is essentially in the
		1189	initialised state, that is it can be reused, moved, modified in any way
		1190	you wish.
		1191
		1192	=back
		1193
		1194	=head2 GENERIC WATCHER FUNCTIONS
		1195
		1196	=over 4
		1197
		1198	=item C<ev_init> (ev_TYPE *watcher, callback)
		1199
		1200	This macro initialises the generic portion of a watcher. The contents
		1201	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1202	the generic parts of the watcher are initialised, you I<need> to call
		1203	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1204	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1205	which rolls both calls into one.
		1206
		1207	You can reinitialise a watcher at any time as long as it has been stopped
		1208	(or never started) and there are no pending events outstanding.
		1209
		1210	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1211	int revents)>.
		1212
		1213	Example: Initialise an C<ev_io> watcher in two steps.
		1214
		1215	ev_io w;
		1216	ev_init (&w, my_cb);
		1217	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1218
		1219	=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
		1220
		1221	This macro initialises the type-specific parts of a watcher. You need to
		1222	call C<ev_init> at least once before you call this macro, but you can
		1223	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1224	macro on a watcher that is active (it can be pending, however, which is a
		1225	difference to the C<ev_init> macro).
		1226
		1227	Although some watcher types do not have type-specific arguments
		1228	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1229
		1230	See C<ev_init>, above, for an example.
		1231
		1232	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1233
		1234	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1235	calls into a single call. This is the most convenient method to initialise
		1236	a watcher. The same limitations apply, of course.
		1237
		1238	Example: Initialise and set an C<ev_io> watcher in one step.
		1239
		1240	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1241
		1242	=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
		1243
		1244	Starts (activates) the given watcher. Only active watchers will receive
		1245	events. If the watcher is already active nothing will happen.
		1246
		1247	Example: Start the C<ev_io> watcher that is being abused as example in this
		1248	whole section.
		1249
		1250	ev_io_start (EV_DEFAULT_UC, &w);
		1251
		1252	=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
		1253
		1254	Stops the given watcher if active, and clears the pending status (whether
		1255	the watcher was active or not).
		1256
		1257	It is possible that stopped watchers are pending - for example,
		1258	non-repeating timers are being stopped when they become pending - but
		1259	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1260	pending. If you want to free or reuse the memory used by the watcher it is
		1261	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1262
		1263	=item bool ev_is_active (ev_TYPE *watcher)
		1264
		1265	Returns a true value iff the watcher is active (i.e. it has been started
		1266	and not yet been stopped). As long as a watcher is active you must not modify
		1267	it.
		1268
		1269	=item bool ev_is_pending (ev_TYPE *watcher)
		1270
		1271	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1272	events but its callback has not yet been invoked). As long as a watcher
		1273	is pending (but not active) you must not call an init function on it (but
		1274	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1275	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1276	it).
		1277
		1278	=item callback ev_cb (ev_TYPE *watcher)
		1279
		1280	Returns the callback currently set on the watcher.
		1281
		1282	=item ev_cb_set (ev_TYPE *watcher, callback)
		1283
		1284	Change the callback. You can change the callback at virtually any time
		1285	(modulo threads).
		1286
		1287	=item ev_set_priority (ev_TYPE *watcher, int priority)
		1288
		1289	=item int ev_priority (ev_TYPE *watcher)
		1290
		1291	Set and query the priority of the watcher. The priority is a small
		1292	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1293	(default: C<-2>). Pending watchers with higher priority will be invoked
		1294	before watchers with lower priority, but priority will not keep watchers
		1295	from being executed (except for C<ev_idle> watchers).
		1296
		1297	If you need to suppress invocation when higher priority events are pending
		1298	you need to look at C<ev_idle> watchers, which provide this functionality.
		1299
		1300	You I<must not> change the priority of a watcher as long as it is active or
		1301	pending.
		1302
		1303	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1304	fine, as long as you do not mind that the priority value you query might
		1305	or might not have been clamped to the valid range.
		1306
		1307	The default priority used by watchers when no priority has been set is
		1308	always C<0>, which is supposed to not be too high and not be too low :).
		1309
		1310	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1311	priorities.
		1312
		1313	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1314
		1315	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1316	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1317	can deal with that fact, as both are simply passed through to the
		1318	callback.
		1319
		1320	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1321
		1322	If the watcher is pending, this function clears its pending status and
		1323	returns its C<revents> bitset (as if its callback was invoked). If the
		1324	watcher isn't pending it does nothing and returns C<0>.
		1325
		1326	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1327	callback to be invoked, which can be accomplished with this function.
		1328
		1329	=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
		1330
		1331	Feeds the given event set into the event loop, as if the specified event
		1332	had happened for the specified watcher (which must be a pointer to an
		1333	initialised but not necessarily started event watcher). Obviously you must
		1334	not free the watcher as long as it has pending events.
		1335
		1336	Stopping the watcher, letting libev invoke it, or calling
		1337	C<ev_clear_pending> will clear the pending event, even if the watcher was
		1338	not started in the first place.
		1339
		1340	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
		1341	functions that do not need a watcher.
		1342
		1343	=back
		1344
		1345
399	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1346	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
400		1347
401	Each watcher has, by default, a member C<void *data> that you can change	1348	Each watcher has, by default, a member C<void *data> that you can change
402	and read at any time, libev will completely ignore it. This can be used	1349	and read at any time: libev will completely ignore it. This can be used
403	to associate arbitrary data with your watcher. If you need more data and	1350	to associate arbitrary data with your watcher. If you need more data and
404	don't want to allocate memory and store a pointer to it in that data	1351	don't want to allocate memory and store a pointer to it in that data
405	member, you can also "subclass" the watcher type and provide your own	1352	member, you can also "subclass" the watcher type and provide your own
406	data:	1353	data:
407		1354
408	struct my_io	1355	struct my_io
409	{	1356	{
410	struct ev_io io;	1357	ev_io io;
411	int otherfd;	1358	int otherfd;
412	void *somedata;	1359	void *somedata;
413	struct whatever *mostinteresting;	1360	struct whatever *mostinteresting;
414	}	1361	};
		1362
		1363	...
		1364	struct my_io w;
		1365	ev_io_init (&w.io, my_cb, fd, EV_READ);
415		1366
416	And since your callback will be called with a pointer to the watcher, you	1367	And since your callback will be called with a pointer to the watcher, you
417	can cast it back to your own type:	1368	can cast it back to your own type:
418		1369
419	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1370	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
420	{	1371	{
421	struct my_io *w = (struct my_io *)w_;	1372	struct my_io *w = (struct my_io *)w_;
422	...	1373	...
423	}	1374	}
424		1375
425	More interesting and less C-conformant ways of catsing your callback type	1376	More interesting and less C-conformant ways of casting your callback type
426	have been omitted....	1377	instead have been omitted.
		1378
		1379	Another common scenario is to use some data structure with multiple
		1380	embedded watchers:
		1381
		1382	struct my_biggy
		1383	{
		1384	int some_data;
		1385	ev_timer t1;
		1386	ev_timer t2;
		1387	}
		1388
		1389	In this case getting the pointer to C<my_biggy> is a bit more
		1390	complicated: Either you store the address of your C<my_biggy> struct
		1391	in the C<data> member of the watcher (for woozies), or you need to use
		1392	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1393	programmers):
		1394
		1395	#include <stddef.h>
		1396
		1397	static void
		1398	t1_cb (EV_P_ ev_timer *w, int revents)
		1399	{
		1400	struct my_biggy big = (struct my_biggy *)
		1401	(((char *)w) - offsetof (struct my_biggy, t1));
		1402	}
		1403
		1404	static void
		1405	t2_cb (EV_P_ ev_timer *w, int revents)
		1406	{
		1407	struct my_biggy big = (struct my_biggy *)
		1408	(((char *)w) - offsetof (struct my_biggy, t2));
		1409	}
		1410
		1411	=head2 WATCHER PRIORITY MODELS
		1412
		1413	Many event loops support I<watcher priorities>, which are usually small
		1414	integers that influence the ordering of event callback invocation
		1415	between watchers in some way, all else being equal.
		1416
		1417	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1418	description for the more technical details such as the actual priority
		1419	range.
		1420
		1421	There are two common ways how these these priorities are being interpreted
		1422	by event loops:
		1423
		1424	In the more common lock-out model, higher priorities "lock out" invocation
		1425	of lower priority watchers, which means as long as higher priority
		1426	watchers receive events, lower priority watchers are not being invoked.
		1427
		1428	The less common only-for-ordering model uses priorities solely to order
		1429	callback invocation within a single event loop iteration: Higher priority
		1430	watchers are invoked before lower priority ones, but they all get invoked
		1431	before polling for new events.
		1432
		1433	Libev uses the second (only-for-ordering) model for all its watchers
		1434	except for idle watchers (which use the lock-out model).
		1435
		1436	The rationale behind this is that implementing the lock-out model for
		1437	watchers is not well supported by most kernel interfaces, and most event
		1438	libraries will just poll for the same events again and again as long as
		1439	their callbacks have not been executed, which is very inefficient in the
		1440	common case of one high-priority watcher locking out a mass of lower
		1441	priority ones.
		1442
		1443	Static (ordering) priorities are most useful when you have two or more
		1444	watchers handling the same resource: a typical usage example is having an
		1445	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1446	timeouts. Under load, data might be received while the program handles
		1447	other jobs, but since timers normally get invoked first, the timeout
		1448	handler will be executed before checking for data. In that case, giving
		1449	the timer a lower priority than the I/O watcher ensures that I/O will be
		1450	handled first even under adverse conditions (which is usually, but not
		1451	always, what you want).
		1452
		1453	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1454	will only be executed when no same or higher priority watchers have
		1455	received events, they can be used to implement the "lock-out" model when
		1456	required.
		1457
		1458	For example, to emulate how many other event libraries handle priorities,
		1459	you can associate an C<ev_idle> watcher to each such watcher, and in
		1460	the normal watcher callback, you just start the idle watcher. The real
		1461	processing is done in the idle watcher callback. This causes libev to
		1462	continuously poll and process kernel event data for the watcher, but when
		1463	the lock-out case is known to be rare (which in turn is rare :), this is
		1464	workable.
		1465
		1466	Usually, however, the lock-out model implemented that way will perform
		1467	miserably under the type of load it was designed to handle. In that case,
		1468	it might be preferable to stop the real watcher before starting the
		1469	idle watcher, so the kernel will not have to process the event in case
		1470	the actual processing will be delayed for considerable time.
		1471
		1472	Here is an example of an I/O watcher that should run at a strictly lower
		1473	priority than the default, and which should only process data when no
		1474	other events are pending:
		1475
		1476	ev_idle idle; // actual processing watcher
		1477	ev_io io; // actual event watcher
		1478
		1479	static void
		1480	io_cb (EV_P_ ev_io *w, int revents)
		1481	{
		1482	// stop the I/O watcher, we received the event, but
		1483	// are not yet ready to handle it.
		1484	ev_io_stop (EV_A_ w);
		1485
		1486	// start the idle watcher to handle the actual event.
		1487	// it will not be executed as long as other watchers
		1488	// with the default priority are receiving events.
		1489	ev_idle_start (EV_A_ &idle);
		1490	}
		1491
		1492	static void
		1493	idle_cb (EV_P_ ev_idle *w, int revents)
		1494	{
		1495	// actual processing
		1496	read (STDIN_FILENO, ...);
		1497
		1498	// have to start the I/O watcher again, as
		1499	// we have handled the event
		1500	ev_io_start (EV_P_ &io);
		1501	}
		1502
		1503	// initialisation
		1504	ev_idle_init (&idle, idle_cb);
		1505	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1506	ev_io_start (EV_DEFAULT_ &io);
		1507
		1508	In the "real" world, it might also be beneficial to start a timer, so that
		1509	low-priority connections can not be locked out forever under load. This
		1510	enables your program to keep a lower latency for important connections
		1511	during short periods of high load, while not completely locking out less
		1512	important ones.
427		1513
428		1514
429	=head1 WATCHER TYPES	1515	=head1 WATCHER TYPES
430		1516
431	This section describes each watcher in detail, but will not repeat	1517	This section describes each watcher in detail, but will not repeat
432	information given in the last section.	1518	information given in the last section. Any initialisation/set macros,
		1519	functions and members specific to the watcher type are explained.
433		1520
		1521	Members are additionally marked with either I<[read-only]>, meaning that,
		1522	while the watcher is active, you can look at the member and expect some
		1523	sensible content, but you must not modify it (you can modify it while the
		1524	watcher is stopped to your hearts content), or I<[read-write]>, which
		1525	means you can expect it to have some sensible content while the watcher
		1526	is active, but you can also modify it. Modifying it may not do something
		1527	sensible or take immediate effect (or do anything at all), but libev will
		1528	not crash or malfunction in any way.
		1529
		1530
434	=head2 C<ev_io> - is this file descriptor readable or writable	1531	=head2 C<ev_io> - is this file descriptor readable or writable?
435		1532
436	I/O watchers check whether a file descriptor is readable or writable	1533	I/O watchers check whether a file descriptor is readable or writable
437	in each iteration of the event loop (This behaviour is called	1534	in each iteration of the event loop, or, more precisely, when reading
438	level-triggering because you keep receiving events as long as the	1535	would not block the process and writing would at least be able to write
439	condition persists. Remember you can stop the watcher if you don't want to	1536	some data. This behaviour is called level-triggering because you keep
440	act on the event and neither want to receive future events).	1537	receiving events as long as the condition persists. Remember you can stop
		1538	the watcher if you don't want to act on the event and neither want to
		1539	receive future events.
441		1540
442	In general you can register as many read and/or write event watchers per	1541	In general you can register as many read and/or write event watchers per
443	fd as you want (as long as you don't confuse yourself). Setting all file	1542	fd as you want (as long as you don't confuse yourself). Setting all file
444	descriptors to non-blocking mode is also usually a good idea (but not	1543	descriptors to non-blocking mode is also usually a good idea (but not
445	required if you know what you are doing).	1544	required if you know what you are doing).
446		1545
447	You have to be careful with dup'ed file descriptors, though. Some backends	1546	If you cannot use non-blocking mode, then force the use of a
448	(the linux epoll backend is a notable example) cannot handle dup'ed file	1547	known-to-be-good backend (at the time of this writing, this includes only
449	descriptors correctly if you register interest in two or more fds pointing	1548	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
450	to the same underlying file/socket etc. description (that is, they share	1549	descriptors for which non-blocking operation makes no sense (such as
451	the same underlying "file open").	1550	files) - libev doesn't guarantee any specific behaviour in that case.
452		1551
453	If you must do this, then force the use of a known-to-be-good backend	1552	Another thing you have to watch out for is that it is quite easy to
454	(at the time of this writing, this includes only EVMETHOD_SELECT and	1553	receive "spurious" readiness notifications, that is your callback might
455	EVMETHOD_POLL).	1554	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1555	because there is no data. Not only are some backends known to create a
		1556	lot of those (for example Solaris ports), it is very easy to get into
		1557	this situation even with a relatively standard program structure. Thus
		1558	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1559	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1560
		1561	If you cannot run the fd in non-blocking mode (for example you should
		1562	not play around with an Xlib connection), then you have to separately
		1563	re-test whether a file descriptor is really ready with a known-to-be good
		1564	interface such as poll (fortunately in our Xlib example, Xlib already
		1565	does this on its own, so its quite safe to use). Some people additionally
		1566	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1567	indefinitely.
		1568
		1569	But really, best use non-blocking mode.
		1570
		1571	=head3 The special problem of disappearing file descriptors
		1572
		1573	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1574	descriptor (either due to calling C<close> explicitly or any other means,
		1575	such as C<dup2>). The reason is that you register interest in some file
		1576	descriptor, but when it goes away, the operating system will silently drop
		1577	this interest. If another file descriptor with the same number then is
		1578	registered with libev, there is no efficient way to see that this is, in
		1579	fact, a different file descriptor.
		1580
		1581	To avoid having to explicitly tell libev about such cases, libev follows
		1582	the following policy: Each time C<ev_io_set> is being called, libev
		1583	will assume that this is potentially a new file descriptor, otherwise
		1584	it is assumed that the file descriptor stays the same. That means that
		1585	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1586	descriptor even if the file descriptor number itself did not change.
		1587
		1588	This is how one would do it normally anyway, the important point is that
		1589	the libev application should not optimise around libev but should leave
		1590	optimisations to libev.
		1591
		1592	=head3 The special problem of dup'ed file descriptors
		1593
		1594	Some backends (e.g. epoll), cannot register events for file descriptors,
		1595	but only events for the underlying file descriptions. That means when you
		1596	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1597	events for them, only one file descriptor might actually receive events.
		1598
		1599	There is no workaround possible except not registering events
		1600	for potentially C<dup ()>'ed file descriptors, or to resort to
		1601	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1602
		1603	=head3 The special problem of fork
		1604
		1605	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1606	useless behaviour. Libev fully supports fork, but needs to be told about
		1607	it in the child.
		1608
		1609	To support fork in your programs, you either have to call
		1610	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1611	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1612	C<EVBACKEND_POLL>.
		1613
		1614	=head3 The special problem of SIGPIPE
		1615
		1616	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1617	when writing to a pipe whose other end has been closed, your program gets
		1618	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1619	this is sensible behaviour, for daemons, this is usually undesirable.
		1620
		1621	So when you encounter spurious, unexplained daemon exits, make sure you
		1622	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1623	somewhere, as that would have given you a big clue).
		1624
		1625	=head3 The special problem of accept()ing when you can't
		1626
		1627	Many implementations of the POSIX C<accept> function (for example,
		1628	found in post-2004 Linux) have the peculiar behaviour of not removing a
		1629	connection from the pending queue in all error cases.
		1630
		1631	For example, larger servers often run out of file descriptors (because
		1632	of resource limits), causing C<accept> to fail with C<ENFILE> but not
		1633	rejecting the connection, leading to libev signalling readiness on
		1634	the next iteration again (the connection still exists after all), and
		1635	typically causing the program to loop at 100% CPU usage.
		1636
		1637	Unfortunately, the set of errors that cause this issue differs between
		1638	operating systems, there is usually little the app can do to remedy the
		1639	situation, and no known thread-safe method of removing the connection to
		1640	cope with overload is known (to me).
		1641
		1642	One of the easiest ways to handle this situation is to just ignore it
		1643	- when the program encounters an overload, it will just loop until the
		1644	situation is over. While this is a form of busy waiting, no OS offers an
		1645	event-based way to handle this situation, so it's the best one can do.
		1646
		1647	A better way to handle the situation is to log any errors other than
		1648	C<EAGAIN> and C<EWOULDBLOCK>, making sure not to flood the log with such
		1649	messages, and continue as usual, which at least gives the user an idea of
		1650	what could be wrong ("raise the ulimit!"). For extra points one could stop
		1651	the C<ev_io> watcher on the listening fd "for a while", which reduces CPU
		1652	usage.
		1653
		1654	If your program is single-threaded, then you could also keep a dummy file
		1655	descriptor for overload situations (e.g. by opening F</dev/null>), and
		1656	when you run into C<ENFILE> or C<EMFILE>, close it, run C<accept>,
		1657	close that fd, and create a new dummy fd. This will gracefully refuse
		1658	clients under typical overload conditions.
		1659
		1660	The last way to handle it is to simply log the error and C<exit>, as
		1661	is often done with C<malloc> failures, but this results in an easy
		1662	opportunity for a DoS attack.
		1663
		1664	=head3 Watcher-Specific Functions
456		1665
457	=over 4	1666	=over 4
458		1667
459	=item ev_io_init (ev_io *, callback, int fd, int events)	1668	=item ev_io_init (ev_io *, callback, int fd, int events)
460		1669
461	=item ev_io_set (ev_io *, int fd, int events)	1670	=item ev_io_set (ev_io *, int fd, int events)
462		1671
463	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1672	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
464	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1673	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
465	EV_WRITE> to receive the given events.	1674	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1675
		1676	=item int fd [read-only]
		1677
		1678	The file descriptor being watched.
		1679
		1680	=item int events [read-only]
		1681
		1682	The events being watched.
466		1683
467	=back	1684	=back
468		1685
		1686	=head3 Examples
		1687
		1688	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1689	readable, but only once. Since it is likely line-buffered, you could
		1690	attempt to read a whole line in the callback.
		1691
		1692	static void
		1693	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1694	{
		1695	ev_io_stop (loop, w);
		1696	.. read from stdin here (or from w->fd) and handle any I/O errors
		1697	}
		1698
		1699	...
		1700	struct ev_loop *loop = ev_default_init (0);
		1701	ev_io stdin_readable;
		1702	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1703	ev_io_start (loop, &stdin_readable);
		1704	ev_run (loop, 0);
		1705
		1706
469	=head2 C<ev_timer> - relative and optionally recurring timeouts	1707	=head2 C<ev_timer> - relative and optionally repeating timeouts
470		1708
471	Timer watchers are simple relative timers that generate an event after a	1709	Timer watchers are simple relative timers that generate an event after a
472	given time, and optionally repeating in regular intervals after that.	1710	given time, and optionally repeating in regular intervals after that.
473		1711
474	The timers are based on real time, that is, if you register an event that	1712	The timers are based on real time, that is, if you register an event that
475	times out after an hour and you reset your system clock to last years	1713	times out after an hour and you reset your system clock to January last
476	time, it will still time out after (roughly) and hour. "Roughly" because	1714	year, it will still time out after (roughly) one hour. "Roughly" because
477	detecting time jumps is hard, and soem inaccuracies are unavoidable (the	1715	detecting time jumps is hard, and some inaccuracies are unavoidable (the
478	monotonic clock option helps a lot here).	1716	monotonic clock option helps a lot here).
		1717
		1718	The callback is guaranteed to be invoked only I<after> its timeout has
		1719	passed (not I<at>, so on systems with very low-resolution clocks this
		1720	might introduce a small delay). If multiple timers become ready during the
		1721	same loop iteration then the ones with earlier time-out values are invoked
		1722	before ones of the same priority with later time-out values (but this is
		1723	no longer true when a callback calls C<ev_run> recursively).
		1724
		1725	=head3 Be smart about timeouts
		1726
		1727	Many real-world problems involve some kind of timeout, usually for error
		1728	recovery. A typical example is an HTTP request - if the other side hangs,
		1729	you want to raise some error after a while.
		1730
		1731	What follows are some ways to handle this problem, from obvious and
		1732	inefficient to smart and efficient.
		1733
		1734	In the following, a 60 second activity timeout is assumed - a timeout that
		1735	gets reset to 60 seconds each time there is activity (e.g. each time some
		1736	data or other life sign was received).
		1737
		1738	=over 4
		1739
		1740	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1741
		1742	This is the most obvious, but not the most simple way: In the beginning,
		1743	start the watcher:
		1744
		1745	ev_timer_init (timer, callback, 60., 0.);
		1746	ev_timer_start (loop, timer);
		1747
		1748	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1749	and start it again:
		1750
		1751	ev_timer_stop (loop, timer);
		1752	ev_timer_set (timer, 60., 0.);
		1753	ev_timer_start (loop, timer);
		1754
		1755	This is relatively simple to implement, but means that each time there is
		1756	some activity, libev will first have to remove the timer from its internal
		1757	data structure and then add it again. Libev tries to be fast, but it's
		1758	still not a constant-time operation.
		1759
		1760	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1761
		1762	This is the easiest way, and involves using C<ev_timer_again> instead of
		1763	C<ev_timer_start>.
		1764
		1765	To implement this, configure an C<ev_timer> with a C<repeat> value
		1766	of C<60> and then call C<ev_timer_again> at start and each time you
		1767	successfully read or write some data. If you go into an idle state where
		1768	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1769	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1770
		1771	That means you can ignore both the C<ev_timer_start> function and the
		1772	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1773	member and C<ev_timer_again>.
		1774
		1775	At start:
		1776
		1777	ev_init (timer, callback);
		1778	timer->repeat = 60.;
		1779	ev_timer_again (loop, timer);
		1780
		1781	Each time there is some activity:
		1782
		1783	ev_timer_again (loop, timer);
		1784
		1785	It is even possible to change the time-out on the fly, regardless of
		1786	whether the watcher is active or not:
		1787
		1788	timer->repeat = 30.;
		1789	ev_timer_again (loop, timer);
		1790
		1791	This is slightly more efficient then stopping/starting the timer each time
		1792	you want to modify its timeout value, as libev does not have to completely
		1793	remove and re-insert the timer from/into its internal data structure.
		1794
		1795	It is, however, even simpler than the "obvious" way to do it.
		1796
		1797	=item 3. Let the timer time out, but then re-arm it as required.
		1798
		1799	This method is more tricky, but usually most efficient: Most timeouts are
		1800	relatively long compared to the intervals between other activity - in
		1801	our example, within 60 seconds, there are usually many I/O events with
		1802	associated activity resets.
		1803
		1804	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1805	but remember the time of last activity, and check for a real timeout only
		1806	within the callback:
		1807
		1808	ev_tstamp last_activity; // time of last activity
		1809
		1810	static void
		1811	callback (EV_P_ ev_timer *w, int revents)
		1812	{
		1813	ev_tstamp now = ev_now (EV_A);
		1814	ev_tstamp timeout = last_activity + 60.;
		1815
		1816	// if last_activity + 60. is older than now, we did time out
		1817	if (timeout < now)
		1818	{
		1819	// timeout occurred, take action
		1820	}
		1821	else
		1822	{
		1823	// callback was invoked, but there was some activity, re-arm
		1824	// the watcher to fire in last_activity + 60, which is
		1825	// guaranteed to be in the future, so "again" is positive:
		1826	w->repeat = timeout - now;
		1827	ev_timer_again (EV_A_ w);
		1828	}
		1829	}
		1830
		1831	To summarise the callback: first calculate the real timeout (defined
		1832	as "60 seconds after the last activity"), then check if that time has
		1833	been reached, which means something I<did>, in fact, time out. Otherwise
		1834	the callback was invoked too early (C<timeout> is in the future), so
		1835	re-schedule the timer to fire at that future time, to see if maybe we have
		1836	a timeout then.
		1837
		1838	Note how C<ev_timer_again> is used, taking advantage of the
		1839	C<ev_timer_again> optimisation when the timer is already running.
		1840
		1841	This scheme causes more callback invocations (about one every 60 seconds
		1842	minus half the average time between activity), but virtually no calls to
		1843	libev to change the timeout.
		1844
		1845	To start the timer, simply initialise the watcher and set C<last_activity>
		1846	to the current time (meaning we just have some activity :), then call the
		1847	callback, which will "do the right thing" and start the timer:
		1848
		1849	ev_init (timer, callback);
		1850	last_activity = ev_now (loop);
		1851	callback (loop, timer, EV_TIMER);
		1852
		1853	And when there is some activity, simply store the current time in
		1854	C<last_activity>, no libev calls at all:
		1855
		1856	last_activity = ev_now (loop);
		1857
		1858	This technique is slightly more complex, but in most cases where the
		1859	time-out is unlikely to be triggered, much more efficient.
		1860
		1861	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1862	callback :) - just change the timeout and invoke the callback, which will
		1863	fix things for you.
		1864
		1865	=item 4. Wee, just use a double-linked list for your timeouts.
		1866
		1867	If there is not one request, but many thousands (millions...), all
		1868	employing some kind of timeout with the same timeout value, then one can
		1869	do even better:
		1870
		1871	When starting the timeout, calculate the timeout value and put the timeout
		1872	at the I<end> of the list.
		1873
		1874	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1875	the list is expected to fire (for example, using the technique #3).
		1876
		1877	When there is some activity, remove the timer from the list, recalculate
		1878	the timeout, append it to the end of the list again, and make sure to
		1879	update the C<ev_timer> if it was taken from the beginning of the list.
		1880
		1881	This way, one can manage an unlimited number of timeouts in O(1) time for
		1882	starting, stopping and updating the timers, at the expense of a major
		1883	complication, and having to use a constant timeout. The constant timeout
		1884	ensures that the list stays sorted.
		1885
		1886	=back
		1887
		1888	So which method the best?
		1889
		1890	Method #2 is a simple no-brain-required solution that is adequate in most
		1891	situations. Method #3 requires a bit more thinking, but handles many cases
		1892	better, and isn't very complicated either. In most case, choosing either
		1893	one is fine, with #3 being better in typical situations.
		1894
		1895	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1896	rather complicated, but extremely efficient, something that really pays
		1897	off after the first million or so of active timers, i.e. it's usually
		1898	overkill :)
		1899
		1900	=head3 The special problem of time updates
		1901
		1902	Establishing the current time is a costly operation (it usually takes at
		1903	least two system calls): EV therefore updates its idea of the current
		1904	time only before and after C<ev_run> collects new events, which causes a
		1905	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1906	lots of events in one iteration.
479		1907
480	The relative timeouts are calculated relative to the C<ev_now ()>	1908	The relative timeouts are calculated relative to the C<ev_now ()>
481	time. This is usually the right thing as this timestamp refers to the time	1909	time. This is usually the right thing as this timestamp refers to the time
482	of the event triggering whatever timeout you are modifying/starting. If	1910	of the event triggering whatever timeout you are modifying/starting. If
483	you suspect event processing to be delayed and you *need* to base the timeout	1911	you suspect event processing to be delayed and you I<need> to base the
484	on the current time, use something like this to adjust for this:	1912	timeout on the current time, use something like this to adjust for this:
485		1913
486	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1914	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
487		1915
		1916	If the event loop is suspended for a long time, you can also force an
		1917	update of the time returned by C<ev_now ()> by calling C<ev_now_update
		1918	()>.
		1919
		1920	=head3 The special problems of suspended animation
		1921
		1922	When you leave the server world it is quite customary to hit machines that
		1923	can suspend/hibernate - what happens to the clocks during such a suspend?
		1924
		1925	Some quick tests made with a Linux 2.6.28 indicate that a suspend freezes
		1926	all processes, while the clocks (C<times>, C<CLOCK_MONOTONIC>) continue
		1927	to run until the system is suspended, but they will not advance while the
		1928	system is suspended. That means, on resume, it will be as if the program
		1929	was frozen for a few seconds, but the suspend time will not be counted
		1930	towards C<ev_timer> when a monotonic clock source is used. The real time
		1931	clock advanced as expected, but if it is used as sole clocksource, then a
		1932	long suspend would be detected as a time jump by libev, and timers would
		1933	be adjusted accordingly.
		1934
		1935	I would not be surprised to see different behaviour in different between
		1936	operating systems, OS versions or even different hardware.
		1937
		1938	The other form of suspend (job control, or sending a SIGSTOP) will see a
		1939	time jump in the monotonic clocks and the realtime clock. If the program
		1940	is suspended for a very long time, and monotonic clock sources are in use,
		1941	then you can expect C<ev_timer>s to expire as the full suspension time
		1942	will be counted towards the timers. When no monotonic clock source is in
		1943	use, then libev will again assume a timejump and adjust accordingly.
		1944
		1945	It might be beneficial for this latter case to call C<ev_suspend>
		1946	and C<ev_resume> in code that handles C<SIGTSTP>, to at least get
		1947	deterministic behaviour in this case (you can do nothing against
		1948	C<SIGSTOP>).
		1949
		1950	=head3 Watcher-Specific Functions and Data Members
		1951
488	=over 4	1952	=over 4
489		1953
490	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1954	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
491		1955
492	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1956	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
493		1957
494	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1958	Configure the timer to trigger after C<after> seconds. If C<repeat>
495	C<0.>, then it will automatically be stopped. If it is positive, then the	1959	is C<0.>, then it will automatically be stopped once the timeout is
496	timer will automatically be configured to trigger again C<repeat> seconds	1960	reached. If it is positive, then the timer will automatically be
497	later, again, and again, until stopped manually.	1961	configured to trigger again C<repeat> seconds later, again, and again,
		1962	until stopped manually.
498		1963
499	The timer itself will do a best-effort at avoiding drift, that is, if you	1964	The timer itself will do a best-effort at avoiding drift, that is, if
500	configure a timer to trigger every 10 seconds, then it will trigger at	1965	you configure a timer to trigger every 10 seconds, then it will normally
501	exactly 10 second intervals. If, however, your program cannot keep up with	1966	trigger at exactly 10 second intervals. If, however, your program cannot
502	the timer (because it takes longer than those 10 seconds to do stuff) the	1967	keep up with the timer (because it takes longer than those 10 seconds to
503	timer will not fire more than once per event loop iteration.	1968	do stuff) the timer will not fire more than once per event loop iteration.
504		1969
505	=item ev_timer_again (loop)	1970	=item ev_timer_again (loop, ev_timer *)
506		1971
507	This will act as if the timer timed out and restart it again if it is	1972	This will act as if the timer timed out and restart it again if it is
508	repeating. The exact semantics are:	1973	repeating. The exact semantics are:
509		1974
		1975	If the timer is pending, its pending status is cleared.
		1976
510	If the timer is started but nonrepeating, stop it.	1977	If the timer is started but non-repeating, stop it (as if it timed out).
511		1978
512	If the timer is repeating, either start it if necessary (with the repeat	1979	If the timer is repeating, either start it if necessary (with the
513	value), or reset the running timer to the repeat value.	1980	C<repeat> value), or reset the running timer to the C<repeat> value.
514		1981
515	This sounds a bit complicated, but here is a useful and typical	1982	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
516	example: Imagine you have a tcp connection and you want a so-called idle	1983	usage example.
517	timeout, that is, you want to be called when there have been, say, 60	1984
518	seconds of inactivity on the socket. The easiest way to do this is to	1985	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
519	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each	1986
520	time you successfully read or write some data. If you go into an idle	1987	Returns the remaining time until a timer fires. If the timer is active,
521	state where you do not expect data to travel on the socket, you can stop	1988	then this time is relative to the current event loop time, otherwise it's
522	the timer, and again will automatically restart it if need be.	1989	the timeout value currently configured.
		1990
		1991	That is, after an C<ev_timer_set (w, 5, 7)>, C<ev_timer_remaining> returns
		1992	C<5>. When the timer is started and one second passes, C<ev_timer_remaining>
		1993	will return C<4>. When the timer expires and is restarted, it will return
		1994	roughly C<7> (likely slightly less as callback invocation takes some time,
		1995	too), and so on.
		1996
		1997	=item ev_tstamp repeat [read-write]
		1998
		1999	The current C<repeat> value. Will be used each time the watcher times out
		2000	or C<ev_timer_again> is called, and determines the next timeout (if any),
		2001	which is also when any modifications are taken into account.
523		2002
524	=back	2003	=back
525		2004
		2005	=head3 Examples
		2006
		2007	Example: Create a timer that fires after 60 seconds.
		2008
		2009	static void
		2010	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2011	{
		2012	.. one minute over, w is actually stopped right here
		2013	}
		2014
		2015	ev_timer mytimer;
		2016	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		2017	ev_timer_start (loop, &mytimer);
		2018
		2019	Example: Create a timeout timer that times out after 10 seconds of
		2020	inactivity.
		2021
		2022	static void
		2023	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		2024	{
		2025	.. ten seconds without any activity
		2026	}
		2027
		2028	ev_timer mytimer;
		2029	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		2030	ev_timer_again (&mytimer); /* start timer */
		2031	ev_run (loop, 0);
		2032
		2033	// and in some piece of code that gets executed on any "activity":
		2034	// reset the timeout to start ticking again at 10 seconds
		2035	ev_timer_again (&mytimer);
		2036
		2037
526	=head2 C<ev_periodic> - to cron or not to cron	2038	=head2 C<ev_periodic> - to cron or not to cron?
527		2039
528	Periodic watchers are also timers of a kind, but they are very versatile	2040	Periodic watchers are also timers of a kind, but they are very versatile
529	(and unfortunately a bit complex).	2041	(and unfortunately a bit complex).
530		2042
531	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	2043	Unlike C<ev_timer>, periodic watchers are not based on real time (or
532	but on wallclock time (absolute time). You can tell a periodic watcher	2044	relative time, the physical time that passes) but on wall clock time
533	to trigger "at" some specific point in time. For example, if you tell a	2045	(absolute time, the thing you can read on your calender or clock). The
534	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	2046	difference is that wall clock time can run faster or slower than real
535	+ 10.>) and then reset your system clock to the last year, then it will	2047	time, and time jumps are not uncommon (e.g. when you adjust your
536	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	2048	wrist-watch).
537	roughly 10 seconds later and of course not if you reset your system time
538	again).
539		2049
540	They can also be used to implement vastly more complex timers, such as	2050	You can tell a periodic watcher to trigger after some specific point
		2051	in time: for example, if you tell a periodic watcher to trigger "in 10
		2052	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		2053	not a delay) and then reset your system clock to January of the previous
		2054	year, then it will take a year or more to trigger the event (unlike an
		2055	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		2056	it, as it uses a relative timeout).
		2057
		2058	C<ev_periodic> watchers can also be used to implement vastly more complex
541	triggering an event on eahc midnight, local time.	2059	timers, such as triggering an event on each "midnight, local time", or
		2060	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		2061	those cannot react to time jumps.
		2062
		2063	As with timers, the callback is guaranteed to be invoked only when the
		2064	point in time where it is supposed to trigger has passed. If multiple
		2065	timers become ready during the same loop iteration then the ones with
		2066	earlier time-out values are invoked before ones with later time-out values
		2067	(but this is no longer true when a callback calls C<ev_run> recursively).
		2068
		2069	=head3 Watcher-Specific Functions and Data Members
542		2070
543	=over 4	2071	=over 4
544		2072
545	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	2073	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
546		2074
547	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	2075	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
548		2076
549	Lots of arguments, lets sort it out... There are basically three modes of	2077	Lots of arguments, let's sort it out... There are basically three modes of
550	operation, and we will explain them from simplest to complex:	2078	operation, and we will explain them from simplest to most complex:
551
552		2079
553	=over 4	2080	=over 4
554		2081
555	=item * absolute timer (interval = reschedule_cb = 0)	2082	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
556		2083
557	In this configuration the watcher triggers an event at the wallclock time	2084	In this configuration the watcher triggers an event after the wall clock
558	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	2085	time C<offset> has passed. It will not repeat and will not adjust when a
559	that is, if it is to be run at January 1st 2011 then it will run when the	2086	time jump occurs, that is, if it is to be run at January 1st 2011 then it
560	system time reaches or surpasses this time.	2087	will be stopped and invoked when the system clock reaches or surpasses
		2088	this point in time.
561		2089
562	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	2090	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
563		2091
564	In this mode the watcher will always be scheduled to time out at the next	2092	In this mode the watcher will always be scheduled to time out at the next
565	C<at + N * interval> time (for some integer N) and then repeat, regardless	2093	C<offset + N * interval> time (for some integer N, which can also be
566	of any time jumps.	2094	negative) and then repeat, regardless of any time jumps. The C<offset>
		2095	argument is merely an offset into the C<interval> periods.
567		2096
568	This can be used to create timers that do not drift with respect to system	2097	This can be used to create timers that do not drift with respect to the
569	time:	2098	system clock, for example, here is an C<ev_periodic> that triggers each
		2099	hour, on the hour (with respect to UTC):
570		2100
571	ev_periodic_set (&periodic, 0., 3600., 0);	2101	ev_periodic_set (&periodic, 0., 3600., 0);
572		2102
573	This doesn't mean there will always be 3600 seconds in between triggers,	2103	This doesn't mean there will always be 3600 seconds in between triggers,
574	but only that the the callback will be called when the system time shows a	2104	but only that the callback will be called when the system time shows a
575	full hour (UTC), or more correctly, when the system time is evenly divisible	2105	full hour (UTC), or more correctly, when the system time is evenly divisible
576	by 3600.	2106	by 3600.
577		2107
578	Another way to think about it (for the mathematically inclined) is that	2108	Another way to think about it (for the mathematically inclined) is that
579	C<ev_periodic> will try to run the callback in this mode at the next possible	2109	C<ev_periodic> will try to run the callback in this mode at the next possible
580	time where C<time = at (mod interval)>, regardless of any time jumps.	2110	time where C<time = offset (mod interval)>, regardless of any time jumps.
581		2111
		2112	For numerical stability it is preferable that the C<offset> value is near
		2113	C<ev_now ()> (the current time), but there is no range requirement for
		2114	this value, and in fact is often specified as zero.
		2115
		2116	Note also that there is an upper limit to how often a timer can fire (CPU
		2117	speed for example), so if C<interval> is very small then timing stability
		2118	will of course deteriorate. Libev itself tries to be exact to be about one
		2119	millisecond (if the OS supports it and the machine is fast enough).
		2120
582	=item * manual reschedule mode (reschedule_cb = callback)	2121	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
583		2122
584	In this mode the values for C<interval> and C<at> are both being	2123	In this mode the values for C<interval> and C<offset> are both being
585	ignored. Instead, each time the periodic watcher gets scheduled, the	2124	ignored. Instead, each time the periodic watcher gets scheduled, the
586	reschedule callback will be called with the watcher as first, and the	2125	reschedule callback will be called with the watcher as first, and the
587	current time as second argument.	2126	current time as second argument.
588		2127
589	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	2128	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
590	ever, or make any event loop modifications>. If you need to stop it,	2129	or make ANY other event loop modifications whatsoever, unless explicitly
591	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	2130	allowed by documentation here>.
592	starting a prepare watcher).
593		2131
		2132	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		2133	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		2134	only event loop modification you are allowed to do).
		2135
594	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	2136	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
595	ev_tstamp now)>, e.g.:	2137	*w, ev_tstamp now)>, e.g.:
596		2138
		2139	static ev_tstamp
597	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	2140	my_rescheduler (ev_periodic *w, ev_tstamp now)
598	{	2141	{
599	return now + 60.;	2142	return now + 60.;
600	}	2143	}
601		2144
602	It must return the next time to trigger, based on the passed time value	2145	It must return the next time to trigger, based on the passed time value
603	(that is, the lowest time value larger than to the second argument). It	2146	(that is, the lowest time value larger than to the second argument). It
604	will usually be called just before the callback will be triggered, but	2147	will usually be called just before the callback will be triggered, but
605	might be called at other times, too.	2148	might be called at other times, too.
606		2149
607	NOTE: I<< This callback must always return a time that is later than the	2150	NOTE: I<< This callback must always return a time that is higher than or
608	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	2151	equal to the passed C<now> value >>.
609		2152
610	This can be used to create very complex timers, such as a timer that	2153	This can be used to create very complex timers, such as a timer that
611	triggers on each midnight, local time. To do this, you would calculate the	2154	triggers on "next midnight, local time". To do this, you would calculate the
612	next midnight after C<now> and return the timestamp value for this. How	2155	next midnight after C<now> and return the timestamp value for this. How
613	you do this is, again, up to you (but it is not trivial, which is the main	2156	you do this is, again, up to you (but it is not trivial, which is the main
614	reason I omitted it as an example).	2157	reason I omitted it as an example).
615		2158
616	=back	2159	=back
…		…
620	Simply stops and restarts the periodic watcher again. This is only useful	2163	Simply stops and restarts the periodic watcher again. This is only useful
621	when you changed some parameters or the reschedule callback would return	2164	when you changed some parameters or the reschedule callback would return
622	a different time than the last time it was called (e.g. in a crond like	2165	a different time than the last time it was called (e.g. in a crond like
623	program when the crontabs have changed).	2166	program when the crontabs have changed).
624		2167
		2168	=item ev_tstamp ev_periodic_at (ev_periodic *)
		2169
		2170	When active, returns the absolute time that the watcher is supposed
		2171	to trigger next. This is not the same as the C<offset> argument to
		2172	C<ev_periodic_set>, but indeed works even in interval and manual
		2173	rescheduling modes.
		2174
		2175	=item ev_tstamp offset [read-write]
		2176
		2177	When repeating, this contains the offset value, otherwise this is the
		2178	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		2179	although libev might modify this value for better numerical stability).
		2180
		2181	Can be modified any time, but changes only take effect when the periodic
		2182	timer fires or C<ev_periodic_again> is being called.
		2183
		2184	=item ev_tstamp interval [read-write]
		2185
		2186	The current interval value. Can be modified any time, but changes only
		2187	take effect when the periodic timer fires or C<ev_periodic_again> is being
		2188	called.
		2189
		2190	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		2191
		2192	The current reschedule callback, or C<0>, if this functionality is
		2193	switched off. Can be changed any time, but changes only take effect when
		2194	the periodic timer fires or C<ev_periodic_again> is being called.
		2195
625	=back	2196	=back
626		2197
		2198	=head3 Examples
		2199
		2200	Example: Call a callback every hour, or, more precisely, whenever the
		2201	system time is divisible by 3600. The callback invocation times have
		2202	potentially a lot of jitter, but good long-term stability.
		2203
		2204	static void
		2205	clock_cb (struct ev_loop *loop, ev_periodic *w, int revents)
		2206	{
		2207	... its now a full hour (UTC, or TAI or whatever your clock follows)
		2208	}
		2209
		2210	ev_periodic hourly_tick;
		2211	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		2212	ev_periodic_start (loop, &hourly_tick);
		2213
		2214	Example: The same as above, but use a reschedule callback to do it:
		2215
		2216	#include <math.h>
		2217
		2218	static ev_tstamp
		2219	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		2220	{
		2221	return now + (3600. - fmod (now, 3600.));
		2222	}
		2223
		2224	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		2225
		2226	Example: Call a callback every hour, starting now:
		2227
		2228	ev_periodic hourly_tick;
		2229	ev_periodic_init (&hourly_tick, clock_cb,
		2230	fmod (ev_now (loop), 3600.), 3600., 0);
		2231	ev_periodic_start (loop, &hourly_tick);
		2232
		2233
627	=head2 C<ev_signal> - signal me when a signal gets signalled	2234	=head2 C<ev_signal> - signal me when a signal gets signalled!
628		2235
629	Signal watchers will trigger an event when the process receives a specific	2236	Signal watchers will trigger an event when the process receives a specific
630	signal one or more times. Even though signals are very asynchronous, libev	2237	signal one or more times. Even though signals are very asynchronous, libev
631	will try it's best to deliver signals synchronously, i.e. as part of the	2238	will try it's best to deliver signals synchronously, i.e. as part of the
632	normal event processing, like any other event.	2239	normal event processing, like any other event.
633		2240
		2241	If you want signals to be delivered truly asynchronously, just use
		2242	C<sigaction> as you would do without libev and forget about sharing
		2243	the signal. You can even use C<ev_async> from a signal handler to
		2244	synchronously wake up an event loop.
		2245
634	You can configure as many watchers as you like per signal. Only when the	2246	You can configure as many watchers as you like for the same signal, but
		2247	only within the same loop, i.e. you can watch for C<SIGINT> in your
		2248	default loop and for C<SIGIO> in another loop, but you cannot watch for
		2249	C<SIGINT> in both the default loop and another loop at the same time. At
		2250	the moment, C<SIGCHLD> is permanently tied to the default loop.
		2251
635	first watcher gets started will libev actually register a signal watcher	2252	When the first watcher gets started will libev actually register something
636	with the kernel (thus it coexists with your own signal handlers as long	2253	with the kernel (thus it coexists with your own signal handlers as long as
637	as you don't register any with libev). Similarly, when the last signal	2254	you don't register any with libev for the same signal).
638	watcher for a signal is stopped libev will reset the signal handler to	2255
639	SIG_DFL (regardless of what it was set to before).	2256	If possible and supported, libev will install its handlers with
		2257	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
		2258	not be unduly interrupted. If you have a problem with system calls getting
		2259	interrupted by signals you can block all signals in an C<ev_check> watcher
		2260	and unblock them in an C<ev_prepare> watcher.
		2261
		2262	=head3 The special problem of inheritance over fork/execve/pthread_create
		2263
		2264	Both the signal mask (C<sigprocmask>) and the signal disposition
		2265	(C<sigaction>) are unspecified after starting a signal watcher (and after
		2266	stopping it again), that is, libev might or might not block the signal,
		2267	and might or might not set or restore the installed signal handler.
		2268
		2269	While this does not matter for the signal disposition (libev never
		2270	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
		2271	C<execve>), this matters for the signal mask: many programs do not expect
		2272	certain signals to be blocked.
		2273
		2274	This means that before calling C<exec> (from the child) you should reset
		2275	the signal mask to whatever "default" you expect (all clear is a good
		2276	choice usually).
		2277
		2278	The simplest way to ensure that the signal mask is reset in the child is
		2279	to install a fork handler with C<pthread_atfork> that resets it. That will
		2280	catch fork calls done by libraries (such as the libc) as well.
		2281
		2282	In current versions of libev, the signal will not be blocked indefinitely
		2283	unless you use the C<signalfd> API (C<EV_SIGNALFD>). While this reduces
		2284	the window of opportunity for problems, it will not go away, as libev
		2285	I<has> to modify the signal mask, at least temporarily.
		2286
		2287	So I can't stress this enough: I<If you do not reset your signal mask when
		2288	you expect it to be empty, you have a race condition in your code>. This
		2289	is not a libev-specific thing, this is true for most event libraries.
		2290
		2291	=head3 Watcher-Specific Functions and Data Members
640		2292
641	=over 4	2293	=over 4
642		2294
643	=item ev_signal_init (ev_signal *, callback, int signum)	2295	=item ev_signal_init (ev_signal *, callback, int signum)
644		2296
645	=item ev_signal_set (ev_signal *, int signum)	2297	=item ev_signal_set (ev_signal *, int signum)
646		2298
647	Configures the watcher to trigger on the given signal number (usually one	2299	Configures the watcher to trigger on the given signal number (usually one
648	of the C<SIGxxx> constants).	2300	of the C<SIGxxx> constants).
649		2301
		2302	=item int signum [read-only]
		2303
		2304	The signal the watcher watches out for.
		2305
650	=back	2306	=back
651		2307
		2308	=head3 Examples
		2309
		2310	Example: Try to exit cleanly on SIGINT.
		2311
		2312	static void
		2313	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		2314	{
		2315	ev_break (loop, EVBREAK_ALL);
		2316	}
		2317
		2318	ev_signal signal_watcher;
		2319	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		2320	ev_signal_start (loop, &signal_watcher);
		2321
		2322
652	=head2 C<ev_child> - wait for pid status changes	2323	=head2 C<ev_child> - watch out for process status changes
653		2324
654	Child watchers trigger when your process receives a SIGCHLD in response to	2325	Child watchers trigger when your process receives a SIGCHLD in response to
655	some child status changes (most typically when a child of yours dies).	2326	some child status changes (most typically when a child of yours dies or
		2327	exits). It is permissible to install a child watcher I<after> the child
		2328	has been forked (which implies it might have already exited), as long
		2329	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2330	forking and then immediately registering a watcher for the child is fine,
		2331	but forking and registering a watcher a few event loop iterations later or
		2332	in the next callback invocation is not.
		2333
		2334	Only the default event loop is capable of handling signals, and therefore
		2335	you can only register child watchers in the default event loop.
		2336
		2337	Due to some design glitches inside libev, child watchers will always be
		2338	handled at maximum priority (their priority is set to C<EV_MAXPRI> by
		2339	libev)
		2340
		2341	=head3 Process Interaction
		2342
		2343	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2344	initialised. This is necessary to guarantee proper behaviour even if the
		2345	first child watcher is started after the child exits. The occurrence
		2346	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2347	synchronously as part of the event loop processing. Libev always reaps all
		2348	children, even ones not watched.
		2349
		2350	=head3 Overriding the Built-In Processing
		2351
		2352	Libev offers no special support for overriding the built-in child
		2353	processing, but if your application collides with libev's default child
		2354	handler, you can override it easily by installing your own handler for
		2355	C<SIGCHLD> after initialising the default loop, and making sure the
		2356	default loop never gets destroyed. You are encouraged, however, to use an
		2357	event-based approach to child reaping and thus use libev's support for
		2358	that, so other libev users can use C<ev_child> watchers freely.
		2359
		2360	=head3 Stopping the Child Watcher
		2361
		2362	Currently, the child watcher never gets stopped, even when the
		2363	child terminates, so normally one needs to stop the watcher in the
		2364	callback. Future versions of libev might stop the watcher automatically
		2365	when a child exit is detected (calling C<ev_child_stop> twice is not a
		2366	problem).
		2367
		2368	=head3 Watcher-Specific Functions and Data Members
656		2369
657	=over 4	2370	=over 4
658		2371
659	=item ev_child_init (ev_child *, callback, int pid)	2372	=item ev_child_init (ev_child *, callback, int pid, int trace)
660		2373
661	=item ev_child_set (ev_child *, int pid)	2374	=item ev_child_set (ev_child *, int pid, int trace)
662		2375
663	Configures the watcher to wait for status changes of process C<pid> (or	2376	Configures the watcher to wait for status changes of process C<pid> (or
664	I<any> process if C<pid> is specified as C<0>). The callback can look	2377	I<any> process if C<pid> is specified as C<0>). The callback can look
665	at the C<rstatus> member of the C<ev_child> watcher structure to see	2378	at the C<rstatus> member of the C<ev_child> watcher structure to see
666	the status word (use the macros from C<sys/wait.h> and see your systems	2379	the status word (use the macros from C<sys/wait.h> and see your systems
667	C<waitpid> documentation). The C<rpid> member contains the pid of the	2380	C<waitpid> documentation). The C<rpid> member contains the pid of the
668	process causing the status change.	2381	process causing the status change. C<trace> must be either C<0> (only
		2382	activate the watcher when the process terminates) or C<1> (additionally
		2383	activate the watcher when the process is stopped or continued).
		2384
		2385	=item int pid [read-only]
		2386
		2387	The process id this watcher watches out for, or C<0>, meaning any process id.
		2388
		2389	=item int rpid [read-write]
		2390
		2391	The process id that detected a status change.
		2392
		2393	=item int rstatus [read-write]
		2394
		2395	The process exit/trace status caused by C<rpid> (see your systems
		2396	C<waitpid> and C<sys/wait.h> documentation for details).
669		2397
670	=back	2398	=back
671		2399
		2400	=head3 Examples
		2401
		2402	Example: C<fork()> a new process and install a child handler to wait for
		2403	its completion.
		2404
		2405	ev_child cw;
		2406
		2407	static void
		2408	child_cb (EV_P_ ev_child *w, int revents)
		2409	{
		2410	ev_child_stop (EV_A_ w);
		2411	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2412	}
		2413
		2414	pid_t pid = fork ();
		2415
		2416	if (pid < 0)
		2417	// error
		2418	else if (pid == 0)
		2419	{
		2420	// the forked child executes here
		2421	exit (1);
		2422	}
		2423	else
		2424	{
		2425	ev_child_init (&cw, child_cb, pid, 0);
		2426	ev_child_start (EV_DEFAULT_ &cw);
		2427	}
		2428
		2429
		2430	=head2 C<ev_stat> - did the file attributes just change?
		2431
		2432	This watches a file system path for attribute changes. That is, it calls
		2433	C<stat> on that path in regular intervals (or when the OS says it changed)
		2434	and sees if it changed compared to the last time, invoking the callback if
		2435	it did.
		2436
		2437	The path does not need to exist: changing from "path exists" to "path does
		2438	not exist" is a status change like any other. The condition "path does not
		2439	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2440	C<st_nlink> field being zero (which is otherwise always forced to be at
		2441	least one) and all the other fields of the stat buffer having unspecified
		2442	contents.
		2443
		2444	The path I<must not> end in a slash or contain special components such as
		2445	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2446	your working directory changes, then the behaviour is undefined.
		2447
		2448	Since there is no portable change notification interface available, the
		2449	portable implementation simply calls C<stat(2)> regularly on the path
		2450	to see if it changed somehow. You can specify a recommended polling
		2451	interval for this case. If you specify a polling interval of C<0> (highly
		2452	recommended!) then a I<suitable, unspecified default> value will be used
		2453	(which you can expect to be around five seconds, although this might
		2454	change dynamically). Libev will also impose a minimum interval which is
		2455	currently around C<0.1>, but that's usually overkill.
		2456
		2457	This watcher type is not meant for massive numbers of stat watchers,
		2458	as even with OS-supported change notifications, this can be
		2459	resource-intensive.
		2460
		2461	At the time of this writing, the only OS-specific interface implemented
		2462	is the Linux inotify interface (implementing kqueue support is left as an
		2463	exercise for the reader. Note, however, that the author sees no way of
		2464	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2465
		2466	=head3 ABI Issues (Largefile Support)
		2467
		2468	Libev by default (unless the user overrides this) uses the default
		2469	compilation environment, which means that on systems with large file
		2470	support disabled by default, you get the 32 bit version of the stat
		2471	structure. When using the library from programs that change the ABI to
		2472	use 64 bit file offsets the programs will fail. In that case you have to
		2473	compile libev with the same flags to get binary compatibility. This is
		2474	obviously the case with any flags that change the ABI, but the problem is
		2475	most noticeably displayed with ev_stat and large file support.
		2476
		2477	The solution for this is to lobby your distribution maker to make large
		2478	file interfaces available by default (as e.g. FreeBSD does) and not
		2479	optional. Libev cannot simply switch on large file support because it has
		2480	to exchange stat structures with application programs compiled using the
		2481	default compilation environment.
		2482
		2483	=head3 Inotify and Kqueue
		2484
		2485	When C<inotify (7)> support has been compiled into libev and present at
		2486	runtime, it will be used to speed up change detection where possible. The
		2487	inotify descriptor will be created lazily when the first C<ev_stat>
		2488	watcher is being started.
		2489
		2490	Inotify presence does not change the semantics of C<ev_stat> watchers
		2491	except that changes might be detected earlier, and in some cases, to avoid
		2492	making regular C<stat> calls. Even in the presence of inotify support
		2493	there are many cases where libev has to resort to regular C<stat> polling,
		2494	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2495	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2496	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2497	xfs are fully working) libev usually gets away without polling.
		2498
		2499	There is no support for kqueue, as apparently it cannot be used to
		2500	implement this functionality, due to the requirement of having a file
		2501	descriptor open on the object at all times, and detecting renames, unlinks
		2502	etc. is difficult.
		2503
		2504	=head3 C<stat ()> is a synchronous operation
		2505
		2506	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2507	the process. The exception are C<ev_stat> watchers - those call C<stat
		2508	()>, which is a synchronous operation.
		2509
		2510	For local paths, this usually doesn't matter: unless the system is very
		2511	busy or the intervals between stat's are large, a stat call will be fast,
		2512	as the path data is usually in memory already (except when starting the
		2513	watcher).
		2514
		2515	For networked file systems, calling C<stat ()> can block an indefinite
		2516	time due to network issues, and even under good conditions, a stat call
		2517	often takes multiple milliseconds.
		2518
		2519	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2520	paths, although this is fully supported by libev.
		2521
		2522	=head3 The special problem of stat time resolution
		2523
		2524	The C<stat ()> system call only supports full-second resolution portably,
		2525	and even on systems where the resolution is higher, most file systems
		2526	still only support whole seconds.
		2527
		2528	That means that, if the time is the only thing that changes, you can
		2529	easily miss updates: on the first update, C<ev_stat> detects a change and
		2530	calls your callback, which does something. When there is another update
		2531	within the same second, C<ev_stat> will be unable to detect unless the
		2532	stat data does change in other ways (e.g. file size).
		2533
		2534	The solution to this is to delay acting on a change for slightly more
		2535	than a second (or till slightly after the next full second boundary), using
		2536	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2537	ev_timer_again (loop, w)>).
		2538
		2539	The C<.02> offset is added to work around small timing inconsistencies
		2540	of some operating systems (where the second counter of the current time
		2541	might be be delayed. One such system is the Linux kernel, where a call to
		2542	C<gettimeofday> might return a timestamp with a full second later than
		2543	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2544	update file times then there will be a small window where the kernel uses
		2545	the previous second to update file times but libev might already execute
		2546	the timer callback).
		2547
		2548	=head3 Watcher-Specific Functions and Data Members
		2549
		2550	=over 4
		2551
		2552	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2553
		2554	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2555
		2556	Configures the watcher to wait for status changes of the given
		2557	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2558	be detected and should normally be specified as C<0> to let libev choose
		2559	a suitable value. The memory pointed to by C<path> must point to the same
		2560	path for as long as the watcher is active.
		2561
		2562	The callback will receive an C<EV_STAT> event when a change was detected,
		2563	relative to the attributes at the time the watcher was started (or the
		2564	last change was detected).
		2565
		2566	=item ev_stat_stat (loop, ev_stat *)
		2567
		2568	Updates the stat buffer immediately with new values. If you change the
		2569	watched path in your callback, you could call this function to avoid
		2570	detecting this change (while introducing a race condition if you are not
		2571	the only one changing the path). Can also be useful simply to find out the
		2572	new values.
		2573
		2574	=item ev_statdata attr [read-only]
		2575
		2576	The most-recently detected attributes of the file. Although the type is
		2577	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2578	suitable for your system, but you can only rely on the POSIX-standardised
		2579	members to be present. If the C<st_nlink> member is C<0>, then there was
		2580	some error while C<stat>ing the file.
		2581
		2582	=item ev_statdata prev [read-only]
		2583
		2584	The previous attributes of the file. The callback gets invoked whenever
		2585	C<prev> != C<attr>, or, more precisely, one or more of these members
		2586	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2587	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2588
		2589	=item ev_tstamp interval [read-only]
		2590
		2591	The specified interval.
		2592
		2593	=item const char *path [read-only]
		2594
		2595	The file system path that is being watched.
		2596
		2597	=back
		2598
		2599	=head3 Examples
		2600
		2601	Example: Watch C</etc/passwd> for attribute changes.
		2602
		2603	static void
		2604	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2605	{
		2606	/* /etc/passwd changed in some way */
		2607	if (w->attr.st_nlink)
		2608	{
		2609	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2610	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2611	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2612	}
		2613	else
		2614	/* you shalt not abuse printf for puts */
		2615	puts ("wow, /etc/passwd is not there, expect problems. "
		2616	"if this is windows, they already arrived\n");
		2617	}
		2618
		2619	...
		2620	ev_stat passwd;
		2621
		2622	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2623	ev_stat_start (loop, &passwd);
		2624
		2625	Example: Like above, but additionally use a one-second delay so we do not
		2626	miss updates (however, frequent updates will delay processing, too, so
		2627	one might do the work both on C<ev_stat> callback invocation I<and> on
		2628	C<ev_timer> callback invocation).
		2629
		2630	static ev_stat passwd;
		2631	static ev_timer timer;
		2632
		2633	static void
		2634	timer_cb (EV_P_ ev_timer *w, int revents)
		2635	{
		2636	ev_timer_stop (EV_A_ w);
		2637
		2638	/* now it's one second after the most recent passwd change */
		2639	}
		2640
		2641	static void
		2642	stat_cb (EV_P_ ev_stat *w, int revents)
		2643	{
		2644	/* reset the one-second timer */
		2645	ev_timer_again (EV_A_ &timer);
		2646	}
		2647
		2648	...
		2649	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2650	ev_stat_start (loop, &passwd);
		2651	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2652
		2653
672	=head2 C<ev_idle> - when you've got nothing better to do	2654	=head2 C<ev_idle> - when you've got nothing better to do...
673		2655
674	Idle watchers trigger events when there are no other events are pending	2656	Idle watchers trigger events when no other events of the same or higher
675	(prepare, check and other idle watchers do not count). That is, as long	2657	priority are pending (prepare, check and other idle watchers do not count
676	as your process is busy handling sockets or timeouts (or even signals,	2658	as receiving "events").
677	imagine) it will not be triggered. But when your process is idle all idle	2659
678	watchers are being called again and again, once per event loop iteration -	2660	That is, as long as your process is busy handling sockets or timeouts
		2661	(or even signals, imagine) of the same or higher priority it will not be
		2662	triggered. But when your process is idle (or only lower-priority watchers
		2663	are pending), the idle watchers are being called once per event loop
679	until stopped, that is, or your process receives more events and becomes	2664	iteration - until stopped, that is, or your process receives more events
680	busy.	2665	and becomes busy again with higher priority stuff.
681		2666
682	The most noteworthy effect is that as long as any idle watchers are	2667	The most noteworthy effect is that as long as any idle watchers are
683	active, the process will not block when waiting for new events.	2668	active, the process will not block when waiting for new events.
684		2669
685	Apart from keeping your process non-blocking (which is a useful	2670	Apart from keeping your process non-blocking (which is a useful
686	effect on its own sometimes), idle watchers are a good place to do	2671	effect on its own sometimes), idle watchers are a good place to do
687	"pseudo-background processing", or delay processing stuff to after the	2672	"pseudo-background processing", or delay processing stuff to after the
688	event loop has handled all outstanding events.	2673	event loop has handled all outstanding events.
689		2674
		2675	=head3 Watcher-Specific Functions and Data Members
		2676
690	=over 4	2677	=over 4
691		2678
692	=item ev_idle_init (ev_signal *, callback)	2679	=item ev_idle_init (ev_idle *, callback)
693		2680
694	Initialises and configures the idle watcher - it has no parameters of any	2681	Initialises and configures the idle watcher - it has no parameters of any
695	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2682	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
696	believe me.	2683	believe me.
697		2684
698	=back	2685	=back
699		2686
		2687	=head3 Examples
		2688
		2689	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2690	callback, free it. Also, use no error checking, as usual.
		2691
		2692	static void
		2693	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2694	{
		2695	free (w);
		2696	// now do something you wanted to do when the program has
		2697	// no longer anything immediate to do.
		2698	}
		2699
		2700	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2701	ev_idle_init (idle_watcher, idle_cb);
		2702	ev_idle_start (loop, idle_watcher);
		2703
		2704
700	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2705	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
701		2706
702	Prepare and check watchers are usually (but not always) used in tandem:	2707	Prepare and check watchers are usually (but not always) used in pairs:
703	prepare watchers get invoked before the process blocks and check watchers	2708	prepare watchers get invoked before the process blocks and check watchers
704	afterwards.	2709	afterwards.
705		2710
		2711	You I<must not> call C<ev_run> or similar functions that enter
		2712	the current event loop from either C<ev_prepare> or C<ev_check>
		2713	watchers. Other loops than the current one are fine, however. The
		2714	rationale behind this is that you do not need to check for recursion in
		2715	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2716	C<ev_check> so if you have one watcher of each kind they will always be
		2717	called in pairs bracketing the blocking call.
		2718
706	Their main purpose is to integrate other event mechanisms into libev. This	2719	Their main purpose is to integrate other event mechanisms into libev and
707	could be used, for example, to track variable changes, implement your own	2720	their use is somewhat advanced. They could be used, for example, to track
708	watchers, integrate net-snmp or a coroutine library and lots more.	2721	variable changes, implement your own watchers, integrate net-snmp or a
		2722	coroutine library and lots more. They are also occasionally useful if
		2723	you cache some data and want to flush it before blocking (for example,
		2724	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2725	watcher).
709		2726
710	This is done by examining in each prepare call which file descriptors need	2727	This is done by examining in each prepare call which file descriptors
711	to be watched by the other library, registering C<ev_io> watchers for	2728	need to be watched by the other library, registering C<ev_io> watchers
712	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2729	for them and starting an C<ev_timer> watcher for any timeouts (many
713	provide just this functionality). Then, in the check watcher you check for	2730	libraries provide exactly this functionality). Then, in the check watcher,
714	any events that occured (by checking the pending status of all watchers	2731	you check for any events that occurred (by checking the pending status
715	and stopping them) and call back into the library. The I/O and timer	2732	of all watchers and stopping them) and call back into the library. The
716	callbacks will never actually be called (but must be valid nevertheless,	2733	I/O and timer callbacks will never actually be called (but must be valid
717	because you never know, you know?).	2734	nevertheless, because you never know, you know?).
718		2735
719	As another example, the Perl Coro module uses these hooks to integrate	2736	As another example, the Perl Coro module uses these hooks to integrate
720	coroutines into libev programs, by yielding to other active coroutines	2737	coroutines into libev programs, by yielding to other active coroutines
721	during each prepare and only letting the process block if no coroutines	2738	during each prepare and only letting the process block if no coroutines
722	are ready to run (it's actually more complicated: it only runs coroutines	2739	are ready to run (it's actually more complicated: it only runs coroutines
723	with priority higher than or equal to the event loop and one coroutine	2740	with priority higher than or equal to the event loop and one coroutine
724	of lower priority, but only once, using idle watchers to keep the event	2741	of lower priority, but only once, using idle watchers to keep the event
725	loop from blocking if lower-priority coroutines are active, thus mapping	2742	loop from blocking if lower-priority coroutines are active, thus mapping
726	low-priority coroutines to idle/background tasks).	2743	low-priority coroutines to idle/background tasks).
727		2744
		2745	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2746	priority, to ensure that they are being run before any other watchers
		2747	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2748
		2749	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2750	activate ("feed") events into libev. While libev fully supports this, they
		2751	might get executed before other C<ev_check> watchers did their job. As
		2752	C<ev_check> watchers are often used to embed other (non-libev) event
		2753	loops those other event loops might be in an unusable state until their
		2754	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2755	others).
		2756
		2757	=head3 Watcher-Specific Functions and Data Members
		2758
728	=over 4	2759	=over 4
729		2760
730	=item ev_prepare_init (ev_prepare *, callback)	2761	=item ev_prepare_init (ev_prepare *, callback)
731		2762
732	=item ev_check_init (ev_check *, callback)	2763	=item ev_check_init (ev_check *, callback)
733		2764
734	Initialises and configures the prepare or check watcher - they have no	2765	Initialises and configures the prepare or check watcher - they have no
735	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2766	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
736	macros, but using them is utterly, utterly and completely pointless.	2767	macros, but using them is utterly, utterly, utterly and completely
		2768	pointless.
737		2769
738	=back	2770	=back
739		2771
		2772	=head3 Examples
		2773
		2774	There are a number of principal ways to embed other event loops or modules
		2775	into libev. Here are some ideas on how to include libadns into libev
		2776	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2777	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2778	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2779	Glib event loop).
		2780
		2781	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2782	and in a check watcher, destroy them and call into libadns. What follows
		2783	is pseudo-code only of course. This requires you to either use a low
		2784	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2785	the callbacks for the IO/timeout watchers might not have been called yet.
		2786
		2787	static ev_io iow [nfd];
		2788	static ev_timer tw;
		2789
		2790	static void
		2791	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2792	{
		2793	}
		2794
		2795	// create io watchers for each fd and a timer before blocking
		2796	static void
		2797	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2798	{
		2799	int timeout = 3600000;
		2800	struct pollfd fds [nfd];
		2801	// actual code will need to loop here and realloc etc.
		2802	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2803
		2804	/* the callback is illegal, but won't be called as we stop during check */
		2805	ev_timer_init (&tw, 0, timeout * 1e-3, 0.);
		2806	ev_timer_start (loop, &tw);
		2807
		2808	// create one ev_io per pollfd
		2809	for (int i = 0; i < nfd; ++i)
		2810	{
		2811	ev_io_init (iow + i, io_cb, fds [i].fd,
		2812	((fds [i].events & POLLIN ? EV_READ : 0)
		2813	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2814
		2815	fds [i].revents = 0;
		2816	ev_io_start (loop, iow + i);
		2817	}
		2818	}
		2819
		2820	// stop all watchers after blocking
		2821	static void
		2822	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2823	{
		2824	ev_timer_stop (loop, &tw);
		2825
		2826	for (int i = 0; i < nfd; ++i)
		2827	{
		2828	// set the relevant poll flags
		2829	// could also call adns_processreadable etc. here
		2830	struct pollfd *fd = fds + i;
		2831	int revents = ev_clear_pending (iow + i);
		2832	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2833	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2834
		2835	// now stop the watcher
		2836	ev_io_stop (loop, iow + i);
		2837	}
		2838
		2839	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2840	}
		2841
		2842	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2843	in the prepare watcher and would dispose of the check watcher.
		2844
		2845	Method 3: If the module to be embedded supports explicit event
		2846	notification (libadns does), you can also make use of the actual watcher
		2847	callbacks, and only destroy/create the watchers in the prepare watcher.
		2848
		2849	static void
		2850	timer_cb (EV_P_ ev_timer *w, int revents)
		2851	{
		2852	adns_state ads = (adns_state)w->data;
		2853	update_now (EV_A);
		2854
		2855	adns_processtimeouts (ads, &tv_now);
		2856	}
		2857
		2858	static void
		2859	io_cb (EV_P_ ev_io *w, int revents)
		2860	{
		2861	adns_state ads = (adns_state)w->data;
		2862	update_now (EV_A);
		2863
		2864	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2865	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2866	}
		2867
		2868	// do not ever call adns_afterpoll
		2869
		2870	Method 4: Do not use a prepare or check watcher because the module you
		2871	want to embed is not flexible enough to support it. Instead, you can
		2872	override their poll function. The drawback with this solution is that the
		2873	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2874	this approach, effectively embedding EV as a client into the horrible
		2875	libglib event loop.
		2876
		2877	static gint
		2878	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2879	{
		2880	int got_events = 0;
		2881
		2882	for (n = 0; n < nfds; ++n)
		2883	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2884
		2885	if (timeout >= 0)
		2886	// create/start timer
		2887
		2888	// poll
		2889	ev_run (EV_A_ 0);
		2890
		2891	// stop timer again
		2892	if (timeout >= 0)
		2893	ev_timer_stop (EV_A_ &to);
		2894
		2895	// stop io watchers again - their callbacks should have set
		2896	for (n = 0; n < nfds; ++n)
		2897	ev_io_stop (EV_A_ iow [n]);
		2898
		2899	return got_events;
		2900	}
		2901
		2902
		2903	=head2 C<ev_embed> - when one backend isn't enough...
		2904
		2905	This is a rather advanced watcher type that lets you embed one event loop
		2906	into another (currently only C<ev_io> events are supported in the embedded
		2907	loop, other types of watchers might be handled in a delayed or incorrect
		2908	fashion and must not be used).
		2909
		2910	There are primarily two reasons you would want that: work around bugs and
		2911	prioritise I/O.
		2912
		2913	As an example for a bug workaround, the kqueue backend might only support
		2914	sockets on some platform, so it is unusable as generic backend, but you
		2915	still want to make use of it because you have many sockets and it scales
		2916	so nicely. In this case, you would create a kqueue-based loop and embed
		2917	it into your default loop (which might use e.g. poll). Overall operation
		2918	will be a bit slower because first libev has to call C<poll> and then
		2919	C<kevent>, but at least you can use both mechanisms for what they are
		2920	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2921
		2922	As for prioritising I/O: under rare circumstances you have the case where
		2923	some fds have to be watched and handled very quickly (with low latency),
		2924	and even priorities and idle watchers might have too much overhead. In
		2925	this case you would put all the high priority stuff in one loop and all
		2926	the rest in a second one, and embed the second one in the first.
		2927
		2928	As long as the watcher is active, the callback will be invoked every
		2929	time there might be events pending in the embedded loop. The callback
		2930	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		2931	sweep and invoke their callbacks (the callback doesn't need to invoke the
		2932	C<ev_embed_sweep> function directly, it could also start an idle watcher
		2933	to give the embedded loop strictly lower priority for example).
		2934
		2935	You can also set the callback to C<0>, in which case the embed watcher
		2936	will automatically execute the embedded loop sweep whenever necessary.
		2937
		2938	Fork detection will be handled transparently while the C<ev_embed> watcher
		2939	is active, i.e., the embedded loop will automatically be forked when the
		2940	embedding loop forks. In other cases, the user is responsible for calling
		2941	C<ev_loop_fork> on the embedded loop.
		2942
		2943	Unfortunately, not all backends are embeddable: only the ones returned by
		2944	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2945	portable one.
		2946
		2947	So when you want to use this feature you will always have to be prepared
		2948	that you cannot get an embeddable loop. The recommended way to get around
		2949	this is to have a separate variables for your embeddable loop, try to
		2950	create it, and if that fails, use the normal loop for everything.
		2951
		2952	=head3 C<ev_embed> and fork
		2953
		2954	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2955	automatically be applied to the embedded loop as well, so no special
		2956	fork handling is required in that case. When the watcher is not running,
		2957	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2958	as applicable.
		2959
		2960	=head3 Watcher-Specific Functions and Data Members
		2961
		2962	=over 4
		2963
		2964	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2965
		2966	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2967
		2968	Configures the watcher to embed the given loop, which must be
		2969	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2970	invoked automatically, otherwise it is the responsibility of the callback
		2971	to invoke it (it will continue to be called until the sweep has been done,
		2972	if you do not want that, you need to temporarily stop the embed watcher).
		2973
		2974	=item ev_embed_sweep (loop, ev_embed *)
		2975
		2976	Make a single, non-blocking sweep over the embedded loop. This works
		2977	similarly to C<ev_run (embedded_loop, EVRUN_NOWAIT)>, but in the most
		2978	appropriate way for embedded loops.
		2979
		2980	=item struct ev_loop *other [read-only]
		2981
		2982	The embedded event loop.
		2983
		2984	=back
		2985
		2986	=head3 Examples
		2987
		2988	Example: Try to get an embeddable event loop and embed it into the default
		2989	event loop. If that is not possible, use the default loop. The default
		2990	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2991	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2992	used).
		2993
		2994	struct ev_loop *loop_hi = ev_default_init (0);
		2995	struct ev_loop *loop_lo = 0;
		2996	ev_embed embed;
		2997
		2998	// see if there is a chance of getting one that works
		2999	// (remember that a flags value of 0 means autodetection)
		3000	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		3001	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		3002	: 0;
		3003
		3004	// if we got one, then embed it, otherwise default to loop_hi
		3005	if (loop_lo)
		3006	{
		3007	ev_embed_init (&embed, 0, loop_lo);
		3008	ev_embed_start (loop_hi, &embed);
		3009	}
		3010	else
		3011	loop_lo = loop_hi;
		3012
		3013	Example: Check if kqueue is available but not recommended and create
		3014	a kqueue backend for use with sockets (which usually work with any
		3015	kqueue implementation). Store the kqueue/socket-only event loop in
		3016	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		3017
		3018	struct ev_loop *loop = ev_default_init (0);
		3019	struct ev_loop *loop_socket = 0;
		3020	ev_embed embed;
		3021
		3022	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		3023	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		3024	{
		3025	ev_embed_init (&embed, 0, loop_socket);
		3026	ev_embed_start (loop, &embed);
		3027	}
		3028
		3029	if (!loop_socket)
		3030	loop_socket = loop;
		3031
		3032	// now use loop_socket for all sockets, and loop for everything else
		3033
		3034
		3035	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		3036
		3037	Fork watchers are called when a C<fork ()> was detected (usually because
		3038	whoever is a good citizen cared to tell libev about it by calling
		3039	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		3040	event loop blocks next and before C<ev_check> watchers are being called,
		3041	and only in the child after the fork. If whoever good citizen calling
		3042	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		3043	handlers will be invoked, too, of course.
		3044
		3045	=head3 The special problem of life after fork - how is it possible?
		3046
		3047	Most uses of C<fork()> consist of forking, then some simple calls to set
		3048	up/change the process environment, followed by a call to C<exec()>. This
		3049	sequence should be handled by libev without any problems.
		3050
		3051	This changes when the application actually wants to do event handling
		3052	in the child, or both parent in child, in effect "continuing" after the
		3053	fork.
		3054
		3055	The default mode of operation (for libev, with application help to detect
		3056	forks) is to duplicate all the state in the child, as would be expected
		3057	when I<either> the parent I<or> the child process continues.
		3058
		3059	When both processes want to continue using libev, then this is usually the
		3060	wrong result. In that case, usually one process (typically the parent) is
		3061	supposed to continue with all watchers in place as before, while the other
		3062	process typically wants to start fresh, i.e. without any active watchers.
		3063
		3064	The cleanest and most efficient way to achieve that with libev is to
		3065	simply create a new event loop, which of course will be "empty", and
		3066	use that for new watchers. This has the advantage of not touching more
		3067	memory than necessary, and thus avoiding the copy-on-write, and the
		3068	disadvantage of having to use multiple event loops (which do not support
		3069	signal watchers).
		3070
		3071	When this is not possible, or you want to use the default loop for
		3072	other reasons, then in the process that wants to start "fresh", call
		3073	C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying
		3074	the default loop will "orphan" (not stop) all registered watchers, so you
		3075	have to be careful not to execute code that modifies those watchers. Note
		3076	also that in that case, you have to re-register any signal watchers.
		3077
		3078	=head3 Watcher-Specific Functions and Data Members
		3079
		3080	=over 4
		3081
		3082	=item ev_fork_init (ev_signal *, callback)
		3083
		3084	Initialises and configures the fork watcher - it has no parameters of any
		3085	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		3086	believe me.
		3087
		3088	=back
		3089
		3090
		3091	=head2 C<ev_async> - how to wake up an event loop
		3092
		3093	In general, you cannot use an C<ev_run> from multiple threads or other
		3094	asynchronous sources such as signal handlers (as opposed to multiple event
		3095	loops - those are of course safe to use in different threads).
		3096
		3097	Sometimes, however, you need to wake up an event loop you do not control,
		3098	for example because it belongs to another thread. This is what C<ev_async>
		3099	watchers do: as long as the C<ev_async> watcher is active, you can signal
		3100	it by calling C<ev_async_send>, which is thread- and signal safe.
		3101
		3102	This functionality is very similar to C<ev_signal> watchers, as signals,
		3103	too, are asynchronous in nature, and signals, too, will be compressed
		3104	(i.e. the number of callback invocations may be less than the number of
		3105	C<ev_async_sent> calls).
		3106
		3107	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		3108	just the default loop.
		3109
		3110	=head3 Queueing
		3111
		3112	C<ev_async> does not support queueing of data in any way. The reason
		3113	is that the author does not know of a simple (or any) algorithm for a
		3114	multiple-writer-single-reader queue that works in all cases and doesn't
		3115	need elaborate support such as pthreads or unportable memory access
		3116	semantics.
		3117
		3118	That means that if you want to queue data, you have to provide your own
		3119	queue. But at least I can tell you how to implement locking around your
		3120	queue:
		3121
		3122	=over 4
		3123
		3124	=item queueing from a signal handler context
		3125
		3126	To implement race-free queueing, you simply add to the queue in the signal
		3127	handler but you block the signal handler in the watcher callback. Here is
		3128	an example that does that for some fictitious SIGUSR1 handler:
		3129
		3130	static ev_async mysig;
		3131
		3132	static void
		3133	sigusr1_handler (void)
		3134	{
		3135	sometype data;
		3136
		3137	// no locking etc.
		3138	queue_put (data);
		3139	ev_async_send (EV_DEFAULT_ &mysig);
		3140	}
		3141
		3142	static void
		3143	mysig_cb (EV_P_ ev_async *w, int revents)
		3144	{
		3145	sometype data;
		3146	sigset_t block, prev;
		3147
		3148	sigemptyset (&block);
		3149	sigaddset (&block, SIGUSR1);
		3150	sigprocmask (SIG_BLOCK, &block, &prev);
		3151
		3152	while (queue_get (&data))
		3153	process (data);
		3154
		3155	if (sigismember (&prev, SIGUSR1)
		3156	sigprocmask (SIG_UNBLOCK, &block, 0);
		3157	}
		3158
		3159	(Note: pthreads in theory requires you to use C<pthread_setmask>
		3160	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		3161	either...).
		3162
		3163	=item queueing from a thread context
		3164
		3165	The strategy for threads is different, as you cannot (easily) block
		3166	threads but you can easily preempt them, so to queue safely you need to
		3167	employ a traditional mutex lock, such as in this pthread example:
		3168
		3169	static ev_async mysig;
		3170	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		3171
		3172	static void
		3173	otherthread (void)
		3174	{
		3175	// only need to lock the actual queueing operation
		3176	pthread_mutex_lock (&mymutex);
		3177	queue_put (data);
		3178	pthread_mutex_unlock (&mymutex);
		3179
		3180	ev_async_send (EV_DEFAULT_ &mysig);
		3181	}
		3182
		3183	static void
		3184	mysig_cb (EV_P_ ev_async *w, int revents)
		3185	{
		3186	pthread_mutex_lock (&mymutex);
		3187
		3188	while (queue_get (&data))
		3189	process (data);
		3190
		3191	pthread_mutex_unlock (&mymutex);
		3192	}
		3193
		3194	=back
		3195
		3196
		3197	=head3 Watcher-Specific Functions and Data Members
		3198
		3199	=over 4
		3200
		3201	=item ev_async_init (ev_async *, callback)
		3202
		3203	Initialises and configures the async watcher - it has no parameters of any
		3204	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		3205	trust me.
		3206
		3207	=item ev_async_send (loop, ev_async *)
		3208
		3209	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		3210	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		3211	C<ev_feed_event>, this call is safe to do from other threads, signal or
		3212	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		3213	section below on what exactly this means).
		3214
		3215	Note that, as with other watchers in libev, multiple events might get
		3216	compressed into a single callback invocation (another way to look at this
		3217	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		3218	reset when the event loop detects that).
		3219
		3220	This call incurs the overhead of a system call only once per event loop
		3221	iteration, so while the overhead might be noticeable, it doesn't apply to
		3222	repeated calls to C<ev_async_send> for the same event loop.
		3223
		3224	=item bool = ev_async_pending (ev_async *)
		3225
		3226	Returns a non-zero value when C<ev_async_send> has been called on the
		3227	watcher but the event has not yet been processed (or even noted) by the
		3228	event loop.
		3229
		3230	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		3231	the loop iterates next and checks for the watcher to have become active,
		3232	it will reset the flag again. C<ev_async_pending> can be used to very
		3233	quickly check whether invoking the loop might be a good idea.
		3234
		3235	Not that this does I<not> check whether the watcher itself is pending,
		3236	only whether it has been requested to make this watcher pending: there
		3237	is a time window between the event loop checking and resetting the async
		3238	notification, and the callback being invoked.
		3239
		3240	=back
		3241
		3242
740	=head1 OTHER FUNCTIONS	3243	=head1 OTHER FUNCTIONS
741		3244
742	There are some other functions of possible interest. Described. Here. Now.	3245	There are some other functions of possible interest. Described. Here. Now.
743		3246
744	=over 4	3247	=over 4
745		3248
746	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3249	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
747		3250
748	This function combines a simple timer and an I/O watcher, calls your	3251	This function combines a simple timer and an I/O watcher, calls your
749	callback on whichever event happens first and automatically stop both	3252	callback on whichever event happens first and automatically stops both
750	watchers. This is useful if you want to wait for a single event on an fd	3253	watchers. This is useful if you want to wait for a single event on an fd
751	or timeout without having to allocate/configure/start/stop/free one or	3254	or timeout without having to allocate/configure/start/stop/free one or
752	more watchers yourself.	3255	more watchers yourself.
753		3256
754	If C<fd> is less than 0, then no I/O watcher will be started and events	3257	If C<fd> is less than 0, then no I/O watcher will be started and the
755	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	3258	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
756	C<events> set will be craeted and started.	3259	the given C<fd> and C<events> set will be created and started.
757		3260
758	If C<timeout> is less than 0, then no timeout watcher will be	3261	If C<timeout> is less than 0, then no timeout watcher will be
759	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	3262	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
760	repeat = 0) will be started. While C<0> is a valid timeout, it is of	3263	repeat = 0) will be started. C<0> is a valid timeout.
761	dubious value.
762		3264
763	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	3265	The callback has the type C<void (*cb)(int revents, void *arg)> and is
764	passed an C<revents> set like normal event callbacks (a combination of	3266	passed an C<revents> set like normal event callbacks (a combination of
765	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	3267	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMER>) and the C<arg>
766	value passed to C<ev_once>:	3268	value passed to C<ev_once>. Note that it is possible to receive I<both>
		3269	a timeout and an io event at the same time - you probably should give io
		3270	events precedence.
767		3271
		3272	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		3273
768	static void stdin_ready (int revents, void *arg)	3274	static void stdin_ready (int revents, void *arg)
769	{	3275	{
770	if (revents & EV_TIMEOUT)
771	/* doh, nothing entered */;
772	else if (revents & EV_READ)	3276	if (revents & EV_READ)
773	/* stdin might have data for us, joy! */;	3277	/* stdin might have data for us, joy! */;
		3278	else if (revents & EV_TIMER)
		3279	/* doh, nothing entered */;
774	}	3280	}
775		3281
776	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3282	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
777
778	=item ev_feed_event (loop, watcher, int events)
779
780	Feeds the given event set into the event loop, as if the specified event
781	had happened for the specified watcher (which must be a pointer to an
782	initialised but not necessarily started event watcher).
783		3283
784	=item ev_feed_fd_event (loop, int fd, int revents)	3284	=item ev_feed_fd_event (loop, int fd, int revents)
785		3285
786	Feed an event on the given fd, as if a file descriptor backend detected	3286	Feed an event on the given fd, as if a file descriptor backend detected
787	the given events it.	3287	the given events it.
788		3288
789	=item ev_feed_signal_event (loop, int signum)	3289	=item ev_feed_signal_event (loop, int signum)
790		3290
791	Feed an event as if the given signal occured (loop must be the default loop!).	3291	Feed an event as if the given signal occurred (C<loop> must be the default
		3292	loop!).
792		3293
793	=back	3294	=back
		3295
794		3296
795	=head1 LIBEVENT EMULATION	3297	=head1 LIBEVENT EMULATION
796		3298
797	Libev offers a compatibility emulation layer for libevent. It cannot	3299	Libev offers a compatibility emulation layer for libevent. It cannot
798	emulate the internals of libevent, so here are some usage hints:	3300	emulate the internals of libevent, so here are some usage hints:
…		…
810		3312
811	=item * Priorities are not currently supported. Initialising priorities	3313	=item * Priorities are not currently supported. Initialising priorities
812	will fail and all watchers will have the same priority, even though there	3314	will fail and all watchers will have the same priority, even though there
813	is an ev_pri field.	3315	is an ev_pri field.
814		3316
		3317	=item * In libevent, the last base created gets the signals, in libev, the
		3318	first base created (== the default loop) gets the signals.
		3319
815	=item * Other members are not supported.	3320	=item * Other members are not supported.
816		3321
817	=item * The libev emulation is I<not> ABI compatible to libevent, you need	3322	=item * The libev emulation is I<not> ABI compatible to libevent, you need
818	to use the libev header file and library.	3323	to use the libev header file and library.
819		3324
820	=back	3325	=back
821		3326
822	=head1 C++ SUPPORT	3327	=head1 C++ SUPPORT
823		3328
824	TBD.	3329	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		3330	you to use some convenience methods to start/stop watchers and also change
		3331	the callback model to a model using method callbacks on objects.
		3332
		3333	To use it,
		3334
		3335	#include <ev++.h>
		3336
		3337	This automatically includes F<ev.h> and puts all of its definitions (many
		3338	of them macros) into the global namespace. All C++ specific things are
		3339	put into the C<ev> namespace. It should support all the same embedding
		3340	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		3341
		3342	Care has been taken to keep the overhead low. The only data member the C++
		3343	classes add (compared to plain C-style watchers) is the event loop pointer
		3344	that the watcher is associated with (or no additional members at all if
		3345	you disable C<EV_MULTIPLICITY> when embedding libev).
		3346
		3347	Currently, functions, and static and non-static member functions can be
		3348	used as callbacks. Other types should be easy to add as long as they only
		3349	need one additional pointer for context. If you need support for other
		3350	types of functors please contact the author (preferably after implementing
		3351	it).
		3352
		3353	Here is a list of things available in the C<ev> namespace:
		3354
		3355	=over 4
		3356
		3357	=item C<ev::READ>, C<ev::WRITE> etc.
		3358
		3359	These are just enum values with the same values as the C<EV_READ> etc.
		3360	macros from F<ev.h>.
		3361
		3362	=item C<ev::tstamp>, C<ev::now>
		3363
		3364	Aliases to the same types/functions as with the C<ev_> prefix.
		3365
		3366	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3367
		3368	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3369	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3370	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3371	defines by many implementations.
		3372
		3373	All of those classes have these methods:
		3374
		3375	=over 4
		3376
		3377	=item ev::TYPE::TYPE ()
		3378
		3379	=item ev::TYPE::TYPE (loop)
		3380
		3381	=item ev::TYPE::~TYPE
		3382
		3383	The constructor (optionally) takes an event loop to associate the watcher
		3384	with. If it is omitted, it will use C<EV_DEFAULT>.
		3385
		3386	The constructor calls C<ev_init> for you, which means you have to call the
		3387	C<set> method before starting it.
		3388
		3389	It will not set a callback, however: You have to call the templated C<set>
		3390	method to set a callback before you can start the watcher.
		3391
		3392	(The reason why you have to use a method is a limitation in C++ which does
		3393	not allow explicit template arguments for constructors).
		3394
		3395	The destructor automatically stops the watcher if it is active.
		3396
		3397	=item w->set<class, &class::method> (object *)
		3398
		3399	This method sets the callback method to call. The method has to have a
		3400	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3401	first argument and the C<revents> as second. The object must be given as
		3402	parameter and is stored in the C<data> member of the watcher.
		3403
		3404	This method synthesizes efficient thunking code to call your method from
		3405	the C callback that libev requires. If your compiler can inline your
		3406	callback (i.e. it is visible to it at the place of the C<set> call and
		3407	your compiler is good :), then the method will be fully inlined into the
		3408	thunking function, making it as fast as a direct C callback.
		3409
		3410	Example: simple class declaration and watcher initialisation
		3411
		3412	struct myclass
		3413	{
		3414	void io_cb (ev::io &w, int revents) { }
		3415	}
		3416
		3417	myclass obj;
		3418	ev::io iow;
		3419	iow.set <myclass, &myclass::io_cb> (&obj);
		3420
		3421	=item w->set (object *)
		3422
		3423	This is a variation of a method callback - leaving out the method to call
		3424	will default the method to C<operator ()>, which makes it possible to use
		3425	functor objects without having to manually specify the C<operator ()> all
		3426	the time. Incidentally, you can then also leave out the template argument
		3427	list.
		3428
		3429	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3430	int revents)>.
		3431
		3432	See the method-C<set> above for more details.
		3433
		3434	Example: use a functor object as callback.
		3435
		3436	struct myfunctor
		3437	{
		3438	void operator() (ev::io &w, int revents)
		3439	{
		3440	...
		3441	}
		3442	}
		3443
		3444	myfunctor f;
		3445
		3446	ev::io w;
		3447	w.set (&f);
		3448
		3449	=item w->set<function> (void *data = 0)
		3450
		3451	Also sets a callback, but uses a static method or plain function as
		3452	callback. The optional C<data> argument will be stored in the watcher's
		3453	C<data> member and is free for you to use.
		3454
		3455	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3456
		3457	See the method-C<set> above for more details.
		3458
		3459	Example: Use a plain function as callback.
		3460
		3461	static void io_cb (ev::io &w, int revents) { }
		3462	iow.set <io_cb> ();
		3463
		3464	=item w->set (loop)
		3465
		3466	Associates a different C<struct ev_loop> with this watcher. You can only
		3467	do this when the watcher is inactive (and not pending either).
		3468
		3469	=item w->set ([arguments])
		3470
		3471	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this
		3472	method or a suitable start method must be called at least once. Unlike the
		3473	C counterpart, an active watcher gets automatically stopped and restarted
		3474	when reconfiguring it with this method.
		3475
		3476	=item w->start ()
		3477
		3478	Starts the watcher. Note that there is no C<loop> argument, as the
		3479	constructor already stores the event loop.
		3480
		3481	=item w->start ([arguments])
		3482
		3483	Instead of calling C<set> and C<start> methods separately, it is often
		3484	convenient to wrap them in one call. Uses the same type of arguments as
		3485	the configure C<set> method of the watcher.
		3486
		3487	=item w->stop ()
		3488
		3489	Stops the watcher if it is active. Again, no C<loop> argument.
		3490
		3491	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3492
		3493	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3494	C<ev_TYPE_again> function.
		3495
		3496	=item w->sweep () (C<ev::embed> only)
		3497
		3498	Invokes C<ev_embed_sweep>.
		3499
		3500	=item w->update () (C<ev::stat> only)
		3501
		3502	Invokes C<ev_stat_stat>.
		3503
		3504	=back
		3505
		3506	=back
		3507
		3508	Example: Define a class with two I/O and idle watchers, start the I/O
		3509	watchers in the constructor.
		3510
		3511	class myclass
		3512	{
		3513	ev::io io ; void io_cb (ev::io &w, int revents);
		3514	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);
		3515	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3516
		3517	myclass (int fd)
		3518	{
		3519	io .set <myclass, &myclass::io_cb > (this);
		3520	io2 .set <myclass, &myclass::io2_cb > (this);
		3521	idle.set <myclass, &myclass::idle_cb> (this);
		3522
		3523	io.set (fd, ev::WRITE); // configure the watcher
		3524	io.start (); // start it whenever convenient
		3525
		3526	io2.start (fd, ev::READ); // set + start in one call
		3527	}
		3528	};
		3529
		3530
		3531	=head1 OTHER LANGUAGE BINDINGS
		3532
		3533	Libev does not offer other language bindings itself, but bindings for a
		3534	number of languages exist in the form of third-party packages. If you know
		3535	any interesting language binding in addition to the ones listed here, drop
		3536	me a note.
		3537
		3538	=over 4
		3539
		3540	=item Perl
		3541
		3542	The EV module implements the full libev API and is actually used to test
		3543	libev. EV is developed together with libev. Apart from the EV core module,
		3544	there are additional modules that implement libev-compatible interfaces
		3545	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3546	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3547	and C<EV::Glib>).
		3548
		3549	It can be found and installed via CPAN, its homepage is at
		3550	L<http://software.schmorp.de/pkg/EV>.
		3551
		3552	=item Python
		3553
		3554	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3555	seems to be quite complete and well-documented.
		3556
		3557	=item Ruby
		3558
		3559	Tony Arcieri has written a ruby extension that offers access to a subset
		3560	of the libev API and adds file handle abstractions, asynchronous DNS and
		3561	more on top of it. It can be found via gem servers. Its homepage is at
		3562	L<http://rev.rubyforge.org/>.
		3563
		3564	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3565	makes rev work even on mingw.
		3566
		3567	=item Haskell
		3568
		3569	A haskell binding to libev is available at
		3570	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3571
		3572	=item D
		3573
		3574	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3575	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3576
		3577	=item Ocaml
		3578
		3579	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3580	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3581
		3582	=item Lua
		3583
		3584	Brian Maher has written a partial interface to libev for lua (at the
		3585	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
		3586	L<http://github.com/brimworks/lua-ev>.
		3587
		3588	=back
		3589
		3590
		3591	=head1 MACRO MAGIC
		3592
		3593	Libev can be compiled with a variety of options, the most fundamental
		3594	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3595	functions and callbacks have an initial C<struct ev_loop *> argument.
		3596
		3597	To make it easier to write programs that cope with either variant, the
		3598	following macros are defined:
		3599
		3600	=over 4
		3601
		3602	=item C<EV_A>, C<EV_A_>
		3603
		3604	This provides the loop I<argument> for functions, if one is required ("ev
		3605	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3606	C<EV_A_> is used when other arguments are following. Example:
		3607
		3608	ev_unref (EV_A);
		3609	ev_timer_add (EV_A_ watcher);
		3610	ev_run (EV_A_ 0);
		3611
		3612	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3613	which is often provided by the following macro.
		3614
		3615	=item C<EV_P>, C<EV_P_>
		3616
		3617	This provides the loop I<parameter> for functions, if one is required ("ev
		3618	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3619	C<EV_P_> is used when other parameters are following. Example:
		3620
		3621	// this is how ev_unref is being declared
		3622	static void ev_unref (EV_P);
		3623
		3624	// this is how you can declare your typical callback
		3625	static void cb (EV_P_ ev_timer *w, int revents)
		3626
		3627	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3628	suitable for use with C<EV_A>.
		3629
		3630	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3631
		3632	Similar to the other two macros, this gives you the value of the default
		3633	loop, if multiple loops are supported ("ev loop default").
		3634
		3635	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3636
		3637	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3638	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3639	is undefined when the default loop has not been initialised by a previous
		3640	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3641
		3642	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3643	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3644
		3645	=back
		3646
		3647	Example: Declare and initialise a check watcher, utilising the above
		3648	macros so it will work regardless of whether multiple loops are supported
		3649	or not.
		3650
		3651	static void
		3652	check_cb (EV_P_ ev_timer *w, int revents)
		3653	{
		3654	ev_check_stop (EV_A_ w);
		3655	}
		3656
		3657	ev_check check;
		3658	ev_check_init (&check, check_cb);
		3659	ev_check_start (EV_DEFAULT_ &check);
		3660	ev_run (EV_DEFAULT_ 0);
		3661
		3662	=head1 EMBEDDING
		3663
		3664	Libev can (and often is) directly embedded into host
		3665	applications. Examples of applications that embed it include the Deliantra
		3666	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3667	and rxvt-unicode.
		3668
		3669	The goal is to enable you to just copy the necessary files into your
		3670	source directory without having to change even a single line in them, so
		3671	you can easily upgrade by simply copying (or having a checked-out copy of
		3672	libev somewhere in your source tree).
		3673
		3674	=head2 FILESETS
		3675
		3676	Depending on what features you need you need to include one or more sets of files
		3677	in your application.
		3678
		3679	=head3 CORE EVENT LOOP
		3680
		3681	To include only the libev core (all the C<ev_*> functions), with manual
		3682	configuration (no autoconf):
		3683
		3684	#define EV_STANDALONE 1
		3685	#include "ev.c"
		3686
		3687	This will automatically include F<ev.h>, too, and should be done in a
		3688	single C source file only to provide the function implementations. To use
		3689	it, do the same for F<ev.h> in all files wishing to use this API (best
		3690	done by writing a wrapper around F<ev.h> that you can include instead and
		3691	where you can put other configuration options):
		3692
		3693	#define EV_STANDALONE 1
		3694	#include "ev.h"
		3695
		3696	Both header files and implementation files can be compiled with a C++
		3697	compiler (at least, that's a stated goal, and breakage will be treated
		3698	as a bug).
		3699
		3700	You need the following files in your source tree, or in a directory
		3701	in your include path (e.g. in libev/ when using -Ilibev):
		3702
		3703	ev.h
		3704	ev.c
		3705	ev_vars.h
		3706	ev_wrap.h
		3707
		3708	ev_win32.c required on win32 platforms only
		3709
		3710	ev_select.c only when select backend is enabled (which is enabled by default)
		3711	ev_poll.c only when poll backend is enabled (disabled by default)
		3712	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3713	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3714	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3715
		3716	F<ev.c> includes the backend files directly when enabled, so you only need
		3717	to compile this single file.
		3718
		3719	=head3 LIBEVENT COMPATIBILITY API
		3720
		3721	To include the libevent compatibility API, also include:
		3722
		3723	#include "event.c"
		3724
		3725	in the file including F<ev.c>, and:
		3726
		3727	#include "event.h"
		3728
		3729	in the files that want to use the libevent API. This also includes F<ev.h>.
		3730
		3731	You need the following additional files for this:
		3732
		3733	event.h
		3734	event.c
		3735
		3736	=head3 AUTOCONF SUPPORT
		3737
		3738	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3739	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3740	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3741	include F<config.h> and configure itself accordingly.
		3742
		3743	For this of course you need the m4 file:
		3744
		3745	libev.m4
		3746
		3747	=head2 PREPROCESSOR SYMBOLS/MACROS
		3748
		3749	Libev can be configured via a variety of preprocessor symbols you have to
		3750	define before including (or compiling) any of its files. The default in
		3751	the absence of autoconf is documented for every option.
		3752
		3753	Symbols marked with "(h)" do not change the ABI, and can have different
		3754	values when compiling libev vs. including F<ev.h>, so it is permissible
		3755	to redefine them before including F<ev.h> without breaking compatibility
		3756	to a compiled library. All other symbols change the ABI, which means all
		3757	users of libev and the libev code itself must be compiled with compatible
		3758	settings.
		3759
		3760	=over 4
		3761
		3762	=item EV_COMPAT3 (h)
		3763
		3764	Backwards compatibility is a major concern for libev. This is why this
		3765	release of libev comes with wrappers for the functions and symbols that
		3766	have been renamed between libev version 3 and 4.
		3767
		3768	You can disable these wrappers (to test compatibility with future
		3769	versions) by defining C<EV_COMPAT3> to C<0> when compiling your
		3770	sources. This has the additional advantage that you can drop the C<struct>
		3771	from C<struct ev_loop> declarations, as libev will provide an C<ev_loop>
		3772	typedef in that case.
		3773
		3774	In some future version, the default for C<EV_COMPAT3> will become C<0>,
		3775	and in some even more future version the compatibility code will be
		3776	removed completely.
		3777
		3778	=item EV_STANDALONE (h)
		3779
		3780	Must always be C<1> if you do not use autoconf configuration, which
		3781	keeps libev from including F<config.h>, and it also defines dummy
		3782	implementations for some libevent functions (such as logging, which is not
		3783	supported). It will also not define any of the structs usually found in
		3784	F<event.h> that are not directly supported by the libev core alone.
		3785
		3786	In standalone mode, libev will still try to automatically deduce the
		3787	configuration, but has to be more conservative.
		3788
		3789	=item EV_USE_MONOTONIC
		3790
		3791	If defined to be C<1>, libev will try to detect the availability of the
		3792	monotonic clock option at both compile time and runtime. Otherwise no
		3793	use of the monotonic clock option will be attempted. If you enable this,
		3794	you usually have to link against librt or something similar. Enabling it
		3795	when the functionality isn't available is safe, though, although you have
		3796	to make sure you link against any libraries where the C<clock_gettime>
		3797	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		3798
		3799	=item EV_USE_REALTIME
		3800
		3801	If defined to be C<1>, libev will try to detect the availability of the
		3802	real-time clock option at compile time (and assume its availability
		3803	at runtime if successful). Otherwise no use of the real-time clock
		3804	option will be attempted. This effectively replaces C<gettimeofday>
		3805	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		3806	correctness. See the note about libraries in the description of
		3807	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3808	C<EV_USE_CLOCK_SYSCALL>.
		3809
		3810	=item EV_USE_CLOCK_SYSCALL
		3811
		3812	If defined to be C<1>, libev will try to use a direct syscall instead
		3813	of calling the system-provided C<clock_gettime> function. This option
		3814	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3815	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3816	programs needlessly. Using a direct syscall is slightly slower (in
		3817	theory), because no optimised vdso implementation can be used, but avoids
		3818	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		3819	higher, as it simplifies linking (no need for C<-lrt>).
		3820
		3821	=item EV_USE_NANOSLEEP
		3822
		3823	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3824	and will use it for delays. Otherwise it will use C<select ()>.
		3825
		3826	=item EV_USE_EVENTFD
		3827
		3828	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3829	available and will probe for kernel support at runtime. This will improve
		3830	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3831	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3832	2.7 or newer, otherwise disabled.
		3833
		3834	=item EV_USE_SELECT
		3835
		3836	If undefined or defined to be C<1>, libev will compile in support for the
		3837	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3838	other method takes over, select will be it. Otherwise the select backend
		3839	will not be compiled in.
		3840
		3841	=item EV_SELECT_USE_FD_SET
		3842
		3843	If defined to C<1>, then the select backend will use the system C<fd_set>
		3844	structure. This is useful if libev doesn't compile due to a missing
		3845	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		3846	on exotic systems. This usually limits the range of file descriptors to
		3847	some low limit such as 1024 or might have other limitations (winsocket
		3848	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		3849	configures the maximum size of the C<fd_set>.
		3850
		3851	=item EV_SELECT_IS_WINSOCKET
		3852
		3853	When defined to C<1>, the select backend will assume that
		3854	select/socket/connect etc. don't understand file descriptors but
		3855	wants osf handles on win32 (this is the case when the select to
		3856	be used is the winsock select). This means that it will call
		3857	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3858	it is assumed that all these functions actually work on fds, even
		3859	on win32. Should not be defined on non-win32 platforms.
		3860
		3861	=item EV_FD_TO_WIN32_HANDLE(fd)
		3862
		3863	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3864	file descriptors to socket handles. When not defining this symbol (the
		3865	default), then libev will call C<_get_osfhandle>, which is usually
		3866	correct. In some cases, programs use their own file descriptor management,
		3867	in which case they can provide this function to map fds to socket handles.
		3868
		3869	=item EV_WIN32_HANDLE_TO_FD(handle)
		3870
		3871	If C<EV_SELECT_IS_WINSOCKET> then libev maps handles to file descriptors
		3872	using the standard C<_open_osfhandle> function. For programs implementing
		3873	their own fd to handle mapping, overwriting this function makes it easier
		3874	to do so. This can be done by defining this macro to an appropriate value.
		3875
		3876	=item EV_WIN32_CLOSE_FD(fd)
		3877
		3878	If programs implement their own fd to handle mapping on win32, then this
		3879	macro can be used to override the C<close> function, useful to unregister
		3880	file descriptors again. Note that the replacement function has to close
		3881	the underlying OS handle.
		3882
		3883	=item EV_USE_POLL
		3884
		3885	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3886	backend. Otherwise it will be enabled on non-win32 platforms. It
		3887	takes precedence over select.
		3888
		3889	=item EV_USE_EPOLL
		3890
		3891	If defined to be C<1>, libev will compile in support for the Linux
		3892	C<epoll>(7) backend. Its availability will be detected at runtime,
		3893	otherwise another method will be used as fallback. This is the preferred
		3894	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3895	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3896
		3897	=item EV_USE_KQUEUE
		3898
		3899	If defined to be C<1>, libev will compile in support for the BSD style
		3900	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3901	otherwise another method will be used as fallback. This is the preferred
		3902	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3903	supports some types of fds correctly (the only platform we found that
		3904	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3905	not be used unless explicitly requested. The best way to use it is to find
		3906	out whether kqueue supports your type of fd properly and use an embedded
		3907	kqueue loop.
		3908
		3909	=item EV_USE_PORT
		3910
		3911	If defined to be C<1>, libev will compile in support for the Solaris
		3912	10 port style backend. Its availability will be detected at runtime,
		3913	otherwise another method will be used as fallback. This is the preferred
		3914	backend for Solaris 10 systems.
		3915
		3916	=item EV_USE_DEVPOLL
		3917
		3918	Reserved for future expansion, works like the USE symbols above.
		3919
		3920	=item EV_USE_INOTIFY
		3921
		3922	If defined to be C<1>, libev will compile in support for the Linux inotify
		3923	interface to speed up C<ev_stat> watchers. Its actual availability will
		3924	be detected at runtime. If undefined, it will be enabled if the headers
		3925	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3926
		3927	=item EV_ATOMIC_T
		3928
		3929	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3930	access is atomic with respect to other threads or signal contexts. No such
		3931	type is easily found in the C language, so you can provide your own type
		3932	that you know is safe for your purposes. It is used both for signal handler "locking"
		3933	as well as for signal and thread safety in C<ev_async> watchers.
		3934
		3935	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3936	(from F<signal.h>), which is usually good enough on most platforms.
		3937
		3938	=item EV_H (h)
		3939
		3940	The name of the F<ev.h> header file used to include it. The default if
		3941	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3942	used to virtually rename the F<ev.h> header file in case of conflicts.
		3943
		3944	=item EV_CONFIG_H (h)
		3945
		3946	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3947	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3948	C<EV_H>, above.
		3949
		3950	=item EV_EVENT_H (h)
		3951
		3952	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3953	of how the F<event.h> header can be found, the default is C<"event.h">.
		3954
		3955	=item EV_PROTOTYPES (h)
		3956
		3957	If defined to be C<0>, then F<ev.h> will not define any function
		3958	prototypes, but still define all the structs and other symbols. This is
		3959	occasionally useful if you want to provide your own wrapper functions
		3960	around libev functions.
		3961
		3962	=item EV_MULTIPLICITY
		3963
		3964	If undefined or defined to C<1>, then all event-loop-specific functions
		3965	will have the C<struct ev_loop *> as first argument, and you can create
		3966	additional independent event loops. Otherwise there will be no support
		3967	for multiple event loops and there is no first event loop pointer
		3968	argument. Instead, all functions act on the single default loop.
		3969
		3970	=item EV_MINPRI
		3971
		3972	=item EV_MAXPRI
		3973
		3974	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3975	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3976	provide for more priorities by overriding those symbols (usually defined
		3977	to be C<-2> and C<2>, respectively).
		3978
		3979	When doing priority-based operations, libev usually has to linearly search
		3980	all the priorities, so having many of them (hundreds) uses a lot of space
		3981	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3982	fine.
		3983
		3984	If your embedding application does not need any priorities, defining these
		3985	both to C<0> will save some memory and CPU.
		3986
		3987	=item EV_PERIODIC_ENABLE, EV_IDLE_ENABLE, EV_EMBED_ENABLE, EV_STAT_ENABLE,
		3988	EV_PREPARE_ENABLE, EV_CHECK_ENABLE, EV_FORK_ENABLE, EV_SIGNAL_ENABLE,
		3989	EV_ASYNC_ENABLE, EV_CHILD_ENABLE.
		3990
		3991	If undefined or defined to be C<1> (and the platform supports it), then
		3992	the respective watcher type is supported. If defined to be C<0>, then it
		3993	is not. Disabling watcher types mainly saves code size.
		3994
		3995	=item EV_FEATURES
		3996
		3997	If you need to shave off some kilobytes of code at the expense of some
		3998	speed (but with the full API), you can define this symbol to request
		3999	certain subsets of functionality. The default is to enable all features
		4000	that can be enabled on the platform.
		4001
		4002	A typical way to use this symbol is to define it to C<0> (or to a bitset
		4003	with some broad features you want) and then selectively re-enable
		4004	additional parts you want, for example if you want everything minimal,
		4005	but multiple event loop support, async and child watchers and the poll
		4006	backend, use this:
		4007
		4008	#define EV_FEATURES 0
		4009	#define EV_MULTIPLICITY 1
		4010	#define EV_USE_POLL 1
		4011	#define EV_CHILD_ENABLE 1
		4012	#define EV_ASYNC_ENABLE 1
		4013
		4014	The actual value is a bitset, it can be a combination of the following
		4015	values:
		4016
		4017	=over 4
		4018
		4019	=item C<1> - faster/larger code
		4020
		4021	Use larger code to speed up some operations.
		4022
		4023	Currently this is used to override some inlining decisions (enlarging the
		4024	code size by roughly 30% on amd64).
		4025
		4026	When optimising for size, use of compiler flags such as C<-Os> with
		4027	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
		4028	assertions.
		4029
		4030	=item C<2> - faster/larger data structures
		4031
		4032	Replaces the small 2-heap for timer management by a faster 4-heap, larger
		4033	hash table sizes and so on. This will usually further increase code size
		4034	and can additionally have an effect on the size of data structures at
		4035	runtime.
		4036
		4037	=item C<4> - full API configuration
		4038
		4039	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
		4040	enables multiplicity (C<EV_MULTIPLICITY>=1).
		4041
		4042	=item C<8> - full API
		4043
		4044	This enables a lot of the "lesser used" API functions. See C<ev.h> for
		4045	details on which parts of the API are still available without this
		4046	feature, and do not complain if this subset changes over time.
		4047
		4048	=item C<16> - enable all optional watcher types
		4049
		4050	Enables all optional watcher types. If you want to selectively enable
		4051	only some watcher types other than I/O and timers (e.g. prepare,
		4052	embed, async, child...) you can enable them manually by defining
		4053	C<EV_watchertype_ENABLE> to C<1> instead.
		4054
		4055	=item C<32> - enable all backends
		4056
		4057	This enables all backends - without this feature, you need to enable at
		4058	least one backend manually (C<EV_USE_SELECT> is a good choice).
		4059
		4060	=item C<64> - enable OS-specific "helper" APIs
		4061
		4062	Enable inotify, eventfd, signalfd and similar OS-specific helper APIs by
		4063	default.
		4064
		4065	=back
		4066
		4067	Compiling with C<gcc -Os -DEV_STANDALONE -DEV_USE_EPOLL=1 -DEV_FEATURES=0>
		4068	reduces the compiled size of libev from 24.7Kb code/2.8Kb data to 6.5Kb
		4069	code/0.3Kb data on my GNU/Linux amd64 system, while still giving you I/O
		4070	watchers, timers and monotonic clock support.
		4071
		4072	With an intelligent-enough linker (gcc+binutils are intelligent enough
		4073	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
		4074	your program might be left out as well - a binary starting a timer and an
		4075	I/O watcher then might come out at only 5Kb.
		4076
		4077	=item EV_AVOID_STDIO
		4078
		4079	If this is set to C<1> at compiletime, then libev will avoid using stdio
		4080	functions (printf, scanf, perror etc.). This will increase the code size
		4081	somewhat, but if your program doesn't otherwise depend on stdio and your
		4082	libc allows it, this avoids linking in the stdio library which is quite
		4083	big.
		4084
		4085	Note that error messages might become less precise when this option is
		4086	enabled.
		4087
		4088	=item EV_NSIG
		4089
		4090	The highest supported signal number, +1 (or, the number of
		4091	signals): Normally, libev tries to deduce the maximum number of signals
		4092	automatically, but sometimes this fails, in which case it can be
		4093	specified. Also, using a lower number than detected (C<32> should be
		4094	good for about any system in existence) can save some memory, as libev
		4095	statically allocates some 12-24 bytes per signal number.
		4096
		4097	=item EV_PID_HASHSIZE
		4098
		4099	C<ev_child> watchers use a small hash table to distribute workload by
		4100	pid. The default size is C<16> (or C<1> with C<EV_FEATURES> disabled),
		4101	usually more than enough. If you need to manage thousands of children you
		4102	might want to increase this value (I<must> be a power of two).
		4103
		4104	=item EV_INOTIFY_HASHSIZE
		4105
		4106	C<ev_stat> watchers use a small hash table to distribute workload by
		4107	inotify watch id. The default size is C<16> (or C<1> with C<EV_FEATURES>
		4108	disabled), usually more than enough. If you need to manage thousands of
		4109	C<ev_stat> watchers you might want to increase this value (I<must> be a
		4110	power of two).
		4111
		4112	=item EV_USE_4HEAP
		4113
		4114	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4115	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		4116	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		4117	faster performance with many (thousands) of watchers.
		4118
		4119	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4120	will be C<0>.
		4121
		4122	=item EV_HEAP_CACHE_AT
		4123
		4124	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		4125	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		4126	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		4127	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		4128	but avoids random read accesses on heap changes. This improves performance
		4129	noticeably with many (hundreds) of watchers.
		4130
		4131	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4132	will be C<0>.
		4133
		4134	=item EV_VERIFY
		4135
		4136	Controls how much internal verification (see C<ev_verify ()>) will
		4137	be done: If set to C<0>, no internal verification code will be compiled
		4138	in. If set to C<1>, then verification code will be compiled in, but not
		4139	called. If set to C<2>, then the internal verification code will be
		4140	called once per loop, which can slow down libev. If set to C<3>, then the
		4141	verification code will be called very frequently, which will slow down
		4142	libev considerably.
		4143
		4144	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
		4145	will be C<0>.
		4146
		4147	=item EV_COMMON
		4148
		4149	By default, all watchers have a C<void *data> member. By redefining
		4150	this macro to something else you can include more and other types of
		4151	members. You have to define it each time you include one of the files,
		4152	though, and it must be identical each time.
		4153
		4154	For example, the perl EV module uses something like this:
		4155
		4156	#define EV_COMMON \
		4157	SV *self; /* contains this struct */ \
		4158	SV *cb_sv, *fh /* note no trailing ";" */
		4159
		4160	=item EV_CB_DECLARE (type)
		4161
		4162	=item EV_CB_INVOKE (watcher, revents)
		4163
		4164	=item ev_set_cb (ev, cb)
		4165
		4166	Can be used to change the callback member declaration in each watcher,
		4167	and the way callbacks are invoked and set. Must expand to a struct member
		4168	definition and a statement, respectively. See the F<ev.h> header file for
		4169	their default definitions. One possible use for overriding these is to
		4170	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		4171	method calls instead of plain function calls in C++.
		4172
		4173	=back
		4174
		4175	=head2 EXPORTED API SYMBOLS
		4176
		4177	If you need to re-export the API (e.g. via a DLL) and you need a list of
		4178	exported symbols, you can use the provided F<Symbol.*> files which list
		4179	all public symbols, one per line:
		4180
		4181	Symbols.ev for libev proper
		4182	Symbols.event for the libevent emulation
		4183
		4184	This can also be used to rename all public symbols to avoid clashes with
		4185	multiple versions of libev linked together (which is obviously bad in
		4186	itself, but sometimes it is inconvenient to avoid this).
		4187
		4188	A sed command like this will create wrapper C<#define>'s that you need to
		4189	include before including F<ev.h>:
		4190
		4191	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		4192
		4193	This would create a file F<wrap.h> which essentially looks like this:
		4194
		4195	#define ev_backend myprefix_ev_backend
		4196	#define ev_check_start myprefix_ev_check_start
		4197	#define ev_check_stop myprefix_ev_check_stop
		4198	...
		4199
		4200	=head2 EXAMPLES
		4201
		4202	For a real-world example of a program the includes libev
		4203	verbatim, you can have a look at the EV perl module
		4204	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		4205	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		4206	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		4207	will be compiled. It is pretty complex because it provides its own header
		4208	file.
		4209
		4210	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		4211	that everybody includes and which overrides some configure choices:
		4212
		4213	#define EV_FEATURES 8
		4214	#define EV_USE_SELECT 1
		4215	#define EV_PREPARE_ENABLE 1
		4216	#define EV_IDLE_ENABLE 1
		4217	#define EV_SIGNAL_ENABLE 1
		4218	#define EV_CHILD_ENABLE 1
		4219	#define EV_USE_STDEXCEPT 0
		4220	#define EV_CONFIG_H <config.h>
		4221
		4222	#include "ev++.h"
		4223
		4224	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		4225
		4226	#include "ev_cpp.h"
		4227	#include "ev.c"
		4228
		4229	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		4230
		4231	=head2 THREADS AND COROUTINES
		4232
		4233	=head3 THREADS
		4234
		4235	All libev functions are reentrant and thread-safe unless explicitly
		4236	documented otherwise, but libev implements no locking itself. This means
		4237	that you can use as many loops as you want in parallel, as long as there
		4238	are no concurrent calls into any libev function with the same loop
		4239	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		4240	of course): libev guarantees that different event loops share no data
		4241	structures that need any locking.
		4242
		4243	Or to put it differently: calls with different loop parameters can be done
		4244	concurrently from multiple threads, calls with the same loop parameter
		4245	must be done serially (but can be done from different threads, as long as
		4246	only one thread ever is inside a call at any point in time, e.g. by using
		4247	a mutex per loop).
		4248
		4249	Specifically to support threads (and signal handlers), libev implements
		4250	so-called C<ev_async> watchers, which allow some limited form of
		4251	concurrency on the same event loop, namely waking it up "from the
		4252	outside".
		4253
		4254	If you want to know which design (one loop, locking, or multiple loops
		4255	without or something else still) is best for your problem, then I cannot
		4256	help you, but here is some generic advice:
		4257
		4258	=over 4
		4259
		4260	=item * most applications have a main thread: use the default libev loop
		4261	in that thread, or create a separate thread running only the default loop.
		4262
		4263	This helps integrating other libraries or software modules that use libev
		4264	themselves and don't care/know about threading.
		4265
		4266	=item * one loop per thread is usually a good model.
		4267
		4268	Doing this is almost never wrong, sometimes a better-performance model
		4269	exists, but it is always a good start.
		4270
		4271	=item * other models exist, such as the leader/follower pattern, where one
		4272	loop is handed through multiple threads in a kind of round-robin fashion.
		4273
		4274	Choosing a model is hard - look around, learn, know that usually you can do
		4275	better than you currently do :-)
		4276
		4277	=item * often you need to talk to some other thread which blocks in the
		4278	event loop.
		4279
		4280	C<ev_async> watchers can be used to wake them up from other threads safely
		4281	(or from signal contexts...).
		4282
		4283	An example use would be to communicate signals or other events that only
		4284	work in the default loop by registering the signal watcher with the
		4285	default loop and triggering an C<ev_async> watcher from the default loop
		4286	watcher callback into the event loop interested in the signal.
		4287
		4288	=back
		4289
		4290	=head4 THREAD LOCKING EXAMPLE
		4291
		4292	Here is a fictitious example of how to run an event loop in a different
		4293	thread than where callbacks are being invoked and watchers are
		4294	created/added/removed.
		4295
		4296	For a real-world example, see the C<EV::Loop::Async> perl module,
		4297	which uses exactly this technique (which is suited for many high-level
		4298	languages).
		4299
		4300	The example uses a pthread mutex to protect the loop data, a condition
		4301	variable to wait for callback invocations, an async watcher to notify the
		4302	event loop thread and an unspecified mechanism to wake up the main thread.
		4303
		4304	First, you need to associate some data with the event loop:
		4305
		4306	typedef struct {
		4307	mutex_t lock; /* global loop lock */
		4308	ev_async async_w;
		4309	thread_t tid;
		4310	cond_t invoke_cv;
		4311	} userdata;
		4312
		4313	void prepare_loop (EV_P)
		4314	{
		4315	// for simplicity, we use a static userdata struct.
		4316	static userdata u;
		4317
		4318	ev_async_init (&u->async_w, async_cb);
		4319	ev_async_start (EV_A_ &u->async_w);
		4320
		4321	pthread_mutex_init (&u->lock, 0);
		4322	pthread_cond_init (&u->invoke_cv, 0);
		4323
		4324	// now associate this with the loop
		4325	ev_set_userdata (EV_A_ u);
		4326	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		4327	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		4328
		4329	// then create the thread running ev_loop
		4330	pthread_create (&u->tid, 0, l_run, EV_A);
		4331	}
		4332
		4333	The callback for the C<ev_async> watcher does nothing: the watcher is used
		4334	solely to wake up the event loop so it takes notice of any new watchers
		4335	that might have been added:
		4336
		4337	static void
		4338	async_cb (EV_P_ ev_async *w, int revents)
		4339	{
		4340	// just used for the side effects
		4341	}
		4342
		4343	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		4344	protecting the loop data, respectively.
		4345
		4346	static void
		4347	l_release (EV_P)
		4348	{
		4349	userdata *u = ev_userdata (EV_A);
		4350	pthread_mutex_unlock (&u->lock);
		4351	}
		4352
		4353	static void
		4354	l_acquire (EV_P)
		4355	{
		4356	userdata *u = ev_userdata (EV_A);
		4357	pthread_mutex_lock (&u->lock);
		4358	}
		4359
		4360	The event loop thread first acquires the mutex, and then jumps straight
		4361	into C<ev_run>:
		4362
		4363	void *
		4364	l_run (void *thr_arg)
		4365	{
		4366	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		4367
		4368	l_acquire (EV_A);
		4369	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		4370	ev_run (EV_A_ 0);
		4371	l_release (EV_A);
		4372
		4373	return 0;
		4374	}
		4375
		4376	Instead of invoking all pending watchers, the C<l_invoke> callback will
		4377	signal the main thread via some unspecified mechanism (signals? pipe
		4378	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		4379	have been called (in a while loop because a) spurious wakeups are possible
		4380	and b) skipping inter-thread-communication when there are no pending
		4381	watchers is very beneficial):
		4382
		4383	static void
		4384	l_invoke (EV_P)
		4385	{
		4386	userdata *u = ev_userdata (EV_A);
		4387
		4388	while (ev_pending_count (EV_A))
		4389	{
		4390	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		4391	pthread_cond_wait (&u->invoke_cv, &u->lock);
		4392	}
		4393	}
		4394
		4395	Now, whenever the main thread gets told to invoke pending watchers, it
		4396	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		4397	thread to continue:
		4398
		4399	static void
		4400	real_invoke_pending (EV_P)
		4401	{
		4402	userdata *u = ev_userdata (EV_A);
		4403
		4404	pthread_mutex_lock (&u->lock);
		4405	ev_invoke_pending (EV_A);
		4406	pthread_cond_signal (&u->invoke_cv);
		4407	pthread_mutex_unlock (&u->lock);
		4408	}
		4409
		4410	Whenever you want to start/stop a watcher or do other modifications to an
		4411	event loop, you will now have to lock:
		4412
		4413	ev_timer timeout_watcher;
		4414	userdata *u = ev_userdata (EV_A);
		4415
		4416	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		4417
		4418	pthread_mutex_lock (&u->lock);
		4419	ev_timer_start (EV_A_ &timeout_watcher);
		4420	ev_async_send (EV_A_ &u->async_w);
		4421	pthread_mutex_unlock (&u->lock);
		4422
		4423	Note that sending the C<ev_async> watcher is required because otherwise
		4424	an event loop currently blocking in the kernel will have no knowledge
		4425	about the newly added timer. By waking up the loop it will pick up any new
		4426	watchers in the next event loop iteration.
		4427
		4428	=head3 COROUTINES
		4429
		4430	Libev is very accommodating to coroutines ("cooperative threads"):
		4431	libev fully supports nesting calls to its functions from different
		4432	coroutines (e.g. you can call C<ev_run> on the same loop from two
		4433	different coroutines, and switch freely between both coroutines running
		4434	the loop, as long as you don't confuse yourself). The only exception is
		4435	that you must not do this from C<ev_periodic> reschedule callbacks.
		4436
		4437	Care has been taken to ensure that libev does not keep local state inside
		4438	C<ev_run>, and other calls do not usually allow for coroutine switches as
		4439	they do not call any callbacks.
		4440
		4441	=head2 COMPILER WARNINGS
		4442
		4443	Depending on your compiler and compiler settings, you might get no or a
		4444	lot of warnings when compiling libev code. Some people are apparently
		4445	scared by this.
		4446
		4447	However, these are unavoidable for many reasons. For one, each compiler
		4448	has different warnings, and each user has different tastes regarding
		4449	warning options. "Warn-free" code therefore cannot be a goal except when
		4450	targeting a specific compiler and compiler-version.
		4451
		4452	Another reason is that some compiler warnings require elaborate
		4453	workarounds, or other changes to the code that make it less clear and less
		4454	maintainable.
		4455
		4456	And of course, some compiler warnings are just plain stupid, or simply
		4457	wrong (because they don't actually warn about the condition their message
		4458	seems to warn about). For example, certain older gcc versions had some
		4459	warnings that resulted in an extreme number of false positives. These have
		4460	been fixed, but some people still insist on making code warn-free with
		4461	such buggy versions.
		4462
		4463	While libev is written to generate as few warnings as possible,
		4464	"warn-free" code is not a goal, and it is recommended not to build libev
		4465	with any compiler warnings enabled unless you are prepared to cope with
		4466	them (e.g. by ignoring them). Remember that warnings are just that:
		4467	warnings, not errors, or proof of bugs.
		4468
		4469
		4470	=head2 VALGRIND
		4471
		4472	Valgrind has a special section here because it is a popular tool that is
		4473	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		4474
		4475	If you think you found a bug (memory leak, uninitialised data access etc.)
		4476	in libev, then check twice: If valgrind reports something like:
		4477
		4478	==2274== definitely lost: 0 bytes in 0 blocks.
		4479	==2274== possibly lost: 0 bytes in 0 blocks.
		4480	==2274== still reachable: 256 bytes in 1 blocks.
		4481
		4482	Then there is no memory leak, just as memory accounted to global variables
		4483	is not a memleak - the memory is still being referenced, and didn't leak.
		4484
		4485	Similarly, under some circumstances, valgrind might report kernel bugs
		4486	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		4487	although an acceptable workaround has been found here), or it might be
		4488	confused.
		4489
		4490	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		4491	make it into some kind of religion.
		4492
		4493	If you are unsure about something, feel free to contact the mailing list
		4494	with the full valgrind report and an explanation on why you think this
		4495	is a bug in libev (best check the archives, too :). However, don't be
		4496	annoyed when you get a brisk "this is no bug" answer and take the chance
		4497	of learning how to interpret valgrind properly.
		4498
		4499	If you need, for some reason, empty reports from valgrind for your project
		4500	I suggest using suppression lists.
		4501
		4502
		4503	=head1 PORTABILITY NOTES
		4504
		4505	=head2 GNU/LINUX 32 BIT LIMITATIONS
		4506
		4507	GNU/Linux is the only common platform that supports 64 bit file/large file
		4508	interfaces but I<disables> them by default.
		4509
		4510	That means that libev compiled in the default environment doesn't support
		4511	files larger than 2GiB or so, which mainly affects C<ev_stat> watchers.
		4512
		4513	Unfortunately, many programs try to work around this GNU/Linux issue
		4514	by enabling the large file API, which makes them incompatible with the
		4515	standard libev compiled for their system.
		4516
		4517	Likewise, libev cannot enable the large file API itself as this would
		4518	suddenly make it incompatible to the default compile time environment,
		4519	i.e. all programs not using special compile switches.
		4520
		4521	=head2 OS/X AND DARWIN BUGS
		4522
		4523	The whole thing is a bug if you ask me - basically any system interface
		4524	you touch is broken, whether it is locales, poll, kqueue or even the
		4525	OpenGL drivers.
		4526
		4527	=head3 C<kqueue> is buggy
		4528
		4529	The kqueue syscall is broken in all known versions - most versions support
		4530	only sockets, many support pipes.
		4531
		4532	Libev tries to work around this by not using C<kqueue> by default on
		4533	this rotten platform, but of course you can still ask for it when creating
		4534	a loop.
		4535
		4536	=head3 C<poll> is buggy
		4537
		4538	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
		4539	implementation by something calling C<kqueue> internally around the 10.5.6
		4540	release, so now C<kqueue> I<and> C<poll> are broken.
		4541
		4542	Libev tries to work around this by not using C<poll> by default on
		4543	this rotten platform, but of course you can still ask for it when creating
		4544	a loop.
		4545
		4546	=head3 C<select> is buggy
		4547
		4548	All that's left is C<select>, and of course Apple found a way to fuck this
		4549	one up as well: On OS/X, C<select> actively limits the number of file
		4550	descriptors you can pass in to 1024 - your program suddenly crashes when
		4551	you use more.
		4552
		4553	There is an undocumented "workaround" for this - defining
		4554	C<_DARWIN_UNLIMITED_SELECT>, which libev tries to use, so select I<should>
		4555	work on OS/X.
		4556
		4557	=head2 SOLARIS PROBLEMS AND WORKAROUNDS
		4558
		4559	=head3 C<errno> reentrancy
		4560
		4561	The default compile environment on Solaris is unfortunately so
		4562	thread-unsafe that you can't even use components/libraries compiled
		4563	without C<-D_REENTRANT> (as long as they use C<errno>), which, of course,
		4564	isn't defined by default.
		4565
		4566	If you want to use libev in threaded environments you have to make sure
		4567	it's compiled with C<_REENTRANT> defined.
		4568
		4569	=head3 Event port backend
		4570
		4571	The scalable event interface for Solaris is called "event ports". Unfortunately,
		4572	this mechanism is very buggy. If you run into high CPU usage, your program
		4573	freezes or you get a large number of spurious wakeups, make sure you have
		4574	all the relevant and latest kernel patches applied. No, I don't know which
		4575	ones, but there are multiple ones.
		4576
		4577	If you can't get it to work, you can try running the program by setting
		4578	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
		4579	C<select> backends.
		4580
		4581	=head2 AIX POLL BUG
		4582
		4583	AIX unfortunately has a broken C<poll.h> header. Libev works around
		4584	this by trying to avoid the poll backend altogether (i.e. it's not even
		4585	compiled in), which normally isn't a big problem as C<select> works fine
		4586	with large bitsets, and AIX is dead anyway.
		4587
		4588	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		4589
		4590	=head3 General issues
		4591
		4592	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		4593	requires, and its I/O model is fundamentally incompatible with the POSIX
		4594	model. Libev still offers limited functionality on this platform in
		4595	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		4596	descriptors. This only applies when using Win32 natively, not when using
		4597	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
		4598	as every compielr comes with a slightly differently broken/incompatible
		4599	environment.
		4600
		4601	Lifting these limitations would basically require the full
		4602	re-implementation of the I/O system. If you are into this kind of thing,
		4603	then note that glib does exactly that for you in a very portable way (note
		4604	also that glib is the slowest event library known to man).
		4605
		4606	There is no supported compilation method available on windows except
		4607	embedding it into other applications.
		4608
		4609	Sensible signal handling is officially unsupported by Microsoft - libev
		4610	tries its best, but under most conditions, signals will simply not work.
		4611
		4612	Not a libev limitation but worth mentioning: windows apparently doesn't
		4613	accept large writes: instead of resulting in a partial write, windows will
		4614	either accept everything or return C<ENOBUFS> if the buffer is too large,
		4615	so make sure you only write small amounts into your sockets (less than a
		4616	megabyte seems safe, but this apparently depends on the amount of memory
		4617	available).
		4618
		4619	Due to the many, low, and arbitrary limits on the win32 platform and
		4620	the abysmal performance of winsockets, using a large number of sockets
		4621	is not recommended (and not reasonable). If your program needs to use
		4622	more than a hundred or so sockets, then likely it needs to use a totally
		4623	different implementation for windows, as libev offers the POSIX readiness
		4624	notification model, which cannot be implemented efficiently on windows
		4625	(due to Microsoft monopoly games).
		4626
		4627	A typical way to use libev under windows is to embed it (see the embedding
		4628	section for details) and use the following F<evwrap.h> header file instead
		4629	of F<ev.h>:
		4630
		4631	#define EV_STANDALONE /* keeps ev from requiring config.h */
		4632	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		4633
		4634	#include "ev.h"
		4635
		4636	And compile the following F<evwrap.c> file into your project (make sure
		4637	you do I<not> compile the F<ev.c> or any other embedded source files!):
		4638
		4639	#include "evwrap.h"
		4640	#include "ev.c"
		4641
		4642	=head3 The winsocket C<select> function
		4643
		4644	The winsocket C<select> function doesn't follow POSIX in that it
		4645	requires socket I<handles> and not socket I<file descriptors> (it is
		4646	also extremely buggy). This makes select very inefficient, and also
		4647	requires a mapping from file descriptors to socket handles (the Microsoft
		4648	C runtime provides the function C<_open_osfhandle> for this). See the
		4649	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		4650	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		4651
		4652	The configuration for a "naked" win32 using the Microsoft runtime
		4653	libraries and raw winsocket select is:
		4654
		4655	#define EV_USE_SELECT 1
		4656	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		4657
		4658	Note that winsockets handling of fd sets is O(n), so you can easily get a
		4659	complexity in the O(n²) range when using win32.
		4660
		4661	=head3 Limited number of file descriptors
		4662
		4663	Windows has numerous arbitrary (and low) limits on things.
		4664
		4665	Early versions of winsocket's select only supported waiting for a maximum
		4666	of C<64> handles (probably owning to the fact that all windows kernels
		4667	can only wait for C<64> things at the same time internally; Microsoft
		4668	recommends spawning a chain of threads and wait for 63 handles and the
		4669	previous thread in each. Sounds great!).
		4670
		4671	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		4672	to some high number (e.g. C<2048>) before compiling the winsocket select
		4673	call (which might be in libev or elsewhere, for example, perl and many
		4674	other interpreters do their own select emulation on windows).
		4675
		4676	Another limit is the number of file descriptors in the Microsoft runtime
		4677	libraries, which by default is C<64> (there must be a hidden I<64>
		4678	fetish or something like this inside Microsoft). You can increase this
		4679	by calling C<_setmaxstdio>, which can increase this limit to C<2048>
		4680	(another arbitrary limit), but is broken in many versions of the Microsoft
		4681	runtime libraries. This might get you to about C<512> or C<2048> sockets
		4682	(depending on windows version and/or the phase of the moon). To get more,
		4683	you need to wrap all I/O functions and provide your own fd management, but
		4684	the cost of calling select (O(n²)) will likely make this unworkable.
		4685
		4686	=head2 PORTABILITY REQUIREMENTS
		4687
		4688	In addition to a working ISO-C implementation and of course the
		4689	backend-specific APIs, libev relies on a few additional extensions:
		4690
		4691	=over 4
		4692
		4693	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		4694	calling conventions regardless of C<ev_watcher_type *>.
		4695
		4696	Libev assumes not only that all watcher pointers have the same internal
		4697	structure (guaranteed by POSIX but not by ISO C for example), but it also
		4698	assumes that the same (machine) code can be used to call any watcher
		4699	callback: The watcher callbacks have different type signatures, but libev
		4700	calls them using an C<ev_watcher *> internally.
		4701
		4702	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4703
		4704	The type C<sig_atomic_t volatile> (or whatever is defined as
		4705	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4706	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4707	believed to be sufficiently portable.
		4708
		4709	=item C<sigprocmask> must work in a threaded environment
		4710
		4711	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4712	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		4713	pthread implementations will either allow C<sigprocmask> in the "main
		4714	thread" or will block signals process-wide, both behaviours would
		4715	be compatible with libev. Interaction between C<sigprocmask> and
		4716	C<pthread_sigmask> could complicate things, however.
		4717
		4718	The most portable way to handle signals is to block signals in all threads
		4719	except the initial one, and run the default loop in the initial thread as
		4720	well.
		4721
		4722	=item C<long> must be large enough for common memory allocation sizes
		4723
		4724	To improve portability and simplify its API, libev uses C<long> internally
		4725	instead of C<size_t> when allocating its data structures. On non-POSIX
		4726	systems (Microsoft...) this might be unexpectedly low, but is still at
		4727	least 31 bits everywhere, which is enough for hundreds of millions of
		4728	watchers.
		4729
		4730	=item C<double> must hold a time value in seconds with enough accuracy
		4731
		4732	The type C<double> is used to represent timestamps. It is required to
		4733	have at least 51 bits of mantissa (and 9 bits of exponent), which is
		4734	good enough for at least into the year 4000 with millisecond accuracy
		4735	(the design goal for libev). This requirement is overfulfilled by
		4736	implementations using IEEE 754, which is basically all existing ones. With
		4737	IEEE 754 doubles, you get microsecond accuracy until at least 2200.
		4738
		4739	=back
		4740
		4741	If you know of other additional requirements drop me a note.
		4742
		4743
		4744	=head1 ALGORITHMIC COMPLEXITIES
		4745
		4746	In this section the complexities of (many of) the algorithms used inside
		4747	libev will be documented. For complexity discussions about backends see
		4748	the documentation for C<ev_default_init>.
		4749
		4750	All of the following are about amortised time: If an array needs to be
		4751	extended, libev needs to realloc and move the whole array, but this
		4752	happens asymptotically rarer with higher number of elements, so O(1) might
		4753	mean that libev does a lengthy realloc operation in rare cases, but on
		4754	average it is much faster and asymptotically approaches constant time.
		4755
		4756	=over 4
		4757
		4758	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		4759
		4760	This means that, when you have a watcher that triggers in one hour and
		4761	there are 100 watchers that would trigger before that, then inserting will
		4762	have to skip roughly seven (C<ld 100>) of these watchers.
		4763
		4764	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		4765
		4766	That means that changing a timer costs less than removing/adding them,
		4767	as only the relative motion in the event queue has to be paid for.
		4768
		4769	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		4770
		4771	These just add the watcher into an array or at the head of a list.
		4772
		4773	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		4774
		4775	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		4776
		4777	These watchers are stored in lists, so they need to be walked to find the
		4778	correct watcher to remove. The lists are usually short (you don't usually
		4779	have many watchers waiting for the same fd or signal: one is typical, two
		4780	is rare).
		4781
		4782	=item Finding the next timer in each loop iteration: O(1)
		4783
		4784	By virtue of using a binary or 4-heap, the next timer is always found at a
		4785	fixed position in the storage array.
		4786
		4787	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		4788
		4789	A change means an I/O watcher gets started or stopped, which requires
		4790	libev to recalculate its status (and possibly tell the kernel, depending
		4791	on backend and whether C<ev_io_set> was used).
		4792
		4793	=item Activating one watcher (putting it into the pending state): O(1)
		4794
		4795	=item Priority handling: O(number_of_priorities)
		4796
		4797	Priorities are implemented by allocating some space for each
		4798	priority. When doing priority-based operations, libev usually has to
		4799	linearly search all the priorities, but starting/stopping and activating
		4800	watchers becomes O(1) with respect to priority handling.
		4801
		4802	=item Sending an ev_async: O(1)
		4803
		4804	=item Processing ev_async_send: O(number_of_async_watchers)
		4805
		4806	=item Processing signals: O(max_signal_number)
		4807
		4808	Sending involves a system call I<iff> there were no other C<ev_async_send>
		4809	calls in the current loop iteration. Checking for async and signal events
		4810	involves iterating over all running async watchers or all signal numbers.
		4811
		4812	=back
		4813
		4814
		4815	=head1 PORTING FROM LIBEV 3.X TO 4.X
		4816
		4817	The major version 4 introduced some minor incompatible changes to the API.
		4818
		4819	At the moment, the C<ev.h> header file tries to implement superficial
		4820	compatibility, so most programs should still compile. Those might be
		4821	removed in later versions of libev, so better update early than late.
		4822
		4823	=over 4
		4824
		4825	=item function/symbol renames
		4826
		4827	A number of functions and symbols have been renamed:
		4828
		4829	ev_loop => ev_run
		4830	EVLOOP_NONBLOCK => EVRUN_NOWAIT
		4831	EVLOOP_ONESHOT => EVRUN_ONCE
		4832
		4833	ev_unloop => ev_break
		4834	EVUNLOOP_CANCEL => EVBREAK_CANCEL
		4835	EVUNLOOP_ONE => EVBREAK_ONE
		4836	EVUNLOOP_ALL => EVBREAK_ALL
		4837
		4838	EV_TIMEOUT => EV_TIMER
		4839
		4840	ev_loop_count => ev_iteration
		4841	ev_loop_depth => ev_depth
		4842	ev_loop_verify => ev_verify
		4843
		4844	Most functions working on C<struct ev_loop> objects don't have an
		4845	C<ev_loop_> prefix, so it was removed; C<ev_loop>, C<ev_unloop> and
		4846	associated constants have been renamed to not collide with the C<struct
		4847	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
		4848	as all other watcher types. Note that C<ev_loop_fork> is still called
		4849	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
		4850	typedef.
		4851
		4852	=item C<EV_COMPAT3> backwards compatibility mechanism
		4853
		4854	The backward compatibility mechanism can be controlled by
		4855	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
		4856	section.
		4857
		4858	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
		4859
		4860	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
		4861	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
		4862	and work, but the library code will of course be larger.
		4863
		4864	=back
		4865
		4866
		4867	=head1 GLOSSARY
		4868
		4869	=over 4
		4870
		4871	=item active
		4872
		4873	A watcher is active as long as it has been started (has been attached to
		4874	an event loop) but not yet stopped (disassociated from the event loop).
		4875
		4876	=item application
		4877
		4878	In this document, an application is whatever is using libev.
		4879
		4880	=item callback
		4881
		4882	The address of a function that is called when some event has been
		4883	detected. Callbacks are being passed the event loop, the watcher that
		4884	received the event, and the actual event bitset.
		4885
		4886	=item callback invocation
		4887
		4888	The act of calling the callback associated with a watcher.
		4889
		4890	=item event
		4891
		4892	A change of state of some external event, such as data now being available
		4893	for reading on a file descriptor, time having passed or simply not having
		4894	any other events happening anymore.
		4895
		4896	In libev, events are represented as single bits (such as C<EV_READ> or
		4897	C<EV_TIMER>).
		4898
		4899	=item event library
		4900
		4901	A software package implementing an event model and loop.
		4902
		4903	=item event loop
		4904
		4905	An entity that handles and processes external events and converts them
		4906	into callback invocations.
		4907
		4908	=item event model
		4909
		4910	The model used to describe how an event loop handles and processes
		4911	watchers and events.
		4912
		4913	=item pending
		4914
		4915	A watcher is pending as soon as the corresponding event has been detected,
		4916	and stops being pending as soon as the watcher will be invoked or its
		4917	pending status is explicitly cleared by the application.
		4918
		4919	A watcher can be pending, but not active. Stopping a watcher also clears
		4920	its pending status.
		4921
		4922	=item real time
		4923
		4924	The physical time that is observed. It is apparently strictly monotonic :)
		4925
		4926	=item wall-clock time
		4927
		4928	The time and date as shown on clocks. Unlike real time, it can actually
		4929	be wrong and jump forwards and backwards, e.g. when the you adjust your
		4930	clock.
		4931
		4932	=item watcher
		4933
		4934	A data structure that describes interest in certain events. Watchers need
		4935	to be started (attached to an event loop) before they can receive events.
		4936
		4937	=item watcher invocation
		4938
		4939	The act of calling the callback associated with a watcher.
		4940
		4941	=back
825		4942
826	=head1 AUTHOR	4943	=head1 AUTHOR
827		4944
828	Marc Lehmann <libev@schmorp.de>.	4945	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
829		4946

Diff Legend

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