ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.8 by root, Mon Nov 12 08:20:02 2007 UTC vs.
Revision 1.213 by root, Wed Nov 5 02:48:45 2008 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	// every watcher type has its own typedef'd struct
		15	// with the name ev_TYPE
		16	ev_io stdin_watcher;
		17	ev_timer timeout_watcher;
		18
		19	// all watcher callbacks have a similar signature
		20	// this callback is called when data is readable on stdin
		21	static void
		22	stdin_cb (EV_P_ ev_io *w, int revents)
		23	{
		24	puts ("stdin ready");
		25	// for one-shot events, one must manually stop the watcher
		26	// with its corresponding stop function.
		27	ev_io_stop (EV_A_ w);
		28
		29	// this causes all nested ev_loop's to stop iterating
		30	ev_unloop (EV_A_ EVUNLOOP_ALL);
		31	}
		32
		33	// another callback, this time for a time-out
		34	static void
		35	timeout_cb (EV_P_ ev_timer *w, int revents)
		36	{
		37	puts ("timeout");
		38	// this causes the innermost ev_loop to stop iterating
		39	ev_unloop (EV_A_ EVUNLOOP_ONE);
		40	}
		41
		42	int
		43	main (void)
		44	{
		45	// use the default event loop unless you have special needs
		46	ev_loop *loop = ev_default_loop (0);
		47
		48	// initialise an io watcher, then start it
		49	// this one will watch for stdin to become readable
		50	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
		51	ev_io_start (loop, &stdin_watcher);
		52
		53	// initialise a timer watcher, then start it
		54	// simple non-repeating 5.5 second timeout
		55	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		56	ev_timer_start (loop, &timeout_watcher);
		57
		58	// now wait for events to arrive
		59	ev_loop (loop, 0);
		60
		61	// unloop was called, so exit
		62	return 0;
		63	}
8		64
9	=head1 DESCRIPTION	65	=head1 DESCRIPTION
10		66
		67	The newest version of this document is also available as an html-formatted
		68	web page you might find easier to navigate when reading it for the first
		69	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		70
11	Libev is an event loop: you register interest in certain events (such as a	71	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	72	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	73	these event sources and provide your program with events.
14		74
15	To do this, it must take more or less complete control over your process	75	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	76	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	77	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	79	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	80	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	81	details of the event, and then hand it over to libev by I<starting> the
22	watcher.	82	watcher.
23		83
24	=head1 FEATURES	84	=head2 FEATURES
25		85
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	86	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27	kqueue mechanisms for file descriptor events, relative timers, absolute	87	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28	timers with customised rescheduling, signal events, process status change	88	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
29	events (related to SIGCHLD), and event watchers dealing with the event	89	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
30	loop mechanism itself (idle, prepare and check watchers). It also is quite	90	with customised rescheduling (C<ev_periodic>), synchronous signals
		91	(C<ev_signal>), process status change events (C<ev_child>), and event
		92	watchers dealing with the event loop mechanism itself (C<ev_idle>,
		93	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as
		94	file watchers (C<ev_stat>) and even limited support for fork events
		95	(C<ev_fork>).
		96
		97	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	98	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	99	for example).
33		100
34	=head1 CONVENTIONS	101	=head2 CONVENTIONS
35		102
36	Libev is very configurable. In this manual the default configuration	103	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	104	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	105	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	106	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	107	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 *>)	108	name C<loop> (which is always of type C<ev_loop *>) will not have
42	will not have this argument.	109	this argument.
43		110
44	=head1 TIME AND OTHER GLOBAL FUNCTIONS	111	=head2 TIME REPRESENTATION
45		112
46	Libev represents time as a single floating point number, representing the	113	Libev represents time as a single floating point number, representing the
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	114	(fractional) number of seconds since the (POSIX) epoch (somewhere near
48	the beginning of 1970, details are complicated, don't ask). This type is	115	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	116	called C<ev_tstamp>, which is what you should use too. It usually aliases
50	to the double type in C.	117	to the C<double> type in C, and when you need to do any calculations on
		118	it, you should treat it as some floating point value. Unlike the name
		119	component C<stamp> might indicate, it is also used for time differences
		120	throughout libev.
		121
		122	=head1 ERROR HANDLING
		123
		124	Libev knows three classes of errors: operating system errors, usage errors
		125	and internal errors (bugs).
		126
		127	When libev catches an operating system error it cannot handle (for example
		128	a system call indicating a condition libev cannot fix), it calls the callback
		129	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		130	abort. The default is to print a diagnostic message and to call C<abort
		131	()>.
		132
		133	When libev detects a usage error such as a negative timer interval, then
		134	it will print a diagnostic message and abort (via the C<assert> mechanism,
		135	so C<NDEBUG> will disable this checking): these are programming errors in
		136	the libev caller and need to be fixed there.
		137
		138	Libev also has a few internal error-checking C<assert>ions, and also has
		139	extensive consistency checking code. These do not trigger under normal
		140	circumstances, as they indicate either a bug in libev or worse.
		141
		142
		143	=head1 GLOBAL FUNCTIONS
		144
		145	These functions can be called anytime, even before initialising the
		146	library in any way.
51		147
52	=over 4	148	=over 4
53		149
54	=item ev_tstamp ev_time ()	150	=item ev_tstamp ev_time ()
55		151
56	Returns the current time as libev would use it.	152	Returns the current time as libev would use it. Please note that the
		153	C<ev_now> function is usually faster and also often returns the timestamp
		154	you actually want to know.
		155
		156	=item ev_sleep (ev_tstamp interval)
		157
		158	Sleep for the given interval: The current thread will be blocked until
		159	either it is interrupted or the given time interval has passed. Basically
		160	this is a sub-second-resolution C<sleep ()>.
57		161
58	=item int ev_version_major ()	162	=item int ev_version_major ()
59		163
60	=item int ev_version_minor ()	164	=item int ev_version_minor ()
61		165
62	You can find out the major and minor version numbers of the library	166	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	167	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	168	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	169	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
66	version of the library your program was compiled against.	170	version of the library your program was compiled against.
67		171
		172	These version numbers refer to the ABI version of the library, not the
		173	release version.
		174
68	Usually, its a good idea to terminate if the major versions mismatch,	175	Usually, it's a good idea to terminate if the major versions mismatch,
69	as this indicates an incompatible change. Minor versions are usually	176	as this indicates an incompatible change. Minor versions are usually
70	compatible to older versions, so a larger minor version alone is usually	177	compatible to older versions, so a larger minor version alone is usually
71	not a problem.	178	not a problem.
72		179
		180	Example: Make sure we haven't accidentally been linked against the wrong
		181	version.
		182
		183	assert (("libev version mismatch",
		184	ev_version_major () == EV_VERSION_MAJOR
		185	&& ev_version_minor () >= EV_VERSION_MINOR));
		186
		187	=item unsigned int ev_supported_backends ()
		188
		189	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		190	value) compiled into this binary of libev (independent of their
		191	availability on the system you are running on). See C<ev_default_loop> for
		192	a description of the set values.
		193
		194	Example: make sure we have the epoll method, because yeah this is cool and
		195	a must have and can we have a torrent of it please!!!11
		196
		197	assert (("sorry, no epoll, no sex",
		198	ev_supported_backends () & EVBACKEND_EPOLL));
		199
		200	=item unsigned int ev_recommended_backends ()
		201
		202	Return the set of all backends compiled into this binary of libev and also
		203	recommended for this platform. This set is often smaller than the one
		204	returned by C<ev_supported_backends>, as for example kqueue is broken on
		205	most BSDs and will not be auto-detected unless you explicitly request it
		206	(assuming you know what you are doing). This is the set of backends that
		207	libev will probe for if you specify no backends explicitly.
		208
		209	=item unsigned int ev_embeddable_backends ()
		210
		211	Returns the set of backends that are embeddable in other event loops. This
		212	is the theoretical, all-platform, value. To find which backends
		213	might be supported on the current system, you would need to look at
		214	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
		215	recommended ones.
		216
		217	See the description of C<ev_embed> watchers for more info.
		218
73	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	219	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
74		220
75	Sets the allocation function to use (the prototype is similar to the	221	Sets the allocation function to use (the prototype is similar - the
76	realloc C function, the semantics are identical). It is used to allocate	222	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
77	and free memory (no surprises here). If it returns zero when memory	223	used to allocate and free memory (no surprises here). If it returns zero
78	needs to be allocated, the library might abort or take some potentially	224	when memory needs to be allocated (C<size != 0>), the library might abort
79	destructive action. The default is your system realloc function.	225	or take some potentially destructive action.
		226
		227	Since some systems (at least OpenBSD and Darwin) fail to implement
		228	correct C<realloc> semantics, libev will use a wrapper around the system
		229	C<realloc> and C<free> functions by default.
80		230
81	You could override this function in high-availability programs to, say,	231	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,	232	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.	233	or even to sleep a while and retry until some memory is available.
84		234
		235	Example: Replace the libev allocator with one that waits a bit and then
		236	retries (example requires a standards-compliant C<realloc>).
		237
		238	static void *
		239	persistent_realloc (void *ptr, size_t size)
		240	{
		241	for (;;)
		242	{
		243	void *newptr = realloc (ptr, size);
		244
		245	if (newptr)
		246	return newptr;
		247
		248	sleep (60);
		249	}
		250	}
		251
		252	...
		253	ev_set_allocator (persistent_realloc);
		254
85	=item ev_set_syserr_cb (void (*cb)(const char *msg));	255	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
86		256
87	Set the callback function to call on a retryable syscall error (such	257	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	258	as failed select, poll, epoll_wait). The message is a printable string
89	indicating the system call or subsystem causing the problem. If this	259	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	260	callback is set, then libev will expect it to remedy the situation, no
91	matter what, when it returns. That is, libev will generally retry the	261	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	262	requested operation, or, if the condition doesn't go away, do bad stuff
93	(such as abort).	263	(such as abort).
94		264
		265	Example: This is basically the same thing that libev does internally, too.
		266
		267	static void
		268	fatal_error (const char *msg)
		269	{
		270	perror (msg);
		271	abort ();
		272	}
		273
		274	...
		275	ev_set_syserr_cb (fatal_error);
		276
95	=back	277	=back
96		278
97	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	279	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
98		280
99	An event loop is described by a C<struct ev_loop *>. The library knows two	281	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	282	is I<not> optional in this case, as there is also an C<ev_loop>
101	events, and dynamically created loops which do not.	283	I<function>).
102		284
103	If you use threads, a common model is to run the default event loop	285	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	286	supports signals and child events, and dynamically created loops which do
105	create, you also create another event loop. Libev itself does no locking	287	not.
106	whatsoever, so if you mix calls to the same event loop in different
107	threads, make sure you lock (this is usually a bad idea, though, even if
108	done correctly, because its hideous and inefficient).
109		288
110	=over 4	289	=over 4
111		290
112	=item struct ev_loop *ev_default_loop (unsigned int flags)	291	=item struct ev_loop *ev_default_loop (unsigned int flags)
113		292
114	This will initialise the default event loop if it hasn't been initialised	293	This will initialise the default event loop if it hasn't been initialised
115	yet and return it. If the default loop could not be initialised, returns	294	yet and return it. If the default loop could not be initialised, returns
116	false. If it already was initialised it simply returns it (and ignores the	295	false. If it already was initialised it simply returns it (and ignores the
117	flags).	296	flags. If that is troubling you, check C<ev_backend ()> afterwards).
118		297
119	If you don't know what event loop to use, use the one returned from this	298	If you don't know what event loop to use, use the one returned from this
120	function.	299	function.
121		300
		301	Note that this function is I<not> thread-safe, so if you want to use it
		302	from multiple threads, you have to lock (note also that this is unlikely,
		303	as loops cannot be shared easily between threads anyway).
		304
		305	The default loop is the only loop that can handle C<ev_signal> and
		306	C<ev_child> watchers, and to do this, it always registers a handler
		307	for C<SIGCHLD>. If this is a problem for your application you can either
		308	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
		309	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
		310	C<ev_default_init>.
		311
122	The flags argument can be used to specify special behaviour or specific	312	The flags argument can be used to specify special behaviour or specific
123	backends to use, and is usually specified as 0 (or EVFLAG_AUTO).	313	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
124		314
125	It supports the following flags:	315	The following flags are supported:
126		316
127	=over 4	317	=over 4
128		318
129	=item EVFLAG_AUTO	319	=item C<EVFLAG_AUTO>
130		320
131	The default flags value. Use this if you have no clue (its the right	321	The default flags value. Use this if you have no clue (it's the right
132	thing, believe me).	322	thing, believe me).
133		323
134	=item EVFLAG_NOENV	324	=item C<EVFLAG_NOENV>
135		325
136	If this flag bit is ored into the flag value (or the program runs setuid	326	If this flag bit is or'ed into the flag value (or the program runs setuid
137	or setgid) then libev will I<not> look at the environment variable	327	or setgid) then libev will I<not> look at the environment variable
138	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	328	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
139	override the flags completely if it is found in the environment. This is	329	override the flags completely if it is found in the environment. This is
140	useful to try out specific backends to test their performance, or to work	330	useful to try out specific backends to test their performance, or to work
141	around bugs.	331	around bugs.
142		332
		333	=item C<EVFLAG_FORKCHECK>
		334
		335	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after
		336	a fork, you can also make libev check for a fork in each iteration by
		337	enabling this flag.
		338
		339	This works by calling C<getpid ()> on every iteration of the loop,
		340	and thus this might slow down your event loop if you do a lot of loop
		341	iterations and little real work, but is usually not noticeable (on my
		342	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		343	without a system call and thus I<very> fast, but my GNU/Linux system also has
		344	C<pthread_atfork> which is even faster).
		345
		346	The big advantage of this flag is that you can forget about fork (and
		347	forget about forgetting to tell libev about forking) when you use this
		348	flag.
		349
		350	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		351	environment variable.
		352
143	=item EVMETHOD_SELECT portable select backend	353	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
144		354
145	=item EVMETHOD_POLL poll backend (everywhere except windows)	355	This is your standard select(2) backend. Not I<completely> standard, as
		356	libev tries to roll its own fd_set with no limits on the number of fds,
		357	but if that fails, expect a fairly low limit on the number of fds when
		358	using this backend. It doesn't scale too well (O(highest_fd)), but its
		359	usually the fastest backend for a low number of (low-numbered :) fds.
146		360
147	=item EVMETHOD_EPOLL linux only	361	To get good performance out of this backend you need a high amount of
		362	parallelism (most of the file descriptors should be busy). If you are
		363	writing a server, you should C<accept ()> in a loop to accept as many
		364	connections as possible during one iteration. You might also want to have
		365	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		366	readiness notifications you get per iteration.
148		367
149	=item EVMETHOD_KQUEUE some bsds only	368	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		369	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		370	C<exceptfds> set on that platform).
150		371
151	=item EVMETHOD_DEVPOLL solaris 8 only	372	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
152		373
153	=item EVMETHOD_PORT solaris 10 only	374	And this is your standard poll(2) backend. It's more complicated
		375	than select, but handles sparse fds better and has no artificial
		376	limit on the number of fds you can use (except it will slow down
		377	considerably with a lot of inactive fds). It scales similarly to select,
		378	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		379	performance tips.
154		380
155	If one or more of these are ored into the flags value, then only these	381	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
156	backends will be tried (in the reverse order as given here). If one are	382	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
157	specified, any backend will do.	383
		384	=item C<EVBACKEND_EPOLL> (value 4, Linux)
		385
		386	For few fds, this backend is a bit little slower than poll and select,
		387	but it scales phenomenally better. While poll and select usually scale
		388	like O(total_fds) where n is the total number of fds (or the highest fd),
		389	epoll scales either O(1) or O(active_fds).
		390
		391	The epoll mechanism deserves honorable mention as the most misdesigned
		392	of the more advanced event mechanisms: mere annoyances include silently
		393	dropping file descriptors, requiring a system call per change per file
		394	descriptor (and unnecessary guessing of parameters), problems with dup and
		395	so on. The biggest issue is fork races, however - if a program forks then
		396	I<both> parent and child process have to recreate the epoll set, which can
		397	take considerable time (one syscall per file descriptor) and is of course
		398	hard to detect.
		399
		400	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		401	of course I<doesn't>, and epoll just loves to report events for totally
		402	I<different> file descriptors (even already closed ones, so one cannot
		403	even remove them from the set) than registered in the set (especially
		404	on SMP systems). Libev tries to counter these spurious notifications by
		405	employing an additional generation counter and comparing that against the
		406	events to filter out spurious ones, recreating the set when required.
		407
		408	While stopping, setting and starting an I/O watcher in the same iteration
		409	will result in some caching, there is still a system call per such
		410	incident (because the same I<file descriptor> could point to a different
		411	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
		412	file descriptors might not work very well if you register events for both
		413	file descriptors.
		414
		415	Best performance from this backend is achieved by not unregistering all
		416	watchers for a file descriptor until it has been closed, if possible,
		417	i.e. keep at least one watcher active per fd at all times. Stopping and
		418	starting a watcher (without re-setting it) also usually doesn't cause
		419	extra overhead. A fork can both result in spurious notifications as well
		420	as in libev having to destroy and recreate the epoll object, which can
		421	take considerable time and thus should be avoided.
		422
		423	All this means that, in practise, C<EVBACKEND_SELECT> is as fast or faster
		424	then epoll for maybe up to a hundred file descriptors. So sad.
		425
		426	While nominally embeddable in other event loops, this feature is broken in
		427	all kernel versions tested so far.
		428
		429	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		430	C<EVBACKEND_POLL>.
		431
		432	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
		433
		434	Kqueue deserves special mention, as at the time of this writing, it
		435	was broken on all BSDs except NetBSD (usually it doesn't work reliably
		436	with anything but sockets and pipes, except on Darwin, where of course
		437	it's completely useless). Unlike epoll, however, whose brokenness
		438	is by design, these kqueue bugs can (and eventually will) be fixed
		439	without API changes to existing programs. For this reason it's not being
		440	"auto-detected" unless you explicitly specify it in the flags (i.e. using
		441	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		442	system like NetBSD.
		443
		444	You still can embed kqueue into a normal poll or select backend and use it
		445	only for sockets (after having made sure that sockets work with kqueue on
		446	the target platform). See C<ev_embed> watchers for more info.
		447
		448	It scales in the same way as the epoll backend, but the interface to the
		449	kernel is more efficient (which says nothing about its actual speed, of
		450	course). While stopping, setting and starting an I/O watcher does never
		451	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
		452	two event changes per incident. Support for C<fork ()> is very bad (but
		453	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		454	cases
		455
		456	This backend usually performs well under most conditions.
		457
		458	While nominally embeddable in other event loops, this doesn't work
		459	everywhere, so you might need to test for this. And since it is broken
		460	almost everywhere, you should only use it when you have a lot of sockets
		461	(for which it usually works), by embedding it into another event loop
		462	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and, did I mention it,
		463	using it only for sockets.
		464
		465	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		466	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		467	C<NOTE_EOF>.
		468
		469	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
		470
		471	This is not implemented yet (and might never be, unless you send me an
		472	implementation). According to reports, C</dev/poll> only supports sockets
		473	and is not embeddable, which would limit the usefulness of this backend
		474	immensely.
		475
		476	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
		477
		478	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
		479	it's really slow, but it still scales very well (O(active_fds)).
		480
		481	Please note that Solaris event ports can deliver a lot of spurious
		482	notifications, so you need to use non-blocking I/O or other means to avoid
		483	blocking when no data (or space) is available.
		484
		485	While this backend scales well, it requires one system call per active
		486	file descriptor per loop iteration. For small and medium numbers of file
		487	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		488	might perform better.
		489
		490	On the positive side, with the exception of the spurious readiness
		491	notifications, this backend actually performed fully to specification
		492	in all tests and is fully embeddable, which is a rare feat among the
		493	OS-specific backends (I vastly prefer correctness over speed hacks).
		494
		495	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		496	C<EVBACKEND_POLL>.
		497
		498	=item C<EVBACKEND_ALL>
		499
		500	Try all backends (even potentially broken ones that wouldn't be tried
		501	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
		502	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		503
		504	It is definitely not recommended to use this flag.
158		505
159	=back	506	=back
		507
		508	If one or more of these are or'ed into the flags value, then only these
		509	backends will be tried (in the reverse order as listed here). If none are
		510	specified, all backends in C<ev_recommended_backends ()> will be tried.
		511
		512	Example: This is the most typical usage.
		513
		514	if (!ev_default_loop (0))
		515	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		516
		517	Example: Restrict libev to the select and poll backends, and do not allow
		518	environment settings to be taken into account:
		519
		520	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		521
		522	Example: Use whatever libev has to offer, but make sure that kqueue is
		523	used if available (warning, breaks stuff, best use only with your own
		524	private event loop and only if you know the OS supports your types of
		525	fds):
		526
		527	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
160		528
161	=item struct ev_loop *ev_loop_new (unsigned int flags)	529	=item struct ev_loop *ev_loop_new (unsigned int flags)
162		530
163	Similar to C<ev_default_loop>, but always creates a new event loop that is	531	Similar to C<ev_default_loop>, but always creates a new event loop that is
164	always distinct from the default loop. Unlike the default loop, it cannot	532	always distinct from the default loop. Unlike the default loop, it cannot
165	handle signal and child watchers, and attempts to do so will be greeted by	533	handle signal and child watchers, and attempts to do so will be greeted by
166	undefined behaviour (or a failed assertion if assertions are enabled).	534	undefined behaviour (or a failed assertion if assertions are enabled).
167		535
		536	Note that this function I<is> thread-safe, and the recommended way to use
		537	libev with threads is indeed to create one loop per thread, and using the
		538	default loop in the "main" or "initial" thread.
		539
		540	Example: Try to create a event loop that uses epoll and nothing else.
		541
		542	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
		543	if (!epoller)
		544	fatal ("no epoll found here, maybe it hides under your chair");
		545
168	=item ev_default_destroy ()	546	=item ev_default_destroy ()
169		547
170	Destroys the default loop again (frees all memory and kernel state	548	Destroys the default loop again (frees all memory and kernel state
171	etc.). This stops all registered event watchers (by not touching them in	549	etc.). None of the active event watchers will be stopped in the normal
172	any way whatsoever, although you cnanot rely on this :).	550	sense, so e.g. C<ev_is_active> might still return true. It is your
		551	responsibility to either stop all watchers cleanly yourself I<before>
		552	calling this function, or cope with the fact afterwards (which is usually
		553	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		554	for example).
		555
		556	Note that certain global state, such as signal state (and installed signal
		557	handlers), will not be freed by this function, and related watchers (such
		558	as signal and child watchers) would need to be stopped manually.
		559
		560	In general it is not advisable to call this function except in the
		561	rare occasion where you really need to free e.g. the signal handling
		562	pipe fds. If you need dynamically allocated loops it is better to use
		563	C<ev_loop_new> and C<ev_loop_destroy>).
173		564
174	=item ev_loop_destroy (loop)	565	=item ev_loop_destroy (loop)
175		566
176	Like C<ev_default_destroy>, but destroys an event loop created by an	567	Like C<ev_default_destroy>, but destroys an event loop created by an
177	earlier call to C<ev_loop_new>.	568	earlier call to C<ev_loop_new>.
178		569
179	=item ev_default_fork ()	570	=item ev_default_fork ()
180		571
		572	This function sets a flag that causes subsequent C<ev_loop> iterations
181	This function reinitialises the kernel state for backends that have	573	to reinitialise the kernel state for backends that have one. Despite the
182	one. Despite the name, you can call it anytime, but it makes most sense	574	name, you can call it anytime, but it makes most sense after forking, in
183	after forking, in either the parent or child process (or both, but that	575	the child process (or both child and parent, but that again makes little
184	again makes little sense).	576	sense). You I<must> call it in the child before using any of the libev
		577	functions, and it will only take effect at the next C<ev_loop> iteration.
185		578
186	You I<must> call this function after forking if and only if you want to	579	On the other hand, you only need to call this function in the child
187	use the event library in both processes. If you just fork+exec, you don't	580	process if and only if you want to use the event library in the child. If
188	have to call it.	581	you just fork+exec, you don't have to call it at all.
189		582
190	The function itself is quite fast and its usually not a problem to call	583	The function itself is quite fast and it's usually not a problem to call
191	it just in case after a fork. To make this easy, the function will fit in	584	it just in case after a fork. To make this easy, the function will fit in
192	quite nicely into a call to C<pthread_atfork>:	585	quite nicely into a call to C<pthread_atfork>:
193		586
194	pthread_atfork (0, 0, ev_default_fork);	587	pthread_atfork (0, 0, ev_default_fork);
195		588
196	=item ev_loop_fork (loop)	589	=item ev_loop_fork (loop)
197		590
198	Like C<ev_default_fork>, but acts on an event loop created by	591	Like C<ev_default_fork>, but acts on an event loop created by
199	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	592	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
200	after fork, and how you do this is entirely your own problem.	593	after fork that you want to re-use in the child, and how you do this is
		594	entirely your own problem.
201		595
		596	=item int ev_is_default_loop (loop)
		597
		598	Returns true when the given loop is, in fact, the default loop, and false
		599	otherwise.
		600
		601	=item unsigned int ev_loop_count (loop)
		602
		603	Returns the count of loop iterations for the loop, which is identical to
		604	the number of times libev did poll for new events. It starts at C<0> and
		605	happily wraps around with enough iterations.
		606
		607	This value can sometimes be useful as a generation counter of sorts (it
		608	"ticks" the number of loop iterations), as it roughly corresponds with
		609	C<ev_prepare> and C<ev_check> calls.
		610
202	=item unsigned int ev_method (loop)	611	=item unsigned int ev_backend (loop)
203		612
204	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	613	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
205	use.	614	use.
206		615
207	=item ev_tstamp = ev_now (loop)	616	=item ev_tstamp ev_now (loop)
208		617
209	Returns the current "event loop time", which is the time the event loop	618	Returns the current "event loop time", which is the time the event loop
210	got events and started processing them. This timestamp does not change	619	received events and started processing them. This timestamp does not
211	as long as callbacks are being processed, and this is also the base time	620	change as long as callbacks are being processed, and this is also the base
212	used for relative timers. You can treat it as the timestamp of the event	621	time used for relative timers. You can treat it as the timestamp of the
213	occuring (or more correctly, the mainloop finding out about it).	622	event occurring (or more correctly, libev finding out about it).
		623
		624	=item ev_now_update (loop)
		625
		626	Establishes the current time by querying the kernel, updating the time
		627	returned by C<ev_now ()> in the progress. This is a costly operation and
		628	is usually done automatically within C<ev_loop ()>.
		629
		630	This function is rarely useful, but when some event callback runs for a
		631	very long time without entering the event loop, updating libev's idea of
		632	the current time is a good idea.
		633
		634	See also "The special problem of time updates" in the C<ev_timer> section.
214		635
215	=item ev_loop (loop, int flags)	636	=item ev_loop (loop, int flags)
216		637
217	Finally, this is it, the event handler. This function usually is called	638	Finally, this is it, the event handler. This function usually is called
218	after you initialised all your watchers and you want to start handling	639	after you initialised all your watchers and you want to start handling
219	events.	640	events.
220		641
221	If the flags argument is specified as 0, it will not return until either	642	If the flags argument is specified as C<0>, it will not return until
222	no event watchers are active anymore or C<ev_unloop> was called.	643	either no event watchers are active anymore or C<ev_unloop> was called.
		644
		645	Please note that an explicit C<ev_unloop> is usually better than
		646	relying on all watchers to be stopped when deciding when a program has
		647	finished (especially in interactive programs), but having a program
		648	that automatically loops as long as it has to and no longer by virtue
		649	of relying on its watchers stopping correctly, that is truly a thing of
		650	beauty.
223		651
224	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	652	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
225	those events and any outstanding ones, but will not block your process in	653	those events and any already outstanding ones, but will not block your
226	case there are no events.	654	process in case there are no events and will return after one iteration of
		655	the loop.
227		656
228	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	657	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
229	neccessary) and will handle those and any outstanding ones. It will block	658	necessary) and will handle those and any already outstanding ones. It
230	your process until at least one new event arrives.	659	will block your process until at least one new event arrives (which could
		660	be an event internal to libev itself, so there is no guarantee that a
		661	user-registered callback will be called), and will return after one
		662	iteration of the loop.
231		663
232	This flags value could be used to implement alternative looping	664	This is useful if you are waiting for some external event in conjunction
233	constructs, but the C<prepare> and C<check> watchers provide a better and	665	with something not expressible using other libev watchers (i.e. "roll your
234	more generic mechanism.	666	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		667	usually a better approach for this kind of thing.
		668
		669	Here are the gory details of what C<ev_loop> does:
		670
		671	- Before the first iteration, call any pending watchers.
		672	* If EVFLAG_FORKCHECK was used, check for a fork.
		673	- If a fork was detected (by any means), queue and call all fork watchers.
		674	- Queue and call all prepare watchers.
		675	- If we have been forked, detach and recreate the kernel state
		676	as to not disturb the other process.
		677	- Update the kernel state with all outstanding changes.
		678	- Update the "event loop time" (ev_now ()).
		679	- Calculate for how long to sleep or block, if at all
		680	(active idle watchers, EVLOOP_NONBLOCK or not having
		681	any active watchers at all will result in not sleeping).
		682	- Sleep if the I/O and timer collect interval say so.
		683	- Block the process, waiting for any events.
		684	- Queue all outstanding I/O (fd) events.
		685	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
		686	- Queue all expired timers.
		687	- Queue all expired periodics.
		688	- Unless any events are pending now, queue all idle watchers.
		689	- Queue all check watchers.
		690	- Call all queued watchers in reverse order (i.e. check watchers first).
		691	Signals and child watchers are implemented as I/O watchers, and will
		692	be handled here by queueing them when their watcher gets executed.
		693	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
		694	were used, or there are no active watchers, return, otherwise
		695	continue with step *.
		696
		697	Example: Queue some jobs and then loop until no events are outstanding
		698	anymore.
		699
		700	... queue jobs here, make sure they register event watchers as long
		701	... as they still have work to do (even an idle watcher will do..)
		702	ev_loop (my_loop, 0);
		703	... jobs done or somebody called unloop. yeah!
235		704
236	=item ev_unloop (loop, how)	705	=item ev_unloop (loop, how)
237		706
238	Can be used to make a call to C<ev_loop> return early. The C<how> argument	707	Can be used to make a call to C<ev_loop> return early (but only after it
		708	has processed all outstanding events). The C<how> argument must be either
239	must be either C<EVUNLOOP_ONCE>, which will make the innermost C<ev_loop>	709	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
240	call return, or C<EVUNLOOP_ALL>, which will make all nested C<ev_loop>	710	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
241	calls return.	711
		712	This "unloop state" will be cleared when entering C<ev_loop> again.
		713
		714	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
242		715
243	=item ev_ref (loop)	716	=item ev_ref (loop)
244		717
245	=item ev_unref (loop)	718	=item ev_unref (loop)
246		719
247	Ref/unref can be used to add or remove a refcount on the event loop: Every	720	Ref/unref can be used to add or remove a reference count on the event
248	watcher keeps one reference. If you have a long-runing watcher you never	721	loop: Every watcher keeps one reference, and as long as the reference
249	unregister that should not keep ev_loop from running, ev_unref() after	722	count is nonzero, C<ev_loop> will not return on its own.
250	starting, and ev_ref() before stopping it. Libev itself uses this for	723
251	example for its internal signal pipe: It is not visible to you as a user	724	If you have a watcher you never unregister that should not keep C<ev_loop>
252	and should not keep C<ev_loop> from exiting if the work is done. It is	725	from returning, call ev_unref() after starting, and ev_ref() before
253	also an excellent way to do this for generic recurring timers or from	726	stopping it.
254	within third-party libraries. Just remember to unref after start and ref	727
255	before stop.	728	As an example, libev itself uses this for its internal signal pipe: It is
		729	not visible to the libev user and should not keep C<ev_loop> from exiting
		730	if no event watchers registered by it are active. It is also an excellent
		731	way to do this for generic recurring timers or from within third-party
		732	libraries. Just remember to I<unref after start> and I<ref before stop>
		733	(but only if the watcher wasn't active before, or was active before,
		734	respectively).
		735
		736	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
		737	running when nothing else is active.
		738
		739	ev_signal exitsig;
		740	ev_signal_init (&exitsig, sig_cb, SIGINT);
		741	ev_signal_start (loop, &exitsig);
		742	evf_unref (loop);
		743
		744	Example: For some weird reason, unregister the above signal handler again.
		745
		746	ev_ref (loop);
		747	ev_signal_stop (loop, &exitsig);
		748
		749	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		750
		751	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		752
		753	These advanced functions influence the time that libev will spend waiting
		754	for events. Both time intervals are by default C<0>, meaning that libev
		755	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		756	latency.
		757
		758	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		759	allows libev to delay invocation of I/O and timer/periodic callbacks
		760	to increase efficiency of loop iterations (or to increase power-saving
		761	opportunities).
		762
		763	The idea is that sometimes your program runs just fast enough to handle
		764	one (or very few) event(s) per loop iteration. While this makes the
		765	program responsive, it also wastes a lot of CPU time to poll for new
		766	events, especially with backends like C<select ()> which have a high
		767	overhead for the actual polling but can deliver many events at once.
		768
		769	By setting a higher I<io collect interval> you allow libev to spend more
		770	time collecting I/O events, so you can handle more events per iteration,
		771	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		772	C<ev_timer>) will be not affected. Setting this to a non-null value will
		773	introduce an additional C<ev_sleep ()> call into most loop iterations.
		774
		775	Likewise, by setting a higher I<timeout collect interval> you allow libev
		776	to spend more time collecting timeouts, at the expense of increased
		777	latency/jitter/inexactness (the watcher callback will be called
		778	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		779	value will not introduce any overhead in libev.
		780
		781	Many (busy) programs can usually benefit by setting the I/O collect
		782	interval to a value near C<0.1> or so, which is often enough for
		783	interactive servers (of course not for games), likewise for timeouts. It
		784	usually doesn't make much sense to set it to a lower value than C<0.01>,
		785	as this approaches the timing granularity of most systems.
		786
		787	Setting the I<timeout collect interval> can improve the opportunity for
		788	saving power, as the program will "bundle" timer callback invocations that
		789	are "near" in time together, by delaying some, thus reducing the number of
		790	times the process sleeps and wakes up again. Another useful technique to
		791	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		792	they fire on, say, one-second boundaries only.
		793
		794	=item ev_loop_verify (loop)
		795
		796	This function only does something when C<EV_VERIFY> support has been
		797	compiled in, which is the default for non-minimal builds. It tries to go
		798	through all internal structures and checks them for validity. If anything
		799	is found to be inconsistent, it will print an error message to standard
		800	error and call C<abort ()>.
		801
		802	This can be used to catch bugs inside libev itself: under normal
		803	circumstances, this function will never abort as of course libev keeps its
		804	data structures consistent.
256		805
257	=back	806	=back
258		807
		808
259	=head1 ANATOMY OF A WATCHER	809	=head1 ANATOMY OF A WATCHER
		810
		811	In the following description, uppercase C<TYPE> in names stands for the
		812	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		813	watchers and C<ev_io_start> for I/O watchers.
260		814
261	A watcher is a structure that you create and register to record your	815	A watcher is a structure that you create and register to record your
262	interest in some event. For instance, if you want to wait for STDIN to	816	interest in some event. For instance, if you want to wait for STDIN to
263	become readable, you would create an ev_io watcher for that:	817	become readable, you would create an C<ev_io> watcher for that:
264		818
265	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	819	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
266	{	820	{
267	ev_io_stop (w);	821	ev_io_stop (w);
268	ev_unloop (loop, EVUNLOOP_ALL);	822	ev_unloop (loop, EVUNLOOP_ALL);
269	}	823	}
270		824
271	struct ev_loop *loop = ev_default_loop (0);	825	struct ev_loop *loop = ev_default_loop (0);
		826
272	struct ev_io stdin_watcher;	827	ev_io stdin_watcher;
		828
273	ev_init (&stdin_watcher, my_cb);	829	ev_init (&stdin_watcher, my_cb);
274	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	830	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
275	ev_io_start (loop, &stdin_watcher);	831	ev_io_start (loop, &stdin_watcher);
		832
276	ev_loop (loop, 0);	833	ev_loop (loop, 0);
277		834
278	As you can see, you are responsible for allocating the memory for your	835	As you can see, you are responsible for allocating the memory for your
279	watcher structures (and it is usually a bad idea to do this on the stack,	836	watcher structures (and it is I<usually> a bad idea to do this on the
280	although this can sometimes be quite valid).	837	stack).
		838
		839	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		840	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
281		841
282	Each watcher structure must be initialised by a call to C<ev_init	842	Each watcher structure must be initialised by a call to C<ev_init
283	(watcher *, callback)>, which expects a callback to be provided. This	843	(watcher *, callback)>, which expects a callback to be provided. This
284	callback gets invoked each time the event occurs (or, in the case of io	844	callback gets invoked each time the event occurs (or, in the case of I/O
285	watchers, each time the event loop detects that the file descriptor given	845	watchers, each time the event loop detects that the file descriptor given
286	is readable and/or writable).	846	is readable and/or writable).
287		847
288	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	848	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
289	with arguments specific to this watcher type. There is also a macro	849	macro to configure it, with arguments specific to the watcher type. There
290	to combine initialisation and setting in one call: C<< ev_<type>_init	850	is also a macro to combine initialisation and setting in one call: C<<
291	(watcher *, callback, ...) >>.	851	ev_TYPE_init (watcher *, callback, ...) >>.
292		852
293	To make the watcher actually watch out for events, you have to start it	853	To make the watcher actually watch out for events, you have to start it
294	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	854	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
295	*) >>), and you can stop watching for events at any time by calling the	855	*) >>), and you can stop watching for events at any time by calling the
296	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	856	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
297		857
298	As long as your watcher is active (has been started but not stopped) you	858	As long as your watcher is active (has been started but not stopped) you
299	must not touch the values stored in it. Most specifically you must never	859	must not touch the values stored in it. Most specifically you must never
300	reinitialise it or call its set method.	860	reinitialise it or call its C<ev_TYPE_set> macro.
301
302	You cna check whether an event is active by calling the C<ev_is_active
303	(watcher *)> macro. To see whether an event is outstanding (but the
304	callback for it has not been called yet) you cna use the C<ev_is_pending
305	(watcher *)> macro.
306		861
307	Each and every callback receives the event loop pointer as first, the	862	Each and every callback receives the event loop pointer as first, the
308	registered watcher structure as second, and a bitset of received events as	863	registered watcher structure as second, and a bitset of received events as
309	third argument.	864	third argument.
310		865
311	The rceeived events usually include a single bit per event type received	866	The received events usually include a single bit per event type received
312	(you can receive multiple events at the same time). The possible bit masks	867	(you can receive multiple events at the same time). The possible bit masks
313	are:	868	are:
314		869
315	=over 4	870	=over 4
316		871
317	=item EV_READ	872	=item C<EV_READ>
318		873
319	=item EV_WRITE	874	=item C<EV_WRITE>
320		875
321	The file descriptor in the ev_io watcher has become readable and/or	876	The file descriptor in the C<ev_io> watcher has become readable and/or
322	writable.	877	writable.
323		878
324	=item EV_TIMEOUT	879	=item C<EV_TIMEOUT>
325		880
326	The ev_timer watcher has timed out.	881	The C<ev_timer> watcher has timed out.
327		882
328	=item EV_PERIODIC	883	=item C<EV_PERIODIC>
329		884
330	The ev_periodic watcher has timed out.	885	The C<ev_periodic> watcher has timed out.
331		886
332	=item EV_SIGNAL	887	=item C<EV_SIGNAL>
333		888
334	The signal specified in the ev_signal watcher has been received by a thread.	889	The signal specified in the C<ev_signal> watcher has been received by a thread.
335		890
336	=item EV_CHILD	891	=item C<EV_CHILD>
337		892
338	The pid specified in the ev_child watcher has received a status change.	893	The pid specified in the C<ev_child> watcher has received a status change.
339		894
		895	=item C<EV_STAT>
		896
		897	The path specified in the C<ev_stat> watcher changed its attributes somehow.
		898
340	=item EV_IDLE	899	=item C<EV_IDLE>
341		900
342	The ev_idle watcher has determined that you have nothing better to do.	901	The C<ev_idle> watcher has determined that you have nothing better to do.
343		902
344	=item EV_PREPARE	903	=item C<EV_PREPARE>
345		904
346	=item EV_CHECK	905	=item C<EV_CHECK>
347		906
348	All ev_prepare watchers are invoked just I<before> C<ev_loop> starts	907	All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts
349	to gather new events, and all ev_check watchers are invoked just after	908	to gather new events, and all C<ev_check> watchers are invoked just after
350	C<ev_loop> has gathered them, but before it invokes any callbacks for any	909	C<ev_loop> has gathered them, but before it invokes any callbacks for any
351	received events. Callbacks of both watcher types can start and stop as	910	received events. Callbacks of both watcher types can start and stop as
352	many watchers as they want, and all of them will be taken into account	911	many watchers as they want, and all of them will be taken into account
353	(for example, a ev_prepare watcher might start an idle watcher to keep	912	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
354	C<ev_loop> from blocking).	913	C<ev_loop> from blocking).
355		914
		915	=item C<EV_EMBED>
		916
		917	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		918
		919	=item C<EV_FORK>
		920
		921	The event loop has been resumed in the child process after fork (see
		922	C<ev_fork>).
		923
		924	=item C<EV_ASYNC>
		925
		926	The given async watcher has been asynchronously notified (see C<ev_async>).
		927
356	=item EV_ERROR	928	=item C<EV_ERROR>
357		929
358	An unspecified error has occured, the watcher has been stopped. This might	930	An unspecified error has occurred, the watcher has been stopped. This might
359	happen because the watcher could not be properly started because libev	931	happen because the watcher could not be properly started because libev
360	ran out of memory, a file descriptor was found to be closed or any other	932	ran out of memory, a file descriptor was found to be closed or any other
		933	problem. Libev considers these application bugs.
		934
361	problem. You best act on it by reporting the problem and somehow coping	935	You best act on it by reporting the problem and somehow coping with the
362	with the watcher being stopped.	936	watcher being stopped. Note that well-written programs should not receive
		937	an error ever, so when your watcher receives it, this usually indicates a
		938	bug in your program.
363		939
364	Libev will usually signal a few "dummy" events together with an error,	940	Libev will usually signal a few "dummy" events together with an error, for
365	for example it might indicate that a fd is readable or writable, and if	941	example it might indicate that a fd is readable or writable, and if your
366	your callbacks is well-written it can just attempt the operation and cope	942	callbacks is well-written it can just attempt the operation and cope with
367	with the error from read() or write(). This will not work in multithreaded	943	the error from read() or write(). This will not work in multi-threaded
368	programs, though, so beware.	944	programs, though, as the fd could already be closed and reused for another
		945	thing, so beware.
369		946
370	=back	947	=back
371		948
		949	=head2 GENERIC WATCHER FUNCTIONS
		950
		951	=over 4
		952
		953	=item C<ev_init> (ev_TYPE *watcher, callback)
		954
		955	This macro initialises the generic portion of a watcher. The contents
		956	of the watcher object can be arbitrary (so C<malloc> will do). Only
		957	the generic parts of the watcher are initialised, you I<need> to call
		958	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		959	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		960	which rolls both calls into one.
		961
		962	You can reinitialise a watcher at any time as long as it has been stopped
		963	(or never started) and there are no pending events outstanding.
		964
		965	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		966	int revents)>.
		967
		968	Example: Initialise an C<ev_io> watcher in two steps.
		969
		970	ev_io w;
		971	ev_init (&w, my_cb);
		972	ev_io_set (&w, STDIN_FILENO, EV_READ);
		973
		974	=item C<ev_TYPE_set> (ev_TYPE *, [args])
		975
		976	This macro initialises the type-specific parts of a watcher. You need to
		977	call C<ev_init> at least once before you call this macro, but you can
		978	call C<ev_TYPE_set> any number of times. You must not, however, call this
		979	macro on a watcher that is active (it can be pending, however, which is a
		980	difference to the C<ev_init> macro).
		981
		982	Although some watcher types do not have type-specific arguments
		983	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		984
		985	See C<ev_init>, above, for an example.
		986
		987	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		988
		989	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		990	calls into a single call. This is the most convenient method to initialise
		991	a watcher. The same limitations apply, of course.
		992
		993	Example: Initialise and set an C<ev_io> watcher in one step.
		994
		995	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		996
		997	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)
		998
		999	Starts (activates) the given watcher. Only active watchers will receive
		1000	events. If the watcher is already active nothing will happen.
		1001
		1002	Example: Start the C<ev_io> watcher that is being abused as example in this
		1003	whole section.
		1004
		1005	ev_io_start (EV_DEFAULT_UC, &w);
		1006
		1007	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)
		1008
		1009	Stops the given watcher if active, and clears the pending status (whether
		1010	the watcher was active or not).
		1011
		1012	It is possible that stopped watchers are pending - for example,
		1013	non-repeating timers are being stopped when they become pending - but
		1014	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1015	pending. If you want to free or reuse the memory used by the watcher it is
		1016	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1017
		1018	=item bool ev_is_active (ev_TYPE *watcher)
		1019
		1020	Returns a true value iff the watcher is active (i.e. it has been started
		1021	and not yet been stopped). As long as a watcher is active you must not modify
		1022	it.
		1023
		1024	=item bool ev_is_pending (ev_TYPE *watcher)
		1025
		1026	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1027	events but its callback has not yet been invoked). As long as a watcher
		1028	is pending (but not active) you must not call an init function on it (but
		1029	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1030	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1031	it).
		1032
		1033	=item callback ev_cb (ev_TYPE *watcher)
		1034
		1035	Returns the callback currently set on the watcher.
		1036
		1037	=item ev_cb_set (ev_TYPE *watcher, callback)
		1038
		1039	Change the callback. You can change the callback at virtually any time
		1040	(modulo threads).
		1041
		1042	=item ev_set_priority (ev_TYPE *watcher, priority)
		1043
		1044	=item int ev_priority (ev_TYPE *watcher)
		1045
		1046	Set and query the priority of the watcher. The priority is a small
		1047	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1048	(default: C<-2>). Pending watchers with higher priority will be invoked
		1049	before watchers with lower priority, but priority will not keep watchers
		1050	from being executed (except for C<ev_idle> watchers).
		1051
		1052	This means that priorities are I<only> used for ordering callback
		1053	invocation after new events have been received. This is useful, for
		1054	example, to reduce latency after idling, or more often, to bind two
		1055	watchers on the same event and make sure one is called first.
		1056
		1057	If you need to suppress invocation when higher priority events are pending
		1058	you need to look at C<ev_idle> watchers, which provide this functionality.
		1059
		1060	You I<must not> change the priority of a watcher as long as it is active or
		1061	pending.
		1062
		1063	The default priority used by watchers when no priority has been set is
		1064	always C<0>, which is supposed to not be too high and not be too low :).
		1065
		1066	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1067	fine, as long as you do not mind that the priority value you query might
		1068	or might not have been clamped to the valid range.
		1069
		1070	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1071
		1072	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1073	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1074	can deal with that fact, as both are simply passed through to the
		1075	callback.
		1076
		1077	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1078
		1079	If the watcher is pending, this function clears its pending status and
		1080	returns its C<revents> bitset (as if its callback was invoked). If the
		1081	watcher isn't pending it does nothing and returns C<0>.
		1082
		1083	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1084	callback to be invoked, which can be accomplished with this function.
		1085
		1086	=back
		1087
		1088
372	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1089	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
373		1090
374	Each watcher has, by default, a member C<void *data> that you can change	1091	Each watcher has, by default, a member C<void *data> that you can change
375	and read at any time, libev will completely ignore it. This cna be used	1092	and read at any time: libev will completely ignore it. This can be used
376	to associate arbitrary data with your watcher. If you need more data and	1093	to associate arbitrary data with your watcher. If you need more data and
377	don't want to allocate memory and store a pointer to it in that data	1094	don't want to allocate memory and store a pointer to it in that data
378	member, you can also "subclass" the watcher type and provide your own	1095	member, you can also "subclass" the watcher type and provide your own
379	data:	1096	data:
380		1097
381	struct my_io	1098	struct my_io
382	{	1099	{
383	struct ev_io io;	1100	ev_io io;
384	int otherfd;	1101	int otherfd;
385	void *somedata;	1102	void *somedata;
386	struct whatever *mostinteresting;	1103	struct whatever *mostinteresting;
387	}	1104	};
		1105
		1106	...
		1107	struct my_io w;
		1108	ev_io_init (&w.io, my_cb, fd, EV_READ);
388		1109
389	And since your callback will be called with a pointer to the watcher, you	1110	And since your callback will be called with a pointer to the watcher, you
390	can cast it back to your own type:	1111	can cast it back to your own type:
391		1112
392	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1113	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
393	{	1114	{
394	struct my_io *w = (struct my_io *)w_;	1115	struct my_io *w = (struct my_io *)w_;
395	...	1116	...
396	}	1117	}
397		1118
398	More interesting and less C-conformant ways of catsing your callback type	1119	More interesting and less C-conformant ways of casting your callback type
399	have been omitted....	1120	instead have been omitted.
		1121
		1122	Another common scenario is to use some data structure with multiple
		1123	embedded watchers:
		1124
		1125	struct my_biggy
		1126	{
		1127	int some_data;
		1128	ev_timer t1;
		1129	ev_timer t2;
		1130	}
		1131
		1132	In this case getting the pointer to C<my_biggy> is a bit more
		1133	complicated: Either you store the address of your C<my_biggy> struct
		1134	in the C<data> member of the watcher (for woozies), or you need to use
		1135	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1136	programmers):
		1137
		1138	#include <stddef.h>
		1139
		1140	static void
		1141	t1_cb (EV_P_ ev_timer *w, int revents)
		1142	{
		1143	struct my_biggy big = (struct my_biggy *
		1144	(((char *)w) - offsetof (struct my_biggy, t1));
		1145	}
		1146
		1147	static void
		1148	t2_cb (EV_P_ ev_timer *w, int revents)
		1149	{
		1150	struct my_biggy big = (struct my_biggy *
		1151	(((char *)w) - offsetof (struct my_biggy, t2));
		1152	}
400		1153
401		1154
402	=head1 WATCHER TYPES	1155	=head1 WATCHER TYPES
403		1156
404	This section describes each watcher in detail, but will not repeat	1157	This section describes each watcher in detail, but will not repeat
405	information given in the last section.	1158	information given in the last section. Any initialisation/set macros,
		1159	functions and members specific to the watcher type are explained.
406		1160
		1161	Members are additionally marked with either I<[read-only]>, meaning that,
		1162	while the watcher is active, you can look at the member and expect some
		1163	sensible content, but you must not modify it (you can modify it while the
		1164	watcher is stopped to your hearts content), or I<[read-write]>, which
		1165	means you can expect it to have some sensible content while the watcher
		1166	is active, but you can also modify it. Modifying it may not do something
		1167	sensible or take immediate effect (or do anything at all), but libev will
		1168	not crash or malfunction in any way.
		1169
		1170
407	=head2 struct ev_io - is my file descriptor readable or writable	1171	=head2 C<ev_io> - is this file descriptor readable or writable?
408		1172
409	I/O watchers check whether a file descriptor is readable or writable	1173	I/O watchers check whether a file descriptor is readable or writable
410	in each iteration of the event loop (This behaviour is called	1174	in each iteration of the event loop, or, more precisely, when reading
411	level-triggering because you keep receiving events as long as the	1175	would not block the process and writing would at least be able to write
412	condition persists. Remember you cna stop the watcher if you don't want to	1176	some data. This behaviour is called level-triggering because you keep
413	act on the event and neither want to receive future events).	1177	receiving events as long as the condition persists. Remember you can stop
		1178	the watcher if you don't want to act on the event and neither want to
		1179	receive future events.
414		1180
415	In general you can register as many read and/or write event watchers oer	1181	In general you can register as many read and/or write event watchers per
416	fd as you want (as long as you don't confuse yourself). Setting all file	1182	fd as you want (as long as you don't confuse yourself). Setting all file
417	descriptors to non-blocking mode is also usually a good idea (but not	1183	descriptors to non-blocking mode is also usually a good idea (but not
418	required if you know what you are doing).	1184	required if you know what you are doing).
419		1185
420	You have to be careful with dup'ed file descriptors, though. Some backends	1186	If you cannot use non-blocking mode, then force the use of a
421	(the linux epoll backend is a notable example) cannot handle dup'ed file	1187	known-to-be-good backend (at the time of this writing, this includes only
422	descriptors correctly if you register interest in two or more fds pointing	1188	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>).
423	to the same file/socket etc. description.
424		1189
425	If you must do this, then force the use of a known-to-be-good backend	1190	Another thing you have to watch out for is that it is quite easy to
426	(at the time of this writing, this includes only EVMETHOD_SELECT and	1191	receive "spurious" readiness notifications, that is your callback might
427	EVMETHOD_POLL).	1192	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1193	because there is no data. Not only are some backends known to create a
		1194	lot of those (for example Solaris ports), it is very easy to get into
		1195	this situation even with a relatively standard program structure. Thus
		1196	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1197	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1198
		1199	If you cannot run the fd in non-blocking mode (for example you should
		1200	not play around with an Xlib connection), then you have to separately
		1201	re-test whether a file descriptor is really ready with a known-to-be good
		1202	interface such as poll (fortunately in our Xlib example, Xlib already
		1203	does this on its own, so its quite safe to use). Some people additionally
		1204	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1205	indefinitely.
		1206
		1207	But really, best use non-blocking mode.
		1208
		1209	=head3 The special problem of disappearing file descriptors
		1210
		1211	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1212	descriptor (either due to calling C<close> explicitly or any other means,
		1213	such as C<dup2>). The reason is that you register interest in some file
		1214	descriptor, but when it goes away, the operating system will silently drop
		1215	this interest. If another file descriptor with the same number then is
		1216	registered with libev, there is no efficient way to see that this is, in
		1217	fact, a different file descriptor.
		1218
		1219	To avoid having to explicitly tell libev about such cases, libev follows
		1220	the following policy: Each time C<ev_io_set> is being called, libev
		1221	will assume that this is potentially a new file descriptor, otherwise
		1222	it is assumed that the file descriptor stays the same. That means that
		1223	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1224	descriptor even if the file descriptor number itself did not change.
		1225
		1226	This is how one would do it normally anyway, the important point is that
		1227	the libev application should not optimise around libev but should leave
		1228	optimisations to libev.
		1229
		1230	=head3 The special problem of dup'ed file descriptors
		1231
		1232	Some backends (e.g. epoll), cannot register events for file descriptors,
		1233	but only events for the underlying file descriptions. That means when you
		1234	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1235	events for them, only one file descriptor might actually receive events.
		1236
		1237	There is no workaround possible except not registering events
		1238	for potentially C<dup ()>'ed file descriptors, or to resort to
		1239	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1240
		1241	=head3 The special problem of fork
		1242
		1243	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1244	useless behaviour. Libev fully supports fork, but needs to be told about
		1245	it in the child.
		1246
		1247	To support fork in your programs, you either have to call
		1248	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1249	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1250	C<EVBACKEND_POLL>.
		1251
		1252	=head3 The special problem of SIGPIPE
		1253
		1254	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1255	when writing to a pipe whose other end has been closed, your program gets
		1256	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1257	this is sensible behaviour, for daemons, this is usually undesirable.
		1258
		1259	So when you encounter spurious, unexplained daemon exits, make sure you
		1260	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1261	somewhere, as that would have given you a big clue).
		1262
		1263
		1264	=head3 Watcher-Specific Functions
428		1265
429	=over 4	1266	=over 4
430		1267
431	=item ev_io_init (ev_io *, callback, int fd, int events)	1268	=item ev_io_init (ev_io *, callback, int fd, int events)
432		1269
433	=item ev_io_set (ev_io *, int fd, int events)	1270	=item ev_io_set (ev_io *, int fd, int events)
434		1271
435	Configures an ev_io watcher. The fd is the file descriptor to rceeive	1272	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
436	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1273	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
437	EV_WRITE> to receive the given events.	1274	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1275
		1276	=item int fd [read-only]
		1277
		1278	The file descriptor being watched.
		1279
		1280	=item int events [read-only]
		1281
		1282	The events being watched.
438		1283
439	=back	1284	=back
440		1285
		1286	=head3 Examples
		1287
		1288	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1289	readable, but only once. Since it is likely line-buffered, you could
		1290	attempt to read a whole line in the callback.
		1291
		1292	static void
		1293	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1294	{
		1295	ev_io_stop (loop, w);
		1296	.. read from stdin here (or from w->fd) and handle any I/O errors
		1297	}
		1298
		1299	...
		1300	struct ev_loop *loop = ev_default_init (0);
		1301	ev_io stdin_readable;
		1302	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1303	ev_io_start (loop, &stdin_readable);
		1304	ev_loop (loop, 0);
		1305
		1306
441	=head2 struct ev_timer - relative and optionally recurring timeouts	1307	=head2 C<ev_timer> - relative and optionally repeating timeouts
442		1308
443	Timer watchers are simple relative timers that generate an event after a	1309	Timer watchers are simple relative timers that generate an event after a
444	given time, and optionally repeating in regular intervals after that.	1310	given time, and optionally repeating in regular intervals after that.
445		1311
446	The timers are based on real time, that is, if you register an event that	1312	The timers are based on real time, that is, if you register an event that
447	times out after an hour and youreset your system clock to last years	1313	times out after an hour and you reset your system clock to January last
448	time, it will still time out after (roughly) and hour. "Roughly" because	1314	year, it will still time out after (roughly) one hour. "Roughly" because
449	detecting time jumps is hard, and soem inaccuracies are unavoidable (the	1315	detecting time jumps is hard, and some inaccuracies are unavoidable (the
450	monotonic clock option helps a lot here).	1316	monotonic clock option helps a lot here).
451		1317
		1318	The callback is guaranteed to be invoked only I<after> its timeout has
		1319	passed, but if multiple timers become ready during the same loop iteration
		1320	then order of execution is undefined.
		1321
		1322	=head3 Be smart about timeouts
		1323
		1324	Many real-world problems involve some kind of timeout, usually for error
		1325	recovery. A typical example is an HTTP request - if the other side hangs,
		1326	you want to raise some error after a while.
		1327
		1328	What follows are some ways to handle this problem, from obvious and
		1329	inefficient to smart and efficient.
		1330
		1331	In the following, a 60 second activity timeout is assumed - a timeout that
		1332	gets reset to 60 seconds each time there is activity (e.g. each time some
		1333	data or other life sign was received).
		1334
452	=over 4	1335	=over 4
453		1336
		1337	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1338
		1339	This is the most obvious, but not the most simple way: In the beginning,
		1340	start the watcher:
		1341
		1342	ev_timer_init (timer, callback, 60., 0.);
		1343	ev_timer_start (loop, timer);
		1344
		1345	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1346	and start it again:
		1347
		1348	ev_timer_stop (loop, timer);
		1349	ev_timer_set (timer, 60., 0.);
		1350	ev_timer_start (loop, timer);
		1351
		1352	This is relatively simple to implement, but means that each time there is
		1353	some activity, libev will first have to remove the timer from its internal
		1354	data structure and then add it again. Libev tries to be fast, but it's
		1355	still not a constant-time operation.
		1356
		1357	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1358
		1359	This is the easiest way, and involves using C<ev_timer_again> instead of
		1360	C<ev_timer_start>.
		1361
		1362	To implement this, configure an C<ev_timer> with a C<repeat> value
		1363	of C<60> and then call C<ev_timer_again> at start and each time you
		1364	successfully read or write some data. If you go into an idle state where
		1365	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1366	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1367
		1368	That means you can ignore both the C<ev_timer_start> function and the
		1369	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1370	member and C<ev_timer_again>.
		1371
		1372	At start:
		1373
		1374	ev_timer_init (timer, callback);
		1375	timer->repeat = 60.;
		1376	ev_timer_again (loop, timer);
		1377
		1378	Each time there is some activity:
		1379
		1380	ev_timer_again (loop, timer);
		1381
		1382	It is even possible to change the time-out on the fly, regardless of
		1383	whether the watcher is active or not:
		1384
		1385	timer->repeat = 30.;
		1386	ev_timer_again (loop, timer);
		1387
		1388	This is slightly more efficient then stopping/starting the timer each time
		1389	you want to modify its timeout value, as libev does not have to completely
		1390	remove and re-insert the timer from/into its internal data structure.
		1391
		1392	It is, however, even simpler than the "obvious" way to do it.
		1393
		1394	=item 3. Let the timer time out, but then re-arm it as required.
		1395
		1396	This method is more tricky, but usually most efficient: Most timeouts are
		1397	relatively long compared to the intervals between other activity - in
		1398	our example, within 60 seconds, there are usually many I/O events with
		1399	associated activity resets.
		1400
		1401	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1402	but remember the time of last activity, and check for a real timeout only
		1403	within the callback:
		1404
		1405	ev_tstamp last_activity; // time of last activity
		1406
		1407	static void
		1408	callback (EV_P_ ev_timer *w, int revents)
		1409	{
		1410	ev_tstamp now = ev_now (EV_A);
		1411	ev_tstamp timeout = last_activity + 60.;
		1412
		1413	// if last_activity + 60. is older than now, we did time out
		1414	if (timeout < now)
		1415	{
		1416	// timeout occured, take action
		1417	}
		1418	else
		1419	{
		1420	// callback was invoked, but there was some activity, re-arm
		1421	// the watcher to fire in last_activity + 60, which is
		1422	// guaranteed to be in the future, so "again" is positive:
		1423	w->again = timeout - now;
		1424	ev_timer_again (EV_A_ w);
		1425	}
		1426	}
		1427
		1428	To summarise the callback: first calculate the real timeout (defined
		1429	as "60 seconds after the last activity"), then check if that time has
		1430	been reached, which means something I<did>, in fact, time out. Otherwise
		1431	the callback was invoked too early (C<timeout> is in the future), so
		1432	re-schedule the timer to fire at that future time, to see if maybe we have
		1433	a timeout then.
		1434
		1435	Note how C<ev_timer_again> is used, taking advantage of the
		1436	C<ev_timer_again> optimisation when the timer is already running.
		1437
		1438	This scheme causes more callback invocations (about one every 60 seconds
		1439	minus half the average time between activity), but virtually no calls to
		1440	libev to change the timeout.
		1441
		1442	To start the timer, simply initialise the watcher and set C<last_activity>
		1443	to the current time (meaning we just have some activity :), then call the
		1444	callback, which will "do the right thing" and start the timer:
		1445
		1446	ev_timer_init (timer, callback);
		1447	last_activity = ev_now (loop);
		1448	callback (loop, timer, EV_TIMEOUT);
		1449
		1450	And when there is some activity, simply store the current time in
		1451	C<last_activity>, no libev calls at all:
		1452
		1453	last_actiivty = ev_now (loop);
		1454
		1455	This technique is slightly more complex, but in most cases where the
		1456	time-out is unlikely to be triggered, much more efficient.
		1457
		1458	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1459	callback :) - just change the timeout and invoke the callback, which will
		1460	fix things for you.
		1461
		1462	=item 4. Wee, just use a double-linked list for your timeouts.
		1463
		1464	If there is not one request, but many thousands (millions...), all
		1465	employing some kind of timeout with the same timeout value, then one can
		1466	do even better:
		1467
		1468	When starting the timeout, calculate the timeout value and put the timeout
		1469	at the I<end> of the list.
		1470
		1471	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1472	the list is expected to fire (for example, using the technique #3).
		1473
		1474	When there is some activity, remove the timer from the list, recalculate
		1475	the timeout, append it to the end of the list again, and make sure to
		1476	update the C<ev_timer> if it was taken from the beginning of the list.
		1477
		1478	This way, one can manage an unlimited number of timeouts in O(1) time for
		1479	starting, stopping and updating the timers, at the expense of a major
		1480	complication, and having to use a constant timeout. The constant timeout
		1481	ensures that the list stays sorted.
		1482
		1483	=back
		1484
		1485	So which method the best?
		1486
		1487	Method #2 is a simple no-brain-required solution that is adequate in most
		1488	situations. Method #3 requires a bit more thinking, but handles many cases
		1489	better, and isn't very complicated either. In most case, choosing either
		1490	one is fine, with #3 being better in typical situations.
		1491
		1492	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1493	rather complicated, but extremely efficient, something that really pays
		1494	off after the first million or so of active timers, i.e. it's usually
		1495	overkill :)
		1496
		1497	=head3 The special problem of time updates
		1498
		1499	Establishing the current time is a costly operation (it usually takes at
		1500	least two system calls): EV therefore updates its idea of the current
		1501	time only before and after C<ev_loop> collects new events, which causes a
		1502	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1503	lots of events in one iteration.
		1504
		1505	The relative timeouts are calculated relative to the C<ev_now ()>
		1506	time. This is usually the right thing as this timestamp refers to the time
		1507	of the event triggering whatever timeout you are modifying/starting. If
		1508	you suspect event processing to be delayed and you I<need> to base the
		1509	timeout on the current time, use something like this to adjust for this:
		1510
		1511	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
		1512
		1513	If the event loop is suspended for a long time, you can also force an
		1514	update of the time returned by C<ev_now ()> by calling C<ev_now_update
		1515	()>.
		1516
		1517	=head3 Watcher-Specific Functions and Data Members
		1518
		1519	=over 4
		1520
454	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1521	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
455		1522
456	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1523	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
457		1524
458	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1525	Configure the timer to trigger after C<after> seconds. If C<repeat>
459	C<0.>, then it will automatically be stopped. If it is positive, then the	1526	is C<0.>, then it will automatically be stopped once the timeout is
460	timer will automatically be configured to trigger again C<repeat> seconds	1527	reached. If it is positive, then the timer will automatically be
461	later, again, and again, until stopped manually.	1528	configured to trigger again C<repeat> seconds later, again, and again,
		1529	until stopped manually.
462		1530
463	The timer itself will do a best-effort at avoiding drift, that is, if you	1531	The timer itself will do a best-effort at avoiding drift, that is, if
464	configure a timer to trigger every 10 seconds, then it will trigger at	1532	you configure a timer to trigger every 10 seconds, then it will normally
465	exactly 10 second intervals. If, however, your program cannot keep up with	1533	trigger at exactly 10 second intervals. If, however, your program cannot
466	the timer (ecause it takes longer than those 10 seconds to do stuff) the	1534	keep up with the timer (because it takes longer than those 10 seconds to
467	timer will not fire more than once per event loop iteration.	1535	do stuff) the timer will not fire more than once per event loop iteration.
468		1536
469	=item ev_timer_again (loop)	1537	=item ev_timer_again (loop, ev_timer *)
470		1538
471	This will act as if the timer timed out and restart it again if it is	1539	This will act as if the timer timed out and restart it again if it is
472	repeating. The exact semantics are:	1540	repeating. The exact semantics are:
473		1541
		1542	If the timer is pending, its pending status is cleared.
		1543
474	If the timer is started but nonrepeating, stop it.	1544	If the timer is started but non-repeating, stop it (as if it timed out).
475		1545
476	If the timer is repeating, either start it if necessary (with the repeat	1546	If the timer is repeating, either start it if necessary (with the
477	value), or reset the running timer to the repeat value.	1547	C<repeat> value), or reset the running timer to the C<repeat> value.
478		1548
479	This sounds a bit complicated, but here is a useful and typical	1549	This sounds a bit complicated, see "Be smart about timeouts", above, for a
480	example: Imagine you have a tcp connection and you want a so-called idle	1550	usage example.
481	timeout, that is, you want to be called when there have been, say, 60	1551
482	seconds of inactivity on the socket. The easiest way to do this is to	1552	=item ev_tstamp repeat [read-write]
483	configure an ev_timer with after=repeat=60 and calling ev_timer_again each	1553
484	time you successfully read or write some data. If you go into an idle	1554	The current C<repeat> value. Will be used each time the watcher times out
485	state where you do not expect data to travel on the socket, you can stop	1555	or C<ev_timer_again> is called, and determines the next timeout (if any),
486	the timer, and again will automatically restart it if need be.	1556	which is also when any modifications are taken into account.
487		1557
488	=back	1558	=back
489		1559
		1560	=head3 Examples
		1561
		1562	Example: Create a timer that fires after 60 seconds.
		1563
		1564	static void
		1565	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1566	{
		1567	.. one minute over, w is actually stopped right here
		1568	}
		1569
		1570	ev_timer mytimer;
		1571	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		1572	ev_timer_start (loop, &mytimer);
		1573
		1574	Example: Create a timeout timer that times out after 10 seconds of
		1575	inactivity.
		1576
		1577	static void
		1578	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1579	{
		1580	.. ten seconds without any activity
		1581	}
		1582
		1583	ev_timer mytimer;
		1584	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		1585	ev_timer_again (&mytimer); /* start timer */
		1586	ev_loop (loop, 0);
		1587
		1588	// and in some piece of code that gets executed on any "activity":
		1589	// reset the timeout to start ticking again at 10 seconds
		1590	ev_timer_again (&mytimer);
		1591
		1592
490	=head2 ev_periodic - to cron or not to cron it	1593	=head2 C<ev_periodic> - to cron or not to cron?
491		1594
492	Periodic watchers are also timers of a kind, but they are very versatile	1595	Periodic watchers are also timers of a kind, but they are very versatile
493	(and unfortunately a bit complex).	1596	(and unfortunately a bit complex).
494		1597
495	Unlike ev_timer's, they are not based on real time (or relative time)	1598	Unlike C<ev_timer>'s, they are not based on real time (or relative time)
496	but on wallclock time (absolute time). You can tell a periodic watcher	1599	but on wall clock time (absolute time). You can tell a periodic watcher
497	to trigger "at" some specific point in time. For example, if you tell a	1600	to trigger after some specific point in time. For example, if you tell a
498	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	1601	periodic watcher to trigger in 10 seconds (by specifying e.g. C<ev_now ()
499	+ 10.>) and then reset your system clock to the last year, then it will	1602	+ 10.>, that is, an absolute time not a delay) and then reset your system
		1603	clock to January of the previous year, then it will take more than year
500	take a year to trigger the event (unlike an ev_timer, which would trigger	1604	to trigger the event (unlike an C<ev_timer>, which would still trigger
501	roughly 10 seconds later and of course not if you reset your system time	1605	roughly 10 seconds later as it uses a relative timeout).
502	again).
503		1606
504	They can also be used to implement vastly more complex timers, such as	1607	C<ev_periodic>s can also be used to implement vastly more complex timers,
505	triggering an event on eahc midnight, local time.	1608	such as triggering an event on each "midnight, local time", or other
		1609	complicated rules.
		1610
		1611	As with timers, the callback is guaranteed to be invoked only when the
		1612	time (C<at>) has passed, but if multiple periodic timers become ready
		1613	during the same loop iteration, then order of execution is undefined.
		1614
		1615	=head3 Watcher-Specific Functions and Data Members
506		1616
507	=over 4	1617	=over 4
508		1618
509	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1619	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
510		1620
511	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1621	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
512		1622
513	Lots of arguments, lets sort it out... There are basically three modes of	1623	Lots of arguments, lets sort it out... There are basically three modes of
514	operation, and we will explain them from simplest to complex:	1624	operation, and we will explain them from simplest to most complex:
515
516		1625
517	=over 4	1626	=over 4
518		1627
519	=item * absolute timer (interval = reschedule_cb = 0)	1628	=item * absolute timer (at = time, interval = reschedule_cb = 0)
520		1629
521	In this configuration the watcher triggers an event at the wallclock time	1630	In this configuration the watcher triggers an event after the wall clock
522	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1631	time C<at> has passed. It will not repeat and will not adjust when a time
523	that is, if it is to be run at January 1st 2011 then it will run when the	1632	jump occurs, that is, if it is to be run at January 1st 2011 then it will
524	system time reaches or surpasses this time.	1633	only run when the system clock reaches or surpasses this time.
525		1634
526	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1635	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
527		1636
528	In this mode the watcher will always be scheduled to time out at the next	1637	In this mode the watcher will always be scheduled to time out at the next
529	C<at + N * interval> time (for some integer N) and then repeat, regardless	1638	C<at + N * interval> time (for some integer N, which can also be negative)
530	of any time jumps.	1639	and then repeat, regardless of any time jumps.
531		1640
532	This can be used to create timers that do not drift with respect to system	1641	This can be used to create timers that do not drift with respect to the
533	time:	1642	system clock, for example, here is a C<ev_periodic> that triggers each
		1643	hour, on the hour:
534		1644
535	ev_periodic_set (&periodic, 0., 3600., 0);	1645	ev_periodic_set (&periodic, 0., 3600., 0);
536		1646
537	This doesn't mean there will always be 3600 seconds in between triggers,	1647	This doesn't mean there will always be 3600 seconds in between triggers,
538	but only that the the callback will be called when the system time shows a	1648	but only that the callback will be called when the system time shows a
539	full hour (UTC), or more correct, when the system time is evenly divisible	1649	full hour (UTC), or more correctly, when the system time is evenly divisible
540	by 3600.	1650	by 3600.
541		1651
542	Another way to think about it (for the mathematically inclined) is that	1652	Another way to think about it (for the mathematically inclined) is that
543	ev_periodic will try to run the callback in this mode at the next possible	1653	C<ev_periodic> will try to run the callback in this mode at the next possible
544	time where C<time = at (mod interval)>, regardless of any time jumps.	1654	time where C<time = at (mod interval)>, regardless of any time jumps.
545		1655
		1656	For numerical stability it is preferable that the C<at> value is near
		1657	C<ev_now ()> (the current time), but there is no range requirement for
		1658	this value, and in fact is often specified as zero.
		1659
		1660	Note also that there is an upper limit to how often a timer can fire (CPU
		1661	speed for example), so if C<interval> is very small then timing stability
		1662	will of course deteriorate. Libev itself tries to be exact to be about one
		1663	millisecond (if the OS supports it and the machine is fast enough).
		1664
546	=item * manual reschedule mode (reschedule_cb = callback)	1665	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
547		1666
548	In this mode the values for C<interval> and C<at> are both being	1667	In this mode the values for C<interval> and C<at> are both being
549	ignored. Instead, each time the periodic watcher gets scheduled, the	1668	ignored. Instead, each time the periodic watcher gets scheduled, the
550	reschedule callback will be called with the watcher as first, and the	1669	reschedule callback will be called with the watcher as first, and the
551	current time as second argument.	1670	current time as second argument.
552		1671
553	NOTE: I<This callback MUST NOT stop or destroy the periodic or any other	1672	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
554	periodic watcher, ever, or make any event loop modificstions>. If you need	1673	ever, or make ANY event loop modifications whatsoever>.
555	to stop it, return 1e30 (or so, fudge fudge) and stop it afterwards.
556		1674
		1675	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		1676	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		1677	only event loop modification you are allowed to do).
		1678
557	Its prototype is c<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	1679	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
558	ev_tstamp now)>, e.g.:	1680	*w, ev_tstamp now)>, e.g.:
559		1681
		1682	static ev_tstamp
560	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1683	my_rescheduler (ev_periodic *w, ev_tstamp now)
561	{	1684	{
562	return now + 60.;	1685	return now + 60.;
563	}	1686	}
564		1687
565	It must return the next time to trigger, based on the passed time value	1688	It must return the next time to trigger, based on the passed time value
566	(that is, the lowest time value larger than to the second argument). It	1689	(that is, the lowest time value larger than to the second argument). It
567	will usually be called just before the callback will be triggered, but	1690	will usually be called just before the callback will be triggered, but
568	might be called at other times, too.	1691	might be called at other times, too.
569		1692
		1693	NOTE: I<< This callback must always return a time that is higher than or
		1694	equal to the passed C<now> value >>.
		1695
570	This can be used to create very complex timers, such as a timer that	1696	This can be used to create very complex timers, such as a timer that
571	triggers on each midnight, local time. To do this, you would calculate the	1697	triggers on "next midnight, local time". To do this, you would calculate the
572	next midnight after C<now> and return the timestamp value for this. How you do this	1698	next midnight after C<now> and return the timestamp value for this. How
573	is, again, up to you (but it is not trivial).	1699	you do this is, again, up to you (but it is not trivial, which is the main
		1700	reason I omitted it as an example).
574		1701
575	=back	1702	=back
576		1703
577	=item ev_periodic_again (loop, ev_periodic *)	1704	=item ev_periodic_again (loop, ev_periodic *)
578		1705
579	Simply stops and restarts the periodic watcher again. This is only useful	1706	Simply stops and restarts the periodic watcher again. This is only useful
580	when you changed some parameters or the reschedule callback would return	1707	when you changed some parameters or the reschedule callback would return
581	a different time than the last time it was called (e.g. in a crond like	1708	a different time than the last time it was called (e.g. in a crond like
582	program when the crontabs have changed).	1709	program when the crontabs have changed).
583		1710
		1711	=item ev_tstamp ev_periodic_at (ev_periodic *)
		1712
		1713	When active, returns the absolute time that the watcher is supposed to
		1714	trigger next.
		1715
		1716	=item ev_tstamp offset [read-write]
		1717
		1718	When repeating, this contains the offset value, otherwise this is the
		1719	absolute point in time (the C<at> value passed to C<ev_periodic_set>).
		1720
		1721	Can be modified any time, but changes only take effect when the periodic
		1722	timer fires or C<ev_periodic_again> is being called.
		1723
		1724	=item ev_tstamp interval [read-write]
		1725
		1726	The current interval value. Can be modified any time, but changes only
		1727	take effect when the periodic timer fires or C<ev_periodic_again> is being
		1728	called.
		1729
		1730	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		1731
		1732	The current reschedule callback, or C<0>, if this functionality is
		1733	switched off. Can be changed any time, but changes only take effect when
		1734	the periodic timer fires or C<ev_periodic_again> is being called.
		1735
584	=back	1736	=back
585		1737
		1738	=head3 Examples
		1739
		1740	Example: Call a callback every hour, or, more precisely, whenever the
		1741	system time is divisible by 3600. The callback invocation times have
		1742	potentially a lot of jitter, but good long-term stability.
		1743
		1744	static void
		1745	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
		1746	{
		1747	... its now a full hour (UTC, or TAI or whatever your clock follows)
		1748	}
		1749
		1750	ev_periodic hourly_tick;
		1751	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		1752	ev_periodic_start (loop, &hourly_tick);
		1753
		1754	Example: The same as above, but use a reschedule callback to do it:
		1755
		1756	#include <math.h>
		1757
		1758	static ev_tstamp
		1759	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		1760	{
		1761	return now + (3600. - fmod (now, 3600.));
		1762	}
		1763
		1764	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		1765
		1766	Example: Call a callback every hour, starting now:
		1767
		1768	ev_periodic hourly_tick;
		1769	ev_periodic_init (&hourly_tick, clock_cb,
		1770	fmod (ev_now (loop), 3600.), 3600., 0);
		1771	ev_periodic_start (loop, &hourly_tick);
		1772
		1773
586	=head2 ev_signal - signal me when a signal gets signalled	1774	=head2 C<ev_signal> - signal me when a signal gets signalled!
587		1775
588	Signal watchers will trigger an event when the process receives a specific	1776	Signal watchers will trigger an event when the process receives a specific
589	signal one or more times. Even though signals are very asynchronous, libev	1777	signal one or more times. Even though signals are very asynchronous, libev
590	will try its best to deliver signals synchronously, i.e. as part of the	1778	will try it's best to deliver signals synchronously, i.e. as part of the
591	normal event processing, like any other event.	1779	normal event processing, like any other event.
592		1780
		1781	If you want signals asynchronously, just use C<sigaction> as you would
		1782	do without libev and forget about sharing the signal. You can even use
		1783	C<ev_async> from a signal handler to synchronously wake up an event loop.
		1784
593	You cna configure as many watchers as you like per signal. Only when the	1785	You can configure as many watchers as you like per signal. Only when the
594	first watcher gets started will libev actually register a signal watcher	1786	first watcher gets started will libev actually register a signal handler
595	with the kernel (thus it coexists with your own signal handlers as long	1787	with the kernel (thus it coexists with your own signal handlers as long as
596	as you don't register any with libev). Similarly, when the last signal	1788	you don't register any with libev for the same signal). Similarly, when
597	watcher for a signal is stopped libev will reset the signal handler to	1789	the last signal watcher for a signal is stopped, libev will reset the
598	SIG_DFL (regardless of what it was set to before).	1790	signal handler to SIG_DFL (regardless of what it was set to before).
		1791
		1792	If possible and supported, libev will install its handlers with
		1793	C<SA_RESTART> behaviour enabled, so system calls should not be unduly
		1794	interrupted. If you have a problem with system calls getting interrupted by
		1795	signals you can block all signals in an C<ev_check> watcher and unblock
		1796	them in an C<ev_prepare> watcher.
		1797
		1798	=head3 Watcher-Specific Functions and Data Members
599		1799
600	=over 4	1800	=over 4
601		1801
602	=item ev_signal_init (ev_signal *, callback, int signum)	1802	=item ev_signal_init (ev_signal *, callback, int signum)
603		1803
604	=item ev_signal_set (ev_signal *, int signum)	1804	=item ev_signal_set (ev_signal *, int signum)
605		1805
606	Configures the watcher to trigger on the given signal number (usually one	1806	Configures the watcher to trigger on the given signal number (usually one
607	of the C<SIGxxx> constants).	1807	of the C<SIGxxx> constants).
608		1808
		1809	=item int signum [read-only]
		1810
		1811	The signal the watcher watches out for.
		1812
609	=back	1813	=back
610		1814
		1815	=head3 Examples
		1816
		1817	Example: Try to exit cleanly on SIGINT.
		1818
		1819	static void
		1820	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		1821	{
		1822	ev_unloop (loop, EVUNLOOP_ALL);
		1823	}
		1824
		1825	ev_signal signal_watcher;
		1826	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		1827	ev_signal_start (loop, &signal_watcher);
		1828
		1829
611	=head2 ev_child - wait for pid status changes	1830	=head2 C<ev_child> - watch out for process status changes
612		1831
613	Child watchers trigger when your process receives a SIGCHLD in response to	1832	Child watchers trigger when your process receives a SIGCHLD in response to
614	some child status changes (most typically when a child of yours dies).	1833	some child status changes (most typically when a child of yours dies or
		1834	exits). It is permissible to install a child watcher I<after> the child
		1835	has been forked (which implies it might have already exited), as long
		1836	as the event loop isn't entered (or is continued from a watcher), i.e.,
		1837	forking and then immediately registering a watcher for the child is fine,
		1838	but forking and registering a watcher a few event loop iterations later is
		1839	not.
		1840
		1841	Only the default event loop is capable of handling signals, and therefore
		1842	you can only register child watchers in the default event loop.
		1843
		1844	=head3 Process Interaction
		1845
		1846	Libev grabs C<SIGCHLD> as soon as the default event loop is
		1847	initialised. This is necessary to guarantee proper behaviour even if
		1848	the first child watcher is started after the child exits. The occurrence
		1849	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		1850	synchronously as part of the event loop processing. Libev always reaps all
		1851	children, even ones not watched.
		1852
		1853	=head3 Overriding the Built-In Processing
		1854
		1855	Libev offers no special support for overriding the built-in child
		1856	processing, but if your application collides with libev's default child
		1857	handler, you can override it easily by installing your own handler for
		1858	C<SIGCHLD> after initialising the default loop, and making sure the
		1859	default loop never gets destroyed. You are encouraged, however, to use an
		1860	event-based approach to child reaping and thus use libev's support for
		1861	that, so other libev users can use C<ev_child> watchers freely.
		1862
		1863	=head3 Stopping the Child Watcher
		1864
		1865	Currently, the child watcher never gets stopped, even when the
		1866	child terminates, so normally one needs to stop the watcher in the
		1867	callback. Future versions of libev might stop the watcher automatically
		1868	when a child exit is detected.
		1869
		1870	=head3 Watcher-Specific Functions and Data Members
615		1871
616	=over 4	1872	=over 4
617		1873
618	=item ev_child_init (ev_child *, callback, int pid)	1874	=item ev_child_init (ev_child *, callback, int pid, int trace)
619		1875
620	=item ev_child_set (ev_child *, int pid)	1876	=item ev_child_set (ev_child *, int pid, int trace)
621		1877
622	Configures the watcher to wait for status changes of process C<pid> (or	1878	Configures the watcher to wait for status changes of process C<pid> (or
623	I<any> process if C<pid> is specified as C<0>). The callback can look	1879	I<any> process if C<pid> is specified as C<0>). The callback can look
624	at the C<rstatus> member of the C<ev_child> watcher structure to see	1880	at the C<rstatus> member of the C<ev_child> watcher structure to see
625	the status word (use the macros from C<sys/wait.h>). The C<rpid> member	1881	the status word (use the macros from C<sys/wait.h> and see your systems
626	contains the pid of the process causing the status change.	1882	C<waitpid> documentation). The C<rpid> member contains the pid of the
		1883	process causing the status change. C<trace> must be either C<0> (only
		1884	activate the watcher when the process terminates) or C<1> (additionally
		1885	activate the watcher when the process is stopped or continued).
		1886
		1887	=item int pid [read-only]
		1888
		1889	The process id this watcher watches out for, or C<0>, meaning any process id.
		1890
		1891	=item int rpid [read-write]
		1892
		1893	The process id that detected a status change.
		1894
		1895	=item int rstatus [read-write]
		1896
		1897	The process exit/trace status caused by C<rpid> (see your systems
		1898	C<waitpid> and C<sys/wait.h> documentation for details).
627		1899
628	=back	1900	=back
629		1901
		1902	=head3 Examples
		1903
		1904	Example: C<fork()> a new process and install a child handler to wait for
		1905	its completion.
		1906
		1907	ev_child cw;
		1908
		1909	static void
		1910	child_cb (EV_P_ ev_child *w, int revents)
		1911	{
		1912	ev_child_stop (EV_A_ w);
		1913	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		1914	}
		1915
		1916	pid_t pid = fork ();
		1917
		1918	if (pid < 0)
		1919	// error
		1920	else if (pid == 0)
		1921	{
		1922	// the forked child executes here
		1923	exit (1);
		1924	}
		1925	else
		1926	{
		1927	ev_child_init (&cw, child_cb, pid, 0);
		1928	ev_child_start (EV_DEFAULT_ &cw);
		1929	}
		1930
		1931
		1932	=head2 C<ev_stat> - did the file attributes just change?
		1933
		1934	This watches a file system path for attribute changes. That is, it calls
		1935	C<stat> on that path in regular intervals (or when the OS says it changed)
		1936	and sees if it changed compared to the last time, invoking the callback if
		1937	it did.
		1938
		1939	The path does not need to exist: changing from "path exists" to "path does
		1940	not exist" is a status change like any other. The condition "path does not
		1941	exist" (or more correctly "path cannot be stat'ed") is signified by the
		1942	C<st_nlink> field being zero (which is otherwise always forced to be at
		1943	least one) and all the other fields of the stat buffer having unspecified
		1944	contents.
		1945
		1946	The path I<must not> end in a slash or contain special components such as
		1947	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		1948	your working directory changes, then the behaviour is undefined.
		1949
		1950	Since there is no portable change notification interface available, the
		1951	portable implementation simply calls C<stat(2)> regularly on the path
		1952	to see if it changed somehow. You can specify a recommended polling
		1953	interval for this case. If you specify a polling interval of C<0> (highly
		1954	recommended!) then a I<suitable, unspecified default> value will be used
		1955	(which you can expect to be around five seconds, although this might
		1956	change dynamically). Libev will also impose a minimum interval which is
		1957	currently around C<0.1>, but that's usually overkill.
		1958
		1959	This watcher type is not meant for massive numbers of stat watchers,
		1960	as even with OS-supported change notifications, this can be
		1961	resource-intensive.
		1962
		1963	At the time of this writing, the only OS-specific interface implemented
		1964	is the Linux inotify interface (implementing kqueue support is left as an
		1965	exercise for the reader. Note, however, that the author sees no way of
		1966	implementing C<ev_stat> semantics with kqueue, except as a hint).
		1967
		1968	=head3 ABI Issues (Largefile Support)
		1969
		1970	Libev by default (unless the user overrides this) uses the default
		1971	compilation environment, which means that on systems with large file
		1972	support disabled by default, you get the 32 bit version of the stat
		1973	structure. When using the library from programs that change the ABI to
		1974	use 64 bit file offsets the programs will fail. In that case you have to
		1975	compile libev with the same flags to get binary compatibility. This is
		1976	obviously the case with any flags that change the ABI, but the problem is
		1977	most noticeably displayed with ev_stat and large file support.
		1978
		1979	The solution for this is to lobby your distribution maker to make large
		1980	file interfaces available by default (as e.g. FreeBSD does) and not
		1981	optional. Libev cannot simply switch on large file support because it has
		1982	to exchange stat structures with application programs compiled using the
		1983	default compilation environment.
		1984
		1985	=head3 Inotify and Kqueue
		1986
		1987	When C<inotify (7)> support has been compiled into libev and present at
		1988	runtime, it will be used to speed up change detection where possible. The
		1989	inotify descriptor will be created lazily when the first C<ev_stat>
		1990	watcher is being started.
		1991
		1992	Inotify presence does not change the semantics of C<ev_stat> watchers
		1993	except that changes might be detected earlier, and in some cases, to avoid
		1994	making regular C<stat> calls. Even in the presence of inotify support
		1995	there are many cases where libev has to resort to regular C<stat> polling,
		1996	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		1997	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		1998	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		1999	xfs are fully working) libev usually gets away without polling.
		2000
		2001	There is no support for kqueue, as apparently it cannot be used to
		2002	implement this functionality, due to the requirement of having a file
		2003	descriptor open on the object at all times, and detecting renames, unlinks
		2004	etc. is difficult.
		2005
		2006	=head3 C<stat ()> is a synchronous operation
		2007
		2008	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2009	the process. The exception are C<ev_stat> watchers - those call C<stat
		2010	()>, which is a synchronous operation.
		2011
		2012	For local paths, this usually doesn't matter: unless the system is very
		2013	busy or the intervals between stat's are large, a stat call will be fast,
		2014	as the path data is suually in memory already (except when starting the
		2015	watcher).
		2016
		2017	For networked file systems, calling C<stat ()> can block an indefinite
		2018	time due to network issues, and even under good conditions, a stat call
		2019	often takes multiple milliseconds.
		2020
		2021	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2022	paths, although this is fully supported by libev.
		2023
		2024	=head3 The special problem of stat time resolution
		2025
		2026	The C<stat ()> system call only supports full-second resolution portably,
		2027	and even on systems where the resolution is higher, most file systems
		2028	still only support whole seconds.
		2029
		2030	That means that, if the time is the only thing that changes, you can
		2031	easily miss updates: on the first update, C<ev_stat> detects a change and
		2032	calls your callback, which does something. When there is another update
		2033	within the same second, C<ev_stat> will be unable to detect unless the
		2034	stat data does change in other ways (e.g. file size).
		2035
		2036	The solution to this is to delay acting on a change for slightly more
		2037	than a second (or till slightly after the next full second boundary), using
		2038	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2039	ev_timer_again (loop, w)>).
		2040
		2041	The C<.02> offset is added to work around small timing inconsistencies
		2042	of some operating systems (where the second counter of the current time
		2043	might be be delayed. One such system is the Linux kernel, where a call to
		2044	C<gettimeofday> might return a timestamp with a full second later than
		2045	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2046	update file times then there will be a small window where the kernel uses
		2047	the previous second to update file times but libev might already execute
		2048	the timer callback).
		2049
		2050	=head3 Watcher-Specific Functions and Data Members
		2051
		2052	=over 4
		2053
		2054	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2055
		2056	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2057
		2058	Configures the watcher to wait for status changes of the given
		2059	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2060	be detected and should normally be specified as C<0> to let libev choose
		2061	a suitable value. The memory pointed to by C<path> must point to the same
		2062	path for as long as the watcher is active.
		2063
		2064	The callback will receive an C<EV_STAT> event when a change was detected,
		2065	relative to the attributes at the time the watcher was started (or the
		2066	last change was detected).
		2067
		2068	=item ev_stat_stat (loop, ev_stat *)
		2069
		2070	Updates the stat buffer immediately with new values. If you change the
		2071	watched path in your callback, you could call this function to avoid
		2072	detecting this change (while introducing a race condition if you are not
		2073	the only one changing the path). Can also be useful simply to find out the
		2074	new values.
		2075
		2076	=item ev_statdata attr [read-only]
		2077
		2078	The most-recently detected attributes of the file. Although the type is
		2079	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2080	suitable for your system, but you can only rely on the POSIX-standardised
		2081	members to be present. If the C<st_nlink> member is C<0>, then there was
		2082	some error while C<stat>ing the file.
		2083
		2084	=item ev_statdata prev [read-only]
		2085
		2086	The previous attributes of the file. The callback gets invoked whenever
		2087	C<prev> != C<attr>, or, more precisely, one or more of these members
		2088	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2089	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2090
		2091	=item ev_tstamp interval [read-only]
		2092
		2093	The specified interval.
		2094
		2095	=item const char *path [read-only]
		2096
		2097	The file system path that is being watched.
		2098
		2099	=back
		2100
		2101	=head3 Examples
		2102
		2103	Example: Watch C</etc/passwd> for attribute changes.
		2104
		2105	static void
		2106	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2107	{
		2108	/* /etc/passwd changed in some way */
		2109	if (w->attr.st_nlink)
		2110	{
		2111	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2112	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2113	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2114	}
		2115	else
		2116	/* you shalt not abuse printf for puts */
		2117	puts ("wow, /etc/passwd is not there, expect problems. "
		2118	"if this is windows, they already arrived\n");
		2119	}
		2120
		2121	...
		2122	ev_stat passwd;
		2123
		2124	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2125	ev_stat_start (loop, &passwd);
		2126
		2127	Example: Like above, but additionally use a one-second delay so we do not
		2128	miss updates (however, frequent updates will delay processing, too, so
		2129	one might do the work both on C<ev_stat> callback invocation I<and> on
		2130	C<ev_timer> callback invocation).
		2131
		2132	static ev_stat passwd;
		2133	static ev_timer timer;
		2134
		2135	static void
		2136	timer_cb (EV_P_ ev_timer *w, int revents)
		2137	{
		2138	ev_timer_stop (EV_A_ w);
		2139
		2140	/* now it's one second after the most recent passwd change */
		2141	}
		2142
		2143	static void
		2144	stat_cb (EV_P_ ev_stat *w, int revents)
		2145	{
		2146	/* reset the one-second timer */
		2147	ev_timer_again (EV_A_ &timer);
		2148	}
		2149
		2150	...
		2151	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2152	ev_stat_start (loop, &passwd);
		2153	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2154
		2155
630	=head2 ev_idle - when you've got nothing better to do	2156	=head2 C<ev_idle> - when you've got nothing better to do...
631		2157
632	Idle watchers trigger events when there are no other I/O or timer (or	2158	Idle watchers trigger events when no other events of the same or higher
633	periodic) events pending. That is, as long as your process is busy	2159	priority are pending (prepare, check and other idle watchers do not count
634	handling sockets or timeouts it will not be called. But when your process	2160	as receiving "events").
635	is idle all idle watchers are being called again and again - until	2161
		2162	That is, as long as your process is busy handling sockets or timeouts
		2163	(or even signals, imagine) of the same or higher priority it will not be
		2164	triggered. But when your process is idle (or only lower-priority watchers
		2165	are pending), the idle watchers are being called once per event loop
636	stopped, that is, or your process receives more events.	2166	iteration - until stopped, that is, or your process receives more events
		2167	and becomes busy again with higher priority stuff.
637		2168
638	The most noteworthy effect is that as long as any idle watchers are	2169	The most noteworthy effect is that as long as any idle watchers are
639	active, the process will not block when waiting for new events.	2170	active, the process will not block when waiting for new events.
640		2171
641	Apart from keeping your process non-blocking (which is a useful	2172	Apart from keeping your process non-blocking (which is a useful
642	effect on its own sometimes), idle watchers are a good place to do	2173	effect on its own sometimes), idle watchers are a good place to do
643	"pseudo-background processing", or delay processing stuff to after the	2174	"pseudo-background processing", or delay processing stuff to after the
644	event loop has handled all outstanding events.	2175	event loop has handled all outstanding events.
645		2176
		2177	=head3 Watcher-Specific Functions and Data Members
		2178
646	=over 4	2179	=over 4
647		2180
648	=item ev_idle_init (ev_signal *, callback)	2181	=item ev_idle_init (ev_signal *, callback)
649		2182
650	Initialises and configures the idle watcher - it has no parameters of any	2183	Initialises and configures the idle watcher - it has no parameters of any
651	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2184	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
652	believe me.	2185	believe me.
653		2186
654	=back	2187	=back
655		2188
656	=head2 prepare and check - your hooks into the event loop	2189	=head3 Examples
657		2190
		2191	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2192	callback, free it. Also, use no error checking, as usual.
		2193
		2194	static void
		2195	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2196	{
		2197	free (w);
		2198	// now do something you wanted to do when the program has
		2199	// no longer anything immediate to do.
		2200	}
		2201
		2202	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2203	ev_idle_init (idle_watcher, idle_cb);
		2204	ev_idle_start (loop, idle_cb);
		2205
		2206
		2207	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
		2208
658	Prepare and check watchers usually (but not always) are used in	2209	Prepare and check watchers are usually (but not always) used in pairs:
659	tandom. Prepare watchers get invoked before the process blocks and check	2210	prepare watchers get invoked before the process blocks and check watchers
660	watchers afterwards.	2211	afterwards.
661		2212
		2213	You I<must not> call C<ev_loop> or similar functions that enter
		2214	the current event loop from either C<ev_prepare> or C<ev_check>
		2215	watchers. Other loops than the current one are fine, however. The
		2216	rationale behind this is that you do not need to check for recursion in
		2217	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2218	C<ev_check> so if you have one watcher of each kind they will always be
		2219	called in pairs bracketing the blocking call.
		2220
662	Their main purpose is to integrate other event mechanisms into libev. This	2221	Their main purpose is to integrate other event mechanisms into libev and
663	could be used, for example, to track variable changes, implement your own	2222	their use is somewhat advanced. They could be used, for example, to track
664	watchers, integrate net-snmp or a coroutine library and lots more.	2223	variable changes, implement your own watchers, integrate net-snmp or a
		2224	coroutine library and lots more. They are also occasionally useful if
		2225	you cache some data and want to flush it before blocking (for example,
		2226	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2227	watcher).
665		2228
666	This is done by examining in each prepare call which file descriptors need	2229	This is done by examining in each prepare call which file descriptors
667	to be watched by the other library, registering ev_io watchers for them	2230	need to be watched by the other library, registering C<ev_io> watchers
668	and starting an ev_timer watcher for any timeouts (many libraries provide	2231	for them and starting an C<ev_timer> watcher for any timeouts (many
669	just this functionality). Then, in the check watcher you check for any	2232	libraries provide exactly this functionality). Then, in the check watcher,
670	events that occured (by making your callbacks set soem flags for example)	2233	you check for any events that occurred (by checking the pending status
671	and call back into the library.	2234	of all watchers and stopping them) and call back into the library. The
		2235	I/O and timer callbacks will never actually be called (but must be valid
		2236	nevertheless, because you never know, you know?).
672		2237
673	As another example, the perl Coro module uses these hooks to integrate	2238	As another example, the Perl Coro module uses these hooks to integrate
674	coroutines into libev programs, by yielding to other active coroutines	2239	coroutines into libev programs, by yielding to other active coroutines
675	during each prepare and only letting the process block if no coroutines	2240	during each prepare and only letting the process block if no coroutines
676	are ready to run.	2241	are ready to run (it's actually more complicated: it only runs coroutines
		2242	with priority higher than or equal to the event loop and one coroutine
		2243	of lower priority, but only once, using idle watchers to keep the event
		2244	loop from blocking if lower-priority coroutines are active, thus mapping
		2245	low-priority coroutines to idle/background tasks).
		2246
		2247	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2248	priority, to ensure that they are being run before any other watchers
		2249	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2250
		2251	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2252	activate ("feed") events into libev. While libev fully supports this, they
		2253	might get executed before other C<ev_check> watchers did their job. As
		2254	C<ev_check> watchers are often used to embed other (non-libev) event
		2255	loops those other event loops might be in an unusable state until their
		2256	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2257	others).
		2258
		2259	=head3 Watcher-Specific Functions and Data Members
677		2260
678	=over 4	2261	=over 4
679		2262
680	=item ev_prepare_init (ev_prepare *, callback)	2263	=item ev_prepare_init (ev_prepare *, callback)
681		2264
682	=item ev_check_init (ev_check *, callback)	2265	=item ev_check_init (ev_check *, callback)
683		2266
684	Initialises and configures the prepare or check watcher - they have no	2267	Initialises and configures the prepare or check watcher - they have no
685	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2268	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
686	macros, but using them is utterly, utterly pointless.	2269	macros, but using them is utterly, utterly, utterly and completely
		2270	pointless.
687		2271
688	=back	2272	=back
689		2273
		2274	=head3 Examples
		2275
		2276	There are a number of principal ways to embed other event loops or modules
		2277	into libev. Here are some ideas on how to include libadns into libev
		2278	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2279	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2280	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2281	Glib event loop).
		2282
		2283	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2284	and in a check watcher, destroy them and call into libadns. What follows
		2285	is pseudo-code only of course. This requires you to either use a low
		2286	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2287	the callbacks for the IO/timeout watchers might not have been called yet.
		2288
		2289	static ev_io iow [nfd];
		2290	static ev_timer tw;
		2291
		2292	static void
		2293	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2294	{
		2295	}
		2296
		2297	// create io watchers for each fd and a timer before blocking
		2298	static void
		2299	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2300	{
		2301	int timeout = 3600000;
		2302	struct pollfd fds [nfd];
		2303	// actual code will need to loop here and realloc etc.
		2304	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2305
		2306	/* the callback is illegal, but won't be called as we stop during check */
		2307	ev_timer_init (&tw, 0, timeout * 1e-3);
		2308	ev_timer_start (loop, &tw);
		2309
		2310	// create one ev_io per pollfd
		2311	for (int i = 0; i < nfd; ++i)
		2312	{
		2313	ev_io_init (iow + i, io_cb, fds [i].fd,
		2314	((fds [i].events & POLLIN ? EV_READ : 0)
		2315	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2316
		2317	fds [i].revents = 0;
		2318	ev_io_start (loop, iow + i);
		2319	}
		2320	}
		2321
		2322	// stop all watchers after blocking
		2323	static void
		2324	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2325	{
		2326	ev_timer_stop (loop, &tw);
		2327
		2328	for (int i = 0; i < nfd; ++i)
		2329	{
		2330	// set the relevant poll flags
		2331	// could also call adns_processreadable etc. here
		2332	struct pollfd *fd = fds + i;
		2333	int revents = ev_clear_pending (iow + i);
		2334	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2335	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2336
		2337	// now stop the watcher
		2338	ev_io_stop (loop, iow + i);
		2339	}
		2340
		2341	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2342	}
		2343
		2344	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2345	in the prepare watcher and would dispose of the check watcher.
		2346
		2347	Method 3: If the module to be embedded supports explicit event
		2348	notification (libadns does), you can also make use of the actual watcher
		2349	callbacks, and only destroy/create the watchers in the prepare watcher.
		2350
		2351	static void
		2352	timer_cb (EV_P_ ev_timer *w, int revents)
		2353	{
		2354	adns_state ads = (adns_state)w->data;
		2355	update_now (EV_A);
		2356
		2357	adns_processtimeouts (ads, &tv_now);
		2358	}
		2359
		2360	static void
		2361	io_cb (EV_P_ ev_io *w, int revents)
		2362	{
		2363	adns_state ads = (adns_state)w->data;
		2364	update_now (EV_A);
		2365
		2366	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2367	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2368	}
		2369
		2370	// do not ever call adns_afterpoll
		2371
		2372	Method 4: Do not use a prepare or check watcher because the module you
		2373	want to embed is not flexible enough to support it. Instead, you can
		2374	override their poll function. The drawback with this solution is that the
		2375	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2376	this approach, effectively embedding EV as a client into the horrible
		2377	libglib event loop.
		2378
		2379	static gint
		2380	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2381	{
		2382	int got_events = 0;
		2383
		2384	for (n = 0; n < nfds; ++n)
		2385	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2386
		2387	if (timeout >= 0)
		2388	// create/start timer
		2389
		2390	// poll
		2391	ev_loop (EV_A_ 0);
		2392
		2393	// stop timer again
		2394	if (timeout >= 0)
		2395	ev_timer_stop (EV_A_ &to);
		2396
		2397	// stop io watchers again - their callbacks should have set
		2398	for (n = 0; n < nfds; ++n)
		2399	ev_io_stop (EV_A_ iow [n]);
		2400
		2401	return got_events;
		2402	}
		2403
		2404
		2405	=head2 C<ev_embed> - when one backend isn't enough...
		2406
		2407	This is a rather advanced watcher type that lets you embed one event loop
		2408	into another (currently only C<ev_io> events are supported in the embedded
		2409	loop, other types of watchers might be handled in a delayed or incorrect
		2410	fashion and must not be used).
		2411
		2412	There are primarily two reasons you would want that: work around bugs and
		2413	prioritise I/O.
		2414
		2415	As an example for a bug workaround, the kqueue backend might only support
		2416	sockets on some platform, so it is unusable as generic backend, but you
		2417	still want to make use of it because you have many sockets and it scales
		2418	so nicely. In this case, you would create a kqueue-based loop and embed
		2419	it into your default loop (which might use e.g. poll). Overall operation
		2420	will be a bit slower because first libev has to call C<poll> and then
		2421	C<kevent>, but at least you can use both mechanisms for what they are
		2422	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2423
		2424	As for prioritising I/O: under rare circumstances you have the case where
		2425	some fds have to be watched and handled very quickly (with low latency),
		2426	and even priorities and idle watchers might have too much overhead. In
		2427	this case you would put all the high priority stuff in one loop and all
		2428	the rest in a second one, and embed the second one in the first.
		2429
		2430	As long as the watcher is active, the callback will be invoked every time
		2431	there might be events pending in the embedded loop. The callback must then
		2432	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke
		2433	their callbacks (you could also start an idle watcher to give the embedded
		2434	loop strictly lower priority for example). You can also set the callback
		2435	to C<0>, in which case the embed watcher will automatically execute the
		2436	embedded loop sweep.
		2437
		2438	As long as the watcher is started it will automatically handle events. The
		2439	callback will be invoked whenever some events have been handled. You can
		2440	set the callback to C<0> to avoid having to specify one if you are not
		2441	interested in that.
		2442
		2443	Also, there have not currently been made special provisions for forking:
		2444	when you fork, you not only have to call C<ev_loop_fork> on both loops,
		2445	but you will also have to stop and restart any C<ev_embed> watchers
		2446	yourself - but you can use a fork watcher to handle this automatically,
		2447	and future versions of libev might do just that.
		2448
		2449	Unfortunately, not all backends are embeddable: only the ones returned by
		2450	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2451	portable one.
		2452
		2453	So when you want to use this feature you will always have to be prepared
		2454	that you cannot get an embeddable loop. The recommended way to get around
		2455	this is to have a separate variables for your embeddable loop, try to
		2456	create it, and if that fails, use the normal loop for everything.
		2457
		2458	=head3 C<ev_embed> and fork
		2459
		2460	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2461	automatically be applied to the embedded loop as well, so no special
		2462	fork handling is required in that case. When the watcher is not running,
		2463	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2464	as applicable.
		2465
		2466	=head3 Watcher-Specific Functions and Data Members
		2467
		2468	=over 4
		2469
		2470	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2471
		2472	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2473
		2474	Configures the watcher to embed the given loop, which must be
		2475	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2476	invoked automatically, otherwise it is the responsibility of the callback
		2477	to invoke it (it will continue to be called until the sweep has been done,
		2478	if you do not want that, you need to temporarily stop the embed watcher).
		2479
		2480	=item ev_embed_sweep (loop, ev_embed *)
		2481
		2482	Make a single, non-blocking sweep over the embedded loop. This works
		2483	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		2484	appropriate way for embedded loops.
		2485
		2486	=item struct ev_loop *other [read-only]
		2487
		2488	The embedded event loop.
		2489
		2490	=back
		2491
		2492	=head3 Examples
		2493
		2494	Example: Try to get an embeddable event loop and embed it into the default
		2495	event loop. If that is not possible, use the default loop. The default
		2496	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2497	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2498	used).
		2499
		2500	struct ev_loop *loop_hi = ev_default_init (0);
		2501	struct ev_loop *loop_lo = 0;
		2502	ev_embed embed;
		2503
		2504	// see if there is a chance of getting one that works
		2505	// (remember that a flags value of 0 means autodetection)
		2506	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2507	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2508	: 0;
		2509
		2510	// if we got one, then embed it, otherwise default to loop_hi
		2511	if (loop_lo)
		2512	{
		2513	ev_embed_init (&embed, 0, loop_lo);
		2514	ev_embed_start (loop_hi, &embed);
		2515	}
		2516	else
		2517	loop_lo = loop_hi;
		2518
		2519	Example: Check if kqueue is available but not recommended and create
		2520	a kqueue backend for use with sockets (which usually work with any
		2521	kqueue implementation). Store the kqueue/socket-only event loop in
		2522	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2523
		2524	struct ev_loop *loop = ev_default_init (0);
		2525	struct ev_loop *loop_socket = 0;
		2526	ev_embed embed;
		2527
		2528	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2529	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2530	{
		2531	ev_embed_init (&embed, 0, loop_socket);
		2532	ev_embed_start (loop, &embed);
		2533	}
		2534
		2535	if (!loop_socket)
		2536	loop_socket = loop;
		2537
		2538	// now use loop_socket for all sockets, and loop for everything else
		2539
		2540
		2541	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		2542
		2543	Fork watchers are called when a C<fork ()> was detected (usually because
		2544	whoever is a good citizen cared to tell libev about it by calling
		2545	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		2546	event loop blocks next and before C<ev_check> watchers are being called,
		2547	and only in the child after the fork. If whoever good citizen calling
		2548	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		2549	handlers will be invoked, too, of course.
		2550
		2551	=head3 Watcher-Specific Functions and Data Members
		2552
		2553	=over 4
		2554
		2555	=item ev_fork_init (ev_signal *, callback)
		2556
		2557	Initialises and configures the fork watcher - it has no parameters of any
		2558	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		2559	believe me.
		2560
		2561	=back
		2562
		2563
		2564	=head2 C<ev_async> - how to wake up another event loop
		2565
		2566	In general, you cannot use an C<ev_loop> from multiple threads or other
		2567	asynchronous sources such as signal handlers (as opposed to multiple event
		2568	loops - those are of course safe to use in different threads).
		2569
		2570	Sometimes, however, you need to wake up another event loop you do not
		2571	control, for example because it belongs to another thread. This is what
		2572	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2573	can signal it by calling C<ev_async_send>, which is thread- and signal
		2574	safe.
		2575
		2576	This functionality is very similar to C<ev_signal> watchers, as signals,
		2577	too, are asynchronous in nature, and signals, too, will be compressed
		2578	(i.e. the number of callback invocations may be less than the number of
		2579	C<ev_async_sent> calls).
		2580
		2581	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2582	just the default loop.
		2583
		2584	=head3 Queueing
		2585
		2586	C<ev_async> does not support queueing of data in any way. The reason
		2587	is that the author does not know of a simple (or any) algorithm for a
		2588	multiple-writer-single-reader queue that works in all cases and doesn't
		2589	need elaborate support such as pthreads.
		2590
		2591	That means that if you want to queue data, you have to provide your own
		2592	queue. But at least I can tell you how to implement locking around your
		2593	queue:
		2594
		2595	=over 4
		2596
		2597	=item queueing from a signal handler context
		2598
		2599	To implement race-free queueing, you simply add to the queue in the signal
		2600	handler but you block the signal handler in the watcher callback. Here is
		2601	an example that does that for some fictitious SIGUSR1 handler:
		2602
		2603	static ev_async mysig;
		2604
		2605	static void
		2606	sigusr1_handler (void)
		2607	{
		2608	sometype data;
		2609
		2610	// no locking etc.
		2611	queue_put (data);
		2612	ev_async_send (EV_DEFAULT_ &mysig);
		2613	}
		2614
		2615	static void
		2616	mysig_cb (EV_P_ ev_async *w, int revents)
		2617	{
		2618	sometype data;
		2619	sigset_t block, prev;
		2620
		2621	sigemptyset (&block);
		2622	sigaddset (&block, SIGUSR1);
		2623	sigprocmask (SIG_BLOCK, &block, &prev);
		2624
		2625	while (queue_get (&data))
		2626	process (data);
		2627
		2628	if (sigismember (&prev, SIGUSR1)
		2629	sigprocmask (SIG_UNBLOCK, &block, 0);
		2630	}
		2631
		2632	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2633	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2634	either...).
		2635
		2636	=item queueing from a thread context
		2637
		2638	The strategy for threads is different, as you cannot (easily) block
		2639	threads but you can easily preempt them, so to queue safely you need to
		2640	employ a traditional mutex lock, such as in this pthread example:
		2641
		2642	static ev_async mysig;
		2643	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2644
		2645	static void
		2646	otherthread (void)
		2647	{
		2648	// only need to lock the actual queueing operation
		2649	pthread_mutex_lock (&mymutex);
		2650	queue_put (data);
		2651	pthread_mutex_unlock (&mymutex);
		2652
		2653	ev_async_send (EV_DEFAULT_ &mysig);
		2654	}
		2655
		2656	static void
		2657	mysig_cb (EV_P_ ev_async *w, int revents)
		2658	{
		2659	pthread_mutex_lock (&mymutex);
		2660
		2661	while (queue_get (&data))
		2662	process (data);
		2663
		2664	pthread_mutex_unlock (&mymutex);
		2665	}
		2666
		2667	=back
		2668
		2669
		2670	=head3 Watcher-Specific Functions and Data Members
		2671
		2672	=over 4
		2673
		2674	=item ev_async_init (ev_async *, callback)
		2675
		2676	Initialises and configures the async watcher - it has no parameters of any
		2677	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		2678	trust me.
		2679
		2680	=item ev_async_send (loop, ev_async *)
		2681
		2682	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2683	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2684	C<ev_feed_event>, this call is safe to do from other threads, signal or
		2685	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		2686	section below on what exactly this means).
		2687
		2688	This call incurs the overhead of a system call only once per loop iteration,
		2689	so while the overhead might be noticeable, it doesn't apply to repeated
		2690	calls to C<ev_async_send>.
		2691
		2692	=item bool = ev_async_pending (ev_async *)
		2693
		2694	Returns a non-zero value when C<ev_async_send> has been called on the
		2695	watcher but the event has not yet been processed (or even noted) by the
		2696	event loop.
		2697
		2698	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		2699	the loop iterates next and checks for the watcher to have become active,
		2700	it will reset the flag again. C<ev_async_pending> can be used to very
		2701	quickly check whether invoking the loop might be a good idea.
		2702
		2703	Not that this does I<not> check whether the watcher itself is pending, only
		2704	whether it has been requested to make this watcher pending.
		2705
		2706	=back
		2707
		2708
690	=head1 OTHER FUNCTIONS	2709	=head1 OTHER FUNCTIONS
691		2710
692	There are some other fucntions of possible interest. Described. Here. Now.	2711	There are some other functions of possible interest. Described. Here. Now.
693		2712
694	=over 4	2713	=over 4
695		2714
696	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2715	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
697		2716
698	This function combines a simple timer and an I/O watcher, calls your	2717	This function combines a simple timer and an I/O watcher, calls your
699	callback on whichever event happens first and automatically stop both	2718	callback on whichever event happens first and automatically stops both
700	watchers. This is useful if you want to wait for a single event on an fd	2719	watchers. This is useful if you want to wait for a single event on an fd
701	or timeout without havign to allocate/configure/start/stop/free one or	2720	or timeout without having to allocate/configure/start/stop/free one or
702	more watchers yourself.	2721	more watchers yourself.
703		2722
704	If C<fd> is less than 0, then no I/O watcher will be started and events is	2723	If C<fd> is less than 0, then no I/O watcher will be started and the
705	ignored. Otherwise, an ev_io watcher for the given C<fd> and C<events> set	2724	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
706	will be craeted and started.	2725	the given C<fd> and C<events> set will be created and started.
707		2726
708	If C<timeout> is less than 0, then no timeout watcher will be	2727	If C<timeout> is less than 0, then no timeout watcher will be
709	started. Otherwise an ev_timer watcher with after = C<timeout> (and repeat	2728	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
710	= 0) will be started.	2729	repeat = 0) will be started. C<0> is a valid timeout.
711		2730
712	The callback has the type C<void (*cb)(int revents, void *arg)> and	2731	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
713	gets passed an events set (normally a combination of EV_ERROR, EV_READ,	2732	passed an C<revents> set like normal event callbacks (a combination of
714	EV_WRITE or EV_TIMEOUT) and the C<arg> value passed to C<ev_once>:	2733	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
		2734	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2735	a timeout and an io event at the same time - you probably should give io
		2736	events precedence.
715		2737
		2738	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		2739
716	static void stdin_ready (int revents, void *arg)	2740	static void stdin_ready (int revents, void *arg)
717	{	2741	{
718	if (revents & EV_TIMEOUT)
719	/* doh, nothing entered */
720	else if (revents & EV_READ)	2742	if (revents & EV_READ)
721	/* stdin might have data for us, joy! */	2743	/* stdin might have data for us, joy! */;
		2744	else if (revents & EV_TIMEOUT)
		2745	/* doh, nothing entered */;
722	}	2746	}
723		2747
724	ev_once (STDIN_FILENO, EV_READm 10., stdin_ready, 0);	2748	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
725		2749
726	=item ev_feed_event (loop, watcher, int events)	2750	=item ev_feed_event (struct ev_loop *, watcher *, int revents)
727		2751
728	Feeds the given event set into the event loop, as if the specified event	2752	Feeds the given event set into the event loop, as if the specified event
729	has happened for the specified watcher (which must be a pointer to an	2753	had happened for the specified watcher (which must be a pointer to an
730	initialised but not necessarily active event watcher).	2754	initialised but not necessarily started event watcher).
731		2755
732	=item ev_feed_fd_event (loop, int fd, int revents)	2756	=item ev_feed_fd_event (struct ev_loop *, int fd, int revents)
733		2757
734	Feed an event on the given fd, as if a file descriptor backend detected it.	2758	Feed an event on the given fd, as if a file descriptor backend detected
		2759	the given events it.
735		2760
736	=item ev_feed_signal_event (loop, int signum)	2761	=item ev_feed_signal_event (struct ev_loop *loop, int signum)
737		2762
738	Feed an event as if the given signal occured (loop must be the default loop!).	2763	Feed an event as if the given signal occurred (C<loop> must be the default
		2764	loop!).
739		2765
740	=back	2766	=back
741		2767
		2768
		2769	=head1 LIBEVENT EMULATION
		2770
		2771	Libev offers a compatibility emulation layer for libevent. It cannot
		2772	emulate the internals of libevent, so here are some usage hints:
		2773
		2774	=over 4
		2775
		2776	=item * Use it by including <event.h>, as usual.
		2777
		2778	=item * The following members are fully supported: ev_base, ev_callback,
		2779	ev_arg, ev_fd, ev_res, ev_events.
		2780
		2781	=item * Avoid using ev_flags and the EVLIST_*-macros, while it is
		2782	maintained by libev, it does not work exactly the same way as in libevent (consider
		2783	it a private API).
		2784
		2785	=item * Priorities are not currently supported. Initialising priorities
		2786	will fail and all watchers will have the same priority, even though there
		2787	is an ev_pri field.
		2788
		2789	=item * In libevent, the last base created gets the signals, in libev, the
		2790	first base created (== the default loop) gets the signals.
		2791
		2792	=item * Other members are not supported.
		2793
		2794	=item * The libev emulation is I<not> ABI compatible to libevent, you need
		2795	to use the libev header file and library.
		2796
		2797	=back
		2798
		2799	=head1 C++ SUPPORT
		2800
		2801	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		2802	you to use some convenience methods to start/stop watchers and also change
		2803	the callback model to a model using method callbacks on objects.
		2804
		2805	To use it,
		2806
		2807	#include <ev++.h>
		2808
		2809	This automatically includes F<ev.h> and puts all of its definitions (many
		2810	of them macros) into the global namespace. All C++ specific things are
		2811	put into the C<ev> namespace. It should support all the same embedding
		2812	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		2813
		2814	Care has been taken to keep the overhead low. The only data member the C++
		2815	classes add (compared to plain C-style watchers) is the event loop pointer
		2816	that the watcher is associated with (or no additional members at all if
		2817	you disable C<EV_MULTIPLICITY> when embedding libev).
		2818
		2819	Currently, functions, and static and non-static member functions can be
		2820	used as callbacks. Other types should be easy to add as long as they only
		2821	need one additional pointer for context. If you need support for other
		2822	types of functors please contact the author (preferably after implementing
		2823	it).
		2824
		2825	Here is a list of things available in the C<ev> namespace:
		2826
		2827	=over 4
		2828
		2829	=item C<ev::READ>, C<ev::WRITE> etc.
		2830
		2831	These are just enum values with the same values as the C<EV_READ> etc.
		2832	macros from F<ev.h>.
		2833
		2834	=item C<ev::tstamp>, C<ev::now>
		2835
		2836	Aliases to the same types/functions as with the C<ev_> prefix.
		2837
		2838	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		2839
		2840	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		2841	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		2842	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		2843	defines by many implementations.
		2844
		2845	All of those classes have these methods:
		2846
		2847	=over 4
		2848
		2849	=item ev::TYPE::TYPE ()
		2850
		2851	=item ev::TYPE::TYPE (struct ev_loop *)
		2852
		2853	=item ev::TYPE::~TYPE
		2854
		2855	The constructor (optionally) takes an event loop to associate the watcher
		2856	with. If it is omitted, it will use C<EV_DEFAULT>.
		2857
		2858	The constructor calls C<ev_init> for you, which means you have to call the
		2859	C<set> method before starting it.
		2860
		2861	It will not set a callback, however: You have to call the templated C<set>
		2862	method to set a callback before you can start the watcher.
		2863
		2864	(The reason why you have to use a method is a limitation in C++ which does
		2865	not allow explicit template arguments for constructors).
		2866
		2867	The destructor automatically stops the watcher if it is active.
		2868
		2869	=item w->set<class, &class::method> (object *)
		2870
		2871	This method sets the callback method to call. The method has to have a
		2872	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		2873	first argument and the C<revents> as second. The object must be given as
		2874	parameter and is stored in the C<data> member of the watcher.
		2875
		2876	This method synthesizes efficient thunking code to call your method from
		2877	the C callback that libev requires. If your compiler can inline your
		2878	callback (i.e. it is visible to it at the place of the C<set> call and
		2879	your compiler is good :), then the method will be fully inlined into the
		2880	thunking function, making it as fast as a direct C callback.
		2881
		2882	Example: simple class declaration and watcher initialisation
		2883
		2884	struct myclass
		2885	{
		2886	void io_cb (ev::io &w, int revents) { }
		2887	}
		2888
		2889	myclass obj;
		2890	ev::io iow;
		2891	iow.set <myclass, &myclass::io_cb> (&obj);
		2892
		2893	=item w->set<function> (void *data = 0)
		2894
		2895	Also sets a callback, but uses a static method or plain function as
		2896	callback. The optional C<data> argument will be stored in the watcher's
		2897	C<data> member and is free for you to use.
		2898
		2899	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		2900
		2901	See the method-C<set> above for more details.
		2902
		2903	Example: Use a plain function as callback.
		2904
		2905	static void io_cb (ev::io &w, int revents) { }
		2906	iow.set <io_cb> ();
		2907
		2908	=item w->set (struct ev_loop *)
		2909
		2910	Associates a different C<struct ev_loop> with this watcher. You can only
		2911	do this when the watcher is inactive (and not pending either).
		2912
		2913	=item w->set ([arguments])
		2914
		2915	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
		2916	called at least once. Unlike the C counterpart, an active watcher gets
		2917	automatically stopped and restarted when reconfiguring it with this
		2918	method.
		2919
		2920	=item w->start ()
		2921
		2922	Starts the watcher. Note that there is no C<loop> argument, as the
		2923	constructor already stores the event loop.
		2924
		2925	=item w->stop ()
		2926
		2927	Stops the watcher if it is active. Again, no C<loop> argument.
		2928
		2929	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		2930
		2931	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		2932	C<ev_TYPE_again> function.
		2933
		2934	=item w->sweep () (C<ev::embed> only)
		2935
		2936	Invokes C<ev_embed_sweep>.
		2937
		2938	=item w->update () (C<ev::stat> only)
		2939
		2940	Invokes C<ev_stat_stat>.
		2941
		2942	=back
		2943
		2944	=back
		2945
		2946	Example: Define a class with an IO and idle watcher, start one of them in
		2947	the constructor.
		2948
		2949	class myclass
		2950	{
		2951	ev::io io ; void io_cb (ev::io &w, int revents);
		2952	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		2953
		2954	myclass (int fd)
		2955	{
		2956	io .set <myclass, &myclass::io_cb > (this);
		2957	idle.set <myclass, &myclass::idle_cb> (this);
		2958
		2959	io.start (fd, ev::READ);
		2960	}
		2961	};
		2962
		2963
		2964	=head1 OTHER LANGUAGE BINDINGS
		2965
		2966	Libev does not offer other language bindings itself, but bindings for a
		2967	number of languages exist in the form of third-party packages. If you know
		2968	any interesting language binding in addition to the ones listed here, drop
		2969	me a note.
		2970
		2971	=over 4
		2972
		2973	=item Perl
		2974
		2975	The EV module implements the full libev API and is actually used to test
		2976	libev. EV is developed together with libev. Apart from the EV core module,
		2977	there are additional modules that implement libev-compatible interfaces
		2978	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		2979	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		2980	and C<EV::Glib>).
		2981
		2982	It can be found and installed via CPAN, its homepage is at
		2983	L<http://software.schmorp.de/pkg/EV>.
		2984
		2985	=item Python
		2986
		2987	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		2988	seems to be quite complete and well-documented. Note, however, that the
		2989	patch they require for libev is outright dangerous as it breaks the ABI
		2990	for everybody else, and therefore, should never be applied in an installed
		2991	libev (if python requires an incompatible ABI then it needs to embed
		2992	libev).
		2993
		2994	=item Ruby
		2995
		2996	Tony Arcieri has written a ruby extension that offers access to a subset
		2997	of the libev API and adds file handle abstractions, asynchronous DNS and
		2998	more on top of it. It can be found via gem servers. Its homepage is at
		2999	L<http://rev.rubyforge.org/>.
		3000
		3001	=item D
		3002
		3003	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3004	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3005
		3006	=item Ocaml
		3007
		3008	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3009	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3010
		3011	=back
		3012
		3013
		3014	=head1 MACRO MAGIC
		3015
		3016	Libev can be compiled with a variety of options, the most fundamental
		3017	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3018	functions and callbacks have an initial C<struct ev_loop *> argument.
		3019
		3020	To make it easier to write programs that cope with either variant, the
		3021	following macros are defined:
		3022
		3023	=over 4
		3024
		3025	=item C<EV_A>, C<EV_A_>
		3026
		3027	This provides the loop I<argument> for functions, if one is required ("ev
		3028	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3029	C<EV_A_> is used when other arguments are following. Example:
		3030
		3031	ev_unref (EV_A);
		3032	ev_timer_add (EV_A_ watcher);
		3033	ev_loop (EV_A_ 0);
		3034
		3035	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3036	which is often provided by the following macro.
		3037
		3038	=item C<EV_P>, C<EV_P_>
		3039
		3040	This provides the loop I<parameter> for functions, if one is required ("ev
		3041	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3042	C<EV_P_> is used when other parameters are following. Example:
		3043
		3044	// this is how ev_unref is being declared
		3045	static void ev_unref (EV_P);
		3046
		3047	// this is how you can declare your typical callback
		3048	static void cb (EV_P_ ev_timer *w, int revents)
		3049
		3050	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3051	suitable for use with C<EV_A>.
		3052
		3053	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3054
		3055	Similar to the other two macros, this gives you the value of the default
		3056	loop, if multiple loops are supported ("ev loop default").
		3057
		3058	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3059
		3060	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3061	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3062	is undefined when the default loop has not been initialised by a previous
		3063	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3064
		3065	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3066	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3067
		3068	=back
		3069
		3070	Example: Declare and initialise a check watcher, utilising the above
		3071	macros so it will work regardless of whether multiple loops are supported
		3072	or not.
		3073
		3074	static void
		3075	check_cb (EV_P_ ev_timer *w, int revents)
		3076	{
		3077	ev_check_stop (EV_A_ w);
		3078	}
		3079
		3080	ev_check check;
		3081	ev_check_init (&check, check_cb);
		3082	ev_check_start (EV_DEFAULT_ &check);
		3083	ev_loop (EV_DEFAULT_ 0);
		3084
		3085	=head1 EMBEDDING
		3086
		3087	Libev can (and often is) directly embedded into host
		3088	applications. Examples of applications that embed it include the Deliantra
		3089	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3090	and rxvt-unicode.
		3091
		3092	The goal is to enable you to just copy the necessary files into your
		3093	source directory without having to change even a single line in them, so
		3094	you can easily upgrade by simply copying (or having a checked-out copy of
		3095	libev somewhere in your source tree).
		3096
		3097	=head2 FILESETS
		3098
		3099	Depending on what features you need you need to include one or more sets of files
		3100	in your application.
		3101
		3102	=head3 CORE EVENT LOOP
		3103
		3104	To include only the libev core (all the C<ev_*> functions), with manual
		3105	configuration (no autoconf):
		3106
		3107	#define EV_STANDALONE 1
		3108	#include "ev.c"
		3109
		3110	This will automatically include F<ev.h>, too, and should be done in a
		3111	single C source file only to provide the function implementations. To use
		3112	it, do the same for F<ev.h> in all files wishing to use this API (best
		3113	done by writing a wrapper around F<ev.h> that you can include instead and
		3114	where you can put other configuration options):
		3115
		3116	#define EV_STANDALONE 1
		3117	#include "ev.h"
		3118
		3119	Both header files and implementation files can be compiled with a C++
		3120	compiler (at least, that's a stated goal, and breakage will be treated
		3121	as a bug).
		3122
		3123	You need the following files in your source tree, or in a directory
		3124	in your include path (e.g. in libev/ when using -Ilibev):
		3125
		3126	ev.h
		3127	ev.c
		3128	ev_vars.h
		3129	ev_wrap.h
		3130
		3131	ev_win32.c required on win32 platforms only
		3132
		3133	ev_select.c only when select backend is enabled (which is enabled by default)
		3134	ev_poll.c only when poll backend is enabled (disabled by default)
		3135	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3136	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3137	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3138
		3139	F<ev.c> includes the backend files directly when enabled, so you only need
		3140	to compile this single file.
		3141
		3142	=head3 LIBEVENT COMPATIBILITY API
		3143
		3144	To include the libevent compatibility API, also include:
		3145
		3146	#include "event.c"
		3147
		3148	in the file including F<ev.c>, and:
		3149
		3150	#include "event.h"
		3151
		3152	in the files that want to use the libevent API. This also includes F<ev.h>.
		3153
		3154	You need the following additional files for this:
		3155
		3156	event.h
		3157	event.c
		3158
		3159	=head3 AUTOCONF SUPPORT
		3160
		3161	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3162	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3163	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3164	include F<config.h> and configure itself accordingly.
		3165
		3166	For this of course you need the m4 file:
		3167
		3168	libev.m4
		3169
		3170	=head2 PREPROCESSOR SYMBOLS/MACROS
		3171
		3172	Libev can be configured via a variety of preprocessor symbols you have to
		3173	define before including any of its files. The default in the absence of
		3174	autoconf is documented for every option.
		3175
		3176	=over 4
		3177
		3178	=item EV_STANDALONE
		3179
		3180	Must always be C<1> if you do not use autoconf configuration, which
		3181	keeps libev from including F<config.h>, and it also defines dummy
		3182	implementations for some libevent functions (such as logging, which is not
		3183	supported). It will also not define any of the structs usually found in
		3184	F<event.h> that are not directly supported by the libev core alone.
		3185
		3186	=item EV_USE_MONOTONIC
		3187
		3188	If defined to be C<1>, libev will try to detect the availability of the
		3189	monotonic clock option at both compile time and runtime. Otherwise no use
		3190	of the monotonic clock option will be attempted. If you enable this, you
		3191	usually have to link against librt or something similar. Enabling it when
		3192	the functionality isn't available is safe, though, although you have
		3193	to make sure you link against any libraries where the C<clock_gettime>
		3194	function is hiding in (often F<-lrt>).
		3195
		3196	=item EV_USE_REALTIME
		3197
		3198	If defined to be C<1>, libev will try to detect the availability of the
		3199	real-time clock option at compile time (and assume its availability at
		3200	runtime if successful). Otherwise no use of the real-time clock option will
		3201	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
		3202	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
		3203	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
		3204
		3205	=item EV_USE_NANOSLEEP
		3206
		3207	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3208	and will use it for delays. Otherwise it will use C<select ()>.
		3209
		3210	=item EV_USE_EVENTFD
		3211
		3212	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3213	available and will probe for kernel support at runtime. This will improve
		3214	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3215	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3216	2.7 or newer, otherwise disabled.
		3217
		3218	=item EV_USE_SELECT
		3219
		3220	If undefined or defined to be C<1>, libev will compile in support for the
		3221	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3222	other method takes over, select will be it. Otherwise the select backend
		3223	will not be compiled in.
		3224
		3225	=item EV_SELECT_USE_FD_SET
		3226
		3227	If defined to C<1>, then the select backend will use the system C<fd_set>
		3228	structure. This is useful if libev doesn't compile due to a missing
		3229	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout on
		3230	exotic systems. This usually limits the range of file descriptors to some
		3231	low limit such as 1024 or might have other limitations (winsocket only
		3232	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might
		3233	influence the size of the C<fd_set> used.
		3234
		3235	=item EV_SELECT_IS_WINSOCKET
		3236
		3237	When defined to C<1>, the select backend will assume that
		3238	select/socket/connect etc. don't understand file descriptors but
		3239	wants osf handles on win32 (this is the case when the select to
		3240	be used is the winsock select). This means that it will call
		3241	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3242	it is assumed that all these functions actually work on fds, even
		3243	on win32. Should not be defined on non-win32 platforms.
		3244
		3245	=item EV_FD_TO_WIN32_HANDLE
		3246
		3247	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3248	file descriptors to socket handles. When not defining this symbol (the
		3249	default), then libev will call C<_get_osfhandle>, which is usually
		3250	correct. In some cases, programs use their own file descriptor management,
		3251	in which case they can provide this function to map fds to socket handles.
		3252
		3253	=item EV_USE_POLL
		3254
		3255	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3256	backend. Otherwise it will be enabled on non-win32 platforms. It
		3257	takes precedence over select.
		3258
		3259	=item EV_USE_EPOLL
		3260
		3261	If defined to be C<1>, libev will compile in support for the Linux
		3262	C<epoll>(7) backend. Its availability will be detected at runtime,
		3263	otherwise another method will be used as fallback. This is the preferred
		3264	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3265	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3266
		3267	=item EV_USE_KQUEUE
		3268
		3269	If defined to be C<1>, libev will compile in support for the BSD style
		3270	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3271	otherwise another method will be used as fallback. This is the preferred
		3272	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3273	supports some types of fds correctly (the only platform we found that
		3274	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3275	not be used unless explicitly requested. The best way to use it is to find
		3276	out whether kqueue supports your type of fd properly and use an embedded
		3277	kqueue loop.
		3278
		3279	=item EV_USE_PORT
		3280
		3281	If defined to be C<1>, libev will compile in support for the Solaris
		3282	10 port style backend. Its availability will be detected at runtime,
		3283	otherwise another method will be used as fallback. This is the preferred
		3284	backend for Solaris 10 systems.
		3285
		3286	=item EV_USE_DEVPOLL
		3287
		3288	Reserved for future expansion, works like the USE symbols above.
		3289
		3290	=item EV_USE_INOTIFY
		3291
		3292	If defined to be C<1>, libev will compile in support for the Linux inotify
		3293	interface to speed up C<ev_stat> watchers. Its actual availability will
		3294	be detected at runtime. If undefined, it will be enabled if the headers
		3295	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3296
		3297	=item EV_ATOMIC_T
		3298
		3299	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3300	access is atomic with respect to other threads or signal contexts. No such
		3301	type is easily found in the C language, so you can provide your own type
		3302	that you know is safe for your purposes. It is used both for signal handler "locking"
		3303	as well as for signal and thread safety in C<ev_async> watchers.
		3304
		3305	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3306	(from F<signal.h>), which is usually good enough on most platforms.
		3307
		3308	=item EV_H
		3309
		3310	The name of the F<ev.h> header file used to include it. The default if
		3311	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3312	used to virtually rename the F<ev.h> header file in case of conflicts.
		3313
		3314	=item EV_CONFIG_H
		3315
		3316	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3317	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3318	C<EV_H>, above.
		3319
		3320	=item EV_EVENT_H
		3321
		3322	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3323	of how the F<event.h> header can be found, the default is C<"event.h">.
		3324
		3325	=item EV_PROTOTYPES
		3326
		3327	If defined to be C<0>, then F<ev.h> will not define any function
		3328	prototypes, but still define all the structs and other symbols. This is
		3329	occasionally useful if you want to provide your own wrapper functions
		3330	around libev functions.
		3331
		3332	=item EV_MULTIPLICITY
		3333
		3334	If undefined or defined to C<1>, then all event-loop-specific functions
		3335	will have the C<struct ev_loop *> as first argument, and you can create
		3336	additional independent event loops. Otherwise there will be no support
		3337	for multiple event loops and there is no first event loop pointer
		3338	argument. Instead, all functions act on the single default loop.
		3339
		3340	=item EV_MINPRI
		3341
		3342	=item EV_MAXPRI
		3343
		3344	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3345	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3346	provide for more priorities by overriding those symbols (usually defined
		3347	to be C<-2> and C<2>, respectively).
		3348
		3349	When doing priority-based operations, libev usually has to linearly search
		3350	all the priorities, so having many of them (hundreds) uses a lot of space
		3351	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3352	fine.
		3353
		3354	If your embedding application does not need any priorities, defining these
		3355	both to C<0> will save some memory and CPU.
		3356
		3357	=item EV_PERIODIC_ENABLE
		3358
		3359	If undefined or defined to be C<1>, then periodic timers are supported. If
		3360	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3361	code.
		3362
		3363	=item EV_IDLE_ENABLE
		3364
		3365	If undefined or defined to be C<1>, then idle watchers are supported. If
		3366	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3367	code.
		3368
		3369	=item EV_EMBED_ENABLE
		3370
		3371	If undefined or defined to be C<1>, then embed watchers are supported. If
		3372	defined to be C<0>, then they are not. Embed watchers rely on most other
		3373	watcher types, which therefore must not be disabled.
		3374
		3375	=item EV_STAT_ENABLE
		3376
		3377	If undefined or defined to be C<1>, then stat watchers are supported. If
		3378	defined to be C<0>, then they are not.
		3379
		3380	=item EV_FORK_ENABLE
		3381
		3382	If undefined or defined to be C<1>, then fork watchers are supported. If
		3383	defined to be C<0>, then they are not.
		3384
		3385	=item EV_ASYNC_ENABLE
		3386
		3387	If undefined or defined to be C<1>, then async watchers are supported. If
		3388	defined to be C<0>, then they are not.
		3389
		3390	=item EV_MINIMAL
		3391
		3392	If you need to shave off some kilobytes of code at the expense of some
		3393	speed, define this symbol to C<1>. Currently this is used to override some
		3394	inlining decisions, saves roughly 30% code size on amd64. It also selects a
		3395	much smaller 2-heap for timer management over the default 4-heap.
		3396
		3397	=item EV_PID_HASHSIZE
		3398
		3399	C<ev_child> watchers use a small hash table to distribute workload by
		3400	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
		3401	than enough. If you need to manage thousands of children you might want to
		3402	increase this value (I<must> be a power of two).
		3403
		3404	=item EV_INOTIFY_HASHSIZE
		3405
		3406	C<ev_stat> watchers use a small hash table to distribute workload by
		3407	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		3408	usually more than enough. If you need to manage thousands of C<ev_stat>
		3409	watchers you might want to increase this value (I<must> be a power of
		3410	two).
		3411
		3412	=item EV_USE_4HEAP
		3413
		3414	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3415	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3416	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3417	faster performance with many (thousands) of watchers.
		3418
		3419	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3420	(disabled).
		3421
		3422	=item EV_HEAP_CACHE_AT
		3423
		3424	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3425	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3426	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3427	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3428	but avoids random read accesses on heap changes. This improves performance
		3429	noticeably with many (hundreds) of watchers.
		3430
		3431	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3432	(disabled).
		3433
		3434	=item EV_VERIFY
		3435
		3436	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3437	be done: If set to C<0>, no internal verification code will be compiled
		3438	in. If set to C<1>, then verification code will be compiled in, but not
		3439	called. If set to C<2>, then the internal verification code will be
		3440	called once per loop, which can slow down libev. If set to C<3>, then the
		3441	verification code will be called very frequently, which will slow down
		3442	libev considerably.
		3443
		3444	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3445	C<0>.
		3446
		3447	=item EV_COMMON
		3448
		3449	By default, all watchers have a C<void *data> member. By redefining
		3450	this macro to a something else you can include more and other types of
		3451	members. You have to define it each time you include one of the files,
		3452	though, and it must be identical each time.
		3453
		3454	For example, the perl EV module uses something like this:
		3455
		3456	#define EV_COMMON \
		3457	SV *self; /* contains this struct */ \
		3458	SV *cb_sv, *fh /* note no trailing ";" */
		3459
		3460	=item EV_CB_DECLARE (type)
		3461
		3462	=item EV_CB_INVOKE (watcher, revents)
		3463
		3464	=item ev_set_cb (ev, cb)
		3465
		3466	Can be used to change the callback member declaration in each watcher,
		3467	and the way callbacks are invoked and set. Must expand to a struct member
		3468	definition and a statement, respectively. See the F<ev.h> header file for
		3469	their default definitions. One possible use for overriding these is to
		3470	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		3471	method calls instead of plain function calls in C++.
		3472
		3473	=back
		3474
		3475	=head2 EXPORTED API SYMBOLS
		3476
		3477	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3478	exported symbols, you can use the provided F<Symbol.*> files which list
		3479	all public symbols, one per line:
		3480
		3481	Symbols.ev for libev proper
		3482	Symbols.event for the libevent emulation
		3483
		3484	This can also be used to rename all public symbols to avoid clashes with
		3485	multiple versions of libev linked together (which is obviously bad in
		3486	itself, but sometimes it is inconvenient to avoid this).
		3487
		3488	A sed command like this will create wrapper C<#define>'s that you need to
		3489	include before including F<ev.h>:
		3490
		3491	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3492
		3493	This would create a file F<wrap.h> which essentially looks like this:
		3494
		3495	#define ev_backend myprefix_ev_backend
		3496	#define ev_check_start myprefix_ev_check_start
		3497	#define ev_check_stop myprefix_ev_check_stop
		3498	...
		3499
		3500	=head2 EXAMPLES
		3501
		3502	For a real-world example of a program the includes libev
		3503	verbatim, you can have a look at the EV perl module
		3504	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		3505	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		3506	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		3507	will be compiled. It is pretty complex because it provides its own header
		3508	file.
		3509
		3510	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		3511	that everybody includes and which overrides some configure choices:
		3512
		3513	#define EV_MINIMAL 1
		3514	#define EV_USE_POLL 0
		3515	#define EV_MULTIPLICITY 0
		3516	#define EV_PERIODIC_ENABLE 0
		3517	#define EV_STAT_ENABLE 0
		3518	#define EV_FORK_ENABLE 0
		3519	#define EV_CONFIG_H <config.h>
		3520	#define EV_MINPRI 0
		3521	#define EV_MAXPRI 0
		3522
		3523	#include "ev++.h"
		3524
		3525	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		3526
		3527	#include "ev_cpp.h"
		3528	#include "ev.c"
		3529
		3530	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		3531
		3532	=head2 THREADS AND COROUTINES
		3533
		3534	=head3 THREADS
		3535
		3536	All libev functions are reentrant and thread-safe unless explicitly
		3537	documented otherwise, but libev implements no locking itself. This means
		3538	that you can use as many loops as you want in parallel, as long as there
		3539	are no concurrent calls into any libev function with the same loop
		3540	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		3541	of course): libev guarantees that different event loops share no data
		3542	structures that need any locking.
		3543
		3544	Or to put it differently: calls with different loop parameters can be done
		3545	concurrently from multiple threads, calls with the same loop parameter
		3546	must be done serially (but can be done from different threads, as long as
		3547	only one thread ever is inside a call at any point in time, e.g. by using
		3548	a mutex per loop).
		3549
		3550	Specifically to support threads (and signal handlers), libev implements
		3551	so-called C<ev_async> watchers, which allow some limited form of
		3552	concurrency on the same event loop, namely waking it up "from the
		3553	outside".
		3554
		3555	If you want to know which design (one loop, locking, or multiple loops
		3556	without or something else still) is best for your problem, then I cannot
		3557	help you, but here is some generic advice:
		3558
		3559	=over 4
		3560
		3561	=item * most applications have a main thread: use the default libev loop
		3562	in that thread, or create a separate thread running only the default loop.
		3563
		3564	This helps integrating other libraries or software modules that use libev
		3565	themselves and don't care/know about threading.
		3566
		3567	=item * one loop per thread is usually a good model.
		3568
		3569	Doing this is almost never wrong, sometimes a better-performance model
		3570	exists, but it is always a good start.
		3571
		3572	=item * other models exist, such as the leader/follower pattern, where one
		3573	loop is handed through multiple threads in a kind of round-robin fashion.
		3574
		3575	Choosing a model is hard - look around, learn, know that usually you can do
		3576	better than you currently do :-)
		3577
		3578	=item * often you need to talk to some other thread which blocks in the
		3579	event loop.
		3580
		3581	C<ev_async> watchers can be used to wake them up from other threads safely
		3582	(or from signal contexts...).
		3583
		3584	An example use would be to communicate signals or other events that only
		3585	work in the default loop by registering the signal watcher with the
		3586	default loop and triggering an C<ev_async> watcher from the default loop
		3587	watcher callback into the event loop interested in the signal.
		3588
		3589	=back
		3590
		3591	=head3 COROUTINES
		3592
		3593	Libev is very accommodating to coroutines ("cooperative threads"):
		3594	libev fully supports nesting calls to its functions from different
		3595	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		3596	different coroutines, and switch freely between both coroutines running the
		3597	loop, as long as you don't confuse yourself). The only exception is that
		3598	you must not do this from C<ev_periodic> reschedule callbacks.
		3599
		3600	Care has been taken to ensure that libev does not keep local state inside
		3601	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		3602	they do not call any callbacks.
		3603
		3604	=head2 COMPILER WARNINGS
		3605
		3606	Depending on your compiler and compiler settings, you might get no or a
		3607	lot of warnings when compiling libev code. Some people are apparently
		3608	scared by this.
		3609
		3610	However, these are unavoidable for many reasons. For one, each compiler
		3611	has different warnings, and each user has different tastes regarding
		3612	warning options. "Warn-free" code therefore cannot be a goal except when
		3613	targeting a specific compiler and compiler-version.
		3614
		3615	Another reason is that some compiler warnings require elaborate
		3616	workarounds, or other changes to the code that make it less clear and less
		3617	maintainable.
		3618
		3619	And of course, some compiler warnings are just plain stupid, or simply
		3620	wrong (because they don't actually warn about the condition their message
		3621	seems to warn about). For example, certain older gcc versions had some
		3622	warnings that resulted an extreme number of false positives. These have
		3623	been fixed, but some people still insist on making code warn-free with
		3624	such buggy versions.
		3625
		3626	While libev is written to generate as few warnings as possible,
		3627	"warn-free" code is not a goal, and it is recommended not to build libev
		3628	with any compiler warnings enabled unless you are prepared to cope with
		3629	them (e.g. by ignoring them). Remember that warnings are just that:
		3630	warnings, not errors, or proof of bugs.
		3631
		3632
		3633	=head2 VALGRIND
		3634
		3635	Valgrind has a special section here because it is a popular tool that is
		3636	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		3637
		3638	If you think you found a bug (memory leak, uninitialised data access etc.)
		3639	in libev, then check twice: If valgrind reports something like:
		3640
		3641	==2274== definitely lost: 0 bytes in 0 blocks.
		3642	==2274== possibly lost: 0 bytes in 0 blocks.
		3643	==2274== still reachable: 256 bytes in 1 blocks.
		3644
		3645	Then there is no memory leak, just as memory accounted to global variables
		3646	is not a memleak - the memory is still being referenced, and didn't leak.
		3647
		3648	Similarly, under some circumstances, valgrind might report kernel bugs
		3649	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3650	although an acceptable workaround has been found here), or it might be
		3651	confused.
		3652
		3653	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		3654	make it into some kind of religion.
		3655
		3656	If you are unsure about something, feel free to contact the mailing list
		3657	with the full valgrind report and an explanation on why you think this
		3658	is a bug in libev (best check the archives, too :). However, don't be
		3659	annoyed when you get a brisk "this is no bug" answer and take the chance
		3660	of learning how to interpret valgrind properly.
		3661
		3662	If you need, for some reason, empty reports from valgrind for your project
		3663	I suggest using suppression lists.
		3664
		3665
		3666	=head1 PORTABILITY NOTES
		3667
		3668	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		3669
		3670	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		3671	requires, and its I/O model is fundamentally incompatible with the POSIX
		3672	model. Libev still offers limited functionality on this platform in
		3673	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		3674	descriptors. This only applies when using Win32 natively, not when using
		3675	e.g. cygwin.
		3676
		3677	Lifting these limitations would basically require the full
		3678	re-implementation of the I/O system. If you are into these kinds of
		3679	things, then note that glib does exactly that for you in a very portable
		3680	way (note also that glib is the slowest event library known to man).
		3681
		3682	There is no supported compilation method available on windows except
		3683	embedding it into other applications.
		3684
		3685	Not a libev limitation but worth mentioning: windows apparently doesn't
		3686	accept large writes: instead of resulting in a partial write, windows will
		3687	either accept everything or return C<ENOBUFS> if the buffer is too large,
		3688	so make sure you only write small amounts into your sockets (less than a
		3689	megabyte seems safe, but this apparently depends on the amount of memory
		3690	available).
		3691
		3692	Due to the many, low, and arbitrary limits on the win32 platform and
		3693	the abysmal performance of winsockets, using a large number of sockets
		3694	is not recommended (and not reasonable). If your program needs to use
		3695	more than a hundred or so sockets, then likely it needs to use a totally
		3696	different implementation for windows, as libev offers the POSIX readiness
		3697	notification model, which cannot be implemented efficiently on windows
		3698	(Microsoft monopoly games).
		3699
		3700	A typical way to use libev under windows is to embed it (see the embedding
		3701	section for details) and use the following F<evwrap.h> header file instead
		3702	of F<ev.h>:
		3703
		3704	#define EV_STANDALONE /* keeps ev from requiring config.h */
		3705	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		3706
		3707	#include "ev.h"
		3708
		3709	And compile the following F<evwrap.c> file into your project (make sure
		3710	you do I<not> compile the F<ev.c> or any other embedded source files!):
		3711
		3712	#include "evwrap.h"
		3713	#include "ev.c"
		3714
		3715	=over 4
		3716
		3717	=item The winsocket select function
		3718
		3719	The winsocket C<select> function doesn't follow POSIX in that it
		3720	requires socket I<handles> and not socket I<file descriptors> (it is
		3721	also extremely buggy). This makes select very inefficient, and also
		3722	requires a mapping from file descriptors to socket handles (the Microsoft
		3723	C runtime provides the function C<_open_osfhandle> for this). See the
		3724	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		3725	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		3726
		3727	The configuration for a "naked" win32 using the Microsoft runtime
		3728	libraries and raw winsocket select is:
		3729
		3730	#define EV_USE_SELECT 1
		3731	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		3732
		3733	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3734	complexity in the O(n²) range when using win32.
		3735
		3736	=item Limited number of file descriptors
		3737
		3738	Windows has numerous arbitrary (and low) limits on things.
		3739
		3740	Early versions of winsocket's select only supported waiting for a maximum
		3741	of C<64> handles (probably owning to the fact that all windows kernels
		3742	can only wait for C<64> things at the same time internally; Microsoft
		3743	recommends spawning a chain of threads and wait for 63 handles and the
		3744	previous thread in each. Great).
		3745
		3746	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3747	to some high number (e.g. C<2048>) before compiling the winsocket select
		3748	call (which might be in libev or elsewhere, for example, perl does its own
		3749	select emulation on windows).
		3750
		3751	Another limit is the number of file descriptors in the Microsoft runtime
		3752	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3753	or something like this inside Microsoft). You can increase this by calling
		3754	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3755	arbitrary limit), but is broken in many versions of the Microsoft runtime
		3756	libraries.
		3757
		3758	This might get you to about C<512> or C<2048> sockets (depending on
		3759	windows version and/or the phase of the moon). To get more, you need to
		3760	wrap all I/O functions and provide your own fd management, but the cost of
		3761	calling select (O(n²)) will likely make this unworkable.
		3762
		3763	=back
		3764
		3765	=head2 PORTABILITY REQUIREMENTS
		3766
		3767	In addition to a working ISO-C implementation and of course the
		3768	backend-specific APIs, libev relies on a few additional extensions:
		3769
		3770	=over 4
		3771
		3772	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		3773	calling conventions regardless of C<ev_watcher_type *>.
		3774
		3775	Libev assumes not only that all watcher pointers have the same internal
		3776	structure (guaranteed by POSIX but not by ISO C for example), but it also
		3777	assumes that the same (machine) code can be used to call any watcher
		3778	callback: The watcher callbacks have different type signatures, but libev
		3779	calls them using an C<ev_watcher *> internally.
		3780
		3781	=item C<sig_atomic_t volatile> must be thread-atomic as well
		3782
		3783	The type C<sig_atomic_t volatile> (or whatever is defined as
		3784	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		3785	threads. This is not part of the specification for C<sig_atomic_t>, but is
		3786	believed to be sufficiently portable.
		3787
		3788	=item C<sigprocmask> must work in a threaded environment
		3789
		3790	Libev uses C<sigprocmask> to temporarily block signals. This is not
		3791	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		3792	pthread implementations will either allow C<sigprocmask> in the "main
		3793	thread" or will block signals process-wide, both behaviours would
		3794	be compatible with libev. Interaction between C<sigprocmask> and
		3795	C<pthread_sigmask> could complicate things, however.
		3796
		3797	The most portable way to handle signals is to block signals in all threads
		3798	except the initial one, and run the default loop in the initial thread as
		3799	well.
		3800
		3801	=item C<long> must be large enough for common memory allocation sizes
		3802
		3803	To improve portability and simplify its API, libev uses C<long> internally
		3804	instead of C<size_t> when allocating its data structures. On non-POSIX
		3805	systems (Microsoft...) this might be unexpectedly low, but is still at
		3806	least 31 bits everywhere, which is enough for hundreds of millions of
		3807	watchers.
		3808
		3809	=item C<double> must hold a time value in seconds with enough accuracy
		3810
		3811	The type C<double> is used to represent timestamps. It is required to
		3812	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		3813	enough for at least into the year 4000. This requirement is fulfilled by
		3814	implementations implementing IEEE 754 (basically all existing ones).
		3815
		3816	=back
		3817
		3818	If you know of other additional requirements drop me a note.
		3819
		3820
		3821	=head1 ALGORITHMIC COMPLEXITIES
		3822
		3823	In this section the complexities of (many of) the algorithms used inside
		3824	libev will be documented. For complexity discussions about backends see
		3825	the documentation for C<ev_default_init>.
		3826
		3827	All of the following are about amortised time: If an array needs to be
		3828	extended, libev needs to realloc and move the whole array, but this
		3829	happens asymptotically rarer with higher number of elements, so O(1) might
		3830	mean that libev does a lengthy realloc operation in rare cases, but on
		3831	average it is much faster and asymptotically approaches constant time.
		3832
		3833	=over 4
		3834
		3835	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		3836
		3837	This means that, when you have a watcher that triggers in one hour and
		3838	there are 100 watchers that would trigger before that, then inserting will
		3839	have to skip roughly seven (C<ld 100>) of these watchers.
		3840
		3841	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		3842
		3843	That means that changing a timer costs less than removing/adding them,
		3844	as only the relative motion in the event queue has to be paid for.
		3845
		3846	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		3847
		3848	These just add the watcher into an array or at the head of a list.
		3849
		3850	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		3851
		3852	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		3853
		3854	These watchers are stored in lists, so they need to be walked to find the
		3855	correct watcher to remove. The lists are usually short (you don't usually
		3856	have many watchers waiting for the same fd or signal: one is typical, two
		3857	is rare).
		3858
		3859	=item Finding the next timer in each loop iteration: O(1)
		3860
		3861	By virtue of using a binary or 4-heap, the next timer is always found at a
		3862	fixed position in the storage array.
		3863
		3864	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		3865
		3866	A change means an I/O watcher gets started or stopped, which requires
		3867	libev to recalculate its status (and possibly tell the kernel, depending
		3868	on backend and whether C<ev_io_set> was used).
		3869
		3870	=item Activating one watcher (putting it into the pending state): O(1)
		3871
		3872	=item Priority handling: O(number_of_priorities)
		3873
		3874	Priorities are implemented by allocating some space for each
		3875	priority. When doing priority-based operations, libev usually has to
		3876	linearly search all the priorities, but starting/stopping and activating
		3877	watchers becomes O(1) with respect to priority handling.
		3878
		3879	=item Sending an ev_async: O(1)
		3880
		3881	=item Processing ev_async_send: O(number_of_async_watchers)
		3882
		3883	=item Processing signals: O(max_signal_number)
		3884
		3885	Sending involves a system call I<iff> there were no other C<ev_async_send>
		3886	calls in the current loop iteration. Checking for async and signal events
		3887	involves iterating over all running async watchers or all signal numbers.
		3888
		3889	=back
		3890
		3891
742	=head1 AUTHOR	3892	=head1 AUTHOR
743		3893
744	Marc Lehmann <libev@schmorp.de>.	3894	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
745		3895

Diff Legend

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