ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.29 by root, Thu Nov 22 12:28:28 2007 UTC vs.
Revision 1.232 by root, Thu Apr 16 06:17:26 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 REPRESENTATION	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
51		144
52	=head1 GLOBAL FUNCTIONS	145	=head1 GLOBAL FUNCTIONS
53		146
54	These functions can be called anytime, even before initialising the	147	These functions can be called anytime, even before initialising the
55	library in any way.	148	library in any way.
…		…
60		153
61	Returns the current time as libev would use it. Please note that the	154	Returns the current time as libev would use it. Please note that the
62	C<ev_now> function is usually faster and also often returns the timestamp	155	C<ev_now> function is usually faster and also often returns the timestamp
63	you actually want to know.	156	you actually want to know.
64		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 ()>.
		163
65	=item int ev_version_major ()	164	=item int ev_version_major ()
66		165
67	=item int ev_version_minor ()	166	=item int ev_version_minor ()
68		167
69	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
70	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
71	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
72	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
73	version of the library your program was compiled against.	172	version of the library your program was compiled against.
74		173
		174	These version numbers refer to the ABI version of the library, not the
		175	release version.
		176
75	Usually, it's a good idea to terminate if the major versions mismatch,	177	Usually, it's a good idea to terminate if the major versions mismatch,
76	as this indicates an incompatible change. Minor versions are usually	178	as this indicates an incompatible change. Minor versions are usually
77	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
78	not a problem.	180	not a problem.
79		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
80	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	221	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
81		222
82	Sets the allocation function to use (the prototype is similar to the	223	Sets the allocation function to use (the prototype is similar - the
83	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
84	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
85	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
86	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.
87		232
88	You could override this function in high-availability programs to, say,	233	You could override this function in high-availability programs to, say,
89	free some memory if it cannot allocate memory, to use a special allocator,	234	free some memory if it cannot allocate memory, to use a special allocator,
90	or even to sleep a while and retry until some memory is available.	235	or even to sleep a while and retry until some memory is available.
91		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
92	=item ev_set_syserr_cb (void (*cb)(const char *msg));	257	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
93		258
94	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
95	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
96	indicating the system call or subsystem causing the problem. If this	261	indicating the system call or subsystem causing the problem. If this
97	callback is set, then libev will expect it to remedy the sitution, no	262	callback is set, then libev will expect it to remedy the situation, no
98	matter what, when it returns. That is, libev will generally retry the	263	matter what, when it returns. That is, libev will generally retry the
99	requested operation, or, if the condition doesn't go away, do bad stuff	264	requested operation, or, if the condition doesn't go away, do bad stuff
100	(such as abort).	265	(such as abort).
101		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
102	=back	279	=back
103		280
104	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	281	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
105		282
106	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>
107	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>
108	events, and dynamically created loops which do not.	285	I<function>).
109		286
110	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
111	in your main thread (or in a separate thread) and for each thread you	288	supports signals and child events, and dynamically created loops which do
112	create, you also create another event loop. Libev itself does no locking	289	not.
113	whatsoever, so if you mix calls to the same event loop in different
114	threads, make sure you lock (this is usually a bad idea, though, even if
115	done correctly, because it's hideous and inefficient).
116		290
117	=over 4	291	=over 4
118		292
119	=item struct ev_loop *ev_default_loop (unsigned int flags)	293	=item struct ev_loop *ev_default_loop (unsigned int flags)
120		294
121	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
122	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
123	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
124	flags).	298	flags. If that is troubling you, check C<ev_backend ()> afterwards).
125		299
126	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
127	function.	301	function.
128		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
129	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
130	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>).
131		316
132	It supports the following flags:	317	The following flags are supported:
133		318
134	=over 4	319	=over 4
135		320
136	=item C<EVFLAG_AUTO>	321	=item C<EVFLAG_AUTO>
137		322
138	The default flags value. Use this if you have no clue (it's the right	323	The default flags value. Use this if you have no clue (it's the right
139	thing, believe me).	324	thing, believe me).
140		325
141	=item C<EVFLAG_NOENV>	326	=item C<EVFLAG_NOENV>
142		327
143	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
144	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
145	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	330	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
146	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
147	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
148	around bugs.	333	around bugs.
149		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
150	=item C<EVMETHOD_SELECT> (value 1, portable select backend)	355	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
151		356
152	This is your standard select(2) backend. Not I<completely> standard, as	357	This is your standard select(2) backend. Not I<completely> standard, as
153	libev tries to roll its own fd_set with no limits on the number of fds,	358	libev tries to roll its own fd_set with no limits on the number of fds,
154	but if that fails, expect a fairly low limit on the number of fds when	359	but if that fails, expect a fairly low limit on the number of fds when
155	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	360	using this backend. It doesn't scale too well (O(highest_fd)), but its
156	the fastest backend for a low number of fds.	361	usually the fastest backend for a low number of (low-numbered :) fds.
157		362
		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.
		369
		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).
		373
158	=item C<EVMETHOD_POLL> (value 2, poll backend, available everywhere except on windows)	374	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
159		375
160	And this is your standard poll(2) backend. It's more complicated than	376	And this is your standard poll(2) backend. It's more complicated
161	select, but handles sparse fds better and has no artificial limit on the	377	than select, but handles sparse fds better and has no artificial
162	number of fds you can use (except it will slow down considerably with a	378	limit on the number of fds you can use (except it will slow down
163	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	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.
164		382
		383	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		384	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		385
165	=item C<EVMETHOD_EPOLL> (value 4, Linux)	386	=item C<EVBACKEND_EPOLL> (value 4, Linux)
166		387
167	For few fds, this backend is a bit little slower than poll and select,	388	For few fds, this backend is a bit little slower than poll and select,
168	but it scales phenomenally better. While poll and select usually scale like	389	but it scales phenomenally better. While poll and select usually scale
169	O(total_fds) where n is the total number of fds (or the highest fd), epoll scales	390	like O(total_fds) where n is the total number of fds (or the highest fd),
170	either O(1) or O(active_fds).	391	epoll scales either O(1) or O(active_fds).
171		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
172	While stopping and starting an I/O watcher in the same iteration will	410	While stopping, setting and starting an I/O watcher in the same iteration
173	result in some caching, there is still a syscall per such incident	411	will result in some caching, there is still a system call per such
174	(because the fd could point to a different file description now), so its	412	incident (because the same I<file descriptor> could point to a different
175	best to avoid that. Also, dup()ed file descriptors might not work very	413	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
176	well if you register events for both fds.	414	file descriptors might not work very well if you register events for both
		415	file descriptors.
177		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
178	=item C<EVMETHOD_KQUEUE> (value 8, most BSD clones)	435	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
179		436
180	Kqueue deserves special mention, as at the time of this writing, it	437	Kqueue deserves special mention, as at the time of this writing, it
181	was broken on all BSDs except NetBSD (usually it doesn't work with	438	was broken on all BSDs except NetBSD (usually it doesn't work reliably
182	anything but sockets and pipes, except on Darwin, where of course its	439	with anything but sockets and pipes, except on Darwin, where of course
183	completely useless). For this reason its not being "autodetected" unless	440	it's completely useless). Unlike epoll, however, whose brokenness
184	you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO).	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.
185		450
186	It scales in the same way as the epoll backend, but the interface to the	451	It scales in the same way as the epoll backend, but the interface to the
187	kernel is more efficient (which says nothing about its actual speed, of	452	kernel is more efficient (which says nothing about its actual speed, of
188	course). While starting and stopping an I/O watcher does not cause an	453	course). While stopping, setting and starting an I/O watcher does never
189	extra syscall as with epoll, it still adds up to four event changes per	454	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
190	incident, so its best to avoid that.	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
191		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
192	=item C<EVMETHOD_DEVPOLL> (value 16, Solaris 8)	472	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
193		473
194	This is not implemented yet (and might never be).	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.
195		478
196	=item C<EVMETHOD_PORT> (value 32, Solaris 10)	479	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
197		480
198	This uses the Solaris 10 port mechanism. As with everything on Solaris,	481	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
199	it's really slow, but it still scales very well (O(active_fds)).	482	it's really slow, but it still scales very well (O(active_fds)).
200		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
201	=item C<EVMETHOD_ALL>	501	=item C<EVBACKEND_ALL>
202		502
203	Try all backends (even potentially broken ones that wouldn't be tried	503	Try all backends (even potentially broken ones that wouldn't be tried
204	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	504	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
205	C<EVMETHOD_ALL & ~EVMETHOD_KQUEUE>.	505	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		506
		507	It is definitely not recommended to use this flag.
206		508
207	=back	509	=back
208		510
209	If one or more of these are ored into the flags value, then only these	511	If one or more of these are or'ed into the flags value, then only these
210	backends will be tried (in the reverse order as given here). If none are	512	backends will be tried (in the reverse order as listed here). If none are
211	specified, most compiled-in backend will be tried, usually in reverse	513	specified, all backends in C<ev_recommended_backends ()> will be tried.
212	order of their flag values :)	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);
213		531
214	=item struct ev_loop *ev_loop_new (unsigned int flags)	532	=item struct ev_loop *ev_loop_new (unsigned int flags)
215		533
216	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
217	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
218	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
219	undefined behaviour (or a failed assertion if assertions are enabled).	537	undefined behaviour (or a failed assertion if assertions are enabled).
220		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
221	=item ev_default_destroy ()	549	=item ev_default_destroy ()
222		550
223	Destroys the default loop again (frees all memory and kernel state	551	Destroys the default loop again (frees all memory and kernel state
224	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
225	any way whatsoever, although you cannot 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>).
226		567
227	=item ev_loop_destroy (loop)	568	=item ev_loop_destroy (loop)
228		569
229	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
230	earlier call to C<ev_loop_new>.	571	earlier call to C<ev_loop_new>.
231		572
232	=item ev_default_fork ()	573	=item ev_default_fork ()
233		574
		575	This function sets a flag that causes subsequent C<ev_loop> iterations
234	This function reinitialises the kernel state for backends that have	576	to reinitialise the kernel state for backends that have one. Despite the
235	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
236	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
237	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.
238		581
239	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
240	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
241	have to call it.	584	you just fork+exec, you don't have to call it at all.
242		585
243	The function itself is quite fast and it's usually not a problem to call	586	The function itself is quite fast and it's usually not a problem to call
244	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
245	quite nicely into a call to C<pthread_atfork>:	588	quite nicely into a call to C<pthread_atfork>:
246		589
…		…
248		591
249	=item ev_loop_fork (loop)	592	=item ev_loop_fork (loop)
250		593
251	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
252	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
253	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.
254		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
255	=item unsigned int ev_method (loop)	614	=item unsigned int ev_backend (loop)
256		615
257	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
258	use.	617	use.
259		618
260	=item ev_tstamp ev_now (loop)	619	=item ev_tstamp ev_now (loop)
261		620
262	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
263	got events and started processing them. This timestamp does not change	622	received events and started processing them. This timestamp does not
264	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
265	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
266	occuring (or more correctly, the mainloop finding out about it).	625	event occurring (or more correctly, libev finding out about it).
		626
		627	=item ev_now_update (loop)
		628
		629	Establishes the current time by querying the kernel, updating the time
		630	returned by C<ev_now ()> in the progress. This is a costly operation and
		631	is usually done automatically within C<ev_loop ()>.
		632
		633	This function is rarely useful, but when some event callback runs for a
		634	very long time without entering the event loop, updating libev's idea of
		635	the current time is a good idea.
		636
		637	See also "The special problem of time updates" in the C<ev_timer> section.
		638
		639	=item ev_suspend (loop)
		640
		641	=item ev_resume (loop)
		642
		643	These two functions suspend and resume a loop, for use when the loop is
		644	not used for a while and timeouts should not be processed.
		645
		646	A typical use case would be an interactive program such as a game: When
		647	the user presses C<^Z> to suspend the game and resumes it an hour later it
		648	would be best to handle timeouts as if no time had actually passed while
		649	the program was suspended. This can be achieved by calling C<ev_suspend>
		650	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		651	C<ev_resume> directly afterwards to resume timer processing.
		652
		653	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		654	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		655	will be rescheduled (that is, they will lose any events that would have
		656	occured while suspended).
		657
		658	After calling C<ev_suspend> you B<must not> call I<any> function on the
		659	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		660	without a previous call to C<ev_suspend>.
		661
		662	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		663	event loop time (see C<ev_now_update>).
267		664
268	=item ev_loop (loop, int flags)	665	=item ev_loop (loop, int flags)
269		666
270	Finally, this is it, the event handler. This function usually is called	667	Finally, this is it, the event handler. This function usually is called
271	after you initialised all your watchers and you want to start handling	668	after you initialised all your watchers and you want to start handling
272	events.	669	events.
273		670
274	If the flags argument is specified as 0, it will not return until either	671	If the flags argument is specified as C<0>, it will not return until
275	no event watchers are active anymore or C<ev_unloop> was called.	672	either no event watchers are active anymore or C<ev_unloop> was called.
		673
		674	Please note that an explicit C<ev_unloop> is usually better than
		675	relying on all watchers to be stopped when deciding when a program has
		676	finished (especially in interactive programs), but having a program
		677	that automatically loops as long as it has to and no longer by virtue
		678	of relying on its watchers stopping correctly, that is truly a thing of
		679	beauty.
276		680
277	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	681	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
278	those events and any outstanding ones, but will not block your process in	682	those events and any already outstanding ones, but will not block your
279	case there are no events and will return after one iteration of the loop.	683	process in case there are no events and will return after one iteration of
		684	the loop.
280		685
281	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	686	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
282	neccessary) and will handle those and any outstanding ones. It will block	687	necessary) and will handle those and any already outstanding ones. It
283	your process until at least one new event arrives, and will return after	688	will block your process until at least one new event arrives (which could
		689	be an event internal to libev itself, so there is no guarantee that a
		690	user-registered callback will be called), and will return after one
284	one iteration of the loop.	691	iteration of the loop.
285		692
286	This flags value could be used to implement alternative looping	693	This is useful if you are waiting for some external event in conjunction
287	constructs, but the C<prepare> and C<check> watchers provide a better and	694	with something not expressible using other libev watchers (i.e. "roll your
288	more generic mechanism.	695	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		696	usually a better approach for this kind of thing.
289		697
290	Here are the gory details of what ev_loop does:	698	Here are the gory details of what C<ev_loop> does:
291		699
292	1. If there are no active watchers (reference count is zero), return.	700	- Before the first iteration, call any pending watchers.
		701	* If EVFLAG_FORKCHECK was used, check for a fork.
		702	- If a fork was detected (by any means), queue and call all fork watchers.
293	2. Queue and immediately call all prepare watchers.	703	- Queue and call all prepare watchers.
294	3. If we have been forked, recreate the kernel state.	704	- If we have been forked, detach and recreate the kernel state
		705	as to not disturb the other process.
295	4. Update the kernel state with all outstanding changes.	706	- Update the kernel state with all outstanding changes.
296	5. Update the "event loop time".	707	- Update the "event loop time" (ev_now ()).
297	6. Calculate for how long to block.	708	- Calculate for how long to sleep or block, if at all
		709	(active idle watchers, EVLOOP_NONBLOCK or not having
		710	any active watchers at all will result in not sleeping).
		711	- Sleep if the I/O and timer collect interval say so.
298	7. Block the process, waiting for events.	712	- Block the process, waiting for any events.
		713	- Queue all outstanding I/O (fd) events.
299	8. Update the "event loop time" and do time jump handling.	714	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
300	9. Queue all outstanding timers.	715	- Queue all expired timers.
301	10. Queue all outstanding periodics.	716	- Queue all expired periodics.
302	11. If no events are pending now, queue all idle watchers.	717	- Unless any events are pending now, queue all idle watchers.
303	12. Queue all check watchers.	718	- Queue all check watchers.
304	13. Call all queued watchers in reverse order (i.e. check watchers first).	719	- Call all queued watchers in reverse order (i.e. check watchers first).
		720	Signals and child watchers are implemented as I/O watchers, and will
		721	be handled here by queueing them when their watcher gets executed.
305	14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	722	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
306	was used, return, otherwise continue with step #1.	723	were used, or there are no active watchers, return, otherwise
		724	continue with step *.
		725
		726	Example: Queue some jobs and then loop until no events are outstanding
		727	anymore.
		728
		729	... queue jobs here, make sure they register event watchers as long
		730	... as they still have work to do (even an idle watcher will do..)
		731	ev_loop (my_loop, 0);
		732	... jobs done or somebody called unloop. yeah!
307		733
308	=item ev_unloop (loop, how)	734	=item ev_unloop (loop, how)
309		735
310	Can be used to make a call to C<ev_loop> return early (but only after it	736	Can be used to make a call to C<ev_loop> return early (but only after it
311	has processed all outstanding events). The C<how> argument must be either	737	has processed all outstanding events). The C<how> argument must be either
312	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	738	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
313	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	739	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
314		740
		741	This "unloop state" will be cleared when entering C<ev_loop> again.
		742
		743	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
		744
315	=item ev_ref (loop)	745	=item ev_ref (loop)
316		746
317	=item ev_unref (loop)	747	=item ev_unref (loop)
318		748
319	Ref/unref can be used to add or remove a reference count on the event	749	Ref/unref can be used to add or remove a reference count on the event
320	loop: Every watcher keeps one reference, and as long as the reference	750	loop: Every watcher keeps one reference, and as long as the reference
321	count is nonzero, C<ev_loop> will not return on its own. If you have	751	count is nonzero, C<ev_loop> will not return on its own.
		752
322	a watcher you never unregister that should not keep C<ev_loop> from	753	If you have a watcher you never unregister that should not keep C<ev_loop>
323	returning, ev_unref() after starting, and ev_ref() before stopping it. For	754	from returning, call ev_unref() after starting, and ev_ref() before
		755	stopping it.
		756
324	example, libev itself uses this for its internal signal pipe: It is not	757	As an example, libev itself uses this for its internal signal pipe: It
325	visible to the libev user and should not keep C<ev_loop> from exiting if	758	is not visible to the libev user and should not keep C<ev_loop> from
326	no event watchers registered by it are active. It is also an excellent	759	exiting if no event watchers registered by it are active. It is also an
327	way to do this for generic recurring timers or from within third-party	760	excellent way to do this for generic recurring timers or from within
328	libraries. Just remember to I<unref after start> and I<ref before stop>.	761	third-party libraries. Just remember to I<unref after start> and I<ref
		762	before stop> (but only if the watcher wasn't active before, or was active
		763	before, respectively. Note also that libev might stop watchers itself
		764	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		765	in the callback).
		766
		767	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
		768	running when nothing else is active.
		769
		770	ev_signal exitsig;
		771	ev_signal_init (&exitsig, sig_cb, SIGINT);
		772	ev_signal_start (loop, &exitsig);
		773	evf_unref (loop);
		774
		775	Example: For some weird reason, unregister the above signal handler again.
		776
		777	ev_ref (loop);
		778	ev_signal_stop (loop, &exitsig);
		779
		780	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		781
		782	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		783
		784	These advanced functions influence the time that libev will spend waiting
		785	for events. Both time intervals are by default C<0>, meaning that libev
		786	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		787	latency.
		788
		789	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		790	allows libev to delay invocation of I/O and timer/periodic callbacks
		791	to increase efficiency of loop iterations (or to increase power-saving
		792	opportunities).
		793
		794	The idea is that sometimes your program runs just fast enough to handle
		795	one (or very few) event(s) per loop iteration. While this makes the
		796	program responsive, it also wastes a lot of CPU time to poll for new
		797	events, especially with backends like C<select ()> which have a high
		798	overhead for the actual polling but can deliver many events at once.
		799
		800	By setting a higher I<io collect interval> you allow libev to spend more
		801	time collecting I/O events, so you can handle more events per iteration,
		802	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		803	C<ev_timer>) will be not affected. Setting this to a non-null value will
		804	introduce an additional C<ev_sleep ()> call into most loop iterations.
		805
		806	Likewise, by setting a higher I<timeout collect interval> you allow libev
		807	to spend more time collecting timeouts, at the expense of increased
		808	latency/jitter/inexactness (the watcher callback will be called
		809	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		810	value will not introduce any overhead in libev.
		811
		812	Many (busy) programs can usually benefit by setting the I/O collect
		813	interval to a value near C<0.1> or so, which is often enough for
		814	interactive servers (of course not for games), likewise for timeouts. It
		815	usually doesn't make much sense to set it to a lower value than C<0.01>,
		816	as this approaches the timing granularity of most systems.
		817
		818	Setting the I<timeout collect interval> can improve the opportunity for
		819	saving power, as the program will "bundle" timer callback invocations that
		820	are "near" in time together, by delaying some, thus reducing the number of
		821	times the process sleeps and wakes up again. Another useful technique to
		822	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		823	they fire on, say, one-second boundaries only.
		824
		825	=item ev_loop_verify (loop)
		826
		827	This function only does something when C<EV_VERIFY> support has been
		828	compiled in, which is the default for non-minimal builds. It tries to go
		829	through all internal structures and checks them for validity. If anything
		830	is found to be inconsistent, it will print an error message to standard
		831	error and call C<abort ()>.
		832
		833	This can be used to catch bugs inside libev itself: under normal
		834	circumstances, this function will never abort as of course libev keeps its
		835	data structures consistent.
329		836
330	=back	837	=back
331		838
		839
332	=head1 ANATOMY OF A WATCHER	840	=head1 ANATOMY OF A WATCHER
		841
		842	In the following description, uppercase C<TYPE> in names stands for the
		843	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		844	watchers and C<ev_io_start> for I/O watchers.
333		845
334	A watcher is a structure that you create and register to record your	846	A watcher is a structure that you create and register to record your
335	interest in some event. For instance, if you want to wait for STDIN to	847	interest in some event. For instance, if you want to wait for STDIN to
336	become readable, you would create an C<ev_io> watcher for that:	848	become readable, you would create an C<ev_io> watcher for that:
337		849
338	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	850	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
339	{	851	{
340	ev_io_stop (w);	852	ev_io_stop (w);
341	ev_unloop (loop, EVUNLOOP_ALL);	853	ev_unloop (loop, EVUNLOOP_ALL);
342	}	854	}
343		855
344	struct ev_loop *loop = ev_default_loop (0);	856	struct ev_loop *loop = ev_default_loop (0);
		857
345	struct ev_io stdin_watcher;	858	ev_io stdin_watcher;
		859
346	ev_init (&stdin_watcher, my_cb);	860	ev_init (&stdin_watcher, my_cb);
347	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	861	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
348	ev_io_start (loop, &stdin_watcher);	862	ev_io_start (loop, &stdin_watcher);
		863
349	ev_loop (loop, 0);	864	ev_loop (loop, 0);
350		865
351	As you can see, you are responsible for allocating the memory for your	866	As you can see, you are responsible for allocating the memory for your
352	watcher structures (and it is usually a bad idea to do this on the stack,	867	watcher structures (and it is I<usually> a bad idea to do this on the
353	although this can sometimes be quite valid).	868	stack).
		869
		870	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		871	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
354		872
355	Each watcher structure must be initialised by a call to C<ev_init	873	Each watcher structure must be initialised by a call to C<ev_init
356	(watcher *, callback)>, which expects a callback to be provided. This	874	(watcher *, callback)>, which expects a callback to be provided. This
357	callback gets invoked each time the event occurs (or, in the case of io	875	callback gets invoked each time the event occurs (or, in the case of I/O
358	watchers, each time the event loop detects that the file descriptor given	876	watchers, each time the event loop detects that the file descriptor given
359	is readable and/or writable).	877	is readable and/or writable).
360		878
361	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	879	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
362	with arguments specific to this watcher type. There is also a macro	880	macro to configure it, with arguments specific to the watcher type. There
363	to combine initialisation and setting in one call: C<< ev_<type>_init	881	is also a macro to combine initialisation and setting in one call: C<<
364	(watcher *, callback, ...) >>.	882	ev_TYPE_init (watcher *, callback, ...) >>.
365		883
366	To make the watcher actually watch out for events, you have to start it	884	To make the watcher actually watch out for events, you have to start it
367	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	885	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
368	*) >>), and you can stop watching for events at any time by calling the	886	*) >>), and you can stop watching for events at any time by calling the
369	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	887	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
370		888
371	As long as your watcher is active (has been started but not stopped) you	889	As long as your watcher is active (has been started but not stopped) you
372	must not touch the values stored in it. Most specifically you must never	890	must not touch the values stored in it. Most specifically you must never
373	reinitialise it or call its set method.	891	reinitialise it or call its C<ev_TYPE_set> macro.
374
375	You can check whether an event is active by calling the C<ev_is_active
376	(watcher *)> macro. To see whether an event is outstanding (but the
377	callback for it has not been called yet) you can use the C<ev_is_pending
378	(watcher *)> macro.
379		892
380	Each and every callback receives the event loop pointer as first, the	893	Each and every callback receives the event loop pointer as first, the
381	registered watcher structure as second, and a bitset of received events as	894	registered watcher structure as second, and a bitset of received events as
382	third argument.	895	third argument.
383		896
…		…
407	The signal specified in the C<ev_signal> watcher has been received by a thread.	920	The signal specified in the C<ev_signal> watcher has been received by a thread.
408		921
409	=item C<EV_CHILD>	922	=item C<EV_CHILD>
410		923
411	The pid specified in the C<ev_child> watcher has received a status change.	924	The pid specified in the C<ev_child> watcher has received a status change.
		925
		926	=item C<EV_STAT>
		927
		928	The path specified in the C<ev_stat> watcher changed its attributes somehow.
412		929
413	=item C<EV_IDLE>	930	=item C<EV_IDLE>
414		931
415	The C<ev_idle> watcher has determined that you have nothing better to do.	932	The C<ev_idle> watcher has determined that you have nothing better to do.
416		933
…		…
424	received events. Callbacks of both watcher types can start and stop as	941	received events. Callbacks of both watcher types can start and stop as
425	many watchers as they want, and all of them will be taken into account	942	many watchers as they want, and all of them will be taken into account
426	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	943	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
427	C<ev_loop> from blocking).	944	C<ev_loop> from blocking).
428		945
		946	=item C<EV_EMBED>
		947
		948	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		949
		950	=item C<EV_FORK>
		951
		952	The event loop has been resumed in the child process after fork (see
		953	C<ev_fork>).
		954
		955	=item C<EV_ASYNC>
		956
		957	The given async watcher has been asynchronously notified (see C<ev_async>).
		958
		959	=item C<EV_CUSTOM>
		960
		961	Not ever sent (or otherwise used) by libev itself, but can be freely used
		962	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		963
429	=item C<EV_ERROR>	964	=item C<EV_ERROR>
430		965
431	An unspecified error has occured, the watcher has been stopped. This might	966	An unspecified error has occurred, the watcher has been stopped. This might
432	happen because the watcher could not be properly started because libev	967	happen because the watcher could not be properly started because libev
433	ran out of memory, a file descriptor was found to be closed or any other	968	ran out of memory, a file descriptor was found to be closed or any other
		969	problem. Libev considers these application bugs.
		970
434	problem. You best act on it by reporting the problem and somehow coping	971	You best act on it by reporting the problem and somehow coping with the
435	with the watcher being stopped.	972	watcher being stopped. Note that well-written programs should not receive
		973	an error ever, so when your watcher receives it, this usually indicates a
		974	bug in your program.
436		975
437	Libev will usually signal a few "dummy" events together with an error,	976	Libev will usually signal a few "dummy" events together with an error, for
438	for example it might indicate that a fd is readable or writable, and if	977	example it might indicate that a fd is readable or writable, and if your
439	your callbacks is well-written it can just attempt the operation and cope	978	callbacks is well-written it can just attempt the operation and cope with
440	with the error from read() or write(). This will not work in multithreaded	979	the error from read() or write(). This will not work in multi-threaded
441	programs, though, so beware.	980	programs, though, as the fd could already be closed and reused for another
		981	thing, so beware.
442		982
443	=back	983	=back
444		984
		985	=head2 GENERIC WATCHER FUNCTIONS
		986
		987	=over 4
		988
		989	=item C<ev_init> (ev_TYPE *watcher, callback)
		990
		991	This macro initialises the generic portion of a watcher. The contents
		992	of the watcher object can be arbitrary (so C<malloc> will do). Only
		993	the generic parts of the watcher are initialised, you I<need> to call
		994	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		995	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		996	which rolls both calls into one.
		997
		998	You can reinitialise a watcher at any time as long as it has been stopped
		999	(or never started) and there are no pending events outstanding.
		1000
		1001	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1002	int revents)>.
		1003
		1004	Example: Initialise an C<ev_io> watcher in two steps.
		1005
		1006	ev_io w;
		1007	ev_init (&w, my_cb);
		1008	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1009
		1010	=item C<ev_TYPE_set> (ev_TYPE *, [args])
		1011
		1012	This macro initialises the type-specific parts of a watcher. You need to
		1013	call C<ev_init> at least once before you call this macro, but you can
		1014	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1015	macro on a watcher that is active (it can be pending, however, which is a
		1016	difference to the C<ev_init> macro).
		1017
		1018	Although some watcher types do not have type-specific arguments
		1019	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1020
		1021	See C<ev_init>, above, for an example.
		1022
		1023	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1024
		1025	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1026	calls into a single call. This is the most convenient method to initialise
		1027	a watcher. The same limitations apply, of course.
		1028
		1029	Example: Initialise and set an C<ev_io> watcher in one step.
		1030
		1031	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1032
		1033	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)
		1034
		1035	Starts (activates) the given watcher. Only active watchers will receive
		1036	events. If the watcher is already active nothing will happen.
		1037
		1038	Example: Start the C<ev_io> watcher that is being abused as example in this
		1039	whole section.
		1040
		1041	ev_io_start (EV_DEFAULT_UC, &w);
		1042
		1043	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)
		1044
		1045	Stops the given watcher if active, and clears the pending status (whether
		1046	the watcher was active or not).
		1047
		1048	It is possible that stopped watchers are pending - for example,
		1049	non-repeating timers are being stopped when they become pending - but
		1050	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1051	pending. If you want to free or reuse the memory used by the watcher it is
		1052	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1053
		1054	=item bool ev_is_active (ev_TYPE *watcher)
		1055
		1056	Returns a true value iff the watcher is active (i.e. it has been started
		1057	and not yet been stopped). As long as a watcher is active you must not modify
		1058	it.
		1059
		1060	=item bool ev_is_pending (ev_TYPE *watcher)
		1061
		1062	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1063	events but its callback has not yet been invoked). As long as a watcher
		1064	is pending (but not active) you must not call an init function on it (but
		1065	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1066	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1067	it).
		1068
		1069	=item callback ev_cb (ev_TYPE *watcher)
		1070
		1071	Returns the callback currently set on the watcher.
		1072
		1073	=item ev_cb_set (ev_TYPE *watcher, callback)
		1074
		1075	Change the callback. You can change the callback at virtually any time
		1076	(modulo threads).
		1077
		1078	=item ev_set_priority (ev_TYPE *watcher, priority)
		1079
		1080	=item int ev_priority (ev_TYPE *watcher)
		1081
		1082	Set and query the priority of the watcher. The priority is a small
		1083	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1084	(default: C<-2>). Pending watchers with higher priority will be invoked
		1085	before watchers with lower priority, but priority will not keep watchers
		1086	from being executed (except for C<ev_idle> watchers).
		1087
		1088	See L<
		1089
		1090	This means that priorities are I<only> used for ordering callback
		1091	invocation after new events have been received. This is useful, for
		1092	example, to reduce latency after idling, or more often, to bind two
		1093	watchers on the same event and make sure one is called first.
		1094
		1095	If you need to suppress invocation when higher priority events are pending
		1096	you need to look at C<ev_idle> watchers, which provide this functionality.
		1097
		1098	You I<must not> change the priority of a watcher as long as it is active or
		1099	pending.
		1100
		1101	The default priority used by watchers when no priority has been set is
		1102	always C<0>, which is supposed to not be too high and not be too low :).
		1103
		1104	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1105	fine, as long as you do not mind that the priority value you query might
		1106	or might not have been clamped to the valid range.
		1107
		1108	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1109
		1110	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1111	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1112	can deal with that fact, as both are simply passed through to the
		1113	callback.
		1114
		1115	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1116
		1117	If the watcher is pending, this function clears its pending status and
		1118	returns its C<revents> bitset (as if its callback was invoked). If the
		1119	watcher isn't pending it does nothing and returns C<0>.
		1120
		1121	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1122	callback to be invoked, which can be accomplished with this function.
		1123
		1124	=back
		1125
		1126
445	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1127	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
446		1128
447	Each watcher has, by default, a member C<void *data> that you can change	1129	Each watcher has, by default, a member C<void *data> that you can change
448	and read at any time, libev will completely ignore it. This can be used	1130	and read at any time: libev will completely ignore it. This can be used
449	to associate arbitrary data with your watcher. If you need more data and	1131	to associate arbitrary data with your watcher. If you need more data and
450	don't want to allocate memory and store a pointer to it in that data	1132	don't want to allocate memory and store a pointer to it in that data
451	member, you can also "subclass" the watcher type and provide your own	1133	member, you can also "subclass" the watcher type and provide your own
452	data:	1134	data:
453		1135
454	struct my_io	1136	struct my_io
455	{	1137	{
456	struct ev_io io;	1138	ev_io io;
457	int otherfd;	1139	int otherfd;
458	void *somedata;	1140	void *somedata;
459	struct whatever *mostinteresting;	1141	struct whatever *mostinteresting;
460	}	1142	};
		1143
		1144	...
		1145	struct my_io w;
		1146	ev_io_init (&w.io, my_cb, fd, EV_READ);
461		1147
462	And since your callback will be called with a pointer to the watcher, you	1148	And since your callback will be called with a pointer to the watcher, you
463	can cast it back to your own type:	1149	can cast it back to your own type:
464		1150
465	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1151	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
466	{	1152	{
467	struct my_io *w = (struct my_io *)w_;	1153	struct my_io *w = (struct my_io *)w_;
468	...	1154	...
469	}	1155	}
470		1156
471	More interesting and less C-conformant ways of catsing your callback type	1157	More interesting and less C-conformant ways of casting your callback type
472	have been omitted....	1158	instead have been omitted.
		1159
		1160	Another common scenario is to use some data structure with multiple
		1161	embedded watchers:
		1162
		1163	struct my_biggy
		1164	{
		1165	int some_data;
		1166	ev_timer t1;
		1167	ev_timer t2;
		1168	}
		1169
		1170	In this case getting the pointer to C<my_biggy> is a bit more
		1171	complicated: Either you store the address of your C<my_biggy> struct
		1172	in the C<data> member of the watcher (for woozies), or you need to use
		1173	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1174	programmers):
		1175
		1176	#include <stddef.h>
		1177
		1178	static void
		1179	t1_cb (EV_P_ ev_timer *w, int revents)
		1180	{
		1181	struct my_biggy big = (struct my_biggy *
		1182	(((char *)w) - offsetof (struct my_biggy, t1));
		1183	}
		1184
		1185	static void
		1186	t2_cb (EV_P_ ev_timer *w, int revents)
		1187	{
		1188	struct my_biggy big = (struct my_biggy *
		1189	(((char *)w) - offsetof (struct my_biggy, t2));
		1190	}
473		1191
474		1192
475	=head1 WATCHER TYPES	1193	=head1 WATCHER TYPES
476		1194
477	This section describes each watcher in detail, but will not repeat	1195	This section describes each watcher in detail, but will not repeat
478	information given in the last section.	1196	information given in the last section. Any initialisation/set macros,
		1197	functions and members specific to the watcher type are explained.
479		1198
		1199	Members are additionally marked with either I<[read-only]>, meaning that,
		1200	while the watcher is active, you can look at the member and expect some
		1201	sensible content, but you must not modify it (you can modify it while the
		1202	watcher is stopped to your hearts content), or I<[read-write]>, which
		1203	means you can expect it to have some sensible content while the watcher
		1204	is active, but you can also modify it. Modifying it may not do something
		1205	sensible or take immediate effect (or do anything at all), but libev will
		1206	not crash or malfunction in any way.
		1207
		1208
480	=head2 C<ev_io> - is this file descriptor readable or writable	1209	=head2 C<ev_io> - is this file descriptor readable or writable?
481		1210
482	I/O watchers check whether a file descriptor is readable or writable	1211	I/O watchers check whether a file descriptor is readable or writable
483	in each iteration of the event loop (This behaviour is called	1212	in each iteration of the event loop, or, more precisely, when reading
484	level-triggering because you keep receiving events as long as the	1213	would not block the process and writing would at least be able to write
485	condition persists. Remember you can stop the watcher if you don't want to	1214	some data. This behaviour is called level-triggering because you keep
486	act on the event and neither want to receive future events).	1215	receiving events as long as the condition persists. Remember you can stop
		1216	the watcher if you don't want to act on the event and neither want to
		1217	receive future events.
487		1218
488	In general you can register as many read and/or write event watchers per	1219	In general you can register as many read and/or write event watchers per
489	fd as you want (as long as you don't confuse yourself). Setting all file	1220	fd as you want (as long as you don't confuse yourself). Setting all file
490	descriptors to non-blocking mode is also usually a good idea (but not	1221	descriptors to non-blocking mode is also usually a good idea (but not
491	required if you know what you are doing).	1222	required if you know what you are doing).
492		1223
493	You have to be careful with dup'ed file descriptors, though. Some backends	1224	If you cannot use non-blocking mode, then force the use of a
494	(the linux epoll backend is a notable example) cannot handle dup'ed file	1225	known-to-be-good backend (at the time of this writing, this includes only
495	descriptors correctly if you register interest in two or more fds pointing	1226	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>).
496	to the same underlying file/socket etc. description (that is, they share
497	the same underlying "file open").
498		1227
499	If you must do this, then force the use of a known-to-be-good backend	1228	Another thing you have to watch out for is that it is quite easy to
500	(at the time of this writing, this includes only EVMETHOD_SELECT and	1229	receive "spurious" readiness notifications, that is your callback might
501	EVMETHOD_POLL).	1230	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1231	because there is no data. Not only are some backends known to create a
		1232	lot of those (for example Solaris ports), it is very easy to get into
		1233	this situation even with a relatively standard program structure. Thus
		1234	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1235	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1236
		1237	If you cannot run the fd in non-blocking mode (for example you should
		1238	not play around with an Xlib connection), then you have to separately
		1239	re-test whether a file descriptor is really ready with a known-to-be good
		1240	interface such as poll (fortunately in our Xlib example, Xlib already
		1241	does this on its own, so its quite safe to use). Some people additionally
		1242	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1243	indefinitely.
		1244
		1245	But really, best use non-blocking mode.
		1246
		1247	=head3 The special problem of disappearing file descriptors
		1248
		1249	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1250	descriptor (either due to calling C<close> explicitly or any other means,
		1251	such as C<dup2>). The reason is that you register interest in some file
		1252	descriptor, but when it goes away, the operating system will silently drop
		1253	this interest. If another file descriptor with the same number then is
		1254	registered with libev, there is no efficient way to see that this is, in
		1255	fact, a different file descriptor.
		1256
		1257	To avoid having to explicitly tell libev about such cases, libev follows
		1258	the following policy: Each time C<ev_io_set> is being called, libev
		1259	will assume that this is potentially a new file descriptor, otherwise
		1260	it is assumed that the file descriptor stays the same. That means that
		1261	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1262	descriptor even if the file descriptor number itself did not change.
		1263
		1264	This is how one would do it normally anyway, the important point is that
		1265	the libev application should not optimise around libev but should leave
		1266	optimisations to libev.
		1267
		1268	=head3 The special problem of dup'ed file descriptors
		1269
		1270	Some backends (e.g. epoll), cannot register events for file descriptors,
		1271	but only events for the underlying file descriptions. That means when you
		1272	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1273	events for them, only one file descriptor might actually receive events.
		1274
		1275	There is no workaround possible except not registering events
		1276	for potentially C<dup ()>'ed file descriptors, or to resort to
		1277	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1278
		1279	=head3 The special problem of fork
		1280
		1281	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1282	useless behaviour. Libev fully supports fork, but needs to be told about
		1283	it in the child.
		1284
		1285	To support fork in your programs, you either have to call
		1286	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1287	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1288	C<EVBACKEND_POLL>.
		1289
		1290	=head3 The special problem of SIGPIPE
		1291
		1292	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1293	when writing to a pipe whose other end has been closed, your program gets
		1294	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1295	this is sensible behaviour, for daemons, this is usually undesirable.
		1296
		1297	So when you encounter spurious, unexplained daemon exits, make sure you
		1298	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1299	somewhere, as that would have given you a big clue).
		1300
		1301
		1302	=head3 Watcher-Specific Functions
502		1303
503	=over 4	1304	=over 4
504		1305
505	=item ev_io_init (ev_io *, callback, int fd, int events)	1306	=item ev_io_init (ev_io *, callback, int fd, int events)
506		1307
507	=item ev_io_set (ev_io *, int fd, int events)	1308	=item ev_io_set (ev_io *, int fd, int events)
508		1309
509	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1310	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
510	events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |	1311	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
511	EV_WRITE> to receive the given events.	1312	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1313
		1314	=item int fd [read-only]
		1315
		1316	The file descriptor being watched.
		1317
		1318	=item int events [read-only]
		1319
		1320	The events being watched.
512		1321
513	=back	1322	=back
514		1323
		1324	=head3 Examples
		1325
		1326	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1327	readable, but only once. Since it is likely line-buffered, you could
		1328	attempt to read a whole line in the callback.
		1329
		1330	static void
		1331	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1332	{
		1333	ev_io_stop (loop, w);
		1334	.. read from stdin here (or from w->fd) and handle any I/O errors
		1335	}
		1336
		1337	...
		1338	struct ev_loop *loop = ev_default_init (0);
		1339	ev_io stdin_readable;
		1340	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1341	ev_io_start (loop, &stdin_readable);
		1342	ev_loop (loop, 0);
		1343
		1344
515	=head2 C<ev_timer> - relative and optionally recurring timeouts	1345	=head2 C<ev_timer> - relative and optionally repeating timeouts
516		1346
517	Timer watchers are simple relative timers that generate an event after a	1347	Timer watchers are simple relative timers that generate an event after a
518	given time, and optionally repeating in regular intervals after that.	1348	given time, and optionally repeating in regular intervals after that.
519		1349
520	The timers are based on real time, that is, if you register an event that	1350	The timers are based on real time, that is, if you register an event that
521	times out after an hour and you reset your system clock to last years	1351	times out after an hour and you reset your system clock to January last
522	time, it will still time out after (roughly) and hour. "Roughly" because	1352	year, it will still time out after (roughly) one hour. "Roughly" because
523	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1353	detecting time jumps is hard, and some inaccuracies are unavoidable (the
524	monotonic clock option helps a lot here).	1354	monotonic clock option helps a lot here).
		1355
		1356	The callback is guaranteed to be invoked only I<after> its timeout has
		1357	passed. If multiple timers become ready during the same loop iteration
		1358	then the ones with earlier time-out values are invoked before ones with
		1359	later time-out values (but this is no longer true when a callback calls
		1360	C<ev_loop> recursively).
		1361
		1362	=head3 Be smart about timeouts
		1363
		1364	Many real-world problems involve some kind of timeout, usually for error
		1365	recovery. A typical example is an HTTP request - if the other side hangs,
		1366	you want to raise some error after a while.
		1367
		1368	What follows are some ways to handle this problem, from obvious and
		1369	inefficient to smart and efficient.
		1370
		1371	In the following, a 60 second activity timeout is assumed - a timeout that
		1372	gets reset to 60 seconds each time there is activity (e.g. each time some
		1373	data or other life sign was received).
		1374
		1375	=over 4
		1376
		1377	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1378
		1379	This is the most obvious, but not the most simple way: In the beginning,
		1380	start the watcher:
		1381
		1382	ev_timer_init (timer, callback, 60., 0.);
		1383	ev_timer_start (loop, timer);
		1384
		1385	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1386	and start it again:
		1387
		1388	ev_timer_stop (loop, timer);
		1389	ev_timer_set (timer, 60., 0.);
		1390	ev_timer_start (loop, timer);
		1391
		1392	This is relatively simple to implement, but means that each time there is
		1393	some activity, libev will first have to remove the timer from its internal
		1394	data structure and then add it again. Libev tries to be fast, but it's
		1395	still not a constant-time operation.
		1396
		1397	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1398
		1399	This is the easiest way, and involves using C<ev_timer_again> instead of
		1400	C<ev_timer_start>.
		1401
		1402	To implement this, configure an C<ev_timer> with a C<repeat> value
		1403	of C<60> and then call C<ev_timer_again> at start and each time you
		1404	successfully read or write some data. If you go into an idle state where
		1405	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1406	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1407
		1408	That means you can ignore both the C<ev_timer_start> function and the
		1409	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1410	member and C<ev_timer_again>.
		1411
		1412	At start:
		1413
		1414	ev_timer_init (timer, callback);
		1415	timer->repeat = 60.;
		1416	ev_timer_again (loop, timer);
		1417
		1418	Each time there is some activity:
		1419
		1420	ev_timer_again (loop, timer);
		1421
		1422	It is even possible to change the time-out on the fly, regardless of
		1423	whether the watcher is active or not:
		1424
		1425	timer->repeat = 30.;
		1426	ev_timer_again (loop, timer);
		1427
		1428	This is slightly more efficient then stopping/starting the timer each time
		1429	you want to modify its timeout value, as libev does not have to completely
		1430	remove and re-insert the timer from/into its internal data structure.
		1431
		1432	It is, however, even simpler than the "obvious" way to do it.
		1433
		1434	=item 3. Let the timer time out, but then re-arm it as required.
		1435
		1436	This method is more tricky, but usually most efficient: Most timeouts are
		1437	relatively long compared to the intervals between other activity - in
		1438	our example, within 60 seconds, there are usually many I/O events with
		1439	associated activity resets.
		1440
		1441	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1442	but remember the time of last activity, and check for a real timeout only
		1443	within the callback:
		1444
		1445	ev_tstamp last_activity; // time of last activity
		1446
		1447	static void
		1448	callback (EV_P_ ev_timer *w, int revents)
		1449	{
		1450	ev_tstamp now = ev_now (EV_A);
		1451	ev_tstamp timeout = last_activity + 60.;
		1452
		1453	// if last_activity + 60. is older than now, we did time out
		1454	if (timeout < now)
		1455	{
		1456	// timeout occured, take action
		1457	}
		1458	else
		1459	{
		1460	// callback was invoked, but there was some activity, re-arm
		1461	// the watcher to fire in last_activity + 60, which is
		1462	// guaranteed to be in the future, so "again" is positive:
		1463	w->repeat = timeout - now;
		1464	ev_timer_again (EV_A_ w);
		1465	}
		1466	}
		1467
		1468	To summarise the callback: first calculate the real timeout (defined
		1469	as "60 seconds after the last activity"), then check if that time has
		1470	been reached, which means something I<did>, in fact, time out. Otherwise
		1471	the callback was invoked too early (C<timeout> is in the future), so
		1472	re-schedule the timer to fire at that future time, to see if maybe we have
		1473	a timeout then.
		1474
		1475	Note how C<ev_timer_again> is used, taking advantage of the
		1476	C<ev_timer_again> optimisation when the timer is already running.
		1477
		1478	This scheme causes more callback invocations (about one every 60 seconds
		1479	minus half the average time between activity), but virtually no calls to
		1480	libev to change the timeout.
		1481
		1482	To start the timer, simply initialise the watcher and set C<last_activity>
		1483	to the current time (meaning we just have some activity :), then call the
		1484	callback, which will "do the right thing" and start the timer:
		1485
		1486	ev_timer_init (timer, callback);
		1487	last_activity = ev_now (loop);
		1488	callback (loop, timer, EV_TIMEOUT);
		1489
		1490	And when there is some activity, simply store the current time in
		1491	C<last_activity>, no libev calls at all:
		1492
		1493	last_actiivty = ev_now (loop);
		1494
		1495	This technique is slightly more complex, but in most cases where the
		1496	time-out is unlikely to be triggered, much more efficient.
		1497
		1498	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1499	callback :) - just change the timeout and invoke the callback, which will
		1500	fix things for you.
		1501
		1502	=item 4. Wee, just use a double-linked list for your timeouts.
		1503
		1504	If there is not one request, but many thousands (millions...), all
		1505	employing some kind of timeout with the same timeout value, then one can
		1506	do even better:
		1507
		1508	When starting the timeout, calculate the timeout value and put the timeout
		1509	at the I<end> of the list.
		1510
		1511	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1512	the list is expected to fire (for example, using the technique #3).
		1513
		1514	When there is some activity, remove the timer from the list, recalculate
		1515	the timeout, append it to the end of the list again, and make sure to
		1516	update the C<ev_timer> if it was taken from the beginning of the list.
		1517
		1518	This way, one can manage an unlimited number of timeouts in O(1) time for
		1519	starting, stopping and updating the timers, at the expense of a major
		1520	complication, and having to use a constant timeout. The constant timeout
		1521	ensures that the list stays sorted.
		1522
		1523	=back
		1524
		1525	So which method the best?
		1526
		1527	Method #2 is a simple no-brain-required solution that is adequate in most
		1528	situations. Method #3 requires a bit more thinking, but handles many cases
		1529	better, and isn't very complicated either. In most case, choosing either
		1530	one is fine, with #3 being better in typical situations.
		1531
		1532	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1533	rather complicated, but extremely efficient, something that really pays
		1534	off after the first million or so of active timers, i.e. it's usually
		1535	overkill :)
		1536
		1537	=head3 The special problem of time updates
		1538
		1539	Establishing the current time is a costly operation (it usually takes at
		1540	least two system calls): EV therefore updates its idea of the current
		1541	time only before and after C<ev_loop> collects new events, which causes a
		1542	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1543	lots of events in one iteration.
525		1544
526	The relative timeouts are calculated relative to the C<ev_now ()>	1545	The relative timeouts are calculated relative to the C<ev_now ()>
527	time. This is usually the right thing as this timestamp refers to the time	1546	time. This is usually the right thing as this timestamp refers to the time
528	of the event triggering whatever timeout you are modifying/starting. If	1547	of the event triggering whatever timeout you are modifying/starting. If
529	you suspect event processing to be delayed and you I<need> to base the timeout	1548	you suspect event processing to be delayed and you I<need> to base the
530	on the current time, use something like this to adjust for this:	1549	timeout on the current time, use something like this to adjust for this:
531		1550
532	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1551	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
533		1552
534	The callback is guarenteed to be invoked only when its timeout has passed,	1553	If the event loop is suspended for a long time, you can also force an
535	but if multiple timers become ready during the same loop iteration then	1554	update of the time returned by C<ev_now ()> by calling C<ev_now_update
536	order of execution is undefined.	1555	()>.
		1556
		1557	=head3 Watcher-Specific Functions and Data Members
537		1558
538	=over 4	1559	=over 4
539		1560
540	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1561	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
541		1562
542	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1563	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
543		1564
544	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1565	Configure the timer to trigger after C<after> seconds. If C<repeat>
545	C<0.>, then it will automatically be stopped. If it is positive, then the	1566	is C<0.>, then it will automatically be stopped once the timeout is
546	timer will automatically be configured to trigger again C<repeat> seconds	1567	reached. If it is positive, then the timer will automatically be
547	later, again, and again, until stopped manually.	1568	configured to trigger again C<repeat> seconds later, again, and again,
		1569	until stopped manually.
548		1570
549	The timer itself will do a best-effort at avoiding drift, that is, if you	1571	The timer itself will do a best-effort at avoiding drift, that is, if
550	configure a timer to trigger every 10 seconds, then it will trigger at	1572	you configure a timer to trigger every 10 seconds, then it will normally
551	exactly 10 second intervals. If, however, your program cannot keep up with	1573	trigger at exactly 10 second intervals. If, however, your program cannot
552	the timer (because it takes longer than those 10 seconds to do stuff) the	1574	keep up with the timer (because it takes longer than those 10 seconds to
553	timer will not fire more than once per event loop iteration.	1575	do stuff) the timer will not fire more than once per event loop iteration.
554		1576
555	=item ev_timer_again (loop)	1577	=item ev_timer_again (loop, ev_timer *)
556		1578
557	This will act as if the timer timed out and restart it again if it is	1579	This will act as if the timer timed out and restart it again if it is
558	repeating. The exact semantics are:	1580	repeating. The exact semantics are:
559		1581
		1582	If the timer is pending, its pending status is cleared.
		1583
560	If the timer is started but nonrepeating, stop it.	1584	If the timer is started but non-repeating, stop it (as if it timed out).
561		1585
562	If the timer is repeating, either start it if necessary (with the repeat	1586	If the timer is repeating, either start it if necessary (with the
563	value), or reset the running timer to the repeat value.	1587	C<repeat> value), or reset the running timer to the C<repeat> value.
564		1588
565	This sounds a bit complicated, but here is a useful and typical	1589	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
566	example: Imagine you have a tcp connection and you want a so-called idle	1590	usage example.
567	timeout, that is, you want to be called when there have been, say, 60	1591
568	seconds of inactivity on the socket. The easiest way to do this is to	1592	=item ev_tstamp repeat [read-write]
569	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each	1593
570	time you successfully read or write some data. If you go into an idle	1594	The current C<repeat> value. Will be used each time the watcher times out
571	state where you do not expect data to travel on the socket, you can stop	1595	or C<ev_timer_again> is called, and determines the next timeout (if any),
572	the timer, and again will automatically restart it if need be.	1596	which is also when any modifications are taken into account.
573		1597
574	=back	1598	=back
575		1599
		1600	=head3 Examples
		1601
		1602	Example: Create a timer that fires after 60 seconds.
		1603
		1604	static void
		1605	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1606	{
		1607	.. one minute over, w is actually stopped right here
		1608	}
		1609
		1610	ev_timer mytimer;
		1611	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		1612	ev_timer_start (loop, &mytimer);
		1613
		1614	Example: Create a timeout timer that times out after 10 seconds of
		1615	inactivity.
		1616
		1617	static void
		1618	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1619	{
		1620	.. ten seconds without any activity
		1621	}
		1622
		1623	ev_timer mytimer;
		1624	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		1625	ev_timer_again (&mytimer); /* start timer */
		1626	ev_loop (loop, 0);
		1627
		1628	// and in some piece of code that gets executed on any "activity":
		1629	// reset the timeout to start ticking again at 10 seconds
		1630	ev_timer_again (&mytimer);
		1631
		1632
576	=head2 C<ev_periodic> - to cron or not to cron	1633	=head2 C<ev_periodic> - to cron or not to cron?
577		1634
578	Periodic watchers are also timers of a kind, but they are very versatile	1635	Periodic watchers are also timers of a kind, but they are very versatile
579	(and unfortunately a bit complex).	1636	(and unfortunately a bit complex).
580		1637
581	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1638	Unlike C<ev_timer>, periodic watchers are not based on real time (or
582	but on wallclock time (absolute time). You can tell a periodic watcher	1639	relative time, the physical time that passes) but on wall clock time
583	to trigger "at" some specific point in time. For example, if you tell a	1640	(absolute time, the thing you can read on your calender or clock). The
584	periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()	1641	difference is that wall clock time can run faster or slower than real
585	+ 10.>) and then reset your system clock to the last year, then it will	1642	time, and time jumps are not uncommon (e.g. when you adjust your
586	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1643	wrist-watch).
587	roughly 10 seconds later and of course not if you reset your system time
588	again).
589		1644
590	They can also be used to implement vastly more complex timers, such as	1645	You can tell a periodic watcher to trigger after some specific point
		1646	in time: for example, if you tell a periodic watcher to trigger "in 10
		1647	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		1648	not a delay) and then reset your system clock to January of the previous
		1649	year, then it will take a year or more to trigger the event (unlike an
		1650	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		1651	it, as it uses a relative timeout).
		1652
		1653	C<ev_periodic> watchers can also be used to implement vastly more complex
591	triggering an event on eahc midnight, local time.	1654	timers, such as triggering an event on each "midnight, local time", or
		1655	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		1656	those cannot react to time jumps.
592		1657
593	As with timers, the callback is guarenteed to be invoked only when the	1658	As with timers, the callback is guaranteed to be invoked only when the
594	time (C<at>) has been passed, but if multiple periodic timers become ready	1659	point in time where it is supposed to trigger has passed. If multiple
595	during the same loop iteration then order of execution is undefined.	1660	timers become ready during the same loop iteration then the ones with
		1661	earlier time-out values are invoked before ones with later time-out values
		1662	(but this is no longer true when a callback calls C<ev_loop> recursively).
		1663
		1664	=head3 Watcher-Specific Functions and Data Members
596		1665
597	=over 4	1666	=over 4
598		1667
599	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1668	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
600		1669
601	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1670	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
602		1671
603	Lots of arguments, lets sort it out... There are basically three modes of	1672	Lots of arguments, let's sort it out... There are basically three modes of
604	operation, and we will explain them from simplest to complex:	1673	operation, and we will explain them from simplest to most complex:
605		1674
606	=over 4	1675	=over 4
607		1676
608	=item * absolute timer (interval = reschedule_cb = 0)	1677	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
609		1678
610	In this configuration the watcher triggers an event at the wallclock time	1679	In this configuration the watcher triggers an event after the wall clock
611	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1680	time C<offset> has passed. It will not repeat and will not adjust when a
612	that is, if it is to be run at January 1st 2011 then it will run when the	1681	time jump occurs, that is, if it is to be run at January 1st 2011 then it
613	system time reaches or surpasses this time.	1682	will be stopped and invoked when the system clock reaches or surpasses
		1683	this point in time.
614		1684
615	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1685	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
616		1686
617	In this mode the watcher will always be scheduled to time out at the next	1687	In this mode the watcher will always be scheduled to time out at the next
618	C<at + N * interval> time (for some integer N) and then repeat, regardless	1688	C<offset + N * interval> time (for some integer N, which can also be
619	of any time jumps.	1689	negative) and then repeat, regardless of any time jumps. The C<offset>
		1690	argument is merely an offset into the C<interval> periods.
620		1691
621	This can be used to create timers that do not drift with respect to system	1692	This can be used to create timers that do not drift with respect to the
622	time:	1693	system clock, for example, here is an C<ev_periodic> that triggers each
		1694	hour, on the hour (with respect to UTC):
623		1695
624	ev_periodic_set (&periodic, 0., 3600., 0);	1696	ev_periodic_set (&periodic, 0., 3600., 0);
625		1697
626	This doesn't mean there will always be 3600 seconds in between triggers,	1698	This doesn't mean there will always be 3600 seconds in between triggers,
627	but only that the the callback will be called when the system time shows a	1699	but only that the callback will be called when the system time shows a
628	full hour (UTC), or more correctly, when the system time is evenly divisible	1700	full hour (UTC), or more correctly, when the system time is evenly divisible
629	by 3600.	1701	by 3600.
630		1702
631	Another way to think about it (for the mathematically inclined) is that	1703	Another way to think about it (for the mathematically inclined) is that
632	C<ev_periodic> will try to run the callback in this mode at the next possible	1704	C<ev_periodic> will try to run the callback in this mode at the next possible
633	time where C<time = at (mod interval)>, regardless of any time jumps.	1705	time where C<time = offset (mod interval)>, regardless of any time jumps.
634		1706
		1707	For numerical stability it is preferable that the C<offset> value is near
		1708	C<ev_now ()> (the current time), but there is no range requirement for
		1709	this value, and in fact is often specified as zero.
		1710
		1711	Note also that there is an upper limit to how often a timer can fire (CPU
		1712	speed for example), so if C<interval> is very small then timing stability
		1713	will of course deteriorate. Libev itself tries to be exact to be about one
		1714	millisecond (if the OS supports it and the machine is fast enough).
		1715
635	=item * manual reschedule mode (reschedule_cb = callback)	1716	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
636		1717
637	In this mode the values for C<interval> and C<at> are both being	1718	In this mode the values for C<interval> and C<offset> are both being
638	ignored. Instead, each time the periodic watcher gets scheduled, the	1719	ignored. Instead, each time the periodic watcher gets scheduled, the
639	reschedule callback will be called with the watcher as first, and the	1720	reschedule callback will be called with the watcher as first, and the
640	current time as second argument.	1721	current time as second argument.
641		1722
642	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1723	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, ever,
643	ever, or make any event loop modifications>. If you need to stop it,	1724	or make ANY other event loop modifications whatsoever, unless explicitly
644	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	1725	allowed by documentation here>.
645	starting a prepare watcher).
646		1726
		1727	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		1728	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		1729	only event loop modification you are allowed to do).
		1730
647	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	1731	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
648	ev_tstamp now)>, e.g.:	1732	*w, ev_tstamp now)>, e.g.:
649		1733
		1734	static ev_tstamp
650	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1735	my_rescheduler (ev_periodic *w, ev_tstamp now)
651	{	1736	{
652	return now + 60.;	1737	return now + 60.;
653	}	1738	}
654		1739
655	It must return the next time to trigger, based on the passed time value	1740	It must return the next time to trigger, based on the passed time value
656	(that is, the lowest time value larger than to the second argument). It	1741	(that is, the lowest time value larger than to the second argument). It
657	will usually be called just before the callback will be triggered, but	1742	will usually be called just before the callback will be triggered, but
658	might be called at other times, too.	1743	might be called at other times, too.
659		1744
660	NOTE: I<< This callback must always return a time that is later than the	1745	NOTE: I<< This callback must always return a time that is higher than or
661	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	1746	equal to the passed C<now> value >>.
662		1747
663	This can be used to create very complex timers, such as a timer that	1748	This can be used to create very complex timers, such as a timer that
664	triggers on each midnight, local time. To do this, you would calculate the	1749	triggers on "next midnight, local time". To do this, you would calculate the
665	next midnight after C<now> and return the timestamp value for this. How	1750	next midnight after C<now> and return the timestamp value for this. How
666	you do this is, again, up to you (but it is not trivial, which is the main	1751	you do this is, again, up to you (but it is not trivial, which is the main
667	reason I omitted it as an example).	1752	reason I omitted it as an example).
668		1753
669	=back	1754	=back
…		…
673	Simply stops and restarts the periodic watcher again. This is only useful	1758	Simply stops and restarts the periodic watcher again. This is only useful
674	when you changed some parameters or the reschedule callback would return	1759	when you changed some parameters or the reschedule callback would return
675	a different time than the last time it was called (e.g. in a crond like	1760	a different time than the last time it was called (e.g. in a crond like
676	program when the crontabs have changed).	1761	program when the crontabs have changed).
677		1762
		1763	=item ev_tstamp ev_periodic_at (ev_periodic *)
		1764
		1765	When active, returns the absolute time that the watcher is supposed
		1766	to trigger next. This is not the same as the C<offset> argument to
		1767	C<ev_periodic_set>, but indeed works even in interval and manual
		1768	rescheduling modes.
		1769
		1770	=item ev_tstamp offset [read-write]
		1771
		1772	When repeating, this contains the offset value, otherwise this is the
		1773	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		1774	although libev might modify this value for better numerical stability).
		1775
		1776	Can be modified any time, but changes only take effect when the periodic
		1777	timer fires or C<ev_periodic_again> is being called.
		1778
		1779	=item ev_tstamp interval [read-write]
		1780
		1781	The current interval value. Can be modified any time, but changes only
		1782	take effect when the periodic timer fires or C<ev_periodic_again> is being
		1783	called.
		1784
		1785	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		1786
		1787	The current reschedule callback, or C<0>, if this functionality is
		1788	switched off. Can be changed any time, but changes only take effect when
		1789	the periodic timer fires or C<ev_periodic_again> is being called.
		1790
678	=back	1791	=back
679		1792
		1793	=head3 Examples
		1794
		1795	Example: Call a callback every hour, or, more precisely, whenever the
		1796	system time is divisible by 3600. The callback invocation times have
		1797	potentially a lot of jitter, but good long-term stability.
		1798
		1799	static void
		1800	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
		1801	{
		1802	... its now a full hour (UTC, or TAI or whatever your clock follows)
		1803	}
		1804
		1805	ev_periodic hourly_tick;
		1806	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		1807	ev_periodic_start (loop, &hourly_tick);
		1808
		1809	Example: The same as above, but use a reschedule callback to do it:
		1810
		1811	#include <math.h>
		1812
		1813	static ev_tstamp
		1814	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		1815	{
		1816	return now + (3600. - fmod (now, 3600.));
		1817	}
		1818
		1819	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		1820
		1821	Example: Call a callback every hour, starting now:
		1822
		1823	ev_periodic hourly_tick;
		1824	ev_periodic_init (&hourly_tick, clock_cb,
		1825	fmod (ev_now (loop), 3600.), 3600., 0);
		1826	ev_periodic_start (loop, &hourly_tick);
		1827
		1828
680	=head2 C<ev_signal> - signal me when a signal gets signalled	1829	=head2 C<ev_signal> - signal me when a signal gets signalled!
681		1830
682	Signal watchers will trigger an event when the process receives a specific	1831	Signal watchers will trigger an event when the process receives a specific
683	signal one or more times. Even though signals are very asynchronous, libev	1832	signal one or more times. Even though signals are very asynchronous, libev
684	will try it's best to deliver signals synchronously, i.e. as part of the	1833	will try it's best to deliver signals synchronously, i.e. as part of the
685	normal event processing, like any other event.	1834	normal event processing, like any other event.
686		1835
		1836	If you want signals asynchronously, just use C<sigaction> as you would
		1837	do without libev and forget about sharing the signal. You can even use
		1838	C<ev_async> from a signal handler to synchronously wake up an event loop.
		1839
687	You can configure as many watchers as you like per signal. Only when the	1840	You can configure as many watchers as you like per signal. Only when the
688	first watcher gets started will libev actually register a signal watcher	1841	first watcher gets started will libev actually register a signal handler
689	with the kernel (thus it coexists with your own signal handlers as long	1842	with the kernel (thus it coexists with your own signal handlers as long as
690	as you don't register any with libev). Similarly, when the last signal	1843	you don't register any with libev for the same signal). Similarly, when
691	watcher for a signal is stopped libev will reset the signal handler to	1844	the last signal watcher for a signal is stopped, libev will reset the
692	SIG_DFL (regardless of what it was set to before).	1845	signal handler to SIG_DFL (regardless of what it was set to before).
		1846
		1847	If possible and supported, libev will install its handlers with
		1848	C<SA_RESTART> behaviour enabled, so system calls should not be unduly
		1849	interrupted. If you have a problem with system calls getting interrupted by
		1850	signals you can block all signals in an C<ev_check> watcher and unblock
		1851	them in an C<ev_prepare> watcher.
		1852
		1853	=head3 Watcher-Specific Functions and Data Members
693		1854
694	=over 4	1855	=over 4
695		1856
696	=item ev_signal_init (ev_signal *, callback, int signum)	1857	=item ev_signal_init (ev_signal *, callback, int signum)
697		1858
698	=item ev_signal_set (ev_signal *, int signum)	1859	=item ev_signal_set (ev_signal *, int signum)
699		1860
700	Configures the watcher to trigger on the given signal number (usually one	1861	Configures the watcher to trigger on the given signal number (usually one
701	of the C<SIGxxx> constants).	1862	of the C<SIGxxx> constants).
702		1863
		1864	=item int signum [read-only]
		1865
		1866	The signal the watcher watches out for.
		1867
703	=back	1868	=back
704		1869
		1870	=head3 Examples
		1871
		1872	Example: Try to exit cleanly on SIGINT.
		1873
		1874	static void
		1875	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		1876	{
		1877	ev_unloop (loop, EVUNLOOP_ALL);
		1878	}
		1879
		1880	ev_signal signal_watcher;
		1881	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		1882	ev_signal_start (loop, &signal_watcher);
		1883
		1884
705	=head2 C<ev_child> - wait for pid status changes	1885	=head2 C<ev_child> - watch out for process status changes
706		1886
707	Child watchers trigger when your process receives a SIGCHLD in response to	1887	Child watchers trigger when your process receives a SIGCHLD in response to
708	some child status changes (most typically when a child of yours dies).	1888	some child status changes (most typically when a child of yours dies or
		1889	exits). It is permissible to install a child watcher I<after> the child
		1890	has been forked (which implies it might have already exited), as long
		1891	as the event loop isn't entered (or is continued from a watcher), i.e.,
		1892	forking and then immediately registering a watcher for the child is fine,
		1893	but forking and registering a watcher a few event loop iterations later is
		1894	not.
		1895
		1896	Only the default event loop is capable of handling signals, and therefore
		1897	you can only register child watchers in the default event loop.
		1898
		1899	=head3 Process Interaction
		1900
		1901	Libev grabs C<SIGCHLD> as soon as the default event loop is
		1902	initialised. This is necessary to guarantee proper behaviour even if
		1903	the first child watcher is started after the child exits. The occurrence
		1904	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		1905	synchronously as part of the event loop processing. Libev always reaps all
		1906	children, even ones not watched.
		1907
		1908	=head3 Overriding the Built-In Processing
		1909
		1910	Libev offers no special support for overriding the built-in child
		1911	processing, but if your application collides with libev's default child
		1912	handler, you can override it easily by installing your own handler for
		1913	C<SIGCHLD> after initialising the default loop, and making sure the
		1914	default loop never gets destroyed. You are encouraged, however, to use an
		1915	event-based approach to child reaping and thus use libev's support for
		1916	that, so other libev users can use C<ev_child> watchers freely.
		1917
		1918	=head3 Stopping the Child Watcher
		1919
		1920	Currently, the child watcher never gets stopped, even when the
		1921	child terminates, so normally one needs to stop the watcher in the
		1922	callback. Future versions of libev might stop the watcher automatically
		1923	when a child exit is detected.
		1924
		1925	=head3 Watcher-Specific Functions and Data Members
709		1926
710	=over 4	1927	=over 4
711		1928
712	=item ev_child_init (ev_child *, callback, int pid)	1929	=item ev_child_init (ev_child *, callback, int pid, int trace)
713		1930
714	=item ev_child_set (ev_child *, int pid)	1931	=item ev_child_set (ev_child *, int pid, int trace)
715		1932
716	Configures the watcher to wait for status changes of process C<pid> (or	1933	Configures the watcher to wait for status changes of process C<pid> (or
717	I<any> process if C<pid> is specified as C<0>). The callback can look	1934	I<any> process if C<pid> is specified as C<0>). The callback can look
718	at the C<rstatus> member of the C<ev_child> watcher structure to see	1935	at the C<rstatus> member of the C<ev_child> watcher structure to see
719	the status word (use the macros from C<sys/wait.h> and see your systems	1936	the status word (use the macros from C<sys/wait.h> and see your systems
720	C<waitpid> documentation). The C<rpid> member contains the pid of the	1937	C<waitpid> documentation). The C<rpid> member contains the pid of the
721	process causing the status change.	1938	process causing the status change. C<trace> must be either C<0> (only
		1939	activate the watcher when the process terminates) or C<1> (additionally
		1940	activate the watcher when the process is stopped or continued).
		1941
		1942	=item int pid [read-only]
		1943
		1944	The process id this watcher watches out for, or C<0>, meaning any process id.
		1945
		1946	=item int rpid [read-write]
		1947
		1948	The process id that detected a status change.
		1949
		1950	=item int rstatus [read-write]
		1951
		1952	The process exit/trace status caused by C<rpid> (see your systems
		1953	C<waitpid> and C<sys/wait.h> documentation for details).
722		1954
723	=back	1955	=back
724		1956
		1957	=head3 Examples
		1958
		1959	Example: C<fork()> a new process and install a child handler to wait for
		1960	its completion.
		1961
		1962	ev_child cw;
		1963
		1964	static void
		1965	child_cb (EV_P_ ev_child *w, int revents)
		1966	{
		1967	ev_child_stop (EV_A_ w);
		1968	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		1969	}
		1970
		1971	pid_t pid = fork ();
		1972
		1973	if (pid < 0)
		1974	// error
		1975	else if (pid == 0)
		1976	{
		1977	// the forked child executes here
		1978	exit (1);
		1979	}
		1980	else
		1981	{
		1982	ev_child_init (&cw, child_cb, pid, 0);
		1983	ev_child_start (EV_DEFAULT_ &cw);
		1984	}
		1985
		1986
		1987	=head2 C<ev_stat> - did the file attributes just change?
		1988
		1989	This watches a file system path for attribute changes. That is, it calls
		1990	C<stat> on that path in regular intervals (or when the OS says it changed)
		1991	and sees if it changed compared to the last time, invoking the callback if
		1992	it did.
		1993
		1994	The path does not need to exist: changing from "path exists" to "path does
		1995	not exist" is a status change like any other. The condition "path does not
		1996	exist" (or more correctly "path cannot be stat'ed") is signified by the
		1997	C<st_nlink> field being zero (which is otherwise always forced to be at
		1998	least one) and all the other fields of the stat buffer having unspecified
		1999	contents.
		2000
		2001	The path I<must not> end in a slash or contain special components such as
		2002	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2003	your working directory changes, then the behaviour is undefined.
		2004
		2005	Since there is no portable change notification interface available, the
		2006	portable implementation simply calls C<stat(2)> regularly on the path
		2007	to see if it changed somehow. You can specify a recommended polling
		2008	interval for this case. If you specify a polling interval of C<0> (highly
		2009	recommended!) then a I<suitable, unspecified default> value will be used
		2010	(which you can expect to be around five seconds, although this might
		2011	change dynamically). Libev will also impose a minimum interval which is
		2012	currently around C<0.1>, but that's usually overkill.
		2013
		2014	This watcher type is not meant for massive numbers of stat watchers,
		2015	as even with OS-supported change notifications, this can be
		2016	resource-intensive.
		2017
		2018	At the time of this writing, the only OS-specific interface implemented
		2019	is the Linux inotify interface (implementing kqueue support is left as an
		2020	exercise for the reader. Note, however, that the author sees no way of
		2021	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2022
		2023	=head3 ABI Issues (Largefile Support)
		2024
		2025	Libev by default (unless the user overrides this) uses the default
		2026	compilation environment, which means that on systems with large file
		2027	support disabled by default, you get the 32 bit version of the stat
		2028	structure. When using the library from programs that change the ABI to
		2029	use 64 bit file offsets the programs will fail. In that case you have to
		2030	compile libev with the same flags to get binary compatibility. This is
		2031	obviously the case with any flags that change the ABI, but the problem is
		2032	most noticeably displayed with ev_stat and large file support.
		2033
		2034	The solution for this is to lobby your distribution maker to make large
		2035	file interfaces available by default (as e.g. FreeBSD does) and not
		2036	optional. Libev cannot simply switch on large file support because it has
		2037	to exchange stat structures with application programs compiled using the
		2038	default compilation environment.
		2039
		2040	=head3 Inotify and Kqueue
		2041
		2042	When C<inotify (7)> support has been compiled into libev and present at
		2043	runtime, it will be used to speed up change detection where possible. The
		2044	inotify descriptor will be created lazily when the first C<ev_stat>
		2045	watcher is being started.
		2046
		2047	Inotify presence does not change the semantics of C<ev_stat> watchers
		2048	except that changes might be detected earlier, and in some cases, to avoid
		2049	making regular C<stat> calls. Even in the presence of inotify support
		2050	there are many cases where libev has to resort to regular C<stat> polling,
		2051	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2052	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2053	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2054	xfs are fully working) libev usually gets away without polling.
		2055
		2056	There is no support for kqueue, as apparently it cannot be used to
		2057	implement this functionality, due to the requirement of having a file
		2058	descriptor open on the object at all times, and detecting renames, unlinks
		2059	etc. is difficult.
		2060
		2061	=head3 C<stat ()> is a synchronous operation
		2062
		2063	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2064	the process. The exception are C<ev_stat> watchers - those call C<stat
		2065	()>, which is a synchronous operation.
		2066
		2067	For local paths, this usually doesn't matter: unless the system is very
		2068	busy or the intervals between stat's are large, a stat call will be fast,
		2069	as the path data is usually in memory already (except when starting the
		2070	watcher).
		2071
		2072	For networked file systems, calling C<stat ()> can block an indefinite
		2073	time due to network issues, and even under good conditions, a stat call
		2074	often takes multiple milliseconds.
		2075
		2076	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2077	paths, although this is fully supported by libev.
		2078
		2079	=head3 The special problem of stat time resolution
		2080
		2081	The C<stat ()> system call only supports full-second resolution portably,
		2082	and even on systems where the resolution is higher, most file systems
		2083	still only support whole seconds.
		2084
		2085	That means that, if the time is the only thing that changes, you can
		2086	easily miss updates: on the first update, C<ev_stat> detects a change and
		2087	calls your callback, which does something. When there is another update
		2088	within the same second, C<ev_stat> will be unable to detect unless the
		2089	stat data does change in other ways (e.g. file size).
		2090
		2091	The solution to this is to delay acting on a change for slightly more
		2092	than a second (or till slightly after the next full second boundary), using
		2093	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2094	ev_timer_again (loop, w)>).
		2095
		2096	The C<.02> offset is added to work around small timing inconsistencies
		2097	of some operating systems (where the second counter of the current time
		2098	might be be delayed. One such system is the Linux kernel, where a call to
		2099	C<gettimeofday> might return a timestamp with a full second later than
		2100	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2101	update file times then there will be a small window where the kernel uses
		2102	the previous second to update file times but libev might already execute
		2103	the timer callback).
		2104
		2105	=head3 Watcher-Specific Functions and Data Members
		2106
		2107	=over 4
		2108
		2109	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2110
		2111	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2112
		2113	Configures the watcher to wait for status changes of the given
		2114	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2115	be detected and should normally be specified as C<0> to let libev choose
		2116	a suitable value. The memory pointed to by C<path> must point to the same
		2117	path for as long as the watcher is active.
		2118
		2119	The callback will receive an C<EV_STAT> event when a change was detected,
		2120	relative to the attributes at the time the watcher was started (or the
		2121	last change was detected).
		2122
		2123	=item ev_stat_stat (loop, ev_stat *)
		2124
		2125	Updates the stat buffer immediately with new values. If you change the
		2126	watched path in your callback, you could call this function to avoid
		2127	detecting this change (while introducing a race condition if you are not
		2128	the only one changing the path). Can also be useful simply to find out the
		2129	new values.
		2130
		2131	=item ev_statdata attr [read-only]
		2132
		2133	The most-recently detected attributes of the file. Although the type is
		2134	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2135	suitable for your system, but you can only rely on the POSIX-standardised
		2136	members to be present. If the C<st_nlink> member is C<0>, then there was
		2137	some error while C<stat>ing the file.
		2138
		2139	=item ev_statdata prev [read-only]
		2140
		2141	The previous attributes of the file. The callback gets invoked whenever
		2142	C<prev> != C<attr>, or, more precisely, one or more of these members
		2143	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2144	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2145
		2146	=item ev_tstamp interval [read-only]
		2147
		2148	The specified interval.
		2149
		2150	=item const char *path [read-only]
		2151
		2152	The file system path that is being watched.
		2153
		2154	=back
		2155
		2156	=head3 Examples
		2157
		2158	Example: Watch C</etc/passwd> for attribute changes.
		2159
		2160	static void
		2161	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2162	{
		2163	/* /etc/passwd changed in some way */
		2164	if (w->attr.st_nlink)
		2165	{
		2166	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2167	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2168	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2169	}
		2170	else
		2171	/* you shalt not abuse printf for puts */
		2172	puts ("wow, /etc/passwd is not there, expect problems. "
		2173	"if this is windows, they already arrived\n");
		2174	}
		2175
		2176	...
		2177	ev_stat passwd;
		2178
		2179	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2180	ev_stat_start (loop, &passwd);
		2181
		2182	Example: Like above, but additionally use a one-second delay so we do not
		2183	miss updates (however, frequent updates will delay processing, too, so
		2184	one might do the work both on C<ev_stat> callback invocation I<and> on
		2185	C<ev_timer> callback invocation).
		2186
		2187	static ev_stat passwd;
		2188	static ev_timer timer;
		2189
		2190	static void
		2191	timer_cb (EV_P_ ev_timer *w, int revents)
		2192	{
		2193	ev_timer_stop (EV_A_ w);
		2194
		2195	/* now it's one second after the most recent passwd change */
		2196	}
		2197
		2198	static void
		2199	stat_cb (EV_P_ ev_stat *w, int revents)
		2200	{
		2201	/* reset the one-second timer */
		2202	ev_timer_again (EV_A_ &timer);
		2203	}
		2204
		2205	...
		2206	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2207	ev_stat_start (loop, &passwd);
		2208	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2209
		2210
725	=head2 C<ev_idle> - when you've got nothing better to do	2211	=head2 C<ev_idle> - when you've got nothing better to do...
726		2212
727	Idle watchers trigger events when there are no other events are pending	2213	Idle watchers trigger events when no other events of the same or higher
728	(prepare, check and other idle watchers do not count). That is, as long	2214	priority are pending (prepare, check and other idle watchers do not count
729	as your process is busy handling sockets or timeouts (or even signals,	2215	as receiving "events").
730	imagine) it will not be triggered. But when your process is idle all idle	2216
731	watchers are being called again and again, once per event loop iteration -	2217	That is, as long as your process is busy handling sockets or timeouts
		2218	(or even signals, imagine) of the same or higher priority it will not be
		2219	triggered. But when your process is idle (or only lower-priority watchers
		2220	are pending), the idle watchers are being called once per event loop
732	until stopped, that is, or your process receives more events and becomes	2221	iteration - until stopped, that is, or your process receives more events
733	busy.	2222	and becomes busy again with higher priority stuff.
734		2223
735	The most noteworthy effect is that as long as any idle watchers are	2224	The most noteworthy effect is that as long as any idle watchers are
736	active, the process will not block when waiting for new events.	2225	active, the process will not block when waiting for new events.
737		2226
738	Apart from keeping your process non-blocking (which is a useful	2227	Apart from keeping your process non-blocking (which is a useful
739	effect on its own sometimes), idle watchers are a good place to do	2228	effect on its own sometimes), idle watchers are a good place to do
740	"pseudo-background processing", or delay processing stuff to after the	2229	"pseudo-background processing", or delay processing stuff to after the
741	event loop has handled all outstanding events.	2230	event loop has handled all outstanding events.
742		2231
		2232	=head3 Watcher-Specific Functions and Data Members
		2233
743	=over 4	2234	=over 4
744		2235
745	=item ev_idle_init (ev_signal *, callback)	2236	=item ev_idle_init (ev_idle *, callback)
746		2237
747	Initialises and configures the idle watcher - it has no parameters of any	2238	Initialises and configures the idle watcher - it has no parameters of any
748	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2239	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
749	believe me.	2240	believe me.
750		2241
751	=back	2242	=back
752		2243
		2244	=head3 Examples
		2245
		2246	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2247	callback, free it. Also, use no error checking, as usual.
		2248
		2249	static void
		2250	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2251	{
		2252	free (w);
		2253	// now do something you wanted to do when the program has
		2254	// no longer anything immediate to do.
		2255	}
		2256
		2257	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2258	ev_idle_init (idle_watcher, idle_cb);
		2259	ev_idle_start (loop, idle_cb);
		2260
		2261
753	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2262	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
754		2263
755	Prepare and check watchers are usually (but not always) used in tandem:	2264	Prepare and check watchers are usually (but not always) used in pairs:
756	prepare watchers get invoked before the process blocks and check watchers	2265	prepare watchers get invoked before the process blocks and check watchers
757	afterwards.	2266	afterwards.
758		2267
		2268	You I<must not> call C<ev_loop> or similar functions that enter
		2269	the current event loop from either C<ev_prepare> or C<ev_check>
		2270	watchers. Other loops than the current one are fine, however. The
		2271	rationale behind this is that you do not need to check for recursion in
		2272	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2273	C<ev_check> so if you have one watcher of each kind they will always be
		2274	called in pairs bracketing the blocking call.
		2275
759	Their main purpose is to integrate other event mechanisms into libev. This	2276	Their main purpose is to integrate other event mechanisms into libev and
760	could be used, for example, to track variable changes, implement your own	2277	their use is somewhat advanced. They could be used, for example, to track
761	watchers, integrate net-snmp or a coroutine library and lots more.	2278	variable changes, implement your own watchers, integrate net-snmp or a
		2279	coroutine library and lots more. They are also occasionally useful if
		2280	you cache some data and want to flush it before blocking (for example,
		2281	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2282	watcher).
762		2283
763	This is done by examining in each prepare call which file descriptors need	2284	This is done by examining in each prepare call which file descriptors
764	to be watched by the other library, registering C<ev_io> watchers for	2285	need to be watched by the other library, registering C<ev_io> watchers
765	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2286	for them and starting an C<ev_timer> watcher for any timeouts (many
766	provide just this functionality). Then, in the check watcher you check for	2287	libraries provide exactly this functionality). Then, in the check watcher,
767	any events that occured (by checking the pending status of all watchers	2288	you check for any events that occurred (by checking the pending status
768	and stopping them) and call back into the library. The I/O and timer	2289	of all watchers and stopping them) and call back into the library. The
769	callbacks will never actually be called (but must be valid nevertheless,	2290	I/O and timer callbacks will never actually be called (but must be valid
770	because you never know, you know?).	2291	nevertheless, because you never know, you know?).
771		2292
772	As another example, the Perl Coro module uses these hooks to integrate	2293	As another example, the Perl Coro module uses these hooks to integrate
773	coroutines into libev programs, by yielding to other active coroutines	2294	coroutines into libev programs, by yielding to other active coroutines
774	during each prepare and only letting the process block if no coroutines	2295	during each prepare and only letting the process block if no coroutines
775	are ready to run (it's actually more complicated: it only runs coroutines	2296	are ready to run (it's actually more complicated: it only runs coroutines
776	with priority higher than or equal to the event loop and one coroutine	2297	with priority higher than or equal to the event loop and one coroutine
777	of lower priority, but only once, using idle watchers to keep the event	2298	of lower priority, but only once, using idle watchers to keep the event
778	loop from blocking if lower-priority coroutines are active, thus mapping	2299	loop from blocking if lower-priority coroutines are active, thus mapping
779	low-priority coroutines to idle/background tasks).	2300	low-priority coroutines to idle/background tasks).
780		2301
		2302	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2303	priority, to ensure that they are being run before any other watchers
		2304	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2305
		2306	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2307	activate ("feed") events into libev. While libev fully supports this, they
		2308	might get executed before other C<ev_check> watchers did their job. As
		2309	C<ev_check> watchers are often used to embed other (non-libev) event
		2310	loops those other event loops might be in an unusable state until their
		2311	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2312	others).
		2313
		2314	=head3 Watcher-Specific Functions and Data Members
		2315
781	=over 4	2316	=over 4
782		2317
783	=item ev_prepare_init (ev_prepare *, callback)	2318	=item ev_prepare_init (ev_prepare *, callback)
784		2319
785	=item ev_check_init (ev_check *, callback)	2320	=item ev_check_init (ev_check *, callback)
786		2321
787	Initialises and configures the prepare or check watcher - they have no	2322	Initialises and configures the prepare or check watcher - they have no
788	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2323	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
789	macros, but using them is utterly, utterly and completely pointless.	2324	macros, but using them is utterly, utterly, utterly and completely
		2325	pointless.
790		2326
791	=back	2327	=back
792		2328
		2329	=head3 Examples
		2330
		2331	There are a number of principal ways to embed other event loops or modules
		2332	into libev. Here are some ideas on how to include libadns into libev
		2333	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2334	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2335	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2336	Glib event loop).
		2337
		2338	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2339	and in a check watcher, destroy them and call into libadns. What follows
		2340	is pseudo-code only of course. This requires you to either use a low
		2341	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2342	the callbacks for the IO/timeout watchers might not have been called yet.
		2343
		2344	static ev_io iow [nfd];
		2345	static ev_timer tw;
		2346
		2347	static void
		2348	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2349	{
		2350	}
		2351
		2352	// create io watchers for each fd and a timer before blocking
		2353	static void
		2354	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2355	{
		2356	int timeout = 3600000;
		2357	struct pollfd fds [nfd];
		2358	// actual code will need to loop here and realloc etc.
		2359	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2360
		2361	/* the callback is illegal, but won't be called as we stop during check */
		2362	ev_timer_init (&tw, 0, timeout * 1e-3);
		2363	ev_timer_start (loop, &tw);
		2364
		2365	// create one ev_io per pollfd
		2366	for (int i = 0; i < nfd; ++i)
		2367	{
		2368	ev_io_init (iow + i, io_cb, fds [i].fd,
		2369	((fds [i].events & POLLIN ? EV_READ : 0)
		2370	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2371
		2372	fds [i].revents = 0;
		2373	ev_io_start (loop, iow + i);
		2374	}
		2375	}
		2376
		2377	// stop all watchers after blocking
		2378	static void
		2379	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2380	{
		2381	ev_timer_stop (loop, &tw);
		2382
		2383	for (int i = 0; i < nfd; ++i)
		2384	{
		2385	// set the relevant poll flags
		2386	// could also call adns_processreadable etc. here
		2387	struct pollfd *fd = fds + i;
		2388	int revents = ev_clear_pending (iow + i);
		2389	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2390	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2391
		2392	// now stop the watcher
		2393	ev_io_stop (loop, iow + i);
		2394	}
		2395
		2396	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2397	}
		2398
		2399	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2400	in the prepare watcher and would dispose of the check watcher.
		2401
		2402	Method 3: If the module to be embedded supports explicit event
		2403	notification (libadns does), you can also make use of the actual watcher
		2404	callbacks, and only destroy/create the watchers in the prepare watcher.
		2405
		2406	static void
		2407	timer_cb (EV_P_ ev_timer *w, int revents)
		2408	{
		2409	adns_state ads = (adns_state)w->data;
		2410	update_now (EV_A);
		2411
		2412	adns_processtimeouts (ads, &tv_now);
		2413	}
		2414
		2415	static void
		2416	io_cb (EV_P_ ev_io *w, int revents)
		2417	{
		2418	adns_state ads = (adns_state)w->data;
		2419	update_now (EV_A);
		2420
		2421	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2422	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2423	}
		2424
		2425	// do not ever call adns_afterpoll
		2426
		2427	Method 4: Do not use a prepare or check watcher because the module you
		2428	want to embed is not flexible enough to support it. Instead, you can
		2429	override their poll function. The drawback with this solution is that the
		2430	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2431	this approach, effectively embedding EV as a client into the horrible
		2432	libglib event loop.
		2433
		2434	static gint
		2435	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2436	{
		2437	int got_events = 0;
		2438
		2439	for (n = 0; n < nfds; ++n)
		2440	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2441
		2442	if (timeout >= 0)
		2443	// create/start timer
		2444
		2445	// poll
		2446	ev_loop (EV_A_ 0);
		2447
		2448	// stop timer again
		2449	if (timeout >= 0)
		2450	ev_timer_stop (EV_A_ &to);
		2451
		2452	// stop io watchers again - their callbacks should have set
		2453	for (n = 0; n < nfds; ++n)
		2454	ev_io_stop (EV_A_ iow [n]);
		2455
		2456	return got_events;
		2457	}
		2458
		2459
		2460	=head2 C<ev_embed> - when one backend isn't enough...
		2461
		2462	This is a rather advanced watcher type that lets you embed one event loop
		2463	into another (currently only C<ev_io> events are supported in the embedded
		2464	loop, other types of watchers might be handled in a delayed or incorrect
		2465	fashion and must not be used).
		2466
		2467	There are primarily two reasons you would want that: work around bugs and
		2468	prioritise I/O.
		2469
		2470	As an example for a bug workaround, the kqueue backend might only support
		2471	sockets on some platform, so it is unusable as generic backend, but you
		2472	still want to make use of it because you have many sockets and it scales
		2473	so nicely. In this case, you would create a kqueue-based loop and embed
		2474	it into your default loop (which might use e.g. poll). Overall operation
		2475	will be a bit slower because first libev has to call C<poll> and then
		2476	C<kevent>, but at least you can use both mechanisms for what they are
		2477	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2478
		2479	As for prioritising I/O: under rare circumstances you have the case where
		2480	some fds have to be watched and handled very quickly (with low latency),
		2481	and even priorities and idle watchers might have too much overhead. In
		2482	this case you would put all the high priority stuff in one loop and all
		2483	the rest in a second one, and embed the second one in the first.
		2484
		2485	As long as the watcher is active, the callback will be invoked every
		2486	time there might be events pending in the embedded loop. The callback
		2487	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		2488	sweep and invoke their callbacks (the callback doesn't need to invoke the
		2489	C<ev_embed_sweep> function directly, it could also start an idle watcher
		2490	to give the embedded loop strictly lower priority for example).
		2491
		2492	You can also set the callback to C<0>, in which case the embed watcher
		2493	will automatically execute the embedded loop sweep whenever necessary.
		2494
		2495	Fork detection will be handled transparently while the C<ev_embed> watcher
		2496	is active, i.e., the embedded loop will automatically be forked when the
		2497	embedding loop forks. In other cases, the user is responsible for calling
		2498	C<ev_loop_fork> on the embedded loop.
		2499
		2500	Unfortunately, not all backends are embeddable: only the ones returned by
		2501	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2502	portable one.
		2503
		2504	So when you want to use this feature you will always have to be prepared
		2505	that you cannot get an embeddable loop. The recommended way to get around
		2506	this is to have a separate variables for your embeddable loop, try to
		2507	create it, and if that fails, use the normal loop for everything.
		2508
		2509	=head3 C<ev_embed> and fork
		2510
		2511	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2512	automatically be applied to the embedded loop as well, so no special
		2513	fork handling is required in that case. When the watcher is not running,
		2514	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2515	as applicable.
		2516
		2517	=head3 Watcher-Specific Functions and Data Members
		2518
		2519	=over 4
		2520
		2521	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2522
		2523	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2524
		2525	Configures the watcher to embed the given loop, which must be
		2526	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2527	invoked automatically, otherwise it is the responsibility of the callback
		2528	to invoke it (it will continue to be called until the sweep has been done,
		2529	if you do not want that, you need to temporarily stop the embed watcher).
		2530
		2531	=item ev_embed_sweep (loop, ev_embed *)
		2532
		2533	Make a single, non-blocking sweep over the embedded loop. This works
		2534	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		2535	appropriate way for embedded loops.
		2536
		2537	=item struct ev_loop *other [read-only]
		2538
		2539	The embedded event loop.
		2540
		2541	=back
		2542
		2543	=head3 Examples
		2544
		2545	Example: Try to get an embeddable event loop and embed it into the default
		2546	event loop. If that is not possible, use the default loop. The default
		2547	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2548	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2549	used).
		2550
		2551	struct ev_loop *loop_hi = ev_default_init (0);
		2552	struct ev_loop *loop_lo = 0;
		2553	ev_embed embed;
		2554
		2555	// see if there is a chance of getting one that works
		2556	// (remember that a flags value of 0 means autodetection)
		2557	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2558	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2559	: 0;
		2560
		2561	// if we got one, then embed it, otherwise default to loop_hi
		2562	if (loop_lo)
		2563	{
		2564	ev_embed_init (&embed, 0, loop_lo);
		2565	ev_embed_start (loop_hi, &embed);
		2566	}
		2567	else
		2568	loop_lo = loop_hi;
		2569
		2570	Example: Check if kqueue is available but not recommended and create
		2571	a kqueue backend for use with sockets (which usually work with any
		2572	kqueue implementation). Store the kqueue/socket-only event loop in
		2573	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2574
		2575	struct ev_loop *loop = ev_default_init (0);
		2576	struct ev_loop *loop_socket = 0;
		2577	ev_embed embed;
		2578
		2579	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2580	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2581	{
		2582	ev_embed_init (&embed, 0, loop_socket);
		2583	ev_embed_start (loop, &embed);
		2584	}
		2585
		2586	if (!loop_socket)
		2587	loop_socket = loop;
		2588
		2589	// now use loop_socket for all sockets, and loop for everything else
		2590
		2591
		2592	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		2593
		2594	Fork watchers are called when a C<fork ()> was detected (usually because
		2595	whoever is a good citizen cared to tell libev about it by calling
		2596	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		2597	event loop blocks next and before C<ev_check> watchers are being called,
		2598	and only in the child after the fork. If whoever good citizen calling
		2599	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		2600	handlers will be invoked, too, of course.
		2601
		2602	=head3 Watcher-Specific Functions and Data Members
		2603
		2604	=over 4
		2605
		2606	=item ev_fork_init (ev_signal *, callback)
		2607
		2608	Initialises and configures the fork watcher - it has no parameters of any
		2609	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		2610	believe me.
		2611
		2612	=back
		2613
		2614
		2615	=head2 C<ev_async> - how to wake up another event loop
		2616
		2617	In general, you cannot use an C<ev_loop> from multiple threads or other
		2618	asynchronous sources such as signal handlers (as opposed to multiple event
		2619	loops - those are of course safe to use in different threads).
		2620
		2621	Sometimes, however, you need to wake up another event loop you do not
		2622	control, for example because it belongs to another thread. This is what
		2623	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2624	can signal it by calling C<ev_async_send>, which is thread- and signal
		2625	safe.
		2626
		2627	This functionality is very similar to C<ev_signal> watchers, as signals,
		2628	too, are asynchronous in nature, and signals, too, will be compressed
		2629	(i.e. the number of callback invocations may be less than the number of
		2630	C<ev_async_sent> calls).
		2631
		2632	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2633	just the default loop.
		2634
		2635	=head3 Queueing
		2636
		2637	C<ev_async> does not support queueing of data in any way. The reason
		2638	is that the author does not know of a simple (or any) algorithm for a
		2639	multiple-writer-single-reader queue that works in all cases and doesn't
		2640	need elaborate support such as pthreads.
		2641
		2642	That means that if you want to queue data, you have to provide your own
		2643	queue. But at least I can tell you how to implement locking around your
		2644	queue:
		2645
		2646	=over 4
		2647
		2648	=item queueing from a signal handler context
		2649
		2650	To implement race-free queueing, you simply add to the queue in the signal
		2651	handler but you block the signal handler in the watcher callback. Here is
		2652	an example that does that for some fictitious SIGUSR1 handler:
		2653
		2654	static ev_async mysig;
		2655
		2656	static void
		2657	sigusr1_handler (void)
		2658	{
		2659	sometype data;
		2660
		2661	// no locking etc.
		2662	queue_put (data);
		2663	ev_async_send (EV_DEFAULT_ &mysig);
		2664	}
		2665
		2666	static void
		2667	mysig_cb (EV_P_ ev_async *w, int revents)
		2668	{
		2669	sometype data;
		2670	sigset_t block, prev;
		2671
		2672	sigemptyset (&block);
		2673	sigaddset (&block, SIGUSR1);
		2674	sigprocmask (SIG_BLOCK, &block, &prev);
		2675
		2676	while (queue_get (&data))
		2677	process (data);
		2678
		2679	if (sigismember (&prev, SIGUSR1)
		2680	sigprocmask (SIG_UNBLOCK, &block, 0);
		2681	}
		2682
		2683	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2684	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2685	either...).
		2686
		2687	=item queueing from a thread context
		2688
		2689	The strategy for threads is different, as you cannot (easily) block
		2690	threads but you can easily preempt them, so to queue safely you need to
		2691	employ a traditional mutex lock, such as in this pthread example:
		2692
		2693	static ev_async mysig;
		2694	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2695
		2696	static void
		2697	otherthread (void)
		2698	{
		2699	// only need to lock the actual queueing operation
		2700	pthread_mutex_lock (&mymutex);
		2701	queue_put (data);
		2702	pthread_mutex_unlock (&mymutex);
		2703
		2704	ev_async_send (EV_DEFAULT_ &mysig);
		2705	}
		2706
		2707	static void
		2708	mysig_cb (EV_P_ ev_async *w, int revents)
		2709	{
		2710	pthread_mutex_lock (&mymutex);
		2711
		2712	while (queue_get (&data))
		2713	process (data);
		2714
		2715	pthread_mutex_unlock (&mymutex);
		2716	}
		2717
		2718	=back
		2719
		2720
		2721	=head3 Watcher-Specific Functions and Data Members
		2722
		2723	=over 4
		2724
		2725	=item ev_async_init (ev_async *, callback)
		2726
		2727	Initialises and configures the async watcher - it has no parameters of any
		2728	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		2729	trust me.
		2730
		2731	=item ev_async_send (loop, ev_async *)
		2732
		2733	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2734	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2735	C<ev_feed_event>, this call is safe to do from other threads, signal or
		2736	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		2737	section below on what exactly this means).
		2738
		2739	Note that, as with other watchers in libev, multiple events might get
		2740	compressed into a single callback invocation (another way to look at this
		2741	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		2742	reset when the event loop detects that).
		2743
		2744	This call incurs the overhead of a system call only once per event loop
		2745	iteration, so while the overhead might be noticeable, it doesn't apply to
		2746	repeated calls to C<ev_async_send> for the same event loop.
		2747
		2748	=item bool = ev_async_pending (ev_async *)
		2749
		2750	Returns a non-zero value when C<ev_async_send> has been called on the
		2751	watcher but the event has not yet been processed (or even noted) by the
		2752	event loop.
		2753
		2754	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		2755	the loop iterates next and checks for the watcher to have become active,
		2756	it will reset the flag again. C<ev_async_pending> can be used to very
		2757	quickly check whether invoking the loop might be a good idea.
		2758
		2759	Not that this does I<not> check whether the watcher itself is pending,
		2760	only whether it has been requested to make this watcher pending: there
		2761	is a time window between the event loop checking and resetting the async
		2762	notification, and the callback being invoked.
		2763
		2764	=back
		2765
		2766
793	=head1 OTHER FUNCTIONS	2767	=head1 OTHER FUNCTIONS
794		2768
795	There are some other functions of possible interest. Described. Here. Now.	2769	There are some other functions of possible interest. Described. Here. Now.
796		2770
797	=over 4	2771	=over 4
798		2772
799	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2773	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
800		2774
801	This function combines a simple timer and an I/O watcher, calls your	2775	This function combines a simple timer and an I/O watcher, calls your
802	callback on whichever event happens first and automatically stop both	2776	callback on whichever event happens first and automatically stops both
803	watchers. This is useful if you want to wait for a single event on an fd	2777	watchers. This is useful if you want to wait for a single event on an fd
804	or timeout without having to allocate/configure/start/stop/free one or	2778	or timeout without having to allocate/configure/start/stop/free one or
805	more watchers yourself.	2779	more watchers yourself.
806		2780
807	If C<fd> is less than 0, then no I/O watcher will be started and events	2781	If C<fd> is less than 0, then no I/O watcher will be started and the
808	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	2782	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
809	C<events> set will be craeted and started.	2783	the given C<fd> and C<events> set will be created and started.
810		2784
811	If C<timeout> is less than 0, then no timeout watcher will be	2785	If C<timeout> is less than 0, then no timeout watcher will be
812	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	2786	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
813	repeat = 0) will be started. While C<0> is a valid timeout, it is of	2787	repeat = 0) will be started. C<0> is a valid timeout.
814	dubious value.
815		2788
816	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	2789	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
817	passed an C<revents> set like normal event callbacks (a combination of	2790	passed an C<revents> set like normal event callbacks (a combination of
818	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	2791	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
819	value passed to C<ev_once>:	2792	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2793	a timeout and an io event at the same time - you probably should give io
		2794	events precedence.
820		2795
		2796	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		2797
821	static void stdin_ready (int revents, void *arg)	2798	static void stdin_ready (int revents, void *arg)
822	{	2799	{
823	if (revents & EV_TIMEOUT)
824	/* doh, nothing entered */;
825	else if (revents & EV_READ)	2800	if (revents & EV_READ)
826	/* stdin might have data for us, joy! */;	2801	/* stdin might have data for us, joy! */;
		2802	else if (revents & EV_TIMEOUT)
		2803	/* doh, nothing entered */;
827	}	2804	}
828		2805
829	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	2806	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
830		2807
831	=item ev_feed_event (loop, watcher, int events)	2808	=item ev_feed_event (struct ev_loop *, watcher *, int revents)
832		2809
833	Feeds the given event set into the event loop, as if the specified event	2810	Feeds the given event set into the event loop, as if the specified event
834	had happened for the specified watcher (which must be a pointer to an	2811	had happened for the specified watcher (which must be a pointer to an
835	initialised but not necessarily started event watcher).	2812	initialised but not necessarily started event watcher).
836		2813
837	=item ev_feed_fd_event (loop, int fd, int revents)	2814	=item ev_feed_fd_event (struct ev_loop *, int fd, int revents)
838		2815
839	Feed an event on the given fd, as if a file descriptor backend detected	2816	Feed an event on the given fd, as if a file descriptor backend detected
840	the given events it.	2817	the given events it.
841		2818
842	=item ev_feed_signal_event (loop, int signum)	2819	=item ev_feed_signal_event (struct ev_loop *loop, int signum)
843		2820
844	Feed an event as if the given signal occured (loop must be the default loop!).	2821	Feed an event as if the given signal occurred (C<loop> must be the default
		2822	loop!).
845		2823
846	=back	2824	=back
		2825
847		2826
848	=head1 LIBEVENT EMULATION	2827	=head1 LIBEVENT EMULATION
849		2828
850	Libev offers a compatibility emulation layer for libevent. It cannot	2829	Libev offers a compatibility emulation layer for libevent. It cannot
851	emulate the internals of libevent, so here are some usage hints:	2830	emulate the internals of libevent, so here are some usage hints:
…		…
863		2842
864	=item * Priorities are not currently supported. Initialising priorities	2843	=item * Priorities are not currently supported. Initialising priorities
865	will fail and all watchers will have the same priority, even though there	2844	will fail and all watchers will have the same priority, even though there
866	is an ev_pri field.	2845	is an ev_pri field.
867		2846
		2847	=item * In libevent, the last base created gets the signals, in libev, the
		2848	first base created (== the default loop) gets the signals.
		2849
868	=item * Other members are not supported.	2850	=item * Other members are not supported.
869		2851
870	=item * The libev emulation is I<not> ABI compatible to libevent, you need	2852	=item * The libev emulation is I<not> ABI compatible to libevent, you need
871	to use the libev header file and library.	2853	to use the libev header file and library.
872		2854
873	=back	2855	=back
874		2856
875	=head1 C++ SUPPORT	2857	=head1 C++ SUPPORT
876		2858
877	TBD.	2859	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		2860	you to use some convenience methods to start/stop watchers and also change
		2861	the callback model to a model using method callbacks on objects.
		2862
		2863	To use it,
		2864
		2865	#include <ev++.h>
		2866
		2867	This automatically includes F<ev.h> and puts all of its definitions (many
		2868	of them macros) into the global namespace. All C++ specific things are
		2869	put into the C<ev> namespace. It should support all the same embedding
		2870	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		2871
		2872	Care has been taken to keep the overhead low. The only data member the C++
		2873	classes add (compared to plain C-style watchers) is the event loop pointer
		2874	that the watcher is associated with (or no additional members at all if
		2875	you disable C<EV_MULTIPLICITY> when embedding libev).
		2876
		2877	Currently, functions, and static and non-static member functions can be
		2878	used as callbacks. Other types should be easy to add as long as they only
		2879	need one additional pointer for context. If you need support for other
		2880	types of functors please contact the author (preferably after implementing
		2881	it).
		2882
		2883	Here is a list of things available in the C<ev> namespace:
		2884
		2885	=over 4
		2886
		2887	=item C<ev::READ>, C<ev::WRITE> etc.
		2888
		2889	These are just enum values with the same values as the C<EV_READ> etc.
		2890	macros from F<ev.h>.
		2891
		2892	=item C<ev::tstamp>, C<ev::now>
		2893
		2894	Aliases to the same types/functions as with the C<ev_> prefix.
		2895
		2896	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		2897
		2898	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		2899	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		2900	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		2901	defines by many implementations.
		2902
		2903	All of those classes have these methods:
		2904
		2905	=over 4
		2906
		2907	=item ev::TYPE::TYPE ()
		2908
		2909	=item ev::TYPE::TYPE (struct ev_loop *)
		2910
		2911	=item ev::TYPE::~TYPE
		2912
		2913	The constructor (optionally) takes an event loop to associate the watcher
		2914	with. If it is omitted, it will use C<EV_DEFAULT>.
		2915
		2916	The constructor calls C<ev_init> for you, which means you have to call the
		2917	C<set> method before starting it.
		2918
		2919	It will not set a callback, however: You have to call the templated C<set>
		2920	method to set a callback before you can start the watcher.
		2921
		2922	(The reason why you have to use a method is a limitation in C++ which does
		2923	not allow explicit template arguments for constructors).
		2924
		2925	The destructor automatically stops the watcher if it is active.
		2926
		2927	=item w->set<class, &class::method> (object *)
		2928
		2929	This method sets the callback method to call. The method has to have a
		2930	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		2931	first argument and the C<revents> as second. The object must be given as
		2932	parameter and is stored in the C<data> member of the watcher.
		2933
		2934	This method synthesizes efficient thunking code to call your method from
		2935	the C callback that libev requires. If your compiler can inline your
		2936	callback (i.e. it is visible to it at the place of the C<set> call and
		2937	your compiler is good :), then the method will be fully inlined into the
		2938	thunking function, making it as fast as a direct C callback.
		2939
		2940	Example: simple class declaration and watcher initialisation
		2941
		2942	struct myclass
		2943	{
		2944	void io_cb (ev::io &w, int revents) { }
		2945	}
		2946
		2947	myclass obj;
		2948	ev::io iow;
		2949	iow.set <myclass, &myclass::io_cb> (&obj);
		2950
		2951	=item w->set (object *)
		2952
		2953	This is an B<experimental> feature that might go away in a future version.
		2954
		2955	This is a variation of a method callback - leaving out the method to call
		2956	will default the method to C<operator ()>, which makes it possible to use
		2957	functor objects without having to manually specify the C<operator ()> all
		2958	the time. Incidentally, you can then also leave out the template argument
		2959	list.
		2960
		2961	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		2962	int revents)>.
		2963
		2964	See the method-C<set> above for more details.
		2965
		2966	Example: use a functor object as callback.
		2967
		2968	struct myfunctor
		2969	{
		2970	void operator() (ev::io &w, int revents)
		2971	{
		2972	...
		2973	}
		2974	}
		2975
		2976	myfunctor f;
		2977
		2978	ev::io w;
		2979	w.set (&f);
		2980
		2981	=item w->set<function> (void *data = 0)
		2982
		2983	Also sets a callback, but uses a static method or plain function as
		2984	callback. The optional C<data> argument will be stored in the watcher's
		2985	C<data> member and is free for you to use.
		2986
		2987	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		2988
		2989	See the method-C<set> above for more details.
		2990
		2991	Example: Use a plain function as callback.
		2992
		2993	static void io_cb (ev::io &w, int revents) { }
		2994	iow.set <io_cb> ();
		2995
		2996	=item w->set (struct ev_loop *)
		2997
		2998	Associates a different C<struct ev_loop> with this watcher. You can only
		2999	do this when the watcher is inactive (and not pending either).
		3000
		3001	=item w->set ([arguments])
		3002
		3003	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
		3004	called at least once. Unlike the C counterpart, an active watcher gets
		3005	automatically stopped and restarted when reconfiguring it with this
		3006	method.
		3007
		3008	=item w->start ()
		3009
		3010	Starts the watcher. Note that there is no C<loop> argument, as the
		3011	constructor already stores the event loop.
		3012
		3013	=item w->stop ()
		3014
		3015	Stops the watcher if it is active. Again, no C<loop> argument.
		3016
		3017	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3018
		3019	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3020	C<ev_TYPE_again> function.
		3021
		3022	=item w->sweep () (C<ev::embed> only)
		3023
		3024	Invokes C<ev_embed_sweep>.
		3025
		3026	=item w->update () (C<ev::stat> only)
		3027
		3028	Invokes C<ev_stat_stat>.
		3029
		3030	=back
		3031
		3032	=back
		3033
		3034	Example: Define a class with an IO and idle watcher, start one of them in
		3035	the constructor.
		3036
		3037	class myclass
		3038	{
		3039	ev::io io ; void io_cb (ev::io &w, int revents);
		3040	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3041
		3042	myclass (int fd)
		3043	{
		3044	io .set <myclass, &myclass::io_cb > (this);
		3045	idle.set <myclass, &myclass::idle_cb> (this);
		3046
		3047	io.start (fd, ev::READ);
		3048	}
		3049	};
		3050
		3051
		3052	=head1 OTHER LANGUAGE BINDINGS
		3053
		3054	Libev does not offer other language bindings itself, but bindings for a
		3055	number of languages exist in the form of third-party packages. If you know
		3056	any interesting language binding in addition to the ones listed here, drop
		3057	me a note.
		3058
		3059	=over 4
		3060
		3061	=item Perl
		3062
		3063	The EV module implements the full libev API and is actually used to test
		3064	libev. EV is developed together with libev. Apart from the EV core module,
		3065	there are additional modules that implement libev-compatible interfaces
		3066	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3067	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3068	and C<EV::Glib>).
		3069
		3070	It can be found and installed via CPAN, its homepage is at
		3071	L<http://software.schmorp.de/pkg/EV>.
		3072
		3073	=item Python
		3074
		3075	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3076	seems to be quite complete and well-documented.
		3077
		3078	=item Ruby
		3079
		3080	Tony Arcieri has written a ruby extension that offers access to a subset
		3081	of the libev API and adds file handle abstractions, asynchronous DNS and
		3082	more on top of it. It can be found via gem servers. Its homepage is at
		3083	L<http://rev.rubyforge.org/>.
		3084
		3085	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3086	makes rev work even on mingw.
		3087
		3088	=item Haskell
		3089
		3090	A haskell binding to libev is available at
		3091	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3092
		3093	=item D
		3094
		3095	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3096	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3097
		3098	=item Ocaml
		3099
		3100	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3101	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3102
		3103	=back
		3104
		3105
		3106	=head1 MACRO MAGIC
		3107
		3108	Libev can be compiled with a variety of options, the most fundamental
		3109	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3110	functions and callbacks have an initial C<struct ev_loop *> argument.
		3111
		3112	To make it easier to write programs that cope with either variant, the
		3113	following macros are defined:
		3114
		3115	=over 4
		3116
		3117	=item C<EV_A>, C<EV_A_>
		3118
		3119	This provides the loop I<argument> for functions, if one is required ("ev
		3120	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3121	C<EV_A_> is used when other arguments are following. Example:
		3122
		3123	ev_unref (EV_A);
		3124	ev_timer_add (EV_A_ watcher);
		3125	ev_loop (EV_A_ 0);
		3126
		3127	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3128	which is often provided by the following macro.
		3129
		3130	=item C<EV_P>, C<EV_P_>
		3131
		3132	This provides the loop I<parameter> for functions, if one is required ("ev
		3133	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3134	C<EV_P_> is used when other parameters are following. Example:
		3135
		3136	// this is how ev_unref is being declared
		3137	static void ev_unref (EV_P);
		3138
		3139	// this is how you can declare your typical callback
		3140	static void cb (EV_P_ ev_timer *w, int revents)
		3141
		3142	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3143	suitable for use with C<EV_A>.
		3144
		3145	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3146
		3147	Similar to the other two macros, this gives you the value of the default
		3148	loop, if multiple loops are supported ("ev loop default").
		3149
		3150	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3151
		3152	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3153	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3154	is undefined when the default loop has not been initialised by a previous
		3155	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3156
		3157	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3158	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3159
		3160	=back
		3161
		3162	Example: Declare and initialise a check watcher, utilising the above
		3163	macros so it will work regardless of whether multiple loops are supported
		3164	or not.
		3165
		3166	static void
		3167	check_cb (EV_P_ ev_timer *w, int revents)
		3168	{
		3169	ev_check_stop (EV_A_ w);
		3170	}
		3171
		3172	ev_check check;
		3173	ev_check_init (&check, check_cb);
		3174	ev_check_start (EV_DEFAULT_ &check);
		3175	ev_loop (EV_DEFAULT_ 0);
		3176
		3177	=head1 EMBEDDING
		3178
		3179	Libev can (and often is) directly embedded into host
		3180	applications. Examples of applications that embed it include the Deliantra
		3181	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3182	and rxvt-unicode.
		3183
		3184	The goal is to enable you to just copy the necessary files into your
		3185	source directory without having to change even a single line in them, so
		3186	you can easily upgrade by simply copying (or having a checked-out copy of
		3187	libev somewhere in your source tree).
		3188
		3189	=head2 FILESETS
		3190
		3191	Depending on what features you need you need to include one or more sets of files
		3192	in your application.
		3193
		3194	=head3 CORE EVENT LOOP
		3195
		3196	To include only the libev core (all the C<ev_*> functions), with manual
		3197	configuration (no autoconf):
		3198
		3199	#define EV_STANDALONE 1
		3200	#include "ev.c"
		3201
		3202	This will automatically include F<ev.h>, too, and should be done in a
		3203	single C source file only to provide the function implementations. To use
		3204	it, do the same for F<ev.h> in all files wishing to use this API (best
		3205	done by writing a wrapper around F<ev.h> that you can include instead and
		3206	where you can put other configuration options):
		3207
		3208	#define EV_STANDALONE 1
		3209	#include "ev.h"
		3210
		3211	Both header files and implementation files can be compiled with a C++
		3212	compiler (at least, that's a stated goal, and breakage will be treated
		3213	as a bug).
		3214
		3215	You need the following files in your source tree, or in a directory
		3216	in your include path (e.g. in libev/ when using -Ilibev):
		3217
		3218	ev.h
		3219	ev.c
		3220	ev_vars.h
		3221	ev_wrap.h
		3222
		3223	ev_win32.c required on win32 platforms only
		3224
		3225	ev_select.c only when select backend is enabled (which is enabled by default)
		3226	ev_poll.c only when poll backend is enabled (disabled by default)
		3227	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3228	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3229	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3230
		3231	F<ev.c> includes the backend files directly when enabled, so you only need
		3232	to compile this single file.
		3233
		3234	=head3 LIBEVENT COMPATIBILITY API
		3235
		3236	To include the libevent compatibility API, also include:
		3237
		3238	#include "event.c"
		3239
		3240	in the file including F<ev.c>, and:
		3241
		3242	#include "event.h"
		3243
		3244	in the files that want to use the libevent API. This also includes F<ev.h>.
		3245
		3246	You need the following additional files for this:
		3247
		3248	event.h
		3249	event.c
		3250
		3251	=head3 AUTOCONF SUPPORT
		3252
		3253	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3254	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3255	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3256	include F<config.h> and configure itself accordingly.
		3257
		3258	For this of course you need the m4 file:
		3259
		3260	libev.m4
		3261
		3262	=head2 PREPROCESSOR SYMBOLS/MACROS
		3263
		3264	Libev can be configured via a variety of preprocessor symbols you have to
		3265	define before including any of its files. The default in the absence of
		3266	autoconf is documented for every option.
		3267
		3268	=over 4
		3269
		3270	=item EV_STANDALONE
		3271
		3272	Must always be C<1> if you do not use autoconf configuration, which
		3273	keeps libev from including F<config.h>, and it also defines dummy
		3274	implementations for some libevent functions (such as logging, which is not
		3275	supported). It will also not define any of the structs usually found in
		3276	F<event.h> that are not directly supported by the libev core alone.
		3277
		3278	In stanbdalone mode, libev will still try to automatically deduce the
		3279	configuration, but has to be more conservative.
		3280
		3281	=item EV_USE_MONOTONIC
		3282
		3283	If defined to be C<1>, libev will try to detect the availability of the
		3284	monotonic clock option at both compile time and runtime. Otherwise no
		3285	use of the monotonic clock option will be attempted. If you enable this,
		3286	you usually have to link against librt or something similar. Enabling it
		3287	when the functionality isn't available is safe, though, although you have
		3288	to make sure you link against any libraries where the C<clock_gettime>
		3289	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		3290
		3291	=item EV_USE_REALTIME
		3292
		3293	If defined to be C<1>, libev will try to detect the availability of the
		3294	real-time clock option at compile time (and assume its availability
		3295	at runtime if successful). Otherwise no use of the real-time clock
		3296	option will be attempted. This effectively replaces C<gettimeofday>
		3297	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		3298	correctness. See the note about libraries in the description of
		3299	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3300	C<EV_USE_CLOCK_SYSCALL>.
		3301
		3302	=item EV_USE_CLOCK_SYSCALL
		3303
		3304	If defined to be C<1>, libev will try to use a direct syscall instead
		3305	of calling the system-provided C<clock_gettime> function. This option
		3306	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3307	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3308	programs needlessly. Using a direct syscall is slightly slower (in
		3309	theory), because no optimised vdso implementation can be used, but avoids
		3310	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		3311	higher, as it simplifies linking (no need for C<-lrt>).
		3312
		3313	=item EV_USE_NANOSLEEP
		3314
		3315	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3316	and will use it for delays. Otherwise it will use C<select ()>.
		3317
		3318	=item EV_USE_EVENTFD
		3319
		3320	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3321	available and will probe for kernel support at runtime. This will improve
		3322	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3323	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3324	2.7 or newer, otherwise disabled.
		3325
		3326	=item EV_USE_SELECT
		3327
		3328	If undefined or defined to be C<1>, libev will compile in support for the
		3329	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3330	other method takes over, select will be it. Otherwise the select backend
		3331	will not be compiled in.
		3332
		3333	=item EV_SELECT_USE_FD_SET
		3334
		3335	If defined to C<1>, then the select backend will use the system C<fd_set>
		3336	structure. This is useful if libev doesn't compile due to a missing
		3337	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		3338	on exotic systems. This usually limits the range of file descriptors to
		3339	some low limit such as 1024 or might have other limitations (winsocket
		3340	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		3341	configures the maximum size of the C<fd_set>.
		3342
		3343	=item EV_SELECT_IS_WINSOCKET
		3344
		3345	When defined to C<1>, the select backend will assume that
		3346	select/socket/connect etc. don't understand file descriptors but
		3347	wants osf handles on win32 (this is the case when the select to
		3348	be used is the winsock select). This means that it will call
		3349	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3350	it is assumed that all these functions actually work on fds, even
		3351	on win32. Should not be defined on non-win32 platforms.
		3352
		3353	=item EV_FD_TO_WIN32_HANDLE
		3354
		3355	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3356	file descriptors to socket handles. When not defining this symbol (the
		3357	default), then libev will call C<_get_osfhandle>, which is usually
		3358	correct. In some cases, programs use their own file descriptor management,
		3359	in which case they can provide this function to map fds to socket handles.
		3360
		3361	=item EV_USE_POLL
		3362
		3363	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3364	backend. Otherwise it will be enabled on non-win32 platforms. It
		3365	takes precedence over select.
		3366
		3367	=item EV_USE_EPOLL
		3368
		3369	If defined to be C<1>, libev will compile in support for the Linux
		3370	C<epoll>(7) backend. Its availability will be detected at runtime,
		3371	otherwise another method will be used as fallback. This is the preferred
		3372	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3373	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3374
		3375	=item EV_USE_KQUEUE
		3376
		3377	If defined to be C<1>, libev will compile in support for the BSD style
		3378	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3379	otherwise another method will be used as fallback. This is the preferred
		3380	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3381	supports some types of fds correctly (the only platform we found that
		3382	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3383	not be used unless explicitly requested. The best way to use it is to find
		3384	out whether kqueue supports your type of fd properly and use an embedded
		3385	kqueue loop.
		3386
		3387	=item EV_USE_PORT
		3388
		3389	If defined to be C<1>, libev will compile in support for the Solaris
		3390	10 port style backend. Its availability will be detected at runtime,
		3391	otherwise another method will be used as fallback. This is the preferred
		3392	backend for Solaris 10 systems.
		3393
		3394	=item EV_USE_DEVPOLL
		3395
		3396	Reserved for future expansion, works like the USE symbols above.
		3397
		3398	=item EV_USE_INOTIFY
		3399
		3400	If defined to be C<1>, libev will compile in support for the Linux inotify
		3401	interface to speed up C<ev_stat> watchers. Its actual availability will
		3402	be detected at runtime. If undefined, it will be enabled if the headers
		3403	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3404
		3405	=item EV_ATOMIC_T
		3406
		3407	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3408	access is atomic with respect to other threads or signal contexts. No such
		3409	type is easily found in the C language, so you can provide your own type
		3410	that you know is safe for your purposes. It is used both for signal handler "locking"
		3411	as well as for signal and thread safety in C<ev_async> watchers.
		3412
		3413	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3414	(from F<signal.h>), which is usually good enough on most platforms.
		3415
		3416	=item EV_H
		3417
		3418	The name of the F<ev.h> header file used to include it. The default if
		3419	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3420	used to virtually rename the F<ev.h> header file in case of conflicts.
		3421
		3422	=item EV_CONFIG_H
		3423
		3424	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3425	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3426	C<EV_H>, above.
		3427
		3428	=item EV_EVENT_H
		3429
		3430	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3431	of how the F<event.h> header can be found, the default is C<"event.h">.
		3432
		3433	=item EV_PROTOTYPES
		3434
		3435	If defined to be C<0>, then F<ev.h> will not define any function
		3436	prototypes, but still define all the structs and other symbols. This is
		3437	occasionally useful if you want to provide your own wrapper functions
		3438	around libev functions.
		3439
		3440	=item EV_MULTIPLICITY
		3441
		3442	If undefined or defined to C<1>, then all event-loop-specific functions
		3443	will have the C<struct ev_loop *> as first argument, and you can create
		3444	additional independent event loops. Otherwise there will be no support
		3445	for multiple event loops and there is no first event loop pointer
		3446	argument. Instead, all functions act on the single default loop.
		3447
		3448	=item EV_MINPRI
		3449
		3450	=item EV_MAXPRI
		3451
		3452	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3453	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3454	provide for more priorities by overriding those symbols (usually defined
		3455	to be C<-2> and C<2>, respectively).
		3456
		3457	When doing priority-based operations, libev usually has to linearly search
		3458	all the priorities, so having many of them (hundreds) uses a lot of space
		3459	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3460	fine.
		3461
		3462	If your embedding application does not need any priorities, defining these
		3463	both to C<0> will save some memory and CPU.
		3464
		3465	=item EV_PERIODIC_ENABLE
		3466
		3467	If undefined or defined to be C<1>, then periodic timers are supported. If
		3468	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3469	code.
		3470
		3471	=item EV_IDLE_ENABLE
		3472
		3473	If undefined or defined to be C<1>, then idle watchers are supported. If
		3474	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3475	code.
		3476
		3477	=item EV_EMBED_ENABLE
		3478
		3479	If undefined or defined to be C<1>, then embed watchers are supported. If
		3480	defined to be C<0>, then they are not. Embed watchers rely on most other
		3481	watcher types, which therefore must not be disabled.
		3482
		3483	=item EV_STAT_ENABLE
		3484
		3485	If undefined or defined to be C<1>, then stat watchers are supported. If
		3486	defined to be C<0>, then they are not.
		3487
		3488	=item EV_FORK_ENABLE
		3489
		3490	If undefined or defined to be C<1>, then fork watchers are supported. If
		3491	defined to be C<0>, then they are not.
		3492
		3493	=item EV_ASYNC_ENABLE
		3494
		3495	If undefined or defined to be C<1>, then async watchers are supported. If
		3496	defined to be C<0>, then they are not.
		3497
		3498	=item EV_MINIMAL
		3499
		3500	If you need to shave off some kilobytes of code at the expense of some
		3501	speed, define this symbol to C<1>. Currently this is used to override some
		3502	inlining decisions, saves roughly 30% code size on amd64. It also selects a
		3503	much smaller 2-heap for timer management over the default 4-heap.
		3504
		3505	=item EV_PID_HASHSIZE
		3506
		3507	C<ev_child> watchers use a small hash table to distribute workload by
		3508	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
		3509	than enough. If you need to manage thousands of children you might want to
		3510	increase this value (I<must> be a power of two).
		3511
		3512	=item EV_INOTIFY_HASHSIZE
		3513
		3514	C<ev_stat> watchers use a small hash table to distribute workload by
		3515	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		3516	usually more than enough. If you need to manage thousands of C<ev_stat>
		3517	watchers you might want to increase this value (I<must> be a power of
		3518	two).
		3519
		3520	=item EV_USE_4HEAP
		3521
		3522	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3523	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3524	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3525	faster performance with many (thousands) of watchers.
		3526
		3527	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3528	(disabled).
		3529
		3530	=item EV_HEAP_CACHE_AT
		3531
		3532	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3533	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3534	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3535	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3536	but avoids random read accesses on heap changes. This improves performance
		3537	noticeably with many (hundreds) of watchers.
		3538
		3539	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3540	(disabled).
		3541
		3542	=item EV_VERIFY
		3543
		3544	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3545	be done: If set to C<0>, no internal verification code will be compiled
		3546	in. If set to C<1>, then verification code will be compiled in, but not
		3547	called. If set to C<2>, then the internal verification code will be
		3548	called once per loop, which can slow down libev. If set to C<3>, then the
		3549	verification code will be called very frequently, which will slow down
		3550	libev considerably.
		3551
		3552	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3553	C<0>.
		3554
		3555	=item EV_COMMON
		3556
		3557	By default, all watchers have a C<void *data> member. By redefining
		3558	this macro to a something else you can include more and other types of
		3559	members. You have to define it each time you include one of the files,
		3560	though, and it must be identical each time.
		3561
		3562	For example, the perl EV module uses something like this:
		3563
		3564	#define EV_COMMON \
		3565	SV *self; /* contains this struct */ \
		3566	SV *cb_sv, *fh /* note no trailing ";" */
		3567
		3568	=item EV_CB_DECLARE (type)
		3569
		3570	=item EV_CB_INVOKE (watcher, revents)
		3571
		3572	=item ev_set_cb (ev, cb)
		3573
		3574	Can be used to change the callback member declaration in each watcher,
		3575	and the way callbacks are invoked and set. Must expand to a struct member
		3576	definition and a statement, respectively. See the F<ev.h> header file for
		3577	their default definitions. One possible use for overriding these is to
		3578	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		3579	method calls instead of plain function calls in C++.
		3580
		3581	=back
		3582
		3583	=head2 EXPORTED API SYMBOLS
		3584
		3585	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3586	exported symbols, you can use the provided F<Symbol.*> files which list
		3587	all public symbols, one per line:
		3588
		3589	Symbols.ev for libev proper
		3590	Symbols.event for the libevent emulation
		3591
		3592	This can also be used to rename all public symbols to avoid clashes with
		3593	multiple versions of libev linked together (which is obviously bad in
		3594	itself, but sometimes it is inconvenient to avoid this).
		3595
		3596	A sed command like this will create wrapper C<#define>'s that you need to
		3597	include before including F<ev.h>:
		3598
		3599	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3600
		3601	This would create a file F<wrap.h> which essentially looks like this:
		3602
		3603	#define ev_backend myprefix_ev_backend
		3604	#define ev_check_start myprefix_ev_check_start
		3605	#define ev_check_stop myprefix_ev_check_stop
		3606	...
		3607
		3608	=head2 EXAMPLES
		3609
		3610	For a real-world example of a program the includes libev
		3611	verbatim, you can have a look at the EV perl module
		3612	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		3613	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		3614	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		3615	will be compiled. It is pretty complex because it provides its own header
		3616	file.
		3617
		3618	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		3619	that everybody includes and which overrides some configure choices:
		3620
		3621	#define EV_MINIMAL 1
		3622	#define EV_USE_POLL 0
		3623	#define EV_MULTIPLICITY 0
		3624	#define EV_PERIODIC_ENABLE 0
		3625	#define EV_STAT_ENABLE 0
		3626	#define EV_FORK_ENABLE 0
		3627	#define EV_CONFIG_H <config.h>
		3628	#define EV_MINPRI 0
		3629	#define EV_MAXPRI 0
		3630
		3631	#include "ev++.h"
		3632
		3633	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		3634
		3635	#include "ev_cpp.h"
		3636	#include "ev.c"
		3637
		3638	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		3639
		3640	=head2 THREADS AND COROUTINES
		3641
		3642	=head3 THREADS
		3643
		3644	All libev functions are reentrant and thread-safe unless explicitly
		3645	documented otherwise, but libev implements no locking itself. This means
		3646	that you can use as many loops as you want in parallel, as long as there
		3647	are no concurrent calls into any libev function with the same loop
		3648	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		3649	of course): libev guarantees that different event loops share no data
		3650	structures that need any locking.
		3651
		3652	Or to put it differently: calls with different loop parameters can be done
		3653	concurrently from multiple threads, calls with the same loop parameter
		3654	must be done serially (but can be done from different threads, as long as
		3655	only one thread ever is inside a call at any point in time, e.g. by using
		3656	a mutex per loop).
		3657
		3658	Specifically to support threads (and signal handlers), libev implements
		3659	so-called C<ev_async> watchers, which allow some limited form of
		3660	concurrency on the same event loop, namely waking it up "from the
		3661	outside".
		3662
		3663	If you want to know which design (one loop, locking, or multiple loops
		3664	without or something else still) is best for your problem, then I cannot
		3665	help you, but here is some generic advice:
		3666
		3667	=over 4
		3668
		3669	=item * most applications have a main thread: use the default libev loop
		3670	in that thread, or create a separate thread running only the default loop.
		3671
		3672	This helps integrating other libraries or software modules that use libev
		3673	themselves and don't care/know about threading.
		3674
		3675	=item * one loop per thread is usually a good model.
		3676
		3677	Doing this is almost never wrong, sometimes a better-performance model
		3678	exists, but it is always a good start.
		3679
		3680	=item * other models exist, such as the leader/follower pattern, where one
		3681	loop is handed through multiple threads in a kind of round-robin fashion.
		3682
		3683	Choosing a model is hard - look around, learn, know that usually you can do
		3684	better than you currently do :-)
		3685
		3686	=item * often you need to talk to some other thread which blocks in the
		3687	event loop.
		3688
		3689	C<ev_async> watchers can be used to wake them up from other threads safely
		3690	(or from signal contexts...).
		3691
		3692	An example use would be to communicate signals or other events that only
		3693	work in the default loop by registering the signal watcher with the
		3694	default loop and triggering an C<ev_async> watcher from the default loop
		3695	watcher callback into the event loop interested in the signal.
		3696
		3697	=back
		3698
		3699	=head3 COROUTINES
		3700
		3701	Libev is very accommodating to coroutines ("cooperative threads"):
		3702	libev fully supports nesting calls to its functions from different
		3703	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		3704	different coroutines, and switch freely between both coroutines running the
		3705	loop, as long as you don't confuse yourself). The only exception is that
		3706	you must not do this from C<ev_periodic> reschedule callbacks.
		3707
		3708	Care has been taken to ensure that libev does not keep local state inside
		3709	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		3710	they do not call any callbacks.
		3711
		3712	=head2 COMPILER WARNINGS
		3713
		3714	Depending on your compiler and compiler settings, you might get no or a
		3715	lot of warnings when compiling libev code. Some people are apparently
		3716	scared by this.
		3717
		3718	However, these are unavoidable for many reasons. For one, each compiler
		3719	has different warnings, and each user has different tastes regarding
		3720	warning options. "Warn-free" code therefore cannot be a goal except when
		3721	targeting a specific compiler and compiler-version.
		3722
		3723	Another reason is that some compiler warnings require elaborate
		3724	workarounds, or other changes to the code that make it less clear and less
		3725	maintainable.
		3726
		3727	And of course, some compiler warnings are just plain stupid, or simply
		3728	wrong (because they don't actually warn about the condition their message
		3729	seems to warn about). For example, certain older gcc versions had some
		3730	warnings that resulted an extreme number of false positives. These have
		3731	been fixed, but some people still insist on making code warn-free with
		3732	such buggy versions.
		3733
		3734	While libev is written to generate as few warnings as possible,
		3735	"warn-free" code is not a goal, and it is recommended not to build libev
		3736	with any compiler warnings enabled unless you are prepared to cope with
		3737	them (e.g. by ignoring them). Remember that warnings are just that:
		3738	warnings, not errors, or proof of bugs.
		3739
		3740
		3741	=head2 VALGRIND
		3742
		3743	Valgrind has a special section here because it is a popular tool that is
		3744	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		3745
		3746	If you think you found a bug (memory leak, uninitialised data access etc.)
		3747	in libev, then check twice: If valgrind reports something like:
		3748
		3749	==2274== definitely lost: 0 bytes in 0 blocks.
		3750	==2274== possibly lost: 0 bytes in 0 blocks.
		3751	==2274== still reachable: 256 bytes in 1 blocks.
		3752
		3753	Then there is no memory leak, just as memory accounted to global variables
		3754	is not a memleak - the memory is still being referenced, and didn't leak.
		3755
		3756	Similarly, under some circumstances, valgrind might report kernel bugs
		3757	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3758	although an acceptable workaround has been found here), or it might be
		3759	confused.
		3760
		3761	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		3762	make it into some kind of religion.
		3763
		3764	If you are unsure about something, feel free to contact the mailing list
		3765	with the full valgrind report and an explanation on why you think this
		3766	is a bug in libev (best check the archives, too :). However, don't be
		3767	annoyed when you get a brisk "this is no bug" answer and take the chance
		3768	of learning how to interpret valgrind properly.
		3769
		3770	If you need, for some reason, empty reports from valgrind for your project
		3771	I suggest using suppression lists.
		3772
		3773
		3774	=head1 PORTABILITY NOTES
		3775
		3776	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		3777
		3778	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		3779	requires, and its I/O model is fundamentally incompatible with the POSIX
		3780	model. Libev still offers limited functionality on this platform in
		3781	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		3782	descriptors. This only applies when using Win32 natively, not when using
		3783	e.g. cygwin.
		3784
		3785	Lifting these limitations would basically require the full
		3786	re-implementation of the I/O system. If you are into these kinds of
		3787	things, then note that glib does exactly that for you in a very portable
		3788	way (note also that glib is the slowest event library known to man).
		3789
		3790	There is no supported compilation method available on windows except
		3791	embedding it into other applications.
		3792
		3793	Not a libev limitation but worth mentioning: windows apparently doesn't
		3794	accept large writes: instead of resulting in a partial write, windows will
		3795	either accept everything or return C<ENOBUFS> if the buffer is too large,
		3796	so make sure you only write small amounts into your sockets (less than a
		3797	megabyte seems safe, but this apparently depends on the amount of memory
		3798	available).
		3799
		3800	Due to the many, low, and arbitrary limits on the win32 platform and
		3801	the abysmal performance of winsockets, using a large number of sockets
		3802	is not recommended (and not reasonable). If your program needs to use
		3803	more than a hundred or so sockets, then likely it needs to use a totally
		3804	different implementation for windows, as libev offers the POSIX readiness
		3805	notification model, which cannot be implemented efficiently on windows
		3806	(Microsoft monopoly games).
		3807
		3808	A typical way to use libev under windows is to embed it (see the embedding
		3809	section for details) and use the following F<evwrap.h> header file instead
		3810	of F<ev.h>:
		3811
		3812	#define EV_STANDALONE /* keeps ev from requiring config.h */
		3813	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		3814
		3815	#include "ev.h"
		3816
		3817	And compile the following F<evwrap.c> file into your project (make sure
		3818	you do I<not> compile the F<ev.c> or any other embedded source files!):
		3819
		3820	#include "evwrap.h"
		3821	#include "ev.c"
		3822
		3823	=over 4
		3824
		3825	=item The winsocket select function
		3826
		3827	The winsocket C<select> function doesn't follow POSIX in that it
		3828	requires socket I<handles> and not socket I<file descriptors> (it is
		3829	also extremely buggy). This makes select very inefficient, and also
		3830	requires a mapping from file descriptors to socket handles (the Microsoft
		3831	C runtime provides the function C<_open_osfhandle> for this). See the
		3832	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		3833	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		3834
		3835	The configuration for a "naked" win32 using the Microsoft runtime
		3836	libraries and raw winsocket select is:
		3837
		3838	#define EV_USE_SELECT 1
		3839	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		3840
		3841	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3842	complexity in the O(n²) range when using win32.
		3843
		3844	=item Limited number of file descriptors
		3845
		3846	Windows has numerous arbitrary (and low) limits on things.
		3847
		3848	Early versions of winsocket's select only supported waiting for a maximum
		3849	of C<64> handles (probably owning to the fact that all windows kernels
		3850	can only wait for C<64> things at the same time internally; Microsoft
		3851	recommends spawning a chain of threads and wait for 63 handles and the
		3852	previous thread in each. Great).
		3853
		3854	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3855	to some high number (e.g. C<2048>) before compiling the winsocket select
		3856	call (which might be in libev or elsewhere, for example, perl does its own
		3857	select emulation on windows).
		3858
		3859	Another limit is the number of file descriptors in the Microsoft runtime
		3860	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3861	or something like this inside Microsoft). You can increase this by calling
		3862	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3863	arbitrary limit), but is broken in many versions of the Microsoft runtime
		3864	libraries.
		3865
		3866	This might get you to about C<512> or C<2048> sockets (depending on
		3867	windows version and/or the phase of the moon). To get more, you need to
		3868	wrap all I/O functions and provide your own fd management, but the cost of
		3869	calling select (O(n²)) will likely make this unworkable.
		3870
		3871	=back
		3872
		3873	=head2 PORTABILITY REQUIREMENTS
		3874
		3875	In addition to a working ISO-C implementation and of course the
		3876	backend-specific APIs, libev relies on a few additional extensions:
		3877
		3878	=over 4
		3879
		3880	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		3881	calling conventions regardless of C<ev_watcher_type *>.
		3882
		3883	Libev assumes not only that all watcher pointers have the same internal
		3884	structure (guaranteed by POSIX but not by ISO C for example), but it also
		3885	assumes that the same (machine) code can be used to call any watcher
		3886	callback: The watcher callbacks have different type signatures, but libev
		3887	calls them using an C<ev_watcher *> internally.
		3888
		3889	=item C<sig_atomic_t volatile> must be thread-atomic as well
		3890
		3891	The type C<sig_atomic_t volatile> (or whatever is defined as
		3892	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		3893	threads. This is not part of the specification for C<sig_atomic_t>, but is
		3894	believed to be sufficiently portable.
		3895
		3896	=item C<sigprocmask> must work in a threaded environment
		3897
		3898	Libev uses C<sigprocmask> to temporarily block signals. This is not
		3899	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		3900	pthread implementations will either allow C<sigprocmask> in the "main
		3901	thread" or will block signals process-wide, both behaviours would
		3902	be compatible with libev. Interaction between C<sigprocmask> and
		3903	C<pthread_sigmask> could complicate things, however.
		3904
		3905	The most portable way to handle signals is to block signals in all threads
		3906	except the initial one, and run the default loop in the initial thread as
		3907	well.
		3908
		3909	=item C<long> must be large enough for common memory allocation sizes
		3910
		3911	To improve portability and simplify its API, libev uses C<long> internally
		3912	instead of C<size_t> when allocating its data structures. On non-POSIX
		3913	systems (Microsoft...) this might be unexpectedly low, but is still at
		3914	least 31 bits everywhere, which is enough for hundreds of millions of
		3915	watchers.
		3916
		3917	=item C<double> must hold a time value in seconds with enough accuracy
		3918
		3919	The type C<double> is used to represent timestamps. It is required to
		3920	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		3921	enough for at least into the year 4000. This requirement is fulfilled by
		3922	implementations implementing IEEE 754 (basically all existing ones).
		3923
		3924	=back
		3925
		3926	If you know of other additional requirements drop me a note.
		3927
		3928
		3929	=head1 ALGORITHMIC COMPLEXITIES
		3930
		3931	In this section the complexities of (many of) the algorithms used inside
		3932	libev will be documented. For complexity discussions about backends see
		3933	the documentation for C<ev_default_init>.
		3934
		3935	All of the following are about amortised time: If an array needs to be
		3936	extended, libev needs to realloc and move the whole array, but this
		3937	happens asymptotically rarer with higher number of elements, so O(1) might
		3938	mean that libev does a lengthy realloc operation in rare cases, but on
		3939	average it is much faster and asymptotically approaches constant time.
		3940
		3941	=over 4
		3942
		3943	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		3944
		3945	This means that, when you have a watcher that triggers in one hour and
		3946	there are 100 watchers that would trigger before that, then inserting will
		3947	have to skip roughly seven (C<ld 100>) of these watchers.
		3948
		3949	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		3950
		3951	That means that changing a timer costs less than removing/adding them,
		3952	as only the relative motion in the event queue has to be paid for.
		3953
		3954	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		3955
		3956	These just add the watcher into an array or at the head of a list.
		3957
		3958	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		3959
		3960	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		3961
		3962	These watchers are stored in lists, so they need to be walked to find the
		3963	correct watcher to remove. The lists are usually short (you don't usually
		3964	have many watchers waiting for the same fd or signal: one is typical, two
		3965	is rare).
		3966
		3967	=item Finding the next timer in each loop iteration: O(1)
		3968
		3969	By virtue of using a binary or 4-heap, the next timer is always found at a
		3970	fixed position in the storage array.
		3971
		3972	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		3973
		3974	A change means an I/O watcher gets started or stopped, which requires
		3975	libev to recalculate its status (and possibly tell the kernel, depending
		3976	on backend and whether C<ev_io_set> was used).
		3977
		3978	=item Activating one watcher (putting it into the pending state): O(1)
		3979
		3980	=item Priority handling: O(number_of_priorities)
		3981
		3982	Priorities are implemented by allocating some space for each
		3983	priority. When doing priority-based operations, libev usually has to
		3984	linearly search all the priorities, but starting/stopping and activating
		3985	watchers becomes O(1) with respect to priority handling.
		3986
		3987	=item Sending an ev_async: O(1)
		3988
		3989	=item Processing ev_async_send: O(number_of_async_watchers)
		3990
		3991	=item Processing signals: O(max_signal_number)
		3992
		3993	Sending involves a system call I<iff> there were no other C<ev_async_send>
		3994	calls in the current loop iteration. Checking for async and signal events
		3995	involves iterating over all running async watchers or all signal numbers.
		3996
		3997	=back
		3998
878		3999
879	=head1 AUTHOR	4000	=head1 AUTHOR
880		4001
881	Marc Lehmann <libev@schmorp.de>.	4002	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
882		4003

Diff Legend

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