ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.74 by root, Sat Dec 8 14:12:08 2007 UTC vs.
Revision 1.212 by root, Mon Nov 3 15:13:53 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		8
9	=head1 EXAMPLE PROGRAM	9	=head2 EXAMPLE PROGRAM
10		10
		11	// a single header file is required
11	#include <ev.h>	12	#include <ev.h>
12		13
		14	// every watcher type has its own typedef'd struct
		15	// with the name ev_TYPE
13	ev_io stdin_watcher;	16	ev_io stdin_watcher;
14	ev_timer timeout_watcher;	17	ev_timer timeout_watcher;
15		18
16	/* called when data readable on stdin */	19	// all watcher callbacks have a similar signature
		20	// this callback is called when data is readable on stdin
17	static void	21	static void
18	stdin_cb (EV_P_ struct ev_io *w, int revents)	22	stdin_cb (EV_P_ ev_io *w, int revents)
19	{	23	{
20	/* puts ("stdin ready"); */	24	puts ("stdin ready");
21	ev_io_stop (EV_A_ w); /* just a syntax example */	25	// for one-shot events, one must manually stop the watcher
22	ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */	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);
23	}	31	}
24		32
		33	// another callback, this time for a time-out
25	static void	34	static void
26	timeout_cb (EV_P_ struct ev_timer *w, int revents)	35	timeout_cb (EV_P_ ev_timer *w, int revents)
27	{	36	{
28	/* puts ("timeout"); */	37	puts ("timeout");
29	ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */	38	// this causes the innermost ev_loop to stop iterating
		39	ev_unloop (EV_A_ EVUNLOOP_ONE);
30	}	40	}
31		41
32	int	42	int
33	main (void)	43	main (void)
34	{	44	{
		45	// use the default event loop unless you have special needs
35	struct ev_loop *loop = ev_default_loop (0);	46	ev_loop *loop = ev_default_loop (0);
36		47
37	/* initialise an io watcher, then start it */	48	// initialise an io watcher, then start it
		49	// this one will watch for stdin to become readable
38	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	50	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
39	ev_io_start (loop, &stdin_watcher);	51	ev_io_start (loop, &stdin_watcher);
40		52
		53	// initialise a timer watcher, then start it
41	/* simple non-repeating 5.5 second timeout */	54	// simple non-repeating 5.5 second timeout
42	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);	55	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
43	ev_timer_start (loop, &timeout_watcher);	56	ev_timer_start (loop, &timeout_watcher);
44		57
45	/* loop till timeout or data ready */	58	// now wait for events to arrive
46	ev_loop (loop, 0);	59	ev_loop (loop, 0);
47		60
		61	// unloop was called, so exit
48	return 0;	62	return 0;
49	}	63	}
50		64
51	=head1 DESCRIPTION	65	=head1 DESCRIPTION
52		66
53	The newest version of this document is also available as a html-formatted	67	The newest version of this document is also available as an html-formatted
54	web page you might find easier to navigate when reading it for the first	68	web page you might find easier to navigate when reading it for the first
55	time: L<http://cvs.schmorp.de/libev/ev.html>.	69	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
56		70
57	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
58	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
59	these event sources and provide your program with events.	73	these event sources and provide your program with events.
60		74
61	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
62	(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
63	communicate events via a callback mechanism.	77	communicate events via a callback mechanism.
…		…
65	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
66	watchers>, which are relatively small C structures you initialise with the	80	watchers>, which are relatively small C structures you initialise with the
67	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
68	watcher.	82	watcher.
69		83
70	=head1 FEATURES	84	=head2 FEATURES
71		85
72	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	86	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	87	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	88	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers	89	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
…		…
82		96
83	It also is quite fast (see this	97	It also is quite fast (see this
84	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent	98	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85	for example).	99	for example).
86		100
87	=head1 CONVENTIONS	101	=head2 CONVENTIONS
88		102
89	Libev is very configurable. In this manual the default configuration will	103	Libev is very configurable. In this manual the default (and most common)
90	be described, which supports multiple event loops. For more info about	104	configuration will be described, which supports multiple event loops. For
91	various configuration options please have a look at B<EMBED> section in	105	more info about various configuration options please have a look at
92	this manual. If libev was configured without support for multiple event	106	B<EMBED> section in this manual. If libev was configured without support
93	loops, then all functions taking an initial argument of name C<loop>	107	for multiple event loops, then all functions taking an initial argument of
94	(which is always of type C<struct ev_loop *>) will not have this argument.	108	name C<loop> (which is always of type C<ev_loop *>) will not have
		109	this argument.
95		110
96	=head1 TIME REPRESENTATION	111	=head2 TIME REPRESENTATION
97		112
98	Libev represents time as a single floating point number, representing the	113	Libev represents time as a single floating point number, representing the
99	(fractional) number of seconds since the (POSIX) epoch (somewhere near	114	(fractional) number of seconds since the (POSIX) epoch (somewhere near
100	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
101	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
102	to the C<double> type in C, and when you need to do any calculations on	117	to the C<double> type in C, and when you need to do any calculations on
103	it, you should treat it as such.	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
104		142
105	=head1 GLOBAL FUNCTIONS	143	=head1 GLOBAL FUNCTIONS
106		144
107	These functions can be called anytime, even before initialising the	145	These functions can be called anytime, even before initialising the
108	library in any way.	146	library in any way.
…		…
113		151
114	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
115	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
116	you actually want to know.	154	you actually want to know.
117		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
118	=item int ev_version_major ()	162	=item int ev_version_major ()
119		163
120	=item int ev_version_minor ()	164	=item int ev_version_minor ()
121		165
122	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
123	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
124	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
125	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
126	version of the library your program was compiled against.	170	version of the library your program was compiled against.
127		171
		172	These version numbers refer to the ABI version of the library, not the
		173	release version.
		174
128	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,
129	as this indicates an incompatible change. Minor versions are usually	176	as this indicates an incompatible change. Minor versions are usually
130	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
131	not a problem.	178	not a problem.
132		179
133	Example: Make sure we haven't accidentally been linked against the wrong	180	Example: Make sure we haven't accidentally been linked against the wrong
134	version.	181	version.
135		182
136	assert (("libev version mismatch",	183	assert (("libev version mismatch",
137	ev_version_major () == EV_VERSION_MAJOR	184	ev_version_major () == EV_VERSION_MAJOR
138	&& ev_version_minor () >= EV_VERSION_MINOR));	185	&& ev_version_minor () >= EV_VERSION_MINOR));
139		186
140	=item unsigned int ev_supported_backends ()	187	=item unsigned int ev_supported_backends ()
141		188
142	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	189	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
143	value) compiled into this binary of libev (independent of their	190	value) compiled into this binary of libev (independent of their
…		…
145	a description of the set values.	192	a description of the set values.
146		193
147	Example: make sure we have the epoll method, because yeah this is cool and	194	Example: make sure we have the epoll method, because yeah this is cool and
148	a must have and can we have a torrent of it please!!!11	195	a must have and can we have a torrent of it please!!!11
149		196
150	assert (("sorry, no epoll, no sex",	197	assert (("sorry, no epoll, no sex",
151	ev_supported_backends () & EVBACKEND_EPOLL));	198	ev_supported_backends () & EVBACKEND_EPOLL));
152		199
153	=item unsigned int ev_recommended_backends ()	200	=item unsigned int ev_recommended_backends ()
154		201
155	Return the set of all backends compiled into this binary of libev and also	202	Return the set of all backends compiled into this binary of libev and also
156	recommended for this platform. This set is often smaller than the one	203	recommended for this platform. This set is often smaller than the one
157	returned by C<ev_supported_backends>, as for example kqueue is broken on	204	returned by C<ev_supported_backends>, as for example kqueue is broken on
158	most BSDs and will not be autodetected unless you explicitly request it	205	most BSDs and will not be auto-detected unless you explicitly request it
159	(assuming you know what you are doing). This is the set of backends that	206	(assuming you know what you are doing). This is the set of backends that
160	libev will probe for if you specify no backends explicitly.	207	libev will probe for if you specify no backends explicitly.
161		208
162	=item unsigned int ev_embeddable_backends ()	209	=item unsigned int ev_embeddable_backends ()
163		210
…		…
167	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	214	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
168	recommended ones.	215	recommended ones.
169		216
170	See the description of C<ev_embed> watchers for more info.	217	See the description of C<ev_embed> watchers for more info.
171		218
172	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	219	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
173		220
174	Sets the allocation function to use (the prototype is similar - the	221	Sets the allocation function to use (the prototype is similar - the
175	semantics is identical - to the realloc C function). It is used to	222	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
176	allocate and free memory (no surprises here). If it returns zero when	223	used to allocate and free memory (no surprises here). If it returns zero
177	memory needs to be allocated, the library might abort or take some	224	when memory needs to be allocated (C<size != 0>), the library might abort
178	potentially destructive action. The default is your system realloc	225	or take some potentially destructive action.
179	function.	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.
180		230
181	You could override this function in high-availability programs to, say,	231	You could override this function in high-availability programs to, say,
182	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,
183	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.
184		234
185	Example: Replace the libev allocator with one that waits a bit and then	235	Example: Replace the libev allocator with one that waits a bit and then
186	retries).	236	retries (example requires a standards-compliant C<realloc>).
187		237
188	static void *	238	static void *
189	persistent_realloc (void *ptr, size_t size)	239	persistent_realloc (void *ptr, size_t size)
190	{	240	{
191	for (;;)	241	for (;;)
…		…
200	}	250	}
201		251
202	...	252	...
203	ev_set_allocator (persistent_realloc);	253	ev_set_allocator (persistent_realloc);
204		254
205	=item ev_set_syserr_cb (void (*cb)(const char *msg));	255	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
206		256
207	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
208	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
209	indicating the system call or subsystem causing the problem. If this	259	indicating the system call or subsystem causing the problem. If this
210	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
211	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
212	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
213	(such as abort).	263	(such as abort).
214		264
215	Example: This is basically the same thing that libev does internally, too.	265	Example: This is basically the same thing that libev does internally, too.
…		…
226		276
227	=back	277	=back
228		278
229	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	279	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
230		280
231	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>
232	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>
233	events, and dynamically created loops which do not.	283	I<function>).
234		284
235	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
236	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
237	create, you also create another event loop. Libev itself does no locking	287	not.
238	whatsoever, so if you mix calls to the same event loop in different
239	threads, make sure you lock (this is usually a bad idea, though, even if
240	done correctly, because it's hideous and inefficient).
241		288
242	=over 4	289	=over 4
243		290
244	=item struct ev_loop *ev_default_loop (unsigned int flags)	291	=item struct ev_loop *ev_default_loop (unsigned int flags)
245		292
…		…
249	flags. If that is troubling you, check C<ev_backend ()> afterwards).	296	flags. If that is troubling you, check C<ev_backend ()> afterwards).
250		297
251	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
252	function.	299	function.
253		300
		301	Note that this function is I<not> thread-safe, so if you want to use it
		302	from multiple threads, you have to lock (note also that this is unlikely,
		303	as loops cannot be shared easily between threads anyway).
		304
		305	The default loop is the only loop that can handle C<ev_signal> and
		306	C<ev_child> watchers, and to do this, it always registers a handler
		307	for C<SIGCHLD>. If this is a problem for your application you can either
		308	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
		309	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
		310	C<ev_default_init>.
		311
254	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
255	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	313	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
256		314
257	The following flags are supported:	315	The following flags are supported:
258		316
…		…
263	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
264	thing, believe me).	322	thing, believe me).
265		323
266	=item C<EVFLAG_NOENV>	324	=item C<EVFLAG_NOENV>
267		325
268	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
269	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
270	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	328	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
271	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
272	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
273	around bugs.	331	around bugs.
…		…
279	enabling this flag.	337	enabling this flag.
280		338
281	This works by calling C<getpid ()> on every iteration of the loop,	339	This works by calling C<getpid ()> on every iteration of the loop,
282	and thus this might slow down your event loop if you do a lot of loop	340	and thus this might slow down your event loop if you do a lot of loop
283	iterations and little real work, but is usually not noticeable (on my	341	iterations and little real work, but is usually not noticeable (on my
284	Linux system for example, C<getpid> is actually a simple 5-insn sequence	342	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
285	without a syscall and thus I<very> fast, but my Linux system also has	343	without a system call and thus I<very> fast, but my GNU/Linux system also has
286	C<pthread_atfork> which is even faster).	344	C<pthread_atfork> which is even faster).
287		345
288	The big advantage of this flag is that you can forget about fork (and	346	The big advantage of this flag is that you can forget about fork (and
289	forget about forgetting to tell libev about forking) when you use this	347	forget about forgetting to tell libev about forking) when you use this
290	flag.	348	flag.
291		349
292	This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS>	350	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
293	environment variable.	351	environment variable.
294		352
295	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	353	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
296		354
297	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
298	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,
299	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
300	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
301	the fastest backend for a low number of fds.	359	usually the fastest backend for a low number of (low-numbered :) fds.
		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).
302		371
303	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	372	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
304		373
305	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
306	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
307	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
308	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.
		380
		381	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		382	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
309		383
310	=item C<EVBACKEND_EPOLL> (value 4, Linux)	384	=item C<EVBACKEND_EPOLL> (value 4, Linux)
311		385
312	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,
313	but it scales phenomenally better. While poll and select usually scale like	387	but it scales phenomenally better. While poll and select usually scale
314	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),
315	either O(1) or O(active_fds).	389	epoll scales either O(1) or O(active_fds).
316		390
		391	The epoll mechanism deserves honorable mention as the most misdesigned
		392	of the more advanced event mechanisms: mere annoyances include silently
		393	dropping file descriptors, requiring a system call per change per file
		394	descriptor (and unnecessary guessing of parameters), problems with dup and
		395	so on. The biggest issue is fork races, however - if a program forks then
		396	I<both> parent and child process have to recreate the epoll set, which can
		397	take considerable time (one syscall per file descriptor) and is of course
		398	hard to detect.
		399
		400	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but
		401	of course I<doesn't>, and epoll just loves to report events for totally
		402	I<different> file descriptors (even already closed ones, so one cannot
		403	even remove them from the set) than registered in the set (especially
		404	on SMP systems). Libev tries to counter these spurious notifications by
		405	employing an additional generation counter and comparing that against the
		406	events to filter out spurious ones, recreating the set when required.
		407
317	While stopping and starting an I/O watcher in the same iteration will	408	While stopping, setting and starting an I/O watcher in the same iteration
318	result in some caching, there is still a syscall per such incident	409	will result in some caching, there is still a system call per such
319	(because the fd could point to a different file description now), so its	410	incident (because the same I<file descriptor> could point to a different
320	best to avoid that. Also, dup()ed file descriptors might not work very	411	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
321	well if you register events for both fds.	412	file descriptors might not work very well if you register events for both
		413	file descriptors.
322		414
323	Please note that epoll sometimes generates spurious notifications, so you	415	Best performance from this backend is achieved by not unregistering all
324	need to use non-blocking I/O or other means to avoid blocking when no data	416	watchers for a file descriptor until it has been closed, if possible,
325	(or space) is available.	417	i.e. keep at least one watcher active per fd at all times. Stopping and
		418	starting a watcher (without re-setting it) also usually doesn't cause
		419	extra overhead. A fork can both result in spurious notifications as well
		420	as in libev having to destroy and recreate the epoll object, which can
		421	take considerable time and thus should be avoided.
		422
		423	While nominally embeddable in other event loops, this feature is broken in
		424	all kernel versions tested so far.
		425
		426	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		427	C<EVBACKEND_POLL>.
326		428
327	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	429	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
328		430
329	Kqueue deserves special mention, as at the time of this writing, it	431	Kqueue deserves special mention, as at the time of this writing, it
330	was broken on all BSDs except NetBSD (usually it doesn't work with	432	was broken on all BSDs except NetBSD (usually it doesn't work reliably
331	anything but sockets and pipes, except on Darwin, where of course its	433	with anything but sockets and pipes, except on Darwin, where of course
332	completely useless). For this reason its not being "autodetected"	434	it's completely useless). Unlike epoll, however, whose brokenness
		435	is by design, these kqueue bugs can (and eventually will) be fixed
		436	without API changes to existing programs. For this reason it's not being
333	unless you explicitly specify it explicitly in the flags (i.e. using	437	"auto-detected" unless you explicitly specify it in the flags (i.e. using
334	C<EVBACKEND_KQUEUE>).	438	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		439	system like NetBSD.
		440
		441	You still can embed kqueue into a normal poll or select backend and use it
		442	only for sockets (after having made sure that sockets work with kqueue on
		443	the target platform). See C<ev_embed> watchers for more info.
335		444
336	It scales in the same way as the epoll backend, but the interface to the	445	It scales in the same way as the epoll backend, but the interface to the
337	kernel is more efficient (which says nothing about its actual speed, of	446	kernel is more efficient (which says nothing about its actual speed, of
338	course). While starting and stopping an I/O watcher does not cause an	447	course). While stopping, setting and starting an I/O watcher does never
339	extra syscall as with epoll, it still adds up to four event changes per	448	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
340	incident, so its best to avoid that.	449	two event changes per incident. Support for C<fork ()> is very bad (but
		450	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		451	cases
		452
		453	This backend usually performs well under most conditions.
		454
		455	While nominally embeddable in other event loops, this doesn't work
		456	everywhere, so you might need to test for this. And since it is broken
		457	almost everywhere, you should only use it when you have a lot of sockets
		458	(for which it usually works), by embedding it into another event loop
		459	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and, did I mention it,
		460	using it only for sockets.
		461
		462	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		463	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		464	C<NOTE_EOF>.
341		465
342	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	466	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
343		467
344	This is not implemented yet (and might never be).	468	This is not implemented yet (and might never be, unless you send me an
		469	implementation). According to reports, C</dev/poll> only supports sockets
		470	and is not embeddable, which would limit the usefulness of this backend
		471	immensely.
345		472
346	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	473	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
347		474
348	This uses the Solaris 10 port mechanism. As with everything on Solaris,	475	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
349	it's really slow, but it still scales very well (O(active_fds)).	476	it's really slow, but it still scales very well (O(active_fds)).
350		477
351	Please note that solaris ports can result in a lot of spurious	478	Please note that Solaris event ports can deliver a lot of spurious
352	notifications, so you need to use non-blocking I/O or other means to avoid	479	notifications, so you need to use non-blocking I/O or other means to avoid
353	blocking when no data (or space) is available.	480	blocking when no data (or space) is available.
		481
		482	While this backend scales well, it requires one system call per active
		483	file descriptor per loop iteration. For small and medium numbers of file
		484	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		485	might perform better.
		486
		487	On the positive side, with the exception of the spurious readiness
		488	notifications, this backend actually performed fully to specification
		489	in all tests and is fully embeddable, which is a rare feat among the
		490	OS-specific backends (I vastly prefer correctness over speed hacks).
		491
		492	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		493	C<EVBACKEND_POLL>.
354		494
355	=item C<EVBACKEND_ALL>	495	=item C<EVBACKEND_ALL>
356		496
357	Try all backends (even potentially broken ones that wouldn't be tried	497	Try all backends (even potentially broken ones that wouldn't be tried
358	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	498	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
359	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	499	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
360		500
		501	It is definitely not recommended to use this flag.
		502
361	=back	503	=back
362		504
363	If one or more of these are ored into the flags value, then only these	505	If one or more of these are or'ed into the flags value, then only these
364	backends will be tried (in the reverse order as given here). If none are	506	backends will be tried (in the reverse order as listed here). If none are
365	specified, most compiled-in backend will be tried, usually in reverse	507	specified, all backends in C<ev_recommended_backends ()> will be tried.
366	order of their flag values :)
367		508
368	The most typical usage is like this:	509	Example: This is the most typical usage.
369		510
370	if (!ev_default_loop (0))	511	if (!ev_default_loop (0))
371	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");	512	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
372		513
373	Restrict libev to the select and poll backends, and do not allow	514	Example: Restrict libev to the select and poll backends, and do not allow
374	environment settings to be taken into account:	515	environment settings to be taken into account:
375		516
376	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);	517	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
377		518
378	Use whatever libev has to offer, but make sure that kqueue is used if	519	Example: Use whatever libev has to offer, but make sure that kqueue is
379	available (warning, breaks stuff, best use only with your own private	520	used if available (warning, breaks stuff, best use only with your own
380	event loop and only if you know the OS supports your types of fds):	521	private event loop and only if you know the OS supports your types of
		522	fds):
381		523
382	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);	524	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
383		525
384	=item struct ev_loop *ev_loop_new (unsigned int flags)	526	=item struct ev_loop *ev_loop_new (unsigned int flags)
385		527
386	Similar to C<ev_default_loop>, but always creates a new event loop that is	528	Similar to C<ev_default_loop>, but always creates a new event loop that is
387	always distinct from the default loop. Unlike the default loop, it cannot	529	always distinct from the default loop. Unlike the default loop, it cannot
388	handle signal and child watchers, and attempts to do so will be greeted by	530	handle signal and child watchers, and attempts to do so will be greeted by
389	undefined behaviour (or a failed assertion if assertions are enabled).	531	undefined behaviour (or a failed assertion if assertions are enabled).
390		532
		533	Note that this function I<is> thread-safe, and the recommended way to use
		534	libev with threads is indeed to create one loop per thread, and using the
		535	default loop in the "main" or "initial" thread.
		536
391	Example: Try to create a event loop that uses epoll and nothing else.	537	Example: Try to create a event loop that uses epoll and nothing else.
392		538
393	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	539	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
394	if (!epoller)	540	if (!epoller)
395	fatal ("no epoll found here, maybe it hides under your chair");	541	fatal ("no epoll found here, maybe it hides under your chair");
396		542
397	=item ev_default_destroy ()	543	=item ev_default_destroy ()
398		544
399	Destroys the default loop again (frees all memory and kernel state	545	Destroys the default loop again (frees all memory and kernel state
400	etc.). None of the active event watchers will be stopped in the normal	546	etc.). None of the active event watchers will be stopped in the normal
401	sense, so e.g. C<ev_is_active> might still return true. It is your	547	sense, so e.g. C<ev_is_active> might still return true. It is your
402	responsibility to either stop all watchers cleanly yoursef I<before>	548	responsibility to either stop all watchers cleanly yourself I<before>
403	calling this function, or cope with the fact afterwards (which is usually	549	calling this function, or cope with the fact afterwards (which is usually
404	the easiest thing, youc na just ignore the watchers and/or C<free ()> them	550	the easiest thing, you can just ignore the watchers and/or C<free ()> them
405	for example).	551	for example).
		552
		553	Note that certain global state, such as signal state (and installed signal
		554	handlers), will not be freed by this function, and related watchers (such
		555	as signal and child watchers) would need to be stopped manually.
		556
		557	In general it is not advisable to call this function except in the
		558	rare occasion where you really need to free e.g. the signal handling
		559	pipe fds. If you need dynamically allocated loops it is better to use
		560	C<ev_loop_new> and C<ev_loop_destroy>).
406		561
407	=item ev_loop_destroy (loop)	562	=item ev_loop_destroy (loop)
408		563
409	Like C<ev_default_destroy>, but destroys an event loop created by an	564	Like C<ev_default_destroy>, but destroys an event loop created by an
410	earlier call to C<ev_loop_new>.	565	earlier call to C<ev_loop_new>.
411		566
412	=item ev_default_fork ()	567	=item ev_default_fork ()
413		568
		569	This function sets a flag that causes subsequent C<ev_loop> iterations
414	This function reinitialises the kernel state for backends that have	570	to reinitialise the kernel state for backends that have one. Despite the
415	one. Despite the name, you can call it anytime, but it makes most sense	571	name, you can call it anytime, but it makes most sense after forking, in
416	after forking, in either the parent or child process (or both, but that	572	the child process (or both child and parent, but that again makes little
417	again makes little sense).	573	sense). You I<must> call it in the child before using any of the libev
		574	functions, and it will only take effect at the next C<ev_loop> iteration.
418		575
419	You I<must> call this function in the child process after forking if and	576	On the other hand, you only need to call this function in the child
420	only if you want to use the event library in both processes. If you just	577	process if and only if you want to use the event library in the child. If
421	fork+exec, you don't have to call it.	578	you just fork+exec, you don't have to call it at all.
422		579
423	The function itself is quite fast and it's usually not a problem to call	580	The function itself is quite fast and it's usually not a problem to call
424	it just in case after a fork. To make this easy, the function will fit in	581	it just in case after a fork. To make this easy, the function will fit in
425	quite nicely into a call to C<pthread_atfork>:	582	quite nicely into a call to C<pthread_atfork>:
426		583
427	pthread_atfork (0, 0, ev_default_fork);	584	pthread_atfork (0, 0, ev_default_fork);
428		585
429	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use
430	without calling this function, so if you force one of those backends you
431	do not need to care.
432
433	=item ev_loop_fork (loop)	586	=item ev_loop_fork (loop)
434		587
435	Like C<ev_default_fork>, but acts on an event loop created by	588	Like C<ev_default_fork>, but acts on an event loop created by
436	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	589	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
437	after fork, and how you do this is entirely your own problem.	590	after fork that you want to re-use in the child, and how you do this is
		591	entirely your own problem.
		592
		593	=item int ev_is_default_loop (loop)
		594
		595	Returns true when the given loop is, in fact, the default loop, and false
		596	otherwise.
438		597
439	=item unsigned int ev_loop_count (loop)	598	=item unsigned int ev_loop_count (loop)
440		599
441	Returns the count of loop iterations for the loop, which is identical to	600	Returns the count of loop iterations for the loop, which is identical to
442	the number of times libev did poll for new events. It starts at C<0> and	601	the number of times libev did poll for new events. It starts at C<0> and
…		…
455		614
456	Returns the current "event loop time", which is the time the event loop	615	Returns the current "event loop time", which is the time the event loop
457	received events and started processing them. This timestamp does not	616	received events and started processing them. This timestamp does not
458	change as long as callbacks are being processed, and this is also the base	617	change as long as callbacks are being processed, and this is also the base
459	time used for relative timers. You can treat it as the timestamp of the	618	time used for relative timers. You can treat it as the timestamp of the
460	event occuring (or more correctly, libev finding out about it).	619	event occurring (or more correctly, libev finding out about it).
		620
		621	=item ev_now_update (loop)
		622
		623	Establishes the current time by querying the kernel, updating the time
		624	returned by C<ev_now ()> in the progress. This is a costly operation and
		625	is usually done automatically within C<ev_loop ()>.
		626
		627	This function is rarely useful, but when some event callback runs for a
		628	very long time without entering the event loop, updating libev's idea of
		629	the current time is a good idea.
		630
		631	See also "The special problem of time updates" in the C<ev_timer> section.
461		632
462	=item ev_loop (loop, int flags)	633	=item ev_loop (loop, int flags)
463		634
464	Finally, this is it, the event handler. This function usually is called	635	Finally, this is it, the event handler. This function usually is called
465	after you initialised all your watchers and you want to start handling	636	after you initialised all your watchers and you want to start handling
…		…
468	If the flags argument is specified as C<0>, it will not return until	639	If the flags argument is specified as C<0>, it will not return until
469	either no event watchers are active anymore or C<ev_unloop> was called.	640	either no event watchers are active anymore or C<ev_unloop> was called.
470		641
471	Please note that an explicit C<ev_unloop> is usually better than	642	Please note that an explicit C<ev_unloop> is usually better than
472	relying on all watchers to be stopped when deciding when a program has	643	relying on all watchers to be stopped when deciding when a program has
473	finished (especially in interactive programs), but having a program that	644	finished (especially in interactive programs), but having a program
474	automatically loops as long as it has to and no longer by virtue of	645	that automatically loops as long as it has to and no longer by virtue
475	relying on its watchers stopping correctly is a thing of beauty.	646	of relying on its watchers stopping correctly, that is truly a thing of
		647	beauty.
476		648
477	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	649	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
478	those events and any outstanding ones, but will not block your process in	650	those events and any already outstanding ones, but will not block your
479	case there are no events and will return after one iteration of the loop.	651	process in case there are no events and will return after one iteration of
		652	the loop.
480		653
481	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	654	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
482	neccessary) and will handle those and any outstanding ones. It will block	655	necessary) and will handle those and any already outstanding ones. It
483	your process until at least one new event arrives, and will return after	656	will block your process until at least one new event arrives (which could
484	one iteration of the loop. This is useful if you are waiting for some	657	be an event internal to libev itself, so there is no guarantee that a
485	external event in conjunction with something not expressible using other	658	user-registered callback will be called), and will return after one
		659	iteration of the loop.
		660
		661	This is useful if you are waiting for some external event in conjunction
		662	with something not expressible using other libev watchers (i.e. "roll your
486	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	663	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
487	usually a better approach for this kind of thing.	664	usually a better approach for this kind of thing.
488		665
489	Here are the gory details of what C<ev_loop> does:	666	Here are the gory details of what C<ev_loop> does:
490		667
491	* If there are no active watchers (reference count is zero), return.	668	- Before the first iteration, call any pending watchers.
492	- Queue prepare watchers and then call all outstanding watchers.	669	* If EVFLAG_FORKCHECK was used, check for a fork.
		670	- If a fork was detected (by any means), queue and call all fork watchers.
		671	- Queue and call all prepare watchers.
493	- If we have been forked, recreate the kernel state.	672	- If we have been forked, detach and recreate the kernel state
		673	as to not disturb the other process.
494	- Update the kernel state with all outstanding changes.	674	- Update the kernel state with all outstanding changes.
495	- Update the "event loop time".	675	- Update the "event loop time" (ev_now ()).
496	- Calculate for how long to block.	676	- Calculate for how long to sleep or block, if at all
		677	(active idle watchers, EVLOOP_NONBLOCK or not having
		678	any active watchers at all will result in not sleeping).
		679	- Sleep if the I/O and timer collect interval say so.
497	- Block the process, waiting for any events.	680	- Block the process, waiting for any events.
498	- Queue all outstanding I/O (fd) events.	681	- Queue all outstanding I/O (fd) events.
499	- Update the "event loop time" and do time jump handling.	682	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
500	- Queue all outstanding timers.	683	- Queue all expired timers.
501	- Queue all outstanding periodics.	684	- Queue all expired periodics.
502	- If no events are pending now, queue all idle watchers.	685	- Unless any events are pending now, queue all idle watchers.
503	- Queue all check watchers.	686	- Queue all check watchers.
504	- Call all queued watchers in reverse order (i.e. check watchers first).	687	- Call all queued watchers in reverse order (i.e. check watchers first).
505	Signals and child watchers are implemented as I/O watchers, and will	688	Signals and child watchers are implemented as I/O watchers, and will
506	be handled here by queueing them when their watcher gets executed.	689	be handled here by queueing them when their watcher gets executed.
507	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	690	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
508	were used, return, otherwise continue with step *.	691	were used, or there are no active watchers, return, otherwise
		692	continue with step *.
509		693
510	Example: Queue some jobs and then loop until no events are outsanding	694	Example: Queue some jobs and then loop until no events are outstanding
511	anymore.	695	anymore.
512		696
513	... queue jobs here, make sure they register event watchers as long	697	... queue jobs here, make sure they register event watchers as long
514	... as they still have work to do (even an idle watcher will do..)	698	... as they still have work to do (even an idle watcher will do..)
515	ev_loop (my_loop, 0);	699	ev_loop (my_loop, 0);
516	... jobs done. yeah!	700	... jobs done or somebody called unloop. yeah!
517		701
518	=item ev_unloop (loop, how)	702	=item ev_unloop (loop, how)
519		703
520	Can be used to make a call to C<ev_loop> return early (but only after it	704	Can be used to make a call to C<ev_loop> return early (but only after it
521	has processed all outstanding events). The C<how> argument must be either	705	has processed all outstanding events). The C<how> argument must be either
522	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	706	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
523	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	707	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
524		708
		709	This "unloop state" will be cleared when entering C<ev_loop> again.
		710
		711	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
		712
525	=item ev_ref (loop)	713	=item ev_ref (loop)
526		714
527	=item ev_unref (loop)	715	=item ev_unref (loop)
528		716
529	Ref/unref can be used to add or remove a reference count on the event	717	Ref/unref can be used to add or remove a reference count on the event
530	loop: Every watcher keeps one reference, and as long as the reference	718	loop: Every watcher keeps one reference, and as long as the reference
531	count is nonzero, C<ev_loop> will not return on its own. If you have	719	count is nonzero, C<ev_loop> will not return on its own.
		720
532	a watcher you never unregister that should not keep C<ev_loop> from	721	If you have a watcher you never unregister that should not keep C<ev_loop>
533	returning, ev_unref() after starting, and ev_ref() before stopping it. For	722	from returning, call ev_unref() after starting, and ev_ref() before
		723	stopping it.
		724
534	example, libev itself uses this for its internal signal pipe: It is not	725	As an example, libev itself uses this for its internal signal pipe: It is
535	visible to the libev user and should not keep C<ev_loop> from exiting if	726	not visible to the libev user and should not keep C<ev_loop> from exiting
536	no event watchers registered by it are active. It is also an excellent	727	if no event watchers registered by it are active. It is also an excellent
537	way to do this for generic recurring timers or from within third-party	728	way to do this for generic recurring timers or from within third-party
538	libraries. Just remember to I<unref after start> and I<ref before stop>.	729	libraries. Just remember to I<unref after start> and I<ref before stop>
		730	(but only if the watcher wasn't active before, or was active before,
		731	respectively).
539		732
540	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	733	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
541	running when nothing else is active.	734	running when nothing else is active.
542		735
543	struct ev_signal exitsig;	736	ev_signal exitsig;
544	ev_signal_init (&exitsig, sig_cb, SIGINT);	737	ev_signal_init (&exitsig, sig_cb, SIGINT);
545	ev_signal_start (loop, &exitsig);	738	ev_signal_start (loop, &exitsig);
546	evf_unref (loop);	739	evf_unref (loop);
547		740
548	Example: For some weird reason, unregister the above signal handler again.	741	Example: For some weird reason, unregister the above signal handler again.
549		742
550	ev_ref (loop);	743	ev_ref (loop);
551	ev_signal_stop (loop, &exitsig);	744	ev_signal_stop (loop, &exitsig);
		745
		746	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		747
		748	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		749
		750	These advanced functions influence the time that libev will spend waiting
		751	for events. Both time intervals are by default C<0>, meaning that libev
		752	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		753	latency.
		754
		755	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		756	allows libev to delay invocation of I/O and timer/periodic callbacks
		757	to increase efficiency of loop iterations (or to increase power-saving
		758	opportunities).
		759
		760	The idea is that sometimes your program runs just fast enough to handle
		761	one (or very few) event(s) per loop iteration. While this makes the
		762	program responsive, it also wastes a lot of CPU time to poll for new
		763	events, especially with backends like C<select ()> which have a high
		764	overhead for the actual polling but can deliver many events at once.
		765
		766	By setting a higher I<io collect interval> you allow libev to spend more
		767	time collecting I/O events, so you can handle more events per iteration,
		768	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		769	C<ev_timer>) will be not affected. Setting this to a non-null value will
		770	introduce an additional C<ev_sleep ()> call into most loop iterations.
		771
		772	Likewise, by setting a higher I<timeout collect interval> you allow libev
		773	to spend more time collecting timeouts, at the expense of increased
		774	latency/jitter/inexactness (the watcher callback will be called
		775	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		776	value will not introduce any overhead in libev.
		777
		778	Many (busy) programs can usually benefit by setting the I/O collect
		779	interval to a value near C<0.1> or so, which is often enough for
		780	interactive servers (of course not for games), likewise for timeouts. It
		781	usually doesn't make much sense to set it to a lower value than C<0.01>,
		782	as this approaches the timing granularity of most systems.
		783
		784	Setting the I<timeout collect interval> can improve the opportunity for
		785	saving power, as the program will "bundle" timer callback invocations that
		786	are "near" in time together, by delaying some, thus reducing the number of
		787	times the process sleeps and wakes up again. Another useful technique to
		788	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		789	they fire on, say, one-second boundaries only.
		790
		791	=item ev_loop_verify (loop)
		792
		793	This function only does something when C<EV_VERIFY> support has been
		794	compiled in, which is the default for non-minimal builds. It tries to go
		795	through all internal structures and checks them for validity. If anything
		796	is found to be inconsistent, it will print an error message to standard
		797	error and call C<abort ()>.
		798
		799	This can be used to catch bugs inside libev itself: under normal
		800	circumstances, this function will never abort as of course libev keeps its
		801	data structures consistent.
552		802
553	=back	803	=back
554		804
555		805
556	=head1 ANATOMY OF A WATCHER	806	=head1 ANATOMY OF A WATCHER
		807
		808	In the following description, uppercase C<TYPE> in names stands for the
		809	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		810	watchers and C<ev_io_start> for I/O watchers.
557		811
558	A watcher is a structure that you create and register to record your	812	A watcher is a structure that you create and register to record your
559	interest in some event. For instance, if you want to wait for STDIN to	813	interest in some event. For instance, if you want to wait for STDIN to
560	become readable, you would create an C<ev_io> watcher for that:	814	become readable, you would create an C<ev_io> watcher for that:
561		815
562	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	816	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
563	{	817	{
564	ev_io_stop (w);	818	ev_io_stop (w);
565	ev_unloop (loop, EVUNLOOP_ALL);	819	ev_unloop (loop, EVUNLOOP_ALL);
566	}	820	}
567		821
568	struct ev_loop *loop = ev_default_loop (0);	822	struct ev_loop *loop = ev_default_loop (0);
		823
569	struct ev_io stdin_watcher;	824	ev_io stdin_watcher;
		825
570	ev_init (&stdin_watcher, my_cb);	826	ev_init (&stdin_watcher, my_cb);
571	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	827	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
572	ev_io_start (loop, &stdin_watcher);	828	ev_io_start (loop, &stdin_watcher);
		829
573	ev_loop (loop, 0);	830	ev_loop (loop, 0);
574		831
575	As you can see, you are responsible for allocating the memory for your	832	As you can see, you are responsible for allocating the memory for your
576	watcher structures (and it is usually a bad idea to do this on the stack,	833	watcher structures (and it is I<usually> a bad idea to do this on the
577	although this can sometimes be quite valid).	834	stack).
		835
		836	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		837	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
578		838
579	Each watcher structure must be initialised by a call to C<ev_init	839	Each watcher structure must be initialised by a call to C<ev_init
580	(watcher *, callback)>, which expects a callback to be provided. This	840	(watcher *, callback)>, which expects a callback to be provided. This
581	callback gets invoked each time the event occurs (or, in the case of io	841	callback gets invoked each time the event occurs (or, in the case of I/O
582	watchers, each time the event loop detects that the file descriptor given	842	watchers, each time the event loop detects that the file descriptor given
583	is readable and/or writable).	843	is readable and/or writable).
584		844
585	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	845	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
586	with arguments specific to this watcher type. There is also a macro	846	macro to configure it, with arguments specific to the watcher type. There
587	to combine initialisation and setting in one call: C<< ev_<type>_init	847	is also a macro to combine initialisation and setting in one call: C<<
588	(watcher *, callback, ...) >>.	848	ev_TYPE_init (watcher *, callback, ...) >>.
589		849
590	To make the watcher actually watch out for events, you have to start it	850	To make the watcher actually watch out for events, you have to start it
591	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	851	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
592	*) >>), and you can stop watching for events at any time by calling the	852	*) >>), and you can stop watching for events at any time by calling the
593	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	853	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
594		854
595	As long as your watcher is active (has been started but not stopped) you	855	As long as your watcher is active (has been started but not stopped) you
596	must not touch the values stored in it. Most specifically you must never	856	must not touch the values stored in it. Most specifically you must never
597	reinitialise it or call its C<set> macro.	857	reinitialise it or call its C<ev_TYPE_set> macro.
598		858
599	Each and every callback receives the event loop pointer as first, the	859	Each and every callback receives the event loop pointer as first, the
600	registered watcher structure as second, and a bitset of received events as	860	registered watcher structure as second, and a bitset of received events as
601	third argument.	861	third argument.
602		862
…		…
656	=item C<EV_FORK>	916	=item C<EV_FORK>
657		917
658	The event loop has been resumed in the child process after fork (see	918	The event loop has been resumed in the child process after fork (see
659	C<ev_fork>).	919	C<ev_fork>).
660		920
		921	=item C<EV_ASYNC>
		922
		923	The given async watcher has been asynchronously notified (see C<ev_async>).
		924
661	=item C<EV_ERROR>	925	=item C<EV_ERROR>
662		926
663	An unspecified error has occured, the watcher has been stopped. This might	927	An unspecified error has occurred, the watcher has been stopped. This might
664	happen because the watcher could not be properly started because libev	928	happen because the watcher could not be properly started because libev
665	ran out of memory, a file descriptor was found to be closed or any other	929	ran out of memory, a file descriptor was found to be closed or any other
		930	problem. Libev considers these application bugs.
		931
666	problem. You best act on it by reporting the problem and somehow coping	932	You best act on it by reporting the problem and somehow coping with the
667	with the watcher being stopped.	933	watcher being stopped. Note that well-written programs should not receive
		934	an error ever, so when your watcher receives it, this usually indicates a
		935	bug in your program.
668		936
669	Libev will usually signal a few "dummy" events together with an error,	937	Libev will usually signal a few "dummy" events together with an error, for
670	for example it might indicate that a fd is readable or writable, and if	938	example it might indicate that a fd is readable or writable, and if your
671	your callbacks is well-written it can just attempt the operation and cope	939	callbacks is well-written it can just attempt the operation and cope with
672	with the error from read() or write(). This will not work in multithreaded	940	the error from read() or write(). This will not work in multi-threaded
673	programs, though, so beware.	941	programs, though, as the fd could already be closed and reused for another
		942	thing, so beware.
674		943
675	=back	944	=back
676		945
677	=head2 GENERIC WATCHER FUNCTIONS	946	=head2 GENERIC WATCHER FUNCTIONS
678
679	In the following description, C<TYPE> stands for the watcher type,
680	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
681		947
682	=over 4	948	=over 4
683		949
684	=item C<ev_init> (ev_TYPE *watcher, callback)	950	=item C<ev_init> (ev_TYPE *watcher, callback)
685		951
…		…
691	which rolls both calls into one.	957	which rolls both calls into one.
692		958
693	You can reinitialise a watcher at any time as long as it has been stopped	959	You can reinitialise a watcher at any time as long as it has been stopped
694	(or never started) and there are no pending events outstanding.	960	(or never started) and there are no pending events outstanding.
695		961
696	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	962	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
697	int revents)>.	963	int revents)>.
		964
		965	Example: Initialise an C<ev_io> watcher in two steps.
		966
		967	ev_io w;
		968	ev_init (&w, my_cb);
		969	ev_io_set (&w, STDIN_FILENO, EV_READ);
698		970
699	=item C<ev_TYPE_set> (ev_TYPE *, [args])	971	=item C<ev_TYPE_set> (ev_TYPE *, [args])
700		972
701	This macro initialises the type-specific parts of a watcher. You need to	973	This macro initialises the type-specific parts of a watcher. You need to
702	call C<ev_init> at least once before you call this macro, but you can	974	call C<ev_init> at least once before you call this macro, but you can
…		…
705	difference to the C<ev_init> macro).	977	difference to the C<ev_init> macro).
706		978
707	Although some watcher types do not have type-specific arguments	979	Although some watcher types do not have type-specific arguments
708	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	980	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
709		981
		982	See C<ev_init>, above, for an example.
		983
710	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	984	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
711		985
712	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	986	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
713	calls into a single call. This is the most convinient method to initialise	987	calls into a single call. This is the most convenient method to initialise
714	a watcher. The same limitations apply, of course.	988	a watcher. The same limitations apply, of course.
		989
		990	Example: Initialise and set an C<ev_io> watcher in one step.
		991
		992	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
715		993
716	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	994	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)
717		995
718	Starts (activates) the given watcher. Only active watchers will receive	996	Starts (activates) the given watcher. Only active watchers will receive
719	events. If the watcher is already active nothing will happen.	997	events. If the watcher is already active nothing will happen.
720		998
		999	Example: Start the C<ev_io> watcher that is being abused as example in this
		1000	whole section.
		1001
		1002	ev_io_start (EV_DEFAULT_UC, &w);
		1003
721	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	1004	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)
722		1005
723	Stops the given watcher again (if active) and clears the pending	1006	Stops the given watcher if active, and clears the pending status (whether
		1007	the watcher was active or not).
		1008
724	status. It is possible that stopped watchers are pending (for example,	1009	It is possible that stopped watchers are pending - for example,
725	non-repeating timers are being stopped when they become pending), but	1010	non-repeating timers are being stopped when they become pending - but
726	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1011	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
727	you want to free or reuse the memory used by the watcher it is therefore a	1012	pending. If you want to free or reuse the memory used by the watcher it is
728	good idea to always call its C<ev_TYPE_stop> function.	1013	therefore a good idea to always call its C<ev_TYPE_stop> function.
729		1014
730	=item bool ev_is_active (ev_TYPE *watcher)	1015	=item bool ev_is_active (ev_TYPE *watcher)
731		1016
732	Returns a true value iff the watcher is active (i.e. it has been started	1017	Returns a true value iff the watcher is active (i.e. it has been started
733	and not yet been stopped). As long as a watcher is active you must not modify	1018	and not yet been stopped). As long as a watcher is active you must not modify
…		…
775	The default priority used by watchers when no priority has been set is	1060	The default priority used by watchers when no priority has been set is
776	always C<0>, which is supposed to not be too high and not be too low :).	1061	always C<0>, which is supposed to not be too high and not be too low :).
777		1062
778	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is	1063	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
779	fine, as long as you do not mind that the priority value you query might	1064	fine, as long as you do not mind that the priority value you query might
780	or might not have been adjusted to be within valid range.	1065	or might not have been clamped to the valid range.
781		1066
782	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1067	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
783		1068
784	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1069	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
785	C<loop> nor C<revents> need to be valid as long as the watcher callback	1070	C<loop> nor C<revents> need to be valid as long as the watcher callback
786	can deal with that fact.	1071	can deal with that fact, as both are simply passed through to the
		1072	callback.
787		1073
788	=item int ev_clear_pending (loop, ev_TYPE *watcher)	1074	=item int ev_clear_pending (loop, ev_TYPE *watcher)
789		1075
790	If the watcher is pending, this function returns clears its pending status	1076	If the watcher is pending, this function clears its pending status and
791	and returns its C<revents> bitset (as if its callback was invoked). If the	1077	returns its C<revents> bitset (as if its callback was invoked). If the
792	watcher isn't pending it does nothing and returns C<0>.	1078	watcher isn't pending it does nothing and returns C<0>.
793		1079
		1080	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1081	callback to be invoked, which can be accomplished with this function.
		1082
794	=back	1083	=back
795		1084
796		1085
797	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1086	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
798		1087
799	Each watcher has, by default, a member C<void *data> that you can change	1088	Each watcher has, by default, a member C<void *data> that you can change
800	and read at any time, libev will completely ignore it. This can be used	1089	and read at any time: libev will completely ignore it. This can be used
801	to associate arbitrary data with your watcher. If you need more data and	1090	to associate arbitrary data with your watcher. If you need more data and
802	don't want to allocate memory and store a pointer to it in that data	1091	don't want to allocate memory and store a pointer to it in that data
803	member, you can also "subclass" the watcher type and provide your own	1092	member, you can also "subclass" the watcher type and provide your own
804	data:	1093	data:
805		1094
806	struct my_io	1095	struct my_io
807	{	1096	{
808	struct ev_io io;	1097	ev_io io;
809	int otherfd;	1098	int otherfd;
810	void *somedata;	1099	void *somedata;
811	struct whatever *mostinteresting;	1100	struct whatever *mostinteresting;
812	}	1101	};
		1102
		1103	...
		1104	struct my_io w;
		1105	ev_io_init (&w.io, my_cb, fd, EV_READ);
813		1106
814	And since your callback will be called with a pointer to the watcher, you	1107	And since your callback will be called with a pointer to the watcher, you
815	can cast it back to your own type:	1108	can cast it back to your own type:
816		1109
817	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1110	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
818	{	1111	{
819	struct my_io *w = (struct my_io *)w_;	1112	struct my_io *w = (struct my_io *)w_;
820	...	1113	...
821	}	1114	}
822		1115
823	More interesting and less C-conformant ways of casting your callback type	1116	More interesting and less C-conformant ways of casting your callback type
824	instead have been omitted.	1117	instead have been omitted.
825		1118
826	Another common scenario is having some data structure with multiple	1119	Another common scenario is to use some data structure with multiple
827	watchers:	1120	embedded watchers:
828		1121
829	struct my_biggy	1122	struct my_biggy
830	{	1123	{
831	int some_data;	1124	int some_data;
832	ev_timer t1;	1125	ev_timer t1;
833	ev_timer t2;	1126	ev_timer t2;
834	}	1127	}
835		1128
836	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1129	In this case getting the pointer to C<my_biggy> is a bit more
837	you need to use C<offsetof>:	1130	complicated: Either you store the address of your C<my_biggy> struct
		1131	in the C<data> member of the watcher (for woozies), or you need to use
		1132	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1133	programmers):
838		1134
839	#include <stddef.h>	1135	#include <stddef.h>
840		1136
841	static void	1137	static void
842	t1_cb (EV_P_ struct ev_timer *w, int revents)	1138	t1_cb (EV_P_ ev_timer *w, int revents)
843	{	1139	{
844	struct my_biggy big = (struct my_biggy *	1140	struct my_biggy big = (struct my_biggy *
845	(((char *)w) - offsetof (struct my_biggy, t1));	1141	(((char *)w) - offsetof (struct my_biggy, t1));
846	}	1142	}
847		1143
848	static void	1144	static void
849	t2_cb (EV_P_ struct ev_timer *w, int revents)	1145	t2_cb (EV_P_ ev_timer *w, int revents)
850	{	1146	{
851	struct my_biggy big = (struct my_biggy *	1147	struct my_biggy big = (struct my_biggy *
852	(((char *)w) - offsetof (struct my_biggy, t2));	1148	(((char *)w) - offsetof (struct my_biggy, t2));
853	}	1149	}
854		1150
855		1151
856	=head1 WATCHER TYPES	1152	=head1 WATCHER TYPES
857		1153
858	This section describes each watcher in detail, but will not repeat	1154	This section describes each watcher in detail, but will not repeat
…		…
882	In general you can register as many read and/or write event watchers per	1178	In general you can register as many read and/or write event watchers per
883	fd as you want (as long as you don't confuse yourself). Setting all file	1179	fd as you want (as long as you don't confuse yourself). Setting all file
884	descriptors to non-blocking mode is also usually a good idea (but not	1180	descriptors to non-blocking mode is also usually a good idea (but not
885	required if you know what you are doing).	1181	required if you know what you are doing).
886		1182
887	You have to be careful with dup'ed file descriptors, though. Some backends	1183	If you cannot use non-blocking mode, then force the use of a
888	(the linux epoll backend is a notable example) cannot handle dup'ed file	1184	known-to-be-good backend (at the time of this writing, this includes only
889	descriptors correctly if you register interest in two or more fds pointing	1185	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>).
890	to the same underlying file/socket/etc. description (that is, they share
891	the same underlying "file open").
892
893	If you must do this, then force the use of a known-to-be-good backend
894	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
895	C<EVBACKEND_POLL>).
896		1186
897	Another thing you have to watch out for is that it is quite easy to	1187	Another thing you have to watch out for is that it is quite easy to
898	receive "spurious" readyness notifications, that is your callback might	1188	receive "spurious" readiness notifications, that is your callback might
899	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1189	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
900	because there is no data. Not only are some backends known to create a	1190	because there is no data. Not only are some backends known to create a
901	lot of those (for example solaris ports), it is very easy to get into	1191	lot of those (for example Solaris ports), it is very easy to get into
902	this situation even with a relatively standard program structure. Thus	1192	this situation even with a relatively standard program structure. Thus
903	it is best to always use non-blocking I/O: An extra C<read>(2) returning	1193	it is best to always use non-blocking I/O: An extra C<read>(2) returning
904	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1194	C<EAGAIN> is far preferable to a program hanging until some data arrives.
905		1195
906	If you cannot run the fd in non-blocking mode (for example you should not	1196	If you cannot run the fd in non-blocking mode (for example you should
907	play around with an Xlib connection), then you have to seperately re-test	1197	not play around with an Xlib connection), then you have to separately
908	whether a file descriptor is really ready with a known-to-be good interface	1198	re-test whether a file descriptor is really ready with a known-to-be good
909	such as poll (fortunately in our Xlib example, Xlib already does this on	1199	interface such as poll (fortunately in our Xlib example, Xlib already
910	its own, so its quite safe to use).	1200	does this on its own, so its quite safe to use). Some people additionally
		1201	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1202	indefinitely.
		1203
		1204	But really, best use non-blocking mode.
		1205
		1206	=head3 The special problem of disappearing file descriptors
		1207
		1208	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1209	descriptor (either due to calling C<close> explicitly or any other means,
		1210	such as C<dup2>). The reason is that you register interest in some file
		1211	descriptor, but when it goes away, the operating system will silently drop
		1212	this interest. If another file descriptor with the same number then is
		1213	registered with libev, there is no efficient way to see that this is, in
		1214	fact, a different file descriptor.
		1215
		1216	To avoid having to explicitly tell libev about such cases, libev follows
		1217	the following policy: Each time C<ev_io_set> is being called, libev
		1218	will assume that this is potentially a new file descriptor, otherwise
		1219	it is assumed that the file descriptor stays the same. That means that
		1220	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1221	descriptor even if the file descriptor number itself did not change.
		1222
		1223	This is how one would do it normally anyway, the important point is that
		1224	the libev application should not optimise around libev but should leave
		1225	optimisations to libev.
		1226
		1227	=head3 The special problem of dup'ed file descriptors
		1228
		1229	Some backends (e.g. epoll), cannot register events for file descriptors,
		1230	but only events for the underlying file descriptions. That means when you
		1231	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1232	events for them, only one file descriptor might actually receive events.
		1233
		1234	There is no workaround possible except not registering events
		1235	for potentially C<dup ()>'ed file descriptors, or to resort to
		1236	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1237
		1238	=head3 The special problem of fork
		1239
		1240	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1241	useless behaviour. Libev fully supports fork, but needs to be told about
		1242	it in the child.
		1243
		1244	To support fork in your programs, you either have to call
		1245	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1246	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1247	C<EVBACKEND_POLL>.
		1248
		1249	=head3 The special problem of SIGPIPE
		1250
		1251	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1252	when writing to a pipe whose other end has been closed, your program gets
		1253	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1254	this is sensible behaviour, for daemons, this is usually undesirable.
		1255
		1256	So when you encounter spurious, unexplained daemon exits, make sure you
		1257	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1258	somewhere, as that would have given you a big clue).
		1259
		1260
		1261	=head3 Watcher-Specific Functions
911		1262
912	=over 4	1263	=over 4
913		1264
914	=item ev_io_init (ev_io *, callback, int fd, int events)	1265	=item ev_io_init (ev_io *, callback, int fd, int events)
915		1266
916	=item ev_io_set (ev_io *, int fd, int events)	1267	=item ev_io_set (ev_io *, int fd, int events)
917		1268
918	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1269	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
919	rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or	1270	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
920	C<EV_READ | EV_WRITE> to receive the given events.	1271	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
921		1272
922	=item int fd [read-only]	1273	=item int fd [read-only]
923		1274
924	The file descriptor being watched.	1275	The file descriptor being watched.
925		1276
926	=item int events [read-only]	1277	=item int events [read-only]
927		1278
928	The events being watched.	1279	The events being watched.
929		1280
930	=back	1281	=back
		1282
		1283	=head3 Examples
931		1284
932	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1285	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
933	readable, but only once. Since it is likely line-buffered, you could	1286	readable, but only once. Since it is likely line-buffered, you could
934	attempt to read a whole line in the callback.	1287	attempt to read a whole line in the callback.
935		1288
936	static void	1289	static void
937	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1290	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
938	{	1291	{
939	ev_io_stop (loop, w);	1292	ev_io_stop (loop, w);
940	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1293	.. read from stdin here (or from w->fd) and handle any I/O errors
941	}	1294	}
942		1295
943	...	1296	...
944	struct ev_loop *loop = ev_default_init (0);	1297	struct ev_loop *loop = ev_default_init (0);
945	struct ev_io stdin_readable;	1298	ev_io stdin_readable;
946	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1299	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
947	ev_io_start (loop, &stdin_readable);	1300	ev_io_start (loop, &stdin_readable);
948	ev_loop (loop, 0);	1301	ev_loop (loop, 0);
949		1302
950		1303
951	=head2 C<ev_timer> - relative and optionally repeating timeouts	1304	=head2 C<ev_timer> - relative and optionally repeating timeouts
952		1305
953	Timer watchers are simple relative timers that generate an event after a	1306	Timer watchers are simple relative timers that generate an event after a
954	given time, and optionally repeating in regular intervals after that.	1307	given time, and optionally repeating in regular intervals after that.
955		1308
956	The timers are based on real time, that is, if you register an event that	1309	The timers are based on real time, that is, if you register an event that
957	times out after an hour and you reset your system clock to last years	1310	times out after an hour and you reset your system clock to January last
958	time, it will still time out after (roughly) and hour. "Roughly" because	1311	year, it will still time out after (roughly) one hour. "Roughly" because
959	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1312	detecting time jumps is hard, and some inaccuracies are unavoidable (the
960	monotonic clock option helps a lot here).	1313	monotonic clock option helps a lot here).
		1314
		1315	The callback is guaranteed to be invoked only I<after> its timeout has
		1316	passed, but if multiple timers become ready during the same loop iteration
		1317	then order of execution is undefined.
		1318
		1319	=head3 Be smart about timeouts
		1320
		1321	Many real-world problems involve some kind of timeout, usually for error
		1322	recovery. A typical example is an HTTP request - if the other side hangs,
		1323	you want to raise some error after a while.
		1324
		1325	What follows are some ways to handle this problem, from obvious and
		1326	inefficient to smart and efficient.
		1327
		1328	In the following, a 60 second activity timeout is assumed - a timeout that
		1329	gets reset to 60 seconds each time there is activity (e.g. each time some
		1330	data or other life sign was received).
		1331
		1332	=over 4
		1333
		1334	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1335
		1336	This is the most obvious, but not the most simple way: In the beginning,
		1337	start the watcher:
		1338
		1339	ev_timer_init (timer, callback, 60., 0.);
		1340	ev_timer_start (loop, timer);
		1341
		1342	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1343	and start it again:
		1344
		1345	ev_timer_stop (loop, timer);
		1346	ev_timer_set (timer, 60., 0.);
		1347	ev_timer_start (loop, timer);
		1348
		1349	This is relatively simple to implement, but means that each time there is
		1350	some activity, libev will first have to remove the timer from its internal
		1351	data structure and then add it again. Libev tries to be fast, but it's
		1352	still not a constant-time operation.
		1353
		1354	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1355
		1356	This is the easiest way, and involves using C<ev_timer_again> instead of
		1357	C<ev_timer_start>.
		1358
		1359	To implement this, configure an C<ev_timer> with a C<repeat> value
		1360	of C<60> and then call C<ev_timer_again> at start and each time you
		1361	successfully read or write some data. If you go into an idle state where
		1362	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1363	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1364
		1365	That means you can ignore both the C<ev_timer_start> function and the
		1366	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1367	member and C<ev_timer_again>.
		1368
		1369	At start:
		1370
		1371	ev_timer_init (timer, callback);
		1372	timer->repeat = 60.;
		1373	ev_timer_again (loop, timer);
		1374
		1375	Each time there is some activity:
		1376
		1377	ev_timer_again (loop, timer);
		1378
		1379	It is even possible to change the time-out on the fly, regardless of
		1380	whether the watcher is active or not:
		1381
		1382	timer->repeat = 30.;
		1383	ev_timer_again (loop, timer);
		1384
		1385	This is slightly more efficient then stopping/starting the timer each time
		1386	you want to modify its timeout value, as libev does not have to completely
		1387	remove and re-insert the timer from/into its internal data structure.
		1388
		1389	It is, however, even simpler than the "obvious" way to do it.
		1390
		1391	=item 3. Let the timer time out, but then re-arm it as required.
		1392
		1393	This method is more tricky, but usually most efficient: Most timeouts are
		1394	relatively long compared to the intervals between other activity - in
		1395	our example, within 60 seconds, there are usually many I/O events with
		1396	associated activity resets.
		1397
		1398	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1399	but remember the time of last activity, and check for a real timeout only
		1400	within the callback:
		1401
		1402	ev_tstamp last_activity; // time of last activity
		1403
		1404	static void
		1405	callback (EV_P_ ev_timer *w, int revents)
		1406	{
		1407	ev_tstamp now = ev_now (EV_A);
		1408	ev_tstamp timeout = last_activity + 60.;
		1409
		1410	// if last_activity + 60. is older than now, we did time out
		1411	if (timeout < now)
		1412	{
		1413	// timeout occured, take action
		1414	}
		1415	else
		1416	{
		1417	// callback was invoked, but there was some activity, re-arm
		1418	// the watcher to fire in last_activity + 60, which is
		1419	// guaranteed to be in the future, so "again" is positive:
		1420	w->again = timeout - now;
		1421	ev_timer_again (EV_A_ w);
		1422	}
		1423	}
		1424
		1425	To summarise the callback: first calculate the real timeout (defined
		1426	as "60 seconds after the last activity"), then check if that time has
		1427	been reached, which means something I<did>, in fact, time out. Otherwise
		1428	the callback was invoked too early (C<timeout> is in the future), so
		1429	re-schedule the timer to fire at that future time, to see if maybe we have
		1430	a timeout then.
		1431
		1432	Note how C<ev_timer_again> is used, taking advantage of the
		1433	C<ev_timer_again> optimisation when the timer is already running.
		1434
		1435	This scheme causes more callback invocations (about one every 60 seconds
		1436	minus half the average time between activity), but virtually no calls to
		1437	libev to change the timeout.
		1438
		1439	To start the timer, simply initialise the watcher and set C<last_activity>
		1440	to the current time (meaning we just have some activity :), then call the
		1441	callback, which will "do the right thing" and start the timer:
		1442
		1443	ev_timer_init (timer, callback);
		1444	last_activity = ev_now (loop);
		1445	callback (loop, timer, EV_TIMEOUT);
		1446
		1447	And when there is some activity, simply store the current time in
		1448	C<last_activity>, no libev calls at all:
		1449
		1450	last_actiivty = ev_now (loop);
		1451
		1452	This technique is slightly more complex, but in most cases where the
		1453	time-out is unlikely to be triggered, much more efficient.
		1454
		1455	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1456	callback :) - just change the timeout and invoke the callback, which will
		1457	fix things for you.
		1458
		1459	=item 4. Wee, just use a double-linked list for your timeouts.
		1460
		1461	If there is not one request, but many thousands (millions...), all
		1462	employing some kind of timeout with the same timeout value, then one can
		1463	do even better:
		1464
		1465	When starting the timeout, calculate the timeout value and put the timeout
		1466	at the I<end> of the list.
		1467
		1468	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1469	the list is expected to fire (for example, using the technique #3).
		1470
		1471	When there is some activity, remove the timer from the list, recalculate
		1472	the timeout, append it to the end of the list again, and make sure to
		1473	update the C<ev_timer> if it was taken from the beginning of the list.
		1474
		1475	This way, one can manage an unlimited number of timeouts in O(1) time for
		1476	starting, stopping and updating the timers, at the expense of a major
		1477	complication, and having to use a constant timeout. The constant timeout
		1478	ensures that the list stays sorted.
		1479
		1480	=back
		1481
		1482	So which method the best?
		1483
		1484	Method #2 is a simple no-brain-required solution that is adequate in most
		1485	situations. Method #3 requires a bit more thinking, but handles many cases
		1486	better, and isn't very complicated either. In most case, choosing either
		1487	one is fine, with #3 being better in typical situations.
		1488
		1489	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1490	rather complicated, but extremely efficient, something that really pays
		1491	off after the first million or so of active timers, i.e. it's usually
		1492	overkill :)
		1493
		1494	=head3 The special problem of time updates
		1495
		1496	Establishing the current time is a costly operation (it usually takes at
		1497	least two system calls): EV therefore updates its idea of the current
		1498	time only before and after C<ev_loop> collects new events, which causes a
		1499	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1500	lots of events in one iteration.
961		1501
962	The relative timeouts are calculated relative to the C<ev_now ()>	1502	The relative timeouts are calculated relative to the C<ev_now ()>
963	time. This is usually the right thing as this timestamp refers to the time	1503	time. This is usually the right thing as this timestamp refers to the time
964	of the event triggering whatever timeout you are modifying/starting. If	1504	of the event triggering whatever timeout you are modifying/starting. If
965	you suspect event processing to be delayed and you I<need> to base the timeout	1505	you suspect event processing to be delayed and you I<need> to base the
966	on the current time, use something like this to adjust for this:	1506	timeout on the current time, use something like this to adjust for this:
967		1507
968	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1508	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
969		1509
970	The callback is guarenteed to be invoked only when its timeout has passed,	1510	If the event loop is suspended for a long time, you can also force an
971	but if multiple timers become ready during the same loop iteration then	1511	update of the time returned by C<ev_now ()> by calling C<ev_now_update
972	order of execution is undefined.	1512	()>.
		1513
		1514	=head3 Watcher-Specific Functions and Data Members
973		1515
974	=over 4	1516	=over 4
975		1517
976	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1518	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
977		1519
978	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1520	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
979		1521
980	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1522	Configure the timer to trigger after C<after> seconds. If C<repeat>
981	C<0.>, then it will automatically be stopped. If it is positive, then the	1523	is C<0.>, then it will automatically be stopped once the timeout is
982	timer will automatically be configured to trigger again C<repeat> seconds	1524	reached. If it is positive, then the timer will automatically be
983	later, again, and again, until stopped manually.	1525	configured to trigger again C<repeat> seconds later, again, and again,
		1526	until stopped manually.
984		1527
985	The timer itself will do a best-effort at avoiding drift, that is, if you	1528	The timer itself will do a best-effort at avoiding drift, that is, if
986	configure a timer to trigger every 10 seconds, then it will trigger at	1529	you configure a timer to trigger every 10 seconds, then it will normally
987	exactly 10 second intervals. If, however, your program cannot keep up with	1530	trigger at exactly 10 second intervals. If, however, your program cannot
988	the timer (because it takes longer than those 10 seconds to do stuff) the	1531	keep up with the timer (because it takes longer than those 10 seconds to
989	timer will not fire more than once per event loop iteration.	1532	do stuff) the timer will not fire more than once per event loop iteration.
990		1533
991	=item ev_timer_again (loop)	1534	=item ev_timer_again (loop, ev_timer *)
992		1535
993	This will act as if the timer timed out and restart it again if it is	1536	This will act as if the timer timed out and restart it again if it is
994	repeating. The exact semantics are:	1537	repeating. The exact semantics are:
995		1538
996	If the timer is pending, its pending status is cleared.	1539	If the timer is pending, its pending status is cleared.
997		1540
998	If the timer is started but nonrepeating, stop it (as if it timed out).	1541	If the timer is started but non-repeating, stop it (as if it timed out).
999		1542
1000	If the timer is repeating, either start it if necessary (with the	1543	If the timer is repeating, either start it if necessary (with the
1001	C<repeat> value), or reset the running timer to the C<repeat> value.	1544	C<repeat> value), or reset the running timer to the C<repeat> value.
1002		1545
1003	This sounds a bit complicated, but here is a useful and typical	1546	This sounds a bit complicated, see "Be smart about timeouts", above, for a
1004	example: Imagine you have a tcp connection and you want a so-called idle	1547	usage example.
1005	timeout, that is, you want to be called when there have been, say, 60
1006	seconds of inactivity on the socket. The easiest way to do this is to
1007	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
1008	C<ev_timer_again> each time you successfully read or write some data. If
1009	you go into an idle state where you do not expect data to travel on the
1010	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
1011	automatically restart it if need be.
1012
1013	That means you can ignore the C<after> value and C<ev_timer_start>
1014	altogether and only ever use the C<repeat> value and C<ev_timer_again>:
1015
1016	ev_timer_init (timer, callback, 0., 5.);
1017	ev_timer_again (loop, timer);
1018	...
1019	timer->again = 17.;
1020	ev_timer_again (loop, timer);
1021	...
1022	timer->again = 10.;
1023	ev_timer_again (loop, timer);
1024
1025	This is more slightly efficient then stopping/starting the timer each time
1026	you want to modify its timeout value.
1027		1548
1028	=item ev_tstamp repeat [read-write]	1549	=item ev_tstamp repeat [read-write]
1029		1550
1030	The current C<repeat> value. Will be used each time the watcher times out	1551	The current C<repeat> value. Will be used each time the watcher times out
1031	or C<ev_timer_again> is called and determines the next timeout (if any),	1552	or C<ev_timer_again> is called, and determines the next timeout (if any),
1032	which is also when any modifications are taken into account.	1553	which is also when any modifications are taken into account.
1033		1554
1034	=back	1555	=back
1035		1556
		1557	=head3 Examples
		1558
1036	Example: Create a timer that fires after 60 seconds.	1559	Example: Create a timer that fires after 60 seconds.
1037		1560
1038	static void	1561	static void
1039	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1562	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1040	{	1563	{
1041	.. one minute over, w is actually stopped right here	1564	.. one minute over, w is actually stopped right here
1042	}	1565	}
1043		1566
1044	struct ev_timer mytimer;	1567	ev_timer mytimer;
1045	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	1568	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1046	ev_timer_start (loop, &mytimer);	1569	ev_timer_start (loop, &mytimer);
1047		1570
1048	Example: Create a timeout timer that times out after 10 seconds of	1571	Example: Create a timeout timer that times out after 10 seconds of
1049	inactivity.	1572	inactivity.
1050		1573
1051	static void	1574	static void
1052	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1575	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1053	{	1576	{
1054	.. ten seconds without any activity	1577	.. ten seconds without any activity
1055	}	1578	}
1056		1579
1057	struct ev_timer mytimer;	1580	ev_timer mytimer;
1058	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	1581	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1059	ev_timer_again (&mytimer); /* start timer */	1582	ev_timer_again (&mytimer); /* start timer */
1060	ev_loop (loop, 0);	1583	ev_loop (loop, 0);
1061		1584
1062	// and in some piece of code that gets executed on any "activity":	1585	// and in some piece of code that gets executed on any "activity":
1063	// reset the timeout to start ticking again at 10 seconds	1586	// reset the timeout to start ticking again at 10 seconds
1064	ev_timer_again (&mytimer);	1587	ev_timer_again (&mytimer);
1065		1588
1066		1589
1067	=head2 C<ev_periodic> - to cron or not to cron?	1590	=head2 C<ev_periodic> - to cron or not to cron?
1068		1591
1069	Periodic watchers are also timers of a kind, but they are very versatile	1592	Periodic watchers are also timers of a kind, but they are very versatile
1070	(and unfortunately a bit complex).	1593	(and unfortunately a bit complex).
1071		1594
1072	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1595	Unlike C<ev_timer>'s, they are not based on real time (or relative time)
1073	but on wallclock time (absolute time). You can tell a periodic watcher	1596	but on wall clock time (absolute time). You can tell a periodic watcher
1074	to trigger "at" some specific point in time. For example, if you tell a	1597	to trigger after some specific point in time. For example, if you tell a
1075	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	1598	periodic watcher to trigger in 10 seconds (by specifying e.g. C<ev_now ()
1076	+ 10.>) and then reset your system clock to the last year, then it will	1599	+ 10.>, that is, an absolute time not a delay) and then reset your system
		1600	clock to January of the previous year, then it will take more than year
1077	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1601	to trigger the event (unlike an C<ev_timer>, which would still trigger
1078	roughly 10 seconds later and of course not if you reset your system time	1602	roughly 10 seconds later as it uses a relative timeout).
1079	again).
1080		1603
1081	They can also be used to implement vastly more complex timers, such as	1604	C<ev_periodic>s can also be used to implement vastly more complex timers,
1082	triggering an event on eahc midnight, local time.	1605	such as triggering an event on each "midnight, local time", or other
		1606	complicated rules.
1083		1607
1084	As with timers, the callback is guarenteed to be invoked only when the	1608	As with timers, the callback is guaranteed to be invoked only when the
1085	time (C<at>) has been passed, but if multiple periodic timers become ready	1609	time (C<at>) has passed, but if multiple periodic timers become ready
1086	during the same loop iteration then order of execution is undefined.	1610	during the same loop iteration, then order of execution is undefined.
		1611
		1612	=head3 Watcher-Specific Functions and Data Members
1087		1613
1088	=over 4	1614	=over 4
1089		1615
1090	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1616	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
1091		1617
1092	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1618	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
1093		1619
1094	Lots of arguments, lets sort it out... There are basically three modes of	1620	Lots of arguments, lets sort it out... There are basically three modes of
1095	operation, and we will explain them from simplest to complex:	1621	operation, and we will explain them from simplest to most complex:
1096		1622
1097	=over 4	1623	=over 4
1098		1624
1099	=item * absolute timer (interval = reschedule_cb = 0)	1625	=item * absolute timer (at = time, interval = reschedule_cb = 0)
1100		1626
1101	In this configuration the watcher triggers an event at the wallclock time	1627	In this configuration the watcher triggers an event after the wall clock
1102	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1628	time C<at> has passed. It will not repeat and will not adjust when a time
1103	that is, if it is to be run at January 1st 2011 then it will run when the	1629	jump occurs, that is, if it is to be run at January 1st 2011 then it will
1104	system time reaches or surpasses this time.	1630	only run when the system clock reaches or surpasses this time.
1105		1631
1106	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1632	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1107		1633
1108	In this mode the watcher will always be scheduled to time out at the next	1634	In this mode the watcher will always be scheduled to time out at the next
1109	C<at + N * interval> time (for some integer N) and then repeat, regardless	1635	C<at + N * interval> time (for some integer N, which can also be negative)
1110	of any time jumps.	1636	and then repeat, regardless of any time jumps.
1111		1637
1112	This can be used to create timers that do not drift with respect to system	1638	This can be used to create timers that do not drift with respect to the
1113	time:	1639	system clock, for example, here is a C<ev_periodic> that triggers each
		1640	hour, on the hour:
1114		1641
1115	ev_periodic_set (&periodic, 0., 3600., 0);	1642	ev_periodic_set (&periodic, 0., 3600., 0);
1116		1643
1117	This doesn't mean there will always be 3600 seconds in between triggers,	1644	This doesn't mean there will always be 3600 seconds in between triggers,
1118	but only that the the callback will be called when the system time shows a	1645	but only that the callback will be called when the system time shows a
1119	full hour (UTC), or more correctly, when the system time is evenly divisible	1646	full hour (UTC), or more correctly, when the system time is evenly divisible
1120	by 3600.	1647	by 3600.
1121		1648
1122	Another way to think about it (for the mathematically inclined) is that	1649	Another way to think about it (for the mathematically inclined) is that
1123	C<ev_periodic> will try to run the callback in this mode at the next possible	1650	C<ev_periodic> will try to run the callback in this mode at the next possible
1124	time where C<time = at (mod interval)>, regardless of any time jumps.	1651	time where C<time = at (mod interval)>, regardless of any time jumps.
1125		1652
		1653	For numerical stability it is preferable that the C<at> value is near
		1654	C<ev_now ()> (the current time), but there is no range requirement for
		1655	this value, and in fact is often specified as zero.
		1656
		1657	Note also that there is an upper limit to how often a timer can fire (CPU
		1658	speed for example), so if C<interval> is very small then timing stability
		1659	will of course deteriorate. Libev itself tries to be exact to be about one
		1660	millisecond (if the OS supports it and the machine is fast enough).
		1661
1126	=item * manual reschedule mode (reschedule_cb = callback)	1662	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1127		1663
1128	In this mode the values for C<interval> and C<at> are both being	1664	In this mode the values for C<interval> and C<at> are both being
1129	ignored. Instead, each time the periodic watcher gets scheduled, the	1665	ignored. Instead, each time the periodic watcher gets scheduled, the
1130	reschedule callback will be called with the watcher as first, and the	1666	reschedule callback will be called with the watcher as first, and the
1131	current time as second argument.	1667	current time as second argument.
1132		1668
1133	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1669	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
1134	ever, or make any event loop modifications>. If you need to stop it,	1670	ever, or make ANY event loop modifications whatsoever>.
1135	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by
1136	starting a prepare watcher).
1137		1671
		1672	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		1673	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		1674	only event loop modification you are allowed to do).
		1675
1138	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	1676	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1139	ev_tstamp now)>, e.g.:	1677	*w, ev_tstamp now)>, e.g.:
1140		1678
		1679	static ev_tstamp
1141	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1680	my_rescheduler (ev_periodic *w, ev_tstamp now)
1142	{	1681	{
1143	return now + 60.;	1682	return now + 60.;
1144	}	1683	}
1145		1684
1146	It must return the next time to trigger, based on the passed time value	1685	It must return the next time to trigger, based on the passed time value
1147	(that is, the lowest time value larger than to the second argument). It	1686	(that is, the lowest time value larger than to the second argument). It
1148	will usually be called just before the callback will be triggered, but	1687	will usually be called just before the callback will be triggered, but
1149	might be called at other times, too.	1688	might be called at other times, too.
1150		1689
1151	NOTE: I<< This callback must always return a time that is later than the	1690	NOTE: I<< This callback must always return a time that is higher than or
1152	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	1691	equal to the passed C<now> value >>.
1153		1692
1154	This can be used to create very complex timers, such as a timer that	1693	This can be used to create very complex timers, such as a timer that
1155	triggers on each midnight, local time. To do this, you would calculate the	1694	triggers on "next midnight, local time". To do this, you would calculate the
1156	next midnight after C<now> and return the timestamp value for this. How	1695	next midnight after C<now> and return the timestamp value for this. How
1157	you do this is, again, up to you (but it is not trivial, which is the main	1696	you do this is, again, up to you (but it is not trivial, which is the main
1158	reason I omitted it as an example).	1697	reason I omitted it as an example).
1159		1698
1160	=back	1699	=back
…		…
1164	Simply stops and restarts the periodic watcher again. This is only useful	1703	Simply stops and restarts the periodic watcher again. This is only useful
1165	when you changed some parameters or the reschedule callback would return	1704	when you changed some parameters or the reschedule callback would return
1166	a different time than the last time it was called (e.g. in a crond like	1705	a different time than the last time it was called (e.g. in a crond like
1167	program when the crontabs have changed).	1706	program when the crontabs have changed).
1168		1707
		1708	=item ev_tstamp ev_periodic_at (ev_periodic *)
		1709
		1710	When active, returns the absolute time that the watcher is supposed to
		1711	trigger next.
		1712
		1713	=item ev_tstamp offset [read-write]
		1714
		1715	When repeating, this contains the offset value, otherwise this is the
		1716	absolute point in time (the C<at> value passed to C<ev_periodic_set>).
		1717
		1718	Can be modified any time, but changes only take effect when the periodic
		1719	timer fires or C<ev_periodic_again> is being called.
		1720
1169	=item ev_tstamp interval [read-write]	1721	=item ev_tstamp interval [read-write]
1170		1722
1171	The current interval value. Can be modified any time, but changes only	1723	The current interval value. Can be modified any time, but changes only
1172	take effect when the periodic timer fires or C<ev_periodic_again> is being	1724	take effect when the periodic timer fires or C<ev_periodic_again> is being
1173	called.	1725	called.
1174		1726
1175	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	1727	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1176		1728
1177	The current reschedule callback, or C<0>, if this functionality is	1729	The current reschedule callback, or C<0>, if this functionality is
1178	switched off. Can be changed any time, but changes only take effect when	1730	switched off. Can be changed any time, but changes only take effect when
1179	the periodic timer fires or C<ev_periodic_again> is being called.	1731	the periodic timer fires or C<ev_periodic_again> is being called.
1180		1732
1181	=back	1733	=back
1182		1734
		1735	=head3 Examples
		1736
1183	Example: Call a callback every hour, or, more precisely, whenever the	1737	Example: Call a callback every hour, or, more precisely, whenever the
1184	system clock is divisible by 3600. The callback invocation times have	1738	system time is divisible by 3600. The callback invocation times have
1185	potentially a lot of jittering, but good long-term stability.	1739	potentially a lot of jitter, but good long-term stability.
1186		1740
1187	static void	1741	static void
1188	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1742	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
1189	{	1743	{
1190	... its now a full hour (UTC, or TAI or whatever your clock follows)	1744	... its now a full hour (UTC, or TAI or whatever your clock follows)
1191	}	1745	}
1192		1746
1193	struct ev_periodic hourly_tick;	1747	ev_periodic hourly_tick;
1194	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	1748	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1195	ev_periodic_start (loop, &hourly_tick);	1749	ev_periodic_start (loop, &hourly_tick);
1196		1750
1197	Example: The same as above, but use a reschedule callback to do it:	1751	Example: The same as above, but use a reschedule callback to do it:
1198		1752
1199	#include <math.h>	1753	#include <math.h>
1200		1754
1201	static ev_tstamp	1755	static ev_tstamp
1202	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	1756	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1203	{	1757	{
1204	return fmod (now, 3600.) + 3600.;	1758	return now + (3600. - fmod (now, 3600.));
1205	}	1759	}
1206		1760
1207	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	1761	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1208		1762
1209	Example: Call a callback every hour, starting now:	1763	Example: Call a callback every hour, starting now:
1210		1764
1211	struct ev_periodic hourly_tick;	1765	ev_periodic hourly_tick;
1212	ev_periodic_init (&hourly_tick, clock_cb,	1766	ev_periodic_init (&hourly_tick, clock_cb,
1213	fmod (ev_now (loop), 3600.), 3600., 0);	1767	fmod (ev_now (loop), 3600.), 3600., 0);
1214	ev_periodic_start (loop, &hourly_tick);	1768	ev_periodic_start (loop, &hourly_tick);
1215		1769
1216		1770
1217	=head2 C<ev_signal> - signal me when a signal gets signalled!	1771	=head2 C<ev_signal> - signal me when a signal gets signalled!
1218		1772
1219	Signal watchers will trigger an event when the process receives a specific	1773	Signal watchers will trigger an event when the process receives a specific
1220	signal one or more times. Even though signals are very asynchronous, libev	1774	signal one or more times. Even though signals are very asynchronous, libev
1221	will try it's best to deliver signals synchronously, i.e. as part of the	1775	will try it's best to deliver signals synchronously, i.e. as part of the
1222	normal event processing, like any other event.	1776	normal event processing, like any other event.
1223		1777
		1778	If you want signals asynchronously, just use C<sigaction> as you would
		1779	do without libev and forget about sharing the signal. You can even use
		1780	C<ev_async> from a signal handler to synchronously wake up an event loop.
		1781
1224	You can configure as many watchers as you like per signal. Only when the	1782	You can configure as many watchers as you like per signal. Only when the
1225	first watcher gets started will libev actually register a signal watcher	1783	first watcher gets started will libev actually register a signal handler
1226	with the kernel (thus it coexists with your own signal handlers as long	1784	with the kernel (thus it coexists with your own signal handlers as long as
1227	as you don't register any with libev). Similarly, when the last signal	1785	you don't register any with libev for the same signal). Similarly, when
1228	watcher for a signal is stopped libev will reset the signal handler to	1786	the last signal watcher for a signal is stopped, libev will reset the
1229	SIG_DFL (regardless of what it was set to before).	1787	signal handler to SIG_DFL (regardless of what it was set to before).
		1788
		1789	If possible and supported, libev will install its handlers with
		1790	C<SA_RESTART> behaviour enabled, so system calls should not be unduly
		1791	interrupted. If you have a problem with system calls getting interrupted by
		1792	signals you can block all signals in an C<ev_check> watcher and unblock
		1793	them in an C<ev_prepare> watcher.
		1794
		1795	=head3 Watcher-Specific Functions and Data Members
1230		1796
1231	=over 4	1797	=over 4
1232		1798
1233	=item ev_signal_init (ev_signal *, callback, int signum)	1799	=item ev_signal_init (ev_signal *, callback, int signum)
1234		1800
…		…
1241		1807
1242	The signal the watcher watches out for.	1808	The signal the watcher watches out for.
1243		1809
1244	=back	1810	=back
1245		1811
		1812	=head3 Examples
		1813
		1814	Example: Try to exit cleanly on SIGINT.
		1815
		1816	static void
		1817	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		1818	{
		1819	ev_unloop (loop, EVUNLOOP_ALL);
		1820	}
		1821
		1822	ev_signal signal_watcher;
		1823	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		1824	ev_signal_start (loop, &signal_watcher);
		1825
1246		1826
1247	=head2 C<ev_child> - watch out for process status changes	1827	=head2 C<ev_child> - watch out for process status changes
1248		1828
1249	Child watchers trigger when your process receives a SIGCHLD in response to	1829	Child watchers trigger when your process receives a SIGCHLD in response to
1250	some child status changes (most typically when a child of yours dies).	1830	some child status changes (most typically when a child of yours dies or
		1831	exits). It is permissible to install a child watcher I<after> the child
		1832	has been forked (which implies it might have already exited), as long
		1833	as the event loop isn't entered (or is continued from a watcher), i.e.,
		1834	forking and then immediately registering a watcher for the child is fine,
		1835	but forking and registering a watcher a few event loop iterations later is
		1836	not.
		1837
		1838	Only the default event loop is capable of handling signals, and therefore
		1839	you can only register child watchers in the default event loop.
		1840
		1841	=head3 Process Interaction
		1842
		1843	Libev grabs C<SIGCHLD> as soon as the default event loop is
		1844	initialised. This is necessary to guarantee proper behaviour even if
		1845	the first child watcher is started after the child exits. The occurrence
		1846	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		1847	synchronously as part of the event loop processing. Libev always reaps all
		1848	children, even ones not watched.
		1849
		1850	=head3 Overriding the Built-In Processing
		1851
		1852	Libev offers no special support for overriding the built-in child
		1853	processing, but if your application collides with libev's default child
		1854	handler, you can override it easily by installing your own handler for
		1855	C<SIGCHLD> after initialising the default loop, and making sure the
		1856	default loop never gets destroyed. You are encouraged, however, to use an
		1857	event-based approach to child reaping and thus use libev's support for
		1858	that, so other libev users can use C<ev_child> watchers freely.
		1859
		1860	=head3 Stopping the Child Watcher
		1861
		1862	Currently, the child watcher never gets stopped, even when the
		1863	child terminates, so normally one needs to stop the watcher in the
		1864	callback. Future versions of libev might stop the watcher automatically
		1865	when a child exit is detected.
		1866
		1867	=head3 Watcher-Specific Functions and Data Members
1251		1868
1252	=over 4	1869	=over 4
1253		1870
1254	=item ev_child_init (ev_child *, callback, int pid)	1871	=item ev_child_init (ev_child *, callback, int pid, int trace)
1255		1872
1256	=item ev_child_set (ev_child *, int pid)	1873	=item ev_child_set (ev_child *, int pid, int trace)
1257		1874
1258	Configures the watcher to wait for status changes of process C<pid> (or	1875	Configures the watcher to wait for status changes of process C<pid> (or
1259	I<any> process if C<pid> is specified as C<0>). The callback can look	1876	I<any> process if C<pid> is specified as C<0>). The callback can look
1260	at the C<rstatus> member of the C<ev_child> watcher structure to see	1877	at the C<rstatus> member of the C<ev_child> watcher structure to see
1261	the status word (use the macros from C<sys/wait.h> and see your systems	1878	the status word (use the macros from C<sys/wait.h> and see your systems
1262	C<waitpid> documentation). The C<rpid> member contains the pid of the	1879	C<waitpid> documentation). The C<rpid> member contains the pid of the
1263	process causing the status change.	1880	process causing the status change. C<trace> must be either C<0> (only
		1881	activate the watcher when the process terminates) or C<1> (additionally
		1882	activate the watcher when the process is stopped or continued).
1264		1883
1265	=item int pid [read-only]	1884	=item int pid [read-only]
1266		1885
1267	The process id this watcher watches out for, or C<0>, meaning any process id.	1886	The process id this watcher watches out for, or C<0>, meaning any process id.
1268		1887
…		…
1275	The process exit/trace status caused by C<rpid> (see your systems	1894	The process exit/trace status caused by C<rpid> (see your systems
1276	C<waitpid> and C<sys/wait.h> documentation for details).	1895	C<waitpid> and C<sys/wait.h> documentation for details).
1277		1896
1278	=back	1897	=back
1279		1898
1280	Example: Try to exit cleanly on SIGINT and SIGTERM.	1899	=head3 Examples
1281		1900
		1901	Example: C<fork()> a new process and install a child handler to wait for
		1902	its completion.
		1903
		1904	ev_child cw;
		1905
1282	static void	1906	static void
1283	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	1907	child_cb (EV_P_ ev_child *w, int revents)
1284	{	1908	{
1285	ev_unloop (loop, EVUNLOOP_ALL);	1909	ev_child_stop (EV_A_ w);
		1910	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1286	}	1911	}
1287		1912
1288	struct ev_signal signal_watcher;	1913	pid_t pid = fork ();
1289	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	1914
1290	ev_signal_start (loop, &sigint_cb);	1915	if (pid < 0)
		1916	// error
		1917	else if (pid == 0)
		1918	{
		1919	// the forked child executes here
		1920	exit (1);
		1921	}
		1922	else
		1923	{
		1924	ev_child_init (&cw, child_cb, pid, 0);
		1925	ev_child_start (EV_DEFAULT_ &cw);
		1926	}
1291		1927
1292		1928
1293	=head2 C<ev_stat> - did the file attributes just change?	1929	=head2 C<ev_stat> - did the file attributes just change?
1294		1930
1295	This watches a filesystem path for attribute changes. That is, it calls	1931	This watches a file system path for attribute changes. That is, it calls
1296	C<stat> regularly (or when the OS says it changed) and sees if it changed	1932	C<stat> on that path in regular intervals (or when the OS says it changed)
1297	compared to the last time, invoking the callback if it did.	1933	and sees if it changed compared to the last time, invoking the callback if
		1934	it did.
1298		1935
1299	The path does not need to exist: changing from "path exists" to "path does	1936	The path does not need to exist: changing from "path exists" to "path does
1300	not exist" is a status change like any other. The condition "path does	1937	not exist" is a status change like any other. The condition "path does not
1301	not exist" is signified by the C<st_nlink> field being zero (which is	1938	exist" (or more correctly "path cannot be stat'ed") is signified by the
1302	otherwise always forced to be at least one) and all the other fields of	1939	C<st_nlink> field being zero (which is otherwise always forced to be at
1303	the stat buffer having unspecified contents.	1940	least one) and all the other fields of the stat buffer having unspecified
		1941	contents.
1304		1942
1305	The path I<should> be absolute and I<must not> end in a slash. If it is	1943	The path I<must not> end in a slash or contain special components such as
		1944	C<.> or C<..>. The path I<should> be absolute: If it is relative and
1306	relative and your working directory changes, the behaviour is undefined.	1945	your working directory changes, then the behaviour is undefined.
1307		1946
1308	Since there is no standard to do this, the portable implementation simply	1947	Since there is no portable change notification interface available, the
1309	calls C<stat (2)> regularly on the path to see if it changed somehow. You	1948	portable implementation simply calls C<stat(2)> regularly on the path
1310	can specify a recommended polling interval for this case. If you specify	1949	to see if it changed somehow. You can specify a recommended polling
1311	a polling interval of C<0> (highly recommended!) then a I<suitable,	1950	interval for this case. If you specify a polling interval of C<0> (highly
1312	unspecified default> value will be used (which you can expect to be around	1951	recommended!) then a I<suitable, unspecified default> value will be used
1313	five seconds, although this might change dynamically). Libev will also	1952	(which you can expect to be around five seconds, although this might
1314	impose a minimum interval which is currently around C<0.1>, but thats	1953	change dynamically). Libev will also impose a minimum interval which is
1315	usually overkill.	1954	currently around C<0.1>, but that's usually overkill.
1316		1955
1317	This watcher type is not meant for massive numbers of stat watchers,	1956	This watcher type is not meant for massive numbers of stat watchers,
1318	as even with OS-supported change notifications, this can be	1957	as even with OS-supported change notifications, this can be
1319	resource-intensive.	1958	resource-intensive.
1320		1959
1321	At the time of this writing, only the Linux inotify interface is	1960	At the time of this writing, the only OS-specific interface implemented
1322	implemented (implementing kqueue support is left as an exercise for the	1961	is the Linux inotify interface (implementing kqueue support is left as an
1323	reader). Inotify will be used to give hints only and should not change the	1962	exercise for the reader. Note, however, that the author sees no way of
1324	semantics of C<ev_stat> watchers, which means that libev sometimes needs	1963	implementing C<ev_stat> semantics with kqueue, except as a hint).
1325	to fall back to regular polling again even with inotify, but changes are	1964
1326	usually detected immediately, and if the file exists there will be no	1965	=head3 ABI Issues (Largefile Support)
1327	polling.	1966
		1967	Libev by default (unless the user overrides this) uses the default
		1968	compilation environment, which means that on systems with large file
		1969	support disabled by default, you get the 32 bit version of the stat
		1970	structure. When using the library from programs that change the ABI to
		1971	use 64 bit file offsets the programs will fail. In that case you have to
		1972	compile libev with the same flags to get binary compatibility. This is
		1973	obviously the case with any flags that change the ABI, but the problem is
		1974	most noticeably displayed with ev_stat and large file support.
		1975
		1976	The solution for this is to lobby your distribution maker to make large
		1977	file interfaces available by default (as e.g. FreeBSD does) and not
		1978	optional. Libev cannot simply switch on large file support because it has
		1979	to exchange stat structures with application programs compiled using the
		1980	default compilation environment.
		1981
		1982	=head3 Inotify and Kqueue
		1983
		1984	When C<inotify (7)> support has been compiled into libev and present at
		1985	runtime, it will be used to speed up change detection where possible. The
		1986	inotify descriptor will be created lazily when the first C<ev_stat>
		1987	watcher is being started.
		1988
		1989	Inotify presence does not change the semantics of C<ev_stat> watchers
		1990	except that changes might be detected earlier, and in some cases, to avoid
		1991	making regular C<stat> calls. Even in the presence of inotify support
		1992	there are many cases where libev has to resort to regular C<stat> polling,
		1993	but as long as kernel 2.6.25 or newer is used (2.6.24 and older have too
		1994	many bugs), the path exists (i.e. stat succeeds), and the path resides on
		1995	a local filesystem (libev currently assumes only ext2/3, jfs, reiserfs and
		1996	xfs are fully working) libev usually gets away without polling.
		1997
		1998	There is no support for kqueue, as apparently it cannot be used to
		1999	implement this functionality, due to the requirement of having a file
		2000	descriptor open on the object at all times, and detecting renames, unlinks
		2001	etc. is difficult.
		2002
		2003	=head3 C<stat ()> is a synchronous operation
		2004
		2005	Libev doesn't normally do any kind of I/O itself, and so is not blocking
		2006	the process. The exception are C<ev_stat> watchers - those call C<stat
		2007	()>, which is a synchronous operation.
		2008
		2009	For local paths, this usually doesn't matter: unless the system is very
		2010	busy or the intervals between stat's are large, a stat call will be fast,
		2011	as the path data is suually in memory already (except when starting the
		2012	watcher).
		2013
		2014	For networked file systems, calling C<stat ()> can block an indefinite
		2015	time due to network issues, and even under good conditions, a stat call
		2016	often takes multiple milliseconds.
		2017
		2018	Therefore, it is best to avoid using C<ev_stat> watchers on networked
		2019	paths, although this is fully supported by libev.
		2020
		2021	=head3 The special problem of stat time resolution
		2022
		2023	The C<stat ()> system call only supports full-second resolution portably,
		2024	and even on systems where the resolution is higher, most file systems
		2025	still only support whole seconds.
		2026
		2027	That means that, if the time is the only thing that changes, you can
		2028	easily miss updates: on the first update, C<ev_stat> detects a change and
		2029	calls your callback, which does something. When there is another update
		2030	within the same second, C<ev_stat> will be unable to detect unless the
		2031	stat data does change in other ways (e.g. file size).
		2032
		2033	The solution to this is to delay acting on a change for slightly more
		2034	than a second (or till slightly after the next full second boundary), using
		2035	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2036	ev_timer_again (loop, w)>).
		2037
		2038	The C<.02> offset is added to work around small timing inconsistencies
		2039	of some operating systems (where the second counter of the current time
		2040	might be be delayed. One such system is the Linux kernel, where a call to
		2041	C<gettimeofday> might return a timestamp with a full second later than
		2042	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2043	update file times then there will be a small window where the kernel uses
		2044	the previous second to update file times but libev might already execute
		2045	the timer callback).
		2046
		2047	=head3 Watcher-Specific Functions and Data Members
1328		2048
1329	=over 4	2049	=over 4
1330		2050
1331	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)	2051	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1332		2052
…		…
1336	C<path>. The C<interval> is a hint on how quickly a change is expected to	2056	C<path>. The C<interval> is a hint on how quickly a change is expected to
1337	be detected and should normally be specified as C<0> to let libev choose	2057	be detected and should normally be specified as C<0> to let libev choose
1338	a suitable value. The memory pointed to by C<path> must point to the same	2058	a suitable value. The memory pointed to by C<path> must point to the same
1339	path for as long as the watcher is active.	2059	path for as long as the watcher is active.
1340		2060
1341	The callback will be receive C<EV_STAT> when a change was detected,	2061	The callback will receive an C<EV_STAT> event when a change was detected,
1342	relative to the attributes at the time the watcher was started (or the	2062	relative to the attributes at the time the watcher was started (or the
1343	last change was detected).	2063	last change was detected).
1344		2064
1345	=item ev_stat_stat (ev_stat *)	2065	=item ev_stat_stat (loop, ev_stat *)
1346		2066
1347	Updates the stat buffer immediately with new values. If you change the	2067	Updates the stat buffer immediately with new values. If you change the
1348	watched path in your callback, you could call this fucntion to avoid	2068	watched path in your callback, you could call this function to avoid
1349	detecting this change (while introducing a race condition). Can also be	2069	detecting this change (while introducing a race condition if you are not
1350	useful simply to find out the new values.	2070	the only one changing the path). Can also be useful simply to find out the
		2071	new values.
1351		2072
1352	=item ev_statdata attr [read-only]	2073	=item ev_statdata attr [read-only]
1353		2074
1354	The most-recently detected attributes of the file. Although the type is of	2075	The most-recently detected attributes of the file. Although the type is
1355	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types	2076	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1356	suitable for your system. If the C<st_nlink> member is C<0>, then there	2077	suitable for your system, but you can only rely on the POSIX-standardised
		2078	members to be present. If the C<st_nlink> member is C<0>, then there was
1357	was some error while C<stat>ing the file.	2079	some error while C<stat>ing the file.
1358		2080
1359	=item ev_statdata prev [read-only]	2081	=item ev_statdata prev [read-only]
1360		2082
1361	The previous attributes of the file. The callback gets invoked whenever	2083	The previous attributes of the file. The callback gets invoked whenever
1362	C<prev> != C<attr>.	2084	C<prev> != C<attr>, or, more precisely, one or more of these members
		2085	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2086	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
1363		2087
1364	=item ev_tstamp interval [read-only]	2088	=item ev_tstamp interval [read-only]
1365		2089
1366	The specified interval.	2090	The specified interval.
1367		2091
1368	=item const char *path [read-only]	2092	=item const char *path [read-only]
1369		2093
1370	The filesystem path that is being watched.	2094	The file system path that is being watched.
1371		2095
1372	=back	2096	=back
1373		2097
		2098	=head3 Examples
		2099
1374	Example: Watch C</etc/passwd> for attribute changes.	2100	Example: Watch C</etc/passwd> for attribute changes.
1375		2101
1376	static void	2102	static void
1377	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2103	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1378	{	2104	{
1379	/* /etc/passwd changed in some way */	2105	/* /etc/passwd changed in some way */
1380	if (w->attr.st_nlink)	2106	if (w->attr.st_nlink)
1381	{	2107	{
1382	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2108	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1383	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2109	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1384	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2110	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1385	}	2111	}
1386	else	2112	else
1387	/* you shalt not abuse printf for puts */	2113	/* you shalt not abuse printf for puts */
1388	puts ("wow, /etc/passwd is not there, expect problems. "	2114	puts ("wow, /etc/passwd is not there, expect problems. "
1389	"if this is windows, they already arrived\n");	2115	"if this is windows, they already arrived\n");
1390	}	2116	}
1391		2117
1392	...	2118	...
1393	ev_stat passwd;	2119	ev_stat passwd;
1394		2120
1395	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	2121	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1396	ev_stat_start (loop, &passwd);	2122	ev_stat_start (loop, &passwd);
		2123
		2124	Example: Like above, but additionally use a one-second delay so we do not
		2125	miss updates (however, frequent updates will delay processing, too, so
		2126	one might do the work both on C<ev_stat> callback invocation I<and> on
		2127	C<ev_timer> callback invocation).
		2128
		2129	static ev_stat passwd;
		2130	static ev_timer timer;
		2131
		2132	static void
		2133	timer_cb (EV_P_ ev_timer *w, int revents)
		2134	{
		2135	ev_timer_stop (EV_A_ w);
		2136
		2137	/* now it's one second after the most recent passwd change */
		2138	}
		2139
		2140	static void
		2141	stat_cb (EV_P_ ev_stat *w, int revents)
		2142	{
		2143	/* reset the one-second timer */
		2144	ev_timer_again (EV_A_ &timer);
		2145	}
		2146
		2147	...
		2148	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2149	ev_stat_start (loop, &passwd);
		2150	ev_timer_init (&timer, timer_cb, 0., 1.02);
1397		2151
1398		2152
1399	=head2 C<ev_idle> - when you've got nothing better to do...	2153	=head2 C<ev_idle> - when you've got nothing better to do...
1400		2154
1401	Idle watchers trigger events when no other events of the same or higher	2155	Idle watchers trigger events when no other events of the same or higher
1402	priority are pending (prepare, check and other idle watchers do not	2156	priority are pending (prepare, check and other idle watchers do not count
1403	count).	2157	as receiving "events").
1404		2158
1405	That is, as long as your process is busy handling sockets or timeouts	2159	That is, as long as your process is busy handling sockets or timeouts
1406	(or even signals, imagine) of the same or higher priority it will not be	2160	(or even signals, imagine) of the same or higher priority it will not be
1407	triggered. But when your process is idle (or only lower-priority watchers	2161	triggered. But when your process is idle (or only lower-priority watchers
1408	are pending), the idle watchers are being called once per event loop	2162	are pending), the idle watchers are being called once per event loop
…		…
1415	Apart from keeping your process non-blocking (which is a useful	2169	Apart from keeping your process non-blocking (which is a useful
1416	effect on its own sometimes), idle watchers are a good place to do	2170	effect on its own sometimes), idle watchers are a good place to do
1417	"pseudo-background processing", or delay processing stuff to after the	2171	"pseudo-background processing", or delay processing stuff to after the
1418	event loop has handled all outstanding events.	2172	event loop has handled all outstanding events.
1419		2173
		2174	=head3 Watcher-Specific Functions and Data Members
		2175
1420	=over 4	2176	=over 4
1421		2177
1422	=item ev_idle_init (ev_signal *, callback)	2178	=item ev_idle_init (ev_signal *, callback)
1423		2179
1424	Initialises and configures the idle watcher - it has no parameters of any	2180	Initialises and configures the idle watcher - it has no parameters of any
1425	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2181	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1426	believe me.	2182	believe me.
1427		2183
1428	=back	2184	=back
1429		2185
		2186	=head3 Examples
		2187
1430	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	2188	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1431	callback, free it. Also, use no error checking, as usual.	2189	callback, free it. Also, use no error checking, as usual.
1432		2190
1433	static void	2191	static void
1434	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2192	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1435	{	2193	{
1436	free (w);	2194	free (w);
1437	// now do something you wanted to do when the program has	2195	// now do something you wanted to do when the program has
1438	// no longer asnything immediate to do.	2196	// no longer anything immediate to do.
1439	}	2197	}
1440		2198
1441	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2199	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1442	ev_idle_init (idle_watcher, idle_cb);	2200	ev_idle_init (idle_watcher, idle_cb);
1443	ev_idle_start (loop, idle_cb);	2201	ev_idle_start (loop, idle_cb);
1444		2202
1445		2203
1446	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	2204	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1447		2205
1448	Prepare and check watchers are usually (but not always) used in tandem:	2206	Prepare and check watchers are usually (but not always) used in pairs:
1449	prepare watchers get invoked before the process blocks and check watchers	2207	prepare watchers get invoked before the process blocks and check watchers
1450	afterwards.	2208	afterwards.
1451		2209
1452	You I<must not> call C<ev_loop> or similar functions that enter	2210	You I<must not> call C<ev_loop> or similar functions that enter
1453	the current event loop from either C<ev_prepare> or C<ev_check>	2211	the current event loop from either C<ev_prepare> or C<ev_check>
…		…
1456	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	2214	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
1457	C<ev_check> so if you have one watcher of each kind they will always be	2215	C<ev_check> so if you have one watcher of each kind they will always be
1458	called in pairs bracketing the blocking call.	2216	called in pairs bracketing the blocking call.
1459		2217
1460	Their main purpose is to integrate other event mechanisms into libev and	2218	Their main purpose is to integrate other event mechanisms into libev and
1461	their use is somewhat advanced. This could be used, for example, to track	2219	their use is somewhat advanced. They could be used, for example, to track
1462	variable changes, implement your own watchers, integrate net-snmp or a	2220	variable changes, implement your own watchers, integrate net-snmp or a
1463	coroutine library and lots more. They are also occasionally useful if	2221	coroutine library and lots more. They are also occasionally useful if
1464	you cache some data and want to flush it before blocking (for example,	2222	you cache some data and want to flush it before blocking (for example,
1465	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	2223	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1466	watcher).	2224	watcher).
1467		2225
1468	This is done by examining in each prepare call which file descriptors need	2226	This is done by examining in each prepare call which file descriptors
1469	to be watched by the other library, registering C<ev_io> watchers for	2227	need to be watched by the other library, registering C<ev_io> watchers
1470	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2228	for them and starting an C<ev_timer> watcher for any timeouts (many
1471	provide just this functionality). Then, in the check watcher you check for	2229	libraries provide exactly this functionality). Then, in the check watcher,
1472	any events that occured (by checking the pending status of all watchers	2230	you check for any events that occurred (by checking the pending status
1473	and stopping them) and call back into the library. The I/O and timer	2231	of all watchers and stopping them) and call back into the library. The
1474	callbacks will never actually be called (but must be valid nevertheless,	2232	I/O and timer callbacks will never actually be called (but must be valid
1475	because you never know, you know?).	2233	nevertheless, because you never know, you know?).
1476		2234
1477	As another example, the Perl Coro module uses these hooks to integrate	2235	As another example, the Perl Coro module uses these hooks to integrate
1478	coroutines into libev programs, by yielding to other active coroutines	2236	coroutines into libev programs, by yielding to other active coroutines
1479	during each prepare and only letting the process block if no coroutines	2237	during each prepare and only letting the process block if no coroutines
1480	are ready to run (it's actually more complicated: it only runs coroutines	2238	are ready to run (it's actually more complicated: it only runs coroutines
1481	with priority higher than or equal to the event loop and one coroutine	2239	with priority higher than or equal to the event loop and one coroutine
1482	of lower priority, but only once, using idle watchers to keep the event	2240	of lower priority, but only once, using idle watchers to keep the event
1483	loop from blocking if lower-priority coroutines are active, thus mapping	2241	loop from blocking if lower-priority coroutines are active, thus mapping
1484	low-priority coroutines to idle/background tasks).	2242	low-priority coroutines to idle/background tasks).
1485		2243
		2244	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2245	priority, to ensure that they are being run before any other watchers
		2246	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2247
		2248	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2249	activate ("feed") events into libev. While libev fully supports this, they
		2250	might get executed before other C<ev_check> watchers did their job. As
		2251	C<ev_check> watchers are often used to embed other (non-libev) event
		2252	loops those other event loops might be in an unusable state until their
		2253	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2254	others).
		2255
		2256	=head3 Watcher-Specific Functions and Data Members
		2257
1486	=over 4	2258	=over 4
1487		2259
1488	=item ev_prepare_init (ev_prepare *, callback)	2260	=item ev_prepare_init (ev_prepare *, callback)
1489		2261
1490	=item ev_check_init (ev_check *, callback)	2262	=item ev_check_init (ev_check *, callback)
1491		2263
1492	Initialises and configures the prepare or check watcher - they have no	2264	Initialises and configures the prepare or check watcher - they have no
1493	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2265	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1494	macros, but using them is utterly, utterly and completely pointless.	2266	macros, but using them is utterly, utterly, utterly and completely
		2267	pointless.
1495		2268
1496	=back	2269	=back
1497		2270
1498	Example: To include a library such as adns, you would add IO watchers	2271	=head3 Examples
1499	and a timeout watcher in a prepare handler, as required by libadns, and	2272
		2273	There are a number of principal ways to embed other event loops or modules
		2274	into libev. Here are some ideas on how to include libadns into libev
		2275	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2276	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2277	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2278	Glib event loop).
		2279
		2280	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1500	in a check watcher, destroy them and call into libadns. What follows is	2281	and in a check watcher, destroy them and call into libadns. What follows
1501	pseudo-code only of course:	2282	is pseudo-code only of course. This requires you to either use a low
		2283	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2284	the callbacks for the IO/timeout watchers might not have been called yet.
1502		2285
1503	static ev_io iow [nfd];	2286	static ev_io iow [nfd];
1504	static ev_timer tw;	2287	static ev_timer tw;
1505		2288
1506	static void	2289	static void
1507	io_cb (ev_loop *loop, ev_io *w, int revents)	2290	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1508	{	2291	{
1509	// set the relevant poll flags
1510	// could also call adns_processreadable etc. here
1511	struct pollfd *fd = (struct pollfd *)w->data;
1512	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1513	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1514	}	2292	}
1515		2293
1516	// create io watchers for each fd and a timer before blocking	2294	// create io watchers for each fd and a timer before blocking
1517	static void	2295	static void
1518	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	2296	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1519	{	2297	{
1520	int timeout = 3600000;	2298	int timeout = 3600000;
1521	struct pollfd fds [nfd];	2299	struct pollfd fds [nfd];
1522	// actual code will need to loop here and realloc etc.	2300	// actual code will need to loop here and realloc etc.
1523	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	2301	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1524		2302
1525	/* the callback is illegal, but won't be called as we stop during check */	2303	/* the callback is illegal, but won't be called as we stop during check */
1526	ev_timer_init (&tw, 0, timeout * 1e-3);	2304	ev_timer_init (&tw, 0, timeout * 1e-3);
1527	ev_timer_start (loop, &tw);	2305	ev_timer_start (loop, &tw);
1528		2306
1529	// create on ev_io per pollfd	2307	// create one ev_io per pollfd
1530	for (int i = 0; i < nfd; ++i)	2308	for (int i = 0; i < nfd; ++i)
1531	{	2309	{
1532	ev_io_init (iow + i, io_cb, fds [i].fd,	2310	ev_io_init (iow + i, io_cb, fds [i].fd,
1533	((fds [i].events & POLLIN ? EV_READ : 0)	2311	((fds [i].events & POLLIN ? EV_READ : 0)
1534	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	2312	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1535		2313
1536	fds [i].revents = 0;	2314	fds [i].revents = 0;
1537	iow [i].data = fds + i;
1538	ev_io_start (loop, iow + i);	2315	ev_io_start (loop, iow + i);
1539	}	2316	}
1540	}	2317	}
1541		2318
1542	// stop all watchers after blocking	2319	// stop all watchers after blocking
1543	static void	2320	static void
1544	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	2321	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1545	{	2322	{
1546	ev_timer_stop (loop, &tw);	2323	ev_timer_stop (loop, &tw);
1547		2324
1548	for (int i = 0; i < nfd; ++i)	2325	for (int i = 0; i < nfd; ++i)
		2326	{
		2327	// set the relevant poll flags
		2328	// could also call adns_processreadable etc. here
		2329	struct pollfd *fd = fds + i;
		2330	int revents = ev_clear_pending (iow + i);
		2331	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2332	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2333
		2334	// now stop the watcher
1549	ev_io_stop (loop, iow + i);	2335	ev_io_stop (loop, iow + i);
		2336	}
1550		2337
1551	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	2338	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1552	}	2339	}
		2340
		2341	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2342	in the prepare watcher and would dispose of the check watcher.
		2343
		2344	Method 3: If the module to be embedded supports explicit event
		2345	notification (libadns does), you can also make use of the actual watcher
		2346	callbacks, and only destroy/create the watchers in the prepare watcher.
		2347
		2348	static void
		2349	timer_cb (EV_P_ ev_timer *w, int revents)
		2350	{
		2351	adns_state ads = (adns_state)w->data;
		2352	update_now (EV_A);
		2353
		2354	adns_processtimeouts (ads, &tv_now);
		2355	}
		2356
		2357	static void
		2358	io_cb (EV_P_ ev_io *w, int revents)
		2359	{
		2360	adns_state ads = (adns_state)w->data;
		2361	update_now (EV_A);
		2362
		2363	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2364	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2365	}
		2366
		2367	// do not ever call adns_afterpoll
		2368
		2369	Method 4: Do not use a prepare or check watcher because the module you
		2370	want to embed is not flexible enough to support it. Instead, you can
		2371	override their poll function. The drawback with this solution is that the
		2372	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2373	this approach, effectively embedding EV as a client into the horrible
		2374	libglib event loop.
		2375
		2376	static gint
		2377	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2378	{
		2379	int got_events = 0;
		2380
		2381	for (n = 0; n < nfds; ++n)
		2382	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2383
		2384	if (timeout >= 0)
		2385	// create/start timer
		2386
		2387	// poll
		2388	ev_loop (EV_A_ 0);
		2389
		2390	// stop timer again
		2391	if (timeout >= 0)
		2392	ev_timer_stop (EV_A_ &to);
		2393
		2394	// stop io watchers again - their callbacks should have set
		2395	for (n = 0; n < nfds; ++n)
		2396	ev_io_stop (EV_A_ iow [n]);
		2397
		2398	return got_events;
		2399	}
1553		2400
1554		2401
1555	=head2 C<ev_embed> - when one backend isn't enough...	2402	=head2 C<ev_embed> - when one backend isn't enough...
1556		2403
1557	This is a rather advanced watcher type that lets you embed one event loop	2404	This is a rather advanced watcher type that lets you embed one event loop
…		…
1563	prioritise I/O.	2410	prioritise I/O.
1564		2411
1565	As an example for a bug workaround, the kqueue backend might only support	2412	As an example for a bug workaround, the kqueue backend might only support
1566	sockets on some platform, so it is unusable as generic backend, but you	2413	sockets on some platform, so it is unusable as generic backend, but you
1567	still want to make use of it because you have many sockets and it scales	2414	still want to make use of it because you have many sockets and it scales
1568	so nicely. In this case, you would create a kqueue-based loop and embed it	2415	so nicely. In this case, you would create a kqueue-based loop and embed
1569	into your default loop (which might use e.g. poll). Overall operation will	2416	it into your default loop (which might use e.g. poll). Overall operation
1570	be a bit slower because first libev has to poll and then call kevent, but	2417	will be a bit slower because first libev has to call C<poll> and then
1571	at least you can use both at what they are best.	2418	C<kevent>, but at least you can use both mechanisms for what they are
		2419	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
1572		2420
1573	As for prioritising I/O: rarely you have the case where some fds have	2421	As for prioritising I/O: under rare circumstances you have the case where
1574	to be watched and handled very quickly (with low latency), and even	2422	some fds have to be watched and handled very quickly (with low latency),
1575	priorities and idle watchers might have too much overhead. In this case	2423	and even priorities and idle watchers might have too much overhead. In
1576	you would put all the high priority stuff in one loop and all the rest in	2424	this case you would put all the high priority stuff in one loop and all
1577	a second one, and embed the second one in the first.	2425	the rest in a second one, and embed the second one in the first.
1578		2426
1579	As long as the watcher is active, the callback will be invoked every time	2427	As long as the watcher is active, the callback will be invoked every time
1580	there might be events pending in the embedded loop. The callback must then	2428	there might be events pending in the embedded loop. The callback must then
1581	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	2429	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke
1582	their callbacks (you could also start an idle watcher to give the embedded	2430	their callbacks (you could also start an idle watcher to give the embedded
…		…
1590	interested in that.	2438	interested in that.
1591		2439
1592	Also, there have not currently been made special provisions for forking:	2440	Also, there have not currently been made special provisions for forking:
1593	when you fork, you not only have to call C<ev_loop_fork> on both loops,	2441	when you fork, you not only have to call C<ev_loop_fork> on both loops,
1594	but you will also have to stop and restart any C<ev_embed> watchers	2442	but you will also have to stop and restart any C<ev_embed> watchers
1595	yourself.	2443	yourself - but you can use a fork watcher to handle this automatically,
		2444	and future versions of libev might do just that.
1596		2445
1597	Unfortunately, not all backends are embeddable, only the ones returned by	2446	Unfortunately, not all backends are embeddable: only the ones returned by
1598	C<ev_embeddable_backends> are, which, unfortunately, does not include any	2447	C<ev_embeddable_backends> are, which, unfortunately, does not include any
1599	portable one.	2448	portable one.
1600		2449
1601	So when you want to use this feature you will always have to be prepared	2450	So when you want to use this feature you will always have to be prepared
1602	that you cannot get an embeddable loop. The recommended way to get around	2451	that you cannot get an embeddable loop. The recommended way to get around
1603	this is to have a separate variables for your embeddable loop, try to	2452	this is to have a separate variables for your embeddable loop, try to
1604	create it, and if that fails, use the normal loop for everything:	2453	create it, and if that fails, use the normal loop for everything.
1605		2454
1606	struct ev_loop *loop_hi = ev_default_init (0);	2455	=head3 C<ev_embed> and fork
1607	struct ev_loop *loop_lo = 0;
1608	struct ev_embed embed;
1609
1610	// see if there is a chance of getting one that works
1611	// (remember that a flags value of 0 means autodetection)
1612	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1613	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1614	: 0;
1615		2456
1616	// if we got one, then embed it, otherwise default to loop_hi	2457	While the C<ev_embed> watcher is running, forks in the embedding loop will
1617	if (loop_lo)	2458	automatically be applied to the embedded loop as well, so no special
1618	{	2459	fork handling is required in that case. When the watcher is not running,
1619	ev_embed_init (&embed, 0, loop_lo);	2460	however, it is still the task of the libev user to call C<ev_loop_fork ()>
1620	ev_embed_start (loop_hi, &embed);	2461	as applicable.
1621	}	2462
1622	else	2463	=head3 Watcher-Specific Functions and Data Members
1623	loop_lo = loop_hi;
1624		2464
1625	=over 4	2465	=over 4
1626		2466
1627	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	2467	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1628		2468
…		…
1630		2470
1631	Configures the watcher to embed the given loop, which must be	2471	Configures the watcher to embed the given loop, which must be
1632	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	2472	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1633	invoked automatically, otherwise it is the responsibility of the callback	2473	invoked automatically, otherwise it is the responsibility of the callback
1634	to invoke it (it will continue to be called until the sweep has been done,	2474	to invoke it (it will continue to be called until the sweep has been done,
1635	if you do not want thta, you need to temporarily stop the embed watcher).	2475	if you do not want that, you need to temporarily stop the embed watcher).
1636		2476
1637	=item ev_embed_sweep (loop, ev_embed *)	2477	=item ev_embed_sweep (loop, ev_embed *)
1638		2478
1639	Make a single, non-blocking sweep over the embedded loop. This works	2479	Make a single, non-blocking sweep over the embedded loop. This works
1640	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	2480	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1641	apropriate way for embedded loops.	2481	appropriate way for embedded loops.
1642		2482
1643	=item struct ev_loop *loop [read-only]	2483	=item struct ev_loop *other [read-only]
1644		2484
1645	The embedded event loop.	2485	The embedded event loop.
1646		2486
1647	=back	2487	=back
		2488
		2489	=head3 Examples
		2490
		2491	Example: Try to get an embeddable event loop and embed it into the default
		2492	event loop. If that is not possible, use the default loop. The default
		2493	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2494	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2495	used).
		2496
		2497	struct ev_loop *loop_hi = ev_default_init (0);
		2498	struct ev_loop *loop_lo = 0;
		2499	ev_embed embed;
		2500
		2501	// see if there is a chance of getting one that works
		2502	// (remember that a flags value of 0 means autodetection)
		2503	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2504	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2505	: 0;
		2506
		2507	// if we got one, then embed it, otherwise default to loop_hi
		2508	if (loop_lo)
		2509	{
		2510	ev_embed_init (&embed, 0, loop_lo);
		2511	ev_embed_start (loop_hi, &embed);
		2512	}
		2513	else
		2514	loop_lo = loop_hi;
		2515
		2516	Example: Check if kqueue is available but not recommended and create
		2517	a kqueue backend for use with sockets (which usually work with any
		2518	kqueue implementation). Store the kqueue/socket-only event loop in
		2519	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2520
		2521	struct ev_loop *loop = ev_default_init (0);
		2522	struct ev_loop *loop_socket = 0;
		2523	ev_embed embed;
		2524
		2525	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2526	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2527	{
		2528	ev_embed_init (&embed, 0, loop_socket);
		2529	ev_embed_start (loop, &embed);
		2530	}
		2531
		2532	if (!loop_socket)
		2533	loop_socket = loop;
		2534
		2535	// now use loop_socket for all sockets, and loop for everything else
1648		2536
1649		2537
1650	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	2538	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1651		2539
1652	Fork watchers are called when a C<fork ()> was detected (usually because	2540	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
1655	event loop blocks next and before C<ev_check> watchers are being called,	2543	event loop blocks next and before C<ev_check> watchers are being called,
1656	and only in the child after the fork. If whoever good citizen calling	2544	and only in the child after the fork. If whoever good citizen calling
1657	C<ev_default_fork> cheats and calls it in the wrong process, the fork	2545	C<ev_default_fork> cheats and calls it in the wrong process, the fork
1658	handlers will be invoked, too, of course.	2546	handlers will be invoked, too, of course.
1659		2547
		2548	=head3 Watcher-Specific Functions and Data Members
		2549
1660	=over 4	2550	=over 4
1661		2551
1662	=item ev_fork_init (ev_signal *, callback)	2552	=item ev_fork_init (ev_signal *, callback)
1663		2553
1664	Initialises and configures the fork watcher - it has no parameters of any	2554	Initialises and configures the fork watcher - it has no parameters of any
…		…
1666	believe me.	2556	believe me.
1667		2557
1668	=back	2558	=back
1669		2559
1670		2560
		2561	=head2 C<ev_async> - how to wake up another event loop
		2562
		2563	In general, you cannot use an C<ev_loop> from multiple threads or other
		2564	asynchronous sources such as signal handlers (as opposed to multiple event
		2565	loops - those are of course safe to use in different threads).
		2566
		2567	Sometimes, however, you need to wake up another event loop you do not
		2568	control, for example because it belongs to another thread. This is what
		2569	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2570	can signal it by calling C<ev_async_send>, which is thread- and signal
		2571	safe.
		2572
		2573	This functionality is very similar to C<ev_signal> watchers, as signals,
		2574	too, are asynchronous in nature, and signals, too, will be compressed
		2575	(i.e. the number of callback invocations may be less than the number of
		2576	C<ev_async_sent> calls).
		2577
		2578	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2579	just the default loop.
		2580
		2581	=head3 Queueing
		2582
		2583	C<ev_async> does not support queueing of data in any way. The reason
		2584	is that the author does not know of a simple (or any) algorithm for a
		2585	multiple-writer-single-reader queue that works in all cases and doesn't
		2586	need elaborate support such as pthreads.
		2587
		2588	That means that if you want to queue data, you have to provide your own
		2589	queue. But at least I can tell you how to implement locking around your
		2590	queue:
		2591
		2592	=over 4
		2593
		2594	=item queueing from a signal handler context
		2595
		2596	To implement race-free queueing, you simply add to the queue in the signal
		2597	handler but you block the signal handler in the watcher callback. Here is
		2598	an example that does that for some fictitious SIGUSR1 handler:
		2599
		2600	static ev_async mysig;
		2601
		2602	static void
		2603	sigusr1_handler (void)
		2604	{
		2605	sometype data;
		2606
		2607	// no locking etc.
		2608	queue_put (data);
		2609	ev_async_send (EV_DEFAULT_ &mysig);
		2610	}
		2611
		2612	static void
		2613	mysig_cb (EV_P_ ev_async *w, int revents)
		2614	{
		2615	sometype data;
		2616	sigset_t block, prev;
		2617
		2618	sigemptyset (&block);
		2619	sigaddset (&block, SIGUSR1);
		2620	sigprocmask (SIG_BLOCK, &block, &prev);
		2621
		2622	while (queue_get (&data))
		2623	process (data);
		2624
		2625	if (sigismember (&prev, SIGUSR1)
		2626	sigprocmask (SIG_UNBLOCK, &block, 0);
		2627	}
		2628
		2629	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2630	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2631	either...).
		2632
		2633	=item queueing from a thread context
		2634
		2635	The strategy for threads is different, as you cannot (easily) block
		2636	threads but you can easily preempt them, so to queue safely you need to
		2637	employ a traditional mutex lock, such as in this pthread example:
		2638
		2639	static ev_async mysig;
		2640	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2641
		2642	static void
		2643	otherthread (void)
		2644	{
		2645	// only need to lock the actual queueing operation
		2646	pthread_mutex_lock (&mymutex);
		2647	queue_put (data);
		2648	pthread_mutex_unlock (&mymutex);
		2649
		2650	ev_async_send (EV_DEFAULT_ &mysig);
		2651	}
		2652
		2653	static void
		2654	mysig_cb (EV_P_ ev_async *w, int revents)
		2655	{
		2656	pthread_mutex_lock (&mymutex);
		2657
		2658	while (queue_get (&data))
		2659	process (data);
		2660
		2661	pthread_mutex_unlock (&mymutex);
		2662	}
		2663
		2664	=back
		2665
		2666
		2667	=head3 Watcher-Specific Functions and Data Members
		2668
		2669	=over 4
		2670
		2671	=item ev_async_init (ev_async *, callback)
		2672
		2673	Initialises and configures the async watcher - it has no parameters of any
		2674	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		2675	trust me.
		2676
		2677	=item ev_async_send (loop, ev_async *)
		2678
		2679	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2680	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2681	C<ev_feed_event>, this call is safe to do from other threads, signal or
		2682	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		2683	section below on what exactly this means).
		2684
		2685	This call incurs the overhead of a system call only once per loop iteration,
		2686	so while the overhead might be noticeable, it doesn't apply to repeated
		2687	calls to C<ev_async_send>.
		2688
		2689	=item bool = ev_async_pending (ev_async *)
		2690
		2691	Returns a non-zero value when C<ev_async_send> has been called on the
		2692	watcher but the event has not yet been processed (or even noted) by the
		2693	event loop.
		2694
		2695	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		2696	the loop iterates next and checks for the watcher to have become active,
		2697	it will reset the flag again. C<ev_async_pending> can be used to very
		2698	quickly check whether invoking the loop might be a good idea.
		2699
		2700	Not that this does I<not> check whether the watcher itself is pending, only
		2701	whether it has been requested to make this watcher pending.
		2702
		2703	=back
		2704
		2705
1671	=head1 OTHER FUNCTIONS	2706	=head1 OTHER FUNCTIONS
1672		2707
1673	There are some other functions of possible interest. Described. Here. Now.	2708	There are some other functions of possible interest. Described. Here. Now.
1674		2709
1675	=over 4	2710	=over 4
1676		2711
1677	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2712	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
1678		2713
1679	This function combines a simple timer and an I/O watcher, calls your	2714	This function combines a simple timer and an I/O watcher, calls your
1680	callback on whichever event happens first and automatically stop both	2715	callback on whichever event happens first and automatically stops both
1681	watchers. This is useful if you want to wait for a single event on an fd	2716	watchers. This is useful if you want to wait for a single event on an fd
1682	or timeout without having to allocate/configure/start/stop/free one or	2717	or timeout without having to allocate/configure/start/stop/free one or
1683	more watchers yourself.	2718	more watchers yourself.
1684		2719
1685	If C<fd> is less than 0, then no I/O watcher will be started and events	2720	If C<fd> is less than 0, then no I/O watcher will be started and the
1686	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	2721	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
1687	C<events> set will be craeted and started.	2722	the given C<fd> and C<events> set will be created and started.
1688		2723
1689	If C<timeout> is less than 0, then no timeout watcher will be	2724	If C<timeout> is less than 0, then no timeout watcher will be
1690	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	2725	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
1691	repeat = 0) will be started. While C<0> is a valid timeout, it is of	2726	repeat = 0) will be started. C<0> is a valid timeout.
1692	dubious value.
1693		2727
1694	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	2728	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
1695	passed an C<revents> set like normal event callbacks (a combination of	2729	passed an C<revents> set like normal event callbacks (a combination of
1696	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	2730	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
1697	value passed to C<ev_once>:	2731	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2732	a timeout and an io event at the same time - you probably should give io
		2733	events precedence.
1698		2734
		2735	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		2736
1699	static void stdin_ready (int revents, void *arg)	2737	static void stdin_ready (int revents, void *arg)
1700	{	2738	{
1701	if (revents & EV_TIMEOUT)
1702	/* doh, nothing entered */;
1703	else if (revents & EV_READ)	2739	if (revents & EV_READ)
1704	/* stdin might have data for us, joy! */;	2740	/* stdin might have data for us, joy! */;
		2741	else if (revents & EV_TIMEOUT)
		2742	/* doh, nothing entered */;
1705	}	2743	}
1706		2744
1707	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	2745	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1708		2746
1709	=item ev_feed_event (ev_loop *, watcher *, int revents)	2747	=item ev_feed_event (struct ev_loop *, watcher *, int revents)
1710		2748
1711	Feeds the given event set into the event loop, as if the specified event	2749	Feeds the given event set into the event loop, as if the specified event
1712	had happened for the specified watcher (which must be a pointer to an	2750	had happened for the specified watcher (which must be a pointer to an
1713	initialised but not necessarily started event watcher).	2751	initialised but not necessarily started event watcher).
1714		2752
1715	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	2753	=item ev_feed_fd_event (struct ev_loop *, int fd, int revents)
1716		2754
1717	Feed an event on the given fd, as if a file descriptor backend detected	2755	Feed an event on the given fd, as if a file descriptor backend detected
1718	the given events it.	2756	the given events it.
1719		2757
1720	=item ev_feed_signal_event (ev_loop *loop, int signum)	2758	=item ev_feed_signal_event (struct ev_loop *loop, int signum)
1721		2759
1722	Feed an event as if the given signal occured (C<loop> must be the default	2760	Feed an event as if the given signal occurred (C<loop> must be the default
1723	loop!).	2761	loop!).
1724		2762
1725	=back	2763	=back
1726		2764
1727		2765
…		…
1743		2781
1744	=item * Priorities are not currently supported. Initialising priorities	2782	=item * Priorities are not currently supported. Initialising priorities
1745	will fail and all watchers will have the same priority, even though there	2783	will fail and all watchers will have the same priority, even though there
1746	is an ev_pri field.	2784	is an ev_pri field.
1747		2785
		2786	=item * In libevent, the last base created gets the signals, in libev, the
		2787	first base created (== the default loop) gets the signals.
		2788
1748	=item * Other members are not supported.	2789	=item * Other members are not supported.
1749		2790
1750	=item * The libev emulation is I<not> ABI compatible to libevent, you need	2791	=item * The libev emulation is I<not> ABI compatible to libevent, you need
1751	to use the libev header file and library.	2792	to use the libev header file and library.
1752		2793
1753	=back	2794	=back
1754		2795
1755	=head1 C++ SUPPORT	2796	=head1 C++ SUPPORT
1756		2797
1757	Libev comes with some simplistic wrapper classes for C++ that mainly allow	2798	Libev comes with some simplistic wrapper classes for C++ that mainly allow
1758	you to use some convinience methods to start/stop watchers and also change	2799	you to use some convenience methods to start/stop watchers and also change
1759	the callback model to a model using method callbacks on objects.	2800	the callback model to a model using method callbacks on objects.
1760		2801
1761	To use it,	2802	To use it,
1762		2803
1763	#include <ev++.h>	2804	#include <ev++.h>
1764		2805
1765	This automatically includes F<ev.h> and puts all of its definitions (many	2806	This automatically includes F<ev.h> and puts all of its definitions (many
1766	of them macros) into the global namespace. All C++ specific things are	2807	of them macros) into the global namespace. All C++ specific things are
1767	put into the C<ev> namespace. It should support all the same embedding	2808	put into the C<ev> namespace. It should support all the same embedding
1768	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.	2809	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
…		…
1835	your compiler is good :), then the method will be fully inlined into the	2876	your compiler is good :), then the method will be fully inlined into the
1836	thunking function, making it as fast as a direct C callback.	2877	thunking function, making it as fast as a direct C callback.
1837		2878
1838	Example: simple class declaration and watcher initialisation	2879	Example: simple class declaration and watcher initialisation
1839		2880
1840	struct myclass	2881	struct myclass
1841	{	2882	{
1842	void io_cb (ev::io &w, int revents) { }	2883	void io_cb (ev::io &w, int revents) { }
1843	}	2884	}
1844		2885
1845	myclass obj;	2886	myclass obj;
1846	ev::io iow;	2887	ev::io iow;
1847	iow.set <myclass, &myclass::io_cb> (&obj);	2888	iow.set <myclass, &myclass::io_cb> (&obj);
1848		2889
1849	=item w->set (void (*function)(watcher &w, int), void *data = 0)	2890	=item w->set<function> (void *data = 0)
1850		2891
1851	Also sets a callback, but uses a static method or plain function as	2892	Also sets a callback, but uses a static method or plain function as
1852	callback. The optional C<data> argument will be stored in the watcher's	2893	callback. The optional C<data> argument will be stored in the watcher's
1853	C<data> member and is free for you to use.	2894	C<data> member and is free for you to use.
1854		2895
		2896	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		2897
1855	See the method-C<set> above for more details.	2898	See the method-C<set> above for more details.
		2899
		2900	Example: Use a plain function as callback.
		2901
		2902	static void io_cb (ev::io &w, int revents) { }
		2903	iow.set <io_cb> ();
1856		2904
1857	=item w->set (struct ev_loop *)	2905	=item w->set (struct ev_loop *)
1858		2906
1859	Associates a different C<struct ev_loop> with this watcher. You can only	2907	Associates a different C<struct ev_loop> with this watcher. You can only
1860	do this when the watcher is inactive (and not pending either).	2908	do this when the watcher is inactive (and not pending either).
1861		2909
1862	=item w->set ([args])	2910	=item w->set ([arguments])
1863		2911
1864	Basically the same as C<ev_TYPE_set>, with the same args. Must be	2912	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
1865	called at least once. Unlike the C counterpart, an active watcher gets	2913	called at least once. Unlike the C counterpart, an active watcher gets
1866	automatically stopped and restarted when reconfiguring it with this	2914	automatically stopped and restarted when reconfiguring it with this
1867	method.	2915	method.
1868		2916
1869	=item w->start ()	2917	=item w->start ()
…		…
1873		2921
1874	=item w->stop ()	2922	=item w->stop ()
1875		2923
1876	Stops the watcher if it is active. Again, no C<loop> argument.	2924	Stops the watcher if it is active. Again, no C<loop> argument.
1877		2925
1878	=item w->again () C<ev::timer>, C<ev::periodic> only	2926	=item w->again () (C<ev::timer>, C<ev::periodic> only)
1879		2927
1880	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding	2928	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
1881	C<ev_TYPE_again> function.	2929	C<ev_TYPE_again> function.
1882		2930
1883	=item w->sweep () C<ev::embed> only	2931	=item w->sweep () (C<ev::embed> only)
1884		2932
1885	Invokes C<ev_embed_sweep>.	2933	Invokes C<ev_embed_sweep>.
1886		2934
1887	=item w->update () C<ev::stat> only	2935	=item w->update () (C<ev::stat> only)
1888		2936
1889	Invokes C<ev_stat_stat>.	2937	Invokes C<ev_stat_stat>.
1890		2938
1891	=back	2939	=back
1892		2940
1893	=back	2941	=back
1894		2942
1895	Example: Define a class with an IO and idle watcher, start one of them in	2943	Example: Define a class with an IO and idle watcher, start one of them in
1896	the constructor.	2944	the constructor.
1897		2945
1898	class myclass	2946	class myclass
1899	{	2947	{
1900	ev_io io; void io_cb (ev::io &w, int revents);	2948	ev::io io ; void io_cb (ev::io &w, int revents);
1901	ev_idle idle void idle_cb (ev::idle &w, int revents);	2949	ev::idle idle; void idle_cb (ev::idle &w, int revents);
1902		2950
1903	myclass ();	2951	myclass (int fd)
1904	}	2952	{
1905
1906	myclass::myclass (int fd)
1907	{
1908	io .set <myclass, &myclass::io_cb > (this);	2953	io .set <myclass, &myclass::io_cb > (this);
1909	idle.set <myclass, &myclass::idle_cb> (this);	2954	idle.set <myclass, &myclass::idle_cb> (this);
1910		2955
1911	io.start (fd, ev::READ);	2956	io.start (fd, ev::READ);
		2957	}
1912	}	2958	};
		2959
		2960
		2961	=head1 OTHER LANGUAGE BINDINGS
		2962
		2963	Libev does not offer other language bindings itself, but bindings for a
		2964	number of languages exist in the form of third-party packages. If you know
		2965	any interesting language binding in addition to the ones listed here, drop
		2966	me a note.
		2967
		2968	=over 4
		2969
		2970	=item Perl
		2971
		2972	The EV module implements the full libev API and is actually used to test
		2973	libev. EV is developed together with libev. Apart from the EV core module,
		2974	there are additional modules that implement libev-compatible interfaces
		2975	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		2976	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		2977	and C<EV::Glib>).
		2978
		2979	It can be found and installed via CPAN, its homepage is at
		2980	L<http://software.schmorp.de/pkg/EV>.
		2981
		2982	=item Python
		2983
		2984	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		2985	seems to be quite complete and well-documented. Note, however, that the
		2986	patch they require for libev is outright dangerous as it breaks the ABI
		2987	for everybody else, and therefore, should never be applied in an installed
		2988	libev (if python requires an incompatible ABI then it needs to embed
		2989	libev).
		2990
		2991	=item Ruby
		2992
		2993	Tony Arcieri has written a ruby extension that offers access to a subset
		2994	of the libev API and adds file handle abstractions, asynchronous DNS and
		2995	more on top of it. It can be found via gem servers. Its homepage is at
		2996	L<http://rev.rubyforge.org/>.
		2997
		2998	=item D
		2999
		3000	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		3001	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		3002
		3003	=item Ocaml
		3004
		3005	Erkki Seppala has written Ocaml bindings for libev, to be found at
		3006	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		3007
		3008	=back
1913		3009
1914		3010
1915	=head1 MACRO MAGIC	3011	=head1 MACRO MAGIC
1916		3012
1917	Libev can be compiled with a variety of options, the most fundemantal is	3013	Libev can be compiled with a variety of options, the most fundamental
1918	C<EV_MULTIPLICITY>. This option determines whether (most) functions and	3014	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
1919	callbacks have an initial C<struct ev_loop *> argument.	3015	functions and callbacks have an initial C<struct ev_loop *> argument.
1920		3016
1921	To make it easier to write programs that cope with either variant, the	3017	To make it easier to write programs that cope with either variant, the
1922	following macros are defined:	3018	following macros are defined:
1923		3019
1924	=over 4	3020	=over 4
…		…
1927		3023
1928	This provides the loop I<argument> for functions, if one is required ("ev	3024	This provides the loop I<argument> for functions, if one is required ("ev
1929	loop argument"). The C<EV_A> form is used when this is the sole argument,	3025	loop argument"). The C<EV_A> form is used when this is the sole argument,
1930	C<EV_A_> is used when other arguments are following. Example:	3026	C<EV_A_> is used when other arguments are following. Example:
1931		3027
1932	ev_unref (EV_A);	3028	ev_unref (EV_A);
1933	ev_timer_add (EV_A_ watcher);	3029	ev_timer_add (EV_A_ watcher);
1934	ev_loop (EV_A_ 0);	3030	ev_loop (EV_A_ 0);
1935		3031
1936	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	3032	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
1937	which is often provided by the following macro.	3033	which is often provided by the following macro.
1938		3034
1939	=item C<EV_P>, C<EV_P_>	3035	=item C<EV_P>, C<EV_P_>
1940		3036
1941	This provides the loop I<parameter> for functions, if one is required ("ev	3037	This provides the loop I<parameter> for functions, if one is required ("ev
1942	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	3038	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
1943	C<EV_P_> is used when other parameters are following. Example:	3039	C<EV_P_> is used when other parameters are following. Example:
1944		3040
1945	// this is how ev_unref is being declared	3041	// this is how ev_unref is being declared
1946	static void ev_unref (EV_P);	3042	static void ev_unref (EV_P);
1947		3043
1948	// this is how you can declare your typical callback	3044	// this is how you can declare your typical callback
1949	static void cb (EV_P_ ev_timer *w, int revents)	3045	static void cb (EV_P_ ev_timer *w, int revents)
1950		3046
1951	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	3047	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
1952	suitable for use with C<EV_A>.	3048	suitable for use with C<EV_A>.
1953		3049
1954	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	3050	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
1955		3051
1956	Similar to the other two macros, this gives you the value of the default	3052	Similar to the other two macros, this gives you the value of the default
1957	loop, if multiple loops are supported ("ev loop default").	3053	loop, if multiple loops are supported ("ev loop default").
		3054
		3055	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3056
		3057	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3058	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3059	is undefined when the default loop has not been initialised by a previous
		3060	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3061
		3062	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3063	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
1958		3064
1959	=back	3065	=back
1960		3066
1961	Example: Declare and initialise a check watcher, utilising the above	3067	Example: Declare and initialise a check watcher, utilising the above
1962	macros so it will work regardless of whether multiple loops are supported	3068	macros so it will work regardless of whether multiple loops are supported
1963	or not.	3069	or not.
1964		3070
1965	static void	3071	static void
1966	check_cb (EV_P_ ev_timer *w, int revents)	3072	check_cb (EV_P_ ev_timer *w, int revents)
1967	{	3073	{
1968	ev_check_stop (EV_A_ w);	3074	ev_check_stop (EV_A_ w);
1969	}	3075	}
1970		3076
1971	ev_check check;	3077	ev_check check;
1972	ev_check_init (&check, check_cb);	3078	ev_check_init (&check, check_cb);
1973	ev_check_start (EV_DEFAULT_ &check);	3079	ev_check_start (EV_DEFAULT_ &check);
1974	ev_loop (EV_DEFAULT_ 0);	3080	ev_loop (EV_DEFAULT_ 0);
1975		3081
1976	=head1 EMBEDDING	3082	=head1 EMBEDDING
1977		3083
1978	Libev can (and often is) directly embedded into host	3084	Libev can (and often is) directly embedded into host
1979	applications. Examples of applications that embed it include the Deliantra	3085	applications. Examples of applications that embed it include the Deliantra
1980	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)	3086	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1981	and rxvt-unicode.	3087	and rxvt-unicode.
1982		3088
1983	The goal is to enable you to just copy the neecssary files into your	3089	The goal is to enable you to just copy the necessary files into your
1984	source directory without having to change even a single line in them, so	3090	source directory without having to change even a single line in them, so
1985	you can easily upgrade by simply copying (or having a checked-out copy of	3091	you can easily upgrade by simply copying (or having a checked-out copy of
1986	libev somewhere in your source tree).	3092	libev somewhere in your source tree).
1987		3093
1988	=head2 FILESETS	3094	=head2 FILESETS
1989		3095
1990	Depending on what features you need you need to include one or more sets of files	3096	Depending on what features you need you need to include one or more sets of files
1991	in your app.	3097	in your application.
1992		3098
1993	=head3 CORE EVENT LOOP	3099	=head3 CORE EVENT LOOP
1994		3100
1995	To include only the libev core (all the C<ev_*> functions), with manual	3101	To include only the libev core (all the C<ev_*> functions), with manual
1996	configuration (no autoconf):	3102	configuration (no autoconf):
1997		3103
1998	#define EV_STANDALONE 1	3104	#define EV_STANDALONE 1
1999	#include "ev.c"	3105	#include "ev.c"
2000		3106
2001	This will automatically include F<ev.h>, too, and should be done in a	3107	This will automatically include F<ev.h>, too, and should be done in a
2002	single C source file only to provide the function implementations. To use	3108	single C source file only to provide the function implementations. To use
2003	it, do the same for F<ev.h> in all files wishing to use this API (best	3109	it, do the same for F<ev.h> in all files wishing to use this API (best
2004	done by writing a wrapper around F<ev.h> that you can include instead and	3110	done by writing a wrapper around F<ev.h> that you can include instead and
2005	where you can put other configuration options):	3111	where you can put other configuration options):
2006		3112
2007	#define EV_STANDALONE 1	3113	#define EV_STANDALONE 1
2008	#include "ev.h"	3114	#include "ev.h"
2009		3115
2010	Both header files and implementation files can be compiled with a C++	3116	Both header files and implementation files can be compiled with a C++
2011	compiler (at least, thats a stated goal, and breakage will be treated	3117	compiler (at least, that's a stated goal, and breakage will be treated
2012	as a bug).	3118	as a bug).
2013		3119
2014	You need the following files in your source tree, or in a directory	3120	You need the following files in your source tree, or in a directory
2015	in your include path (e.g. in libev/ when using -Ilibev):	3121	in your include path (e.g. in libev/ when using -Ilibev):
2016		3122
2017	ev.h	3123	ev.h
2018	ev.c	3124	ev.c
2019	ev_vars.h	3125	ev_vars.h
2020	ev_wrap.h	3126	ev_wrap.h
2021		3127
2022	ev_win32.c required on win32 platforms only	3128	ev_win32.c required on win32 platforms only
2023		3129
2024	ev_select.c only when select backend is enabled (which is enabled by default)	3130	ev_select.c only when select backend is enabled (which is enabled by default)
2025	ev_poll.c only when poll backend is enabled (disabled by default)	3131	ev_poll.c only when poll backend is enabled (disabled by default)
2026	ev_epoll.c only when the epoll backend is enabled (disabled by default)	3132	ev_epoll.c only when the epoll backend is enabled (disabled by default)
2027	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	3133	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2028	ev_port.c only when the solaris port backend is enabled (disabled by default)	3134	ev_port.c only when the solaris port backend is enabled (disabled by default)
2029		3135
2030	F<ev.c> includes the backend files directly when enabled, so you only need	3136	F<ev.c> includes the backend files directly when enabled, so you only need
2031	to compile this single file.	3137	to compile this single file.
2032		3138
2033	=head3 LIBEVENT COMPATIBILITY API	3139	=head3 LIBEVENT COMPATIBILITY API
2034		3140
2035	To include the libevent compatibility API, also include:	3141	To include the libevent compatibility API, also include:
2036		3142
2037	#include "event.c"	3143	#include "event.c"
2038		3144
2039	in the file including F<ev.c>, and:	3145	in the file including F<ev.c>, and:
2040		3146
2041	#include "event.h"	3147	#include "event.h"
2042		3148
2043	in the files that want to use the libevent API. This also includes F<ev.h>.	3149	in the files that want to use the libevent API. This also includes F<ev.h>.
2044		3150
2045	You need the following additional files for this:	3151	You need the following additional files for this:
2046		3152
2047	event.h	3153	event.h
2048	event.c	3154	event.c
2049		3155
2050	=head3 AUTOCONF SUPPORT	3156	=head3 AUTOCONF SUPPORT
2051		3157
2052	Instead of using C<EV_STANDALONE=1> and providing your config in	3158	Instead of using C<EV_STANDALONE=1> and providing your configuration in
2053	whatever way you want, you can also C<m4_include([libev.m4])> in your	3159	whatever way you want, you can also C<m4_include([libev.m4])> in your
2054	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	3160	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
2055	include F<config.h> and configure itself accordingly.	3161	include F<config.h> and configure itself accordingly.
2056		3162
2057	For this of course you need the m4 file:	3163	For this of course you need the m4 file:
2058		3164
2059	libev.m4	3165	libev.m4
2060		3166
2061	=head2 PREPROCESSOR SYMBOLS/MACROS	3167	=head2 PREPROCESSOR SYMBOLS/MACROS
2062		3168
2063	Libev can be configured via a variety of preprocessor symbols you have to define	3169	Libev can be configured via a variety of preprocessor symbols you have to
2064	before including any of its files. The default is not to build for multiplicity	3170	define before including any of its files. The default in the absence of
2065	and only include the select backend.	3171	autoconf is documented for every option.
2066		3172
2067	=over 4	3173	=over 4
2068		3174
2069	=item EV_STANDALONE	3175	=item EV_STANDALONE
2070		3176
…		…
2075	F<event.h> that are not directly supported by the libev core alone.	3181	F<event.h> that are not directly supported by the libev core alone.
2076		3182
2077	=item EV_USE_MONOTONIC	3183	=item EV_USE_MONOTONIC
2078		3184
2079	If defined to be C<1>, libev will try to detect the availability of the	3185	If defined to be C<1>, libev will try to detect the availability of the
2080	monotonic clock option at both compiletime and runtime. Otherwise no use	3186	monotonic clock option at both compile time and runtime. Otherwise no use
2081	of the monotonic clock option will be attempted. If you enable this, you	3187	of the monotonic clock option will be attempted. If you enable this, you
2082	usually have to link against librt or something similar. Enabling it when	3188	usually have to link against librt or something similar. Enabling it when
2083	the functionality isn't available is safe, though, althoguh you have	3189	the functionality isn't available is safe, though, although you have
2084	to make sure you link against any libraries where the C<clock_gettime>	3190	to make sure you link against any libraries where the C<clock_gettime>
2085	function is hiding in (often F<-lrt>).	3191	function is hiding in (often F<-lrt>).
2086		3192
2087	=item EV_USE_REALTIME	3193	=item EV_USE_REALTIME
2088		3194
2089	If defined to be C<1>, libev will try to detect the availability of the	3195	If defined to be C<1>, libev will try to detect the availability of the
2090	realtime clock option at compiletime (and assume its availability at	3196	real-time clock option at compile time (and assume its availability at
2091	runtime if successful). Otherwise no use of the realtime clock option will	3197	runtime if successful). Otherwise no use of the real-time clock option will
2092	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	3198	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2093	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries	3199	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2094	in the description of C<EV_USE_MONOTONIC>, though.	3200	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
		3201
		3202	=item EV_USE_NANOSLEEP
		3203
		3204	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3205	and will use it for delays. Otherwise it will use C<select ()>.
		3206
		3207	=item EV_USE_EVENTFD
		3208
		3209	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3210	available and will probe for kernel support at runtime. This will improve
		3211	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3212	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3213	2.7 or newer, otherwise disabled.
2095		3214
2096	=item EV_USE_SELECT	3215	=item EV_USE_SELECT
2097		3216
2098	If undefined or defined to be C<1>, libev will compile in support for the	3217	If undefined or defined to be C<1>, libev will compile in support for the
2099	C<select>(2) backend. No attempt at autodetection will be done: if no	3218	C<select>(2) backend. No attempt at auto-detection will be done: if no
2100	other method takes over, select will be it. Otherwise the select backend	3219	other method takes over, select will be it. Otherwise the select backend
2101	will not be compiled in.	3220	will not be compiled in.
2102		3221
2103	=item EV_SELECT_USE_FD_SET	3222	=item EV_SELECT_USE_FD_SET
2104		3223
2105	If defined to C<1>, then the select backend will use the system C<fd_set>	3224	If defined to C<1>, then the select backend will use the system C<fd_set>
2106	structure. This is useful if libev doesn't compile due to a missing	3225	structure. This is useful if libev doesn't compile due to a missing
2107	C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on	3226	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout on
2108	exotic systems. This usually limits the range of file descriptors to some	3227	exotic systems. This usually limits the range of file descriptors to some
2109	low limit such as 1024 or might have other limitations (winsocket only	3228	low limit such as 1024 or might have other limitations (winsocket only
2110	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	3229	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might
2111	influence the size of the C<fd_set> used.	3230	influence the size of the C<fd_set> used.
2112		3231
…		…
2118	be used is the winsock select). This means that it will call	3237	be used is the winsock select). This means that it will call
2119	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	3238	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2120	it is assumed that all these functions actually work on fds, even	3239	it is assumed that all these functions actually work on fds, even
2121	on win32. Should not be defined on non-win32 platforms.	3240	on win32. Should not be defined on non-win32 platforms.
2122		3241
		3242	=item EV_FD_TO_WIN32_HANDLE
		3243
		3244	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3245	file descriptors to socket handles. When not defining this symbol (the
		3246	default), then libev will call C<_get_osfhandle>, which is usually
		3247	correct. In some cases, programs use their own file descriptor management,
		3248	in which case they can provide this function to map fds to socket handles.
		3249
2123	=item EV_USE_POLL	3250	=item EV_USE_POLL
2124		3251
2125	If defined to be C<1>, libev will compile in support for the C<poll>(2)	3252	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2126	backend. Otherwise it will be enabled on non-win32 platforms. It	3253	backend. Otherwise it will be enabled on non-win32 platforms. It
2127	takes precedence over select.	3254	takes precedence over select.
2128		3255
2129	=item EV_USE_EPOLL	3256	=item EV_USE_EPOLL
2130		3257
2131	If defined to be C<1>, libev will compile in support for the Linux	3258	If defined to be C<1>, libev will compile in support for the Linux
2132	C<epoll>(7) backend. Its availability will be detected at runtime,	3259	C<epoll>(7) backend. Its availability will be detected at runtime,
2133	otherwise another method will be used as fallback. This is the	3260	otherwise another method will be used as fallback. This is the preferred
2134	preferred backend for GNU/Linux systems.	3261	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3262	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
2135		3263
2136	=item EV_USE_KQUEUE	3264	=item EV_USE_KQUEUE
2137		3265
2138	If defined to be C<1>, libev will compile in support for the BSD style	3266	If defined to be C<1>, libev will compile in support for the BSD style
2139	C<kqueue>(2) backend. Its actual availability will be detected at runtime,	3267	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
…		…
2152	otherwise another method will be used as fallback. This is the preferred	3280	otherwise another method will be used as fallback. This is the preferred
2153	backend for Solaris 10 systems.	3281	backend for Solaris 10 systems.
2154		3282
2155	=item EV_USE_DEVPOLL	3283	=item EV_USE_DEVPOLL
2156		3284
2157	reserved for future expansion, works like the USE symbols above.	3285	Reserved for future expansion, works like the USE symbols above.
2158		3286
2159	=item EV_USE_INOTIFY	3287	=item EV_USE_INOTIFY
2160		3288
2161	If defined to be C<1>, libev will compile in support for the Linux inotify	3289	If defined to be C<1>, libev will compile in support for the Linux inotify
2162	interface to speed up C<ev_stat> watchers. Its actual availability will	3290	interface to speed up C<ev_stat> watchers. Its actual availability will
2163	be detected at runtime.	3291	be detected at runtime. If undefined, it will be enabled if the headers
		3292	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3293
		3294	=item EV_ATOMIC_T
		3295
		3296	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3297	access is atomic with respect to other threads or signal contexts. No such
		3298	type is easily found in the C language, so you can provide your own type
		3299	that you know is safe for your purposes. It is used both for signal handler "locking"
		3300	as well as for signal and thread safety in C<ev_async> watchers.
		3301
		3302	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3303	(from F<signal.h>), which is usually good enough on most platforms.
2164		3304
2165	=item EV_H	3305	=item EV_H
2166		3306
2167	The name of the F<ev.h> header file used to include it. The default if	3307	The name of the F<ev.h> header file used to include it. The default if
2168	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	3308	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2169	can be used to virtually rename the F<ev.h> header file in case of conflicts.	3309	used to virtually rename the F<ev.h> header file in case of conflicts.
2170		3310
2171	=item EV_CONFIG_H	3311	=item EV_CONFIG_H
2172		3312
2173	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	3313	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2174	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	3314	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2175	C<EV_H>, above.	3315	C<EV_H>, above.
2176		3316
2177	=item EV_EVENT_H	3317	=item EV_EVENT_H
2178		3318
2179	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	3319	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2180	of how the F<event.h> header can be found.	3320	of how the F<event.h> header can be found, the default is C<"event.h">.
2181		3321
2182	=item EV_PROTOTYPES	3322	=item EV_PROTOTYPES
2183		3323
2184	If defined to be C<0>, then F<ev.h> will not define any function	3324	If defined to be C<0>, then F<ev.h> will not define any function
2185	prototypes, but still define all the structs and other symbols. This is	3325	prototypes, but still define all the structs and other symbols. This is
…		…
2206	When doing priority-based operations, libev usually has to linearly search	3346	When doing priority-based operations, libev usually has to linearly search
2207	all the priorities, so having many of them (hundreds) uses a lot of space	3347	all the priorities, so having many of them (hundreds) uses a lot of space
2208	and time, so using the defaults of five priorities (-2 .. +2) is usually	3348	and time, so using the defaults of five priorities (-2 .. +2) is usually
2209	fine.	3349	fine.
2210		3350
2211	If your embedding app does not need any priorities, defining these both to	3351	If your embedding application does not need any priorities, defining these
2212	C<0> will save some memory and cpu.	3352	both to C<0> will save some memory and CPU.
2213		3353
2214	=item EV_PERIODIC_ENABLE	3354	=item EV_PERIODIC_ENABLE
2215		3355
2216	If undefined or defined to be C<1>, then periodic timers are supported. If	3356	If undefined or defined to be C<1>, then periodic timers are supported. If
2217	defined to be C<0>, then they are not. Disabling them saves a few kB of	3357	defined to be C<0>, then they are not. Disabling them saves a few kB of
…		…
2224	code.	3364	code.
2225		3365
2226	=item EV_EMBED_ENABLE	3366	=item EV_EMBED_ENABLE
2227		3367
2228	If undefined or defined to be C<1>, then embed watchers are supported. If	3368	If undefined or defined to be C<1>, then embed watchers are supported. If
2229	defined to be C<0>, then they are not.	3369	defined to be C<0>, then they are not. Embed watchers rely on most other
		3370	watcher types, which therefore must not be disabled.
2230		3371
2231	=item EV_STAT_ENABLE	3372	=item EV_STAT_ENABLE
2232		3373
2233	If undefined or defined to be C<1>, then stat watchers are supported. If	3374	If undefined or defined to be C<1>, then stat watchers are supported. If
2234	defined to be C<0>, then they are not.	3375	defined to be C<0>, then they are not.
…		…
2236	=item EV_FORK_ENABLE	3377	=item EV_FORK_ENABLE
2237		3378
2238	If undefined or defined to be C<1>, then fork watchers are supported. If	3379	If undefined or defined to be C<1>, then fork watchers are supported. If
2239	defined to be C<0>, then they are not.	3380	defined to be C<0>, then they are not.
2240		3381
		3382	=item EV_ASYNC_ENABLE
		3383
		3384	If undefined or defined to be C<1>, then async watchers are supported. If
		3385	defined to be C<0>, then they are not.
		3386
2241	=item EV_MINIMAL	3387	=item EV_MINIMAL
2242		3388
2243	If you need to shave off some kilobytes of code at the expense of some	3389	If you need to shave off some kilobytes of code at the expense of some
2244	speed, define this symbol to C<1>. Currently only used for gcc to override	3390	speed, define this symbol to C<1>. Currently this is used to override some
2245	some inlining decisions, saves roughly 30% codesize of amd64.	3391	inlining decisions, saves roughly 30% code size on amd64. It also selects a
		3392	much smaller 2-heap for timer management over the default 4-heap.
2246		3393
2247	=item EV_PID_HASHSIZE	3394	=item EV_PID_HASHSIZE
2248		3395
2249	C<ev_child> watchers use a small hash table to distribute workload by	3396	C<ev_child> watchers use a small hash table to distribute workload by
2250	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	3397	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
2251	than enough. If you need to manage thousands of children you might want to	3398	than enough. If you need to manage thousands of children you might want to
2252	increase this value (I<must> be a power of two).	3399	increase this value (I<must> be a power of two).
2253		3400
2254	=item EV_INOTIFY_HASHSIZE	3401	=item EV_INOTIFY_HASHSIZE
2255		3402
2256	C<ev_staz> watchers use a small hash table to distribute workload by	3403	C<ev_stat> watchers use a small hash table to distribute workload by
2257	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	3404	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2258	usually more than enough. If you need to manage thousands of C<ev_stat>	3405	usually more than enough. If you need to manage thousands of C<ev_stat>
2259	watchers you might want to increase this value (I<must> be a power of	3406	watchers you might want to increase this value (I<must> be a power of
2260	two).	3407	two).
2261		3408
		3409	=item EV_USE_4HEAP
		3410
		3411	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3412	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3413	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3414	faster performance with many (thousands) of watchers.
		3415
		3416	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3417	(disabled).
		3418
		3419	=item EV_HEAP_CACHE_AT
		3420
		3421	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3422	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3423	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3424	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3425	but avoids random read accesses on heap changes. This improves performance
		3426	noticeably with many (hundreds) of watchers.
		3427
		3428	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3429	(disabled).
		3430
		3431	=item EV_VERIFY
		3432
		3433	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3434	be done: If set to C<0>, no internal verification code will be compiled
		3435	in. If set to C<1>, then verification code will be compiled in, but not
		3436	called. If set to C<2>, then the internal verification code will be
		3437	called once per loop, which can slow down libev. If set to C<3>, then the
		3438	verification code will be called very frequently, which will slow down
		3439	libev considerably.
		3440
		3441	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3442	C<0>.
		3443
2262	=item EV_COMMON	3444	=item EV_COMMON
2263		3445
2264	By default, all watchers have a C<void *data> member. By redefining	3446	By default, all watchers have a C<void *data> member. By redefining
2265	this macro to a something else you can include more and other types of	3447	this macro to a something else you can include more and other types of
2266	members. You have to define it each time you include one of the files,	3448	members. You have to define it each time you include one of the files,
2267	though, and it must be identical each time.	3449	though, and it must be identical each time.
2268		3450
2269	For example, the perl EV module uses something like this:	3451	For example, the perl EV module uses something like this:
2270		3452
2271	#define EV_COMMON \	3453	#define EV_COMMON \
2272	SV *self; /* contains this struct */ \	3454	SV *self; /* contains this struct */ \
2273	SV *cb_sv, *fh /* note no trailing ";" */	3455	SV *cb_sv, *fh /* note no trailing ";" */
2274		3456
2275	=item EV_CB_DECLARE (type)	3457	=item EV_CB_DECLARE (type)
2276		3458
2277	=item EV_CB_INVOKE (watcher, revents)	3459	=item EV_CB_INVOKE (watcher, revents)
2278		3460
2279	=item ev_set_cb (ev, cb)	3461	=item ev_set_cb (ev, cb)
2280		3462
2281	Can be used to change the callback member declaration in each watcher,	3463	Can be used to change the callback member declaration in each watcher,
2282	and the way callbacks are invoked and set. Must expand to a struct member	3464	and the way callbacks are invoked and set. Must expand to a struct member
2283	definition and a statement, respectively. See the F<ev.v> header file for	3465	definition and a statement, respectively. See the F<ev.h> header file for
2284	their default definitions. One possible use for overriding these is to	3466	their default definitions. One possible use for overriding these is to
2285	avoid the C<struct ev_loop *> as first argument in all cases, or to use	3467	avoid the C<struct ev_loop *> as first argument in all cases, or to use
2286	method calls instead of plain function calls in C++.	3468	method calls instead of plain function calls in C++.
		3469
		3470	=back
		3471
		3472	=head2 EXPORTED API SYMBOLS
		3473
		3474	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3475	exported symbols, you can use the provided F<Symbol.*> files which list
		3476	all public symbols, one per line:
		3477
		3478	Symbols.ev for libev proper
		3479	Symbols.event for the libevent emulation
		3480
		3481	This can also be used to rename all public symbols to avoid clashes with
		3482	multiple versions of libev linked together (which is obviously bad in
		3483	itself, but sometimes it is inconvenient to avoid this).
		3484
		3485	A sed command like this will create wrapper C<#define>'s that you need to
		3486	include before including F<ev.h>:
		3487
		3488	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3489
		3490	This would create a file F<wrap.h> which essentially looks like this:
		3491
		3492	#define ev_backend myprefix_ev_backend
		3493	#define ev_check_start myprefix_ev_check_start
		3494	#define ev_check_stop myprefix_ev_check_stop
		3495	...
2287		3496
2288	=head2 EXAMPLES	3497	=head2 EXAMPLES
2289		3498
2290	For a real-world example of a program the includes libev	3499	For a real-world example of a program the includes libev
2291	verbatim, you can have a look at the EV perl module	3500	verbatim, you can have a look at the EV perl module
…		…
2296	file.	3505	file.
2297		3506
2298	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	3507	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
2299	that everybody includes and which overrides some configure choices:	3508	that everybody includes and which overrides some configure choices:
2300		3509
2301	#define EV_MINIMAL 1	3510	#define EV_MINIMAL 1
2302	#define EV_USE_POLL 0	3511	#define EV_USE_POLL 0
2303	#define EV_MULTIPLICITY 0	3512	#define EV_MULTIPLICITY 0
2304	#define EV_PERIODIC_ENABLE 0	3513	#define EV_PERIODIC_ENABLE 0
2305	#define EV_STAT_ENABLE 0	3514	#define EV_STAT_ENABLE 0
2306	#define EV_FORK_ENABLE 0	3515	#define EV_FORK_ENABLE 0
2307	#define EV_CONFIG_H <config.h>	3516	#define EV_CONFIG_H <config.h>
2308	#define EV_MINPRI 0	3517	#define EV_MINPRI 0
2309	#define EV_MAXPRI 0	3518	#define EV_MAXPRI 0
2310		3519
2311	#include "ev++.h"	3520	#include "ev++.h"
2312		3521
2313	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	3522	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
2314		3523
2315	#include "ev_cpp.h"	3524	#include "ev_cpp.h"
2316	#include "ev.c"	3525	#include "ev.c"
2317		3526
		3527	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
2318		3528
		3529	=head2 THREADS AND COROUTINES
		3530
		3531	=head3 THREADS
		3532
		3533	All libev functions are reentrant and thread-safe unless explicitly
		3534	documented otherwise, but libev implements no locking itself. This means
		3535	that you can use as many loops as you want in parallel, as long as there
		3536	are no concurrent calls into any libev function with the same loop
		3537	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		3538	of course): libev guarantees that different event loops share no data
		3539	structures that need any locking.
		3540
		3541	Or to put it differently: calls with different loop parameters can be done
		3542	concurrently from multiple threads, calls with the same loop parameter
		3543	must be done serially (but can be done from different threads, as long as
		3544	only one thread ever is inside a call at any point in time, e.g. by using
		3545	a mutex per loop).
		3546
		3547	Specifically to support threads (and signal handlers), libev implements
		3548	so-called C<ev_async> watchers, which allow some limited form of
		3549	concurrency on the same event loop, namely waking it up "from the
		3550	outside".
		3551
		3552	If you want to know which design (one loop, locking, or multiple loops
		3553	without or something else still) is best for your problem, then I cannot
		3554	help you, but here is some generic advice:
		3555
		3556	=over 4
		3557
		3558	=item * most applications have a main thread: use the default libev loop
		3559	in that thread, or create a separate thread running only the default loop.
		3560
		3561	This helps integrating other libraries or software modules that use libev
		3562	themselves and don't care/know about threading.
		3563
		3564	=item * one loop per thread is usually a good model.
		3565
		3566	Doing this is almost never wrong, sometimes a better-performance model
		3567	exists, but it is always a good start.
		3568
		3569	=item * other models exist, such as the leader/follower pattern, where one
		3570	loop is handed through multiple threads in a kind of round-robin fashion.
		3571
		3572	Choosing a model is hard - look around, learn, know that usually you can do
		3573	better than you currently do :-)
		3574
		3575	=item * often you need to talk to some other thread which blocks in the
		3576	event loop.
		3577
		3578	C<ev_async> watchers can be used to wake them up from other threads safely
		3579	(or from signal contexts...).
		3580
		3581	An example use would be to communicate signals or other events that only
		3582	work in the default loop by registering the signal watcher with the
		3583	default loop and triggering an C<ev_async> watcher from the default loop
		3584	watcher callback into the event loop interested in the signal.
		3585
		3586	=back
		3587
		3588	=head3 COROUTINES
		3589
		3590	Libev is very accommodating to coroutines ("cooperative threads"):
		3591	libev fully supports nesting calls to its functions from different
		3592	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		3593	different coroutines, and switch freely between both coroutines running the
		3594	loop, as long as you don't confuse yourself). The only exception is that
		3595	you must not do this from C<ev_periodic> reschedule callbacks.
		3596
		3597	Care has been taken to ensure that libev does not keep local state inside
		3598	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		3599	they do not call any callbacks.
		3600
		3601	=head2 COMPILER WARNINGS
		3602
		3603	Depending on your compiler and compiler settings, you might get no or a
		3604	lot of warnings when compiling libev code. Some people are apparently
		3605	scared by this.
		3606
		3607	However, these are unavoidable for many reasons. For one, each compiler
		3608	has different warnings, and each user has different tastes regarding
		3609	warning options. "Warn-free" code therefore cannot be a goal except when
		3610	targeting a specific compiler and compiler-version.
		3611
		3612	Another reason is that some compiler warnings require elaborate
		3613	workarounds, or other changes to the code that make it less clear and less
		3614	maintainable.
		3615
		3616	And of course, some compiler warnings are just plain stupid, or simply
		3617	wrong (because they don't actually warn about the condition their message
		3618	seems to warn about). For example, certain older gcc versions had some
		3619	warnings that resulted an extreme number of false positives. These have
		3620	been fixed, but some people still insist on making code warn-free with
		3621	such buggy versions.
		3622
		3623	While libev is written to generate as few warnings as possible,
		3624	"warn-free" code is not a goal, and it is recommended not to build libev
		3625	with any compiler warnings enabled unless you are prepared to cope with
		3626	them (e.g. by ignoring them). Remember that warnings are just that:
		3627	warnings, not errors, or proof of bugs.
		3628
		3629
		3630	=head2 VALGRIND
		3631
		3632	Valgrind has a special section here because it is a popular tool that is
		3633	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		3634
		3635	If you think you found a bug (memory leak, uninitialised data access etc.)
		3636	in libev, then check twice: If valgrind reports something like:
		3637
		3638	==2274== definitely lost: 0 bytes in 0 blocks.
		3639	==2274== possibly lost: 0 bytes in 0 blocks.
		3640	==2274== still reachable: 256 bytes in 1 blocks.
		3641
		3642	Then there is no memory leak, just as memory accounted to global variables
		3643	is not a memleak - the memory is still being referenced, and didn't leak.
		3644
		3645	Similarly, under some circumstances, valgrind might report kernel bugs
		3646	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3647	although an acceptable workaround has been found here), or it might be
		3648	confused.
		3649
		3650	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		3651	make it into some kind of religion.
		3652
		3653	If you are unsure about something, feel free to contact the mailing list
		3654	with the full valgrind report and an explanation on why you think this
		3655	is a bug in libev (best check the archives, too :). However, don't be
		3656	annoyed when you get a brisk "this is no bug" answer and take the chance
		3657	of learning how to interpret valgrind properly.
		3658
		3659	If you need, for some reason, empty reports from valgrind for your project
		3660	I suggest using suppression lists.
		3661
		3662
		3663	=head1 PORTABILITY NOTES
		3664
		3665	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		3666
		3667	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		3668	requires, and its I/O model is fundamentally incompatible with the POSIX
		3669	model. Libev still offers limited functionality on this platform in
		3670	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		3671	descriptors. This only applies when using Win32 natively, not when using
		3672	e.g. cygwin.
		3673
		3674	Lifting these limitations would basically require the full
		3675	re-implementation of the I/O system. If you are into these kinds of
		3676	things, then note that glib does exactly that for you in a very portable
		3677	way (note also that glib is the slowest event library known to man).
		3678
		3679	There is no supported compilation method available on windows except
		3680	embedding it into other applications.
		3681
		3682	Not a libev limitation but worth mentioning: windows apparently doesn't
		3683	accept large writes: instead of resulting in a partial write, windows will
		3684	either accept everything or return C<ENOBUFS> if the buffer is too large,
		3685	so make sure you only write small amounts into your sockets (less than a
		3686	megabyte seems safe, but this apparently depends on the amount of memory
		3687	available).
		3688
		3689	Due to the many, low, and arbitrary limits on the win32 platform and
		3690	the abysmal performance of winsockets, using a large number of sockets
		3691	is not recommended (and not reasonable). If your program needs to use
		3692	more than a hundred or so sockets, then likely it needs to use a totally
		3693	different implementation for windows, as libev offers the POSIX readiness
		3694	notification model, which cannot be implemented efficiently on windows
		3695	(Microsoft monopoly games).
		3696
		3697	A typical way to use libev under windows is to embed it (see the embedding
		3698	section for details) and use the following F<evwrap.h> header file instead
		3699	of F<ev.h>:
		3700
		3701	#define EV_STANDALONE /* keeps ev from requiring config.h */
		3702	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		3703
		3704	#include "ev.h"
		3705
		3706	And compile the following F<evwrap.c> file into your project (make sure
		3707	you do I<not> compile the F<ev.c> or any other embedded source files!):
		3708
		3709	#include "evwrap.h"
		3710	#include "ev.c"
		3711
		3712	=over 4
		3713
		3714	=item The winsocket select function
		3715
		3716	The winsocket C<select> function doesn't follow POSIX in that it
		3717	requires socket I<handles> and not socket I<file descriptors> (it is
		3718	also extremely buggy). This makes select very inefficient, and also
		3719	requires a mapping from file descriptors to socket handles (the Microsoft
		3720	C runtime provides the function C<_open_osfhandle> for this). See the
		3721	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		3722	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		3723
		3724	The configuration for a "naked" win32 using the Microsoft runtime
		3725	libraries and raw winsocket select is:
		3726
		3727	#define EV_USE_SELECT 1
		3728	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		3729
		3730	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3731	complexity in the O(n²) range when using win32.
		3732
		3733	=item Limited number of file descriptors
		3734
		3735	Windows has numerous arbitrary (and low) limits on things.
		3736
		3737	Early versions of winsocket's select only supported waiting for a maximum
		3738	of C<64> handles (probably owning to the fact that all windows kernels
		3739	can only wait for C<64> things at the same time internally; Microsoft
		3740	recommends spawning a chain of threads and wait for 63 handles and the
		3741	previous thread in each. Great).
		3742
		3743	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3744	to some high number (e.g. C<2048>) before compiling the winsocket select
		3745	call (which might be in libev or elsewhere, for example, perl does its own
		3746	select emulation on windows).
		3747
		3748	Another limit is the number of file descriptors in the Microsoft runtime
		3749	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3750	or something like this inside Microsoft). You can increase this by calling
		3751	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3752	arbitrary limit), but is broken in many versions of the Microsoft runtime
		3753	libraries.
		3754
		3755	This might get you to about C<512> or C<2048> sockets (depending on
		3756	windows version and/or the phase of the moon). To get more, you need to
		3757	wrap all I/O functions and provide your own fd management, but the cost of
		3758	calling select (O(n²)) will likely make this unworkable.
		3759
		3760	=back
		3761
		3762	=head2 PORTABILITY REQUIREMENTS
		3763
		3764	In addition to a working ISO-C implementation and of course the
		3765	backend-specific APIs, libev relies on a few additional extensions:
		3766
		3767	=over 4
		3768
		3769	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		3770	calling conventions regardless of C<ev_watcher_type *>.
		3771
		3772	Libev assumes not only that all watcher pointers have the same internal
		3773	structure (guaranteed by POSIX but not by ISO C for example), but it also
		3774	assumes that the same (machine) code can be used to call any watcher
		3775	callback: The watcher callbacks have different type signatures, but libev
		3776	calls them using an C<ev_watcher *> internally.
		3777
		3778	=item C<sig_atomic_t volatile> must be thread-atomic as well
		3779
		3780	The type C<sig_atomic_t volatile> (or whatever is defined as
		3781	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		3782	threads. This is not part of the specification for C<sig_atomic_t>, but is
		3783	believed to be sufficiently portable.
		3784
		3785	=item C<sigprocmask> must work in a threaded environment
		3786
		3787	Libev uses C<sigprocmask> to temporarily block signals. This is not
		3788	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		3789	pthread implementations will either allow C<sigprocmask> in the "main
		3790	thread" or will block signals process-wide, both behaviours would
		3791	be compatible with libev. Interaction between C<sigprocmask> and
		3792	C<pthread_sigmask> could complicate things, however.
		3793
		3794	The most portable way to handle signals is to block signals in all threads
		3795	except the initial one, and run the default loop in the initial thread as
		3796	well.
		3797
		3798	=item C<long> must be large enough for common memory allocation sizes
		3799
		3800	To improve portability and simplify its API, libev uses C<long> internally
		3801	instead of C<size_t> when allocating its data structures. On non-POSIX
		3802	systems (Microsoft...) this might be unexpectedly low, but is still at
		3803	least 31 bits everywhere, which is enough for hundreds of millions of
		3804	watchers.
		3805
		3806	=item C<double> must hold a time value in seconds with enough accuracy
		3807
		3808	The type C<double> is used to represent timestamps. It is required to
		3809	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		3810	enough for at least into the year 4000. This requirement is fulfilled by
		3811	implementations implementing IEEE 754 (basically all existing ones).
		3812
		3813	=back
		3814
		3815	If you know of other additional requirements drop me a note.
		3816
		3817
2319	=head1 COMPLEXITIES	3818	=head1 ALGORITHMIC COMPLEXITIES
2320		3819
2321	In this section the complexities of (many of) the algorithms used inside	3820	In this section the complexities of (many of) the algorithms used inside
2322	libev will be explained. For complexity discussions about backends see the	3821	libev will be documented. For complexity discussions about backends see
2323	documentation for C<ev_default_init>.	3822	the documentation for C<ev_default_init>.
2324		3823
2325	All of the following are about amortised time: If an array needs to be	3824	All of the following are about amortised time: If an array needs to be
2326	extended, libev needs to realloc and move the whole array, but this	3825	extended, libev needs to realloc and move the whole array, but this
2327	happens asymptotically never with higher number of elements, so O(1) might	3826	happens asymptotically rarer with higher number of elements, so O(1) might
2328	mean it might do a lengthy realloc operation in rare cases, but on average	3827	mean that libev does a lengthy realloc operation in rare cases, but on
2329	it is much faster and asymptotically approaches constant time.	3828	average it is much faster and asymptotically approaches constant time.
2330		3829
2331	=over 4	3830	=over 4
2332		3831
2333	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	3832	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2334		3833
2335	This means that, when you have a watcher that triggers in one hour and	3834	This means that, when you have a watcher that triggers in one hour and
2336	there are 100 watchers that would trigger before that then inserting will	3835	there are 100 watchers that would trigger before that, then inserting will
2337	have to skip those 100 watchers.	3836	have to skip roughly seven (C<ld 100>) of these watchers.
2338		3837
2339	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	3838	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2340		3839
2341	That means that for changing a timer costs less than removing/adding them	3840	That means that changing a timer costs less than removing/adding them,
2342	as only the relative motion in the event queue has to be paid for.	3841	as only the relative motion in the event queue has to be paid for.
2343		3842
2344	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	3843	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2345		3844
2346	These just add the watcher into an array or at the head of a list.	3845	These just add the watcher into an array or at the head of a list.
		3846
2347	=item Stopping check/prepare/idle watchers: O(1)	3847	=item Stopping check/prepare/idle/fork/async watchers: O(1)
2348		3848
2349	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	3849	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2350		3850
2351	These watchers are stored in lists then need to be walked to find the	3851	These watchers are stored in lists, so they need to be walked to find the
2352	correct watcher to remove. The lists are usually short (you don't usually	3852	correct watcher to remove. The lists are usually short (you don't usually
2353	have many watchers waiting for the same fd or signal).	3853	have many watchers waiting for the same fd or signal: one is typical, two
		3854	is rare).
2354		3855
2355	=item Finding the next timer per loop iteration: O(1)	3856	=item Finding the next timer in each loop iteration: O(1)
		3857
		3858	By virtue of using a binary or 4-heap, the next timer is always found at a
		3859	fixed position in the storage array.
2356		3860
2357	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	3861	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2358		3862
2359	A change means an I/O watcher gets started or stopped, which requires	3863	A change means an I/O watcher gets started or stopped, which requires
2360	libev to recalculate its status (and possibly tell the kernel).	3864	libev to recalculate its status (and possibly tell the kernel, depending
		3865	on backend and whether C<ev_io_set> was used).
2361		3866
2362	=item Activating one watcher: O(1)	3867	=item Activating one watcher (putting it into the pending state): O(1)
2363		3868
2364	=item Priority handling: O(number_of_priorities)	3869	=item Priority handling: O(number_of_priorities)
2365		3870
2366	Priorities are implemented by allocating some space for each	3871	Priorities are implemented by allocating some space for each
2367	priority. When doing priority-based operations, libev usually has to	3872	priority. When doing priority-based operations, libev usually has to
2368	linearly search all the priorities.	3873	linearly search all the priorities, but starting/stopping and activating
		3874	watchers becomes O(1) with respect to priority handling.
		3875
		3876	=item Sending an ev_async: O(1)
		3877
		3878	=item Processing ev_async_send: O(number_of_async_watchers)
		3879
		3880	=item Processing signals: O(max_signal_number)
		3881
		3882	Sending involves a system call I<iff> there were no other C<ev_async_send>
		3883	calls in the current loop iteration. Checking for async and signal events
		3884	involves iterating over all running async watchers or all signal numbers.
2369		3885
2370	=back	3886	=back
2371		3887
2372		3888
2373	=head1 AUTHOR	3889	=head1 AUTHOR
2374		3890
2375	Marc Lehmann <libev@schmorp.de>.	3891	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.
2376		3892

Diff Legend

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