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 1.229 by root, Wed Apr 15 17:49:27 2009 UTC

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

Diff Legend

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