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 1.206 by root, Tue Oct 28 12:31:38 2008 UTC

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

Diff Legend

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