ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.55 by root, Tue Nov 27 20:38:07 2007 UTC vs.
Revision 1.208 by root, Wed Oct 29 10:24:23 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
		67	The newest version of this document is also available as an html-formatted
		68	web page you might find easier to navigate when reading it for the first
		69	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
		70
53	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
54	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
55	these event sources and provide your program with events.	73	these event sources and provide your program with events.
56		74
57	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
58	(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
59	communicate events via a callback mechanism.	77	communicate events via a callback mechanism.
…		…
61	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
62	watchers>, which are relatively small C structures you initialise with the	80	watchers>, which are relatively small C structures you initialise with the
63	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
64	watcher.	82	watcher.
65		83
66	=head1 FEATURES	84	=head2 FEATURES
67		85
68	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
69	bsd-specific C<kqueue> and the solaris-specific event port mechanisms	87	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
70	for file descriptor events (C<ev_io>), relative timers (C<ev_timer>),	88	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
		89	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
71	absolute timers with customised rescheduling (C<ev_periodic>), synchronous	90	with customised rescheduling (C<ev_periodic>), synchronous signals
72	signals (C<ev_signal>), process status change events (C<ev_child>), and	91	(C<ev_signal>), process status change events (C<ev_child>), and event
73	event watchers dealing with the event loop mechanism itself (C<ev_idle>,	92	watchers dealing with the event loop mechanism itself (C<ev_idle>,
74	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as	93	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as
75	file watchers (C<ev_stat>) and even limited support for fork events	94	file watchers (C<ev_stat>) and even limited support for fork events
76	(C<ev_fork>).	95	(C<ev_fork>).
77		96
78	It also is quite fast (see this	97	It also is quite fast (see this
79	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
80	for example).	99	for example).
81		100
82	=head1 CONVENTIONS	101	=head2 CONVENTIONS
83		102
84	Libev is very configurable. In this manual the default configuration will	103	Libev is very configurable. In this manual the default (and most common)
85	be described, which supports multiple event loops. For more info about	104	configuration will be described, which supports multiple event loops. For
86	various configuration options please have a look at B<EMBED> section in	105	more info about various configuration options please have a look at
87	this manual. If libev was configured without support for multiple event	106	B<EMBED> section in this manual. If libev was configured without support
88	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
89	(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.
90		110
91	=head1 TIME REPRESENTATION	111	=head2 TIME REPRESENTATION
92		112
93	Libev represents time as a single floating point number, representing the	113	Libev represents time as a single floating point number, representing the
94	(fractional) number of seconds since the (POSIX) epoch (somewhere near	114	(fractional) number of seconds since the (POSIX) epoch (somewhere near
95	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
96	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
97	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
98	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
99		142
100	=head1 GLOBAL FUNCTIONS	143	=head1 GLOBAL FUNCTIONS
101		144
102	These functions can be called anytime, even before initialising the	145	These functions can be called anytime, even before initialising the
103	library in any way.	146	library in any way.
…		…
108		151
109	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
110	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
111	you actually want to know.	154	you actually want to know.
112		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
113	=item int ev_version_major ()	162	=item int ev_version_major ()
114		163
115	=item int ev_version_minor ()	164	=item int ev_version_minor ()
116		165
117	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
118	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
119	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
120	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
121	version of the library your program was compiled against.	170	version of the library your program was compiled against.
122		171
		172	These version numbers refer to the ABI version of the library, not the
		173	release version.
		174
123	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,
124	as this indicates an incompatible change. Minor versions are usually	176	as this indicates an incompatible change. Minor versions are usually
125	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
126	not a problem.	178	not a problem.
127		179
128	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
129	version.	181	version.
130		182
131	assert (("libev version mismatch",	183	assert (("libev version mismatch",
132	ev_version_major () == EV_VERSION_MAJOR	184	ev_version_major () == EV_VERSION_MAJOR
133	&& ev_version_minor () >= EV_VERSION_MINOR));	185	&& ev_version_minor () >= EV_VERSION_MINOR));
134		186
135	=item unsigned int ev_supported_backends ()	187	=item unsigned int ev_supported_backends ()
136		188
137	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_*>
138	value) compiled into this binary of libev (independent of their	190	value) compiled into this binary of libev (independent of their
…		…
140	a description of the set values.	192	a description of the set values.
141		193
142	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
143	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
144		196
145	assert (("sorry, no epoll, no sex",	197	assert (("sorry, no epoll, no sex",
146	ev_supported_backends () & EVBACKEND_EPOLL));	198	ev_supported_backends () & EVBACKEND_EPOLL));
147		199
148	=item unsigned int ev_recommended_backends ()	200	=item unsigned int ev_recommended_backends ()
149		201
150	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
151	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
152	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
153	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
154	(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
155	libev will probe for if you specify no backends explicitly.	207	libev will probe for if you specify no backends explicitly.
156		208
157	=item unsigned int ev_embeddable_backends ()	209	=item unsigned int ev_embeddable_backends ()
158		210
…		…
162	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	214	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
163	recommended ones.	215	recommended ones.
164		216
165	See the description of C<ev_embed> watchers for more info.	217	See the description of C<ev_embed> watchers for more info.
166		218
167	=item ev_set_allocator (void *(*cb)(void *ptr, size_t size))	219	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
168		220
169	Sets the allocation function to use (the prototype and semantics are	221	Sets the allocation function to use (the prototype is similar - the
170	identical to the realloc C function). It is used to allocate and free	222	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
171	memory (no surprises here). If it returns zero when memory needs to be	223	used to allocate and free memory (no surprises here). If it returns zero
172	allocated, the library might abort or take some potentially destructive	224	when memory needs to be allocated (C<size != 0>), the library might abort
173	action. The default is your system realloc function.	225	or take some potentially destructive action.
		226
		227	Since some systems (at least OpenBSD and Darwin) fail to implement
		228	correct C<realloc> semantics, libev will use a wrapper around the system
		229	C<realloc> and C<free> functions by default.
174		230
175	You could override this function in high-availability programs to, say,	231	You could override this function in high-availability programs to, say,
176	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,
177	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.
178		234
179	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
180	retries).	236	retries (example requires a standards-compliant C<realloc>).
181		237
182	static void *	238	static void *
183	persistent_realloc (void *ptr, size_t size)	239	persistent_realloc (void *ptr, size_t size)
184	{	240	{
185	for (;;)	241	for (;;)
…		…
194	}	250	}
195		251
196	...	252	...
197	ev_set_allocator (persistent_realloc);	253	ev_set_allocator (persistent_realloc);
198		254
199	=item ev_set_syserr_cb (void (*cb)(const char *msg));	255	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
200		256
201	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
202	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
203	indicating the system call or subsystem causing the problem. If this	259	indicating the system call or subsystem causing the problem. If this
204	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
205	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
206	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
207	(such as abort).	263	(such as abort).
208		264
209	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.
…		…
220		276
221	=back	277	=back
222		278
223	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	279	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
224		280
225	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>
226	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>
227	events, and dynamically created loops which do not.	283	I<function>).
228		284
229	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
230	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
231	create, you also create another event loop. Libev itself does no locking	287	not.
232	whatsoever, so if you mix calls to the same event loop in different
233	threads, make sure you lock (this is usually a bad idea, though, even if
234	done correctly, because it's hideous and inefficient).
235		288
236	=over 4	289	=over 4
237		290
238	=item struct ev_loop *ev_default_loop (unsigned int flags)	291	=item struct ev_loop *ev_default_loop (unsigned int flags)
239		292
…		…
243	flags. If that is troubling you, check C<ev_backend ()> afterwards).	296	flags. If that is troubling you, check C<ev_backend ()> afterwards).
244		297
245	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
246	function.	299	function.
247		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
248	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
249	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>).
250		314
251	The following flags are supported:	315	The following flags are supported:
252		316
…		…
257	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
258	thing, believe me).	322	thing, believe me).
259		323
260	=item C<EVFLAG_NOENV>	324	=item C<EVFLAG_NOENV>
261		325
262	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
263	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
264	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	328	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
265	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
266	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
267	around bugs.	331	around bugs.
268		332
		333	=item C<EVFLAG_FORKCHECK>
		334
		335	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after
		336	a fork, you can also make libev check for a fork in each iteration by
		337	enabling this flag.
		338
		339	This works by calling C<getpid ()> on every iteration of the loop,
		340	and thus this might slow down your event loop if you do a lot of loop
		341	iterations and little real work, but is usually not noticeable (on my
		342	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
		343	without a system call and thus I<very> fast, but my GNU/Linux system also has
		344	C<pthread_atfork> which is even faster).
		345
		346	The big advantage of this flag is that you can forget about fork (and
		347	forget about forgetting to tell libev about forking) when you use this
		348	flag.
		349
		350	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
		351	environment variable.
		352
269	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	353	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
270		354
271	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
272	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,
273	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
274	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
275	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).
276		371
277	=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)
278		373
279	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
280	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
281	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
282	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>.
283		383
284	=item C<EVBACKEND_EPOLL> (value 4, Linux)	384	=item C<EVBACKEND_EPOLL> (value 4, Linux)
285		385
286	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,
287	but it scales phenomenally better. While poll and select usually scale like	387	but it scales phenomenally better. While poll and select usually scale
288	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),
289	either O(1) or O(active_fds).	389	epoll scales either O(1) or O(active_fds).
290		390
		391	The epoll syscalls are the most misdesigned of the more advanced event
		392	mechanisms: problems include silently dropping fds, requiring a system
		393	call per change per fd (and unnecessary guessing of parameters), problems
		394	with dup and so on. The biggest issue is fork races, however - if a
		395	program forks then I<both> parent and child process have to recreate the
		396	epoll set, which can take considerable time (one syscall per fd) and is of
		397	course hard to detect.
		398
		399	Epoll is also notoriously buggy - embedding epoll fds should work, but
		400	of course doesn't, and epoll just loves to report events for totally
		401	I<different> file descriptors (even already closed ones, so one cannot
		402	even remove them from the set) than registered in the set (especially
		403	on SMP systems). Libev tries to counter these spurious notifications by
		404	employing an additional generation counter and comparing that against the
		405	events to filter out spurious ones.
		406
291	While stopping and starting an I/O watcher in the same iteration will	407	While stopping, setting and starting an I/O watcher in the same iteration
292	result in some caching, there is still a syscall per such incident	408	will result in some caching, there is still a system call per such incident
293	(because the fd could point to a different file description now), so its	409	(because the fd could point to a different file description now), so its
294	best to avoid that. Also, dup()ed file descriptors might not work very	410	best to avoid that. Also, C<dup ()>'ed file descriptors might not work
295	well if you register events for both fds.	411	very well if you register events for both fds.
296		412
297	Please note that epoll sometimes generates spurious notifications, so you	413	Best performance from this backend is achieved by not unregistering all
298	need to use non-blocking I/O or other means to avoid blocking when no data	414	watchers for a file descriptor until it has been closed, if possible,
299	(or space) is available.	415	i.e. keep at least one watcher active per fd at all times. Stopping and
		416	starting a watcher (without re-setting it) also usually doesn't cause
		417	extra overhead. A fork can both result in spurious notifications as well
		418	as in libev having to destroy and recreate the epoll object, which can
		419	take considerable time and thus should be avoided.
		420
		421	While nominally embeddable in other event loops, this feature is broken in
		422	all kernel versions tested so far.
		423
		424	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		425	C<EVBACKEND_POLL>.
300		426
301	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	427	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
302		428
303	Kqueue deserves special mention, as at the time of this writing, it	429	Kqueue deserves special mention, as at the time of this writing, it was
304	was broken on all BSDs except NetBSD (usually it doesn't work with	430	broken on all BSDs except NetBSD (usually it doesn't work reliably with
305	anything but sockets and pipes, except on Darwin, where of course its	431	anything but sockets and pipes, except on Darwin, where of course it's
306	completely useless). For this reason its not being "autodetected"	432	completely useless). For this reason it's not being "auto-detected" unless
307	unless you explicitly specify it explicitly in the flags (i.e. using	433	you explicitly specify it in the flags (i.e. using C<EVBACKEND_KQUEUE>) or
308	C<EVBACKEND_KQUEUE>).	434	libev was compiled on a known-to-be-good (-enough) system like NetBSD.
		435
		436	You still can embed kqueue into a normal poll or select backend and use it
		437	only for sockets (after having made sure that sockets work with kqueue on
		438	the target platform). See C<ev_embed> watchers for more info.
309		439
310	It scales in the same way as the epoll backend, but the interface to the	440	It scales in the same way as the epoll backend, but the interface to the
311	kernel is more efficient (which says nothing about its actual speed, of	441	kernel is more efficient (which says nothing about its actual speed, of
312	course). While starting and stopping an I/O watcher does not cause an	442	course). While stopping, setting and starting an I/O watcher does never
313	extra syscall as with epoll, it still adds up to four event changes per	443	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
314	incident, so its best to avoid that.	444	two event changes per incident. Support for C<fork ()> is very bad (but
		445	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect
		446	cases
		447
		448	This backend usually performs well under most conditions.
		449
		450	While nominally embeddable in other event loops, this doesn't work
		451	everywhere, so you might need to test for this. And since it is broken
		452	almost everywhere, you should only use it when you have a lot of sockets
		453	(for which it usually works), by embedding it into another event loop
		454	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and, did I mention it,
		455	using it only for sockets.
		456
		457	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		458	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		459	C<NOTE_EOF>.
315		460
316	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	461	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
317		462
318	This is not implemented yet (and might never be).	463	This is not implemented yet (and might never be, unless you send me an
		464	implementation). According to reports, C</dev/poll> only supports sockets
		465	and is not embeddable, which would limit the usefulness of this backend
		466	immensely.
319		467
320	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	468	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
321		469
322	This uses the Solaris 10 port mechanism. As with everything on Solaris,	470	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
323	it's really slow, but it still scales very well (O(active_fds)).	471	it's really slow, but it still scales very well (O(active_fds)).
324		472
325	Please note that solaris ports can result in a lot of spurious	473	Please note that Solaris event ports can deliver a lot of spurious
326	notifications, so you need to use non-blocking I/O or other means to avoid	474	notifications, so you need to use non-blocking I/O or other means to avoid
327	blocking when no data (or space) is available.	475	blocking when no data (or space) is available.
		476
		477	While this backend scales well, it requires one system call per active
		478	file descriptor per loop iteration. For small and medium numbers of file
		479	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		480	might perform better.
		481
		482	On the positive side, with the exception of the spurious readiness
		483	notifications, this backend actually performed fully to specification
		484	in all tests and is fully embeddable, which is a rare feat among the
		485	OS-specific backends (I vastly prefer correctness over speed hacks).
		486
		487	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		488	C<EVBACKEND_POLL>.
328		489
329	=item C<EVBACKEND_ALL>	490	=item C<EVBACKEND_ALL>
330		491
331	Try all backends (even potentially broken ones that wouldn't be tried	492	Try all backends (even potentially broken ones that wouldn't be tried
332	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	493	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
333	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	494	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
334		495
		496	It is definitely not recommended to use this flag.
		497
335	=back	498	=back
336		499
337	If one or more of these are ored into the flags value, then only these	500	If one or more of these are or'ed into the flags value, then only these
338	backends will be tried (in the reverse order as given here). If none are	501	backends will be tried (in the reverse order as listed here). If none are
339	specified, most compiled-in backend will be tried, usually in reverse	502	specified, all backends in C<ev_recommended_backends ()> will be tried.
340	order of their flag values :)
341		503
342	The most typical usage is like this:	504	Example: This is the most typical usage.
343		505
344	if (!ev_default_loop (0))	506	if (!ev_default_loop (0))
345	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");	507	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
346		508
347	Restrict libev to the select and poll backends, and do not allow	509	Example: Restrict libev to the select and poll backends, and do not allow
348	environment settings to be taken into account:	510	environment settings to be taken into account:
349		511
350	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);	512	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
351		513
352	Use whatever libev has to offer, but make sure that kqueue is used if	514	Example: Use whatever libev has to offer, but make sure that kqueue is
353	available (warning, breaks stuff, best use only with your own private	515	used if available (warning, breaks stuff, best use only with your own
354	event loop and only if you know the OS supports your types of fds):	516	private event loop and only if you know the OS supports your types of
		517	fds):
355		518
356	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);	519	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
357		520
358	=item struct ev_loop *ev_loop_new (unsigned int flags)	521	=item struct ev_loop *ev_loop_new (unsigned int flags)
359		522
360	Similar to C<ev_default_loop>, but always creates a new event loop that is	523	Similar to C<ev_default_loop>, but always creates a new event loop that is
361	always distinct from the default loop. Unlike the default loop, it cannot	524	always distinct from the default loop. Unlike the default loop, it cannot
362	handle signal and child watchers, and attempts to do so will be greeted by	525	handle signal and child watchers, and attempts to do so will be greeted by
363	undefined behaviour (or a failed assertion if assertions are enabled).	526	undefined behaviour (or a failed assertion if assertions are enabled).
364		527
		528	Note that this function I<is> thread-safe, and the recommended way to use
		529	libev with threads is indeed to create one loop per thread, and using the
		530	default loop in the "main" or "initial" thread.
		531
365	Example: Try to create a event loop that uses epoll and nothing else.	532	Example: Try to create a event loop that uses epoll and nothing else.
366		533
367	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	534	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
368	if (!epoller)	535	if (!epoller)
369	fatal ("no epoll found here, maybe it hides under your chair");	536	fatal ("no epoll found here, maybe it hides under your chair");
370		537
371	=item ev_default_destroy ()	538	=item ev_default_destroy ()
372		539
373	Destroys the default loop again (frees all memory and kernel state	540	Destroys the default loop again (frees all memory and kernel state
374	etc.). None of the active event watchers will be stopped in the normal	541	etc.). None of the active event watchers will be stopped in the normal
375	sense, so e.g. C<ev_is_active> might still return true. It is your	542	sense, so e.g. C<ev_is_active> might still return true. It is your
376	responsibility to either stop all watchers cleanly yoursef I<before>	543	responsibility to either stop all watchers cleanly yourself I<before>
377	calling this function, or cope with the fact afterwards (which is usually	544	calling this function, or cope with the fact afterwards (which is usually
378	the easiest thing, youc na just ignore the watchers and/or C<free ()> them	545	the easiest thing, you can just ignore the watchers and/or C<free ()> them
379	for example).	546	for example).
		547
		548	Note that certain global state, such as signal state (and installed signal
		549	handlers), will not be freed by this function, and related watchers (such
		550	as signal and child watchers) would need to be stopped manually.
		551
		552	In general it is not advisable to call this function except in the
		553	rare occasion where you really need to free e.g. the signal handling
		554	pipe fds. If you need dynamically allocated loops it is better to use
		555	C<ev_loop_new> and C<ev_loop_destroy>).
380		556
381	=item ev_loop_destroy (loop)	557	=item ev_loop_destroy (loop)
382		558
383	Like C<ev_default_destroy>, but destroys an event loop created by an	559	Like C<ev_default_destroy>, but destroys an event loop created by an
384	earlier call to C<ev_loop_new>.	560	earlier call to C<ev_loop_new>.
385		561
386	=item ev_default_fork ()	562	=item ev_default_fork ()
387		563
		564	This function sets a flag that causes subsequent C<ev_loop> iterations
388	This function reinitialises the kernel state for backends that have	565	to reinitialise the kernel state for backends that have one. Despite the
389	one. Despite the name, you can call it anytime, but it makes most sense	566	name, you can call it anytime, but it makes most sense after forking, in
390	after forking, in either the parent or child process (or both, but that	567	the child process (or both child and parent, but that again makes little
391	again makes little sense).	568	sense). You I<must> call it in the child before using any of the libev
		569	functions, and it will only take effect at the next C<ev_loop> iteration.
392		570
393	You I<must> call this function in the child process after forking if and	571	On the other hand, you only need to call this function in the child
394	only if you want to use the event library in both processes. If you just	572	process if and only if you want to use the event library in the child. If
395	fork+exec, you don't have to call it.	573	you just fork+exec, you don't have to call it at all.
396		574
397	The function itself is quite fast and it's usually not a problem to call	575	The function itself is quite fast and it's usually not a problem to call
398	it just in case after a fork. To make this easy, the function will fit in	576	it just in case after a fork. To make this easy, the function will fit in
399	quite nicely into a call to C<pthread_atfork>:	577	quite nicely into a call to C<pthread_atfork>:
400		578
401	pthread_atfork (0, 0, ev_default_fork);	579	pthread_atfork (0, 0, ev_default_fork);
402		580
403	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use
404	without calling this function, so if you force one of those backends you
405	do not need to care.
406
407	=item ev_loop_fork (loop)	581	=item ev_loop_fork (loop)
408		582
409	Like C<ev_default_fork>, but acts on an event loop created by	583	Like C<ev_default_fork>, but acts on an event loop created by
410	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	584	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
411	after fork, and how you do this is entirely your own problem.	585	after fork that you want to re-use in the child, and how you do this is
		586	entirely your own problem.
		587
		588	=item int ev_is_default_loop (loop)
		589
		590	Returns true when the given loop is, in fact, the default loop, and false
		591	otherwise.
		592
		593	=item unsigned int ev_loop_count (loop)
		594
		595	Returns the count of loop iterations for the loop, which is identical to
		596	the number of times libev did poll for new events. It starts at C<0> and
		597	happily wraps around with enough iterations.
		598
		599	This value can sometimes be useful as a generation counter of sorts (it
		600	"ticks" the number of loop iterations), as it roughly corresponds with
		601	C<ev_prepare> and C<ev_check> calls.
412		602
413	=item unsigned int ev_backend (loop)	603	=item unsigned int ev_backend (loop)
414		604
415	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	605	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
416	use.	606	use.
…		…
419		609
420	Returns the current "event loop time", which is the time the event loop	610	Returns the current "event loop time", which is the time the event loop
421	received events and started processing them. This timestamp does not	611	received events and started processing them. This timestamp does not
422	change as long as callbacks are being processed, and this is also the base	612	change as long as callbacks are being processed, and this is also the base
423	time used for relative timers. You can treat it as the timestamp of the	613	time used for relative timers. You can treat it as the timestamp of the
424	event occuring (or more correctly, libev finding out about it).	614	event occurring (or more correctly, libev finding out about it).
		615
		616	=item ev_now_update (loop)
		617
		618	Establishes the current time by querying the kernel, updating the time
		619	returned by C<ev_now ()> in the progress. This is a costly operation and
		620	is usually done automatically within C<ev_loop ()>.
		621
		622	This function is rarely useful, but when some event callback runs for a
		623	very long time without entering the event loop, updating libev's idea of
		624	the current time is a good idea.
		625
		626	See also "The special problem of time updates" in the C<ev_timer> section.
425		627
426	=item ev_loop (loop, int flags)	628	=item ev_loop (loop, int flags)
427		629
428	Finally, this is it, the event handler. This function usually is called	630	Finally, this is it, the event handler. This function usually is called
429	after you initialised all your watchers and you want to start handling	631	after you initialised all your watchers and you want to start handling
…		…
432	If the flags argument is specified as C<0>, it will not return until	634	If the flags argument is specified as C<0>, it will not return until
433	either no event watchers are active anymore or C<ev_unloop> was called.	635	either no event watchers are active anymore or C<ev_unloop> was called.
434		636
435	Please note that an explicit C<ev_unloop> is usually better than	637	Please note that an explicit C<ev_unloop> is usually better than
436	relying on all watchers to be stopped when deciding when a program has	638	relying on all watchers to be stopped when deciding when a program has
437	finished (especially in interactive programs), but having a program that	639	finished (especially in interactive programs), but having a program
438	automatically loops as long as it has to and no longer by virtue of	640	that automatically loops as long as it has to and no longer by virtue
439	relying on its watchers stopping correctly is a thing of beauty.	641	of relying on its watchers stopping correctly, that is truly a thing of
		642	beauty.
440		643
441	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	644	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
442	those events and any outstanding ones, but will not block your process in	645	those events and any already outstanding ones, but will not block your
443	case there are no events and will return after one iteration of the loop.	646	process in case there are no events and will return after one iteration of
		647	the loop.
444		648
445	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	649	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
446	neccessary) and will handle those and any outstanding ones. It will block	650	necessary) and will handle those and any already outstanding ones. It
447	your process until at least one new event arrives, and will return after	651	will block your process until at least one new event arrives (which could
448	one iteration of the loop. This is useful if you are waiting for some	652	be an event internal to libev itself, so there is no guarantee that a
449	external event in conjunction with something not expressible using other	653	user-registered callback will be called), and will return after one
		654	iteration of the loop.
		655
		656	This is useful if you are waiting for some external event in conjunction
		657	with something not expressible using other libev watchers (i.e. "roll your
450	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	658	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
451	usually a better approach for this kind of thing.	659	usually a better approach for this kind of thing.
452		660
453	Here are the gory details of what C<ev_loop> does:	661	Here are the gory details of what C<ev_loop> does:
454		662
455	* If there are no active watchers (reference count is zero), return.	663	- Before the first iteration, call any pending watchers.
456	- Queue prepare watchers and then call all outstanding watchers.	664	* If EVFLAG_FORKCHECK was used, check for a fork.
		665	- If a fork was detected (by any means), queue and call all fork watchers.
		666	- Queue and call all prepare watchers.
457	- If we have been forked, recreate the kernel state.	667	- If we have been forked, detach and recreate the kernel state
		668	as to not disturb the other process.
458	- Update the kernel state with all outstanding changes.	669	- Update the kernel state with all outstanding changes.
459	- Update the "event loop time".	670	- Update the "event loop time" (ev_now ()).
460	- Calculate for how long to block.	671	- Calculate for how long to sleep or block, if at all
		672	(active idle watchers, EVLOOP_NONBLOCK or not having
		673	any active watchers at all will result in not sleeping).
		674	- Sleep if the I/O and timer collect interval say so.
461	- Block the process, waiting for any events.	675	- Block the process, waiting for any events.
462	- Queue all outstanding I/O (fd) events.	676	- Queue all outstanding I/O (fd) events.
463	- Update the "event loop time" and do time jump handling.	677	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
464	- Queue all outstanding timers.	678	- Queue all expired timers.
465	- Queue all outstanding periodics.	679	- Queue all expired periodics.
466	- If no events are pending now, queue all idle watchers.	680	- Unless any events are pending now, queue all idle watchers.
467	- Queue all check watchers.	681	- Queue all check watchers.
468	- Call all queued watchers in reverse order (i.e. check watchers first).	682	- Call all queued watchers in reverse order (i.e. check watchers first).
469	Signals and child watchers are implemented as I/O watchers, and will	683	Signals and child watchers are implemented as I/O watchers, and will
470	be handled here by queueing them when their watcher gets executed.	684	be handled here by queueing them when their watcher gets executed.
471	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	685	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
472	were used, return, otherwise continue with step *.	686	were used, or there are no active watchers, return, otherwise
		687	continue with step *.
473		688
474	Example: Queue some jobs and then loop until no events are outsanding	689	Example: Queue some jobs and then loop until no events are outstanding
475	anymore.	690	anymore.
476		691
477	... queue jobs here, make sure they register event watchers as long	692	... queue jobs here, make sure they register event watchers as long
478	... as they still have work to do (even an idle watcher will do..)	693	... as they still have work to do (even an idle watcher will do..)
479	ev_loop (my_loop, 0);	694	ev_loop (my_loop, 0);
480	... jobs done. yeah!	695	... jobs done or somebody called unloop. yeah!
481		696
482	=item ev_unloop (loop, how)	697	=item ev_unloop (loop, how)
483		698
484	Can be used to make a call to C<ev_loop> return early (but only after it	699	Can be used to make a call to C<ev_loop> return early (but only after it
485	has processed all outstanding events). The C<how> argument must be either	700	has processed all outstanding events). The C<how> argument must be either
486	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	701	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
487	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	702	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
488		703
		704	This "unloop state" will be cleared when entering C<ev_loop> again.
		705
		706	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
		707
489	=item ev_ref (loop)	708	=item ev_ref (loop)
490		709
491	=item ev_unref (loop)	710	=item ev_unref (loop)
492		711
493	Ref/unref can be used to add or remove a reference count on the event	712	Ref/unref can be used to add or remove a reference count on the event
494	loop: Every watcher keeps one reference, and as long as the reference	713	loop: Every watcher keeps one reference, and as long as the reference
495	count is nonzero, C<ev_loop> will not return on its own. If you have	714	count is nonzero, C<ev_loop> will not return on its own.
		715
496	a watcher you never unregister that should not keep C<ev_loop> from	716	If you have a watcher you never unregister that should not keep C<ev_loop>
497	returning, ev_unref() after starting, and ev_ref() before stopping it. For	717	from returning, call ev_unref() after starting, and ev_ref() before
		718	stopping it.
		719
498	example, libev itself uses this for its internal signal pipe: It is not	720	As an example, libev itself uses this for its internal signal pipe: It is
499	visible to the libev user and should not keep C<ev_loop> from exiting if	721	not visible to the libev user and should not keep C<ev_loop> from exiting
500	no event watchers registered by it are active. It is also an excellent	722	if no event watchers registered by it are active. It is also an excellent
501	way to do this for generic recurring timers or from within third-party	723	way to do this for generic recurring timers or from within third-party
502	libraries. Just remember to I<unref after start> and I<ref before stop>.	724	libraries. Just remember to I<unref after start> and I<ref before stop>
		725	(but only if the watcher wasn't active before, or was active before,
		726	respectively).
503		727
504	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	728	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
505	running when nothing else is active.	729	running when nothing else is active.
506		730
507	struct ev_signal exitsig;	731	ev_signal exitsig;
508	ev_signal_init (&exitsig, sig_cb, SIGINT);	732	ev_signal_init (&exitsig, sig_cb, SIGINT);
509	ev_signal_start (loop, &exitsig);	733	ev_signal_start (loop, &exitsig);
510	evf_unref (loop);	734	evf_unref (loop);
511		735
512	Example: For some weird reason, unregister the above signal handler again.	736	Example: For some weird reason, unregister the above signal handler again.
513		737
514	ev_ref (loop);	738	ev_ref (loop);
515	ev_signal_stop (loop, &exitsig);	739	ev_signal_stop (loop, &exitsig);
		740
		741	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		742
		743	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		744
		745	These advanced functions influence the time that libev will spend waiting
		746	for events. Both time intervals are by default C<0>, meaning that libev
		747	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		748	latency.
		749
		750	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		751	allows libev to delay invocation of I/O and timer/periodic callbacks
		752	to increase efficiency of loop iterations (or to increase power-saving
		753	opportunities).
		754
		755	The idea is that sometimes your program runs just fast enough to handle
		756	one (or very few) event(s) per loop iteration. While this makes the
		757	program responsive, it also wastes a lot of CPU time to poll for new
		758	events, especially with backends like C<select ()> which have a high
		759	overhead for the actual polling but can deliver many events at once.
		760
		761	By setting a higher I<io collect interval> you allow libev to spend more
		762	time collecting I/O events, so you can handle more events per iteration,
		763	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		764	C<ev_timer>) will be not affected. Setting this to a non-null value will
		765	introduce an additional C<ev_sleep ()> call into most loop iterations.
		766
		767	Likewise, by setting a higher I<timeout collect interval> you allow libev
		768	to spend more time collecting timeouts, at the expense of increased
		769	latency/jitter/inexactness (the watcher callback will be called
		770	later). C<ev_io> watchers will not be affected. Setting this to a non-null
		771	value will not introduce any overhead in libev.
		772
		773	Many (busy) programs can usually benefit by setting the I/O collect
		774	interval to a value near C<0.1> or so, which is often enough for
		775	interactive servers (of course not for games), likewise for timeouts. It
		776	usually doesn't make much sense to set it to a lower value than C<0.01>,
		777	as this approaches the timing granularity of most systems.
		778
		779	Setting the I<timeout collect interval> can improve the opportunity for
		780	saving power, as the program will "bundle" timer callback invocations that
		781	are "near" in time together, by delaying some, thus reducing the number of
		782	times the process sleeps and wakes up again. Another useful technique to
		783	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		784	they fire on, say, one-second boundaries only.
		785
		786	=item ev_loop_verify (loop)
		787
		788	This function only does something when C<EV_VERIFY> support has been
		789	compiled in, which is the default for non-minimal builds. It tries to go
		790	through all internal structures and checks them for validity. If anything
		791	is found to be inconsistent, it will print an error message to standard
		792	error and call C<abort ()>.
		793
		794	This can be used to catch bugs inside libev itself: under normal
		795	circumstances, this function will never abort as of course libev keeps its
		796	data structures consistent.
516		797
517	=back	798	=back
518		799
519		800
520	=head1 ANATOMY OF A WATCHER	801	=head1 ANATOMY OF A WATCHER
		802
		803	In the following description, uppercase C<TYPE> in names stands for the
		804	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		805	watchers and C<ev_io_start> for I/O watchers.
521		806
522	A watcher is a structure that you create and register to record your	807	A watcher is a structure that you create and register to record your
523	interest in some event. For instance, if you want to wait for STDIN to	808	interest in some event. For instance, if you want to wait for STDIN to
524	become readable, you would create an C<ev_io> watcher for that:	809	become readable, you would create an C<ev_io> watcher for that:
525		810
526	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	811	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
527	{	812	{
528	ev_io_stop (w);	813	ev_io_stop (w);
529	ev_unloop (loop, EVUNLOOP_ALL);	814	ev_unloop (loop, EVUNLOOP_ALL);
530	}	815	}
531		816
532	struct ev_loop *loop = ev_default_loop (0);	817	struct ev_loop *loop = ev_default_loop (0);
		818
533	struct ev_io stdin_watcher;	819	ev_io stdin_watcher;
		820
534	ev_init (&stdin_watcher, my_cb);	821	ev_init (&stdin_watcher, my_cb);
535	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	822	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
536	ev_io_start (loop, &stdin_watcher);	823	ev_io_start (loop, &stdin_watcher);
		824
537	ev_loop (loop, 0);	825	ev_loop (loop, 0);
538		826
539	As you can see, you are responsible for allocating the memory for your	827	As you can see, you are responsible for allocating the memory for your
540	watcher structures (and it is usually a bad idea to do this on the stack,	828	watcher structures (and it is I<usually> a bad idea to do this on the
541	although this can sometimes be quite valid).	829	stack).
		830
		831	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		832	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
542		833
543	Each watcher structure must be initialised by a call to C<ev_init	834	Each watcher structure must be initialised by a call to C<ev_init
544	(watcher *, callback)>, which expects a callback to be provided. This	835	(watcher *, callback)>, which expects a callback to be provided. This
545	callback gets invoked each time the event occurs (or, in the case of io	836	callback gets invoked each time the event occurs (or, in the case of I/O
546	watchers, each time the event loop detects that the file descriptor given	837	watchers, each time the event loop detects that the file descriptor given
547	is readable and/or writable).	838	is readable and/or writable).
548		839
549	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	840	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
550	with arguments specific to this watcher type. There is also a macro	841	macro to configure it, with arguments specific to the watcher type. There
551	to combine initialisation and setting in one call: C<< ev_<type>_init	842	is also a macro to combine initialisation and setting in one call: C<<
552	(watcher *, callback, ...) >>.	843	ev_TYPE_init (watcher *, callback, ...) >>.
553		844
554	To make the watcher actually watch out for events, you have to start it	845	To make the watcher actually watch out for events, you have to start it
555	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	846	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
556	*) >>), and you can stop watching for events at any time by calling the	847	*) >>), and you can stop watching for events at any time by calling the
557	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	848	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
558		849
559	As long as your watcher is active (has been started but not stopped) you	850	As long as your watcher is active (has been started but not stopped) you
560	must not touch the values stored in it. Most specifically you must never	851	must not touch the values stored in it. Most specifically you must never
561	reinitialise it or call its C<set> macro.	852	reinitialise it or call its C<ev_TYPE_set> macro.
562		853
563	Each and every callback receives the event loop pointer as first, the	854	Each and every callback receives the event loop pointer as first, the
564	registered watcher structure as second, and a bitset of received events as	855	registered watcher structure as second, and a bitset of received events as
565	third argument.	856	third argument.
566		857
…		…
620	=item C<EV_FORK>	911	=item C<EV_FORK>
621		912
622	The event loop has been resumed in the child process after fork (see	913	The event loop has been resumed in the child process after fork (see
623	C<ev_fork>).	914	C<ev_fork>).
624		915
		916	=item C<EV_ASYNC>
		917
		918	The given async watcher has been asynchronously notified (see C<ev_async>).
		919
625	=item C<EV_ERROR>	920	=item C<EV_ERROR>
626		921
627	An unspecified error has occured, the watcher has been stopped. This might	922	An unspecified error has occurred, the watcher has been stopped. This might
628	happen because the watcher could not be properly started because libev	923	happen because the watcher could not be properly started because libev
629	ran out of memory, a file descriptor was found to be closed or any other	924	ran out of memory, a file descriptor was found to be closed or any other
		925	problem. Libev considers these application bugs.
		926
630	problem. You best act on it by reporting the problem and somehow coping	927	You best act on it by reporting the problem and somehow coping with the
631	with the watcher being stopped.	928	watcher being stopped. Note that well-written programs should not receive
		929	an error ever, so when your watcher receives it, this usually indicates a
		930	bug in your program.
632		931
633	Libev will usually signal a few "dummy" events together with an error,	932	Libev will usually signal a few "dummy" events together with an error, for
634	for example it might indicate that a fd is readable or writable, and if	933	example it might indicate that a fd is readable or writable, and if your
635	your callbacks is well-written it can just attempt the operation and cope	934	callbacks is well-written it can just attempt the operation and cope with
636	with the error from read() or write(). This will not work in multithreaded	935	the error from read() or write(). This will not work in multi-threaded
637	programs, though, so beware.	936	programs, though, as the fd could already be closed and reused for another
		937	thing, so beware.
638		938
639	=back	939	=back
640		940
641	=head2 GENERIC WATCHER FUNCTIONS	941	=head2 GENERIC WATCHER FUNCTIONS
642
643	In the following description, C<TYPE> stands for the watcher type,
644	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
645		942
646	=over 4	943	=over 4
647		944
648	=item C<ev_init> (ev_TYPE *watcher, callback)	945	=item C<ev_init> (ev_TYPE *watcher, callback)
649		946
…		…
655	which rolls both calls into one.	952	which rolls both calls into one.
656		953
657	You can reinitialise a watcher at any time as long as it has been stopped	954	You can reinitialise a watcher at any time as long as it has been stopped
658	(or never started) and there are no pending events outstanding.	955	(or never started) and there are no pending events outstanding.
659		956
660	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	957	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
661	int revents)>.	958	int revents)>.
		959
		960	Example: Initialise an C<ev_io> watcher in two steps.
		961
		962	ev_io w;
		963	ev_init (&w, my_cb);
		964	ev_io_set (&w, STDIN_FILENO, EV_READ);
662		965
663	=item C<ev_TYPE_set> (ev_TYPE *, [args])	966	=item C<ev_TYPE_set> (ev_TYPE *, [args])
664		967
665	This macro initialises the type-specific parts of a watcher. You need to	968	This macro initialises the type-specific parts of a watcher. You need to
666	call C<ev_init> at least once before you call this macro, but you can	969	call C<ev_init> at least once before you call this macro, but you can
…		…
669	difference to the C<ev_init> macro).	972	difference to the C<ev_init> macro).
670		973
671	Although some watcher types do not have type-specific arguments	974	Although some watcher types do not have type-specific arguments
672	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	975	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
673		976
		977	See C<ev_init>, above, for an example.
		978
674	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	979	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
675		980
676	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	981	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
677	calls into a single call. This is the most convinient method to initialise	982	calls into a single call. This is the most convenient method to initialise
678	a watcher. The same limitations apply, of course.	983	a watcher. The same limitations apply, of course.
		984
		985	Example: Initialise and set an C<ev_io> watcher in one step.
		986
		987	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
679		988
680	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	989	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)
681		990
682	Starts (activates) the given watcher. Only active watchers will receive	991	Starts (activates) the given watcher. Only active watchers will receive
683	events. If the watcher is already active nothing will happen.	992	events. If the watcher is already active nothing will happen.
684		993
		994	Example: Start the C<ev_io> watcher that is being abused as example in this
		995	whole section.
		996
		997	ev_io_start (EV_DEFAULT_UC, &w);
		998
685	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	999	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)
686		1000
687	Stops the given watcher again (if active) and clears the pending	1001	Stops the given watcher if active, and clears the pending status (whether
		1002	the watcher was active or not).
		1003
688	status. It is possible that stopped watchers are pending (for example,	1004	It is possible that stopped watchers are pending - for example,
689	non-repeating timers are being stopped when they become pending), but	1005	non-repeating timers are being stopped when they become pending - but
690	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	1006	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
691	you want to free or reuse the memory used by the watcher it is therefore a	1007	pending. If you want to free or reuse the memory used by the watcher it is
692	good idea to always call its C<ev_TYPE_stop> function.	1008	therefore a good idea to always call its C<ev_TYPE_stop> function.
693		1009
694	=item bool ev_is_active (ev_TYPE *watcher)	1010	=item bool ev_is_active (ev_TYPE *watcher)
695		1011
696	Returns a true value iff the watcher is active (i.e. it has been started	1012	Returns a true value iff the watcher is active (i.e. it has been started
697	and not yet been stopped). As long as a watcher is active you must not modify	1013	and not yet been stopped). As long as a watcher is active you must not modify
…		…
700	=item bool ev_is_pending (ev_TYPE *watcher)	1016	=item bool ev_is_pending (ev_TYPE *watcher)
701		1017
702	Returns a true value iff the watcher is pending, (i.e. it has outstanding	1018	Returns a true value iff the watcher is pending, (i.e. it has outstanding
703	events but its callback has not yet been invoked). As long as a watcher	1019	events but its callback has not yet been invoked). As long as a watcher
704	is pending (but not active) you must not call an init function on it (but	1020	is pending (but not active) you must not call an init function on it (but
705	C<ev_TYPE_set> is safe) and you must make sure the watcher is available to	1021	C<ev_TYPE_set> is safe), you must not change its priority, and you must
706	libev (e.g. you cnanot C<free ()> it).	1022	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		1023	it).
707		1024
708	=item callback ev_cb (ev_TYPE *watcher)	1025	=item callback ev_cb (ev_TYPE *watcher)
709		1026
710	Returns the callback currently set on the watcher.	1027	Returns the callback currently set on the watcher.
711		1028
712	=item ev_cb_set (ev_TYPE *watcher, callback)	1029	=item ev_cb_set (ev_TYPE *watcher, callback)
713		1030
714	Change the callback. You can change the callback at virtually any time	1031	Change the callback. You can change the callback at virtually any time
715	(modulo threads).	1032	(modulo threads).
716		1033
		1034	=item ev_set_priority (ev_TYPE *watcher, priority)
		1035
		1036	=item int ev_priority (ev_TYPE *watcher)
		1037
		1038	Set and query the priority of the watcher. The priority is a small
		1039	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		1040	(default: C<-2>). Pending watchers with higher priority will be invoked
		1041	before watchers with lower priority, but priority will not keep watchers
		1042	from being executed (except for C<ev_idle> watchers).
		1043
		1044	This means that priorities are I<only> used for ordering callback
		1045	invocation after new events have been received. This is useful, for
		1046	example, to reduce latency after idling, or more often, to bind two
		1047	watchers on the same event and make sure one is called first.
		1048
		1049	If you need to suppress invocation when higher priority events are pending
		1050	you need to look at C<ev_idle> watchers, which provide this functionality.
		1051
		1052	You I<must not> change the priority of a watcher as long as it is active or
		1053	pending.
		1054
		1055	The default priority used by watchers when no priority has been set is
		1056	always C<0>, which is supposed to not be too high and not be too low :).
		1057
		1058	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		1059	fine, as long as you do not mind that the priority value you query might
		1060	or might not have been clamped to the valid range.
		1061
		1062	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		1063
		1064	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		1065	C<loop> nor C<revents> need to be valid as long as the watcher callback
		1066	can deal with that fact, as both are simply passed through to the
		1067	callback.
		1068
		1069	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		1070
		1071	If the watcher is pending, this function clears its pending status and
		1072	returns its C<revents> bitset (as if its callback was invoked). If the
		1073	watcher isn't pending it does nothing and returns C<0>.
		1074
		1075	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1076	callback to be invoked, which can be accomplished with this function.
		1077
717	=back	1078	=back
718		1079
719		1080
720	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1081	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
721		1082
722	Each watcher has, by default, a member C<void *data> that you can change	1083	Each watcher has, by default, a member C<void *data> that you can change
723	and read at any time, libev will completely ignore it. This can be used	1084	and read at any time: libev will completely ignore it. This can be used
724	to associate arbitrary data with your watcher. If you need more data and	1085	to associate arbitrary data with your watcher. If you need more data and
725	don't want to allocate memory and store a pointer to it in that data	1086	don't want to allocate memory and store a pointer to it in that data
726	member, you can also "subclass" the watcher type and provide your own	1087	member, you can also "subclass" the watcher type and provide your own
727	data:	1088	data:
728		1089
729	struct my_io	1090	struct my_io
730	{	1091	{
731	struct ev_io io;	1092	ev_io io;
732	int otherfd;	1093	int otherfd;
733	void *somedata;	1094	void *somedata;
734	struct whatever *mostinteresting;	1095	struct whatever *mostinteresting;
735	}	1096	};
		1097
		1098	...
		1099	struct my_io w;
		1100	ev_io_init (&w.io, my_cb, fd, EV_READ);
736		1101
737	And since your callback will be called with a pointer to the watcher, you	1102	And since your callback will be called with a pointer to the watcher, you
738	can cast it back to your own type:	1103	can cast it back to your own type:
739		1104
740	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1105	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
741	{	1106	{
742	struct my_io *w = (struct my_io *)w_;	1107	struct my_io *w = (struct my_io *)w_;
743	...	1108	...
744	}	1109	}
745		1110
746	More interesting and less C-conformant ways of casting your callback type	1111	More interesting and less C-conformant ways of casting your callback type
747	instead have been omitted.	1112	instead have been omitted.
748		1113
749	Another common scenario is having some data structure with multiple	1114	Another common scenario is to use some data structure with multiple
750	watchers:	1115	embedded watchers:
751		1116
752	struct my_biggy	1117	struct my_biggy
753	{	1118	{
754	int some_data;	1119	int some_data;
755	ev_timer t1;	1120	ev_timer t1;
756	ev_timer t2;	1121	ev_timer t2;
757	}	1122	}
758		1123
759	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1124	In this case getting the pointer to C<my_biggy> is a bit more
760	you need to use C<offsetof>:	1125	complicated: Either you store the address of your C<my_biggy> struct
		1126	in the C<data> member of the watcher (for woozies), or you need to use
		1127	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1128	programmers):
761		1129
762	#include <stddef.h>	1130	#include <stddef.h>
763		1131
764	static void	1132	static void
765	t1_cb (EV_P_ struct ev_timer *w, int revents)	1133	t1_cb (EV_P_ ev_timer *w, int revents)
766	{	1134	{
767	struct my_biggy big = (struct my_biggy *	1135	struct my_biggy big = (struct my_biggy *
768	(((char *)w) - offsetof (struct my_biggy, t1));	1136	(((char *)w) - offsetof (struct my_biggy, t1));
769	}	1137	}
770		1138
771	static void	1139	static void
772	t2_cb (EV_P_ struct ev_timer *w, int revents)	1140	t2_cb (EV_P_ ev_timer *w, int revents)
773	{	1141	{
774	struct my_biggy big = (struct my_biggy *	1142	struct my_biggy big = (struct my_biggy *
775	(((char *)w) - offsetof (struct my_biggy, t2));	1143	(((char *)w) - offsetof (struct my_biggy, t2));
776	}	1144	}
777		1145
778		1146
779	=head1 WATCHER TYPES	1147	=head1 WATCHER TYPES
780		1148
781	This section describes each watcher in detail, but will not repeat	1149	This section describes each watcher in detail, but will not repeat
…		…
805	In general you can register as many read and/or write event watchers per	1173	In general you can register as many read and/or write event watchers per
806	fd as you want (as long as you don't confuse yourself). Setting all file	1174	fd as you want (as long as you don't confuse yourself). Setting all file
807	descriptors to non-blocking mode is also usually a good idea (but not	1175	descriptors to non-blocking mode is also usually a good idea (but not
808	required if you know what you are doing).	1176	required if you know what you are doing).
809		1177
810	You have to be careful with dup'ed file descriptors, though. Some backends	1178	If you cannot use non-blocking mode, then force the use of a
811	(the linux epoll backend is a notable example) cannot handle dup'ed file	1179	known-to-be-good backend (at the time of this writing, this includes only
812	descriptors correctly if you register interest in two or more fds pointing	1180	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>).
813	to the same underlying file/socket/etc. description (that is, they share
814	the same underlying "file open").
815
816	If you must do this, then force the use of a known-to-be-good backend
817	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
818	C<EVBACKEND_POLL>).
819		1181
820	Another thing you have to watch out for is that it is quite easy to	1182	Another thing you have to watch out for is that it is quite easy to
821	receive "spurious" readyness notifications, that is your callback might	1183	receive "spurious" readiness notifications, that is your callback might
822	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1184	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
823	because there is no data. Not only are some backends known to create a	1185	because there is no data. Not only are some backends known to create a
824	lot of those (for example solaris ports), it is very easy to get into	1186	lot of those (for example Solaris ports), it is very easy to get into
825	this situation even with a relatively standard program structure. Thus	1187	this situation even with a relatively standard program structure. Thus
826	it is best to always use non-blocking I/O: An extra C<read>(2) returning	1188	it is best to always use non-blocking I/O: An extra C<read>(2) returning
827	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1189	C<EAGAIN> is far preferable to a program hanging until some data arrives.
828		1190
829	If you cannot run the fd in non-blocking mode (for example you should not	1191	If you cannot run the fd in non-blocking mode (for example you should
830	play around with an Xlib connection), then you have to seperately re-test	1192	not play around with an Xlib connection), then you have to separately
831	wether a file descriptor is really ready with a known-to-be good interface	1193	re-test whether a file descriptor is really ready with a known-to-be good
832	such as poll (fortunately in our Xlib example, Xlib already does this on	1194	interface such as poll (fortunately in our Xlib example, Xlib already
833	its own, so its quite safe to use).	1195	does this on its own, so its quite safe to use). Some people additionally
		1196	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1197	indefinitely.
		1198
		1199	But really, best use non-blocking mode.
		1200
		1201	=head3 The special problem of disappearing file descriptors
		1202
		1203	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1204	descriptor (either due to calling C<close> explicitly or any other means,
		1205	such as C<dup2>). The reason is that you register interest in some file
		1206	descriptor, but when it goes away, the operating system will silently drop
		1207	this interest. If another file descriptor with the same number then is
		1208	registered with libev, there is no efficient way to see that this is, in
		1209	fact, a different file descriptor.
		1210
		1211	To avoid having to explicitly tell libev about such cases, libev follows
		1212	the following policy: Each time C<ev_io_set> is being called, libev
		1213	will assume that this is potentially a new file descriptor, otherwise
		1214	it is assumed that the file descriptor stays the same. That means that
		1215	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1216	descriptor even if the file descriptor number itself did not change.
		1217
		1218	This is how one would do it normally anyway, the important point is that
		1219	the libev application should not optimise around libev but should leave
		1220	optimisations to libev.
		1221
		1222	=head3 The special problem of dup'ed file descriptors
		1223
		1224	Some backends (e.g. epoll), cannot register events for file descriptors,
		1225	but only events for the underlying file descriptions. That means when you
		1226	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1227	events for them, only one file descriptor might actually receive events.
		1228
		1229	There is no workaround possible except not registering events
		1230	for potentially C<dup ()>'ed file descriptors, or to resort to
		1231	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1232
		1233	=head3 The special problem of fork
		1234
		1235	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1236	useless behaviour. Libev fully supports fork, but needs to be told about
		1237	it in the child.
		1238
		1239	To support fork in your programs, you either have to call
		1240	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1241	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1242	C<EVBACKEND_POLL>.
		1243
		1244	=head3 The special problem of SIGPIPE
		1245
		1246	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
		1247	when writing to a pipe whose other end has been closed, your program gets
		1248	sent a SIGPIPE, which, by default, aborts your program. For most programs
		1249	this is sensible behaviour, for daemons, this is usually undesirable.
		1250
		1251	So when you encounter spurious, unexplained daemon exits, make sure you
		1252	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
		1253	somewhere, as that would have given you a big clue).
		1254
		1255
		1256	=head3 Watcher-Specific Functions
834		1257
835	=over 4	1258	=over 4
836		1259
837	=item ev_io_init (ev_io *, callback, int fd, int events)	1260	=item ev_io_init (ev_io *, callback, int fd, int events)
838		1261
839	=item ev_io_set (ev_io *, int fd, int events)	1262	=item ev_io_set (ev_io *, int fd, int events)
840		1263
841	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1264	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
842	rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or	1265	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
843	C<EV_READ | EV_WRITE> to receive the given events.	1266	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
844		1267
845	=item int fd [read-only]	1268	=item int fd [read-only]
846		1269
847	The file descriptor being watched.	1270	The file descriptor being watched.
848		1271
849	=item int events [read-only]	1272	=item int events [read-only]
850		1273
851	The events being watched.	1274	The events being watched.
852		1275
853	=back	1276	=back
		1277
		1278	=head3 Examples
854		1279
855	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1280	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
856	readable, but only once. Since it is likely line-buffered, you could	1281	readable, but only once. Since it is likely line-buffered, you could
857	attempt to read a whole line in the callback.	1282	attempt to read a whole line in the callback.
858		1283
859	static void	1284	static void
860	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1285	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
861	{	1286	{
862	ev_io_stop (loop, w);	1287	ev_io_stop (loop, w);
863	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1288	.. read from stdin here (or from w->fd) and handle any I/O errors
864	}	1289	}
865		1290
866	...	1291	...
867	struct ev_loop *loop = ev_default_init (0);	1292	struct ev_loop *loop = ev_default_init (0);
868	struct ev_io stdin_readable;	1293	ev_io stdin_readable;
869	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1294	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
870	ev_io_start (loop, &stdin_readable);	1295	ev_io_start (loop, &stdin_readable);
871	ev_loop (loop, 0);	1296	ev_loop (loop, 0);
872		1297
873		1298
874	=head2 C<ev_timer> - relative and optionally repeating timeouts	1299	=head2 C<ev_timer> - relative and optionally repeating timeouts
875		1300
876	Timer watchers are simple relative timers that generate an event after a	1301	Timer watchers are simple relative timers that generate an event after a
877	given time, and optionally repeating in regular intervals after that.	1302	given time, and optionally repeating in regular intervals after that.
878		1303
879	The timers are based on real time, that is, if you register an event that	1304	The timers are based on real time, that is, if you register an event that
880	times out after an hour and you reset your system clock to last years	1305	times out after an hour and you reset your system clock to January last
881	time, it will still time out after (roughly) and hour. "Roughly" because	1306	year, it will still time out after (roughly) one hour. "Roughly" because
882	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1307	detecting time jumps is hard, and some inaccuracies are unavoidable (the
883	monotonic clock option helps a lot here).	1308	monotonic clock option helps a lot here).
		1309
		1310	The callback is guaranteed to be invoked only I<after> its timeout has
		1311	passed, but if multiple timers become ready during the same loop iteration
		1312	then order of execution is undefined.
		1313
		1314	=head3 Be smart about timeouts
		1315
		1316	Many real-world problems involve some kind of timeout, usually for error
		1317	recovery. A typical example is an HTTP request - if the other side hangs,
		1318	you want to raise some error after a while.
		1319
		1320	What follows are some ways to handle this problem, from obvious and
		1321	inefficient to smart and efficient.
		1322
		1323	In the following, a 60 second activity timeout is assumed - a timeout that
		1324	gets reset to 60 seconds each time there is activity (e.g. each time some
		1325	data or other life sign was received).
		1326
		1327	=over 4
		1328
		1329	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1330
		1331	This is the most obvious, but not the most simple way: In the beginning,
		1332	start the watcher:
		1333
		1334	ev_timer_init (timer, callback, 60., 0.);
		1335	ev_timer_start (loop, timer);
		1336
		1337	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1338	and start it again:
		1339
		1340	ev_timer_stop (loop, timer);
		1341	ev_timer_set (timer, 60., 0.);
		1342	ev_timer_start (loop, timer);
		1343
		1344	This is relatively simple to implement, but means that each time there is
		1345	some activity, libev will first have to remove the timer from its internal
		1346	data structure and then add it again. Libev tries to be fast, but it's
		1347	still not a constant-time operation.
		1348
		1349	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1350
		1351	This is the easiest way, and involves using C<ev_timer_again> instead of
		1352	C<ev_timer_start>.
		1353
		1354	To implement this, configure an C<ev_timer> with a C<repeat> value
		1355	of C<60> and then call C<ev_timer_again> at start and each time you
		1356	successfully read or write some data. If you go into an idle state where
		1357	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1358	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1359
		1360	That means you can ignore both the C<ev_timer_start> function and the
		1361	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1362	member and C<ev_timer_again>.
		1363
		1364	At start:
		1365
		1366	ev_timer_init (timer, callback);
		1367	timer->repeat = 60.;
		1368	ev_timer_again (loop, timer);
		1369
		1370	Each time there is some activity:
		1371
		1372	ev_timer_again (loop, timer);
		1373
		1374	It is even possible to change the time-out on the fly, regardless of
		1375	whether the watcher is active or not:
		1376
		1377	timer->repeat = 30.;
		1378	ev_timer_again (loop, timer);
		1379
		1380	This is slightly more efficient then stopping/starting the timer each time
		1381	you want to modify its timeout value, as libev does not have to completely
		1382	remove and re-insert the timer from/into its internal data structure.
		1383
		1384	It is, however, even simpler than the "obvious" way to do it.
		1385
		1386	=item 3. Let the timer time out, but then re-arm it as required.
		1387
		1388	This method is more tricky, but usually most efficient: Most timeouts are
		1389	relatively long compared to the intervals between other activity - in
		1390	our example, within 60 seconds, there are usually many I/O events with
		1391	associated activity resets.
		1392
		1393	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1394	but remember the time of last activity, and check for a real timeout only
		1395	within the callback:
		1396
		1397	ev_tstamp last_activity; // time of last activity
		1398
		1399	static void
		1400	callback (EV_P_ ev_timer *w, int revents)
		1401	{
		1402	ev_tstamp now = ev_now (EV_A);
		1403	ev_tstamp timeout = last_activity + 60.;
		1404
		1405	// if last_activity + 60. is older than now, we did time out
		1406	if (timeout < now)
		1407	{
		1408	// timeout occured, take action
		1409	}
		1410	else
		1411	{
		1412	// callback was invoked, but there was some activity, re-arm
		1413	// the watcher to fire in last_activity + 60, which is
		1414	// guaranteed to be in the future, so "again" is positive:
		1415	w->again = timeout - now;
		1416	ev_timer_again (EV_A_ w);
		1417	}
		1418	}
		1419
		1420	To summarise the callback: first calculate the real timeout (defined
		1421	as "60 seconds after the last activity"), then check if that time has
		1422	been reached, which means something I<did>, in fact, time out. Otherwise
		1423	the callback was invoked too early (C<timeout> is in the future), so
		1424	re-schedule the timer to fire at that future time, to see if maybe we have
		1425	a timeout then.
		1426
		1427	Note how C<ev_timer_again> is used, taking advantage of the
		1428	C<ev_timer_again> optimisation when the timer is already running.
		1429
		1430	This scheme causes more callback invocations (about one every 60 seconds
		1431	minus half the average time between activity), but virtually no calls to
		1432	libev to change the timeout.
		1433
		1434	To start the timer, simply initialise the watcher and set C<last_activity>
		1435	to the current time (meaning we just have some activity :), then call the
		1436	callback, which will "do the right thing" and start the timer:
		1437
		1438	ev_timer_init (timer, callback);
		1439	last_activity = ev_now (loop);
		1440	callback (loop, timer, EV_TIMEOUT);
		1441
		1442	And when there is some activity, simply store the current time in
		1443	C<last_activity>, no libev calls at all:
		1444
		1445	last_actiivty = ev_now (loop);
		1446
		1447	This technique is slightly more complex, but in most cases where the
		1448	time-out is unlikely to be triggered, much more efficient.
		1449
		1450	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1451	callback :) - just change the timeout and invoke the callback, which will
		1452	fix things for you.
		1453
		1454	=item 4. Wee, just use a double-linked list for your timeouts.
		1455
		1456	If there is not one request, but many thousands (millions...), all
		1457	employing some kind of timeout with the same timeout value, then one can
		1458	do even better:
		1459
		1460	When starting the timeout, calculate the timeout value and put the timeout
		1461	at the I<end> of the list.
		1462
		1463	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1464	the list is expected to fire (for example, using the technique #3).
		1465
		1466	When there is some activity, remove the timer from the list, recalculate
		1467	the timeout, append it to the end of the list again, and make sure to
		1468	update the C<ev_timer> if it was taken from the beginning of the list.
		1469
		1470	This way, one can manage an unlimited number of timeouts in O(1) time for
		1471	starting, stopping and updating the timers, at the expense of a major
		1472	complication, and having to use a constant timeout. The constant timeout
		1473	ensures that the list stays sorted.
		1474
		1475	=back
		1476
		1477	So which method the best?
		1478
		1479	Method #2 is a simple no-brain-required solution that is adequate in most
		1480	situations. Method #3 requires a bit more thinking, but handles many cases
		1481	better, and isn't very complicated either. In most case, choosing either
		1482	one is fine, with #3 being better in typical situations.
		1483
		1484	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1485	rather complicated, but extremely efficient, something that really pays
		1486	off after the first million or so of active timers, i.e. it's usually
		1487	overkill :)
		1488
		1489	=head3 The special problem of time updates
		1490
		1491	Establishing the current time is a costly operation (it usually takes at
		1492	least two system calls): EV therefore updates its idea of the current
		1493	time only before and after C<ev_loop> collects new events, which causes a
		1494	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1495	lots of events in one iteration.
884		1496
885	The relative timeouts are calculated relative to the C<ev_now ()>	1497	The relative timeouts are calculated relative to the C<ev_now ()>
886	time. This is usually the right thing as this timestamp refers to the time	1498	time. This is usually the right thing as this timestamp refers to the time
887	of the event triggering whatever timeout you are modifying/starting. If	1499	of the event triggering whatever timeout you are modifying/starting. If
888	you suspect event processing to be delayed and you I<need> to base the timeout	1500	you suspect event processing to be delayed and you I<need> to base the
889	on the current time, use something like this to adjust for this:	1501	timeout on the current time, use something like this to adjust for this:
890		1502
891	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1503	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
892		1504
893	The callback is guarenteed to be invoked only when its timeout has passed,	1505	If the event loop is suspended for a long time, you can also force an
894	but if multiple timers become ready during the same loop iteration then	1506	update of the time returned by C<ev_now ()> by calling C<ev_now_update
895	order of execution is undefined.	1507	()>.
		1508
		1509	=head3 Watcher-Specific Functions and Data Members
896		1510
897	=over 4	1511	=over 4
898		1512
899	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1513	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
900		1514
901	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1515	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
902		1516
903	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1517	Configure the timer to trigger after C<after> seconds. If C<repeat>
904	C<0.>, then it will automatically be stopped. If it is positive, then the	1518	is C<0.>, then it will automatically be stopped once the timeout is
905	timer will automatically be configured to trigger again C<repeat> seconds	1519	reached. If it is positive, then the timer will automatically be
906	later, again, and again, until stopped manually.	1520	configured to trigger again C<repeat> seconds later, again, and again,
		1521	until stopped manually.
907		1522
908	The timer itself will do a best-effort at avoiding drift, that is, if you	1523	The timer itself will do a best-effort at avoiding drift, that is, if
909	configure a timer to trigger every 10 seconds, then it will trigger at	1524	you configure a timer to trigger every 10 seconds, then it will normally
910	exactly 10 second intervals. If, however, your program cannot keep up with	1525	trigger at exactly 10 second intervals. If, however, your program cannot
911	the timer (because it takes longer than those 10 seconds to do stuff) the	1526	keep up with the timer (because it takes longer than those 10 seconds to
912	timer will not fire more than once per event loop iteration.	1527	do stuff) the timer will not fire more than once per event loop iteration.
913		1528
914	=item ev_timer_again (loop)	1529	=item ev_timer_again (loop, ev_timer *)
915		1530
916	This will act as if the timer timed out and restart it again if it is	1531	This will act as if the timer timed out and restart it again if it is
917	repeating. The exact semantics are:	1532	repeating. The exact semantics are:
918		1533
		1534	If the timer is pending, its pending status is cleared.
		1535
919	If the timer is started but nonrepeating, stop it.	1536	If the timer is started but non-repeating, stop it (as if it timed out).
920		1537
921	If the timer is repeating, either start it if necessary (with the repeat	1538	If the timer is repeating, either start it if necessary (with the
922	value), or reset the running timer to the repeat value.	1539	C<repeat> value), or reset the running timer to the C<repeat> value.
923		1540
924	This sounds a bit complicated, but here is a useful and typical	1541	This sounds a bit complicated, see "Be smart about timeouts", above, for a
925	example: Imagine you have a tcp connection and you want a so-called	1542	usage example.
926	idle timeout, that is, you want to be called when there have been,
927	say, 60 seconds of inactivity on the socket. The easiest way to do
928	this is to configure an C<ev_timer> with C<after>=C<repeat>=C<60> and calling
929	C<ev_timer_again> each time you successfully read or write some data. If
930	you go into an idle state where you do not expect data to travel on the
931	socket, you can stop the timer, and again will automatically restart it if
932	need be.
933
934	You can also ignore the C<after> value and C<ev_timer_start> altogether
935	and only ever use the C<repeat> value:
936
937	ev_timer_init (timer, callback, 0., 5.);
938	ev_timer_again (loop, timer);
939	...
940	timer->again = 17.;
941	ev_timer_again (loop, timer);
942	...
943	timer->again = 10.;
944	ev_timer_again (loop, timer);
945
946	This is more efficient then stopping/starting the timer eahc time you want
947	to modify its timeout value.
948		1543
949	=item ev_tstamp repeat [read-write]	1544	=item ev_tstamp repeat [read-write]
950		1545
951	The current C<repeat> value. Will be used each time the watcher times out	1546	The current C<repeat> value. Will be used each time the watcher times out
952	or C<ev_timer_again> is called and determines the next timeout (if any),	1547	or C<ev_timer_again> is called, and determines the next timeout (if any),
953	which is also when any modifications are taken into account.	1548	which is also when any modifications are taken into account.
954		1549
955	=back	1550	=back
956		1551
		1552	=head3 Examples
		1553
957	Example: Create a timer that fires after 60 seconds.	1554	Example: Create a timer that fires after 60 seconds.
958		1555
959	static void	1556	static void
960	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1557	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
961	{	1558	{
962	.. one minute over, w is actually stopped right here	1559	.. one minute over, w is actually stopped right here
963	}	1560	}
964		1561
965	struct ev_timer mytimer;	1562	ev_timer mytimer;
966	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	1563	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
967	ev_timer_start (loop, &mytimer);	1564	ev_timer_start (loop, &mytimer);
968		1565
969	Example: Create a timeout timer that times out after 10 seconds of	1566	Example: Create a timeout timer that times out after 10 seconds of
970	inactivity.	1567	inactivity.
971		1568
972	static void	1569	static void
973	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1570	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
974	{	1571	{
975	.. ten seconds without any activity	1572	.. ten seconds without any activity
976	}	1573	}
977		1574
978	struct ev_timer mytimer;	1575	ev_timer mytimer;
979	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	1576	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
980	ev_timer_again (&mytimer); /* start timer */	1577	ev_timer_again (&mytimer); /* start timer */
981	ev_loop (loop, 0);	1578	ev_loop (loop, 0);
982		1579
983	// and in some piece of code that gets executed on any "activity":	1580	// and in some piece of code that gets executed on any "activity":
984	// reset the timeout to start ticking again at 10 seconds	1581	// reset the timeout to start ticking again at 10 seconds
985	ev_timer_again (&mytimer);	1582	ev_timer_again (&mytimer);
986		1583
987		1584
988	=head2 C<ev_periodic> - to cron or not to cron?	1585	=head2 C<ev_periodic> - to cron or not to cron?
989		1586
990	Periodic watchers are also timers of a kind, but they are very versatile	1587	Periodic watchers are also timers of a kind, but they are very versatile
991	(and unfortunately a bit complex).	1588	(and unfortunately a bit complex).
992		1589
993	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1590	Unlike C<ev_timer>'s, they are not based on real time (or relative time)
994	but on wallclock time (absolute time). You can tell a periodic watcher	1591	but on wall clock time (absolute time). You can tell a periodic watcher
995	to trigger "at" some specific point in time. For example, if you tell a	1592	to trigger after some specific point in time. For example, if you tell a
996	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	1593	periodic watcher to trigger in 10 seconds (by specifying e.g. C<ev_now ()
997	+ 10.>) and then reset your system clock to the last year, then it will	1594	+ 10.>, that is, an absolute time not a delay) and then reset your system
		1595	clock to January of the previous year, then it will take more than year
998	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1596	to trigger the event (unlike an C<ev_timer>, which would still trigger
999	roughly 10 seconds later and of course not if you reset your system time	1597	roughly 10 seconds later as it uses a relative timeout).
1000	again).
1001		1598
1002	They can also be used to implement vastly more complex timers, such as	1599	C<ev_periodic>s can also be used to implement vastly more complex timers,
1003	triggering an event on eahc midnight, local time.	1600	such as triggering an event on each "midnight, local time", or other
		1601	complicated rules.
1004		1602
1005	As with timers, the callback is guarenteed to be invoked only when the	1603	As with timers, the callback is guaranteed to be invoked only when the
1006	time (C<at>) has been passed, but if multiple periodic timers become ready	1604	time (C<at>) has passed, but if multiple periodic timers become ready
1007	during the same loop iteration then order of execution is undefined.	1605	during the same loop iteration, then order of execution is undefined.
		1606
		1607	=head3 Watcher-Specific Functions and Data Members
1008		1608
1009	=over 4	1609	=over 4
1010		1610
1011	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1611	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
1012		1612
1013	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1613	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
1014		1614
1015	Lots of arguments, lets sort it out... There are basically three modes of	1615	Lots of arguments, lets sort it out... There are basically three modes of
1016	operation, and we will explain them from simplest to complex:	1616	operation, and we will explain them from simplest to most complex:
1017		1617
1018	=over 4	1618	=over 4
1019		1619
1020	=item * absolute timer (interval = reschedule_cb = 0)	1620	=item * absolute timer (at = time, interval = reschedule_cb = 0)
1021		1621
1022	In this configuration the watcher triggers an event at the wallclock time	1622	In this configuration the watcher triggers an event after the wall clock
1023	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1623	time C<at> has passed. It will not repeat and will not adjust when a time
1024	that is, if it is to be run at January 1st 2011 then it will run when the	1624	jump occurs, that is, if it is to be run at January 1st 2011 then it will
1025	system time reaches or surpasses this time.	1625	only run when the system clock reaches or surpasses this time.
1026		1626
1027	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1627	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1028		1628
1029	In this mode the watcher will always be scheduled to time out at the next	1629	In this mode the watcher will always be scheduled to time out at the next
1030	C<at + N * interval> time (for some integer N) and then repeat, regardless	1630	C<at + N * interval> time (for some integer N, which can also be negative)
1031	of any time jumps.	1631	and then repeat, regardless of any time jumps.
1032		1632
1033	This can be used to create timers that do not drift with respect to system	1633	This can be used to create timers that do not drift with respect to the
1034	time:	1634	system clock, for example, here is a C<ev_periodic> that triggers each
		1635	hour, on the hour:
1035		1636
1036	ev_periodic_set (&periodic, 0., 3600., 0);	1637	ev_periodic_set (&periodic, 0., 3600., 0);
1037		1638
1038	This doesn't mean there will always be 3600 seconds in between triggers,	1639	This doesn't mean there will always be 3600 seconds in between triggers,
1039	but only that the the callback will be called when the system time shows a	1640	but only that the callback will be called when the system time shows a
1040	full hour (UTC), or more correctly, when the system time is evenly divisible	1641	full hour (UTC), or more correctly, when the system time is evenly divisible
1041	by 3600.	1642	by 3600.
1042		1643
1043	Another way to think about it (for the mathematically inclined) is that	1644	Another way to think about it (for the mathematically inclined) is that
1044	C<ev_periodic> will try to run the callback in this mode at the next possible	1645	C<ev_periodic> will try to run the callback in this mode at the next possible
1045	time where C<time = at (mod interval)>, regardless of any time jumps.	1646	time where C<time = at (mod interval)>, regardless of any time jumps.
1046		1647
		1648	For numerical stability it is preferable that the C<at> value is near
		1649	C<ev_now ()> (the current time), but there is no range requirement for
		1650	this value, and in fact is often specified as zero.
		1651
		1652	Note also that there is an upper limit to how often a timer can fire (CPU
		1653	speed for example), so if C<interval> is very small then timing stability
		1654	will of course deteriorate. Libev itself tries to be exact to be about one
		1655	millisecond (if the OS supports it and the machine is fast enough).
		1656
1047	=item * manual reschedule mode (reschedule_cb = callback)	1657	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1048		1658
1049	In this mode the values for C<interval> and C<at> are both being	1659	In this mode the values for C<interval> and C<at> are both being
1050	ignored. Instead, each time the periodic watcher gets scheduled, the	1660	ignored. Instead, each time the periodic watcher gets scheduled, the
1051	reschedule callback will be called with the watcher as first, and the	1661	reschedule callback will be called with the watcher as first, and the
1052	current time as second argument.	1662	current time as second argument.
1053		1663
1054	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1664	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
1055	ever, or make any event loop modifications>. If you need to stop it,	1665	ever, or make ANY event loop modifications whatsoever>.
1056	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by
1057	starting a prepare watcher).
1058		1666
		1667	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		1668	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		1669	only event loop modification you are allowed to do).
		1670
1059	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	1671	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1060	ev_tstamp now)>, e.g.:	1672	*w, ev_tstamp now)>, e.g.:
1061		1673
		1674	static ev_tstamp
1062	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1675	my_rescheduler (ev_periodic *w, ev_tstamp now)
1063	{	1676	{
1064	return now + 60.;	1677	return now + 60.;
1065	}	1678	}
1066		1679
1067	It must return the next time to trigger, based on the passed time value	1680	It must return the next time to trigger, based on the passed time value
1068	(that is, the lowest time value larger than to the second argument). It	1681	(that is, the lowest time value larger than to the second argument). It
1069	will usually be called just before the callback will be triggered, but	1682	will usually be called just before the callback will be triggered, but
1070	might be called at other times, too.	1683	might be called at other times, too.
1071		1684
1072	NOTE: I<< This callback must always return a time that is later than the	1685	NOTE: I<< This callback must always return a time that is higher than or
1073	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	1686	equal to the passed C<now> value >>.
1074		1687
1075	This can be used to create very complex timers, such as a timer that	1688	This can be used to create very complex timers, such as a timer that
1076	triggers on each midnight, local time. To do this, you would calculate the	1689	triggers on "next midnight, local time". To do this, you would calculate the
1077	next midnight after C<now> and return the timestamp value for this. How	1690	next midnight after C<now> and return the timestamp value for this. How
1078	you do this is, again, up to you (but it is not trivial, which is the main	1691	you do this is, again, up to you (but it is not trivial, which is the main
1079	reason I omitted it as an example).	1692	reason I omitted it as an example).
1080		1693
1081	=back	1694	=back
…		…
1085	Simply stops and restarts the periodic watcher again. This is only useful	1698	Simply stops and restarts the periodic watcher again. This is only useful
1086	when you changed some parameters or the reschedule callback would return	1699	when you changed some parameters or the reschedule callback would return
1087	a different time than the last time it was called (e.g. in a crond like	1700	a different time than the last time it was called (e.g. in a crond like
1088	program when the crontabs have changed).	1701	program when the crontabs have changed).
1089		1702
		1703	=item ev_tstamp ev_periodic_at (ev_periodic *)
		1704
		1705	When active, returns the absolute time that the watcher is supposed to
		1706	trigger next.
		1707
		1708	=item ev_tstamp offset [read-write]
		1709
		1710	When repeating, this contains the offset value, otherwise this is the
		1711	absolute point in time (the C<at> value passed to C<ev_periodic_set>).
		1712
		1713	Can be modified any time, but changes only take effect when the periodic
		1714	timer fires or C<ev_periodic_again> is being called.
		1715
1090	=item ev_tstamp interval [read-write]	1716	=item ev_tstamp interval [read-write]
1091		1717
1092	The current interval value. Can be modified any time, but changes only	1718	The current interval value. Can be modified any time, but changes only
1093	take effect when the periodic timer fires or C<ev_periodic_again> is being	1719	take effect when the periodic timer fires or C<ev_periodic_again> is being
1094	called.	1720	called.
1095		1721
1096	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	1722	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1097		1723
1098	The current reschedule callback, or C<0>, if this functionality is	1724	The current reschedule callback, or C<0>, if this functionality is
1099	switched off. Can be changed any time, but changes only take effect when	1725	switched off. Can be changed any time, but changes only take effect when
1100	the periodic timer fires or C<ev_periodic_again> is being called.	1726	the periodic timer fires or C<ev_periodic_again> is being called.
1101		1727
1102	=back	1728	=back
1103		1729
		1730	=head3 Examples
		1731
1104	Example: Call a callback every hour, or, more precisely, whenever the	1732	Example: Call a callback every hour, or, more precisely, whenever the
1105	system clock is divisible by 3600. The callback invocation times have	1733	system time is divisible by 3600. The callback invocation times have
1106	potentially a lot of jittering, but good long-term stability.	1734	potentially a lot of jitter, but good long-term stability.
1107		1735
1108	static void	1736	static void
1109	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1737	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
1110	{	1738	{
1111	... its now a full hour (UTC, or TAI or whatever your clock follows)	1739	... its now a full hour (UTC, or TAI or whatever your clock follows)
1112	}	1740	}
1113		1741
1114	struct ev_periodic hourly_tick;	1742	ev_periodic hourly_tick;
1115	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	1743	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1116	ev_periodic_start (loop, &hourly_tick);	1744	ev_periodic_start (loop, &hourly_tick);
1117		1745
1118	Example: The same as above, but use a reschedule callback to do it:	1746	Example: The same as above, but use a reschedule callback to do it:
1119		1747
1120	#include <math.h>	1748	#include <math.h>
1121		1749
1122	static ev_tstamp	1750	static ev_tstamp
1123	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	1751	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1124	{	1752	{
1125	return fmod (now, 3600.) + 3600.;	1753	return now + (3600. - fmod (now, 3600.));
1126	}	1754	}
1127		1755
1128	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	1756	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1129		1757
1130	Example: Call a callback every hour, starting now:	1758	Example: Call a callback every hour, starting now:
1131		1759
1132	struct ev_periodic hourly_tick;	1760	ev_periodic hourly_tick;
1133	ev_periodic_init (&hourly_tick, clock_cb,	1761	ev_periodic_init (&hourly_tick, clock_cb,
1134	fmod (ev_now (loop), 3600.), 3600., 0);	1762	fmod (ev_now (loop), 3600.), 3600., 0);
1135	ev_periodic_start (loop, &hourly_tick);	1763	ev_periodic_start (loop, &hourly_tick);
1136		1764
1137		1765
1138	=head2 C<ev_signal> - signal me when a signal gets signalled!	1766	=head2 C<ev_signal> - signal me when a signal gets signalled!
1139		1767
1140	Signal watchers will trigger an event when the process receives a specific	1768	Signal watchers will trigger an event when the process receives a specific
1141	signal one or more times. Even though signals are very asynchronous, libev	1769	signal one or more times. Even though signals are very asynchronous, libev
1142	will try it's best to deliver signals synchronously, i.e. as part of the	1770	will try it's best to deliver signals synchronously, i.e. as part of the
1143	normal event processing, like any other event.	1771	normal event processing, like any other event.
1144		1772
		1773	If you want signals asynchronously, just use C<sigaction> as you would
		1774	do without libev and forget about sharing the signal. You can even use
		1775	C<ev_async> from a signal handler to synchronously wake up an event loop.
		1776
1145	You can configure as many watchers as you like per signal. Only when the	1777	You can configure as many watchers as you like per signal. Only when the
1146	first watcher gets started will libev actually register a signal watcher	1778	first watcher gets started will libev actually register a signal handler
1147	with the kernel (thus it coexists with your own signal handlers as long	1779	with the kernel (thus it coexists with your own signal handlers as long as
1148	as you don't register any with libev). Similarly, when the last signal	1780	you don't register any with libev for the same signal). Similarly, when
1149	watcher for a signal is stopped libev will reset the signal handler to	1781	the last signal watcher for a signal is stopped, libev will reset the
1150	SIG_DFL (regardless of what it was set to before).	1782	signal handler to SIG_DFL (regardless of what it was set to before).
		1783
		1784	If possible and supported, libev will install its handlers with
		1785	C<SA_RESTART> behaviour enabled, so system calls should not be unduly
		1786	interrupted. If you have a problem with system calls getting interrupted by
		1787	signals you can block all signals in an C<ev_check> watcher and unblock
		1788	them in an C<ev_prepare> watcher.
		1789
		1790	=head3 Watcher-Specific Functions and Data Members
1151		1791
1152	=over 4	1792	=over 4
1153		1793
1154	=item ev_signal_init (ev_signal *, callback, int signum)	1794	=item ev_signal_init (ev_signal *, callback, int signum)
1155		1795
…		…
1162		1802
1163	The signal the watcher watches out for.	1803	The signal the watcher watches out for.
1164		1804
1165	=back	1805	=back
1166		1806
		1807	=head3 Examples
		1808
		1809	Example: Try to exit cleanly on SIGINT.
		1810
		1811	static void
		1812	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
		1813	{
		1814	ev_unloop (loop, EVUNLOOP_ALL);
		1815	}
		1816
		1817	ev_signal signal_watcher;
		1818	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		1819	ev_signal_start (loop, &signal_watcher);
		1820
1167		1821
1168	=head2 C<ev_child> - watch out for process status changes	1822	=head2 C<ev_child> - watch out for process status changes
1169		1823
1170	Child watchers trigger when your process receives a SIGCHLD in response to	1824	Child watchers trigger when your process receives a SIGCHLD in response to
1171	some child status changes (most typically when a child of yours dies).	1825	some child status changes (most typically when a child of yours dies or
		1826	exits). It is permissible to install a child watcher I<after> the child
		1827	has been forked (which implies it might have already exited), as long
		1828	as the event loop isn't entered (or is continued from a watcher), i.e.,
		1829	forking and then immediately registering a watcher for the child is fine,
		1830	but forking and registering a watcher a few event loop iterations later is
		1831	not.
		1832
		1833	Only the default event loop is capable of handling signals, and therefore
		1834	you can only register child watchers in the default event loop.
		1835
		1836	=head3 Process Interaction
		1837
		1838	Libev grabs C<SIGCHLD> as soon as the default event loop is
		1839	initialised. This is necessary to guarantee proper behaviour even if
		1840	the first child watcher is started after the child exits. The occurrence
		1841	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		1842	synchronously as part of the event loop processing. Libev always reaps all
		1843	children, even ones not watched.
		1844
		1845	=head3 Overriding the Built-In Processing
		1846
		1847	Libev offers no special support for overriding the built-in child
		1848	processing, but if your application collides with libev's default child
		1849	handler, you can override it easily by installing your own handler for
		1850	C<SIGCHLD> after initialising the default loop, and making sure the
		1851	default loop never gets destroyed. You are encouraged, however, to use an
		1852	event-based approach to child reaping and thus use libev's support for
		1853	that, so other libev users can use C<ev_child> watchers freely.
		1854
		1855	=head3 Stopping the Child Watcher
		1856
		1857	Currently, the child watcher never gets stopped, even when the
		1858	child terminates, so normally one needs to stop the watcher in the
		1859	callback. Future versions of libev might stop the watcher automatically
		1860	when a child exit is detected.
		1861
		1862	=head3 Watcher-Specific Functions and Data Members
1172		1863
1173	=over 4	1864	=over 4
1174		1865
1175	=item ev_child_init (ev_child *, callback, int pid)	1866	=item ev_child_init (ev_child *, callback, int pid, int trace)
1176		1867
1177	=item ev_child_set (ev_child *, int pid)	1868	=item ev_child_set (ev_child *, int pid, int trace)
1178		1869
1179	Configures the watcher to wait for status changes of process C<pid> (or	1870	Configures the watcher to wait for status changes of process C<pid> (or
1180	I<any> process if C<pid> is specified as C<0>). The callback can look	1871	I<any> process if C<pid> is specified as C<0>). The callback can look
1181	at the C<rstatus> member of the C<ev_child> watcher structure to see	1872	at the C<rstatus> member of the C<ev_child> watcher structure to see
1182	the status word (use the macros from C<sys/wait.h> and see your systems	1873	the status word (use the macros from C<sys/wait.h> and see your systems
1183	C<waitpid> documentation). The C<rpid> member contains the pid of the	1874	C<waitpid> documentation). The C<rpid> member contains the pid of the
1184	process causing the status change.	1875	process causing the status change. C<trace> must be either C<0> (only
		1876	activate the watcher when the process terminates) or C<1> (additionally
		1877	activate the watcher when the process is stopped or continued).
1185		1878
1186	=item int pid [read-only]	1879	=item int pid [read-only]
1187		1880
1188	The process id this watcher watches out for, or C<0>, meaning any process id.	1881	The process id this watcher watches out for, or C<0>, meaning any process id.
1189		1882
…		…
1196	The process exit/trace status caused by C<rpid> (see your systems	1889	The process exit/trace status caused by C<rpid> (see your systems
1197	C<waitpid> and C<sys/wait.h> documentation for details).	1890	C<waitpid> and C<sys/wait.h> documentation for details).
1198		1891
1199	=back	1892	=back
1200		1893
1201	Example: Try to exit cleanly on SIGINT and SIGTERM.	1894	=head3 Examples
1202		1895
		1896	Example: C<fork()> a new process and install a child handler to wait for
		1897	its completion.
		1898
		1899	ev_child cw;
		1900
1203	static void	1901	static void
1204	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	1902	child_cb (EV_P_ ev_child *w, int revents)
1205	{	1903	{
1206	ev_unloop (loop, EVUNLOOP_ALL);	1904	ev_child_stop (EV_A_ w);
		1905	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1207	}	1906	}
1208		1907
1209	struct ev_signal signal_watcher;	1908	pid_t pid = fork ();
1210	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	1909
1211	ev_signal_start (loop, &sigint_cb);	1910	if (pid < 0)
		1911	// error
		1912	else if (pid == 0)
		1913	{
		1914	// the forked child executes here
		1915	exit (1);
		1916	}
		1917	else
		1918	{
		1919	ev_child_init (&cw, child_cb, pid, 0);
		1920	ev_child_start (EV_DEFAULT_ &cw);
		1921	}
1212		1922
1213		1923
1214	=head2 C<ev_stat> - did the file attributes just change?	1924	=head2 C<ev_stat> - did the file attributes just change?
1215		1925
1216	This watches a filesystem path for attribute changes. That is, it calls	1926	This watches a file system path for attribute changes. That is, it calls
1217	C<stat> regularly (or when the OS says it changed) and sees if it changed	1927	C<stat> on that path in regular intervals (or when the OS says it changed)
1218	compared to the last time, invoking the callback if it did.	1928	and sees if it changed compared to the last time, invoking the callback if
		1929	it did.
1219		1930
1220	The path does not need to exist: changing from "path exists" to "path does	1931	The path does not need to exist: changing from "path exists" to "path does
1221	not exist" is a status change like any other. The condition "path does	1932	not exist" is a status change like any other. The condition "path does
1222	not exist" is signified by the C<st_nlink> field being zero (which is	1933	not exist" is signified by the C<st_nlink> field being zero (which is
1223	otherwise always forced to be at least one) and all the other fields of	1934	otherwise always forced to be at least one) and all the other fields of
1224	the stat buffer having unspecified contents.	1935	the stat buffer having unspecified contents.
1225		1936
1226	Since there is no standard to do this, the portable implementation simply	1937	The path I<must not> end in a slash or contain special components such as
1227	calls C<stat (2)> regulalry on the path to see if it changed somehow. You	1938	C<.> or C<..>. The path I<should> be absolute: If it is relative and
1228	can specify a recommended polling interval for this case. If you specify	1939	your working directory changes, then the behaviour is undefined.
1229	a polling interval of C<0> (highly recommended!) then a I<suitable,	1940
1230	unspecified default> value will be used (which you can expect to be around	1941	Since there is no portable change notification interface available, the
1231	five seconds, although this might change dynamically). Libev will also	1942	portable implementation simply calls C<stat(2)> regularly on the path
1232	impose a minimum interval which is currently around C<0.1>, but thats	1943	to see if it changed somehow. You can specify a recommended polling
1233	usually overkill.	1944	interval for this case. If you specify a polling interval of C<0> (highly
		1945	recommended!) then a I<suitable, unspecified default> value will be used
		1946	(which you can expect to be around five seconds, although this might
		1947	change dynamically). Libev will also impose a minimum interval which is
		1948	currently around C<0.1>, but that's usually overkill.
1234		1949
1235	This watcher type is not meant for massive numbers of stat watchers,	1950	This watcher type is not meant for massive numbers of stat watchers,
1236	as even with OS-supported change notifications, this can be	1951	as even with OS-supported change notifications, this can be
1237	resource-intensive.	1952	resource-intensive.
1238		1953
1239	At the time of this writing, no specific OS backends are implemented, but	1954	At the time of this writing, the only OS-specific interface implemented
1240	if demand increases, at least a kqueue and inotify backend will be added.	1955	is the Linux inotify interface (implementing kqueue support is left as
		1956	an exercise for the reader. Note, however, that the author sees no way
		1957	of implementing C<ev_stat> semantics with kqueue).
		1958
		1959	=head3 ABI Issues (Largefile Support)
		1960
		1961	Libev by default (unless the user overrides this) uses the default
		1962	compilation environment, which means that on systems with large file
		1963	support disabled by default, you get the 32 bit version of the stat
		1964	structure. When using the library from programs that change the ABI to
		1965	use 64 bit file offsets the programs will fail. In that case you have to
		1966	compile libev with the same flags to get binary compatibility. This is
		1967	obviously the case with any flags that change the ABI, but the problem is
		1968	most noticeably displayed with ev_stat and large file support.
		1969
		1970	The solution for this is to lobby your distribution maker to make large
		1971	file interfaces available by default (as e.g. FreeBSD does) and not
		1972	optional. Libev cannot simply switch on large file support because it has
		1973	to exchange stat structures with application programs compiled using the
		1974	default compilation environment.
		1975
		1976	=head3 Inotify and Kqueue
		1977
		1978	When C<inotify (7)> support has been compiled into libev (generally
		1979	only available with Linux 2.6.25 or above due to bugs in earlier
		1980	implementations) and present at runtime, it will be used to speed up
		1981	change detection where possible. The inotify descriptor will be created
		1982	lazily when the first C<ev_stat> watcher is being started.
		1983
		1984	Inotify presence does not change the semantics of C<ev_stat> watchers
		1985	except that changes might be detected earlier, and in some cases, to avoid
		1986	making regular C<stat> calls. Even in the presence of inotify support
		1987	there are many cases where libev has to resort to regular C<stat> polling,
		1988	but as long as the path exists, libev usually gets away without polling.
		1989
		1990	There is no support for kqueue, as apparently it cannot be used to
		1991	implement this functionality, due to the requirement of having a file
		1992	descriptor open on the object at all times, and detecting renames, unlinks
		1993	etc. is difficult.
		1994
		1995	=head3 The special problem of stat time resolution
		1996
		1997	The C<stat ()> system call only supports full-second resolution portably,
		1998	and even on systems where the resolution is higher, most file systems
		1999	still only support whole seconds.
		2000
		2001	That means that, if the time is the only thing that changes, you can
		2002	easily miss updates: on the first update, C<ev_stat> detects a change and
		2003	calls your callback, which does something. When there is another update
		2004	within the same second, C<ev_stat> will be unable to detect unless the
		2005	stat data does change in other ways (e.g. file size).
		2006
		2007	The solution to this is to delay acting on a change for slightly more
		2008	than a second (or till slightly after the next full second boundary), using
		2009	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
		2010	ev_timer_again (loop, w)>).
		2011
		2012	The C<.02> offset is added to work around small timing inconsistencies
		2013	of some operating systems (where the second counter of the current time
		2014	might be be delayed. One such system is the Linux kernel, where a call to
		2015	C<gettimeofday> might return a timestamp with a full second later than
		2016	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		2017	update file times then there will be a small window where the kernel uses
		2018	the previous second to update file times but libev might already execute
		2019	the timer callback).
		2020
		2021	=head3 Watcher-Specific Functions and Data Members
1241		2022
1242	=over 4	2023	=over 4
1243		2024
1244	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)	2025	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1245		2026
…		…
1249	C<path>. The C<interval> is a hint on how quickly a change is expected to	2030	C<path>. The C<interval> is a hint on how quickly a change is expected to
1250	be detected and should normally be specified as C<0> to let libev choose	2031	be detected and should normally be specified as C<0> to let libev choose
1251	a suitable value. The memory pointed to by C<path> must point to the same	2032	a suitable value. The memory pointed to by C<path> must point to the same
1252	path for as long as the watcher is active.	2033	path for as long as the watcher is active.
1253		2034
1254	The callback will be receive C<EV_STAT> when a change was detected,	2035	The callback will receive an C<EV_STAT> event when a change was detected,
1255	relative to the attributes at the time the watcher was started (or the	2036	relative to the attributes at the time the watcher was started (or the
1256	last change was detected).	2037	last change was detected).
1257		2038
1258	=item ev_stat_stat (ev_stat *)	2039	=item ev_stat_stat (loop, ev_stat *)
1259		2040
1260	Updates the stat buffer immediately with new values. If you change the	2041	Updates the stat buffer immediately with new values. If you change the
1261	watched path in your callback, you could call this fucntion to avoid	2042	watched path in your callback, you could call this function to avoid
1262	detecting this change (while introducing a race condition). Can also be	2043	detecting this change (while introducing a race condition if you are not
1263	useful simply to find out the new values.	2044	the only one changing the path). Can also be useful simply to find out the
		2045	new values.
1264		2046
1265	=item ev_statdata attr [read-only]	2047	=item ev_statdata attr [read-only]
1266		2048
1267	The most-recently detected attributes of the file. Although the type is of	2049	The most-recently detected attributes of the file. Although the type is
1268	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types	2050	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1269	suitable for your system. If the C<st_nlink> member is C<0>, then there	2051	suitable for your system, but you can only rely on the POSIX-standardised
		2052	members to be present. If the C<st_nlink> member is C<0>, then there was
1270	was some error while C<stat>ing the file.	2053	some error while C<stat>ing the file.
1271		2054
1272	=item ev_statdata prev [read-only]	2055	=item ev_statdata prev [read-only]
1273		2056
1274	The previous attributes of the file. The callback gets invoked whenever	2057	The previous attributes of the file. The callback gets invoked whenever
1275	C<prev> != C<attr>.	2058	C<prev> != C<attr>, or, more precisely, one or more of these members
		2059	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		2060	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
1276		2061
1277	=item ev_tstamp interval [read-only]	2062	=item ev_tstamp interval [read-only]
1278		2063
1279	The specified interval.	2064	The specified interval.
1280		2065
1281	=item const char *path [read-only]	2066	=item const char *path [read-only]
1282		2067
1283	The filesystem path that is being watched.	2068	The file system path that is being watched.
1284		2069
1285	=back	2070	=back
1286		2071
		2072	=head3 Examples
		2073
1287	Example: Watch C</etc/passwd> for attribute changes.	2074	Example: Watch C</etc/passwd> for attribute changes.
1288		2075
1289	static void	2076	static void
1290	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2077	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1291	{	2078	{
1292	/* /etc/passwd changed in some way */	2079	/* /etc/passwd changed in some way */
1293	if (w->attr.st_nlink)	2080	if (w->attr.st_nlink)
1294	{	2081	{
1295	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2082	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1296	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2083	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1297	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2084	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1298	}	2085	}
1299	else	2086	else
1300	/* you shalt not abuse printf for puts */	2087	/* you shalt not abuse printf for puts */
1301	puts ("wow, /etc/passwd is not there, expect problems. "	2088	puts ("wow, /etc/passwd is not there, expect problems. "
1302	"if this is windows, they already arrived\n");	2089	"if this is windows, they already arrived\n");
1303	}	2090	}
1304		2091
1305	...	2092	...
1306	ev_stat passwd;	2093	ev_stat passwd;
1307		2094
1308	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	2095	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1309	ev_stat_start (loop, &passwd);	2096	ev_stat_start (loop, &passwd);
		2097
		2098	Example: Like above, but additionally use a one-second delay so we do not
		2099	miss updates (however, frequent updates will delay processing, too, so
		2100	one might do the work both on C<ev_stat> callback invocation I<and> on
		2101	C<ev_timer> callback invocation).
		2102
		2103	static ev_stat passwd;
		2104	static ev_timer timer;
		2105
		2106	static void
		2107	timer_cb (EV_P_ ev_timer *w, int revents)
		2108	{
		2109	ev_timer_stop (EV_A_ w);
		2110
		2111	/* now it's one second after the most recent passwd change */
		2112	}
		2113
		2114	static void
		2115	stat_cb (EV_P_ ev_stat *w, int revents)
		2116	{
		2117	/* reset the one-second timer */
		2118	ev_timer_again (EV_A_ &timer);
		2119	}
		2120
		2121	...
		2122	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		2123	ev_stat_start (loop, &passwd);
		2124	ev_timer_init (&timer, timer_cb, 0., 1.02);
1310		2125
1311		2126
1312	=head2 C<ev_idle> - when you've got nothing better to do...	2127	=head2 C<ev_idle> - when you've got nothing better to do...
1313		2128
1314	Idle watchers trigger events when there are no other events are pending	2129	Idle watchers trigger events when no other events of the same or higher
1315	(prepare, check and other idle watchers do not count). That is, as long	2130	priority are pending (prepare, check and other idle watchers do not count
1316	as your process is busy handling sockets or timeouts (or even signals,	2131	as receiving "events").
1317	imagine) it will not be triggered. But when your process is idle all idle	2132
1318	watchers are being called again and again, once per event loop iteration -	2133	That is, as long as your process is busy handling sockets or timeouts
		2134	(or even signals, imagine) of the same or higher priority it will not be
		2135	triggered. But when your process is idle (or only lower-priority watchers
		2136	are pending), the idle watchers are being called once per event loop
1319	until stopped, that is, or your process receives more events and becomes	2137	iteration - until stopped, that is, or your process receives more events
1320	busy.	2138	and becomes busy again with higher priority stuff.
1321		2139
1322	The most noteworthy effect is that as long as any idle watchers are	2140	The most noteworthy effect is that as long as any idle watchers are
1323	active, the process will not block when waiting for new events.	2141	active, the process will not block when waiting for new events.
1324		2142
1325	Apart from keeping your process non-blocking (which is a useful	2143	Apart from keeping your process non-blocking (which is a useful
1326	effect on its own sometimes), idle watchers are a good place to do	2144	effect on its own sometimes), idle watchers are a good place to do
1327	"pseudo-background processing", or delay processing stuff to after the	2145	"pseudo-background processing", or delay processing stuff to after the
1328	event loop has handled all outstanding events.	2146	event loop has handled all outstanding events.
1329		2147
		2148	=head3 Watcher-Specific Functions and Data Members
		2149
1330	=over 4	2150	=over 4
1331		2151
1332	=item ev_idle_init (ev_signal *, callback)	2152	=item ev_idle_init (ev_signal *, callback)
1333		2153
1334	Initialises and configures the idle watcher - it has no parameters of any	2154	Initialises and configures the idle watcher - it has no parameters of any
1335	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	2155	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1336	believe me.	2156	believe me.
1337		2157
1338	=back	2158	=back
1339		2159
		2160	=head3 Examples
		2161
1340	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	2162	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1341	callback, free it. Also, use no error checking, as usual.	2163	callback, free it. Also, use no error checking, as usual.
1342		2164
1343	static void	2165	static void
1344	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2166	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1345	{	2167	{
1346	free (w);	2168	free (w);
1347	// now do something you wanted to do when the program has	2169	// now do something you wanted to do when the program has
1348	// no longer asnything immediate to do.	2170	// no longer anything immediate to do.
1349	}	2171	}
1350		2172
1351	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2173	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1352	ev_idle_init (idle_watcher, idle_cb);	2174	ev_idle_init (idle_watcher, idle_cb);
1353	ev_idle_start (loop, idle_cb);	2175	ev_idle_start (loop, idle_cb);
1354		2176
1355		2177
1356	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	2178	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1357		2179
1358	Prepare and check watchers are usually (but not always) used in tandem:	2180	Prepare and check watchers are usually (but not always) used in pairs:
1359	prepare watchers get invoked before the process blocks and check watchers	2181	prepare watchers get invoked before the process blocks and check watchers
1360	afterwards.	2182	afterwards.
1361		2183
1362	You I<must not> call C<ev_loop> or similar functions that enter	2184	You I<must not> call C<ev_loop> or similar functions that enter
1363	the current event loop from either C<ev_prepare> or C<ev_check>	2185	the current event loop from either C<ev_prepare> or C<ev_check>
…		…
1366	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	2188	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
1367	C<ev_check> so if you have one watcher of each kind they will always be	2189	C<ev_check> so if you have one watcher of each kind they will always be
1368	called in pairs bracketing the blocking call.	2190	called in pairs bracketing the blocking call.
1369		2191
1370	Their main purpose is to integrate other event mechanisms into libev and	2192	Their main purpose is to integrate other event mechanisms into libev and
1371	their use is somewhat advanced. This could be used, for example, to track	2193	their use is somewhat advanced. They could be used, for example, to track
1372	variable changes, implement your own watchers, integrate net-snmp or a	2194	variable changes, implement your own watchers, integrate net-snmp or a
1373	coroutine library and lots more. They are also occasionally useful if	2195	coroutine library and lots more. They are also occasionally useful if
1374	you cache some data and want to flush it before blocking (for example,	2196	you cache some data and want to flush it before blocking (for example,
1375	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	2197	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1376	watcher).	2198	watcher).
1377		2199
1378	This is done by examining in each prepare call which file descriptors need	2200	This is done by examining in each prepare call which file descriptors
1379	to be watched by the other library, registering C<ev_io> watchers for	2201	need to be watched by the other library, registering C<ev_io> watchers
1380	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2202	for them and starting an C<ev_timer> watcher for any timeouts (many
1381	provide just this functionality). Then, in the check watcher you check for	2203	libraries provide exactly this functionality). Then, in the check watcher,
1382	any events that occured (by checking the pending status of all watchers	2204	you check for any events that occurred (by checking the pending status
1383	and stopping them) and call back into the library. The I/O and timer	2205	of all watchers and stopping them) and call back into the library. The
1384	callbacks will never actually be called (but must be valid nevertheless,	2206	I/O and timer callbacks will never actually be called (but must be valid
1385	because you never know, you know?).	2207	nevertheless, because you never know, you know?).
1386		2208
1387	As another example, the Perl Coro module uses these hooks to integrate	2209	As another example, the Perl Coro module uses these hooks to integrate
1388	coroutines into libev programs, by yielding to other active coroutines	2210	coroutines into libev programs, by yielding to other active coroutines
1389	during each prepare and only letting the process block if no coroutines	2211	during each prepare and only letting the process block if no coroutines
1390	are ready to run (it's actually more complicated: it only runs coroutines	2212	are ready to run (it's actually more complicated: it only runs coroutines
1391	with priority higher than or equal to the event loop and one coroutine	2213	with priority higher than or equal to the event loop and one coroutine
1392	of lower priority, but only once, using idle watchers to keep the event	2214	of lower priority, but only once, using idle watchers to keep the event
1393	loop from blocking if lower-priority coroutines are active, thus mapping	2215	loop from blocking if lower-priority coroutines are active, thus mapping
1394	low-priority coroutines to idle/background tasks).	2216	low-priority coroutines to idle/background tasks).
1395		2217
		2218	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		2219	priority, to ensure that they are being run before any other watchers
		2220	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2221
		2222	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
		2223	activate ("feed") events into libev. While libev fully supports this, they
		2224	might get executed before other C<ev_check> watchers did their job. As
		2225	C<ev_check> watchers are often used to embed other (non-libev) event
		2226	loops those other event loops might be in an unusable state until their
		2227	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		2228	others).
		2229
		2230	=head3 Watcher-Specific Functions and Data Members
		2231
1396	=over 4	2232	=over 4
1397		2233
1398	=item ev_prepare_init (ev_prepare *, callback)	2234	=item ev_prepare_init (ev_prepare *, callback)
1399		2235
1400	=item ev_check_init (ev_check *, callback)	2236	=item ev_check_init (ev_check *, callback)
1401		2237
1402	Initialises and configures the prepare or check watcher - they have no	2238	Initialises and configures the prepare or check watcher - they have no
1403	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2239	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1404	macros, but using them is utterly, utterly and completely pointless.	2240	macros, but using them is utterly, utterly, utterly and completely
		2241	pointless.
1405		2242
1406	=back	2243	=back
1407		2244
1408	Example: To include a library such as adns, you would add IO watchers	2245	=head3 Examples
1409	and a timeout watcher in a prepare handler, as required by libadns, and	2246
		2247	There are a number of principal ways to embed other event loops or modules
		2248	into libev. Here are some ideas on how to include libadns into libev
		2249	(there is a Perl module named C<EV::ADNS> that does this, which you could
		2250	use as a working example. Another Perl module named C<EV::Glib> embeds a
		2251	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
		2252	Glib event loop).
		2253
		2254	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1410	in a check watcher, destroy them and call into libadns. What follows is	2255	and in a check watcher, destroy them and call into libadns. What follows
1411	pseudo-code only of course:	2256	is pseudo-code only of course. This requires you to either use a low
		2257	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		2258	the callbacks for the IO/timeout watchers might not have been called yet.
1412		2259
1413	static ev_io iow [nfd];	2260	static ev_io iow [nfd];
1414	static ev_timer tw;	2261	static ev_timer tw;
1415		2262
1416	static void	2263	static void
1417	io_cb (ev_loop *loop, ev_io *w, int revents)	2264	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1418	{	2265	{
1419	// set the relevant poll flags
1420	// could also call adns_processreadable etc. here
1421	struct pollfd *fd = (struct pollfd *)w->data;
1422	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1423	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1424	}	2266	}
1425		2267
1426	// create io watchers for each fd and a timer before blocking	2268	// create io watchers for each fd and a timer before blocking
1427	static void	2269	static void
1428	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	2270	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1429	{	2271	{
1430	int timeout = 3600000;truct pollfd fds [nfd];	2272	int timeout = 3600000;
		2273	struct pollfd fds [nfd];
1431	// actual code will need to loop here and realloc etc.	2274	// actual code will need to loop here and realloc etc.
1432	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	2275	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1433		2276
1434	/* the callback is illegal, but won't be called as we stop during check */	2277	/* the callback is illegal, but won't be called as we stop during check */
1435	ev_timer_init (&tw, 0, timeout * 1e-3);	2278	ev_timer_init (&tw, 0, timeout * 1e-3);
1436	ev_timer_start (loop, &tw);	2279	ev_timer_start (loop, &tw);
1437		2280
1438	// create on ev_io per pollfd	2281	// create one ev_io per pollfd
1439	for (int i = 0; i < nfd; ++i)	2282	for (int i = 0; i < nfd; ++i)
1440	{	2283	{
1441	ev_io_init (iow + i, io_cb, fds [i].fd,	2284	ev_io_init (iow + i, io_cb, fds [i].fd,
1442	((fds [i].events & POLLIN ? EV_READ : 0)	2285	((fds [i].events & POLLIN ? EV_READ : 0)
1443	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	2286	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1444		2287
1445	fds [i].revents = 0;	2288	fds [i].revents = 0;
1446	iow [i].data = fds + i;
1447	ev_io_start (loop, iow + i);	2289	ev_io_start (loop, iow + i);
1448	}	2290	}
1449	}	2291	}
1450		2292
1451	// stop all watchers after blocking	2293	// stop all watchers after blocking
1452	static void	2294	static void
1453	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	2295	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1454	{	2296	{
1455	ev_timer_stop (loop, &tw);	2297	ev_timer_stop (loop, &tw);
1456		2298
1457	for (int i = 0; i < nfd; ++i)	2299	for (int i = 0; i < nfd; ++i)
		2300	{
		2301	// set the relevant poll flags
		2302	// could also call adns_processreadable etc. here
		2303	struct pollfd *fd = fds + i;
		2304	int revents = ev_clear_pending (iow + i);
		2305	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		2306	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		2307
		2308	// now stop the watcher
1458	ev_io_stop (loop, iow + i);	2309	ev_io_stop (loop, iow + i);
		2310	}
1459		2311
1460	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	2312	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1461	}	2313	}
		2314
		2315	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		2316	in the prepare watcher and would dispose of the check watcher.
		2317
		2318	Method 3: If the module to be embedded supports explicit event
		2319	notification (libadns does), you can also make use of the actual watcher
		2320	callbacks, and only destroy/create the watchers in the prepare watcher.
		2321
		2322	static void
		2323	timer_cb (EV_P_ ev_timer *w, int revents)
		2324	{
		2325	adns_state ads = (adns_state)w->data;
		2326	update_now (EV_A);
		2327
		2328	adns_processtimeouts (ads, &tv_now);
		2329	}
		2330
		2331	static void
		2332	io_cb (EV_P_ ev_io *w, int revents)
		2333	{
		2334	adns_state ads = (adns_state)w->data;
		2335	update_now (EV_A);
		2336
		2337	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		2338	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		2339	}
		2340
		2341	// do not ever call adns_afterpoll
		2342
		2343	Method 4: Do not use a prepare or check watcher because the module you
		2344	want to embed is not flexible enough to support it. Instead, you can
		2345	override their poll function. The drawback with this solution is that the
		2346	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
		2347	this approach, effectively embedding EV as a client into the horrible
		2348	libglib event loop.
		2349
		2350	static gint
		2351	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		2352	{
		2353	int got_events = 0;
		2354
		2355	for (n = 0; n < nfds; ++n)
		2356	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		2357
		2358	if (timeout >= 0)
		2359	// create/start timer
		2360
		2361	// poll
		2362	ev_loop (EV_A_ 0);
		2363
		2364	// stop timer again
		2365	if (timeout >= 0)
		2366	ev_timer_stop (EV_A_ &to);
		2367
		2368	// stop io watchers again - their callbacks should have set
		2369	for (n = 0; n < nfds; ++n)
		2370	ev_io_stop (EV_A_ iow [n]);
		2371
		2372	return got_events;
		2373	}
1462		2374
1463		2375
1464	=head2 C<ev_embed> - when one backend isn't enough...	2376	=head2 C<ev_embed> - when one backend isn't enough...
1465		2377
1466	This is a rather advanced watcher type that lets you embed one event loop	2378	This is a rather advanced watcher type that lets you embed one event loop
…		…
1472	prioritise I/O.	2384	prioritise I/O.
1473		2385
1474	As an example for a bug workaround, the kqueue backend might only support	2386	As an example for a bug workaround, the kqueue backend might only support
1475	sockets on some platform, so it is unusable as generic backend, but you	2387	sockets on some platform, so it is unusable as generic backend, but you
1476	still want to make use of it because you have many sockets and it scales	2388	still want to make use of it because you have many sockets and it scales
1477	so nicely. In this case, you would create a kqueue-based loop and embed it	2389	so nicely. In this case, you would create a kqueue-based loop and embed
1478	into your default loop (which might use e.g. poll). Overall operation will	2390	it into your default loop (which might use e.g. poll). Overall operation
1479	be a bit slower because first libev has to poll and then call kevent, but	2391	will be a bit slower because first libev has to call C<poll> and then
1480	at least you can use both at what they are best.	2392	C<kevent>, but at least you can use both mechanisms for what they are
		2393	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
1481		2394
1482	As for prioritising I/O: rarely you have the case where some fds have	2395	As for prioritising I/O: under rare circumstances you have the case where
1483	to be watched and handled very quickly (with low latency), and even	2396	some fds have to be watched and handled very quickly (with low latency),
1484	priorities and idle watchers might have too much overhead. In this case	2397	and even priorities and idle watchers might have too much overhead. In
1485	you would put all the high priority stuff in one loop and all the rest in	2398	this case you would put all the high priority stuff in one loop and all
1486	a second one, and embed the second one in the first.	2399	the rest in a second one, and embed the second one in the first.
1487		2400
1488	As long as the watcher is active, the callback will be invoked every time	2401	As long as the watcher is active, the callback will be invoked every time
1489	there might be events pending in the embedded loop. The callback must then	2402	there might be events pending in the embedded loop. The callback must then
1490	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	2403	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke
1491	their callbacks (you could also start an idle watcher to give the embedded	2404	their callbacks (you could also start an idle watcher to give the embedded
…		…
1499	interested in that.	2412	interested in that.
1500		2413
1501	Also, there have not currently been made special provisions for forking:	2414	Also, there have not currently been made special provisions for forking:
1502	when you fork, you not only have to call C<ev_loop_fork> on both loops,	2415	when you fork, you not only have to call C<ev_loop_fork> on both loops,
1503	but you will also have to stop and restart any C<ev_embed> watchers	2416	but you will also have to stop and restart any C<ev_embed> watchers
1504	yourself.	2417	yourself - but you can use a fork watcher to handle this automatically,
		2418	and future versions of libev might do just that.
1505		2419
1506	Unfortunately, not all backends are embeddable, only the ones returned by	2420	Unfortunately, not all backends are embeddable: only the ones returned by
1507	C<ev_embeddable_backends> are, which, unfortunately, does not include any	2421	C<ev_embeddable_backends> are, which, unfortunately, does not include any
1508	portable one.	2422	portable one.
1509		2423
1510	So when you want to use this feature you will always have to be prepared	2424	So when you want to use this feature you will always have to be prepared
1511	that you cannot get an embeddable loop. The recommended way to get around	2425	that you cannot get an embeddable loop. The recommended way to get around
1512	this is to have a separate variables for your embeddable loop, try to	2426	this is to have a separate variables for your embeddable loop, try to
1513	create it, and if that fails, use the normal loop for everything:	2427	create it, and if that fails, use the normal loop for everything.
1514		2428
1515	struct ev_loop *loop_hi = ev_default_init (0);	2429	=head3 C<ev_embed> and fork
1516	struct ev_loop *loop_lo = 0;
1517	struct ev_embed embed;
1518
1519	// see if there is a chance of getting one that works
1520	// (remember that a flags value of 0 means autodetection)
1521	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1522	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1523	: 0;
1524		2430
1525	// if we got one, then embed it, otherwise default to loop_hi	2431	While the C<ev_embed> watcher is running, forks in the embedding loop will
1526	if (loop_lo)	2432	automatically be applied to the embedded loop as well, so no special
1527	{	2433	fork handling is required in that case. When the watcher is not running,
1528	ev_embed_init (&embed, 0, loop_lo);	2434	however, it is still the task of the libev user to call C<ev_loop_fork ()>
1529	ev_embed_start (loop_hi, &embed);	2435	as applicable.
1530	}	2436
1531	else	2437	=head3 Watcher-Specific Functions and Data Members
1532	loop_lo = loop_hi;
1533		2438
1534	=over 4	2439	=over 4
1535		2440
1536	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	2441	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1537		2442
…		…
1539		2444
1540	Configures the watcher to embed the given loop, which must be	2445	Configures the watcher to embed the given loop, which must be
1541	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	2446	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1542	invoked automatically, otherwise it is the responsibility of the callback	2447	invoked automatically, otherwise it is the responsibility of the callback
1543	to invoke it (it will continue to be called until the sweep has been done,	2448	to invoke it (it will continue to be called until the sweep has been done,
1544	if you do not want thta, you need to temporarily stop the embed watcher).	2449	if you do not want that, you need to temporarily stop the embed watcher).
1545		2450
1546	=item ev_embed_sweep (loop, ev_embed *)	2451	=item ev_embed_sweep (loop, ev_embed *)
1547		2452
1548	Make a single, non-blocking sweep over the embedded loop. This works	2453	Make a single, non-blocking sweep over the embedded loop. This works
1549	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	2454	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1550	apropriate way for embedded loops.	2455	appropriate way for embedded loops.
1551		2456
1552	=item struct ev_loop *loop [read-only]	2457	=item struct ev_loop *other [read-only]
1553		2458
1554	The embedded event loop.	2459	The embedded event loop.
1555		2460
1556	=back	2461	=back
		2462
		2463	=head3 Examples
		2464
		2465	Example: Try to get an embeddable event loop and embed it into the default
		2466	event loop. If that is not possible, use the default loop. The default
		2467	loop is stored in C<loop_hi>, while the embeddable loop is stored in
		2468	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
		2469	used).
		2470
		2471	struct ev_loop *loop_hi = ev_default_init (0);
		2472	struct ev_loop *loop_lo = 0;
		2473	ev_embed embed;
		2474
		2475	// see if there is a chance of getting one that works
		2476	// (remember that a flags value of 0 means autodetection)
		2477	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
		2478	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
		2479	: 0;
		2480
		2481	// if we got one, then embed it, otherwise default to loop_hi
		2482	if (loop_lo)
		2483	{
		2484	ev_embed_init (&embed, 0, loop_lo);
		2485	ev_embed_start (loop_hi, &embed);
		2486	}
		2487	else
		2488	loop_lo = loop_hi;
		2489
		2490	Example: Check if kqueue is available but not recommended and create
		2491	a kqueue backend for use with sockets (which usually work with any
		2492	kqueue implementation). Store the kqueue/socket-only event loop in
		2493	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
		2494
		2495	struct ev_loop *loop = ev_default_init (0);
		2496	struct ev_loop *loop_socket = 0;
		2497	ev_embed embed;
		2498
		2499	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2500	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2501	{
		2502	ev_embed_init (&embed, 0, loop_socket);
		2503	ev_embed_start (loop, &embed);
		2504	}
		2505
		2506	if (!loop_socket)
		2507	loop_socket = loop;
		2508
		2509	// now use loop_socket for all sockets, and loop for everything else
1557		2510
1558		2511
1559	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	2512	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1560		2513
1561	Fork watchers are called when a C<fork ()> was detected (usually because	2514	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
1564	event loop blocks next and before C<ev_check> watchers are being called,	2517	event loop blocks next and before C<ev_check> watchers are being called,
1565	and only in the child after the fork. If whoever good citizen calling	2518	and only in the child after the fork. If whoever good citizen calling
1566	C<ev_default_fork> cheats and calls it in the wrong process, the fork	2519	C<ev_default_fork> cheats and calls it in the wrong process, the fork
1567	handlers will be invoked, too, of course.	2520	handlers will be invoked, too, of course.
1568		2521
		2522	=head3 Watcher-Specific Functions and Data Members
		2523
1569	=over 4	2524	=over 4
1570		2525
1571	=item ev_fork_init (ev_signal *, callback)	2526	=item ev_fork_init (ev_signal *, callback)
1572		2527
1573	Initialises and configures the fork watcher - it has no parameters of any	2528	Initialises and configures the fork watcher - it has no parameters of any
…		…
1575	believe me.	2530	believe me.
1576		2531
1577	=back	2532	=back
1578		2533
1579		2534
		2535	=head2 C<ev_async> - how to wake up another event loop
		2536
		2537	In general, you cannot use an C<ev_loop> from multiple threads or other
		2538	asynchronous sources such as signal handlers (as opposed to multiple event
		2539	loops - those are of course safe to use in different threads).
		2540
		2541	Sometimes, however, you need to wake up another event loop you do not
		2542	control, for example because it belongs to another thread. This is what
		2543	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2544	can signal it by calling C<ev_async_send>, which is thread- and signal
		2545	safe.
		2546
		2547	This functionality is very similar to C<ev_signal> watchers, as signals,
		2548	too, are asynchronous in nature, and signals, too, will be compressed
		2549	(i.e. the number of callback invocations may be less than the number of
		2550	C<ev_async_sent> calls).
		2551
		2552	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2553	just the default loop.
		2554
		2555	=head3 Queueing
		2556
		2557	C<ev_async> does not support queueing of data in any way. The reason
		2558	is that the author does not know of a simple (or any) algorithm for a
		2559	multiple-writer-single-reader queue that works in all cases and doesn't
		2560	need elaborate support such as pthreads.
		2561
		2562	That means that if you want to queue data, you have to provide your own
		2563	queue. But at least I can tell you how to implement locking around your
		2564	queue:
		2565
		2566	=over 4
		2567
		2568	=item queueing from a signal handler context
		2569
		2570	To implement race-free queueing, you simply add to the queue in the signal
		2571	handler but you block the signal handler in the watcher callback. Here is
		2572	an example that does that for some fictitious SIGUSR1 handler:
		2573
		2574	static ev_async mysig;
		2575
		2576	static void
		2577	sigusr1_handler (void)
		2578	{
		2579	sometype data;
		2580
		2581	// no locking etc.
		2582	queue_put (data);
		2583	ev_async_send (EV_DEFAULT_ &mysig);
		2584	}
		2585
		2586	static void
		2587	mysig_cb (EV_P_ ev_async *w, int revents)
		2588	{
		2589	sometype data;
		2590	sigset_t block, prev;
		2591
		2592	sigemptyset (&block);
		2593	sigaddset (&block, SIGUSR1);
		2594	sigprocmask (SIG_BLOCK, &block, &prev);
		2595
		2596	while (queue_get (&data))
		2597	process (data);
		2598
		2599	if (sigismember (&prev, SIGUSR1)
		2600	sigprocmask (SIG_UNBLOCK, &block, 0);
		2601	}
		2602
		2603	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2604	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2605	either...).
		2606
		2607	=item queueing from a thread context
		2608
		2609	The strategy for threads is different, as you cannot (easily) block
		2610	threads but you can easily preempt them, so to queue safely you need to
		2611	employ a traditional mutex lock, such as in this pthread example:
		2612
		2613	static ev_async mysig;
		2614	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2615
		2616	static void
		2617	otherthread (void)
		2618	{
		2619	// only need to lock the actual queueing operation
		2620	pthread_mutex_lock (&mymutex);
		2621	queue_put (data);
		2622	pthread_mutex_unlock (&mymutex);
		2623
		2624	ev_async_send (EV_DEFAULT_ &mysig);
		2625	}
		2626
		2627	static void
		2628	mysig_cb (EV_P_ ev_async *w, int revents)
		2629	{
		2630	pthread_mutex_lock (&mymutex);
		2631
		2632	while (queue_get (&data))
		2633	process (data);
		2634
		2635	pthread_mutex_unlock (&mymutex);
		2636	}
		2637
		2638	=back
		2639
		2640
		2641	=head3 Watcher-Specific Functions and Data Members
		2642
		2643	=over 4
		2644
		2645	=item ev_async_init (ev_async *, callback)
		2646
		2647	Initialises and configures the async watcher - it has no parameters of any
		2648	kind. There is a C<ev_async_set> macro, but using it is utterly pointless,
		2649	trust me.
		2650
		2651	=item ev_async_send (loop, ev_async *)
		2652
		2653	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2654	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2655	C<ev_feed_event>, this call is safe to do from other threads, signal or
		2656	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
		2657	section below on what exactly this means).
		2658
		2659	This call incurs the overhead of a system call only once per loop iteration,
		2660	so while the overhead might be noticeable, it doesn't apply to repeated
		2661	calls to C<ev_async_send>.
		2662
		2663	=item bool = ev_async_pending (ev_async *)
		2664
		2665	Returns a non-zero value when C<ev_async_send> has been called on the
		2666	watcher but the event has not yet been processed (or even noted) by the
		2667	event loop.
		2668
		2669	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
		2670	the loop iterates next and checks for the watcher to have become active,
		2671	it will reset the flag again. C<ev_async_pending> can be used to very
		2672	quickly check whether invoking the loop might be a good idea.
		2673
		2674	Not that this does I<not> check whether the watcher itself is pending, only
		2675	whether it has been requested to make this watcher pending.
		2676
		2677	=back
		2678
		2679
1580	=head1 OTHER FUNCTIONS	2680	=head1 OTHER FUNCTIONS
1581		2681
1582	There are some other functions of possible interest. Described. Here. Now.	2682	There are some other functions of possible interest. Described. Here. Now.
1583		2683
1584	=over 4	2684	=over 4
1585		2685
1586	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2686	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
1587		2687
1588	This function combines a simple timer and an I/O watcher, calls your	2688	This function combines a simple timer and an I/O watcher, calls your
1589	callback on whichever event happens first and automatically stop both	2689	callback on whichever event happens first and automatically stops both
1590	watchers. This is useful if you want to wait for a single event on an fd	2690	watchers. This is useful if you want to wait for a single event on an fd
1591	or timeout without having to allocate/configure/start/stop/free one or	2691	or timeout without having to allocate/configure/start/stop/free one or
1592	more watchers yourself.	2692	more watchers yourself.
1593		2693
1594	If C<fd> is less than 0, then no I/O watcher will be started and events	2694	If C<fd> is less than 0, then no I/O watcher will be started and the
1595	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	2695	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
1596	C<events> set will be craeted and started.	2696	the given C<fd> and C<events> set will be created and started.
1597		2697
1598	If C<timeout> is less than 0, then no timeout watcher will be	2698	If C<timeout> is less than 0, then no timeout watcher will be
1599	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	2699	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
1600	repeat = 0) will be started. While C<0> is a valid timeout, it is of	2700	repeat = 0) will be started. C<0> is a valid timeout.
1601	dubious value.
1602		2701
1603	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	2702	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
1604	passed an C<revents> set like normal event callbacks (a combination of	2703	passed an C<revents> set like normal event callbacks (a combination of
1605	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	2704	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
1606	value passed to C<ev_once>:	2705	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2706	a timeout and an io event at the same time - you probably should give io
		2707	events precedence.
1607		2708
		2709	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		2710
1608	static void stdin_ready (int revents, void *arg)	2711	static void stdin_ready (int revents, void *arg)
1609	{	2712	{
1610	if (revents & EV_TIMEOUT)
1611	/* doh, nothing entered */;
1612	else if (revents & EV_READ)	2713	if (revents & EV_READ)
1613	/* stdin might have data for us, joy! */;	2714	/* stdin might have data for us, joy! */;
		2715	else if (revents & EV_TIMEOUT)
		2716	/* doh, nothing entered */;
1614	}	2717	}
1615		2718
1616	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	2719	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1617		2720
1618	=item ev_feed_event (ev_loop *, watcher *, int revents)	2721	=item ev_feed_event (struct ev_loop *, watcher *, int revents)
1619		2722
1620	Feeds the given event set into the event loop, as if the specified event	2723	Feeds the given event set into the event loop, as if the specified event
1621	had happened for the specified watcher (which must be a pointer to an	2724	had happened for the specified watcher (which must be a pointer to an
1622	initialised but not necessarily started event watcher).	2725	initialised but not necessarily started event watcher).
1623		2726
1624	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	2727	=item ev_feed_fd_event (struct ev_loop *, int fd, int revents)
1625		2728
1626	Feed an event on the given fd, as if a file descriptor backend detected	2729	Feed an event on the given fd, as if a file descriptor backend detected
1627	the given events it.	2730	the given events it.
1628		2731
1629	=item ev_feed_signal_event (ev_loop *loop, int signum)	2732	=item ev_feed_signal_event (struct ev_loop *loop, int signum)
1630		2733
1631	Feed an event as if the given signal occured (C<loop> must be the default	2734	Feed an event as if the given signal occurred (C<loop> must be the default
1632	loop!).	2735	loop!).
1633		2736
1634	=back	2737	=back
1635		2738
1636		2739
…		…
1652		2755
1653	=item * Priorities are not currently supported. Initialising priorities	2756	=item * Priorities are not currently supported. Initialising priorities
1654	will fail and all watchers will have the same priority, even though there	2757	will fail and all watchers will have the same priority, even though there
1655	is an ev_pri field.	2758	is an ev_pri field.
1656		2759
		2760	=item * In libevent, the last base created gets the signals, in libev, the
		2761	first base created (== the default loop) gets the signals.
		2762
1657	=item * Other members are not supported.	2763	=item * Other members are not supported.
1658		2764
1659	=item * The libev emulation is I<not> ABI compatible to libevent, you need	2765	=item * The libev emulation is I<not> ABI compatible to libevent, you need
1660	to use the libev header file and library.	2766	to use the libev header file and library.
1661		2767
1662	=back	2768	=back
1663		2769
1664	=head1 C++ SUPPORT	2770	=head1 C++ SUPPORT
1665		2771
1666	Libev comes with some simplistic wrapper classes for C++ that mainly allow	2772	Libev comes with some simplistic wrapper classes for C++ that mainly allow
1667	you to use some convinience methods to start/stop watchers and also change	2773	you to use some convenience methods to start/stop watchers and also change
1668	the callback model to a model using method callbacks on objects.	2774	the callback model to a model using method callbacks on objects.
1669		2775
1670	To use it,	2776	To use it,
1671		2777
1672	#include <ev++.h>	2778	#include <ev++.h>
1673		2779
1674	(it is not installed by default). This automatically includes F<ev.h>	2780	This automatically includes F<ev.h> and puts all of its definitions (many
1675	and puts all of its definitions (many of them macros) into the global	2781	of them macros) into the global namespace. All C++ specific things are
1676	namespace. All C++ specific things are put into the C<ev> namespace.	2782	put into the C<ev> namespace. It should support all the same embedding
		2783	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
1677		2784
1678	It should support all the same embedding options as F<ev.h>, most notably	2785	Care has been taken to keep the overhead low. The only data member the C++
1679	C<EV_MULTIPLICITY>.	2786	classes add (compared to plain C-style watchers) is the event loop pointer
		2787	that the watcher is associated with (or no additional members at all if
		2788	you disable C<EV_MULTIPLICITY> when embedding libev).
		2789
		2790	Currently, functions, and static and non-static member functions can be
		2791	used as callbacks. Other types should be easy to add as long as they only
		2792	need one additional pointer for context. If you need support for other
		2793	types of functors please contact the author (preferably after implementing
		2794	it).
1680		2795
1681	Here is a list of things available in the C<ev> namespace:	2796	Here is a list of things available in the C<ev> namespace:
1682		2797
1683	=over 4	2798	=over 4
1684		2799
…		…
1700		2815
1701	All of those classes have these methods:	2816	All of those classes have these methods:
1702		2817
1703	=over 4	2818	=over 4
1704		2819
1705	=item ev::TYPE::TYPE (object *, object::method *)	2820	=item ev::TYPE::TYPE ()
1706		2821
1707	=item ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)	2822	=item ev::TYPE::TYPE (struct ev_loop *)
1708		2823
1709	=item ev::TYPE::~TYPE	2824	=item ev::TYPE::~TYPE
1710		2825
1711	The constructor takes a pointer to an object and a method pointer to	2826	The constructor (optionally) takes an event loop to associate the watcher
1712	the event handler callback to call in this class. The constructor calls	2827	with. If it is omitted, it will use C<EV_DEFAULT>.
1713	C<ev_init> for you, which means you have to call the C<set> method	2828
1714	before starting it. If you do not specify a loop then the constructor	2829	The constructor calls C<ev_init> for you, which means you have to call the
1715	automatically associates the default loop with this watcher.	2830	C<set> method before starting it.
		2831
		2832	It will not set a callback, however: You have to call the templated C<set>
		2833	method to set a callback before you can start the watcher.
		2834
		2835	(The reason why you have to use a method is a limitation in C++ which does
		2836	not allow explicit template arguments for constructors).
1716		2837
1717	The destructor automatically stops the watcher if it is active.	2838	The destructor automatically stops the watcher if it is active.
		2839
		2840	=item w->set<class, &class::method> (object *)
		2841
		2842	This method sets the callback method to call. The method has to have a
		2843	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		2844	first argument and the C<revents> as second. The object must be given as
		2845	parameter and is stored in the C<data> member of the watcher.
		2846
		2847	This method synthesizes efficient thunking code to call your method from
		2848	the C callback that libev requires. If your compiler can inline your
		2849	callback (i.e. it is visible to it at the place of the C<set> call and
		2850	your compiler is good :), then the method will be fully inlined into the
		2851	thunking function, making it as fast as a direct C callback.
		2852
		2853	Example: simple class declaration and watcher initialisation
		2854
		2855	struct myclass
		2856	{
		2857	void io_cb (ev::io &w, int revents) { }
		2858	}
		2859
		2860	myclass obj;
		2861	ev::io iow;
		2862	iow.set <myclass, &myclass::io_cb> (&obj);
		2863
		2864	=item w->set<function> (void *data = 0)
		2865
		2866	Also sets a callback, but uses a static method or plain function as
		2867	callback. The optional C<data> argument will be stored in the watcher's
		2868	C<data> member and is free for you to use.
		2869
		2870	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		2871
		2872	See the method-C<set> above for more details.
		2873
		2874	Example: Use a plain function as callback.
		2875
		2876	static void io_cb (ev::io &w, int revents) { }
		2877	iow.set <io_cb> ();
1718		2878
1719	=item w->set (struct ev_loop *)	2879	=item w->set (struct ev_loop *)
1720		2880
1721	Associates a different C<struct ev_loop> with this watcher. You can only	2881	Associates a different C<struct ev_loop> with this watcher. You can only
1722	do this when the watcher is inactive (and not pending either).	2882	do this when the watcher is inactive (and not pending either).
1723		2883
1724	=item w->set ([args])	2884	=item w->set ([arguments])
1725		2885
1726	Basically the same as C<ev_TYPE_set>, with the same args. Must be	2886	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
1727	called at least once. Unlike the C counterpart, an active watcher gets	2887	called at least once. Unlike the C counterpart, an active watcher gets
1728	automatically stopped and restarted.	2888	automatically stopped and restarted when reconfiguring it with this
		2889	method.
1729		2890
1730	=item w->start ()	2891	=item w->start ()
1731		2892
1732	Starts the watcher. Note that there is no C<loop> argument as the	2893	Starts the watcher. Note that there is no C<loop> argument, as the
1733	constructor already takes the loop.	2894	constructor already stores the event loop.
1734		2895
1735	=item w->stop ()	2896	=item w->stop ()
1736		2897
1737	Stops the watcher if it is active. Again, no C<loop> argument.	2898	Stops the watcher if it is active. Again, no C<loop> argument.
1738		2899
1739	=item w->again () C<ev::timer>, C<ev::periodic> only	2900	=item w->again () (C<ev::timer>, C<ev::periodic> only)
1740		2901
1741	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding	2902	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
1742	C<ev_TYPE_again> function.	2903	C<ev_TYPE_again> function.
1743		2904
1744	=item w->sweep () C<ev::embed> only	2905	=item w->sweep () (C<ev::embed> only)
1745		2906
1746	Invokes C<ev_embed_sweep>.	2907	Invokes C<ev_embed_sweep>.
1747		2908
1748	=item w->update () C<ev::stat> only	2909	=item w->update () (C<ev::stat> only)
1749		2910
1750	Invokes C<ev_stat_stat>.	2911	Invokes C<ev_stat_stat>.
1751		2912
1752	=back	2913	=back
1753		2914
1754	=back	2915	=back
1755		2916
1756	Example: Define a class with an IO and idle watcher, start one of them in	2917	Example: Define a class with an IO and idle watcher, start one of them in
1757	the constructor.	2918	the constructor.
1758		2919
1759	class myclass	2920	class myclass
1760	{	2921	{
1761	ev_io io; void io_cb (ev::io &w, int revents);	2922	ev::io io ; void io_cb (ev::io &w, int revents);
1762	ev_idle idle void idle_cb (ev::idle &w, int revents);	2923	ev::idle idle; void idle_cb (ev::idle &w, int revents);
1763		2924
1764	myclass ();	2925	myclass (int fd)
1765	}	2926	{
		2927	io .set <myclass, &myclass::io_cb > (this);
		2928	idle.set <myclass, &myclass::idle_cb> (this);
1766		2929
1767	myclass::myclass (int fd)
1768	: io (this, &myclass::io_cb),
1769	idle (this, &myclass::idle_cb)
1770	{
1771	io.start (fd, ev::READ);	2930	io.start (fd, ev::READ);
		2931	}
1772	}	2932	};
		2933
		2934
		2935	=head1 OTHER LANGUAGE BINDINGS
		2936
		2937	Libev does not offer other language bindings itself, but bindings for a
		2938	number of languages exist in the form of third-party packages. If you know
		2939	any interesting language binding in addition to the ones listed here, drop
		2940	me a note.
		2941
		2942	=over 4
		2943
		2944	=item Perl
		2945
		2946	The EV module implements the full libev API and is actually used to test
		2947	libev. EV is developed together with libev. Apart from the EV core module,
		2948	there are additional modules that implement libev-compatible interfaces
		2949	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
		2950	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		2951	and C<EV::Glib>).
		2952
		2953	It can be found and installed via CPAN, its homepage is at
		2954	L<http://software.schmorp.de/pkg/EV>.
		2955
		2956	=item Python
		2957
		2958	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		2959	seems to be quite complete and well-documented. Note, however, that the
		2960	patch they require for libev is outright dangerous as it breaks the ABI
		2961	for everybody else, and therefore, should never be applied in an installed
		2962	libev (if python requires an incompatible ABI then it needs to embed
		2963	libev).
		2964
		2965	=item Ruby
		2966
		2967	Tony Arcieri has written a ruby extension that offers access to a subset
		2968	of the libev API and adds file handle abstractions, asynchronous DNS and
		2969	more on top of it. It can be found via gem servers. Its homepage is at
		2970	L<http://rev.rubyforge.org/>.
		2971
		2972	=item D
		2973
		2974	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
		2975	be found at L<http://proj.llucax.com.ar/wiki/evd>.
		2976
		2977	=item Ocaml
		2978
		2979	Erkki Seppala has written Ocaml bindings for libev, to be found at
		2980	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
		2981
		2982	=back
1773		2983
1774		2984
1775	=head1 MACRO MAGIC	2985	=head1 MACRO MAGIC
1776		2986
1777	Libev can be compiled with a variety of options, the most fundemantal is	2987	Libev can be compiled with a variety of options, the most fundamental
1778	C<EV_MULTIPLICITY>. This option determines wether (most) functions and	2988	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
1779	callbacks have an initial C<struct ev_loop *> argument.	2989	functions and callbacks have an initial C<struct ev_loop *> argument.
1780		2990
1781	To make it easier to write programs that cope with either variant, the	2991	To make it easier to write programs that cope with either variant, the
1782	following macros are defined:	2992	following macros are defined:
1783		2993
1784	=over 4	2994	=over 4
…		…
1787		2997
1788	This provides the loop I<argument> for functions, if one is required ("ev	2998	This provides the loop I<argument> for functions, if one is required ("ev
1789	loop argument"). The C<EV_A> form is used when this is the sole argument,	2999	loop argument"). The C<EV_A> form is used when this is the sole argument,
1790	C<EV_A_> is used when other arguments are following. Example:	3000	C<EV_A_> is used when other arguments are following. Example:
1791		3001
1792	ev_unref (EV_A);	3002	ev_unref (EV_A);
1793	ev_timer_add (EV_A_ watcher);	3003	ev_timer_add (EV_A_ watcher);
1794	ev_loop (EV_A_ 0);	3004	ev_loop (EV_A_ 0);
1795		3005
1796	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	3006	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
1797	which is often provided by the following macro.	3007	which is often provided by the following macro.
1798		3008
1799	=item C<EV_P>, C<EV_P_>	3009	=item C<EV_P>, C<EV_P_>
1800		3010
1801	This provides the loop I<parameter> for functions, if one is required ("ev	3011	This provides the loop I<parameter> for functions, if one is required ("ev
1802	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	3012	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
1803	C<EV_P_> is used when other parameters are following. Example:	3013	C<EV_P_> is used when other parameters are following. Example:
1804		3014
1805	// this is how ev_unref is being declared	3015	// this is how ev_unref is being declared
1806	static void ev_unref (EV_P);	3016	static void ev_unref (EV_P);
1807		3017
1808	// this is how you can declare your typical callback	3018	// this is how you can declare your typical callback
1809	static void cb (EV_P_ ev_timer *w, int revents)	3019	static void cb (EV_P_ ev_timer *w, int revents)
1810		3020
1811	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	3021	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
1812	suitable for use with C<EV_A>.	3022	suitable for use with C<EV_A>.
1813		3023
1814	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	3024	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
1815		3025
1816	Similar to the other two macros, this gives you the value of the default	3026	Similar to the other two macros, this gives you the value of the default
1817	loop, if multiple loops are supported ("ev loop default").	3027	loop, if multiple loops are supported ("ev loop default").
1818		3028
		3029	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
		3030
		3031	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
		3032	default loop has been initialised (C<UC> == unchecked). Their behaviour
		3033	is undefined when the default loop has not been initialised by a previous
		3034	execution of C<EV_DEFAULT>, C<EV_DEFAULT_> or C<ev_default_init (...)>.
		3035
		3036	It is often prudent to use C<EV_DEFAULT> when initialising the first
		3037	watcher in a function but use C<EV_DEFAULT_UC> afterwards.
		3038
1819	=back	3039	=back
1820		3040
1821	Example: Declare and initialise a check watcher, working regardless of	3041	Example: Declare and initialise a check watcher, utilising the above
1822	wether multiple loops are supported or not.	3042	macros so it will work regardless of whether multiple loops are supported
		3043	or not.
1823		3044
1824	static void	3045	static void
1825	check_cb (EV_P_ ev_timer *w, int revents)	3046	check_cb (EV_P_ ev_timer *w, int revents)
1826	{	3047	{
1827	ev_check_stop (EV_A_ w);	3048	ev_check_stop (EV_A_ w);
1828	}	3049	}
1829		3050
1830	ev_check check;	3051	ev_check check;
1831	ev_check_init (&check, check_cb);	3052	ev_check_init (&check, check_cb);
1832	ev_check_start (EV_DEFAULT_ &check);	3053	ev_check_start (EV_DEFAULT_ &check);
1833	ev_loop (EV_DEFAULT_ 0);	3054	ev_loop (EV_DEFAULT_ 0);
1834
1835		3055
1836	=head1 EMBEDDING	3056	=head1 EMBEDDING
1837		3057
1838	Libev can (and often is) directly embedded into host	3058	Libev can (and often is) directly embedded into host
1839	applications. Examples of applications that embed it include the Deliantra	3059	applications. Examples of applications that embed it include the Deliantra
1840	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)	3060	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1841	and rxvt-unicode.	3061	and rxvt-unicode.
1842		3062
1843	The goal is to enable you to just copy the neecssary files into your	3063	The goal is to enable you to just copy the necessary files into your
1844	source directory without having to change even a single line in them, so	3064	source directory without having to change even a single line in them, so
1845	you can easily upgrade by simply copying (or having a checked-out copy of	3065	you can easily upgrade by simply copying (or having a checked-out copy of
1846	libev somewhere in your source tree).	3066	libev somewhere in your source tree).
1847		3067
1848	=head2 FILESETS	3068	=head2 FILESETS
1849		3069
1850	Depending on what features you need you need to include one or more sets of files	3070	Depending on what features you need you need to include one or more sets of files
1851	in your app.	3071	in your application.
1852		3072
1853	=head3 CORE EVENT LOOP	3073	=head3 CORE EVENT LOOP
1854		3074
1855	To include only the libev core (all the C<ev_*> functions), with manual	3075	To include only the libev core (all the C<ev_*> functions), with manual
1856	configuration (no autoconf):	3076	configuration (no autoconf):
1857		3077
1858	#define EV_STANDALONE 1	3078	#define EV_STANDALONE 1
1859	#include "ev.c"	3079	#include "ev.c"
1860		3080
1861	This will automatically include F<ev.h>, too, and should be done in a	3081	This will automatically include F<ev.h>, too, and should be done in a
1862	single C source file only to provide the function implementations. To use	3082	single C source file only to provide the function implementations. To use
1863	it, do the same for F<ev.h> in all files wishing to use this API (best	3083	it, do the same for F<ev.h> in all files wishing to use this API (best
1864	done by writing a wrapper around F<ev.h> that you can include instead and	3084	done by writing a wrapper around F<ev.h> that you can include instead and
1865	where you can put other configuration options):	3085	where you can put other configuration options):
1866		3086
1867	#define EV_STANDALONE 1	3087	#define EV_STANDALONE 1
1868	#include "ev.h"	3088	#include "ev.h"
1869		3089
1870	Both header files and implementation files can be compiled with a C++	3090	Both header files and implementation files can be compiled with a C++
1871	compiler (at least, thats a stated goal, and breakage will be treated	3091	compiler (at least, that's a stated goal, and breakage will be treated
1872	as a bug).	3092	as a bug).
1873		3093
1874	You need the following files in your source tree, or in a directory	3094	You need the following files in your source tree, or in a directory
1875	in your include path (e.g. in libev/ when using -Ilibev):	3095	in your include path (e.g. in libev/ when using -Ilibev):
1876		3096
1877	ev.h	3097	ev.h
1878	ev.c	3098	ev.c
1879	ev_vars.h	3099	ev_vars.h
1880	ev_wrap.h	3100	ev_wrap.h
1881		3101
1882	ev_win32.c required on win32 platforms only	3102	ev_win32.c required on win32 platforms only
1883		3103
1884	ev_select.c only when select backend is enabled (which is by default)	3104	ev_select.c only when select backend is enabled (which is enabled by default)
1885	ev_poll.c only when poll backend is enabled (disabled by default)	3105	ev_poll.c only when poll backend is enabled (disabled by default)
1886	ev_epoll.c only when the epoll backend is enabled (disabled by default)	3106	ev_epoll.c only when the epoll backend is enabled (disabled by default)
1887	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	3107	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
1888	ev_port.c only when the solaris port backend is enabled (disabled by default)	3108	ev_port.c only when the solaris port backend is enabled (disabled by default)
1889		3109
1890	F<ev.c> includes the backend files directly when enabled, so you only need	3110	F<ev.c> includes the backend files directly when enabled, so you only need
1891	to compile this single file.	3111	to compile this single file.
1892		3112
1893	=head3 LIBEVENT COMPATIBILITY API	3113	=head3 LIBEVENT COMPATIBILITY API
1894		3114
1895	To include the libevent compatibility API, also include:	3115	To include the libevent compatibility API, also include:
1896		3116
1897	#include "event.c"	3117	#include "event.c"
1898		3118
1899	in the file including F<ev.c>, and:	3119	in the file including F<ev.c>, and:
1900		3120
1901	#include "event.h"	3121	#include "event.h"
1902		3122
1903	in the files that want to use the libevent API. This also includes F<ev.h>.	3123	in the files that want to use the libevent API. This also includes F<ev.h>.
1904		3124
1905	You need the following additional files for this:	3125	You need the following additional files for this:
1906		3126
1907	event.h	3127	event.h
1908	event.c	3128	event.c
1909		3129
1910	=head3 AUTOCONF SUPPORT	3130	=head3 AUTOCONF SUPPORT
1911		3131
1912	Instead of using C<EV_STANDALONE=1> and providing your config in	3132	Instead of using C<EV_STANDALONE=1> and providing your configuration in
1913	whatever way you want, you can also C<m4_include([libev.m4])> in your	3133	whatever way you want, you can also C<m4_include([libev.m4])> in your
1914	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	3134	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
1915	include F<config.h> and configure itself accordingly.	3135	include F<config.h> and configure itself accordingly.
1916		3136
1917	For this of course you need the m4 file:	3137	For this of course you need the m4 file:
1918		3138
1919	libev.m4	3139	libev.m4
1920		3140
1921	=head2 PREPROCESSOR SYMBOLS/MACROS	3141	=head2 PREPROCESSOR SYMBOLS/MACROS
1922		3142
1923	Libev can be configured via a variety of preprocessor symbols you have to define	3143	Libev can be configured via a variety of preprocessor symbols you have to
1924	before including any of its files. The default is not to build for multiplicity	3144	define before including any of its files. The default in the absence of
1925	and only include the select backend.	3145	autoconf is documented for every option.
1926		3146
1927	=over 4	3147	=over 4
1928		3148
1929	=item EV_STANDALONE	3149	=item EV_STANDALONE
1930		3150
…		…
1935	F<event.h> that are not directly supported by the libev core alone.	3155	F<event.h> that are not directly supported by the libev core alone.
1936		3156
1937	=item EV_USE_MONOTONIC	3157	=item EV_USE_MONOTONIC
1938		3158
1939	If defined to be C<1>, libev will try to detect the availability of the	3159	If defined to be C<1>, libev will try to detect the availability of the
1940	monotonic clock option at both compiletime and runtime. Otherwise no use	3160	monotonic clock option at both compile time and runtime. Otherwise no use
1941	of the monotonic clock option will be attempted. If you enable this, you	3161	of the monotonic clock option will be attempted. If you enable this, you
1942	usually have to link against librt or something similar. Enabling it when	3162	usually have to link against librt or something similar. Enabling it when
1943	the functionality isn't available is safe, though, althoguh you have	3163	the functionality isn't available is safe, though, although you have
1944	to make sure you link against any libraries where the C<clock_gettime>	3164	to make sure you link against any libraries where the C<clock_gettime>
1945	function is hiding in (often F<-lrt>).	3165	function is hiding in (often F<-lrt>).
1946		3166
1947	=item EV_USE_REALTIME	3167	=item EV_USE_REALTIME
1948		3168
1949	If defined to be C<1>, libev will try to detect the availability of the	3169	If defined to be C<1>, libev will try to detect the availability of the
1950	realtime clock option at compiletime (and assume its availability at	3170	real-time clock option at compile time (and assume its availability at
1951	runtime if successful). Otherwise no use of the realtime clock option will	3171	runtime if successful). Otherwise no use of the real-time clock option will
1952	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	3172	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
1953	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries	3173	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
1954	in the description of C<EV_USE_MONOTONIC>, though.	3174	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
		3175
		3176	=item EV_USE_NANOSLEEP
		3177
		3178	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		3179	and will use it for delays. Otherwise it will use C<select ()>.
		3180
		3181	=item EV_USE_EVENTFD
		3182
		3183	If defined to be C<1>, then libev will assume that C<eventfd ()> is
		3184	available and will probe for kernel support at runtime. This will improve
		3185	C<ev_signal> and C<ev_async> performance and reduce resource consumption.
		3186	If undefined, it will be enabled if the headers indicate GNU/Linux + Glibc
		3187	2.7 or newer, otherwise disabled.
1955		3188
1956	=item EV_USE_SELECT	3189	=item EV_USE_SELECT
1957		3190
1958	If undefined or defined to be C<1>, libev will compile in support for the	3191	If undefined or defined to be C<1>, libev will compile in support for the
1959	C<select>(2) backend. No attempt at autodetection will be done: if no	3192	C<select>(2) backend. No attempt at auto-detection will be done: if no
1960	other method takes over, select will be it. Otherwise the select backend	3193	other method takes over, select will be it. Otherwise the select backend
1961	will not be compiled in.	3194	will not be compiled in.
1962		3195
1963	=item EV_SELECT_USE_FD_SET	3196	=item EV_SELECT_USE_FD_SET
1964		3197
1965	If defined to C<1>, then the select backend will use the system C<fd_set>	3198	If defined to C<1>, then the select backend will use the system C<fd_set>
1966	structure. This is useful if libev doesn't compile due to a missing	3199	structure. This is useful if libev doesn't compile due to a missing
1967	C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on	3200	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout on
1968	exotic systems. This usually limits the range of file descriptors to some	3201	exotic systems. This usually limits the range of file descriptors to some
1969	low limit such as 1024 or might have other limitations (winsocket only	3202	low limit such as 1024 or might have other limitations (winsocket only
1970	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	3203	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might
1971	influence the size of the C<fd_set> used.	3204	influence the size of the C<fd_set> used.
1972		3205
…		…
1978	be used is the winsock select). This means that it will call	3211	be used is the winsock select). This means that it will call
1979	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	3212	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
1980	it is assumed that all these functions actually work on fds, even	3213	it is assumed that all these functions actually work on fds, even
1981	on win32. Should not be defined on non-win32 platforms.	3214	on win32. Should not be defined on non-win32 platforms.
1982		3215
		3216	=item EV_FD_TO_WIN32_HANDLE
		3217
		3218	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		3219	file descriptors to socket handles. When not defining this symbol (the
		3220	default), then libev will call C<_get_osfhandle>, which is usually
		3221	correct. In some cases, programs use their own file descriptor management,
		3222	in which case they can provide this function to map fds to socket handles.
		3223
1983	=item EV_USE_POLL	3224	=item EV_USE_POLL
1984		3225
1985	If defined to be C<1>, libev will compile in support for the C<poll>(2)	3226	If defined to be C<1>, libev will compile in support for the C<poll>(2)
1986	backend. Otherwise it will be enabled on non-win32 platforms. It	3227	backend. Otherwise it will be enabled on non-win32 platforms. It
1987	takes precedence over select.	3228	takes precedence over select.
1988		3229
1989	=item EV_USE_EPOLL	3230	=item EV_USE_EPOLL
1990		3231
1991	If defined to be C<1>, libev will compile in support for the Linux	3232	If defined to be C<1>, libev will compile in support for the Linux
1992	C<epoll>(7) backend. Its availability will be detected at runtime,	3233	C<epoll>(7) backend. Its availability will be detected at runtime,
1993	otherwise another method will be used as fallback. This is the	3234	otherwise another method will be used as fallback. This is the preferred
1994	preferred backend for GNU/Linux systems.	3235	backend for GNU/Linux systems. If undefined, it will be enabled if the
		3236	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
1995		3237
1996	=item EV_USE_KQUEUE	3238	=item EV_USE_KQUEUE
1997		3239
1998	If defined to be C<1>, libev will compile in support for the BSD style	3240	If defined to be C<1>, libev will compile in support for the BSD style
1999	C<kqueue>(2) backend. Its actual availability will be detected at runtime,	3241	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
…		…
2012	otherwise another method will be used as fallback. This is the preferred	3254	otherwise another method will be used as fallback. This is the preferred
2013	backend for Solaris 10 systems.	3255	backend for Solaris 10 systems.
2014		3256
2015	=item EV_USE_DEVPOLL	3257	=item EV_USE_DEVPOLL
2016		3258
2017	reserved for future expansion, works like the USE symbols above.	3259	Reserved for future expansion, works like the USE symbols above.
		3260
		3261	=item EV_USE_INOTIFY
		3262
		3263	If defined to be C<1>, libev will compile in support for the Linux inotify
		3264	interface to speed up C<ev_stat> watchers. Its actual availability will
		3265	be detected at runtime. If undefined, it will be enabled if the headers
		3266	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		3267
		3268	=item EV_ATOMIC_T
		3269
		3270	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		3271	access is atomic with respect to other threads or signal contexts. No such
		3272	type is easily found in the C language, so you can provide your own type
		3273	that you know is safe for your purposes. It is used both for signal handler "locking"
		3274	as well as for signal and thread safety in C<ev_async> watchers.
		3275
		3276	In the absence of this define, libev will use C<sig_atomic_t volatile>
		3277	(from F<signal.h>), which is usually good enough on most platforms.
2018		3278
2019	=item EV_H	3279	=item EV_H
2020		3280
2021	The name of the F<ev.h> header file used to include it. The default if	3281	The name of the F<ev.h> header file used to include it. The default if
2022	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	3282	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2023	can be used to virtually rename the F<ev.h> header file in case of conflicts.	3283	used to virtually rename the F<ev.h> header file in case of conflicts.
2024		3284
2025	=item EV_CONFIG_H	3285	=item EV_CONFIG_H
2026		3286
2027	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	3287	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2028	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	3288	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2029	C<EV_H>, above.	3289	C<EV_H>, above.
2030		3290
2031	=item EV_EVENT_H	3291	=item EV_EVENT_H
2032		3292
2033	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	3293	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2034	of how the F<event.h> header can be found.	3294	of how the F<event.h> header can be found, the default is C<"event.h">.
2035		3295
2036	=item EV_PROTOTYPES	3296	=item EV_PROTOTYPES
2037		3297
2038	If defined to be C<0>, then F<ev.h> will not define any function	3298	If defined to be C<0>, then F<ev.h> will not define any function
2039	prototypes, but still define all the structs and other symbols. This is	3299	prototypes, but still define all the structs and other symbols. This is
…		…
2046	will have the C<struct ev_loop *> as first argument, and you can create	3306	will have the C<struct ev_loop *> as first argument, and you can create
2047	additional independent event loops. Otherwise there will be no support	3307	additional independent event loops. Otherwise there will be no support
2048	for multiple event loops and there is no first event loop pointer	3308	for multiple event loops and there is no first event loop pointer
2049	argument. Instead, all functions act on the single default loop.	3309	argument. Instead, all functions act on the single default loop.
2050		3310
		3311	=item EV_MINPRI
		3312
		3313	=item EV_MAXPRI
		3314
		3315	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		3316	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		3317	provide for more priorities by overriding those symbols (usually defined
		3318	to be C<-2> and C<2>, respectively).
		3319
		3320	When doing priority-based operations, libev usually has to linearly search
		3321	all the priorities, so having many of them (hundreds) uses a lot of space
		3322	and time, so using the defaults of five priorities (-2 .. +2) is usually
		3323	fine.
		3324
		3325	If your embedding application does not need any priorities, defining these
		3326	both to C<0> will save some memory and CPU.
		3327
2051	=item EV_PERIODIC_ENABLE	3328	=item EV_PERIODIC_ENABLE
2052		3329
2053	If undefined or defined to be C<1>, then periodic timers are supported. If	3330	If undefined or defined to be C<1>, then periodic timers are supported. If
2054	defined to be C<0>, then they are not. Disabling them saves a few kB of	3331	defined to be C<0>, then they are not. Disabling them saves a few kB of
2055	code.	3332	code.
2056		3333
		3334	=item EV_IDLE_ENABLE
		3335
		3336	If undefined or defined to be C<1>, then idle watchers are supported. If
		3337	defined to be C<0>, then they are not. Disabling them saves a few kB of
		3338	code.
		3339
2057	=item EV_EMBED_ENABLE	3340	=item EV_EMBED_ENABLE
2058		3341
2059	If undefined or defined to be C<1>, then embed watchers are supported. If	3342	If undefined or defined to be C<1>, then embed watchers are supported. If
2060	defined to be C<0>, then they are not.	3343	defined to be C<0>, then they are not. Embed watchers rely on most other
		3344	watcher types, which therefore must not be disabled.
2061		3345
2062	=item EV_STAT_ENABLE	3346	=item EV_STAT_ENABLE
2063		3347
2064	If undefined or defined to be C<1>, then stat watchers are supported. If	3348	If undefined or defined to be C<1>, then stat watchers are supported. If
2065	defined to be C<0>, then they are not.	3349	defined to be C<0>, then they are not.
…		…
2067	=item EV_FORK_ENABLE	3351	=item EV_FORK_ENABLE
2068		3352
2069	If undefined or defined to be C<1>, then fork watchers are supported. If	3353	If undefined or defined to be C<1>, then fork watchers are supported. If
2070	defined to be C<0>, then they are not.	3354	defined to be C<0>, then they are not.
2071		3355
		3356	=item EV_ASYNC_ENABLE
		3357
		3358	If undefined or defined to be C<1>, then async watchers are supported. If
		3359	defined to be C<0>, then they are not.
		3360
2072	=item EV_MINIMAL	3361	=item EV_MINIMAL
2073		3362
2074	If you need to shave off some kilobytes of code at the expense of some	3363	If you need to shave off some kilobytes of code at the expense of some
2075	speed, define this symbol to C<1>. Currently only used for gcc to override	3364	speed, define this symbol to C<1>. Currently this is used to override some
2076	some inlining decisions, saves roughly 30% codesize of amd64.	3365	inlining decisions, saves roughly 30% code size on amd64. It also selects a
		3366	much smaller 2-heap for timer management over the default 4-heap.
2077		3367
2078	=item EV_PID_HASHSIZE	3368	=item EV_PID_HASHSIZE
2079		3369
2080	C<ev_child> watchers use a small hash table to distribute workload by	3370	C<ev_child> watchers use a small hash table to distribute workload by
2081	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	3371	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
2082	than enough. If you need to manage thousands of children you might want to	3372	than enough. If you need to manage thousands of children you might want to
2083	increase this value.	3373	increase this value (I<must> be a power of two).
		3374
		3375	=item EV_INOTIFY_HASHSIZE
		3376
		3377	C<ev_stat> watchers use a small hash table to distribute workload by
		3378	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		3379	usually more than enough. If you need to manage thousands of C<ev_stat>
		3380	watchers you might want to increase this value (I<must> be a power of
		3381	two).
		3382
		3383	=item EV_USE_4HEAP
		3384
		3385	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3386	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
		3387	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
		3388	faster performance with many (thousands) of watchers.
		3389
		3390	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3391	(disabled).
		3392
		3393	=item EV_HEAP_CACHE_AT
		3394
		3395	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3396	timer and periodics heaps, libev can cache the timestamp (I<at>) within
		3397	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3398	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3399	but avoids random read accesses on heap changes. This improves performance
		3400	noticeably with many (hundreds) of watchers.
		3401
		3402	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3403	(disabled).
		3404
		3405	=item EV_VERIFY
		3406
		3407	Controls how much internal verification (see C<ev_loop_verify ()>) will
		3408	be done: If set to C<0>, no internal verification code will be compiled
		3409	in. If set to C<1>, then verification code will be compiled in, but not
		3410	called. If set to C<2>, then the internal verification code will be
		3411	called once per loop, which can slow down libev. If set to C<3>, then the
		3412	verification code will be called very frequently, which will slow down
		3413	libev considerably.
		3414
		3415	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
		3416	C<0>.
2084		3417
2085	=item EV_COMMON	3418	=item EV_COMMON
2086		3419
2087	By default, all watchers have a C<void *data> member. By redefining	3420	By default, all watchers have a C<void *data> member. By redefining
2088	this macro to a something else you can include more and other types of	3421	this macro to a something else you can include more and other types of
2089	members. You have to define it each time you include one of the files,	3422	members. You have to define it each time you include one of the files,
2090	though, and it must be identical each time.	3423	though, and it must be identical each time.
2091		3424
2092	For example, the perl EV module uses something like this:	3425	For example, the perl EV module uses something like this:
2093		3426
2094	#define EV_COMMON \	3427	#define EV_COMMON \
2095	SV *self; /* contains this struct */ \	3428	SV *self; /* contains this struct */ \
2096	SV *cb_sv, *fh /* note no trailing ";" */	3429	SV *cb_sv, *fh /* note no trailing ";" */
2097		3430
2098	=item EV_CB_DECLARE (type)	3431	=item EV_CB_DECLARE (type)
2099		3432
2100	=item EV_CB_INVOKE (watcher, revents)	3433	=item EV_CB_INVOKE (watcher, revents)
2101		3434
2102	=item ev_set_cb (ev, cb)	3435	=item ev_set_cb (ev, cb)
2103		3436
2104	Can be used to change the callback member declaration in each watcher,	3437	Can be used to change the callback member declaration in each watcher,
2105	and the way callbacks are invoked and set. Must expand to a struct member	3438	and the way callbacks are invoked and set. Must expand to a struct member
2106	definition and a statement, respectively. See the F<ev.v> header file for	3439	definition and a statement, respectively. See the F<ev.h> header file for
2107	their default definitions. One possible use for overriding these is to	3440	their default definitions. One possible use for overriding these is to
2108	avoid the C<struct ev_loop *> as first argument in all cases, or to use	3441	avoid the C<struct ev_loop *> as first argument in all cases, or to use
2109	method calls instead of plain function calls in C++.	3442	method calls instead of plain function calls in C++.
		3443
		3444	=back
		3445
		3446	=head2 EXPORTED API SYMBOLS
		3447
		3448	If you need to re-export the API (e.g. via a DLL) and you need a list of
		3449	exported symbols, you can use the provided F<Symbol.*> files which list
		3450	all public symbols, one per line:
		3451
		3452	Symbols.ev for libev proper
		3453	Symbols.event for the libevent emulation
		3454
		3455	This can also be used to rename all public symbols to avoid clashes with
		3456	multiple versions of libev linked together (which is obviously bad in
		3457	itself, but sometimes it is inconvenient to avoid this).
		3458
		3459	A sed command like this will create wrapper C<#define>'s that you need to
		3460	include before including F<ev.h>:
		3461
		3462	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		3463
		3464	This would create a file F<wrap.h> which essentially looks like this:
		3465
		3466	#define ev_backend myprefix_ev_backend
		3467	#define ev_check_start myprefix_ev_check_start
		3468	#define ev_check_stop myprefix_ev_check_stop
		3469	...
2110		3470
2111	=head2 EXAMPLES	3471	=head2 EXAMPLES
2112		3472
2113	For a real-world example of a program the includes libev	3473	For a real-world example of a program the includes libev
2114	verbatim, you can have a look at the EV perl module	3474	verbatim, you can have a look at the EV perl module
…		…
2117	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file	3477	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
2118	will be compiled. It is pretty complex because it provides its own header	3478	will be compiled. It is pretty complex because it provides its own header
2119	file.	3479	file.
2120		3480
2121	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	3481	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
2122	that everybody includes and which overrides some autoconf choices:	3482	that everybody includes and which overrides some configure choices:
2123		3483
		3484	#define EV_MINIMAL 1
2124	#define EV_USE_POLL 0	3485	#define EV_USE_POLL 0
2125	#define EV_MULTIPLICITY 0	3486	#define EV_MULTIPLICITY 0
2126	#define EV_PERIODICS 0	3487	#define EV_PERIODIC_ENABLE 0
		3488	#define EV_STAT_ENABLE 0
		3489	#define EV_FORK_ENABLE 0
2127	#define EV_CONFIG_H <config.h>	3490	#define EV_CONFIG_H <config.h>
		3491	#define EV_MINPRI 0
		3492	#define EV_MAXPRI 0
2128		3493
2129	#include "ev++.h"	3494	#include "ev++.h"
2130		3495
2131	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	3496	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
2132		3497
2133	#include "ev_cpp.h"	3498	#include "ev_cpp.h"
2134	#include "ev.c"	3499	#include "ev.c"
2135		3500
		3501	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
2136		3502
		3503	=head2 THREADS AND COROUTINES
		3504
		3505	=head3 THREADS
		3506
		3507	All libev functions are reentrant and thread-safe unless explicitly
		3508	documented otherwise, but libev implements no locking itself. This means
		3509	that you can use as many loops as you want in parallel, as long as there
		3510	are no concurrent calls into any libev function with the same loop
		3511	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		3512	of course): libev guarantees that different event loops share no data
		3513	structures that need any locking.
		3514
		3515	Or to put it differently: calls with different loop parameters can be done
		3516	concurrently from multiple threads, calls with the same loop parameter
		3517	must be done serially (but can be done from different threads, as long as
		3518	only one thread ever is inside a call at any point in time, e.g. by using
		3519	a mutex per loop).
		3520
		3521	Specifically to support threads (and signal handlers), libev implements
		3522	so-called C<ev_async> watchers, which allow some limited form of
		3523	concurrency on the same event loop, namely waking it up "from the
		3524	outside".
		3525
		3526	If you want to know which design (one loop, locking, or multiple loops
		3527	without or something else still) is best for your problem, then I cannot
		3528	help you, but here is some generic advice:
		3529
		3530	=over 4
		3531
		3532	=item * most applications have a main thread: use the default libev loop
		3533	in that thread, or create a separate thread running only the default loop.
		3534
		3535	This helps integrating other libraries or software modules that use libev
		3536	themselves and don't care/know about threading.
		3537
		3538	=item * one loop per thread is usually a good model.
		3539
		3540	Doing this is almost never wrong, sometimes a better-performance model
		3541	exists, but it is always a good start.
		3542
		3543	=item * other models exist, such as the leader/follower pattern, where one
		3544	loop is handed through multiple threads in a kind of round-robin fashion.
		3545
		3546	Choosing a model is hard - look around, learn, know that usually you can do
		3547	better than you currently do :-)
		3548
		3549	=item * often you need to talk to some other thread which blocks in the
		3550	event loop.
		3551
		3552	C<ev_async> watchers can be used to wake them up from other threads safely
		3553	(or from signal contexts...).
		3554
		3555	An example use would be to communicate signals or other events that only
		3556	work in the default loop by registering the signal watcher with the
		3557	default loop and triggering an C<ev_async> watcher from the default loop
		3558	watcher callback into the event loop interested in the signal.
		3559
		3560	=back
		3561
		3562	=head3 COROUTINES
		3563
		3564	Libev is very accommodating to coroutines ("cooperative threads"):
		3565	libev fully supports nesting calls to its functions from different
		3566	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		3567	different coroutines, and switch freely between both coroutines running the
		3568	loop, as long as you don't confuse yourself). The only exception is that
		3569	you must not do this from C<ev_periodic> reschedule callbacks.
		3570
		3571	Care has been taken to ensure that libev does not keep local state inside
		3572	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		3573	they do not call any callbacks.
		3574
		3575	=head2 COMPILER WARNINGS
		3576
		3577	Depending on your compiler and compiler settings, you might get no or a
		3578	lot of warnings when compiling libev code. Some people are apparently
		3579	scared by this.
		3580
		3581	However, these are unavoidable for many reasons. For one, each compiler
		3582	has different warnings, and each user has different tastes regarding
		3583	warning options. "Warn-free" code therefore cannot be a goal except when
		3584	targeting a specific compiler and compiler-version.
		3585
		3586	Another reason is that some compiler warnings require elaborate
		3587	workarounds, or other changes to the code that make it less clear and less
		3588	maintainable.
		3589
		3590	And of course, some compiler warnings are just plain stupid, or simply
		3591	wrong (because they don't actually warn about the condition their message
		3592	seems to warn about). For example, certain older gcc versions had some
		3593	warnings that resulted an extreme number of false positives. These have
		3594	been fixed, but some people still insist on making code warn-free with
		3595	such buggy versions.
		3596
		3597	While libev is written to generate as few warnings as possible,
		3598	"warn-free" code is not a goal, and it is recommended not to build libev
		3599	with any compiler warnings enabled unless you are prepared to cope with
		3600	them (e.g. by ignoring them). Remember that warnings are just that:
		3601	warnings, not errors, or proof of bugs.
		3602
		3603
		3604	=head2 VALGRIND
		3605
		3606	Valgrind has a special section here because it is a popular tool that is
		3607	highly useful. Unfortunately, valgrind reports are very hard to interpret.
		3608
		3609	If you think you found a bug (memory leak, uninitialised data access etc.)
		3610	in libev, then check twice: If valgrind reports something like:
		3611
		3612	==2274== definitely lost: 0 bytes in 0 blocks.
		3613	==2274== possibly lost: 0 bytes in 0 blocks.
		3614	==2274== still reachable: 256 bytes in 1 blocks.
		3615
		3616	Then there is no memory leak, just as memory accounted to global variables
		3617	is not a memleak - the memory is still being referenced, and didn't leak.
		3618
		3619	Similarly, under some circumstances, valgrind might report kernel bugs
		3620	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3621	although an acceptable workaround has been found here), or it might be
		3622	confused.
		3623
		3624	Keep in mind that valgrind is a very good tool, but only a tool. Don't
		3625	make it into some kind of religion.
		3626
		3627	If you are unsure about something, feel free to contact the mailing list
		3628	with the full valgrind report and an explanation on why you think this
		3629	is a bug in libev (best check the archives, too :). However, don't be
		3630	annoyed when you get a brisk "this is no bug" answer and take the chance
		3631	of learning how to interpret valgrind properly.
		3632
		3633	If you need, for some reason, empty reports from valgrind for your project
		3634	I suggest using suppression lists.
		3635
		3636
		3637	=head1 PORTABILITY NOTES
		3638
		3639	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
		3640
		3641	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		3642	requires, and its I/O model is fundamentally incompatible with the POSIX
		3643	model. Libev still offers limited functionality on this platform in
		3644	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		3645	descriptors. This only applies when using Win32 natively, not when using
		3646	e.g. cygwin.
		3647
		3648	Lifting these limitations would basically require the full
		3649	re-implementation of the I/O system. If you are into these kinds of
		3650	things, then note that glib does exactly that for you in a very portable
		3651	way (note also that glib is the slowest event library known to man).
		3652
		3653	There is no supported compilation method available on windows except
		3654	embedding it into other applications.
		3655
		3656	Not a libev limitation but worth mentioning: windows apparently doesn't
		3657	accept large writes: instead of resulting in a partial write, windows will
		3658	either accept everything or return C<ENOBUFS> if the buffer is too large,
		3659	so make sure you only write small amounts into your sockets (less than a
		3660	megabyte seems safe, but this apparently depends on the amount of memory
		3661	available).
		3662
		3663	Due to the many, low, and arbitrary limits on the win32 platform and
		3664	the abysmal performance of winsockets, using a large number of sockets
		3665	is not recommended (and not reasonable). If your program needs to use
		3666	more than a hundred or so sockets, then likely it needs to use a totally
		3667	different implementation for windows, as libev offers the POSIX readiness
		3668	notification model, which cannot be implemented efficiently on windows
		3669	(Microsoft monopoly games).
		3670
		3671	A typical way to use libev under windows is to embed it (see the embedding
		3672	section for details) and use the following F<evwrap.h> header file instead
		3673	of F<ev.h>:
		3674
		3675	#define EV_STANDALONE /* keeps ev from requiring config.h */
		3676	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		3677
		3678	#include "ev.h"
		3679
		3680	And compile the following F<evwrap.c> file into your project (make sure
		3681	you do I<not> compile the F<ev.c> or any other embedded source files!):
		3682
		3683	#include "evwrap.h"
		3684	#include "ev.c"
		3685
		3686	=over 4
		3687
		3688	=item The winsocket select function
		3689
		3690	The winsocket C<select> function doesn't follow POSIX in that it
		3691	requires socket I<handles> and not socket I<file descriptors> (it is
		3692	also extremely buggy). This makes select very inefficient, and also
		3693	requires a mapping from file descriptors to socket handles (the Microsoft
		3694	C runtime provides the function C<_open_osfhandle> for this). See the
		3695	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		3696	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
		3697
		3698	The configuration for a "naked" win32 using the Microsoft runtime
		3699	libraries and raw winsocket select is:
		3700
		3701	#define EV_USE_SELECT 1
		3702	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		3703
		3704	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3705	complexity in the O(n²) range when using win32.
		3706
		3707	=item Limited number of file descriptors
		3708
		3709	Windows has numerous arbitrary (and low) limits on things.
		3710
		3711	Early versions of winsocket's select only supported waiting for a maximum
		3712	of C<64> handles (probably owning to the fact that all windows kernels
		3713	can only wait for C<64> things at the same time internally; Microsoft
		3714	recommends spawning a chain of threads and wait for 63 handles and the
		3715	previous thread in each. Great).
		3716
		3717	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3718	to some high number (e.g. C<2048>) before compiling the winsocket select
		3719	call (which might be in libev or elsewhere, for example, perl does its own
		3720	select emulation on windows).
		3721
		3722	Another limit is the number of file descriptors in the Microsoft runtime
		3723	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3724	or something like this inside Microsoft). You can increase this by calling
		3725	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3726	arbitrary limit), but is broken in many versions of the Microsoft runtime
		3727	libraries.
		3728
		3729	This might get you to about C<512> or C<2048> sockets (depending on
		3730	windows version and/or the phase of the moon). To get more, you need to
		3731	wrap all I/O functions and provide your own fd management, but the cost of
		3732	calling select (O(n²)) will likely make this unworkable.
		3733
		3734	=back
		3735
		3736	=head2 PORTABILITY REQUIREMENTS
		3737
		3738	In addition to a working ISO-C implementation and of course the
		3739	backend-specific APIs, libev relies on a few additional extensions:
		3740
		3741	=over 4
		3742
		3743	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		3744	calling conventions regardless of C<ev_watcher_type *>.
		3745
		3746	Libev assumes not only that all watcher pointers have the same internal
		3747	structure (guaranteed by POSIX but not by ISO C for example), but it also
		3748	assumes that the same (machine) code can be used to call any watcher
		3749	callback: The watcher callbacks have different type signatures, but libev
		3750	calls them using an C<ev_watcher *> internally.
		3751
		3752	=item C<sig_atomic_t volatile> must be thread-atomic as well
		3753
		3754	The type C<sig_atomic_t volatile> (or whatever is defined as
		3755	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
		3756	threads. This is not part of the specification for C<sig_atomic_t>, but is
		3757	believed to be sufficiently portable.
		3758
		3759	=item C<sigprocmask> must work in a threaded environment
		3760
		3761	Libev uses C<sigprocmask> to temporarily block signals. This is not
		3762	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		3763	pthread implementations will either allow C<sigprocmask> in the "main
		3764	thread" or will block signals process-wide, both behaviours would
		3765	be compatible with libev. Interaction between C<sigprocmask> and
		3766	C<pthread_sigmask> could complicate things, however.
		3767
		3768	The most portable way to handle signals is to block signals in all threads
		3769	except the initial one, and run the default loop in the initial thread as
		3770	well.
		3771
		3772	=item C<long> must be large enough for common memory allocation sizes
		3773
		3774	To improve portability and simplify its API, libev uses C<long> internally
		3775	instead of C<size_t> when allocating its data structures. On non-POSIX
		3776	systems (Microsoft...) this might be unexpectedly low, but is still at
		3777	least 31 bits everywhere, which is enough for hundreds of millions of
		3778	watchers.
		3779
		3780	=item C<double> must hold a time value in seconds with enough accuracy
		3781
		3782	The type C<double> is used to represent timestamps. It is required to
		3783	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		3784	enough for at least into the year 4000. This requirement is fulfilled by
		3785	implementations implementing IEEE 754 (basically all existing ones).
		3786
		3787	=back
		3788
		3789	If you know of other additional requirements drop me a note.
		3790
		3791
2137	=head1 COMPLEXITIES	3792	=head1 ALGORITHMIC COMPLEXITIES
2138		3793
2139	In this section the complexities of (many of) the algorithms used inside	3794	In this section the complexities of (many of) the algorithms used inside
2140	libev will be explained. For complexity discussions about backends see the	3795	libev will be documented. For complexity discussions about backends see
2141	documentation for C<ev_default_init>.	3796	the documentation for C<ev_default_init>.
		3797
		3798	All of the following are about amortised time: If an array needs to be
		3799	extended, libev needs to realloc and move the whole array, but this
		3800	happens asymptotically rarer with higher number of elements, so O(1) might
		3801	mean that libev does a lengthy realloc operation in rare cases, but on
		3802	average it is much faster and asymptotically approaches constant time.
2142		3803
2143	=over 4	3804	=over 4
2144		3805
2145	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	3806	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2146		3807
		3808	This means that, when you have a watcher that triggers in one hour and
		3809	there are 100 watchers that would trigger before that, then inserting will
		3810	have to skip roughly seven (C<ld 100>) of these watchers.
		3811
2147	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	3812	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2148		3813
		3814	That means that changing a timer costs less than removing/adding them,
		3815	as only the relative motion in the event queue has to be paid for.
		3816
2149	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	3817	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2150		3818
		3819	These just add the watcher into an array or at the head of a list.
		3820
2151	=item Stopping check/prepare/idle watchers: O(1)	3821	=item Stopping check/prepare/idle/fork/async watchers: O(1)
2152		3822
2153	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % 16))	3823	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2154		3824
		3825	These watchers are stored in lists, so they need to be walked to find the
		3826	correct watcher to remove. The lists are usually short (you don't usually
		3827	have many watchers waiting for the same fd or signal: one is typical, two
		3828	is rare).
		3829
2155	=item Finding the next timer per loop iteration: O(1)	3830	=item Finding the next timer in each loop iteration: O(1)
		3831
		3832	By virtue of using a binary or 4-heap, the next timer is always found at a
		3833	fixed position in the storage array.
2156		3834
2157	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	3835	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2158		3836
2159	=item Activating one watcher: O(1)	3837	A change means an I/O watcher gets started or stopped, which requires
		3838	libev to recalculate its status (and possibly tell the kernel, depending
		3839	on backend and whether C<ev_io_set> was used).
		3840
		3841	=item Activating one watcher (putting it into the pending state): O(1)
		3842
		3843	=item Priority handling: O(number_of_priorities)
		3844
		3845	Priorities are implemented by allocating some space for each
		3846	priority. When doing priority-based operations, libev usually has to
		3847	linearly search all the priorities, but starting/stopping and activating
		3848	watchers becomes O(1) with respect to priority handling.
		3849
		3850	=item Sending an ev_async: O(1)
		3851
		3852	=item Processing ev_async_send: O(number_of_async_watchers)
		3853
		3854	=item Processing signals: O(max_signal_number)
		3855
		3856	Sending involves a system call I<iff> there were no other C<ev_async_send>
		3857	calls in the current loop iteration. Checking for async and signal events
		3858	involves iterating over all running async watchers or all signal numbers.
2160		3859
2161	=back	3860	=back
2162		3861
2163		3862
2164	=head1 AUTHOR	3863	=head1 AUTHOR

Diff Legend

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