ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.30 by root, Fri Nov 23 04:36:03 2007 UTC vs.
Revision 1.237 by root, Thu Apr 16 07:58:03 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		8
9	=head1 DESCRIPTION	9	=head2 EXAMPLE PROGRAM
		10
		11	// a single header file is required
		12	#include <ev.h>
		13
		14	#include <stdio.h> // for puts
		15
		16	// every watcher type has its own typedef'd struct
		17	// with the name ev_TYPE
		18	ev_io stdin_watcher;
		19	ev_timer timeout_watcher;
		20
		21	// all watcher callbacks have a similar signature
		22	// this callback is called when data is readable on stdin
		23	static void
		24	stdin_cb (EV_P_ ev_io *w, int revents)
		25	{
		26	puts ("stdin ready");
		27	// for one-shot events, one must manually stop the watcher
		28	// with its corresponding stop function.
		29	ev_io_stop (EV_A_ w);
		30
		31	// this causes all nested ev_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	}
		66
		67	=head1 ABOUT THIS DOCUMENT
		68
		69	This document documents the libev software package.
		70
		71	The newest version of this document is also available as an html-formatted
		72	web page you might find easier to navigate when reading it for the first
		73	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		74
		75	While this document tries to be as complete as possible in documenting
		76	libev, its usage and the rationale behind its design, it is not a tutorial
		77	on event-based programming, nor will it introduce event-based programming
		78	with libev.
		79
		80	Familarity with event based programming techniques in general is assumed
		81	throughout this document.
		82
		83	=head1 ABOUT LIBEV
10		84
11	Libev is an event loop: you register interest in certain events (such as a	85	Libev is an event loop: you register interest in certain events (such as a
12	file descriptor being readable or a timeout occuring), and it will manage	86	file descriptor being readable or a timeout occurring), and it will manage
13	these event sources and provide your program with events.	87	these event sources and provide your program with events.
14		88
15	To do this, it must take more or less complete control over your process	89	To do this, it must take more or less complete control over your process
16	(or thread) by executing the I<event loop> handler, and will then	90	(or thread) by executing the I<event loop> handler, and will then
17	communicate events via a callback mechanism.	91	communicate events via a callback mechanism.
…		…
19	You register interest in certain events by registering so-called I<event	93	You register interest in certain events by registering so-called I<event
20	watchers>, which are relatively small C structures you initialise with the	94	watchers>, which are relatively small C structures you initialise with the
21	details of the event, and then hand it over to libev by I<starting> the	95	details of the event, and then hand it over to libev by I<starting> the
22	watcher.	96	watcher.
23		97
24	=head1 FEATURES	98	=head2 FEATURES
25		99
26	Libev supports select, poll, the linux-specific epoll and the bsd-specific	100	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
27	kqueue mechanisms for file descriptor events, relative timers, absolute	101	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
28	timers with customised rescheduling, signal events, process status change	102	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
29	events (related to SIGCHLD), and event watchers dealing with the event	103	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
30	loop mechanism itself (idle, prepare and check watchers). It also is quite	104	with customised rescheduling (C<ev_periodic>), synchronous signals
		105	(C<ev_signal>), process status change events (C<ev_child>), and event
		106	watchers dealing with the event loop mechanism itself (C<ev_idle>,
		107	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as
		108	file watchers (C<ev_stat>) and even limited support for fork events
		109	(C<ev_fork>).
		110
		111	It also is quite fast (see this
31	fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing	112	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
32	it to libevent for example).	113	for example).
33		114
34	=head1 CONVENTIONS	115	=head2 CONVENTIONS
35		116
36	Libev is very configurable. In this manual the default configuration	117	Libev is very configurable. In this manual the default (and most common)
37	will be described, which supports multiple event loops. For more info	118	configuration will be described, which supports multiple event loops. For
38	about various configuration options please have a look at the file	119	more info about various configuration options please have a look at
39	F<README.embed> in the libev distribution. If libev was configured without	120	B<EMBED> section in this manual. If libev was configured without support
40	support for multiple event loops, then all functions taking an initial	121	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 *>)	122	name C<loop> (which is always of type C<ev_loop *>) will not have
42	will not have this argument.	123	this argument.
43		124
44	=head1 TIME REPRESENTATION	125	=head2 TIME REPRESENTATION
45		126
46	Libev represents time as a single floating point number, representing the	127	Libev represents time as a single floating point number, representing
47	(fractional) number of seconds since the (POSIX) epoch (somewhere near	128	the (fractional) number of seconds since the (POSIX) epoch (somewhere
48	the beginning of 1970, details are complicated, don't ask). This type is	129	near the beginning of 1970, details are complicated, don't ask). This
49	called C<ev_tstamp>, which is what you should use too. It usually aliases	130	type is called C<ev_tstamp>, which is what you should use too. It usually
50	to the double type in C.	131	aliases to the C<double> type in C. When you need to do any calculations
		132	on it, you should treat it as some floating point value. Unlike the name
		133	component C<stamp> might indicate, it is also used for time differences
		134	throughout libev.
		135
		136	=head1 ERROR HANDLING
		137
		138	Libev knows three classes of errors: operating system errors, usage errors
		139	and internal errors (bugs).
		140
		141	When libev catches an operating system error it cannot handle (for example
		142	a system call indicating a condition libev cannot fix), it calls the callback
		143	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		144	abort. The default is to print a diagnostic message and to call C<abort
		145	()>.
		146
		147	When libev detects a usage error such as a negative timer interval, then
		148	it will print a diagnostic message and abort (via the C<assert> mechanism,
		149	so C<NDEBUG> will disable this checking): these are programming errors in
		150	the libev caller and need to be fixed there.
		151
		152	Libev also has a few internal error-checking C<assert>ions, and also has
		153	extensive consistency checking code. These do not trigger under normal
		154	circumstances, as they indicate either a bug in libev or worse.
		155
51		156
52	=head1 GLOBAL FUNCTIONS	157	=head1 GLOBAL FUNCTIONS
53		158
54	These functions can be called anytime, even before initialising the	159	These functions can be called anytime, even before initialising the
55	library in any way.	160	library in any way.
…		…
60		165
61	Returns the current time as libev would use it. Please note that the	166	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	167	C<ev_now> function is usually faster and also often returns the timestamp
63	you actually want to know.	168	you actually want to know.
64		169
		170	=item ev_sleep (ev_tstamp interval)
		171
		172	Sleep for the given interval: The current thread will be blocked until
		173	either it is interrupted or the given time interval has passed. Basically
		174	this is a sub-second-resolution C<sleep ()>.
		175
65	=item int ev_version_major ()	176	=item int ev_version_major ()
66		177
67	=item int ev_version_minor ()	178	=item int ev_version_minor ()
68		179
69	You can find out the major and minor version numbers of the library	180	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	181	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	182	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	183	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
73	version of the library your program was compiled against.	184	version of the library your program was compiled against.
74		185
		186	These version numbers refer to the ABI version of the library, not the
		187	release version.
		188
75	Usually, it's a good idea to terminate if the major versions mismatch,	189	Usually, it's a good idea to terminate if the major versions mismatch,
76	as this indicates an incompatible change. Minor versions are usually	190	as this indicates an incompatible change. Minor versions are usually
77	compatible to older versions, so a larger minor version alone is usually	191	compatible to older versions, so a larger minor version alone is usually
78	not a problem.	192	not a problem.
79		193
		194	Example: Make sure we haven't accidentally been linked against the wrong
		195	version.
		196
		197	assert (("libev version mismatch",
		198	ev_version_major () == EV_VERSION_MAJOR
		199	&& ev_version_minor () >= EV_VERSION_MINOR));
		200
		201	=item unsigned int ev_supported_backends ()
		202
		203	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
		204	value) compiled into this binary of libev (independent of their
		205	availability on the system you are running on). See C<ev_default_loop> for
		206	a description of the set values.
		207
		208	Example: make sure we have the epoll method, because yeah this is cool and
		209	a must have and can we have a torrent of it please!!!11
		210
		211	assert (("sorry, no epoll, no sex",
		212	ev_supported_backends () & EVBACKEND_EPOLL));
		213
		214	=item unsigned int ev_recommended_backends ()
		215
		216	Return the set of all backends compiled into this binary of libev and also
		217	recommended for this platform. This set is often smaller than the one
		218	returned by C<ev_supported_backends>, as for example kqueue is broken on
		219	most BSDs and will not be auto-detected unless you explicitly request it
		220	(assuming you know what you are doing). This is the set of backends that
		221	libev will probe for if you specify no backends explicitly.
		222
		223	=item unsigned int ev_embeddable_backends ()
		224
		225	Returns the set of backends that are embeddable in other event loops. This
		226	is the theoretical, all-platform, value. To find which backends
		227	might be supported on the current system, you would need to look at
		228	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
		229	recommended ones.
		230
		231	See the description of C<ev_embed> watchers for more info.
		232
80	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	233	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
81		234
82	Sets the allocation function to use (the prototype is similar to the	235	Sets the allocation function to use (the prototype is similar - the
83	realloc C function, the semantics are identical). It is used to allocate	236	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	237	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	238	when memory needs to be allocated (C<size != 0>), the library might abort
86	destructive action. The default is your system realloc function.	239	or take some potentially destructive action.
		240
		241	Since some systems (at least OpenBSD and Darwin) fail to implement
		242	correct C<realloc> semantics, libev will use a wrapper around the system
		243	C<realloc> and C<free> functions by default.
87		244
88	You could override this function in high-availability programs to, say,	245	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,	246	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.	247	or even to sleep a while and retry until some memory is available.
91		248
		249	Example: Replace the libev allocator with one that waits a bit and then
		250	retries (example requires a standards-compliant C<realloc>).
		251
		252	static void *
		253	persistent_realloc (void *ptr, size_t size)
		254	{
		255	for (;;)
		256	{
		257	void *newptr = realloc (ptr, size);
		258
		259	if (newptr)
		260	return newptr;
		261
		262	sleep (60);
		263	}
		264	}
		265
		266	...
		267	ev_set_allocator (persistent_realloc);
		268
92	=item ev_set_syserr_cb (void (*cb)(const char *msg));	269	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
93		270
94	Set the callback function to call on a retryable syscall error (such	271	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	272	as failed select, poll, epoll_wait). The message is a printable string
96	indicating the system call or subsystem causing the problem. If this	273	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	274	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	275	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	276	requested operation, or, if the condition doesn't go away, do bad stuff
100	(such as abort).	277	(such as abort).
101		278
		279	Example: This is basically the same thing that libev does internally, too.
		280
		281	static void
		282	fatal_error (const char *msg)
		283	{
		284	perror (msg);
		285	abort ();
		286	}
		287
		288	...
		289	ev_set_syserr_cb (fatal_error);
		290
102	=back	291	=back
103		292
104	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	293	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
105		294
106	An event loop is described by a C<struct ev_loop *>. The library knows two	295	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	296	is I<not> optional in this case, as there is also an C<ev_loop>
108	events, and dynamically created loops which do not.	297	I<function>).
109		298
110	If you use threads, a common model is to run the default event loop	299	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	300	supports signals and child events, and dynamically created loops which do
112	create, you also create another event loop. Libev itself does no locking	301	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		302
117	=over 4	303	=over 4
118		304
119	=item struct ev_loop *ev_default_loop (unsigned int flags)	305	=item struct ev_loop *ev_default_loop (unsigned int flags)
120		306
121	This will initialise the default event loop if it hasn't been initialised	307	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	308	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	309	false. If it already was initialised it simply returns it (and ignores the
124	flags).	310	flags. If that is troubling you, check C<ev_backend ()> afterwards).
125		311
126	If you don't know what event loop to use, use the one returned from this	312	If you don't know what event loop to use, use the one returned from this
127	function.	313	function.
128		314
		315	Note that this function is I<not> thread-safe, so if you want to use it
		316	from multiple threads, you have to lock (note also that this is unlikely,
		317	as loops cannot be shared easily between threads anyway).
		318
		319	The default loop is the only loop that can handle C<ev_signal> and
		320	C<ev_child> watchers, and to do this, it always registers a handler
		321	for C<SIGCHLD>. If this is a problem for your application you can either
		322	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
		323	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
		324	C<ev_default_init>.
		325
129	The flags argument can be used to specify special behaviour or specific	326	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).	327	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
131		328
132	It supports the following flags:	329	The following flags are supported:
133		330
134	=over 4	331	=over 4
135		332
136	=item C<EVFLAG_AUTO>	333	=item C<EVFLAG_AUTO>
137		334
138	The default flags value. Use this if you have no clue (it's the right	335	The default flags value. Use this if you have no clue (it's the right
139	thing, believe me).	336	thing, believe me).
140		337
141	=item C<EVFLAG_NOENV>	338	=item C<EVFLAG_NOENV>
142		339
143	If this flag bit is ored into the flag value (or the program runs setuid	340	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	341	or setgid) then libev will I<not> look at the environment variable
145	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	342	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
146	override the flags completely if it is found in the environment. This is	343	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	344	useful to try out specific backends to test their performance, or to work
148	around bugs.	345	around bugs.
149		346
		347	=item C<EVFLAG_FORKCHECK>
		348
		349	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after
		350	a fork, you can also make libev check for a fork in each iteration by
		351	enabling this flag.
		352
		353	This works by calling C<getpid ()> on every iteration of the loop,
		354	and thus this might slow down your event loop if you do a lot of loop
		355	iterations and little real work, but is usually not noticeable (on my
		356	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		357	without a system call and thus I<very> fast, but my GNU/Linux system also has
		358	C<pthread_atfork> which is even faster).
		359
		360	The big advantage of this flag is that you can forget about fork (and
		361	forget about forgetting to tell libev about forking) when you use this
		362	flag.
		363
		364	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		365	environment variable.
		366
150	=item C<EVMETHOD_SELECT> (value 1, portable select backend)	367	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
151		368
152	This is your standard select(2) backend. Not I<completely> standard, as	369	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,	370	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	371	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	372	using this backend. It doesn't scale too well (O(highest_fd)), but its
156	the fastest backend for a low number of fds.	373	usually the fastest backend for a low number of (low-numbered :) fds.
157		374
		375	To get good performance out of this backend you need a high amount of
		376	parallelism (most of the file descriptors should be busy). If you are
		377	writing a server, you should C<accept ()> in a loop to accept as many
		378	connections as possible during one iteration. You might also want to have
		379	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		380	readiness notifications you get per iteration.
		381
		382	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		383	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		384	C<exceptfds> set on that platform).
		385
158	=item C<EVMETHOD_POLL> (value 2, poll backend, available everywhere except on windows)	386	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
159		387
160	And this is your standard poll(2) backend. It's more complicated than	388	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	389	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	390	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).	391	considerably with a lot of inactive fds). It scales similarly to select,
		392	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		393	performance tips.
164		394
		395	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		396	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		397
165	=item C<EVMETHOD_EPOLL> (value 4, Linux)	398	=item C<EVBACKEND_EPOLL> (value 4, Linux)
166		399
167	For few fds, this backend is a bit little slower than poll and select,	400	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	401	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	402	like O(total_fds) where n is the total number of fds (or the highest fd),
170	either O(1) or O(active_fds).	403	epoll scales either O(1) or O(active_fds).
171		404
		405	The epoll mechanism deserves honorable mention as the most misdesigned
		406	of the more advanced event mechanisms: mere annoyances include silently
		407	dropping file descriptors, requiring a system call per change per file
		408	descriptor (and unnecessary guessing of parameters), problems with dup and
		409	so on. The biggest issue is fork races, however - if a program forks then
		410	I<both> parent and child process have to recreate the epoll set, which can
		411	take considerable time (one syscall per file descriptor) and is of course
		412	hard to detect.
		413
		414	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		415	of course I<doesn't>, and epoll just loves to report events for totally
		416	I<different> file descriptors (even already closed ones, so one cannot
		417	even remove them from the set) than registered in the set (especially
		418	on SMP systems). Libev tries to counter these spurious notifications by
		419	employing an additional generation counter and comparing that against the
		420	events to filter out spurious ones, recreating the set when required.
		421
172	While stopping and starting an I/O watcher in the same iteration will	422	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	423	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	424	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	425	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
176	well if you register events for both fds.	426	file descriptors might not work very well if you register events for both
		427	file descriptors.
177		428
		429	Best performance from this backend is achieved by not unregistering all
		430	watchers for a file descriptor until it has been closed, if possible,
		431	i.e. keep at least one watcher active per fd at all times. Stopping and
		432	starting a watcher (without re-setting it) also usually doesn't cause
		433	extra overhead. A fork can both result in spurious notifications as well
		434	as in libev having to destroy and recreate the epoll object, which can
		435	take considerable time and thus should be avoided.
		436
		437	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
		438	faster than epoll for maybe up to a hundred file descriptors, depending on
		439	the usage. So sad.
		440
		441	While nominally embeddable in other event loops, this feature is broken in
		442	all kernel versions tested so far.
		443
		444	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		445	C<EVBACKEND_POLL>.
		446
178	=item C<EVMETHOD_KQUEUE> (value 8, most BSD clones)	447	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
179		448
180	Kqueue deserves special mention, as at the time of this writing, it	449	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	450	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	451	with anything but sockets and pipes, except on Darwin, where of course
183	completely useless). For this reason its not being "autodetected" unless	452	it's completely useless). Unlike epoll, however, whose brokenness
184	you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO).	453	is by design, these kqueue bugs can (and eventually will) be fixed
		454	without API changes to existing programs. For this reason it's not being
		455	"auto-detected" unless you explicitly specify it in the flags (i.e. using
		456	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		457	system like NetBSD.
		458
		459	You still can embed kqueue into a normal poll or select backend and use it
		460	only for sockets (after having made sure that sockets work with kqueue on
		461	the target platform). See C<ev_embed> watchers for more info.
185		462
186	It scales in the same way as the epoll backend, but the interface to the	463	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	464	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	465	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	466	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
190	incident, so its best to avoid that.	467	two event changes per incident. Support for C<fork ()> is very bad (but
		468	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		469	cases
191		470
		471	This backend usually performs well under most conditions.
		472
		473	While nominally embeddable in other event loops, this doesn't work
		474	everywhere, so you might need to test for this. And since it is broken
		475	almost everywhere, you should only use it when you have a lot of sockets
		476	(for which it usually works), by embedding it into another event loop
		477	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> (but C<poll> is of course
		478	also broken on OS X)) and, did I mention it, using it only for sockets.
		479
		480	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		481	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		482	C<NOTE_EOF>.
		483
192	=item C<EVMETHOD_DEVPOLL> (value 16, Solaris 8)	484	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
193		485
194	This is not implemented yet (and might never be).	486	This is not implemented yet (and might never be, unless you send me an
		487	implementation). According to reports, C</dev/poll> only supports sockets
		488	and is not embeddable, which would limit the usefulness of this backend
		489	immensely.
195		490
196	=item C<EVMETHOD_PORT> (value 32, Solaris 10)	491	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
197		492
198	This uses the Solaris 10 port mechanism. As with everything on Solaris,	493	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)).	494	it's really slow, but it still scales very well (O(active_fds)).
200		495
		496	Please note that Solaris event ports can deliver a lot of spurious
		497	notifications, so you need to use non-blocking I/O or other means to avoid
		498	blocking when no data (or space) is available.
		499
		500	While this backend scales well, it requires one system call per active
		501	file descriptor per loop iteration. For small and medium numbers of file
		502	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		503	might perform better.
		504
		505	On the positive side, with the exception of the spurious readiness
		506	notifications, this backend actually performed fully to specification
		507	in all tests and is fully embeddable, which is a rare feat among the
		508	OS-specific backends (I vastly prefer correctness over speed hacks).
		509
		510	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		511	C<EVBACKEND_POLL>.
		512
201	=item C<EVMETHOD_ALL>	513	=item C<EVBACKEND_ALL>
202		514
203	Try all backends (even potentially broken ones that wouldn't be tried	515	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	516	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
205	C<EVMETHOD_ALL & ~EVMETHOD_KQUEUE>.	517	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		518
		519	It is definitely not recommended to use this flag.
206		520
207	=back	521	=back
208		522
209	If one or more of these are ored into the flags value, then only these	523	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	524	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	525	specified, all backends in C<ev_recommended_backends ()> will be tried.
212	order of their flag values :)	526
		527	Example: This is the most typical usage.
		528
		529	if (!ev_default_loop (0))
		530	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		531
		532	Example: Restrict libev to the select and poll backends, and do not allow
		533	environment settings to be taken into account:
		534
		535	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		536
		537	Example: Use whatever libev has to offer, but make sure that kqueue is
		538	used if available (warning, breaks stuff, best use only with your own
		539	private event loop and only if you know the OS supports your types of
		540	fds):
		541
		542	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
213		543
214	=item struct ev_loop *ev_loop_new (unsigned int flags)	544	=item struct ev_loop *ev_loop_new (unsigned int flags)
215		545
216	Similar to C<ev_default_loop>, but always creates a new event loop that is	546	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	547	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	548	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).	549	undefined behaviour (or a failed assertion if assertions are enabled).
220		550
		551	Note that this function I<is> thread-safe, and the recommended way to use
		552	libev with threads is indeed to create one loop per thread, and using the
		553	default loop in the "main" or "initial" thread.
		554
		555	Example: Try to create a event loop that uses epoll and nothing else.
		556
		557	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
		558	if (!epoller)
		559	fatal ("no epoll found here, maybe it hides under your chair");
		560
221	=item ev_default_destroy ()	561	=item ev_default_destroy ()
222		562
223	Destroys the default loop again (frees all memory and kernel state	563	Destroys the default loop again (frees all memory and kernel state
224	etc.). This stops all registered event watchers (by not touching them in	564	etc.). None of the active event watchers will be stopped in the normal
225	any way whatsoever, although you cannot rely on this :).	565	sense, so e.g. C<ev_is_active> might still return true. It is your
		566	responsibility to either stop all watchers cleanly yourself I<before>
		567	calling this function, or cope with the fact afterwards (which is usually
		568	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		569	for example).
		570
		571	Note that certain global state, such as signal state (and installed signal
		572	handlers), will not be freed by this function, and related watchers (such
		573	as signal and child watchers) would need to be stopped manually.
		574
		575	In general it is not advisable to call this function except in the
		576	rare occasion where you really need to free e.g. the signal handling
		577	pipe fds. If you need dynamically allocated loops it is better to use
		578	C<ev_loop_new> and C<ev_loop_destroy>).
226		579
227	=item ev_loop_destroy (loop)	580	=item ev_loop_destroy (loop)
228		581
229	Like C<ev_default_destroy>, but destroys an event loop created by an	582	Like C<ev_default_destroy>, but destroys an event loop created by an
230	earlier call to C<ev_loop_new>.	583	earlier call to C<ev_loop_new>.
231		584
232	=item ev_default_fork ()	585	=item ev_default_fork ()
233		586
		587	This function sets a flag that causes subsequent C<ev_loop> iterations
234	This function reinitialises the kernel state for backends that have	588	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	589	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	590	the child process (or both child and parent, but that again makes little
237	again makes little sense).	591	sense). You I<must> call it in the child before using any of the libev
		592	functions, and it will only take effect at the next C<ev_loop> iteration.
238		593
239	You I<must> call this function in the child process after forking if and	594	On the other hand, you only need to call this function in the child
240	only if you want to use the event library in both processes. If you just	595	process if and only if you want to use the event library in the child. If
241	fork+exec, you don't have to call it.	596	you just fork+exec, you don't have to call it at all.
242		597
243	The function itself is quite fast and it's usually not a problem to call	598	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	599	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>:	600	quite nicely into a call to C<pthread_atfork>:
246		601
…		…
248		603
249	=item ev_loop_fork (loop)	604	=item ev_loop_fork (loop)
250		605
251	Like C<ev_default_fork>, but acts on an event loop created by	606	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	607	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.	608	after fork that you want to re-use in the child, and how you do this is
		609	entirely your own problem.
254		610
		611	=item int ev_is_default_loop (loop)
		612
		613	Returns true when the given loop is, in fact, the default loop, and false
		614	otherwise.
		615
		616	=item unsigned int ev_loop_count (loop)
		617
		618	Returns the count of loop iterations for the loop, which is identical to
		619	the number of times libev did poll for new events. It starts at C<0> and
		620	happily wraps around with enough iterations.
		621
		622	This value can sometimes be useful as a generation counter of sorts (it
		623	"ticks" the number of loop iterations), as it roughly corresponds with
		624	C<ev_prepare> and C<ev_check> calls.
		625
255	=item unsigned int ev_method (loop)	626	=item unsigned int ev_backend (loop)
256		627
257	Returns one of the C<EVMETHOD_*> flags indicating the event backend in	628	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
258	use.	629	use.
259		630
260	=item ev_tstamp ev_now (loop)	631	=item ev_tstamp ev_now (loop)
261		632
262	Returns the current "event loop time", which is the time the event loop	633	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	634	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	635	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	636	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).	637	event occurring (or more correctly, libev finding out about it).
		638
		639	=item ev_now_update (loop)
		640
		641	Establishes the current time by querying the kernel, updating the time
		642	returned by C<ev_now ()> in the progress. This is a costly operation and
		643	is usually done automatically within C<ev_loop ()>.
		644
		645	This function is rarely useful, but when some event callback runs for a
		646	very long time without entering the event loop, updating libev's idea of
		647	the current time is a good idea.
		648
		649	See also "The special problem of time updates" in the C<ev_timer> section.
		650
		651	=item ev_suspend (loop)
		652
		653	=item ev_resume (loop)
		654
		655	These two functions suspend and resume a loop, for use when the loop is
		656	not used for a while and timeouts should not be processed.
		657
		658	A typical use case would be an interactive program such as a game: When
		659	the user presses C<^Z> to suspend the game and resumes it an hour later it
		660	would be best to handle timeouts as if no time had actually passed while
		661	the program was suspended. This can be achieved by calling C<ev_suspend>
		662	in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
		663	C<ev_resume> directly afterwards to resume timer processing.
		664
		665	Effectively, all C<ev_timer> watchers will be delayed by the time spend
		666	between C<ev_suspend> and C<ev_resume>, and all C<ev_periodic> watchers
		667	will be rescheduled (that is, they will lose any events that would have
		668	occured while suspended).
		669
		670	After calling C<ev_suspend> you B<must not> call I<any> function on the
		671	given loop other than C<ev_resume>, and you B<must not> call C<ev_resume>
		672	without a previous call to C<ev_suspend>.
		673
		674	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
		675	event loop time (see C<ev_now_update>).
267		676
268	=item ev_loop (loop, int flags)	677	=item ev_loop (loop, int flags)
269		678
270	Finally, this is it, the event handler. This function usually is called	679	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	680	after you initialised all your watchers and you want to start handling
272	events.	681	events.
273		682
274	If the flags argument is specified as 0, it will not return until either	683	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.	684	either no event watchers are active anymore or C<ev_unloop> was called.
		685
		686	Please note that an explicit C<ev_unloop> is usually better than
		687	relying on all watchers to be stopped when deciding when a program has
		688	finished (especially in interactive programs), but having a program
		689	that automatically loops as long as it has to and no longer by virtue
		690	of relying on its watchers stopping correctly, that is truly a thing of
		691	beauty.
276		692
277	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	693	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	694	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.	695	process in case there are no events and will return after one iteration of
		696	the loop.
280		697
281	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	698	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	699	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	700	will block your process until at least one new event arrives (which could
		701	be an event internal to libev itself, so there is no guarantee that a
		702	user-registered callback will be called), and will return after one
284	one iteration of the loop.	703	iteration of the loop.
285		704
286	This flags value could be used to implement alternative looping	705	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	706	with something not expressible using other libev watchers (i.e. "roll your
288	more generic mechanism.	707	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
		708	usually a better approach for this kind of thing.
289		709
290	Here are the gory details of what ev_loop does:	710	Here are the gory details of what C<ev_loop> does:
291		711
292	1. If there are no active watchers (reference count is zero), return.	712	- Before the first iteration, call any pending watchers.
		713	* If EVFLAG_FORKCHECK was used, check for a fork.
		714	- If a fork was detected (by any means), queue and call all fork watchers.
293	2. Queue and immediately call all prepare watchers.	715	- Queue and call all prepare watchers.
294	3. If we have been forked, recreate the kernel state.	716	- If we have been forked, detach and recreate the kernel state
		717	as to not disturb the other process.
295	4. Update the kernel state with all outstanding changes.	718	- Update the kernel state with all outstanding changes.
296	5. Update the "event loop time".	719	- Update the "event loop time" (ev_now ()).
297	6. Calculate for how long to block.	720	- Calculate for how long to sleep or block, if at all
		721	(active idle watchers, EVLOOP_NONBLOCK or not having
		722	any active watchers at all will result in not sleeping).
		723	- Sleep if the I/O and timer collect interval say so.
298	7. Block the process, waiting for events.	724	- Block the process, waiting for any events.
		725	- Queue all outstanding I/O (fd) events.
299	8. Update the "event loop time" and do time jump handling.	726	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
300	9. Queue all outstanding timers.	727	- Queue all expired timers.
301	10. Queue all outstanding periodics.	728	- Queue all expired periodics.
302	11. If no events are pending now, queue all idle watchers.	729	- Unless any events are pending now, queue all idle watchers.
303	12. Queue all check watchers.	730	- Queue all check watchers.
304	13. Call all queued watchers in reverse order (i.e. check watchers first).	731	- Call all queued watchers in reverse order (i.e. check watchers first).
		732	Signals and child watchers are implemented as I/O watchers, and will
		733	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	734	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
306	was used, return, otherwise continue with step #1.	735	were used, or there are no active watchers, return, otherwise
		736	continue with step *.
		737
		738	Example: Queue some jobs and then loop until no events are outstanding
		739	anymore.
		740
		741	... queue jobs here, make sure they register event watchers as long
		742	... as they still have work to do (even an idle watcher will do..)
		743	ev_loop (my_loop, 0);
		744	... jobs done or somebody called unloop. yeah!
307		745
308	=item ev_unloop (loop, how)	746	=item ev_unloop (loop, how)
309		747
310	Can be used to make a call to C<ev_loop> return early (but only after it	748	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	749	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	750	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.	751	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
314		752
		753	This "unloop state" will be cleared when entering C<ev_loop> again.
		754
		755	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
		756
315	=item ev_ref (loop)	757	=item ev_ref (loop)
316		758
317	=item ev_unref (loop)	759	=item ev_unref (loop)
318		760
319	Ref/unref can be used to add or remove a reference count on the event	761	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	762	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	763	count is nonzero, C<ev_loop> will not return on its own.
		764
322	a watcher you never unregister that should not keep C<ev_loop> from	765	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	766	from returning, call ev_unref() after starting, and ev_ref() before
		767	stopping it.
		768
324	example, libev itself uses this for its internal signal pipe: It is not	769	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	770	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	771	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	772	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>.	773	third-party libraries. Just remember to I<unref after start> and I<ref
		774	before stop> (but only if the watcher wasn't active before, or was active
		775	before, respectively. Note also that libev might stop watchers itself
		776	(e.g. non-repeating timers) in which case you have to C<ev_ref>
		777	in the callback).
		778
		779	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
		780	running when nothing else is active.
		781
		782	ev_signal exitsig;
		783	ev_signal_init (&exitsig, sig_cb, SIGINT);
		784	ev_signal_start (loop, &exitsig);
		785	evf_unref (loop);
		786
		787	Example: For some weird reason, unregister the above signal handler again.
		788
		789	ev_ref (loop);
		790	ev_signal_stop (loop, &exitsig);
		791
		792	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		793
		794	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		795
		796	These advanced functions influence the time that libev will spend waiting
		797	for events. Both time intervals are by default C<0>, meaning that libev
		798	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		799	latency.
		800
		801	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		802	allows libev to delay invocation of I/O and timer/periodic callbacks
		803	to increase efficiency of loop iterations (or to increase power-saving
		804	opportunities).
		805
		806	The idea is that sometimes your program runs just fast enough to handle
		807	one (or very few) event(s) per loop iteration. While this makes the
		808	program responsive, it also wastes a lot of CPU time to poll for new
		809	events, especially with backends like C<select ()> which have a high
		810	overhead for the actual polling but can deliver many events at once.
		811
		812	By setting a higher I<io collect interval> you allow libev to spend more
		813	time collecting I/O events, so you can handle more events per iteration,
		814	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		815	C<ev_timer>) will be not affected. Setting this to a non-null value will
		816	introduce an additional C<ev_sleep ()> call into most loop iterations.
		817
		818	Likewise, by setting a higher I<timeout collect interval> you allow libev
		819	to spend more time collecting timeouts, at the expense of increased
		820	latency/jitter/inexactness (the watcher callback will be called
		821	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		822	value will not introduce any overhead in libev.
		823
		824	Many (busy) programs can usually benefit by setting the I/O collect
		825	interval to a value near C<0.1> or so, which is often enough for
		826	interactive servers (of course not for games), likewise for timeouts. It
		827	usually doesn't make much sense to set it to a lower value than C<0.01>,
		828	as this approaches the timing granularity of most systems.
		829
		830	Setting the I<timeout collect interval> can improve the opportunity for
		831	saving power, as the program will "bundle" timer callback invocations that
		832	are "near" in time together, by delaying some, thus reducing the number of
		833	times the process sleeps and wakes up again. Another useful technique to
		834	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		835	they fire on, say, one-second boundaries only.
		836
		837	=item ev_loop_verify (loop)
		838
		839	This function only does something when C<EV_VERIFY> support has been
		840	compiled in, which is the default for non-minimal builds. It tries to go
		841	through all internal structures and checks them for validity. If anything
		842	is found to be inconsistent, it will print an error message to standard
		843	error and call C<abort ()>.
		844
		845	This can be used to catch bugs inside libev itself: under normal
		846	circumstances, this function will never abort as of course libev keeps its
		847	data structures consistent.
329		848
330	=back	849	=back
331		850
		851
332	=head1 ANATOMY OF A WATCHER	852	=head1 ANATOMY OF A WATCHER
		853
		854	In the following description, uppercase C<TYPE> in names stands for the
		855	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		856	watchers and C<ev_io_start> for I/O watchers.
333		857
334	A watcher is a structure that you create and register to record your	858	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	859	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:	860	become readable, you would create an C<ev_io> watcher for that:
337		861
338	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	862	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
339	{	863	{
340	ev_io_stop (w);	864	ev_io_stop (w);
341	ev_unloop (loop, EVUNLOOP_ALL);	865	ev_unloop (loop, EVUNLOOP_ALL);
342	}	866	}
343		867
344	struct ev_loop *loop = ev_default_loop (0);	868	struct ev_loop *loop = ev_default_loop (0);
		869
345	struct ev_io stdin_watcher;	870	ev_io stdin_watcher;
		871
346	ev_init (&stdin_watcher, my_cb);	872	ev_init (&stdin_watcher, my_cb);
347	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	873	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
348	ev_io_start (loop, &stdin_watcher);	874	ev_io_start (loop, &stdin_watcher);
		875
349	ev_loop (loop, 0);	876	ev_loop (loop, 0);
350		877
351	As you can see, you are responsible for allocating the memory for your	878	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,	879	watcher structures (and it is I<usually> a bad idea to do this on the
353	although this can sometimes be quite valid).	880	stack).
		881
		882	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		883	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
354		884
355	Each watcher structure must be initialised by a call to C<ev_init	885	Each watcher structure must be initialised by a call to C<ev_init
356	(watcher *, callback)>, which expects a callback to be provided. This	886	(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	887	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	888	watchers, each time the event loop detects that the file descriptor given
359	is readable and/or writable).	889	is readable and/or writable).
360		890
361	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	891	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	892	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	893	is also a macro to combine initialisation and setting in one call: C<<
364	(watcher *, callback, ...) >>.	894	ev_TYPE_init (watcher *, callback, ...) >>.
365		895
366	To make the watcher actually watch out for events, you have to start it	896	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	897	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	898	*) >>), and you can stop watching for events at any time by calling the
369	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	899	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
370		900
371	As long as your watcher is active (has been started but not stopped) you	901	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	902	must not touch the values stored in it. Most specifically you must never
373	reinitialise it or call its set method.	903	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		904
380	Each and every callback receives the event loop pointer as first, the	905	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	906	registered watcher structure as second, and a bitset of received events as
382	third argument.	907	third argument.
383		908
…		…
407	The signal specified in the C<ev_signal> watcher has been received by a thread.	932	The signal specified in the C<ev_signal> watcher has been received by a thread.
408		933
409	=item C<EV_CHILD>	934	=item C<EV_CHILD>
410		935
411	The pid specified in the C<ev_child> watcher has received a status change.	936	The pid specified in the C<ev_child> watcher has received a status change.
		937
		938	=item C<EV_STAT>
		939
		940	The path specified in the C<ev_stat> watcher changed its attributes somehow.
412		941
413	=item C<EV_IDLE>	942	=item C<EV_IDLE>
414		943
415	The C<ev_idle> watcher has determined that you have nothing better to do.	944	The C<ev_idle> watcher has determined that you have nothing better to do.
416		945
…		…
424	received events. Callbacks of both watcher types can start and stop as	953	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	954	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	955	(for example, a C<ev_prepare> watcher might start an idle watcher to keep
427	C<ev_loop> from blocking).	956	C<ev_loop> from blocking).
428		957
		958	=item C<EV_EMBED>
		959
		960	The embedded event loop specified in the C<ev_embed> watcher needs attention.
		961
		962	=item C<EV_FORK>
		963
		964	The event loop has been resumed in the child process after fork (see
		965	C<ev_fork>).
		966
		967	=item C<EV_ASYNC>
		968
		969	The given async watcher has been asynchronously notified (see C<ev_async>).
		970
		971	=item C<EV_CUSTOM>
		972
		973	Not ever sent (or otherwise used) by libev itself, but can be freely used
		974	by libev users to signal watchers (e.g. via C<ev_feed_event>).
		975
429	=item C<EV_ERROR>	976	=item C<EV_ERROR>
430		977
431	An unspecified error has occured, the watcher has been stopped. This might	978	An unspecified error has occurred, the watcher has been stopped. This might
432	happen because the watcher could not be properly started because libev	979	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	980	ran out of memory, a file descriptor was found to be closed or any other
		981	problem. Libev considers these application bugs.
		982
434	problem. You best act on it by reporting the problem and somehow coping	983	You best act on it by reporting the problem and somehow coping with the
435	with the watcher being stopped.	984	watcher being stopped. Note that well-written programs should not receive
		985	an error ever, so when your watcher receives it, this usually indicates a
		986	bug in your program.
436		987
437	Libev will usually signal a few "dummy" events together with an error,	988	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	989	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	990	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	991	the error from read() or write(). This will not work in multi-threaded
441	programs, though, so beware.	992	programs, though, as the fd could already be closed and reused for another
		993	thing, so beware.
442		994
443	=back	995	=back
444		996
		997	=head2 GENERIC WATCHER FUNCTIONS
		998
		999	=over 4
		1000
		1001	=item C<ev_init> (ev_TYPE *watcher, callback)
		1002
		1003	This macro initialises the generic portion of a watcher. The contents
		1004	of the watcher object can be arbitrary (so C<malloc> will do). Only
		1005	the generic parts of the watcher are initialised, you I<need> to call
		1006	the type-specific C<ev_TYPE_set> macro afterwards to initialise the
		1007	type-specific parts. For each type there is also a C<ev_TYPE_init> macro
		1008	which rolls both calls into one.
		1009
		1010	You can reinitialise a watcher at any time as long as it has been stopped
		1011	(or never started) and there are no pending events outstanding.
		1012
		1013	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
		1014	int revents)>.
		1015
		1016	Example: Initialise an C<ev_io> watcher in two steps.
		1017
		1018	ev_io w;
		1019	ev_init (&w, my_cb);
		1020	ev_io_set (&w, STDIN_FILENO, EV_READ);
		1021
		1022	=item C<ev_TYPE_set> (ev_TYPE *, [args])
		1023
		1024	This macro initialises the type-specific parts of a watcher. You need to
		1025	call C<ev_init> at least once before you call this macro, but you can
		1026	call C<ev_TYPE_set> any number of times. You must not, however, call this
		1027	macro on a watcher that is active (it can be pending, however, which is a
		1028	difference to the C<ev_init> macro).
		1029
		1030	Although some watcher types do not have type-specific arguments
		1031	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
		1032
		1033	See C<ev_init>, above, for an example.
		1034
		1035	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
		1036
		1037	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
		1038	calls into a single call. This is the most convenient method to initialise
		1039	a watcher. The same limitations apply, of course.
		1040
		1041	Example: Initialise and set an C<ev_io> watcher in one step.
		1042
		1043	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
		1044
		1045	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)
		1046
		1047	Starts (activates) the given watcher. Only active watchers will receive
		1048	events. If the watcher is already active nothing will happen.
		1049
		1050	Example: Start the C<ev_io> watcher that is being abused as example in this
		1051	whole section.
		1052
		1053	ev_io_start (EV_DEFAULT_UC, &w);
		1054
		1055	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)
		1056
		1057	Stops the given watcher if active, and clears the pending status (whether
		1058	the watcher was active or not).
		1059
		1060	It is possible that stopped watchers are pending - for example,
		1061	non-repeating timers are being stopped when they become pending - but
		1062	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
		1063	pending. If you want to free or reuse the memory used by the watcher it is
		1064	therefore a good idea to always call its C<ev_TYPE_stop> function.
		1065
		1066	=item bool ev_is_active (ev_TYPE *watcher)
		1067
		1068	Returns a true value iff the watcher is active (i.e. it has been started
		1069	and not yet been stopped). As long as a watcher is active you must not modify
		1070	it.
		1071
		1072	=item bool ev_is_pending (ev_TYPE *watcher)
		1073
		1074	Returns a true value iff the watcher is pending, (i.e. it has outstanding
		1075	events but its callback has not yet been invoked). As long as a watcher
		1076	is pending (but not active) you must not call an init function on it (but
		1077	C<ev_TYPE_set> is safe), you must not change its priority, and you must
		1078	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1079	it).
		1080
		1081	=item callback ev_cb (ev_TYPE *watcher)
		1082
		1083	Returns the callback currently set on the watcher.
		1084
		1085	=item ev_cb_set (ev_TYPE *watcher, callback)
		1086
		1087	Change the callback. You can change the callback at virtually any time
		1088	(modulo threads).
		1089
		1090	=item ev_set_priority (ev_TYPE *watcher, priority)
		1091
		1092	=item int ev_priority (ev_TYPE *watcher)
		1093
		1094	Set and query the priority of the watcher. The priority is a small
		1095	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1096	(default: C<-2>). Pending watchers with higher priority will be invoked
		1097	before watchers with lower priority, but priority will not keep watchers
		1098	from being executed (except for C<ev_idle> watchers).
		1099
		1100	If you need to suppress invocation when higher priority events are pending
		1101	you need to look at C<ev_idle> watchers, which provide this functionality.
		1102
		1103	You I<must not> change the priority of a watcher as long as it is active or
		1104	pending.
		1105
		1106	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1107	fine, as long as you do not mind that the priority value you query might
		1108	or might not have been clamped to the valid range.
		1109
		1110	The default priority used by watchers when no priority has been set is
		1111	always C<0>, which is supposed to not be too high and not be too low :).
		1112
		1113	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
		1114	priorities.
		1115
		1116	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1117
		1118	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1119	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1120	can deal with that fact, as both are simply passed through to the
		1121	callback.
		1122
		1123	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1124
		1125	If the watcher is pending, this function clears its pending status and
		1126	returns its C<revents> bitset (as if its callback was invoked). If the
		1127	watcher isn't pending it does nothing and returns C<0>.
		1128
		1129	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1130	callback to be invoked, which can be accomplished with this function.
		1131
		1132	=back
		1133
		1134
445	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1135	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
446		1136
447	Each watcher has, by default, a member C<void *data> that you can change	1137	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	1138	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	1139	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	1140	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	1141	member, you can also "subclass" the watcher type and provide your own
452	data:	1142	data:
453		1143
454	struct my_io	1144	struct my_io
455	{	1145	{
456	struct ev_io io;	1146	ev_io io;
457	int otherfd;	1147	int otherfd;
458	void *somedata;	1148	void *somedata;
459	struct whatever *mostinteresting;	1149	struct whatever *mostinteresting;
460	}	1150	};
		1151
		1152	...
		1153	struct my_io w;
		1154	ev_io_init (&w.io, my_cb, fd, EV_READ);
461		1155
462	And since your callback will be called with a pointer to the watcher, you	1156	And since your callback will be called with a pointer to the watcher, you
463	can cast it back to your own type:	1157	can cast it back to your own type:
464		1158
465	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1159	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
466	{	1160	{
467	struct my_io *w = (struct my_io *)w_;	1161	struct my_io *w = (struct my_io *)w_;
468	...	1162	...
469	}	1163	}
470		1164
471	More interesting and less C-conformant ways of catsing your callback type	1165	More interesting and less C-conformant ways of casting your callback type
472	have been omitted....	1166	instead have been omitted.
		1167
		1168	Another common scenario is to use some data structure with multiple
		1169	embedded watchers:
		1170
		1171	struct my_biggy
		1172	{
		1173	int some_data;
		1174	ev_timer t1;
		1175	ev_timer t2;
		1176	}
		1177
		1178	In this case getting the pointer to C<my_biggy> is a bit more
		1179	complicated: Either you store the address of your C<my_biggy> struct
		1180	in the C<data> member of the watcher (for woozies), or you need to use
		1181	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1182	programmers):
		1183
		1184	#include <stddef.h>
		1185
		1186	static void
		1187	t1_cb (EV_P_ ev_timer *w, int revents)
		1188	{
		1189	struct my_biggy big = (struct my_biggy *
		1190	(((char *)w) - offsetof (struct my_biggy, t1));
		1191	}
		1192
		1193	static void
		1194	t2_cb (EV_P_ ev_timer *w, int revents)
		1195	{
		1196	struct my_biggy big = (struct my_biggy *
		1197	(((char *)w) - offsetof (struct my_biggy, t2));
		1198	}
		1199
		1200	=head2 WATCHER PRIORITY MODELS
		1201
		1202	Many event loops support I<watcher priorities>, which are usually small
		1203	integers that influence the ordering of event callback invocation
		1204	between watchers in some way, all else being equal.
		1205
		1206	In libev, Watcher priorities can be set using C<ev_set_priority>. See its
		1207	description for the more technical details such as the actual priority
		1208	range.
		1209
		1210	There are two common ways how these these priorities are being interpreted
		1211	by event loops:
		1212
		1213	In the more common lock-out model, higher priorities "lock out" invocation
		1214	of lower priority watchers, which means as long as higher priority
		1215	watchers receive events, lower priority watchers are not being invoked.
		1216
		1217	The less common only-for-ordering model uses priorities solely to order
		1218	callback invocation within a single event loop iteration: Higher priority
		1219	watchers are invoked before lower priority ones, but they all get invoked
		1220	before polling for new events.
		1221
		1222	Libev uses the second (only-for-ordering) model for all its watchers
		1223	except for idle watchers (which use the lock-out model).
		1224
		1225	The rationale behind this is that implementing the lock-out model for
		1226	watchers is not well supported by most kernel interfaces, and most event
		1227	libraries will just poll for the same events again and again as long as
		1228	their callbacks have not been executed, which is very inefficient in the
		1229	common case of one high-priority watcher locking out a mass of lower
		1230	priority ones.
		1231
		1232	Static (ordering) priorities are most useful when you have two or more
		1233	watchers handling the same resource: a typical usage example is having an
		1234	C<ev_io> watcher to receive data, and an associated C<ev_timer> to handle
		1235	timeouts. Under load, data might be received while the program handles
		1236	other jobs, but since timers normally get invoked first, the timeout
		1237	handler will be executed before checking for data. In that case, giving
		1238	the timer a lower priority than the I/O watcher ensures that I/O will be
		1239	handled first even under adverse conditions (which is usually, but not
		1240	always, what you want).
		1241
		1242	Since idle watchers use the "lock-out" model, meaning that idle watchers
		1243	will only be executed when no same or higher priority watchers have
		1244	received events, they can be used to implement the "lock-out" model when
		1245	required.
		1246
		1247	For example, to emulate how many other event libraries handle priorities,
		1248	you can associate an C<ev_idle> watcher to each such watcher, and in
		1249	the normal watcher callback, you just start the idle watcher. The real
		1250	processing is done in the idle watcher callback. This causes libev to
		1251	continously poll and process kernel event data for the watcher, but when
		1252	the lock-out case is known to be rare (which in turn is rare :), this is
		1253	workable.
		1254
		1255	Usually, however, the lock-out model implemented that way will perform
		1256	miserably under the type of load it was designed to handle. In that case,
		1257	it might be preferable to stop the real watcher before starting the
		1258	idle watcher, so the kernel will not have to process the event in case
		1259	the actual processing will be delayed for considerable time.
		1260
		1261	Here is an example of an I/O watcher that should run at a strictly lower
		1262	priority than the default, and which should only process data when no
		1263	other events are pending:
		1264
		1265	ev_idle idle; // actual processing watcher
		1266	ev_io io; // actual event watcher
		1267
		1268	static void
		1269	io_cb (EV_P_ ev_io *w, int revents)
		1270	{
		1271	// stop the I/O watcher, we received the event, but
		1272	// are not yet ready to handle it.
		1273	ev_io_stop (EV_A_ w);
		1274
		1275	// start the idle watcher to ahndle the actual event.
		1276	// it will not be executed as long as other watchers
		1277	// with the default priority are receiving events.
		1278	ev_idle_start (EV_A_ &idle);
		1279	}
		1280
		1281	static void
		1282	idle-cb (EV_P_ ev_idle *w, int revents)
		1283	{
		1284	// actual processing
		1285	read (STDIN_FILENO, ...);
		1286
		1287	// have to start the I/O watcher again, as
		1288	// we have handled the event
		1289	ev_io_start (EV_P_ &io);
		1290	}
		1291
		1292	// initialisation
		1293	ev_idle_init (&idle, idle_cb);
		1294	ev_io_init (&io, io_cb, STDIN_FILENO, EV_READ);
		1295	ev_io_start (EV_DEFAULT_ &io);
		1296
		1297	In the "real" world, it might also be beneficial to start a timer, so that
		1298	low-priority connections can not be locked out forever under load. This
		1299	enables your program to keep a lower latency for important connections
		1300	during short periods of high load, while not completely locking out less
		1301	important ones.
473		1302
474		1303
475	=head1 WATCHER TYPES	1304	=head1 WATCHER TYPES
476		1305
477	This section describes each watcher in detail, but will not repeat	1306	This section describes each watcher in detail, but will not repeat
478	information given in the last section.	1307	information given in the last section. Any initialisation/set macros,
		1308	functions and members specific to the watcher type are explained.
479		1309
		1310	Members are additionally marked with either I<[read-only]>, meaning that,
		1311	while the watcher is active, you can look at the member and expect some
		1312	sensible content, but you must not modify it (you can modify it while the
		1313	watcher is stopped to your hearts content), or I<[read-write]>, which
		1314	means you can expect it to have some sensible content while the watcher
		1315	is active, but you can also modify it. Modifying it may not do something
		1316	sensible or take immediate effect (or do anything at all), but libev will
		1317	not crash or malfunction in any way.
		1318
		1319
480	=head2 C<ev_io> - is this file descriptor readable or writable	1320	=head2 C<ev_io> - is this file descriptor readable or writable?
481		1321
482	I/O watchers check whether a file descriptor is readable or writable	1322	I/O watchers check whether a file descriptor is readable or writable
483	in each iteration of the event loop (This behaviour is called	1323	in each iteration of the event loop, or, more precisely, when reading
484	level-triggering because you keep receiving events as long as the	1324	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	1325	some data. This behaviour is called level-triggering because you keep
486	act on the event and neither want to receive future events).	1326	receiving events as long as the condition persists. Remember you can stop
		1327	the watcher if you don't want to act on the event and neither want to
		1328	receive future events.
487		1329
488	In general you can register as many read and/or write event watchers per	1330	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	1331	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	1332	descriptors to non-blocking mode is also usually a good idea (but not
491	required if you know what you are doing).	1333	required if you know what you are doing).
492		1334
493	You have to be careful with dup'ed file descriptors, though. Some backends	1335	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	1336	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	1337	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		1338
499	If you must do this, then force the use of a known-to-be-good backend	1339	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	1340	receive "spurious" readiness notifications, that is your callback might
501	EVMETHOD_POLL).	1341	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
		1342	because there is no data. Not only are some backends known to create a
		1343	lot of those (for example Solaris ports), it is very easy to get into
		1344	this situation even with a relatively standard program structure. Thus
		1345	it is best to always use non-blocking I/O: An extra C<read>(2) returning
		1346	C<EAGAIN> is far preferable to a program hanging until some data arrives.
		1347
		1348	If you cannot run the fd in non-blocking mode (for example you should
		1349	not play around with an Xlib connection), then you have to separately
		1350	re-test whether a file descriptor is really ready with a known-to-be good
		1351	interface such as poll (fortunately in our Xlib example, Xlib already
		1352	does this on its own, so its quite safe to use). Some people additionally
		1353	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1354	indefinitely.
		1355
		1356	But really, best use non-blocking mode.
		1357
		1358	=head3 The special problem of disappearing file descriptors
		1359
		1360	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1361	descriptor (either due to calling C<close> explicitly or any other means,
		1362	such as C<dup2>). The reason is that you register interest in some file
		1363	descriptor, but when it goes away, the operating system will silently drop
		1364	this interest. If another file descriptor with the same number then is
		1365	registered with libev, there is no efficient way to see that this is, in
		1366	fact, a different file descriptor.
		1367
		1368	To avoid having to explicitly tell libev about such cases, libev follows
		1369	the following policy: Each time C<ev_io_set> is being called, libev
		1370	will assume that this is potentially a new file descriptor, otherwise
		1371	it is assumed that the file descriptor stays the same. That means that
		1372	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1373	descriptor even if the file descriptor number itself did not change.
		1374
		1375	This is how one would do it normally anyway, the important point is that
		1376	the libev application should not optimise around libev but should leave
		1377	optimisations to libev.
		1378
		1379	=head3 The special problem of dup'ed file descriptors
		1380
		1381	Some backends (e.g. epoll), cannot register events for file descriptors,
		1382	but only events for the underlying file descriptions. That means when you
		1383	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1384	events for them, only one file descriptor might actually receive events.
		1385
		1386	There is no workaround possible except not registering events
		1387	for potentially C<dup ()>'ed file descriptors, or to resort to
		1388	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1389
		1390	=head3 The special problem of fork
		1391
		1392	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1393	useless behaviour. Libev fully supports fork, but needs to be told about
		1394	it in the child.
		1395
		1396	To support fork in your programs, you either have to call
		1397	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1398	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1399	C<EVBACKEND_POLL>.
		1400
		1401	=head3 The special problem of SIGPIPE
		1402
		1403	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1404	when writing to a pipe whose other end has been closed, your program gets
		1405	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1406	this is sensible behaviour, for daemons, this is usually undesirable.
		1407
		1408	So when you encounter spurious, unexplained daemon exits, make sure you
		1409	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1410	somewhere, as that would have given you a big clue).
		1411
		1412
		1413	=head3 Watcher-Specific Functions
502		1414
503	=over 4	1415	=over 4
504		1416
505	=item ev_io_init (ev_io *, callback, int fd, int events)	1417	=item ev_io_init (ev_io *, callback, int fd, int events)
506		1418
507	=item ev_io_set (ev_io *, int fd, int events)	1419	=item ev_io_set (ev_io *, int fd, int events)
508		1420
509	Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive	1421	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 |	1422	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
511	EV_WRITE> to receive the given events.	1423	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
		1424
		1425	=item int fd [read-only]
		1426
		1427	The file descriptor being watched.
		1428
		1429	=item int events [read-only]
		1430
		1431	The events being watched.
512		1432
513	=back	1433	=back
514		1434
		1435	=head3 Examples
		1436
		1437	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
		1438	readable, but only once. Since it is likely line-buffered, you could
		1439	attempt to read a whole line in the callback.
		1440
		1441	static void
		1442	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
		1443	{
		1444	ev_io_stop (loop, w);
		1445	.. read from stdin here (or from w->fd) and handle any I/O errors
		1446	}
		1447
		1448	...
		1449	struct ev_loop *loop = ev_default_init (0);
		1450	ev_io stdin_readable;
		1451	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
		1452	ev_io_start (loop, &stdin_readable);
		1453	ev_loop (loop, 0);
		1454
		1455
515	=head2 C<ev_timer> - relative and optionally recurring timeouts	1456	=head2 C<ev_timer> - relative and optionally repeating timeouts
516		1457
517	Timer watchers are simple relative timers that generate an event after a	1458	Timer watchers are simple relative timers that generate an event after a
518	given time, and optionally repeating in regular intervals after that.	1459	given time, and optionally repeating in regular intervals after that.
519		1460
520	The timers are based on real time, that is, if you register an event that	1461	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	1462	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	1463	year, it will still time out after (roughly) one hour. "Roughly" because
523	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1464	detecting time jumps is hard, and some inaccuracies are unavoidable (the
524	monotonic clock option helps a lot here).	1465	monotonic clock option helps a lot here).
		1466
		1467	The callback is guaranteed to be invoked only I<after> its timeout has
		1468	passed. If multiple timers become ready during the same loop iteration
		1469	then the ones with earlier time-out values are invoked before ones with
		1470	later time-out values (but this is no longer true when a callback calls
		1471	C<ev_loop> recursively).
		1472
		1473	=head3 Be smart about timeouts
		1474
		1475	Many real-world problems involve some kind of timeout, usually for error
		1476	recovery. A typical example is an HTTP request - if the other side hangs,
		1477	you want to raise some error after a while.
		1478
		1479	What follows are some ways to handle this problem, from obvious and
		1480	inefficient to smart and efficient.
		1481
		1482	In the following, a 60 second activity timeout is assumed - a timeout that
		1483	gets reset to 60 seconds each time there is activity (e.g. each time some
		1484	data or other life sign was received).
		1485
		1486	=over 4
		1487
		1488	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1489
		1490	This is the most obvious, but not the most simple way: In the beginning,
		1491	start the watcher:
		1492
		1493	ev_timer_init (timer, callback, 60., 0.);
		1494	ev_timer_start (loop, timer);
		1495
		1496	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1497	and start it again:
		1498
		1499	ev_timer_stop (loop, timer);
		1500	ev_timer_set (timer, 60., 0.);
		1501	ev_timer_start (loop, timer);
		1502
		1503	This is relatively simple to implement, but means that each time there is
		1504	some activity, libev will first have to remove the timer from its internal
		1505	data structure and then add it again. Libev tries to be fast, but it's
		1506	still not a constant-time operation.
		1507
		1508	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1509
		1510	This is the easiest way, and involves using C<ev_timer_again> instead of
		1511	C<ev_timer_start>.
		1512
		1513	To implement this, configure an C<ev_timer> with a C<repeat> value
		1514	of C<60> and then call C<ev_timer_again> at start and each time you
		1515	successfully read or write some data. If you go into an idle state where
		1516	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1517	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1518
		1519	That means you can ignore both the C<ev_timer_start> function and the
		1520	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1521	member and C<ev_timer_again>.
		1522
		1523	At start:
		1524
		1525	ev_timer_init (timer, callback);
		1526	timer->repeat = 60.;
		1527	ev_timer_again (loop, timer);
		1528
		1529	Each time there is some activity:
		1530
		1531	ev_timer_again (loop, timer);
		1532
		1533	It is even possible to change the time-out on the fly, regardless of
		1534	whether the watcher is active or not:
		1535
		1536	timer->repeat = 30.;
		1537	ev_timer_again (loop, timer);
		1538
		1539	This is slightly more efficient then stopping/starting the timer each time
		1540	you want to modify its timeout value, as libev does not have to completely
		1541	remove and re-insert the timer from/into its internal data structure.
		1542
		1543	It is, however, even simpler than the "obvious" way to do it.
		1544
		1545	=item 3. Let the timer time out, but then re-arm it as required.
		1546
		1547	This method is more tricky, but usually most efficient: Most timeouts are
		1548	relatively long compared to the intervals between other activity - in
		1549	our example, within 60 seconds, there are usually many I/O events with
		1550	associated activity resets.
		1551
		1552	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1553	but remember the time of last activity, and check for a real timeout only
		1554	within the callback:
		1555
		1556	ev_tstamp last_activity; // time of last activity
		1557
		1558	static void
		1559	callback (EV_P_ ev_timer *w, int revents)
		1560	{
		1561	ev_tstamp now = ev_now (EV_A);
		1562	ev_tstamp timeout = last_activity + 60.;
		1563
		1564	// if last_activity + 60. is older than now, we did time out
		1565	if (timeout < now)
		1566	{
		1567	// timeout occured, take action
		1568	}
		1569	else
		1570	{
		1571	// callback was invoked, but there was some activity, re-arm
		1572	// the watcher to fire in last_activity + 60, which is
		1573	// guaranteed to be in the future, so "again" is positive:
		1574	w->repeat = timeout - now;
		1575	ev_timer_again (EV_A_ w);
		1576	}
		1577	}
		1578
		1579	To summarise the callback: first calculate the real timeout (defined
		1580	as "60 seconds after the last activity"), then check if that time has
		1581	been reached, which means something I<did>, in fact, time out. Otherwise
		1582	the callback was invoked too early (C<timeout> is in the future), so
		1583	re-schedule the timer to fire at that future time, to see if maybe we have
		1584	a timeout then.
		1585
		1586	Note how C<ev_timer_again> is used, taking advantage of the
		1587	C<ev_timer_again> optimisation when the timer is already running.
		1588
		1589	This scheme causes more callback invocations (about one every 60 seconds
		1590	minus half the average time between activity), but virtually no calls to
		1591	libev to change the timeout.
		1592
		1593	To start the timer, simply initialise the watcher and set C<last_activity>
		1594	to the current time (meaning we just have some activity :), then call the
		1595	callback, which will "do the right thing" and start the timer:
		1596
		1597	ev_timer_init (timer, callback);
		1598	last_activity = ev_now (loop);
		1599	callback (loop, timer, EV_TIMEOUT);
		1600
		1601	And when there is some activity, simply store the current time in
		1602	C<last_activity>, no libev calls at all:
		1603
		1604	last_actiivty = ev_now (loop);
		1605
		1606	This technique is slightly more complex, but in most cases where the
		1607	time-out is unlikely to be triggered, much more efficient.
		1608
		1609	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1610	callback :) - just change the timeout and invoke the callback, which will
		1611	fix things for you.
		1612
		1613	=item 4. Wee, just use a double-linked list for your timeouts.
		1614
		1615	If there is not one request, but many thousands (millions...), all
		1616	employing some kind of timeout with the same timeout value, then one can
		1617	do even better:
		1618
		1619	When starting the timeout, calculate the timeout value and put the timeout
		1620	at the I<end> of the list.
		1621
		1622	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1623	the list is expected to fire (for example, using the technique #3).
		1624
		1625	When there is some activity, remove the timer from the list, recalculate
		1626	the timeout, append it to the end of the list again, and make sure to
		1627	update the C<ev_timer> if it was taken from the beginning of the list.
		1628
		1629	This way, one can manage an unlimited number of timeouts in O(1) time for
		1630	starting, stopping and updating the timers, at the expense of a major
		1631	complication, and having to use a constant timeout. The constant timeout
		1632	ensures that the list stays sorted.
		1633
		1634	=back
		1635
		1636	So which method the best?
		1637
		1638	Method #2 is a simple no-brain-required solution that is adequate in most
		1639	situations. Method #3 requires a bit more thinking, but handles many cases
		1640	better, and isn't very complicated either. In most case, choosing either
		1641	one is fine, with #3 being better in typical situations.
		1642
		1643	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1644	rather complicated, but extremely efficient, something that really pays
		1645	off after the first million or so of active timers, i.e. it's usually
		1646	overkill :)
		1647
		1648	=head3 The special problem of time updates
		1649
		1650	Establishing the current time is a costly operation (it usually takes at
		1651	least two system calls): EV therefore updates its idea of the current
		1652	time only before and after C<ev_loop> collects new events, which causes a
		1653	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1654	lots of events in one iteration.
525		1655
526	The relative timeouts are calculated relative to the C<ev_now ()>	1656	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	1657	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	1658	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	1659	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:	1660	timeout on the current time, use something like this to adjust for this:
531		1661
532	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1662	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
533		1663
534	The callback is guarenteed to be invoked only when its timeout has passed,	1664	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	1665	update of the time returned by C<ev_now ()> by calling C<ev_now_update
536	order of execution is undefined.	1666	()>.
		1667
		1668	=head3 Watcher-Specific Functions and Data Members
537		1669
538	=over 4	1670	=over 4
539		1671
540	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1672	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
541		1673
542	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1674	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
543		1675
544	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1676	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	1677	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	1678	reached. If it is positive, then the timer will automatically be
547	later, again, and again, until stopped manually.	1679	configured to trigger again C<repeat> seconds later, again, and again,
		1680	until stopped manually.
548		1681
549	The timer itself will do a best-effort at avoiding drift, that is, if you	1682	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	1683	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	1684	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	1685	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.	1686	do stuff) the timer will not fire more than once per event loop iteration.
554		1687
555	=item ev_timer_again (loop)	1688	=item ev_timer_again (loop, ev_timer *)
556		1689
557	This will act as if the timer timed out and restart it again if it is	1690	This will act as if the timer timed out and restart it again if it is
558	repeating. The exact semantics are:	1691	repeating. The exact semantics are:
559		1692
		1693	If the timer is pending, its pending status is cleared.
		1694
560	If the timer is started but nonrepeating, stop it.	1695	If the timer is started but non-repeating, stop it (as if it timed out).
561		1696
562	If the timer is repeating, either start it if necessary (with the repeat	1697	If the timer is repeating, either start it if necessary (with the
563	value), or reset the running timer to the repeat value.	1698	C<repeat> value), or reset the running timer to the C<repeat> value.
564		1699
565	This sounds a bit complicated, but here is a useful and typical	1700	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	1701	usage example.
567	timeout, that is, you want to be called when there have been, say, 60	1702
568	seconds of inactivity on the socket. The easiest way to do this is to	1703	=item ev_tstamp repeat [read-write]
569	configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each	1704
570	time you successfully read or write some data. If you go into an idle	1705	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	1706	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.	1707	which is also when any modifications are taken into account.
573		1708
574	=back	1709	=back
575		1710
		1711	=head3 Examples
		1712
		1713	Example: Create a timer that fires after 60 seconds.
		1714
		1715	static void
		1716	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1717	{
		1718	.. one minute over, w is actually stopped right here
		1719	}
		1720
		1721	ev_timer mytimer;
		1722	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
		1723	ev_timer_start (loop, &mytimer);
		1724
		1725	Example: Create a timeout timer that times out after 10 seconds of
		1726	inactivity.
		1727
		1728	static void
		1729	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
		1730	{
		1731	.. ten seconds without any activity
		1732	}
		1733
		1734	ev_timer mytimer;
		1735	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
		1736	ev_timer_again (&mytimer); /* start timer */
		1737	ev_loop (loop, 0);
		1738
		1739	// and in some piece of code that gets executed on any "activity":
		1740	// reset the timeout to start ticking again at 10 seconds
		1741	ev_timer_again (&mytimer);
		1742
		1743
576	=head2 C<ev_periodic> - to cron or not to cron	1744	=head2 C<ev_periodic> - to cron or not to cron?
577		1745
578	Periodic watchers are also timers of a kind, but they are very versatile	1746	Periodic watchers are also timers of a kind, but they are very versatile
579	(and unfortunately a bit complex).	1747	(and unfortunately a bit complex).
580		1748
581	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1749	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	1750	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	1751	(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 ()	1752	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	1753	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	1754	wrist-watch).
587	roughly 10 seconds later and of course not if you reset your system time
588	again).
589		1755
590	They can also be used to implement vastly more complex timers, such as	1756	You can tell a periodic watcher to trigger after some specific point
		1757	in time: for example, if you tell a periodic watcher to trigger "in 10
		1758	seconds" (by specifying e.g. C<ev_now () + 10.>, that is, an absolute time
		1759	not a delay) and then reset your system clock to January of the previous
		1760	year, then it will take a year or more to trigger the event (unlike an
		1761	C<ev_timer>, which would still trigger roughly 10 seconds after starting
		1762	it, as it uses a relative timeout).
		1763
		1764	C<ev_periodic> watchers can also be used to implement vastly more complex
591	triggering an event on eahc midnight, local time.	1765	timers, such as triggering an event on each "midnight, local time", or
		1766	other complicated rules. This cannot be done with C<ev_timer> watchers, as
		1767	those cannot react to time jumps.
592		1768
593	As with timers, the callback is guarenteed to be invoked only when the	1769	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	1770	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.	1771	timers become ready during the same loop iteration then the ones with
		1772	earlier time-out values are invoked before ones with later time-out values
		1773	(but this is no longer true when a callback calls C<ev_loop> recursively).
		1774
		1775	=head3 Watcher-Specific Functions and Data Members
596		1776
597	=over 4	1777	=over 4
598		1778
599	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1779	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
600		1780
601	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1781	=item ev_periodic_set (ev_periodic *, ev_tstamp offset, ev_tstamp interval, reschedule_cb)
602		1782
603	Lots of arguments, lets sort it out... There are basically three modes of	1783	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:	1784	operation, and we will explain them from simplest to most complex:
605		1785
606	=over 4	1786	=over 4
607		1787
608	=item * absolute timer (interval = reschedule_cb = 0)	1788	=item * absolute timer (offset = absolute time, interval = 0, reschedule_cb = 0)
609		1789
610	In this configuration the watcher triggers an event at the wallclock time	1790	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,	1791	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	1792	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.	1793	will be stopped and invoked when the system clock reaches or surpasses
		1794	this point in time.
614		1795
615	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1796	=item * repeating interval timer (offset = offset within interval, interval > 0, reschedule_cb = 0)
616		1797
617	In this mode the watcher will always be scheduled to time out at the next	1798	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	1799	C<offset + N * interval> time (for some integer N, which can also be
619	of any time jumps.	1800	negative) and then repeat, regardless of any time jumps. The C<offset>
		1801	argument is merely an offset into the C<interval> periods.
620		1802
621	This can be used to create timers that do not drift with respect to system	1803	This can be used to create timers that do not drift with respect to the
622	time:	1804	system clock, for example, here is an C<ev_periodic> that triggers each
		1805	hour, on the hour (with respect to UTC):
623		1806
624	ev_periodic_set (&periodic, 0., 3600., 0);	1807	ev_periodic_set (&periodic, 0., 3600., 0);
625		1808
626	This doesn't mean there will always be 3600 seconds in between triggers,	1809	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	1810	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	1811	full hour (UTC), or more correctly, when the system time is evenly divisible
629	by 3600.	1812	by 3600.
630		1813
631	Another way to think about it (for the mathematically inclined) is that	1814	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	1815	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.	1816	time where C<time = offset (mod interval)>, regardless of any time jumps.
634		1817
		1818	For numerical stability it is preferable that the C<offset> value is near
		1819	C<ev_now ()> (the current time), but there is no range requirement for
		1820	this value, and in fact is often specified as zero.
		1821
		1822	Note also that there is an upper limit to how often a timer can fire (CPU
		1823	speed for example), so if C<interval> is very small then timing stability
		1824	will of course deteriorate. Libev itself tries to be exact to be about one
		1825	millisecond (if the OS supports it and the machine is fast enough).
		1826
635	=item * manual reschedule mode (reschedule_cb = callback)	1827	=item * manual reschedule mode (offset ignored, interval ignored, reschedule_cb = callback)
636		1828
637	In this mode the values for C<interval> and C<at> are both being	1829	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	1830	ignored. Instead, each time the periodic watcher gets scheduled, the
639	reschedule callback will be called with the watcher as first, and the	1831	reschedule callback will be called with the watcher as first, and the
640	current time as second argument.	1832	current time as second argument.
641		1833
642	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1834	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,	1835	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	1836	allowed by documentation here>.
645	starting a prepare watcher).
646		1837
		1838	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		1839	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		1840	only event loop modification you are allowed to do).
		1841
647	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	1842	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
648	ev_tstamp now)>, e.g.:	1843	*w, ev_tstamp now)>, e.g.:
649		1844
		1845	static ev_tstamp
650	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1846	my_rescheduler (ev_periodic *w, ev_tstamp now)
651	{	1847	{
652	return now + 60.;	1848	return now + 60.;
653	}	1849	}
654		1850
655	It must return the next time to trigger, based on the passed time value	1851	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	1852	(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	1853	will usually be called just before the callback will be triggered, but
658	might be called at other times, too.	1854	might be called at other times, too.
659		1855
660	NOTE: I<< This callback must always return a time that is later than the	1856	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.	1857	equal to the passed C<now> value >>.
662		1858
663	This can be used to create very complex timers, such as a timer that	1859	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	1860	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	1861	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	1862	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).	1863	reason I omitted it as an example).
668		1864
669	=back	1865	=back
…		…
673	Simply stops and restarts the periodic watcher again. This is only useful	1869	Simply stops and restarts the periodic watcher again. This is only useful
674	when you changed some parameters or the reschedule callback would return	1870	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	1871	a different time than the last time it was called (e.g. in a crond like
676	program when the crontabs have changed).	1872	program when the crontabs have changed).
677		1873
		1874	=item ev_tstamp ev_periodic_at (ev_periodic *)
		1875
		1876	When active, returns the absolute time that the watcher is supposed
		1877	to trigger next. This is not the same as the C<offset> argument to
		1878	C<ev_periodic_set>, but indeed works even in interval and manual
		1879	rescheduling modes.
		1880
		1881	=item ev_tstamp offset [read-write]
		1882
		1883	When repeating, this contains the offset value, otherwise this is the
		1884	absolute point in time (the C<offset> value passed to C<ev_periodic_set>,
		1885	although libev might modify this value for better numerical stability).
		1886
		1887	Can be modified any time, but changes only take effect when the periodic
		1888	timer fires or C<ev_periodic_again> is being called.
		1889
		1890	=item ev_tstamp interval [read-write]
		1891
		1892	The current interval value. Can be modified any time, but changes only
		1893	take effect when the periodic timer fires or C<ev_periodic_again> is being
		1894	called.
		1895
		1896	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
		1897
		1898	The current reschedule callback, or C<0>, if this functionality is
		1899	switched off. Can be changed any time, but changes only take effect when
		1900	the periodic timer fires or C<ev_periodic_again> is being called.
		1901
678	=back	1902	=back
679		1903
		1904	=head3 Examples
		1905
		1906	Example: Call a callback every hour, or, more precisely, whenever the
		1907	system time is divisible by 3600. The callback invocation times have
		1908	potentially a lot of jitter, but good long-term stability.
		1909
		1910	static void
		1911	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
		1912	{
		1913	... its now a full hour (UTC, or TAI or whatever your clock follows)
		1914	}
		1915
		1916	ev_periodic hourly_tick;
		1917	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
		1918	ev_periodic_start (loop, &hourly_tick);
		1919
		1920	Example: The same as above, but use a reschedule callback to do it:
		1921
		1922	#include <math.h>
		1923
		1924	static ev_tstamp
		1925	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
		1926	{
		1927	return now + (3600. - fmod (now, 3600.));
		1928	}
		1929
		1930	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
		1931
		1932	Example: Call a callback every hour, starting now:
		1933
		1934	ev_periodic hourly_tick;
		1935	ev_periodic_init (&hourly_tick, clock_cb,
		1936	fmod (ev_now (loop), 3600.), 3600., 0);
		1937	ev_periodic_start (loop, &hourly_tick);
		1938
		1939
680	=head2 C<ev_signal> - signal me when a signal gets signalled	1940	=head2 C<ev_signal> - signal me when a signal gets signalled!
681		1941
682	Signal watchers will trigger an event when the process receives a specific	1942	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	1943	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	1944	will try it's best to deliver signals synchronously, i.e. as part of the
685	normal event processing, like any other event.	1945	normal event processing, like any other event.
686		1946
		1947	If you want signals asynchronously, just use C<sigaction> as you would
		1948	do without libev and forget about sharing the signal. You can even use
		1949	C<ev_async> from a signal handler to synchronously wake up an event loop.
		1950
687	You can configure as many watchers as you like per signal. Only when the	1951	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	1952	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	1953	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	1954	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	1955	the last signal watcher for a signal is stopped, libev will reset the
692	SIG_DFL (regardless of what it was set to before).	1956	signal handler to SIG_DFL (regardless of what it was set to before).
		1957
		1958	If possible and supported, libev will install its handlers with
		1959	C<SA_RESTART> behaviour enabled, so system calls should not be unduly
		1960	interrupted. If you have a problem with system calls getting interrupted by
		1961	signals you can block all signals in an C<ev_check> watcher and unblock
		1962	them in an C<ev_prepare> watcher.
		1963
		1964	=head3 Watcher-Specific Functions and Data Members
693		1965
694	=over 4	1966	=over 4
695		1967
696	=item ev_signal_init (ev_signal *, callback, int signum)	1968	=item ev_signal_init (ev_signal *, callback, int signum)
697		1969
698	=item ev_signal_set (ev_signal *, int signum)	1970	=item ev_signal_set (ev_signal *, int signum)
699		1971
700	Configures the watcher to trigger on the given signal number (usually one	1972	Configures the watcher to trigger on the given signal number (usually one
701	of the C<SIGxxx> constants).	1973	of the C<SIGxxx> constants).
702		1974
		1975	=item int signum [read-only]
		1976
		1977	The signal the watcher watches out for.
		1978
703	=back	1979	=back
704		1980
		1981	=head3 Examples
		1982
		1983	Example: Try to exit cleanly on SIGINT.
		1984
		1985	static void
		1986	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		1987	{
		1988	ev_unloop (loop, EVUNLOOP_ALL);
		1989	}
		1990
		1991	ev_signal signal_watcher;
		1992	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		1993	ev_signal_start (loop, &signal_watcher);
		1994
		1995
705	=head2 C<ev_child> - wait for pid status changes	1996	=head2 C<ev_child> - watch out for process status changes
706		1997
707	Child watchers trigger when your process receives a SIGCHLD in response to	1998	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).	1999	some child status changes (most typically when a child of yours dies or
		2000	exits). It is permissible to install a child watcher I<after> the child
		2001	has been forked (which implies it might have already exited), as long
		2002	as the event loop isn't entered (or is continued from a watcher), i.e.,
		2003	forking and then immediately registering a watcher for the child is fine,
		2004	but forking and registering a watcher a few event loop iterations later is
		2005	not.
		2006
		2007	Only the default event loop is capable of handling signals, and therefore
		2008	you can only register child watchers in the default event loop.
		2009
		2010	=head3 Process Interaction
		2011
		2012	Libev grabs C<SIGCHLD> as soon as the default event loop is
		2013	initialised. This is necessary to guarantee proper behaviour even if
		2014	the first child watcher is started after the child exits. The occurrence
		2015	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		2016	synchronously as part of the event loop processing. Libev always reaps all
		2017	children, even ones not watched.
		2018
		2019	=head3 Overriding the Built-In Processing
		2020
		2021	Libev offers no special support for overriding the built-in child
		2022	processing, but if your application collides with libev's default child
		2023	handler, you can override it easily by installing your own handler for
		2024	C<SIGCHLD> after initialising the default loop, and making sure the
		2025	default loop never gets destroyed. You are encouraged, however, to use an
		2026	event-based approach to child reaping and thus use libev's support for
		2027	that, so other libev users can use C<ev_child> watchers freely.
		2028
		2029	=head3 Stopping the Child Watcher
		2030
		2031	Currently, the child watcher never gets stopped, even when the
		2032	child terminates, so normally one needs to stop the watcher in the
		2033	callback. Future versions of libev might stop the watcher automatically
		2034	when a child exit is detected.
		2035
		2036	=head3 Watcher-Specific Functions and Data Members
709		2037
710	=over 4	2038	=over 4
711		2039
712	=item ev_child_init (ev_child *, callback, int pid)	2040	=item ev_child_init (ev_child *, callback, int pid, int trace)
713		2041
714	=item ev_child_set (ev_child *, int pid)	2042	=item ev_child_set (ev_child *, int pid, int trace)
715		2043
716	Configures the watcher to wait for status changes of process C<pid> (or	2044	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	2045	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	2046	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	2047	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	2048	C<waitpid> documentation). The C<rpid> member contains the pid of the
721	process causing the status change.	2049	process causing the status change. C<trace> must be either C<0> (only
		2050	activate the watcher when the process terminates) or C<1> (additionally
		2051	activate the watcher when the process is stopped or continued).
		2052
		2053	=item int pid [read-only]
		2054
		2055	The process id this watcher watches out for, or C<0>, meaning any process id.
		2056
		2057	=item int rpid [read-write]
		2058
		2059	The process id that detected a status change.
		2060
		2061	=item int rstatus [read-write]
		2062
		2063	The process exit/trace status caused by C<rpid> (see your systems
		2064	C<waitpid> and C<sys/wait.h> documentation for details).
722		2065
723	=back	2066	=back
724		2067
		2068	=head3 Examples
		2069
		2070	Example: C<fork()> a new process and install a child handler to wait for
		2071	its completion.
		2072
		2073	ev_child cw;
		2074
		2075	static void
		2076	child_cb (EV_P_ ev_child *w, int revents)
		2077	{
		2078	ev_child_stop (EV_A_ w);
		2079	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
		2080	}
		2081
		2082	pid_t pid = fork ();
		2083
		2084	if (pid < 0)
		2085	// error
		2086	else if (pid == 0)
		2087	{
		2088	// the forked child executes here
		2089	exit (1);
		2090	}
		2091	else
		2092	{
		2093	ev_child_init (&cw, child_cb, pid, 0);
		2094	ev_child_start (EV_DEFAULT_ &cw);
		2095	}
		2096
		2097
		2098	=head2 C<ev_stat> - did the file attributes just change?
		2099
		2100	This watches a file system path for attribute changes. That is, it calls
		2101	C<stat> on that path in regular intervals (or when the OS says it changed)
		2102	and sees if it changed compared to the last time, invoking the callback if
		2103	it did.
		2104
		2105	The path does not need to exist: changing from "path exists" to "path does
		2106	not exist" is a status change like any other. The condition "path does not
		2107	exist" (or more correctly "path cannot be stat'ed") is signified by the
		2108	C<st_nlink> field being zero (which is otherwise always forced to be at
		2109	least one) and all the other fields of the stat buffer having unspecified
		2110	contents.
		2111
		2112	The path I<must not> end in a slash or contain special components such as
		2113	C<.> or C<..>. The path I<should> be absolute: If it is relative and
		2114	your working directory changes, then the behaviour is undefined.
		2115
		2116	Since there is no portable change notification interface available, the
		2117	portable implementation simply calls C<stat(2)> regularly on the path
		2118	to see if it changed somehow. You can specify a recommended polling
		2119	interval for this case. If you specify a polling interval of C<0> (highly
		2120	recommended!) then a I<suitable, unspecified default> value will be used
		2121	(which you can expect to be around five seconds, although this might
		2122	change dynamically). Libev will also impose a minimum interval which is
		2123	currently around C<0.1>, but that's usually overkill.
		2124
		2125	This watcher type is not meant for massive numbers of stat watchers,
		2126	as even with OS-supported change notifications, this can be
		2127	resource-intensive.
		2128
		2129	At the time of this writing, the only OS-specific interface implemented
		2130	is the Linux inotify interface (implementing kqueue support is left as an
		2131	exercise for the reader. Note, however, that the author sees no way of
		2132	implementing C<ev_stat> semantics with kqueue, except as a hint).
		2133
		2134	=head3 ABI Issues (Largefile Support)
		2135
		2136	Libev by default (unless the user overrides this) uses the default
		2137	compilation environment, which means that on systems with large file
		2138	support disabled by default, you get the 32 bit version of the stat
		2139	structure. When using the library from programs that change the ABI to
		2140	use 64 bit file offsets the programs will fail. In that case you have to
		2141	compile libev with the same flags to get binary compatibility. This is
		2142	obviously the case with any flags that change the ABI, but the problem is
		2143	most noticeably displayed with ev_stat and large file support.
		2144
		2145	The solution for this is to lobby your distribution maker to make large
		2146	file interfaces available by default (as e.g. FreeBSD does) and not
		2147	optional. Libev cannot simply switch on large file support because it has
		2148	to exchange stat structures with application programs compiled using the
		2149	default compilation environment.
		2150
		2151	=head3 Inotify and Kqueue
		2152
		2153	When C<inotify (7)> support has been compiled into libev and present at
		2154	runtime, it will be used to speed up change detection where possible. The
		2155	inotify descriptor will be created lazily when the first C<ev_stat>
		2156	watcher is being started.
		2157
		2158	Inotify presence does not change the semantics of C<ev_stat> watchers
		2159	except that changes might be detected earlier, and in some cases, to avoid
		2160	making regular C<stat> calls. Even in the presence of inotify support
		2161	there are many cases where libev has to resort to regular C<stat> polling,
		2162	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		2163	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		2164	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		2165	xfs are fully working) libev usually gets away without polling.
		2166
		2167	There is no support for kqueue, as apparently it cannot be used to
		2168	implement this functionality, due to the requirement of having a file
		2169	descriptor open on the object at all times, and detecting renames, unlinks
		2170	etc. is difficult.
		2171
		2172	=head3 C<stat ()> is a synchronous operation
		2173
		2174	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2175	the process. The exception are C<ev_stat> watchers - those call C<stat
		2176	()>, which is a synchronous operation.
		2177
		2178	For local paths, this usually doesn't matter: unless the system is very
		2179	busy or the intervals between stat's are large, a stat call will be fast,
		2180	as the path data is usually in memory already (except when starting the
		2181	watcher).
		2182
		2183	For networked file systems, calling C<stat ()> can block an indefinite
		2184	time due to network issues, and even under good conditions, a stat call
		2185	often takes multiple milliseconds.
		2186
		2187	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2188	paths, although this is fully supported by libev.
		2189
		2190	=head3 The special problem of stat time resolution
		2191
		2192	The C<stat ()> system call only supports full-second resolution portably,
		2193	and even on systems where the resolution is higher, most file systems
		2194	still only support whole seconds.
		2195
		2196	That means that, if the time is the only thing that changes, you can
		2197	easily miss updates: on the first update, C<ev_stat> detects a change and
		2198	calls your callback, which does something. When there is another update
		2199	within the same second, C<ev_stat> will be unable to detect unless the
		2200	stat data does change in other ways (e.g. file size).
		2201
		2202	The solution to this is to delay acting on a change for slightly more
		2203	than a second (or till slightly after the next full second boundary), using
		2204	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2205	ev_timer_again (loop, w)>).
		2206
		2207	The C<.02> offset is added to work around small timing inconsistencies
		2208	of some operating systems (where the second counter of the current time
		2209	might be be delayed. One such system is the Linux kernel, where a call to
		2210	C<gettimeofday> might return a timestamp with a full second later than
		2211	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2212	update file times then there will be a small window where the kernel uses
		2213	the previous second to update file times but libev might already execute
		2214	the timer callback).
		2215
		2216	=head3 Watcher-Specific Functions and Data Members
		2217
		2218	=over 4
		2219
		2220	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
		2221
		2222	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
		2223
		2224	Configures the watcher to wait for status changes of the given
		2225	C<path>. The C<interval> is a hint on how quickly a change is expected to
		2226	be detected and should normally be specified as C<0> to let libev choose
		2227	a suitable value. The memory pointed to by C<path> must point to the same
		2228	path for as long as the watcher is active.
		2229
		2230	The callback will receive an C<EV_STAT> event when a change was detected,
		2231	relative to the attributes at the time the watcher was started (or the
		2232	last change was detected).
		2233
		2234	=item ev_stat_stat (loop, ev_stat *)
		2235
		2236	Updates the stat buffer immediately with new values. If you change the
		2237	watched path in your callback, you could call this function to avoid
		2238	detecting this change (while introducing a race condition if you are not
		2239	the only one changing the path). Can also be useful simply to find out the
		2240	new values.
		2241
		2242	=item ev_statdata attr [read-only]
		2243
		2244	The most-recently detected attributes of the file. Although the type is
		2245	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
		2246	suitable for your system, but you can only rely on the POSIX-standardised
		2247	members to be present. If the C<st_nlink> member is C<0>, then there was
		2248	some error while C<stat>ing the file.
		2249
		2250	=item ev_statdata prev [read-only]
		2251
		2252	The previous attributes of the file. The callback gets invoked whenever
		2253	C<prev> != C<attr>, or, more precisely, one or more of these members
		2254	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2255	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
		2256
		2257	=item ev_tstamp interval [read-only]
		2258
		2259	The specified interval.
		2260
		2261	=item const char *path [read-only]
		2262
		2263	The file system path that is being watched.
		2264
		2265	=back
		2266
		2267	=head3 Examples
		2268
		2269	Example: Watch C</etc/passwd> for attribute changes.
		2270
		2271	static void
		2272	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
		2273	{
		2274	/* /etc/passwd changed in some way */
		2275	if (w->attr.st_nlink)
		2276	{
		2277	printf ("passwd current size %ld\n", (long)w->attr.st_size);
		2278	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
		2279	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
		2280	}
		2281	else
		2282	/* you shalt not abuse printf for puts */
		2283	puts ("wow, /etc/passwd is not there, expect problems. "
		2284	"if this is windows, they already arrived\n");
		2285	}
		2286
		2287	...
		2288	ev_stat passwd;
		2289
		2290	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
		2291	ev_stat_start (loop, &passwd);
		2292
		2293	Example: Like above, but additionally use a one-second delay so we do not
		2294	miss updates (however, frequent updates will delay processing, too, so
		2295	one might do the work both on C<ev_stat> callback invocation I<and> on
		2296	C<ev_timer> callback invocation).
		2297
		2298	static ev_stat passwd;
		2299	static ev_timer timer;
		2300
		2301	static void
		2302	timer_cb (EV_P_ ev_timer *w, int revents)
		2303	{
		2304	ev_timer_stop (EV_A_ w);
		2305
		2306	/* now it's one second after the most recent passwd change */
		2307	}
		2308
		2309	static void
		2310	stat_cb (EV_P_ ev_stat *w, int revents)
		2311	{
		2312	/* reset the one-second timer */
		2313	ev_timer_again (EV_A_ &timer);
		2314	}
		2315
		2316	...
		2317	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2318	ev_stat_start (loop, &passwd);
		2319	ev_timer_init (&timer, timer_cb, 0., 1.02);
		2320
		2321
725	=head2 C<ev_idle> - when you've got nothing better to do	2322	=head2 C<ev_idle> - when you've got nothing better to do...
726		2323
727	Idle watchers trigger events when there are no other events are pending	2324	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	2325	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,	2326	as receiving "events").
730	imagine) it will not be triggered. But when your process is idle all idle	2327
731	watchers are being called again and again, once per event loop iteration -	2328	That is, as long as your process is busy handling sockets or timeouts
		2329	(or even signals, imagine) of the same or higher priority it will not be
		2330	triggered. But when your process is idle (or only lower-priority watchers
		2331	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	2332	iteration - until stopped, that is, or your process receives more events
733	busy.	2333	and becomes busy again with higher priority stuff.
734		2334
735	The most noteworthy effect is that as long as any idle watchers are	2335	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.	2336	active, the process will not block when waiting for new events.
737		2337
738	Apart from keeping your process non-blocking (which is a useful	2338	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	2339	effect on its own sometimes), idle watchers are a good place to do
740	"pseudo-background processing", or delay processing stuff to after the	2340	"pseudo-background processing", or delay processing stuff to after the
741	event loop has handled all outstanding events.	2341	event loop has handled all outstanding events.
742		2342
		2343	=head3 Watcher-Specific Functions and Data Members
		2344
743	=over 4	2345	=over 4
744		2346
745	=item ev_idle_init (ev_signal *, callback)	2347	=item ev_idle_init (ev_idle *, callback)
746		2348
747	Initialises and configures the idle watcher - it has no parameters of any	2349	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,	2350	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
749	believe me.	2351	believe me.
750		2352
751	=back	2353	=back
752		2354
		2355	=head3 Examples
		2356
		2357	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
		2358	callback, free it. Also, use no error checking, as usual.
		2359
		2360	static void
		2361	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
		2362	{
		2363	free (w);
		2364	// now do something you wanted to do when the program has
		2365	// no longer anything immediate to do.
		2366	}
		2367
		2368	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
		2369	ev_idle_init (idle_watcher, idle_cb);
		2370	ev_idle_start (loop, idle_cb);
		2371
		2372
753	=head2 C<ev_prepare> and C<ev_check> - customise your event loop	2373	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
754		2374
755	Prepare and check watchers are usually (but not always) used in tandem:	2375	Prepare and check watchers are usually (but not always) used in pairs:
756	prepare watchers get invoked before the process blocks and check watchers	2376	prepare watchers get invoked before the process blocks and check watchers
757	afterwards.	2377	afterwards.
758		2378
		2379	You I<must not> call C<ev_loop> or similar functions that enter
		2380	the current event loop from either C<ev_prepare> or C<ev_check>
		2381	watchers. Other loops than the current one are fine, however. The
		2382	rationale behind this is that you do not need to check for recursion in
		2383	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
		2384	C<ev_check> so if you have one watcher of each kind they will always be
		2385	called in pairs bracketing the blocking call.
		2386
759	Their main purpose is to integrate other event mechanisms into libev. This	2387	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	2388	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.	2389	variable changes, implement your own watchers, integrate net-snmp or a
		2390	coroutine library and lots more. They are also occasionally useful if
		2391	you cache some data and want to flush it before blocking (for example,
		2392	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
		2393	watcher).
762		2394
763	This is done by examining in each prepare call which file descriptors need	2395	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	2396	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	2397	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	2398	libraries provide exactly this functionality). Then, in the check watcher,
767	any events that occured (by checking the pending status of all watchers	2399	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	2400	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,	2401	I/O and timer callbacks will never actually be called (but must be valid
770	because you never know, you know?).	2402	nevertheless, because you never know, you know?).
771		2403
772	As another example, the Perl Coro module uses these hooks to integrate	2404	As another example, the Perl Coro module uses these hooks to integrate
773	coroutines into libev programs, by yielding to other active coroutines	2405	coroutines into libev programs, by yielding to other active coroutines
774	during each prepare and only letting the process block if no coroutines	2406	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	2407	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	2408	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	2409	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	2410	loop from blocking if lower-priority coroutines are active, thus mapping
779	low-priority coroutines to idle/background tasks).	2411	low-priority coroutines to idle/background tasks).
780		2412
		2413	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2414	priority, to ensure that they are being run before any other watchers
		2415	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2416
		2417	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2418	activate ("feed") events into libev. While libev fully supports this, they
		2419	might get executed before other C<ev_check> watchers did their job. As
		2420	C<ev_check> watchers are often used to embed other (non-libev) event
		2421	loops those other event loops might be in an unusable state until their
		2422	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2423	others).
		2424
		2425	=head3 Watcher-Specific Functions and Data Members
		2426
781	=over 4	2427	=over 4
782		2428
783	=item ev_prepare_init (ev_prepare *, callback)	2429	=item ev_prepare_init (ev_prepare *, callback)
784		2430
785	=item ev_check_init (ev_check *, callback)	2431	=item ev_check_init (ev_check *, callback)
786		2432
787	Initialises and configures the prepare or check watcher - they have no	2433	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>	2434	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.	2435	macros, but using them is utterly, utterly, utterly and completely
		2436	pointless.
790		2437
791	=back	2438	=back
792		2439
		2440	=head3 Examples
		2441
		2442	There are a number of principal ways to embed other event loops or modules
		2443	into libev. Here are some ideas on how to include libadns into libev
		2444	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2445	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2446	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2447	Glib event loop).
		2448
		2449	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
		2450	and in a check watcher, destroy them and call into libadns. What follows
		2451	is pseudo-code only of course. This requires you to either use a low
		2452	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2453	the callbacks for the IO/timeout watchers might not have been called yet.
		2454
		2455	static ev_io iow [nfd];
		2456	static ev_timer tw;
		2457
		2458	static void
		2459	io_cb (struct ev_loop *loop, ev_io *w, int revents)
		2460	{
		2461	}
		2462
		2463	// create io watchers for each fd and a timer before blocking
		2464	static void
		2465	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
		2466	{
		2467	int timeout = 3600000;
		2468	struct pollfd fds [nfd];
		2469	// actual code will need to loop here and realloc etc.
		2470	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
		2471
		2472	/* the callback is illegal, but won't be called as we stop during check */
		2473	ev_timer_init (&tw, 0, timeout * 1e-3);
		2474	ev_timer_start (loop, &tw);
		2475
		2476	// create one ev_io per pollfd
		2477	for (int i = 0; i < nfd; ++i)
		2478	{
		2479	ev_io_init (iow + i, io_cb, fds [i].fd,
		2480	((fds [i].events & POLLIN ? EV_READ : 0)
		2481	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
		2482
		2483	fds [i].revents = 0;
		2484	ev_io_start (loop, iow + i);
		2485	}
		2486	}
		2487
		2488	// stop all watchers after blocking
		2489	static void
		2490	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
		2491	{
		2492	ev_timer_stop (loop, &tw);
		2493
		2494	for (int i = 0; i < nfd; ++i)
		2495	{
		2496	// set the relevant poll flags
		2497	// could also call adns_processreadable etc. here
		2498	struct pollfd *fd = fds + i;
		2499	int revents = ev_clear_pending (iow + i);
		2500	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2501	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2502
		2503	// now stop the watcher
		2504	ev_io_stop (loop, iow + i);
		2505	}
		2506
		2507	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		2508	}
		2509
		2510	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2511	in the prepare watcher and would dispose of the check watcher.
		2512
		2513	Method 3: If the module to be embedded supports explicit event
		2514	notification (libadns does), you can also make use of the actual watcher
		2515	callbacks, and only destroy/create the watchers in the prepare watcher.
		2516
		2517	static void
		2518	timer_cb (EV_P_ ev_timer *w, int revents)
		2519	{
		2520	adns_state ads = (adns_state)w->data;
		2521	update_now (EV_A);
		2522
		2523	adns_processtimeouts (ads, &tv_now);
		2524	}
		2525
		2526	static void
		2527	io_cb (EV_P_ ev_io *w, int revents)
		2528	{
		2529	adns_state ads = (adns_state)w->data;
		2530	update_now (EV_A);
		2531
		2532	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2533	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2534	}
		2535
		2536	// do not ever call adns_afterpoll
		2537
		2538	Method 4: Do not use a prepare or check watcher because the module you
		2539	want to embed is not flexible enough to support it. Instead, you can
		2540	override their poll function. The drawback with this solution is that the
		2541	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2542	this approach, effectively embedding EV as a client into the horrible
		2543	libglib event loop.
		2544
		2545	static gint
		2546	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2547	{
		2548	int got_events = 0;
		2549
		2550	for (n = 0; n < nfds; ++n)
		2551	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2552
		2553	if (timeout >= 0)
		2554	// create/start timer
		2555
		2556	// poll
		2557	ev_loop (EV_A_ 0);
		2558
		2559	// stop timer again
		2560	if (timeout >= 0)
		2561	ev_timer_stop (EV_A_ &to);
		2562
		2563	// stop io watchers again - their callbacks should have set
		2564	for (n = 0; n < nfds; ++n)
		2565	ev_io_stop (EV_A_ iow [n]);
		2566
		2567	return got_events;
		2568	}
		2569
		2570
		2571	=head2 C<ev_embed> - when one backend isn't enough...
		2572
		2573	This is a rather advanced watcher type that lets you embed one event loop
		2574	into another (currently only C<ev_io> events are supported in the embedded
		2575	loop, other types of watchers might be handled in a delayed or incorrect
		2576	fashion and must not be used).
		2577
		2578	There are primarily two reasons you would want that: work around bugs and
		2579	prioritise I/O.
		2580
		2581	As an example for a bug workaround, the kqueue backend might only support
		2582	sockets on some platform, so it is unusable as generic backend, but you
		2583	still want to make use of it because you have many sockets and it scales
		2584	so nicely. In this case, you would create a kqueue-based loop and embed
		2585	it into your default loop (which might use e.g. poll). Overall operation
		2586	will be a bit slower because first libev has to call C<poll> and then
		2587	C<kevent>, but at least you can use both mechanisms for what they are
		2588	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
		2589
		2590	As for prioritising I/O: under rare circumstances you have the case where
		2591	some fds have to be watched and handled very quickly (with low latency),
		2592	and even priorities and idle watchers might have too much overhead. In
		2593	this case you would put all the high priority stuff in one loop and all
		2594	the rest in a second one, and embed the second one in the first.
		2595
		2596	As long as the watcher is active, the callback will be invoked every
		2597	time there might be events pending in the embedded loop. The callback
		2598	must then call C<ev_embed_sweep (mainloop, watcher)> to make a single
		2599	sweep and invoke their callbacks (the callback doesn't need to invoke the
		2600	C<ev_embed_sweep> function directly, it could also start an idle watcher
		2601	to give the embedded loop strictly lower priority for example).
		2602
		2603	You can also set the callback to C<0>, in which case the embed watcher
		2604	will automatically execute the embedded loop sweep whenever necessary.
		2605
		2606	Fork detection will be handled transparently while the C<ev_embed> watcher
		2607	is active, i.e., the embedded loop will automatically be forked when the
		2608	embedding loop forks. In other cases, the user is responsible for calling
		2609	C<ev_loop_fork> on the embedded loop.
		2610
		2611	Unfortunately, not all backends are embeddable: only the ones returned by
		2612	C<ev_embeddable_backends> are, which, unfortunately, does not include any
		2613	portable one.
		2614
		2615	So when you want to use this feature you will always have to be prepared
		2616	that you cannot get an embeddable loop. The recommended way to get around
		2617	this is to have a separate variables for your embeddable loop, try to
		2618	create it, and if that fails, use the normal loop for everything.
		2619
		2620	=head3 C<ev_embed> and fork
		2621
		2622	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2623	automatically be applied to the embedded loop as well, so no special
		2624	fork handling is required in that case. When the watcher is not running,
		2625	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2626	as applicable.
		2627
		2628	=head3 Watcher-Specific Functions and Data Members
		2629
		2630	=over 4
		2631
		2632	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		2633
		2634	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		2635
		2636	Configures the watcher to embed the given loop, which must be
		2637	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2638	invoked automatically, otherwise it is the responsibility of the callback
		2639	to invoke it (it will continue to be called until the sweep has been done,
		2640	if you do not want that, you need to temporarily stop the embed watcher).
		2641
		2642	=item ev_embed_sweep (loop, ev_embed *)
		2643
		2644	Make a single, non-blocking sweep over the embedded loop. This works
		2645	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		2646	appropriate way for embedded loops.
		2647
		2648	=item struct ev_loop *other [read-only]
		2649
		2650	The embedded event loop.
		2651
		2652	=back
		2653
		2654	=head3 Examples
		2655
		2656	Example: Try to get an embeddable event loop and embed it into the default
		2657	event loop. If that is not possible, use the default loop. The default
		2658	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2659	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2660	used).
		2661
		2662	struct ev_loop *loop_hi = ev_default_init (0);
		2663	struct ev_loop *loop_lo = 0;
		2664	ev_embed embed;
		2665
		2666	// see if there is a chance of getting one that works
		2667	// (remember that a flags value of 0 means autodetection)
		2668	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2669	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2670	: 0;
		2671
		2672	// if we got one, then embed it, otherwise default to loop_hi
		2673	if (loop_lo)
		2674	{
		2675	ev_embed_init (&embed, 0, loop_lo);
		2676	ev_embed_start (loop_hi, &embed);
		2677	}
		2678	else
		2679	loop_lo = loop_hi;
		2680
		2681	Example: Check if kqueue is available but not recommended and create
		2682	a kqueue backend for use with sockets (which usually work with any
		2683	kqueue implementation). Store the kqueue/socket-only event loop in
		2684	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2685
		2686	struct ev_loop *loop = ev_default_init (0);
		2687	struct ev_loop *loop_socket = 0;
		2688	ev_embed embed;
		2689
		2690	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2691	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2692	{
		2693	ev_embed_init (&embed, 0, loop_socket);
		2694	ev_embed_start (loop, &embed);
		2695	}
		2696
		2697	if (!loop_socket)
		2698	loop_socket = loop;
		2699
		2700	// now use loop_socket for all sockets, and loop for everything else
		2701
		2702
		2703	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
		2704
		2705	Fork watchers are called when a C<fork ()> was detected (usually because
		2706	whoever is a good citizen cared to tell libev about it by calling
		2707	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the
		2708	event loop blocks next and before C<ev_check> watchers are being called,
		2709	and only in the child after the fork. If whoever good citizen calling
		2710	C<ev_default_fork> cheats and calls it in the wrong process, the fork
		2711	handlers will be invoked, too, of course.
		2712
		2713	=head3 Watcher-Specific Functions and Data Members
		2714
		2715	=over 4
		2716
		2717	=item ev_fork_init (ev_signal *, callback)
		2718
		2719	Initialises and configures the fork watcher - it has no parameters of any
		2720	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
		2721	believe me.
		2722
		2723	=back
		2724
		2725
		2726	=head2 C<ev_async> - how to wake up another event loop
		2727
		2728	In general, you cannot use an C<ev_loop> from multiple threads or other
		2729	asynchronous sources such as signal handlers (as opposed to multiple event
		2730	loops - those are of course safe to use in different threads).
		2731
		2732	Sometimes, however, you need to wake up another event loop you do not
		2733	control, for example because it belongs to another thread. This is what
		2734	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2735	can signal it by calling C<ev_async_send>, which is thread- and signal
		2736	safe.
		2737
		2738	This functionality is very similar to C<ev_signal> watchers, as signals,
		2739	too, are asynchronous in nature, and signals, too, will be compressed
		2740	(i.e. the number of callback invocations may be less than the number of
		2741	C<ev_async_sent> calls).
		2742
		2743	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2744	just the default loop.
		2745
		2746	=head3 Queueing
		2747
		2748	C<ev_async> does not support queueing of data in any way. The reason
		2749	is that the author does not know of a simple (or any) algorithm for a
		2750	multiple-writer-single-reader queue that works in all cases and doesn't
		2751	need elaborate support such as pthreads.
		2752
		2753	That means that if you want to queue data, you have to provide your own
		2754	queue. But at least I can tell you how to implement locking around your
		2755	queue:
		2756
		2757	=over 4
		2758
		2759	=item queueing from a signal handler context
		2760
		2761	To implement race-free queueing, you simply add to the queue in the signal
		2762	handler but you block the signal handler in the watcher callback. Here is
		2763	an example that does that for some fictitious SIGUSR1 handler:
		2764
		2765	static ev_async mysig;
		2766
		2767	static void
		2768	sigusr1_handler (void)
		2769	{
		2770	sometype data;
		2771
		2772	// no locking etc.
		2773	queue_put (data);
		2774	ev_async_send (EV_DEFAULT_ &mysig);
		2775	}
		2776
		2777	static void
		2778	mysig_cb (EV_P_ ev_async *w, int revents)
		2779	{
		2780	sometype data;
		2781	sigset_t block, prev;
		2782
		2783	sigemptyset (&block);
		2784	sigaddset (&block, SIGUSR1);
		2785	sigprocmask (SIG_BLOCK, &block, &prev);
		2786
		2787	while (queue_get (&data))
		2788	process (data);
		2789
		2790	if (sigismember (&prev, SIGUSR1)
		2791	sigprocmask (SIG_UNBLOCK, &block, 0);
		2792	}
		2793
		2794	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2795	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2796	either...).
		2797
		2798	=item queueing from a thread context
		2799
		2800	The strategy for threads is different, as you cannot (easily) block
		2801	threads but you can easily preempt them, so to queue safely you need to
		2802	employ a traditional mutex lock, such as in this pthread example:
		2803
		2804	static ev_async mysig;
		2805	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2806
		2807	static void
		2808	otherthread (void)
		2809	{
		2810	// only need to lock the actual queueing operation
		2811	pthread_mutex_lock (&mymutex);
		2812	queue_put (data);
		2813	pthread_mutex_unlock (&mymutex);
		2814
		2815	ev_async_send (EV_DEFAULT_ &mysig);
		2816	}
		2817
		2818	static void
		2819	mysig_cb (EV_P_ ev_async *w, int revents)
		2820	{
		2821	pthread_mutex_lock (&mymutex);
		2822
		2823	while (queue_get (&data))
		2824	process (data);
		2825
		2826	pthread_mutex_unlock (&mymutex);
		2827	}
		2828
		2829	=back
		2830
		2831
		2832	=head3 Watcher-Specific Functions and Data Members
		2833
		2834	=over 4
		2835
		2836	=item ev_async_init (ev_async *, callback)
		2837
		2838	Initialises and configures the async watcher - it has no parameters of any
		2839	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		2840	trust me.
		2841
		2842	=item ev_async_send (loop, ev_async *)
		2843
		2844	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2845	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2846	C<ev_feed_event>, this call is safe to do from other threads, signal or
		2847	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		2848	section below on what exactly this means).
		2849
		2850	Note that, as with other watchers in libev, multiple events might get
		2851	compressed into a single callback invocation (another way to look at this
		2852	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,
		2853	reset when the event loop detects that).
		2854
		2855	This call incurs the overhead of a system call only once per event loop
		2856	iteration, so while the overhead might be noticeable, it doesn't apply to
		2857	repeated calls to C<ev_async_send> for the same event loop.
		2858
		2859	=item bool = ev_async_pending (ev_async *)
		2860
		2861	Returns a non-zero value when C<ev_async_send> has been called on the
		2862	watcher but the event has not yet been processed (or even noted) by the
		2863	event loop.
		2864
		2865	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		2866	the loop iterates next and checks for the watcher to have become active,
		2867	it will reset the flag again. C<ev_async_pending> can be used to very
		2868	quickly check whether invoking the loop might be a good idea.
		2869
		2870	Not that this does I<not> check whether the watcher itself is pending,
		2871	only whether it has been requested to make this watcher pending: there
		2872	is a time window between the event loop checking and resetting the async
		2873	notification, and the callback being invoked.
		2874
		2875	=back
		2876
		2877
793	=head1 OTHER FUNCTIONS	2878	=head1 OTHER FUNCTIONS
794		2879
795	There are some other functions of possible interest. Described. Here. Now.	2880	There are some other functions of possible interest. Described. Here. Now.
796		2881
797	=over 4	2882	=over 4
798		2883
799	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2884	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
800		2885
801	This function combines a simple timer and an I/O watcher, calls your	2886	This function combines a simple timer and an I/O watcher, calls your
802	callback on whichever event happens first and automatically stop both	2887	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	2888	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	2889	or timeout without having to allocate/configure/start/stop/free one or
805	more watchers yourself.	2890	more watchers yourself.
806		2891
807	If C<fd> is less than 0, then no I/O watcher will be started and events	2892	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	2893	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
809	C<events> set will be craeted and started.	2894	the given C<fd> and C<events> set will be created and started.
810		2895
811	If C<timeout> is less than 0, then no timeout watcher will be	2896	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	2897	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	2898	repeat = 0) will be started. C<0> is a valid timeout.
814	dubious value.
815		2899
816	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	2900	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	2901	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>	2902	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
819	value passed to C<ev_once>:	2903	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2904	a timeout and an io event at the same time - you probably should give io
		2905	events precedence.
820		2906
		2907	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		2908
821	static void stdin_ready (int revents, void *arg)	2909	static void stdin_ready (int revents, void *arg)
822	{	2910	{
823	if (revents & EV_TIMEOUT)
824	/* doh, nothing entered */;
825	else if (revents & EV_READ)	2911	if (revents & EV_READ)
826	/* stdin might have data for us, joy! */;	2912	/* stdin might have data for us, joy! */;
		2913	else if (revents & EV_TIMEOUT)
		2914	/* doh, nothing entered */;
827	}	2915	}
828		2916
829	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	2917	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
830		2918
831	=item ev_feed_event (loop, watcher, int events)	2919	=item ev_feed_event (struct ev_loop *, watcher *, int revents)
832		2920
833	Feeds the given event set into the event loop, as if the specified event	2921	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	2922	had happened for the specified watcher (which must be a pointer to an
835	initialised but not necessarily started event watcher).	2923	initialised but not necessarily started event watcher).
836		2924
837	=item ev_feed_fd_event (loop, int fd, int revents)	2925	=item ev_feed_fd_event (struct ev_loop *, int fd, int revents)
838		2926
839	Feed an event on the given fd, as if a file descriptor backend detected	2927	Feed an event on the given fd, as if a file descriptor backend detected
840	the given events it.	2928	the given events it.
841		2929
842	=item ev_feed_signal_event (loop, int signum)	2930	=item ev_feed_signal_event (struct ev_loop *loop, int signum)
843		2931
844	Feed an event as if the given signal occured (loop must be the default loop!).	2932	Feed an event as if the given signal occurred (C<loop> must be the default
		2933	loop!).
845		2934
846	=back	2935	=back
		2936
847		2937
848	=head1 LIBEVENT EMULATION	2938	=head1 LIBEVENT EMULATION
849		2939
850	Libev offers a compatibility emulation layer for libevent. It cannot	2940	Libev offers a compatibility emulation layer for libevent. It cannot
851	emulate the internals of libevent, so here are some usage hints:	2941	emulate the internals of libevent, so here are some usage hints:
…		…
863		2953
864	=item * Priorities are not currently supported. Initialising priorities	2954	=item * Priorities are not currently supported. Initialising priorities
865	will fail and all watchers will have the same priority, even though there	2955	will fail and all watchers will have the same priority, even though there
866	is an ev_pri field.	2956	is an ev_pri field.
867		2957
		2958	=item * In libevent, the last base created gets the signals, in libev, the
		2959	first base created (== the default loop) gets the signals.
		2960
868	=item * Other members are not supported.	2961	=item * Other members are not supported.
869		2962
870	=item * The libev emulation is I<not> ABI compatible to libevent, you need	2963	=item * The libev emulation is I<not> ABI compatible to libevent, you need
871	to use the libev header file and library.	2964	to use the libev header file and library.
872		2965
873	=back	2966	=back
874		2967
875	=head1 C++ SUPPORT	2968	=head1 C++ SUPPORT
876		2969
877	TBD.	2970	Libev comes with some simplistic wrapper classes for C++ that mainly allow
		2971	you to use some convenience methods to start/stop watchers and also change
		2972	the callback model to a model using method callbacks on objects.
		2973
		2974	To use it,
		2975
		2976	#include <ev++.h>
		2977
		2978	This automatically includes F<ev.h> and puts all of its definitions (many
		2979	of them macros) into the global namespace. All C++ specific things are
		2980	put into the C<ev> namespace. It should support all the same embedding
		2981	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
		2982
		2983	Care has been taken to keep the overhead low. The only data member the C++
		2984	classes add (compared to plain C-style watchers) is the event loop pointer
		2985	that the watcher is associated with (or no additional members at all if
		2986	you disable C<EV_MULTIPLICITY> when embedding libev).
		2987
		2988	Currently, functions, and static and non-static member functions can be
		2989	used as callbacks. Other types should be easy to add as long as they only
		2990	need one additional pointer for context. If you need support for other
		2991	types of functors please contact the author (preferably after implementing
		2992	it).
		2993
		2994	Here is a list of things available in the C<ev> namespace:
		2995
		2996	=over 4
		2997
		2998	=item C<ev::READ>, C<ev::WRITE> etc.
		2999
		3000	These are just enum values with the same values as the C<EV_READ> etc.
		3001	macros from F<ev.h>.
		3002
		3003	=item C<ev::tstamp>, C<ev::now>
		3004
		3005	Aliases to the same types/functions as with the C<ev_> prefix.
		3006
		3007	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
		3008
		3009	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
		3010	the same name in the C<ev> namespace, with the exception of C<ev_signal>
		3011	which is called C<ev::sig> to avoid clashes with the C<signal> macro
		3012	defines by many implementations.
		3013
		3014	All of those classes have these methods:
		3015
		3016	=over 4
		3017
		3018	=item ev::TYPE::TYPE ()
		3019
		3020	=item ev::TYPE::TYPE (struct ev_loop *)
		3021
		3022	=item ev::TYPE::~TYPE
		3023
		3024	The constructor (optionally) takes an event loop to associate the watcher
		3025	with. If it is omitted, it will use C<EV_DEFAULT>.
		3026
		3027	The constructor calls C<ev_init> for you, which means you have to call the
		3028	C<set> method before starting it.
		3029
		3030	It will not set a callback, however: You have to call the templated C<set>
		3031	method to set a callback before you can start the watcher.
		3032
		3033	(The reason why you have to use a method is a limitation in C++ which does
		3034	not allow explicit template arguments for constructors).
		3035
		3036	The destructor automatically stops the watcher if it is active.
		3037
		3038	=item w->set<class, &class::method> (object *)
		3039
		3040	This method sets the callback method to call. The method has to have a
		3041	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		3042	first argument and the C<revents> as second. The object must be given as
		3043	parameter and is stored in the C<data> member of the watcher.
		3044
		3045	This method synthesizes efficient thunking code to call your method from
		3046	the C callback that libev requires. If your compiler can inline your
		3047	callback (i.e. it is visible to it at the place of the C<set> call and
		3048	your compiler is good :), then the method will be fully inlined into the
		3049	thunking function, making it as fast as a direct C callback.
		3050
		3051	Example: simple class declaration and watcher initialisation
		3052
		3053	struct myclass
		3054	{
		3055	void io_cb (ev::io &w, int revents) { }
		3056	}
		3057
		3058	myclass obj;
		3059	ev::io iow;
		3060	iow.set <myclass, &myclass::io_cb> (&obj);
		3061
		3062	=item w->set (object *)
		3063
		3064	This is an B<experimental> feature that might go away in a future version.
		3065
		3066	This is a variation of a method callback - leaving out the method to call
		3067	will default the method to C<operator ()>, which makes it possible to use
		3068	functor objects without having to manually specify the C<operator ()> all
		3069	the time. Incidentally, you can then also leave out the template argument
		3070	list.
		3071
		3072	The C<operator ()> method prototype must be C<void operator ()(watcher &w,
		3073	int revents)>.
		3074
		3075	See the method-C<set> above for more details.
		3076
		3077	Example: use a functor object as callback.
		3078
		3079	struct myfunctor
		3080	{
		3081	void operator() (ev::io &w, int revents)
		3082	{
		3083	...
		3084	}
		3085	}
		3086
		3087	myfunctor f;
		3088
		3089	ev::io w;
		3090	w.set (&f);
		3091
		3092	=item w->set<function> (void *data = 0)
		3093
		3094	Also sets a callback, but uses a static method or plain function as
		3095	callback. The optional C<data> argument will be stored in the watcher's
		3096	C<data> member and is free for you to use.
		3097
		3098	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		3099
		3100	See the method-C<set> above for more details.
		3101
		3102	Example: Use a plain function as callback.
		3103
		3104	static void io_cb (ev::io &w, int revents) { }
		3105	iow.set <io_cb> ();
		3106
		3107	=item w->set (struct ev_loop *)
		3108
		3109	Associates a different C<struct ev_loop> with this watcher. You can only
		3110	do this when the watcher is inactive (and not pending either).
		3111
		3112	=item w->set ([arguments])
		3113
		3114	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
		3115	called at least once. Unlike the C counterpart, an active watcher gets
		3116	automatically stopped and restarted when reconfiguring it with this
		3117	method.
		3118
		3119	=item w->start ()
		3120
		3121	Starts the watcher. Note that there is no C<loop> argument, as the
		3122	constructor already stores the event loop.
		3123
		3124	=item w->stop ()
		3125
		3126	Stops the watcher if it is active. Again, no C<loop> argument.
		3127
		3128	=item w->again () (C<ev::timer>, C<ev::periodic> only)
		3129
		3130	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
		3131	C<ev_TYPE_again> function.
		3132
		3133	=item w->sweep () (C<ev::embed> only)
		3134
		3135	Invokes C<ev_embed_sweep>.
		3136
		3137	=item w->update () (C<ev::stat> only)
		3138
		3139	Invokes C<ev_stat_stat>.
		3140
		3141	=back
		3142
		3143	=back
		3144
		3145	Example: Define a class with an IO and idle watcher, start one of them in
		3146	the constructor.
		3147
		3148	class myclass
		3149	{
		3150	ev::io io ; void io_cb (ev::io &w, int revents);
		3151	ev::idle idle; void idle_cb (ev::idle &w, int revents);
		3152
		3153	myclass (int fd)
		3154	{
		3155	io .set <myclass, &myclass::io_cb > (this);
		3156	idle.set <myclass, &myclass::idle_cb> (this);
		3157
		3158	io.start (fd, ev::READ);
		3159	}
		3160	};
		3161
		3162
		3163	=head1 OTHER LANGUAGE BINDINGS
		3164
		3165	Libev does not offer other language bindings itself, but bindings for a
		3166	number of languages exist in the form of third-party packages. If you know
		3167	any interesting language binding in addition to the ones listed here, drop
		3168	me a note.
		3169
		3170	=over 4
		3171
		3172	=item Perl
		3173
		3174	The EV module implements the full libev API and is actually used to test
		3175	libev. EV is developed together with libev. Apart from the EV core module,
		3176	there are additional modules that implement libev-compatible interfaces
		3177	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		3178	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		3179	and C<EV::Glib>).
		3180
		3181	It can be found and installed via CPAN, its homepage is at
		3182	L<http://software.schmorp.de/pkg/EV>.
		3183
		3184	=item Python
		3185
		3186	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		3187	seems to be quite complete and well-documented.
		3188
		3189	=item Ruby
		3190
		3191	Tony Arcieri has written a ruby extension that offers access to a subset
		3192	of the libev API and adds file handle abstractions, asynchronous DNS and
		3193	more on top of it. It can be found via gem servers. Its homepage is at
		3194	L<http://rev.rubyforge.org/>.
		3195
		3196	Roger Pack reports that using the link order C<-lws2_32 -lmsvcrt-ruby-190>
		3197	makes rev work even on mingw.
		3198
		3199	=item Haskell
		3200
		3201	A haskell binding to libev is available at
		3202	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
		3203
		3204	=item D
		3205
		3206	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3207	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3208
		3209	=item Ocaml
		3210
		3211	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3212	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3213
		3214	=back
		3215
		3216
		3217	=head1 MACRO MAGIC
		3218
		3219	Libev can be compiled with a variety of options, the most fundamental
		3220	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
		3221	functions and callbacks have an initial C<struct ev_loop *> argument.
		3222
		3223	To make it easier to write programs that cope with either variant, the
		3224	following macros are defined:
		3225
		3226	=over 4
		3227
		3228	=item C<EV_A>, C<EV_A_>
		3229
		3230	This provides the loop I<argument> for functions, if one is required ("ev
		3231	loop argument"). The C<EV_A> form is used when this is the sole argument,
		3232	C<EV_A_> is used when other arguments are following. Example:
		3233
		3234	ev_unref (EV_A);
		3235	ev_timer_add (EV_A_ watcher);
		3236	ev_loop (EV_A_ 0);
		3237
		3238	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
		3239	which is often provided by the following macro.
		3240
		3241	=item C<EV_P>, C<EV_P_>
		3242
		3243	This provides the loop I<parameter> for functions, if one is required ("ev
		3244	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
		3245	C<EV_P_> is used when other parameters are following. Example:
		3246
		3247	// this is how ev_unref is being declared
		3248	static void ev_unref (EV_P);
		3249
		3250	// this is how you can declare your typical callback
		3251	static void cb (EV_P_ ev_timer *w, int revents)
		3252
		3253	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
		3254	suitable for use with C<EV_A>.
		3255
		3256	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
		3257
		3258	Similar to the other two macros, this gives you the value of the default
		3259	loop, if multiple loops are supported ("ev loop default").
		3260
		3261	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3262
		3263	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3264	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3265	is undefined when the default loop has not been initialised by a previous
		3266	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3267
		3268	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3269	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3270
		3271	=back
		3272
		3273	Example: Declare and initialise a check watcher, utilising the above
		3274	macros so it will work regardless of whether multiple loops are supported
		3275	or not.
		3276
		3277	static void
		3278	check_cb (EV_P_ ev_timer *w, int revents)
		3279	{
		3280	ev_check_stop (EV_A_ w);
		3281	}
		3282
		3283	ev_check check;
		3284	ev_check_init (&check, check_cb);
		3285	ev_check_start (EV_DEFAULT_ &check);
		3286	ev_loop (EV_DEFAULT_ 0);
		3287
		3288	=head1 EMBEDDING
		3289
		3290	Libev can (and often is) directly embedded into host
		3291	applications. Examples of applications that embed it include the Deliantra
		3292	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
		3293	and rxvt-unicode.
		3294
		3295	The goal is to enable you to just copy the necessary files into your
		3296	source directory without having to change even a single line in them, so
		3297	you can easily upgrade by simply copying (or having a checked-out copy of
		3298	libev somewhere in your source tree).
		3299
		3300	=head2 FILESETS
		3301
		3302	Depending on what features you need you need to include one or more sets of files
		3303	in your application.
		3304
		3305	=head3 CORE EVENT LOOP
		3306
		3307	To include only the libev core (all the C<ev_*> functions), with manual
		3308	configuration (no autoconf):
		3309
		3310	#define EV_STANDALONE 1
		3311	#include "ev.c"
		3312
		3313	This will automatically include F<ev.h>, too, and should be done in a
		3314	single C source file only to provide the function implementations. To use
		3315	it, do the same for F<ev.h> in all files wishing to use this API (best
		3316	done by writing a wrapper around F<ev.h> that you can include instead and
		3317	where you can put other configuration options):
		3318
		3319	#define EV_STANDALONE 1
		3320	#include "ev.h"
		3321
		3322	Both header files and implementation files can be compiled with a C++
		3323	compiler (at least, that's a stated goal, and breakage will be treated
		3324	as a bug).
		3325
		3326	You need the following files in your source tree, or in a directory
		3327	in your include path (e.g. in libev/ when using -Ilibev):
		3328
		3329	ev.h
		3330	ev.c
		3331	ev_vars.h
		3332	ev_wrap.h
		3333
		3334	ev_win32.c required on win32 platforms only
		3335
		3336	ev_select.c only when select backend is enabled (which is enabled by default)
		3337	ev_poll.c only when poll backend is enabled (disabled by default)
		3338	ev_epoll.c only when the epoll backend is enabled (disabled by default)
		3339	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
		3340	ev_port.c only when the solaris port backend is enabled (disabled by default)
		3341
		3342	F<ev.c> includes the backend files directly when enabled, so you only need
		3343	to compile this single file.
		3344
		3345	=head3 LIBEVENT COMPATIBILITY API
		3346
		3347	To include the libevent compatibility API, also include:
		3348
		3349	#include "event.c"
		3350
		3351	in the file including F<ev.c>, and:
		3352
		3353	#include "event.h"
		3354
		3355	in the files that want to use the libevent API. This also includes F<ev.h>.
		3356
		3357	You need the following additional files for this:
		3358
		3359	event.h
		3360	event.c
		3361
		3362	=head3 AUTOCONF SUPPORT
		3363
		3364	Instead of using C<EV_STANDALONE=1> and providing your configuration in
		3365	whatever way you want, you can also C<m4_include([libev.m4])> in your
		3366	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
		3367	include F<config.h> and configure itself accordingly.
		3368
		3369	For this of course you need the m4 file:
		3370
		3371	libev.m4
		3372
		3373	=head2 PREPROCESSOR SYMBOLS/MACROS
		3374
		3375	Libev can be configured via a variety of preprocessor symbols you have to
		3376	define before including any of its files. The default in the absence of
		3377	autoconf is documented for every option.
		3378
		3379	=over 4
		3380
		3381	=item EV_STANDALONE
		3382
		3383	Must always be C<1> if you do not use autoconf configuration, which
		3384	keeps libev from including F<config.h>, and it also defines dummy
		3385	implementations for some libevent functions (such as logging, which is not
		3386	supported). It will also not define any of the structs usually found in
		3387	F<event.h> that are not directly supported by the libev core alone.
		3388
		3389	In stanbdalone mode, libev will still try to automatically deduce the
		3390	configuration, but has to be more conservative.
		3391
		3392	=item EV_USE_MONOTONIC
		3393
		3394	If defined to be C<1>, libev will try to detect the availability of the
		3395	monotonic clock option at both compile time and runtime. Otherwise no
		3396	use of the monotonic clock option will be attempted. If you enable this,
		3397	you usually have to link against librt or something similar. Enabling it
		3398	when the functionality isn't available is safe, though, although you have
		3399	to make sure you link against any libraries where the C<clock_gettime>
		3400	function is hiding in (often F<-lrt>). See also C<EV_USE_CLOCK_SYSCALL>.
		3401
		3402	=item EV_USE_REALTIME
		3403
		3404	If defined to be C<1>, libev will try to detect the availability of the
		3405	real-time clock option at compile time (and assume its availability
		3406	at runtime if successful). Otherwise no use of the real-time clock
		3407	option will be attempted. This effectively replaces C<gettimeofday>
		3408	by C<clock_get (CLOCK_REALTIME, ...)> and will not normally affect
		3409	correctness. See the note about libraries in the description of
		3410	C<EV_USE_MONOTONIC>, though. Defaults to the opposite value of
		3411	C<EV_USE_CLOCK_SYSCALL>.
		3412
		3413	=item EV_USE_CLOCK_SYSCALL
		3414
		3415	If defined to be C<1>, libev will try to use a direct syscall instead
		3416	of calling the system-provided C<clock_gettime> function. This option
		3417	exists because on GNU/Linux, C<clock_gettime> is in C<librt>, but C<librt>
		3418	unconditionally pulls in C<libpthread>, slowing down single-threaded
		3419	programs needlessly. Using a direct syscall is slightly slower (in
		3420	theory), because no optimised vdso implementation can be used, but avoids
		3421	the pthread dependency. Defaults to C<1> on GNU/Linux with glibc 2.x or
		3422	higher, as it simplifies linking (no need for C<-lrt>).
		3423
		3424	=item EV_USE_NANOSLEEP
		3425
		3426	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3427	and will use it for delays. Otherwise it will use C<select ()>.
		3428
		3429	=item EV_USE_EVENTFD
		3430
		3431	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3432	available and will probe for kernel support at runtime. This will improve
		3433	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3434	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3435	2.7 or newer, otherwise disabled.
		3436
		3437	=item EV_USE_SELECT
		3438
		3439	If undefined or defined to be C<1>, libev will compile in support for the
		3440	C<select>(2) backend. No attempt at auto-detection will be done: if no
		3441	other method takes over, select will be it. Otherwise the select backend
		3442	will not be compiled in.
		3443
		3444	=item EV_SELECT_USE_FD_SET
		3445
		3446	If defined to C<1>, then the select backend will use the system C<fd_set>
		3447	structure. This is useful if libev doesn't compile due to a missing
		3448	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout
		3449	on exotic systems. This usually limits the range of file descriptors to
		3450	some low limit such as 1024 or might have other limitations (winsocket
		3451	only allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation,
		3452	configures the maximum size of the C<fd_set>.
		3453
		3454	=item EV_SELECT_IS_WINSOCKET
		3455
		3456	When defined to C<1>, the select backend will assume that
		3457	select/socket/connect etc. don't understand file descriptors but
		3458	wants osf handles on win32 (this is the case when the select to
		3459	be used is the winsock select). This means that it will call
		3460	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
		3461	it is assumed that all these functions actually work on fds, even
		3462	on win32. Should not be defined on non-win32 platforms.
		3463
		3464	=item EV_FD_TO_WIN32_HANDLE
		3465
		3466	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3467	file descriptors to socket handles. When not defining this symbol (the
		3468	default), then libev will call C<_get_osfhandle>, which is usually
		3469	correct. In some cases, programs use their own file descriptor management,
		3470	in which case they can provide this function to map fds to socket handles.
		3471
		3472	=item EV_USE_POLL
		3473
		3474	If defined to be C<1>, libev will compile in support for the C<poll>(2)
		3475	backend. Otherwise it will be enabled on non-win32 platforms. It
		3476	takes precedence over select.
		3477
		3478	=item EV_USE_EPOLL
		3479
		3480	If defined to be C<1>, libev will compile in support for the Linux
		3481	C<epoll>(7) backend. Its availability will be detected at runtime,
		3482	otherwise another method will be used as fallback. This is the preferred
		3483	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3484	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3485
		3486	=item EV_USE_KQUEUE
		3487
		3488	If defined to be C<1>, libev will compile in support for the BSD style
		3489	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
		3490	otherwise another method will be used as fallback. This is the preferred
		3491	backend for BSD and BSD-like systems, although on most BSDs kqueue only
		3492	supports some types of fds correctly (the only platform we found that
		3493	supports ptys for example was NetBSD), so kqueue might be compiled in, but
		3494	not be used unless explicitly requested. The best way to use it is to find
		3495	out whether kqueue supports your type of fd properly and use an embedded
		3496	kqueue loop.
		3497
		3498	=item EV_USE_PORT
		3499
		3500	If defined to be C<1>, libev will compile in support for the Solaris
		3501	10 port style backend. Its availability will be detected at runtime,
		3502	otherwise another method will be used as fallback. This is the preferred
		3503	backend for Solaris 10 systems.
		3504
		3505	=item EV_USE_DEVPOLL
		3506
		3507	Reserved for future expansion, works like the USE symbols above.
		3508
		3509	=item EV_USE_INOTIFY
		3510
		3511	If defined to be C<1>, libev will compile in support for the Linux inotify
		3512	interface to speed up C<ev_stat> watchers. Its actual availability will
		3513	be detected at runtime. If undefined, it will be enabled if the headers
		3514	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3515
		3516	=item EV_ATOMIC_T
		3517
		3518	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3519	access is atomic with respect to other threads or signal contexts. No such
		3520	type is easily found in the C language, so you can provide your own type
		3521	that you know is safe for your purposes. It is used both for signal handler "locking"
		3522	as well as for signal and thread safety in C<ev_async> watchers.
		3523
		3524	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3525	(from F<signal.h>), which is usually good enough on most platforms.
		3526
		3527	=item EV_H
		3528
		3529	The name of the F<ev.h> header file used to include it. The default if
		3530	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
		3531	used to virtually rename the F<ev.h> header file in case of conflicts.
		3532
		3533	=item EV_CONFIG_H
		3534
		3535	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
		3536	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
		3537	C<EV_H>, above.
		3538
		3539	=item EV_EVENT_H
		3540
		3541	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
		3542	of how the F<event.h> header can be found, the default is C<"event.h">.
		3543
		3544	=item EV_PROTOTYPES
		3545
		3546	If defined to be C<0>, then F<ev.h> will not define any function
		3547	prototypes, but still define all the structs and other symbols. This is
		3548	occasionally useful if you want to provide your own wrapper functions
		3549	around libev functions.
		3550
		3551	=item EV_MULTIPLICITY
		3552
		3553	If undefined or defined to C<1>, then all event-loop-specific functions
		3554	will have the C<struct ev_loop *> as first argument, and you can create
		3555	additional independent event loops. Otherwise there will be no support
		3556	for multiple event loops and there is no first event loop pointer
		3557	argument. Instead, all functions act on the single default loop.
		3558
		3559	=item EV_MINPRI
		3560
		3561	=item EV_MAXPRI
		3562
		3563	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3564	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3565	provide for more priorities by overriding those symbols (usually defined
		3566	to be C<-2> and C<2>, respectively).
		3567
		3568	When doing priority-based operations, libev usually has to linearly search
		3569	all the priorities, so having many of them (hundreds) uses a lot of space
		3570	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3571	fine.
		3572
		3573	If your embedding application does not need any priorities, defining these
		3574	both to C<0> will save some memory and CPU.
		3575
		3576	=item EV_PERIODIC_ENABLE
		3577
		3578	If undefined or defined to be C<1>, then periodic timers are supported. If
		3579	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3580	code.
		3581
		3582	=item EV_IDLE_ENABLE
		3583
		3584	If undefined or defined to be C<1>, then idle watchers are supported. If
		3585	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3586	code.
		3587
		3588	=item EV_EMBED_ENABLE
		3589
		3590	If undefined or defined to be C<1>, then embed watchers are supported. If
		3591	defined to be C<0>, then they are not. Embed watchers rely on most other
		3592	watcher types, which therefore must not be disabled.
		3593
		3594	=item EV_STAT_ENABLE
		3595
		3596	If undefined or defined to be C<1>, then stat watchers are supported. If
		3597	defined to be C<0>, then they are not.
		3598
		3599	=item EV_FORK_ENABLE
		3600
		3601	If undefined or defined to be C<1>, then fork watchers are supported. If
		3602	defined to be C<0>, then they are not.
		3603
		3604	=item EV_ASYNC_ENABLE
		3605
		3606	If undefined or defined to be C<1>, then async watchers are supported. If
		3607	defined to be C<0>, then they are not.
		3608
		3609	=item EV_MINIMAL
		3610
		3611	If you need to shave off some kilobytes of code at the expense of some
		3612	speed, define this symbol to C<1>. Currently this is used to override some
		3613	inlining decisions, saves roughly 30% code size on amd64. It also selects a
		3614	much smaller 2-heap for timer management over the default 4-heap.
		3615
		3616	=item EV_PID_HASHSIZE
		3617
		3618	C<ev_child> watchers use a small hash table to distribute workload by
		3619	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
		3620	than enough. If you need to manage thousands of children you might want to
		3621	increase this value (I<must> be a power of two).
		3622
		3623	=item EV_INOTIFY_HASHSIZE
		3624
		3625	C<ev_stat> watchers use a small hash table to distribute workload by
		3626	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		3627	usually more than enough. If you need to manage thousands of C<ev_stat>
		3628	watchers you might want to increase this value (I<must> be a power of
		3629	two).
		3630
		3631	=item EV_USE_4HEAP
		3632
		3633	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3634	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3635	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3636	faster performance with many (thousands) of watchers.
		3637
		3638	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3639	(disabled).
		3640
		3641	=item EV_HEAP_CACHE_AT
		3642
		3643	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3644	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3645	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3646	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3647	but avoids random read accesses on heap changes. This improves performance
		3648	noticeably with many (hundreds) of watchers.
		3649
		3650	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3651	(disabled).
		3652
		3653	=item EV_VERIFY
		3654
		3655	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3656	be done: If set to C<0>, no internal verification code will be compiled
		3657	in. If set to C<1>, then verification code will be compiled in, but not
		3658	called. If set to C<2>, then the internal verification code will be
		3659	called once per loop, which can slow down libev. If set to C<3>, then the
		3660	verification code will be called very frequently, which will slow down
		3661	libev considerably.
		3662
		3663	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3664	C<0>.
		3665
		3666	=item EV_COMMON
		3667
		3668	By default, all watchers have a C<void *data> member. By redefining
		3669	this macro to a something else you can include more and other types of
		3670	members. You have to define it each time you include one of the files,
		3671	though, and it must be identical each time.
		3672
		3673	For example, the perl EV module uses something like this:
		3674
		3675	#define EV_COMMON \
		3676	SV *self; /* contains this struct */ \
		3677	SV *cb_sv, *fh /* note no trailing ";" */
		3678
		3679	=item EV_CB_DECLARE (type)
		3680
		3681	=item EV_CB_INVOKE (watcher, revents)
		3682
		3683	=item ev_set_cb (ev, cb)
		3684
		3685	Can be used to change the callback member declaration in each watcher,
		3686	and the way callbacks are invoked and set. Must expand to a struct member
		3687	definition and a statement, respectively. See the F<ev.h> header file for
		3688	their default definitions. One possible use for overriding these is to
		3689	avoid the C<struct ev_loop *> as first argument in all cases, or to use
		3690	method calls instead of plain function calls in C++.
		3691
		3692	=back
		3693
		3694	=head2 EXPORTED API SYMBOLS
		3695
		3696	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3697	exported symbols, you can use the provided F<Symbol.*> files which list
		3698	all public symbols, one per line:
		3699
		3700	Symbols.ev for libev proper
		3701	Symbols.event for the libevent emulation
		3702
		3703	This can also be used to rename all public symbols to avoid clashes with
		3704	multiple versions of libev linked together (which is obviously bad in
		3705	itself, but sometimes it is inconvenient to avoid this).
		3706
		3707	A sed command like this will create wrapper C<#define>'s that you need to
		3708	include before including F<ev.h>:
		3709
		3710	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3711
		3712	This would create a file F<wrap.h> which essentially looks like this:
		3713
		3714	#define ev_backend myprefix_ev_backend
		3715	#define ev_check_start myprefix_ev_check_start
		3716	#define ev_check_stop myprefix_ev_check_stop
		3717	...
		3718
		3719	=head2 EXAMPLES
		3720
		3721	For a real-world example of a program the includes libev
		3722	verbatim, you can have a look at the EV perl module
		3723	(L<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
		3724	the F<libev/> subdirectory and includes them in the F<EV/EVAPI.h> (public
		3725	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
		3726	will be compiled. It is pretty complex because it provides its own header
		3727	file.
		3728
		3729	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
		3730	that everybody includes and which overrides some configure choices:
		3731
		3732	#define EV_MINIMAL 1
		3733	#define EV_USE_POLL 0
		3734	#define EV_MULTIPLICITY 0
		3735	#define EV_PERIODIC_ENABLE 0
		3736	#define EV_STAT_ENABLE 0
		3737	#define EV_FORK_ENABLE 0
		3738	#define EV_CONFIG_H <config.h>
		3739	#define EV_MINPRI 0
		3740	#define EV_MAXPRI 0
		3741
		3742	#include "ev++.h"
		3743
		3744	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
		3745
		3746	#include "ev_cpp.h"
		3747	#include "ev.c"
		3748
		3749	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
		3750
		3751	=head2 THREADS AND COROUTINES
		3752
		3753	=head3 THREADS
		3754
		3755	All libev functions are reentrant and thread-safe unless explicitly
		3756	documented otherwise, but libev implements no locking itself. This means
		3757	that you can use as many loops as you want in parallel, as long as there
		3758	are no concurrent calls into any libev function with the same loop
		3759	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		3760	of course): libev guarantees that different event loops share no data
		3761	structures that need any locking.
		3762
		3763	Or to put it differently: calls with different loop parameters can be done
		3764	concurrently from multiple threads, calls with the same loop parameter
		3765	must be done serially (but can be done from different threads, as long as
		3766	only one thread ever is inside a call at any point in time, e.g. by using
		3767	a mutex per loop).
		3768
		3769	Specifically to support threads (and signal handlers), libev implements
		3770	so-called C<ev_async> watchers, which allow some limited form of
		3771	concurrency on the same event loop, namely waking it up "from the
		3772	outside".
		3773
		3774	If you want to know which design (one loop, locking, or multiple loops
		3775	without or something else still) is best for your problem, then I cannot
		3776	help you, but here is some generic advice:
		3777
		3778	=over 4
		3779
		3780	=item * most applications have a main thread: use the default libev loop
		3781	in that thread, or create a separate thread running only the default loop.
		3782
		3783	This helps integrating other libraries or software modules that use libev
		3784	themselves and don't care/know about threading.
		3785
		3786	=item * one loop per thread is usually a good model.
		3787
		3788	Doing this is almost never wrong, sometimes a better-performance model
		3789	exists, but it is always a good start.
		3790
		3791	=item * other models exist, such as the leader/follower pattern, where one
		3792	loop is handed through multiple threads in a kind of round-robin fashion.
		3793
		3794	Choosing a model is hard - look around, learn, know that usually you can do
		3795	better than you currently do :-)
		3796
		3797	=item * often you need to talk to some other thread which blocks in the
		3798	event loop.
		3799
		3800	C<ev_async> watchers can be used to wake them up from other threads safely
		3801	(or from signal contexts...).
		3802
		3803	An example use would be to communicate signals or other events that only
		3804	work in the default loop by registering the signal watcher with the
		3805	default loop and triggering an C<ev_async> watcher from the default loop
		3806	watcher callback into the event loop interested in the signal.
		3807
		3808	=back
		3809
		3810	=head3 COROUTINES
		3811
		3812	Libev is very accommodating to coroutines ("cooperative threads"):
		3813	libev fully supports nesting calls to its functions from different
		3814	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		3815	different coroutines, and switch freely between both coroutines running the
		3816	loop, as long as you don't confuse yourself). The only exception is that
		3817	you must not do this from C<ev_periodic> reschedule callbacks.
		3818
		3819	Care has been taken to ensure that libev does not keep local state inside
		3820	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		3821	they do not call any callbacks.
		3822
		3823	=head2 COMPILER WARNINGS
		3824
		3825	Depending on your compiler and compiler settings, you might get no or a
		3826	lot of warnings when compiling libev code. Some people are apparently
		3827	scared by this.
		3828
		3829	However, these are unavoidable for many reasons. For one, each compiler
		3830	has different warnings, and each user has different tastes regarding
		3831	warning options. "Warn-free" code therefore cannot be a goal except when
		3832	targeting a specific compiler and compiler-version.
		3833
		3834	Another reason is that some compiler warnings require elaborate
		3835	workarounds, or other changes to the code that make it less clear and less
		3836	maintainable.
		3837
		3838	And of course, some compiler warnings are just plain stupid, or simply
		3839	wrong (because they don't actually warn about the condition their message
		3840	seems to warn about). For example, certain older gcc versions had some
		3841	warnings that resulted an extreme number of false positives. These have
		3842	been fixed, but some people still insist on making code warn-free with
		3843	such buggy versions.
		3844
		3845	While libev is written to generate as few warnings as possible,
		3846	"warn-free" code is not a goal, and it is recommended not to build libev
		3847	with any compiler warnings enabled unless you are prepared to cope with
		3848	them (e.g. by ignoring them). Remember that warnings are just that:
		3849	warnings, not errors, or proof of bugs.
		3850
		3851
		3852	=head2 VALGRIND
		3853
		3854	Valgrind has a special section here because it is a popular tool that is
		3855	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		3856
		3857	If you think you found a bug (memory leak, uninitialised data access etc.)
		3858	in libev, then check twice: If valgrind reports something like:
		3859
		3860	==2274== definitely lost: 0 bytes in 0 blocks.
		3861	==2274== possibly lost: 0 bytes in 0 blocks.
		3862	==2274== still reachable: 256 bytes in 1 blocks.
		3863
		3864	Then there is no memory leak, just as memory accounted to global variables
		3865	is not a memleak - the memory is still being referenced, and didn't leak.
		3866
		3867	Similarly, under some circumstances, valgrind might report kernel bugs
		3868	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3869	although an acceptable workaround has been found here), or it might be
		3870	confused.
		3871
		3872	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		3873	make it into some kind of religion.
		3874
		3875	If you are unsure about something, feel free to contact the mailing list
		3876	with the full valgrind report and an explanation on why you think this
		3877	is a bug in libev (best check the archives, too :). However, don't be
		3878	annoyed when you get a brisk "this is no bug" answer and take the chance
		3879	of learning how to interpret valgrind properly.
		3880
		3881	If you need, for some reason, empty reports from valgrind for your project
		3882	I suggest using suppression lists.
		3883
		3884
		3885	=head1 PORTABILITY NOTES
		3886
		3887	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		3888
		3889	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		3890	requires, and its I/O model is fundamentally incompatible with the POSIX
		3891	model. Libev still offers limited functionality on this platform in
		3892	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		3893	descriptors. This only applies when using Win32 natively, not when using
		3894	e.g. cygwin.
		3895
		3896	Lifting these limitations would basically require the full
		3897	re-implementation of the I/O system. If you are into these kinds of
		3898	things, then note that glib does exactly that for you in a very portable
		3899	way (note also that glib is the slowest event library known to man).
		3900
		3901	There is no supported compilation method available on windows except
		3902	embedding it into other applications.
		3903
		3904	Not a libev limitation but worth mentioning: windows apparently doesn't
		3905	accept large writes: instead of resulting in a partial write, windows will
		3906	either accept everything or return C<ENOBUFS> if the buffer is too large,
		3907	so make sure you only write small amounts into your sockets (less than a
		3908	megabyte seems safe, but this apparently depends on the amount of memory
		3909	available).
		3910
		3911	Due to the many, low, and arbitrary limits on the win32 platform and
		3912	the abysmal performance of winsockets, using a large number of sockets
		3913	is not recommended (and not reasonable). If your program needs to use
		3914	more than a hundred or so sockets, then likely it needs to use a totally
		3915	different implementation for windows, as libev offers the POSIX readiness
		3916	notification model, which cannot be implemented efficiently on windows
		3917	(Microsoft monopoly games).
		3918
		3919	A typical way to use libev under windows is to embed it (see the embedding
		3920	section for details) and use the following F<evwrap.h> header file instead
		3921	of F<ev.h>:
		3922
		3923	#define EV_STANDALONE /* keeps ev from requiring config.h */
		3924	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		3925
		3926	#include "ev.h"
		3927
		3928	And compile the following F<evwrap.c> file into your project (make sure
		3929	you do I<not> compile the F<ev.c> or any other embedded source files!):
		3930
		3931	#include "evwrap.h"
		3932	#include "ev.c"
		3933
		3934	=over 4
		3935
		3936	=item The winsocket select function
		3937
		3938	The winsocket C<select> function doesn't follow POSIX in that it
		3939	requires socket I<handles> and not socket I<file descriptors> (it is
		3940	also extremely buggy). This makes select very inefficient, and also
		3941	requires a mapping from file descriptors to socket handles (the Microsoft
		3942	C runtime provides the function C<_open_osfhandle> for this). See the
		3943	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		3944	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		3945
		3946	The configuration for a "naked" win32 using the Microsoft runtime
		3947	libraries and raw winsocket select is:
		3948
		3949	#define EV_USE_SELECT 1
		3950	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		3951
		3952	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3953	complexity in the O(n²) range when using win32.
		3954
		3955	=item Limited number of file descriptors
		3956
		3957	Windows has numerous arbitrary (and low) limits on things.
		3958
		3959	Early versions of winsocket's select only supported waiting for a maximum
		3960	of C<64> handles (probably owning to the fact that all windows kernels
		3961	can only wait for C<64> things at the same time internally; Microsoft
		3962	recommends spawning a chain of threads and wait for 63 handles and the
		3963	previous thread in each. Great).
		3964
		3965	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3966	to some high number (e.g. C<2048>) before compiling the winsocket select
		3967	call (which might be in libev or elsewhere, for example, perl does its own
		3968	select emulation on windows).
		3969
		3970	Another limit is the number of file descriptors in the Microsoft runtime
		3971	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3972	or something like this inside Microsoft). You can increase this by calling
		3973	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3974	arbitrary limit), but is broken in many versions of the Microsoft runtime
		3975	libraries.
		3976
		3977	This might get you to about C<512> or C<2048> sockets (depending on
		3978	windows version and/or the phase of the moon). To get more, you need to
		3979	wrap all I/O functions and provide your own fd management, but the cost of
		3980	calling select (O(n²)) will likely make this unworkable.
		3981
		3982	=back
		3983
		3984	=head2 PORTABILITY REQUIREMENTS
		3985
		3986	In addition to a working ISO-C implementation and of course the
		3987	backend-specific APIs, libev relies on a few additional extensions:
		3988
		3989	=over 4
		3990
		3991	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		3992	calling conventions regardless of C<ev_watcher_type *>.
		3993
		3994	Libev assumes not only that all watcher pointers have the same internal
		3995	structure (guaranteed by POSIX but not by ISO C for example), but it also
		3996	assumes that the same (machine) code can be used to call any watcher
		3997	callback: The watcher callbacks have different type signatures, but libev
		3998	calls them using an C<ev_watcher *> internally.
		3999
		4000	=item C<sig_atomic_t volatile> must be thread-atomic as well
		4001
		4002	The type C<sig_atomic_t volatile> (or whatever is defined as
		4003	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		4004	threads. This is not part of the specification for C<sig_atomic_t>, but is
		4005	believed to be sufficiently portable.
		4006
		4007	=item C<sigprocmask> must work in a threaded environment
		4008
		4009	Libev uses C<sigprocmask> to temporarily block signals. This is not
		4010	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		4011	pthread implementations will either allow C<sigprocmask> in the "main
		4012	thread" or will block signals process-wide, both behaviours would
		4013	be compatible with libev. Interaction between C<sigprocmask> and
		4014	C<pthread_sigmask> could complicate things, however.
		4015
		4016	The most portable way to handle signals is to block signals in all threads
		4017	except the initial one, and run the default loop in the initial thread as
		4018	well.
		4019
		4020	=item C<long> must be large enough for common memory allocation sizes
		4021
		4022	To improve portability and simplify its API, libev uses C<long> internally
		4023	instead of C<size_t> when allocating its data structures. On non-POSIX
		4024	systems (Microsoft...) this might be unexpectedly low, but is still at
		4025	least 31 bits everywhere, which is enough for hundreds of millions of
		4026	watchers.
		4027
		4028	=item C<double> must hold a time value in seconds with enough accuracy
		4029
		4030	The type C<double> is used to represent timestamps. It is required to
		4031	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		4032	enough for at least into the year 4000. This requirement is fulfilled by
		4033	implementations implementing IEEE 754 (basically all existing ones).
		4034
		4035	=back
		4036
		4037	If you know of other additional requirements drop me a note.
		4038
		4039
		4040	=head1 ALGORITHMIC COMPLEXITIES
		4041
		4042	In this section the complexities of (many of) the algorithms used inside
		4043	libev will be documented. For complexity discussions about backends see
		4044	the documentation for C<ev_default_init>.
		4045
		4046	All of the following are about amortised time: If an array needs to be
		4047	extended, libev needs to realloc and move the whole array, but this
		4048	happens asymptotically rarer with higher number of elements, so O(1) might
		4049	mean that libev does a lengthy realloc operation in rare cases, but on
		4050	average it is much faster and asymptotically approaches constant time.
		4051
		4052	=over 4
		4053
		4054	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
		4055
		4056	This means that, when you have a watcher that triggers in one hour and
		4057	there are 100 watchers that would trigger before that, then inserting will
		4058	have to skip roughly seven (C<ld 100>) of these watchers.
		4059
		4060	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
		4061
		4062	That means that changing a timer costs less than removing/adding them,
		4063	as only the relative motion in the event queue has to be paid for.
		4064
		4065	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		4066
		4067	These just add the watcher into an array or at the head of a list.
		4068
		4069	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		4070
		4071	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		4072
		4073	These watchers are stored in lists, so they need to be walked to find the
		4074	correct watcher to remove. The lists are usually short (you don't usually
		4075	have many watchers waiting for the same fd or signal: one is typical, two
		4076	is rare).
		4077
		4078	=item Finding the next timer in each loop iteration: O(1)
		4079
		4080	By virtue of using a binary or 4-heap, the next timer is always found at a
		4081	fixed position in the storage array.
		4082
		4083	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		4084
		4085	A change means an I/O watcher gets started or stopped, which requires
		4086	libev to recalculate its status (and possibly tell the kernel, depending
		4087	on backend and whether C<ev_io_set> was used).
		4088
		4089	=item Activating one watcher (putting it into the pending state): O(1)
		4090
		4091	=item Priority handling: O(number_of_priorities)
		4092
		4093	Priorities are implemented by allocating some space for each
		4094	priority. When doing priority-based operations, libev usually has to
		4095	linearly search all the priorities, but starting/stopping and activating
		4096	watchers becomes O(1) with respect to priority handling.
		4097
		4098	=item Sending an ev_async: O(1)
		4099
		4100	=item Processing ev_async_send: O(number_of_async_watchers)
		4101
		4102	=item Processing signals: O(max_signal_number)
		4103
		4104	Sending involves a system call I<iff> there were no other C<ev_async_send>
		4105	calls in the current loop iteration. Checking for async and signal events
		4106	involves iterating over all running async watchers or all signal numbers.
		4107
		4108	=back
		4109
		4110
		4111	=head1 GLOSSARY
		4112
		4113	=over 4
		4114
		4115	=item active
		4116
		4117	A watcher is active as long as it has been started (has been attached to
		4118	an event loop) but not yet stopped (disassociated from the event loop).
		4119
		4120	=item application
		4121
		4122	In this document, an application is whatever is using libev.
		4123
		4124	=item callback
		4125
		4126	The address of a function that is called when some event has been
		4127	detected. Callbacks are being passed the event loop, the watcher that
		4128	received the event, and the actual event bitset.
		4129
		4130	=item callback invocation
		4131
		4132	The act of calling the callback associated with a watcher.
		4133
		4134	=item event
		4135
		4136	A change of state of some external event, such as data now being available
		4137	for reading on a file descriptor, time having passed or simply not having
		4138	any other events happening anymore.
		4139
		4140	In libev, events are represented as single bits (such as C<EV_READ> or
		4141	C<EV_TIMEOUT>).
		4142
		4143	=item event library
		4144
		4145	A software package implementing an event model and loop.
		4146
		4147	=item event loop
		4148
		4149	An entity that handles and processes external events and converts them
		4150	into callback invocations.
		4151
		4152	=item event model
		4153
		4154	The model used to describe how an event loop handles and processes
		4155	watchers and events.
		4156
		4157	=item pending
		4158
		4159	A watcher is pending as soon as the corresponding event has been detected,
		4160	and stops being pending as soon as the watcher will be invoked or its
		4161	pending status is explicitly cleared by the application.
		4162
		4163	A watcher can be pending, but not active. Stopping a watcher also clears
		4164	its pending status.
		4165
		4166	=item real time
		4167
		4168	The physical time that is observed. It is apparently strictly monotonic :)
		4169
		4170	=item wall-clock time
		4171
		4172	The time and date as shown on clocks. Unlike real time, it can actually
		4173	be wrong and jump forwards and backwards, e.g. when the you adjust your
		4174	clock.
		4175
		4176	=item watcher
		4177
		4178	A data structure that describes interest in certain events. Watchers need
		4179	to be started (attached to an event loop) before they can receive events.
		4180
		4181	=item watcher invocation
		4182
		4183	The act of calling the callback associated with a watcher.
		4184
		4185	=back
878		4186
879	=head1 AUTHOR	4187	=head1 AUTHOR
880		4188
881	Marc Lehmann <libev@schmorp.de>.	4189	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
882		4190

Diff Legend

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