ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.5 by root, Mon Nov 12 08:11:57 2007 UTC vs.
Revision 1.231 by root, Wed Apr 15 19:35:53 2009 UTC

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

Diff Legend

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