ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.310 by root, Thu Oct 21 12:32:47 2010 UTC vs.
Revision 1.457 by root, Fri Sep 6 22:17:52 2019 UTC

		1	=encoding utf-8
		2
1	=head1 NAME	3	=head1 NAME
2		4
3	libev - a high performance full-featured event loop written in C	5	libev - a high performance full-featured event loop written in C
4		6
5	=head1 SYNOPSIS	7	=head1 SYNOPSIS
…		…
43		45
44	int	46	int
45	main (void)	47	main (void)
46	{	48	{
47	// use the default event loop unless you have special needs	49	// use the default event loop unless you have special needs
48	struct ev_loop *loop = ev_default_loop (0);	50	struct ev_loop *loop = EV_DEFAULT;
49		51
50	// initialise an io watcher, then start it	52	// initialise an io watcher, then start it
51	// this one will watch for stdin to become readable	53	// this one will watch for stdin to become readable
52	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	54	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
53	ev_io_start (loop, &stdin_watcher);	55	ev_io_start (loop, &stdin_watcher);
…		…
58	ev_timer_start (loop, &timeout_watcher);	60	ev_timer_start (loop, &timeout_watcher);
59		61
60	// now wait for events to arrive	62	// now wait for events to arrive
61	ev_run (loop, 0);	63	ev_run (loop, 0);
62		64
63	// unloop was called, so exit	65	// break was called, so exit
64	return 0;	66	return 0;
65	}	67	}
66		68
67	=head1 ABOUT THIS DOCUMENT	69	=head1 ABOUT THIS DOCUMENT
68		70
…		…
78	with libev.	80	with libev.
79		81
80	Familiarity with event based programming techniques in general is assumed	82	Familiarity with event based programming techniques in general is assumed
81	throughout this document.	83	throughout this document.
82		84
		85	=head1 WHAT TO READ WHEN IN A HURRY
		86
		87	This manual tries to be very detailed, but unfortunately, this also makes
		88	it very long. If you just want to know the basics of libev, I suggest
		89	reading L</ANATOMY OF A WATCHER>, then the L</EXAMPLE PROGRAM> above and
		90	look up the missing functions in L</GLOBAL FUNCTIONS> and the C<ev_io> and
		91	C<ev_timer> sections in L</WATCHER TYPES>.
		92
83	=head1 ABOUT LIBEV	93	=head1 ABOUT LIBEV
84		94
85	Libev is an event loop: you register interest in certain events (such as a	95	Libev is an event loop: you register interest in certain events (such as a
86	file descriptor being readable or a timeout occurring), and it will manage	96	file descriptor being readable or a timeout occurring), and it will manage
87	these event sources and provide your program with events.	97	these event sources and provide your program with events.
…		…
95	details of the event, and then hand it over to libev by I<starting> the	105	details of the event, and then hand it over to libev by I<starting> the
96	watcher.	106	watcher.
97		107
98	=head2 FEATURES	108	=head2 FEATURES
99		109
100	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	110	Libev supports C<select>, C<poll>, the Linux-specific aio and C<epoll>
101	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	111	interfaces, the BSD-specific C<kqueue> and the Solaris-specific event port
102	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	112	mechanisms for file descriptor events (C<ev_io>), the Linux C<inotify>
103	(for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner	113	interface (for C<ev_stat>), Linux eventfd/signalfd (for faster and cleaner
104	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative	114	inter-thread wakeup (C<ev_async>)/signal handling (C<ev_signal>)) relative
105	timers (C<ev_timer>), absolute timers with customised rescheduling	115	timers (C<ev_timer>), absolute timers with customised rescheduling
106	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status	116	(C<ev_periodic>), synchronous signals (C<ev_signal>), process status
107	change events (C<ev_child>), and event watchers dealing with the event	117	change events (C<ev_child>), and event watchers dealing with the event
108	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and	118	loop mechanism itself (C<ev_idle>, C<ev_embed>, C<ev_prepare> and
…		…
124	this argument.	134	this argument.
125		135
126	=head2 TIME REPRESENTATION	136	=head2 TIME REPRESENTATION
127		137
128	Libev represents time as a single floating point number, representing	138	Libev represents time as a single floating point number, representing
129	the (fractional) number of seconds since the (POSIX) epoch (in practise	139	the (fractional) number of seconds since the (POSIX) epoch (in practice
130	somewhere near the beginning of 1970, details are complicated, don't	140	somewhere near the beginning of 1970, details are complicated, don't
131	ask). This type is called C<ev_tstamp>, which is what you should use	141	ask). This type is called C<ev_tstamp>, which is what you should use
132	too. It usually aliases to the C<double> type in C. When you need to do	142	too. It usually aliases to the C<double> type in C. When you need to do
133	any calculations on it, you should treat it as some floating point value.	143	any calculations on it, you should treat it as some floating point value.
134		144
…		…
149	When libev detects a usage error such as a negative timer interval, then	159	When libev detects a usage error such as a negative timer interval, then
150	it will print a diagnostic message and abort (via the C<assert> mechanism,	160	it will print a diagnostic message and abort (via the C<assert> mechanism,
151	so C<NDEBUG> will disable this checking): these are programming errors in	161	so C<NDEBUG> will disable this checking): these are programming errors in
152	the libev caller and need to be fixed there.	162	the libev caller and need to be fixed there.
153		163
		164	Via the C<EV_FREQUENT> macro you can compile in and/or enable extensive
		165	consistency checking code inside libev that can be used to check for
		166	internal inconsistencies, suually caused by application bugs.
		167
154	Libev also has a few internal error-checking C<assert>ions, and also has	168	Libev also has a few internal error-checking C<assert>ions. These do not
155	extensive consistency checking code. These do not trigger under normal
156	circumstances, as they indicate either a bug in libev or worse.	169	trigger under normal circumstances, as they indicate either a bug in libev
		170	or worse.
157		171
158		172
159	=head1 GLOBAL FUNCTIONS	173	=head1 GLOBAL FUNCTIONS
160		174
161	These functions can be called anytime, even before initialising the	175	These functions can be called anytime, even before initialising the
…		…
165		179
166	=item ev_tstamp ev_time ()	180	=item ev_tstamp ev_time ()
167		181
168	Returns the current time as libev would use it. Please note that the	182	Returns the current time as libev would use it. Please note that the
169	C<ev_now> function is usually faster and also often returns the timestamp	183	C<ev_now> function is usually faster and also often returns the timestamp
170	you actually want to know.	184	you actually want to know. Also interesting is the combination of
		185	C<ev_now_update> and C<ev_now>.
171		186
172	=item ev_sleep (ev_tstamp interval)	187	=item ev_sleep (ev_tstamp interval)
173		188
174	Sleep for the given interval: The current thread will be blocked until	189	Sleep for the given interval: The current thread will be blocked
175	either it is interrupted or the given time interval has passed. Basically	190	until either it is interrupted or the given time interval has
		191	passed (approximately - it might return a bit earlier even if not
		192	interrupted). Returns immediately if C<< interval <= 0 >>.
		193
176	this is a sub-second-resolution C<sleep ()>.	194	Basically this is a sub-second-resolution C<sleep ()>.
		195
		196	The range of the C<interval> is limited - libev only guarantees to work
		197	with sleep times of up to one day (C<< interval <= 86400 >>).
177		198
178	=item int ev_version_major ()	199	=item int ev_version_major ()
179		200
180	=item int ev_version_minor ()	201	=item int ev_version_minor ()
181		202
…		…
192	as this indicates an incompatible change. Minor versions are usually	213	as this indicates an incompatible change. Minor versions are usually
193	compatible to older versions, so a larger minor version alone is usually	214	compatible to older versions, so a larger minor version alone is usually
194	not a problem.	215	not a problem.
195		216
196	Example: Make sure we haven't accidentally been linked against the wrong	217	Example: Make sure we haven't accidentally been linked against the wrong
197	version (note, however, that this will not detect ABI mismatches :).	218	version (note, however, that this will not detect other ABI mismatches,
		219	such as LFS or reentrancy).
198		220
199	assert (("libev version mismatch",	221	assert (("libev version mismatch",
200	ev_version_major () == EV_VERSION_MAJOR	222	ev_version_major () == EV_VERSION_MAJOR
201	&& ev_version_minor () >= EV_VERSION_MINOR));	223	&& ev_version_minor () >= EV_VERSION_MINOR));
202		224
…		…
213	assert (("sorry, no epoll, no sex",	235	assert (("sorry, no epoll, no sex",
214	ev_supported_backends () & EVBACKEND_EPOLL));	236	ev_supported_backends () & EVBACKEND_EPOLL));
215		237
216	=item unsigned int ev_recommended_backends ()	238	=item unsigned int ev_recommended_backends ()
217		239
218	Return the set of all backends compiled into this binary of libev and also	240	Return the set of all backends compiled into this binary of libev and
219	recommended for this platform. This set is often smaller than the one	241	also recommended for this platform, meaning it will work for most file
		242	descriptor types. This set is often smaller than the one returned by
220	returned by C<ev_supported_backends>, as for example kqueue is broken on	243	C<ev_supported_backends>, as for example kqueue is broken on most BSDs
221	most BSDs and will not be auto-detected unless you explicitly request it	244	and will not be auto-detected unless you explicitly request it (assuming
222	(assuming you know what you are doing). This is the set of backends that	245	you know what you are doing). This is the set of backends that libev will
223	libev will probe for if you specify no backends explicitly.	246	probe for if you specify no backends explicitly.
224		247
225	=item unsigned int ev_embeddable_backends ()	248	=item unsigned int ev_embeddable_backends ()
226		249
227	Returns the set of backends that are embeddable in other event loops. This	250	Returns the set of backends that are embeddable in other event loops. This
228	is the theoretical, all-platform, value. To find which backends	251	value is platform-specific but can include backends not available on the
229	might be supported on the current system, you would need to look at	252	current system. To find which embeddable backends might be supported on
230	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	253	the current system, you would need to look at C<ev_embeddable_backends ()
231	recommended ones.	254	& ev_supported_backends ()>, likewise for recommended ones.
232		255
233	See the description of C<ev_embed> watchers for more info.	256	See the description of C<ev_embed> watchers for more info.
234		257
235	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]	258	=item ev_set_allocator (void *(*cb)(void *ptr, long size) throw ())
236		259
237	Sets the allocation function to use (the prototype is similar - the	260	Sets the allocation function to use (the prototype is similar - the
238	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is	261	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
239	used to allocate and free memory (no surprises here). If it returns zero	262	used to allocate and free memory (no surprises here). If it returns zero
240	when memory needs to be allocated (C<size != 0>), the library might abort	263	when memory needs to be allocated (C<size != 0>), the library might abort
…		…
246		269
247	You could override this function in high-availability programs to, say,	270	You could override this function in high-availability programs to, say,
248	free some memory if it cannot allocate memory, to use a special allocator,	271	free some memory if it cannot allocate memory, to use a special allocator,
249	or even to sleep a while and retry until some memory is available.	272	or even to sleep a while and retry until some memory is available.
250		273
		274	Example: The following is the C<realloc> function that libev itself uses
		275	which should work with C<realloc> and C<free> functions of all kinds and
		276	is probably a good basis for your own implementation.
		277
		278	static void *
		279	ev_realloc_emul (void *ptr, long size) EV_NOEXCEPT
		280	{
		281	if (size)
		282	return realloc (ptr, size);
		283
		284	free (ptr);
		285	return 0;
		286	}
		287
251	Example: Replace the libev allocator with one that waits a bit and then	288	Example: Replace the libev allocator with one that waits a bit and then
252	retries (example requires a standards-compliant C<realloc>).	289	retries.
253		290
254	static void *	291	static void *
255	persistent_realloc (void *ptr, size_t size)	292	persistent_realloc (void *ptr, size_t size)
256	{	293	{
		294	if (!size)
		295	{
		296	free (ptr);
		297	return 0;
		298	}
		299
257	for (;;)	300	for (;;)
258	{	301	{
259	void *newptr = realloc (ptr, size);	302	void *newptr = realloc (ptr, size);
260		303
261	if (newptr)	304	if (newptr)
…		…
266	}	309	}
267		310
268	...	311	...
269	ev_set_allocator (persistent_realloc);	312	ev_set_allocator (persistent_realloc);
270		313
271	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]	314	=item ev_set_syserr_cb (void (*cb)(const char *msg) throw ())
272		315
273	Set the callback function to call on a retryable system call error (such	316	Set the callback function to call on a retryable system call error (such
274	as failed select, poll, epoll_wait). The message is a printable string	317	as failed select, poll, epoll_wait). The message is a printable string
275	indicating the system call or subsystem causing the problem. If this	318	indicating the system call or subsystem causing the problem. If this
276	callback is set, then libev will expect it to remedy the situation, no	319	callback is set, then libev will expect it to remedy the situation, no
…		…
288	}	331	}
289		332
290	...	333	...
291	ev_set_syserr_cb (fatal_error);	334	ev_set_syserr_cb (fatal_error);
292		335
		336	=item ev_feed_signal (int signum)
		337
		338	This function can be used to "simulate" a signal receive. It is completely
		339	safe to call this function at any time, from any context, including signal
		340	handlers or random threads.
		341
		342	Its main use is to customise signal handling in your process, especially
		343	in the presence of threads. For example, you could block signals
		344	by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
		345	creating any loops), and in one thread, use C<sigwait> or any other
		346	mechanism to wait for signals, then "deliver" them to libev by calling
		347	C<ev_feed_signal>.
		348
293	=back	349	=back
294		350
295	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	351	=head1 FUNCTIONS CONTROLLING EVENT LOOPS
296		352
297	An event loop is described by a C<struct ev_loop *> (the C<struct> is	353	An event loop is described by a C<struct ev_loop *> (the C<struct> is
298	I<not> optional in case unless libev 3 compatibility is disabled, as libev	354	I<not> optional in this case unless libev 3 compatibility is disabled, as
299	3 had an C<ev_loop> function colliding with the struct name).	355	libev 3 had an C<ev_loop> function colliding with the struct name).
300		356
301	The library knows two types of such loops, the I<default> loop, which	357	The library knows two types of such loops, the I<default> loop, which
302	supports signals and child events, and dynamically created event loops	358	supports child process events, and dynamically created event loops which
303	which do not.	359	do not.
304		360
305	=over 4	361	=over 4
306		362
307	=item struct ev_loop *ev_default_loop (unsigned int flags)	363	=item struct ev_loop *ev_default_loop (unsigned int flags)
308		364
309	This will initialise the default event loop if it hasn't been initialised	365	This returns the "default" event loop object, which is what you should
310	yet and return it. If the default loop could not be initialised, returns	366	normally use when you just need "the event loop". Event loop objects and
311	false. If it already was initialised it simply returns it (and ignores the	367	the C<flags> parameter are described in more detail in the entry for
312	flags. If that is troubling you, check C<ev_backend ()> afterwards).	368	C<ev_loop_new>.
		369
		370	If the default loop is already initialised then this function simply
		371	returns it (and ignores the flags. If that is troubling you, check
		372	C<ev_backend ()> afterwards). Otherwise it will create it with the given
		373	flags, which should almost always be C<0>, unless the caller is also the
		374	one calling C<ev_run> or otherwise qualifies as "the main program".
313		375
314	If you don't know what event loop to use, use the one returned from this	376	If you don't know what event loop to use, use the one returned from this
315	function.	377	function (or via the C<EV_DEFAULT> macro).
316		378
317	Note that this function is I<not> thread-safe, so if you want to use it	379	Note that this function is I<not> thread-safe, so if you want to use it
318	from multiple threads, you have to lock (note also that this is unlikely,	380	from multiple threads, you have to employ some kind of mutex (note also
319	as loops cannot be shared easily between threads anyway).	381	that this case is unlikely, as loops cannot be shared easily between
		382	threads anyway).
320		383
321	The default loop is the only loop that can handle C<ev_signal> and	384	The default loop is the only loop that can handle C<ev_child> watchers,
322	C<ev_child> watchers, and to do this, it always registers a handler	385	and to do this, it always registers a handler for C<SIGCHLD>. If this is
323	for C<SIGCHLD>. If this is a problem for your application you can either	386	a problem for your application you can either create a dynamic loop with
324	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you	387	C<ev_loop_new> which doesn't do that, or you can simply overwrite the
325	can simply overwrite the C<SIGCHLD> signal handler I<after> calling	388	C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
326	C<ev_default_init>.	389
		390	Example: This is the most typical usage.
		391
		392	if (!ev_default_loop (0))
		393	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
		394
		395	Example: Restrict libev to the select and poll backends, and do not allow
		396	environment settings to be taken into account:
		397
		398	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
		399
		400	=item struct ev_loop *ev_loop_new (unsigned int flags)
		401
		402	This will create and initialise a new event loop object. If the loop
		403	could not be initialised, returns false.
		404
		405	This function is thread-safe, and one common way to use libev with
		406	threads is indeed to create one loop per thread, and using the default
		407	loop in the "main" or "initial" thread.
327		408
328	The flags argument can be used to specify special behaviour or specific	409	The flags argument can be used to specify special behaviour or specific
329	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	410	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
330		411
331	The following flags are supported:	412	The following flags are supported:
…		…
341		422
342	If this flag bit is or'ed into the flag value (or the program runs setuid	423	If this flag bit is or'ed into the flag value (or the program runs setuid
343	or setgid) then libev will I<not> look at the environment variable	424	or setgid) then libev will I<not> look at the environment variable
344	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	425	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
345	override the flags completely if it is found in the environment. This is	426	override the flags completely if it is found in the environment. This is
346	useful to try out specific backends to test their performance, or to work	427	useful to try out specific backends to test their performance, to work
347	around bugs.	428	around bugs, or to make libev threadsafe (accessing environment variables
		429	cannot be done in a threadsafe way, but usually it works if no other
		430	thread modifies them).
348		431
349	=item C<EVFLAG_FORKCHECK>	432	=item C<EVFLAG_FORKCHECK>
350		433
351	Instead of calling C<ev_loop_fork> manually after a fork, you can also	434	Instead of calling C<ev_loop_fork> manually after a fork, you can also
352	make libev check for a fork in each iteration by enabling this flag.	435	make libev check for a fork in each iteration by enabling this flag.
353		436
354	This works by calling C<getpid ()> on every iteration of the loop,	437	This works by calling C<getpid ()> on every iteration of the loop,
355	and thus this might slow down your event loop if you do a lot of loop	438	and thus this might slow down your event loop if you do a lot of loop
356	iterations and little real work, but is usually not noticeable (on my	439	iterations and little real work, but is usually not noticeable (on my
357	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence	440	GNU/Linux system for example, C<getpid> is actually a simple 5-insn
358	without a system call and thus I<very> fast, but my GNU/Linux system also has	441	sequence without a system call and thus I<very> fast, but my GNU/Linux
359	C<pthread_atfork> which is even faster).	442	system also has C<pthread_atfork> which is even faster). (Update: glibc
		443	versions 2.25 apparently removed the C<getpid> optimisation again).
360		444
361	The big advantage of this flag is that you can forget about fork (and	445	The big advantage of this flag is that you can forget about fork (and
362	forget about forgetting to tell libev about forking) when you use this	446	forget about forgetting to tell libev about forking, although you still
363	flag.	447	have to ignore C<SIGPIPE>) when you use this flag.
364		448
365	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>	449	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
366	environment variable.	450	environment variable.
367		451
368	=item C<EVFLAG_NOINOTIFY>	452	=item C<EVFLAG_NOINOTIFY>
369		453
370	When this flag is specified, then libev will not attempt to use the	454	When this flag is specified, then libev will not attempt to use the
371	I<inotify> API for it's C<ev_stat> watchers. Apart from debugging and	455	I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
372	testing, this flag can be useful to conserve inotify file descriptors, as	456	testing, this flag can be useful to conserve inotify file descriptors, as
373	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.	457	otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
374		458
375	=item C<EVFLAG_SIGNALFD>	459	=item C<EVFLAG_SIGNALFD>
376		460
377	When this flag is specified, then libev will attempt to use the	461	When this flag is specified, then libev will attempt to use the
378	I<signalfd> API for it's C<ev_signal> (and C<ev_child>) watchers. This API	462	I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
379	delivers signals synchronously, which makes it both faster and might make	463	delivers signals synchronously, which makes it both faster and might make
380	it possible to get the queued signal data. It can also simplify signal	464	it possible to get the queued signal data. It can also simplify signal
381	handling with threads, as long as you properly block signals in your	465	handling with threads, as long as you properly block signals in your
382	threads that are not interested in handling them.	466	threads that are not interested in handling them.
383		467
384	Signalfd will not be used by default as this changes your signal mask, and	468	Signalfd will not be used by default as this changes your signal mask, and
385	there are a lot of shoddy libraries and programs (glib's threadpool for	469	there are a lot of shoddy libraries and programs (glib's threadpool for
386	example) that can't properly initialise their signal masks.	470	example) that can't properly initialise their signal masks.
		471
		472	=item C<EVFLAG_NOSIGMASK>
		473
		474	When this flag is specified, then libev will avoid to modify the signal
		475	mask. Specifically, this means you have to make sure signals are unblocked
		476	when you want to receive them.
		477
		478	This behaviour is useful when you want to do your own signal handling, or
		479	want to handle signals only in specific threads and want to avoid libev
		480	unblocking the signals.
		481
		482	It's also required by POSIX in a threaded program, as libev calls
		483	C<sigprocmask>, whose behaviour is officially unspecified.
		484
		485	This flag's behaviour will become the default in future versions of libev.
387		486
388	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	487	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
389		488
390	This is your standard select(2) backend. Not I<completely> standard, as	489	This is your standard select(2) backend. Not I<completely> standard, as
391	libev tries to roll its own fd_set with no limits on the number of fds,	490	libev tries to roll its own fd_set with no limits on the number of fds,
…		…
416	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and	515	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
417	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.	516	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
418		517
419	=item C<EVBACKEND_EPOLL> (value 4, Linux)	518	=item C<EVBACKEND_EPOLL> (value 4, Linux)
420		519
421	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9	520	Use the Linux-specific epoll(7) interface (for both pre- and post-2.6.9
422	kernels).	521	kernels).
423		522
424	For few fds, this backend is a bit little slower than poll and select,	523	For few fds, this backend is a bit little slower than poll and select, but
425	but it scales phenomenally better. While poll and select usually scale	524	it scales phenomenally better. While poll and select usually scale like
426	like O(total_fds) where n is the total number of fds (or the highest fd),	525	O(total_fds) where total_fds is the total number of fds (or the highest
427	epoll scales either O(1) or O(active_fds).	526	fd), epoll scales either O(1) or O(active_fds).
428		527
429	The epoll mechanism deserves honorable mention as the most misdesigned	528	The epoll mechanism deserves honorable mention as the most misdesigned
430	of the more advanced event mechanisms: mere annoyances include silently	529	of the more advanced event mechanisms: mere annoyances include silently
431	dropping file descriptors, requiring a system call per change per file	530	dropping file descriptors, requiring a system call per change per file
432	descriptor (and unnecessary guessing of parameters), problems with dup and	531	descriptor (and unnecessary guessing of parameters), problems with dup,
		532	returning before the timeout value, resulting in additional iterations
		533	(and only giving 5ms accuracy while select on the same platform gives
433	so on. The biggest issue is fork races, however - if a program forks then	534	0.1ms) and so on. The biggest issue is fork races, however - if a program
434	I<both> parent and child process have to recreate the epoll set, which can	535	forks then I<both> parent and child process have to recreate the epoll
435	take considerable time (one syscall per file descriptor) and is of course	536	set, which can take considerable time (one syscall per file descriptor)
436	hard to detect.	537	and is of course hard to detect.
437		538
438	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but	539	Epoll is also notoriously buggy - embedding epoll fds I<should> work,
439	of course I<doesn't>, and epoll just loves to report events for totally	540	but of course I<doesn't>, and epoll just loves to report events for
440	I<different> file descriptors (even already closed ones, so one cannot	541	totally I<different> file descriptors (even already closed ones, so
441	even remove them from the set) than registered in the set (especially	542	one cannot even remove them from the set) than registered in the set
442	on SMP systems). Libev tries to counter these spurious notifications by	543	(especially on SMP systems). Libev tries to counter these spurious
443	employing an additional generation counter and comparing that against the	544	notifications by employing an additional generation counter and comparing
444	events to filter out spurious ones, recreating the set when required. Last	545	that against the events to filter out spurious ones, recreating the set
		546	when required. Epoll also erroneously rounds down timeouts, but gives you
		547	no way to know when and by how much, so sometimes you have to busy-wait
		548	because epoll returns immediately despite a nonzero timeout. And last
445	not least, it also refuses to work with some file descriptors which work	549	not least, it also refuses to work with some file descriptors which work
446	perfectly fine with C<select> (files, many character devices...).	550	perfectly fine with C<select> (files, many character devices...).
		551
		552	Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
		553	cobbled together in a hurry, no thought to design or interaction with
		554	others. Oh, the pain, will it ever stop...
447		555
448	While stopping, setting and starting an I/O watcher in the same iteration	556	While stopping, setting and starting an I/O watcher in the same iteration
449	will result in some caching, there is still a system call per such	557	will result in some caching, there is still a system call per such
450	incident (because the same I<file descriptor> could point to a different	558	incident (because the same I<file descriptor> could point to a different
451	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed	559	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
…		…
463	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or	571	All this means that, in practice, C<EVBACKEND_SELECT> can be as fast or
464	faster than epoll for maybe up to a hundred file descriptors, depending on	572	faster than epoll for maybe up to a hundred file descriptors, depending on
465	the usage. So sad.	573	the usage. So sad.
466		574
467	While nominally embeddable in other event loops, this feature is broken in	575	While nominally embeddable in other event loops, this feature is broken in
468	all kernel versions tested so far.	576	a lot of kernel revisions, but probably(!) works in current versions.
469		577
470	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as	578	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
471	C<EVBACKEND_POLL>.	579	C<EVBACKEND_POLL>.
472		580
		581	=item C<EVBACKEND_LINUXAIO> (value 64, Linux)
		582
		583	Use the Linux-specific Linux AIO (I<not> C<< aio(7) >> but C<<
		584	io_submit(2) >>) event interface available in post-4.18 kernels (but libev
		585	only tries to use it in 4.19+).
		586
		587	This is another Linux train wreck of an event interface.
		588
		589	If this backend works for you (as of this writing, it was very
		590	experimental), it is the best event interface available on Linux and might
		591	be well worth enabling it - if it isn't available in your kernel this will
		592	be detected and this backend will be skipped.
		593
		594	This backend can batch oneshot requests and supports a user-space ring
		595	buffer to receive events. It also doesn't suffer from most of the design
		596	problems of epoll (such as not being able to remove event sources from
		597	the epoll set), and generally sounds too good to be true. Because, this
		598	being the Linux kernel, of course it suffers from a whole new set of
		599	limitations, forcing you to fall back to epoll, inheriting all its design
		600	issues.
		601
		602	For one, it is not easily embeddable (but probably could be done using
		603	an event fd at some extra overhead). It also is subject to a system wide
		604	limit that can be configured in F</proc/sys/fs/aio-max-nr>. If no AIO
		605	requests are left, this backend will be skipped during initialisation, and
		606	will switch to epoll when the loop is active.
		607
		608	Most problematic in practice, however, is that not all file descriptors
		609	work with it. For example, in Linux 5.1, TCP sockets, pipes, event fds,
		610	files, F</dev/null> and many others are supported, but ttys do not work
		611	properly (a known bug that the kernel developers don't care about, see
		612	L<https://lore.kernel.org/patchwork/patch/1047453/>), so this is not
		613	(yet?) a generic event polling interface.
		614
		615	Overall, it seems the Linux developers just don't want it to have a
		616	generic event handling mechanism other than C<select> or C<poll>.
		617
		618	To work around all these problem, the current version of libev uses its
		619	epoll backend as a fallback for file descriptor types that do not work. Or
		620	falls back completely to epoll if the kernel acts up.
		621
		622	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		623	C<EVBACKEND_POLL>.
		624
473	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	625	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
474		626
475	Kqueue deserves special mention, as at the time of this writing, it	627	Kqueue deserves special mention, as at the time this backend was
476	was broken on all BSDs except NetBSD (usually it doesn't work reliably	628	implemented, it was broken on all BSDs except NetBSD (usually it doesn't
477	with anything but sockets and pipes, except on Darwin, where of course	629	work reliably with anything but sockets and pipes, except on Darwin,
478	it's completely useless). Unlike epoll, however, whose brokenness	630	where of course it's completely useless). Unlike epoll, however, whose
479	is by design, these kqueue bugs can (and eventually will) be fixed	631	brokenness is by design, these kqueue bugs can be (and mostly have been)
480	without API changes to existing programs. For this reason it's not being	632	fixed without API changes to existing programs. For this reason it's not
481	"auto-detected" unless you explicitly specify it in the flags (i.e. using	633	being "auto-detected" on all platforms unless you explicitly specify it
482	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	634	in the flags (i.e. using C<EVBACKEND_KQUEUE>) or libev was compiled on a
483	system like NetBSD.	635	known-to-be-good (-enough) system like NetBSD.
484		636
485	You still can embed kqueue into a normal poll or select backend and use it	637	You still can embed kqueue into a normal poll or select backend and use it
486	only for sockets (after having made sure that sockets work with kqueue on	638	only for sockets (after having made sure that sockets work with kqueue on
487	the target platform). See C<ev_embed> watchers for more info.	639	the target platform). See C<ev_embed> watchers for more info.
488		640
489	It scales in the same way as the epoll backend, but the interface to the	641	It scales in the same way as the epoll backend, but the interface to the
490	kernel is more efficient (which says nothing about its actual speed, of	642	kernel is more efficient (which says nothing about its actual speed, of
491	course). While stopping, setting and starting an I/O watcher does never	643	course). While stopping, setting and starting an I/O watcher does never
492	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to	644	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
493	two event changes per incident. Support for C<fork ()> is very bad (but	645	two event changes per incident. Support for C<fork ()> is very bad (you
494	sane, unlike epoll) and it drops fds silently in similarly hard-to-detect	646	might have to leak fds on fork, but it's more sane than epoll) and it
495	cases	647	drops fds silently in similarly hard-to-detect cases.
496		648
497	This backend usually performs well under most conditions.	649	This backend usually performs well under most conditions.
498		650
499	While nominally embeddable in other event loops, this doesn't work	651	While nominally embeddable in other event loops, this doesn't work
500	everywhere, so you might need to test for this. And since it is broken	652	everywhere, so you might need to test for this. And since it is broken
…		…
517	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	669	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
518		670
519	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	671	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
520	it's really slow, but it still scales very well (O(active_fds)).	672	it's really slow, but it still scales very well (O(active_fds)).
521		673
522	Please note that Solaris event ports can deliver a lot of spurious
523	notifications, so you need to use non-blocking I/O or other means to avoid
524	blocking when no data (or space) is available.
525
526	While this backend scales well, it requires one system call per active	674	While this backend scales well, it requires one system call per active
527	file descriptor per loop iteration. For small and medium numbers of file	675	file descriptor per loop iteration. For small and medium numbers of file
528	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend	676	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
529	might perform better.	677	might perform better.
530		678
531	On the positive side, with the exception of the spurious readiness	679	On the positive side, this backend actually performed fully to
532	notifications, this backend actually performed fully to specification
533	in all tests and is fully embeddable, which is a rare feat among the	680	specification in all tests and is fully embeddable, which is a rare feat
534	OS-specific backends (I vastly prefer correctness over speed hacks).	681	among the OS-specific backends (I vastly prefer correctness over speed
		682	hacks).
		683
		684	On the negative side, the interface is I<bizarre> - so bizarre that
		685	even sun itself gets it wrong in their code examples: The event polling
		686	function sometimes returns events to the caller even though an error
		687	occurred, but with no indication whether it has done so or not (yes, it's
		688	even documented that way) - deadly for edge-triggered interfaces where you
		689	absolutely have to know whether an event occurred or not because you have
		690	to re-arm the watcher.
		691
		692	Fortunately libev seems to be able to work around these idiocies.
535		693
536	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as	694	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
537	C<EVBACKEND_POLL>.	695	C<EVBACKEND_POLL>.
538		696
539	=item C<EVBACKEND_ALL>	697	=item C<EVBACKEND_ALL>
540		698
541	Try all backends (even potentially broken ones that wouldn't be tried	699	Try all backends (even potentially broken ones that wouldn't be tried
542	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	700	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
543	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	701	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
544		702
545	It is definitely not recommended to use this flag.	703	It is definitely not recommended to use this flag, use whatever
		704	C<ev_recommended_backends ()> returns, or simply do not specify a backend
		705	at all.
		706
		707	=item C<EVBACKEND_MASK>
		708
		709	Not a backend at all, but a mask to select all backend bits from a
		710	C<flags> value, in case you want to mask out any backends from a flags
		711	value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
546		712
547	=back	713	=back
548		714
549	If one or more of the backend flags are or'ed into the flags value,	715	If one or more of the backend flags are or'ed into the flags value,
550	then only these backends will be tried (in the reverse order as listed	716	then only these backends will be tried (in the reverse order as listed
551	here). If none are specified, all backends in C<ev_recommended_backends	717	here). If none are specified, all backends in C<ev_recommended_backends
552	()> will be tried.	718	()> will be tried.
553		719
554	Example: This is the most typical usage.
555
556	if (!ev_default_loop (0))
557	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
558
559	Example: Restrict libev to the select and poll backends, and do not allow
560	environment settings to be taken into account:
561
562	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
563
564	Example: Use whatever libev has to offer, but make sure that kqueue is
565	used if available (warning, breaks stuff, best use only with your own
566	private event loop and only if you know the OS supports your types of
567	fds):
568
569	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
570
571	=item struct ev_loop *ev_loop_new (unsigned int flags)
572
573	Similar to C<ev_default_loop>, but always creates a new event loop that is
574	always distinct from the default loop.
575
576	Note that this function I<is> thread-safe, and one common way to use
577	libev with threads is indeed to create one loop per thread, and using the
578	default loop in the "main" or "initial" thread.
579
580	Example: Try to create a event loop that uses epoll and nothing else.	720	Example: Try to create a event loop that uses epoll and nothing else.
581		721
582	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	722	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
583	if (!epoller)	723	if (!epoller)
584	fatal ("no epoll found here, maybe it hides under your chair");	724	fatal ("no epoll found here, maybe it hides under your chair");
585		725
		726	Example: Use whatever libev has to offer, but make sure that kqueue is
		727	used if available.
		728
		729	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
		730
		731	Example: Similarly, on linux, you mgiht want to take advantage of the
		732	linux aio backend if possible, but fall back to something else if that
		733	isn't available.
		734
		735	struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_LINUXAIO);
		736
586	=item ev_default_destroy ()	737	=item ev_loop_destroy (loop)
587		738
588	Destroys the default loop (frees all memory and kernel state etc.). None	739	Destroys an event loop object (frees all memory and kernel state
589	of the active event watchers will be stopped in the normal sense, so	740	etc.). None of the active event watchers will be stopped in the normal
590	e.g. C<ev_is_active> might still return true. It is your responsibility to	741	sense, so e.g. C<ev_is_active> might still return true. It is your
591	either stop all watchers cleanly yourself I<before> calling this function,	742	responsibility to either stop all watchers cleanly yourself I<before>
592	or cope with the fact afterwards (which is usually the easiest thing, you	743	calling this function, or cope with the fact afterwards (which is usually
593	can just ignore the watchers and/or C<free ()> them for example).	744	the easiest thing, you can just ignore the watchers and/or C<free ()> them
		745	for example).
594		746
595	Note that certain global state, such as signal state (and installed signal	747	Note that certain global state, such as signal state (and installed signal
596	handlers), will not be freed by this function, and related watchers (such	748	handlers), will not be freed by this function, and related watchers (such
597	as signal and child watchers) would need to be stopped manually.	749	as signal and child watchers) would need to be stopped manually.
598		750
599	In general it is not advisable to call this function except in the	751	This function is normally used on loop objects allocated by
600	rare occasion where you really need to free e.g. the signal handling	752	C<ev_loop_new>, but it can also be used on the default loop returned by
		753	C<ev_default_loop>, in which case it is not thread-safe.
		754
		755	Note that it is not advisable to call this function on the default loop
		756	except in the rare occasion where you really need to free its resources.
601	pipe fds. If you need dynamically allocated loops it is better to use	757	If you need dynamically allocated loops it is better to use C<ev_loop_new>
602	C<ev_loop_new> and C<ev_loop_destroy>.	758	and C<ev_loop_destroy>.
603		759
604	=item ev_loop_destroy (loop)	760	=item ev_loop_fork (loop)
605
606	Like C<ev_default_destroy>, but destroys an event loop created by an
607	earlier call to C<ev_loop_new>.
608
609	=item ev_default_fork ()
610		761
611	This function sets a flag that causes subsequent C<ev_run> iterations	762	This function sets a flag that causes subsequent C<ev_run> iterations
612	to reinitialise the kernel state for backends that have one. Despite the	763	to reinitialise the kernel state for backends that have one. Despite
613	name, you can call it anytime, but it makes most sense after forking, in	764	the name, you can call it anytime you are allowed to start or stop
614	the child process (or both child and parent, but that again makes little	765	watchers (except inside an C<ev_prepare> callback), but it makes most
615	sense). You I<must> call it in the child before using any of the libev	766	sense after forking, in the child process. You I<must> call it (or use
616	functions, and it will only take effect at the next C<ev_run> iteration.	767	C<EVFLAG_FORKCHECK>) in the child before resuming or calling C<ev_run>.
617		768
		769	In addition, if you want to reuse a loop (via this function or
		770	C<EVFLAG_FORKCHECK>), you I<also> have to ignore C<SIGPIPE>.
		771
618	Again, you I<have> to call it on I<any> loop that you want to re-use after	772	Again, you I<have> to call it on I<any> loop that you want to re-use after
619	a fork, I<even if you do not plan to use the loop in the parent>. This is	773	a fork, I<even if you do not plan to use the loop in the parent>. This is
620	because some kernel interfaces *cough* I<kqueue> *cough* do funny things	774	because some kernel interfaces *cough* I<kqueue> *cough* do funny things
621	during fork.	775	during fork.
622		776
623	On the other hand, you only need to call this function in the child	777	On the other hand, you only need to call this function in the child
…		…
626	call it at all (in fact, C<epoll> is so badly broken that it makes a	780	call it at all (in fact, C<epoll> is so badly broken that it makes a
627	difference, but libev will usually detect this case on its own and do a	781	difference, but libev will usually detect this case on its own and do a
628	costly reset of the backend).	782	costly reset of the backend).
629		783
630	The function itself is quite fast and it's usually not a problem to call	784	The function itself is quite fast and it's usually not a problem to call
631	it just in case after a fork. To make this easy, the function will fit in	785	it just in case after a fork.
632	quite nicely into a call to C<pthread_atfork>:
633		786
		787	Example: Automate calling C<ev_loop_fork> on the default loop when
		788	using pthreads.
		789
		790	static void
		791	post_fork_child (void)
		792	{
		793	ev_loop_fork (EV_DEFAULT);
		794	}
		795
		796	...
634	pthread_atfork (0, 0, ev_default_fork);	797	pthread_atfork (0, 0, post_fork_child);
635
636	=item ev_loop_fork (loop)
637
638	Like C<ev_default_fork>, but acts on an event loop created by
639	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
640	after fork that you want to re-use in the child, and how you keep track of
641	them is entirely your own problem.
642		798
643	=item int ev_is_default_loop (loop)	799	=item int ev_is_default_loop (loop)
644		800
645	Returns true when the given loop is, in fact, the default loop, and false	801	Returns true when the given loop is, in fact, the default loop, and false
646	otherwise.	802	otherwise.
…		…
657	prepare and check phases.	813	prepare and check phases.
658		814
659	=item unsigned int ev_depth (loop)	815	=item unsigned int ev_depth (loop)
660		816
661	Returns the number of times C<ev_run> was entered minus the number of	817	Returns the number of times C<ev_run> was entered minus the number of
662	times C<ev_run> was exited, in other words, the recursion depth.	818	times C<ev_run> was exited normally, in other words, the recursion depth.
663		819
664	Outside C<ev_run>, this number is zero. In a callback, this number is	820	Outside C<ev_run>, this number is zero. In a callback, this number is
665	C<1>, unless C<ev_run> was invoked recursively (or from another thread),	821	C<1>, unless C<ev_run> was invoked recursively (or from another thread),
666	in which case it is higher.	822	in which case it is higher.
667		823
668	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread	824	Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
669	etc.), doesn't count as "exit" - consider this as a hint to avoid such	825	throwing an exception etc.), doesn't count as "exit" - consider this
670	ungentleman-like behaviour unless it's really convenient.	826	as a hint to avoid such ungentleman-like behaviour unless it's really
		827	convenient, in which case it is fully supported.
671		828
672	=item unsigned int ev_backend (loop)	829	=item unsigned int ev_backend (loop)
673		830
674	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	831	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
675	use.	832	use.
…		…
690		847
691	This function is rarely useful, but when some event callback runs for a	848	This function is rarely useful, but when some event callback runs for a
692	very long time without entering the event loop, updating libev's idea of	849	very long time without entering the event loop, updating libev's idea of
693	the current time is a good idea.	850	the current time is a good idea.
694		851
695	See also L<The special problem of time updates> in the C<ev_timer> section.	852	See also L</The special problem of time updates> in the C<ev_timer> section.
696		853
697	=item ev_suspend (loop)	854	=item ev_suspend (loop)
698		855
699	=item ev_resume (loop)	856	=item ev_resume (loop)
700		857
…		…
718	without a previous call to C<ev_suspend>.	875	without a previous call to C<ev_suspend>.
719		876
720	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the	877	Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
721	event loop time (see C<ev_now_update>).	878	event loop time (see C<ev_now_update>).
722		879
723	=item ev_run (loop, int flags)	880	=item bool ev_run (loop, int flags)
724		881
725	Finally, this is it, the event handler. This function usually is called	882	Finally, this is it, the event handler. This function usually is called
726	after you have initialised all your watchers and you want to start	883	after you have initialised all your watchers and you want to start
727	handling events. It will ask the operating system for any new events, call	884	handling events. It will ask the operating system for any new events, call
728	the watcher callbacks, an then repeat the whole process indefinitely: This	885	the watcher callbacks, and then repeat the whole process indefinitely: This
729	is why event loops are called I<loops>.	886	is why event loops are called I<loops>.
730		887
731	If the flags argument is specified as C<0>, it will keep handling events	888	If the flags argument is specified as C<0>, it will keep handling events
732	until either no event watchers are active anymore or C<ev_break> was	889	until either no event watchers are active anymore or C<ev_break> was
733	called.	890	called.
		891
		892	The return value is false if there are no more active watchers (which
		893	usually means "all jobs done" or "deadlock"), and true in all other cases
		894	(which usually means " you should call C<ev_run> again").
734		895
735	Please note that an explicit C<ev_break> is usually better than	896	Please note that an explicit C<ev_break> is usually better than
736	relying on all watchers to be stopped when deciding when a program has	897	relying on all watchers to be stopped when deciding when a program has
737	finished (especially in interactive programs), but having a program	898	finished (especially in interactive programs), but having a program
738	that automatically loops as long as it has to and no longer by virtue	899	that automatically loops as long as it has to and no longer by virtue
739	of relying on its watchers stopping correctly, that is truly a thing of	900	of relying on its watchers stopping correctly, that is truly a thing of
740	beauty.	901	beauty.
741		902
		903	This function is I<mostly> exception-safe - you can break out of a
		904	C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
		905	exception and so on. This does not decrement the C<ev_depth> value, nor
		906	will it clear any outstanding C<EVBREAK_ONE> breaks.
		907
742	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle	908	A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
743	those events and any already outstanding ones, but will not wait and	909	those events and any already outstanding ones, but will not wait and
744	block your process in case there are no events and will return after one	910	block your process in case there are no events and will return after one
745	iteration of the loop. This is sometimes useful to poll and handle new	911	iteration of the loop. This is sometimes useful to poll and handle new
746	events while doing lengthy calculations, to keep the program responsive.	912	events while doing lengthy calculations, to keep the program responsive.
…		…
755	This is useful if you are waiting for some external event in conjunction	921	This is useful if you are waiting for some external event in conjunction
756	with something not expressible using other libev watchers (i.e. "roll your	922	with something not expressible using other libev watchers (i.e. "roll your
757	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is	923	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
758	usually a better approach for this kind of thing.	924	usually a better approach for this kind of thing.
759		925
760	Here are the gory details of what C<ev_run> does:	926	Here are the gory details of what C<ev_run> does (this is for your
		927	understanding, not a guarantee that things will work exactly like this in
		928	future versions):
761		929
762	- Increment loop depth.	930	- Increment loop depth.
763	- Reset the ev_break status.	931	- Reset the ev_break status.
764	- Before the first iteration, call any pending watchers.	932	- Before the first iteration, call any pending watchers.
765	LOOP:	933	LOOP:
…		…
798	anymore.	966	anymore.
799		967
800	... queue jobs here, make sure they register event watchers as long	968	... queue jobs here, make sure they register event watchers as long
801	... as they still have work to do (even an idle watcher will do..)	969	... as they still have work to do (even an idle watcher will do..)
802	ev_run (my_loop, 0);	970	ev_run (my_loop, 0);
803	... jobs done or somebody called unloop. yeah!	971	... jobs done or somebody called break. yeah!
804		972
805	=item ev_break (loop, how)	973	=item ev_break (loop, how)
806		974
807	Can be used to make a call to C<ev_run> return early (but only after it	975	Can be used to make a call to C<ev_run> return early (but only after it
808	has processed all outstanding events). The C<how> argument must be either	976	has processed all outstanding events). The C<how> argument must be either
809	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or	977	C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
810	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.	978	C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
811		979
812	This "unloop state" will be cleared when entering C<ev_run> again.	980	This "break state" will be cleared on the next call to C<ev_run>.
813		981
814	It is safe to call C<ev_break> from outside any C<ev_run> calls. ##TODO##	982	It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
		983	which case it will have no effect.
815		984
816	=item ev_ref (loop)	985	=item ev_ref (loop)
817		986
818	=item ev_unref (loop)	987	=item ev_unref (loop)
819		988
…		…
840	running when nothing else is active.	1009	running when nothing else is active.
841		1010
842	ev_signal exitsig;	1011	ev_signal exitsig;
843	ev_signal_init (&exitsig, sig_cb, SIGINT);	1012	ev_signal_init (&exitsig, sig_cb, SIGINT);
844	ev_signal_start (loop, &exitsig);	1013	ev_signal_start (loop, &exitsig);
845	evf_unref (loop);	1014	ev_unref (loop);
846		1015
847	Example: For some weird reason, unregister the above signal handler again.	1016	Example: For some weird reason, unregister the above signal handler again.
848		1017
849	ev_ref (loop);	1018	ev_ref (loop);
850	ev_signal_stop (loop, &exitsig);	1019	ev_signal_stop (loop, &exitsig);
…		…
870	overhead for the actual polling but can deliver many events at once.	1039	overhead for the actual polling but can deliver many events at once.
871		1040
872	By setting a higher I<io collect interval> you allow libev to spend more	1041	By setting a higher I<io collect interval> you allow libev to spend more
873	time collecting I/O events, so you can handle more events per iteration,	1042	time collecting I/O events, so you can handle more events per iteration,
874	at the cost of increasing latency. Timeouts (both C<ev_periodic> and	1043	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
875	C<ev_timer>) will be not affected. Setting this to a non-null value will	1044	C<ev_timer>) will not be affected. Setting this to a non-null value will
876	introduce an additional C<ev_sleep ()> call into most loop iterations. The	1045	introduce an additional C<ev_sleep ()> call into most loop iterations. The
877	sleep time ensures that libev will not poll for I/O events more often then	1046	sleep time ensures that libev will not poll for I/O events more often then
878	once per this interval, on average.	1047	once per this interval, on average (as long as the host time resolution is
		1048	good enough).
879		1049
880	Likewise, by setting a higher I<timeout collect interval> you allow libev	1050	Likewise, by setting a higher I<timeout collect interval> you allow libev
881	to spend more time collecting timeouts, at the expense of increased	1051	to spend more time collecting timeouts, at the expense of increased
882	latency/jitter/inexactness (the watcher callback will be called	1052	latency/jitter/inexactness (the watcher callback will be called
883	later). C<ev_io> watchers will not be affected. Setting this to a non-null	1053	later). C<ev_io> watchers will not be affected. Setting this to a non-null
…		…
908		1078
909	=item ev_invoke_pending (loop)	1079	=item ev_invoke_pending (loop)
910		1080
911	This call will simply invoke all pending watchers while resetting their	1081	This call will simply invoke all pending watchers while resetting their
912	pending state. Normally, C<ev_run> does this automatically when required,	1082	pending state. Normally, C<ev_run> does this automatically when required,
913	but when overriding the invoke callback this call comes handy.	1083	but when overriding the invoke callback this call comes handy. This
		1084	function can be invoked from a watcher - this can be useful for example
		1085	when you want to do some lengthy calculation and want to pass further
		1086	event handling to another thread (you still have to make sure only one
		1087	thread executes within C<ev_invoke_pending> or C<ev_run> of course).
914		1088
915	=item int ev_pending_count (loop)	1089	=item int ev_pending_count (loop)
916		1090
917	Returns the number of pending watchers - zero indicates that no watchers	1091	Returns the number of pending watchers - zero indicates that no watchers
918	are pending.	1092	are pending.
…		…
925	invoke the actual watchers inside another context (another thread etc.).	1099	invoke the actual watchers inside another context (another thread etc.).
926		1100
927	If you want to reset the callback, use C<ev_invoke_pending> as new	1101	If you want to reset the callback, use C<ev_invoke_pending> as new
928	callback.	1102	callback.
929		1103
930	=item ev_set_loop_release_cb (loop, void (*release)(EV_P), void (*acquire)(EV_P))	1104	=item ev_set_loop_release_cb (loop, void (*release)(EV_P) throw (), void (*acquire)(EV_P) throw ())
931		1105
932	Sometimes you want to share the same loop between multiple threads. This	1106	Sometimes you want to share the same loop between multiple threads. This
933	can be done relatively simply by putting mutex_lock/unlock calls around	1107	can be done relatively simply by putting mutex_lock/unlock calls around
934	each call to a libev function.	1108	each call to a libev function.
935		1109
936	However, C<ev_run> can run an indefinite time, so it is not feasible	1110	However, C<ev_run> can run an indefinite time, so it is not feasible
937	to wait for it to return. One way around this is to wake up the event	1111	to wait for it to return. One way around this is to wake up the event
938	loop via C<ev_break> and C<av_async_send>, another way is to set these	1112	loop via C<ev_break> and C<ev_async_send>, another way is to set these
939	I<release> and I<acquire> callbacks on the loop.	1113	I<release> and I<acquire> callbacks on the loop.
940		1114
941	When set, then C<release> will be called just before the thread is	1115	When set, then C<release> will be called just before the thread is
942	suspended waiting for new events, and C<acquire> is called just	1116	suspended waiting for new events, and C<acquire> is called just
943	afterwards.	1117	afterwards.
…		…
958	See also the locking example in the C<THREADS> section later in this	1132	See also the locking example in the C<THREADS> section later in this
959	document.	1133	document.
960		1134
961	=item ev_set_userdata (loop, void *data)	1135	=item ev_set_userdata (loop, void *data)
962		1136
963	=item ev_userdata (loop)	1137	=item void *ev_userdata (loop)
964		1138
965	Set and retrieve a single C<void *> associated with a loop. When	1139	Set and retrieve a single C<void *> associated with a loop. When
966	C<ev_set_userdata> has never been called, then C<ev_userdata> returns	1140	C<ev_set_userdata> has never been called, then C<ev_userdata> returns
967	C<0.>	1141	C<0>.
968		1142
969	These two functions can be used to associate arbitrary data with a loop,	1143	These two functions can be used to associate arbitrary data with a loop,
970	and are intended solely for the C<invoke_pending_cb>, C<release> and	1144	and are intended solely for the C<invoke_pending_cb>, C<release> and
971	C<acquire> callbacks described above, but of course can be (ab-)used for	1145	C<acquire> callbacks described above, but of course can be (ab-)used for
972	any other purpose as well.	1146	any other purpose as well.
…		…
990		1164
991	In the following description, uppercase C<TYPE> in names stands for the	1165	In the following description, uppercase C<TYPE> in names stands for the
992	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer	1166	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
993	watchers and C<ev_io_start> for I/O watchers.	1167	watchers and C<ev_io_start> for I/O watchers.
994		1168
995	A watcher is a structure that you create and register to record your	1169	A watcher is an opaque structure that you allocate and register to record
996	interest in some event. For instance, if you want to wait for STDIN to	1170	your interest in some event. To make a concrete example, imagine you want
997	become readable, you would create an C<ev_io> watcher for that:	1171	to wait for STDIN to become readable, you would create an C<ev_io> watcher
		1172	for that:
998		1173
999	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)	1174	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
1000	{	1175	{
1001	ev_io_stop (w);	1176	ev_io_stop (w);
1002	ev_break (loop, EVBREAK_ALL);	1177	ev_break (loop, EVBREAK_ALL);
…		…
1017	stack).	1192	stack).
1018		1193
1019	Each watcher has an associated watcher structure (called C<struct ev_TYPE>	1194	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
1020	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).	1195	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
1021		1196
1022	Each watcher structure must be initialised by a call to C<ev_init	1197	Each watcher structure must be initialised by a call to C<ev_init (watcher
1023	(watcher *, callback)>, which expects a callback to be provided. This	1198	*, callback)>, which expects a callback to be provided. This callback is
1024	callback gets invoked each time the event occurs (or, in the case of I/O	1199	invoked each time the event occurs (or, in the case of I/O watchers, each
1025	watchers, each time the event loop detects that the file descriptor given	1200	time the event loop detects that the file descriptor given is readable
1026	is readable and/or writable).	1201	and/or writable).
1027		1202
1028	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>	1203	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
1029	macro to configure it, with arguments specific to the watcher type. There	1204	macro to configure it, with arguments specific to the watcher type. There
1030	is also a macro to combine initialisation and setting in one call: C<<	1205	is also a macro to combine initialisation and setting in one call: C<<
1031	ev_TYPE_init (watcher *, callback, ...) >>.	1206	ev_TYPE_init (watcher *, callback, ...) >>.
…		…
1082		1257
1083	=item C<EV_PREPARE>	1258	=item C<EV_PREPARE>
1084		1259
1085	=item C<EV_CHECK>	1260	=item C<EV_CHECK>
1086		1261
1087	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts	1262	All C<ev_prepare> watchers are invoked just I<before> C<ev_run> starts to
1088	to gather new events, and all C<ev_check> watchers are invoked just after	1263	gather new events, and all C<ev_check> watchers are queued (not invoked)
1089	C<ev_run> has gathered them, but before it invokes any callbacks for any	1264	just after C<ev_run> has gathered them, but before it queues any callbacks
		1265	for any received events. That means C<ev_prepare> watchers are the last
		1266	watchers invoked before the event loop sleeps or polls for new events, and
		1267	C<ev_check> watchers will be invoked before any other watchers of the same
		1268	or lower priority within an event loop iteration.
		1269
1090	received events. Callbacks of both watcher types can start and stop as	1270	Callbacks of both watcher types can start and stop as many watchers as
1091	many watchers as they want, and all of them will be taken into account	1271	they want, and all of them will be taken into account (for example, a
1092	(for example, a C<ev_prepare> watcher might start an idle watcher to keep	1272	C<ev_prepare> watcher might start an idle watcher to keep C<ev_run> from
1093	C<ev_run> from blocking).	1273	blocking).
1094		1274
1095	=item C<EV_EMBED>	1275	=item C<EV_EMBED>
1096		1276
1097	The embedded event loop specified in the C<ev_embed> watcher needs attention.	1277	The embedded event loop specified in the C<ev_embed> watcher needs attention.
1098		1278
1099	=item C<EV_FORK>	1279	=item C<EV_FORK>
1100		1280
1101	The event loop has been resumed in the child process after fork (see	1281	The event loop has been resumed in the child process after fork (see
1102	C<ev_fork>).	1282	C<ev_fork>).
		1283
		1284	=item C<EV_CLEANUP>
		1285
		1286	The event loop is about to be destroyed (see C<ev_cleanup>).
1103		1287
1104	=item C<EV_ASYNC>	1288	=item C<EV_ASYNC>
1105		1289
1106	The given async watcher has been asynchronously notified (see C<ev_async>).	1290	The given async watcher has been asynchronously notified (see C<ev_async>).
1107		1291
…		…
1217		1401
1218	=item callback ev_cb (ev_TYPE *watcher)	1402	=item callback ev_cb (ev_TYPE *watcher)
1219		1403
1220	Returns the callback currently set on the watcher.	1404	Returns the callback currently set on the watcher.
1221		1405
1222	=item ev_cb_set (ev_TYPE *watcher, callback)	1406	=item ev_set_cb (ev_TYPE *watcher, callback)
1223		1407
1224	Change the callback. You can change the callback at virtually any time	1408	Change the callback. You can change the callback at virtually any time
1225	(modulo threads).	1409	(modulo threads).
1226		1410
1227	=item ev_set_priority (ev_TYPE *watcher, int priority)	1411	=item ev_set_priority (ev_TYPE *watcher, int priority)
…		…
1245	or might not have been clamped to the valid range.	1429	or might not have been clamped to the valid range.
1246		1430
1247	The default priority used by watchers when no priority has been set is	1431	The default priority used by watchers when no priority has been set is
1248	always C<0>, which is supposed to not be too high and not be too low :).	1432	always C<0>, which is supposed to not be too high and not be too low :).
1249		1433
1250	See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of	1434	See L</WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
1251	priorities.	1435	priorities.
1252		1436
1253	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1437	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
1254		1438
1255	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1439	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
…		…
1280	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related	1464	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1281	functions that do not need a watcher.	1465	functions that do not need a watcher.
1282		1466
1283	=back	1467	=back
1284		1468
		1469	See also the L</ASSOCIATING CUSTOM DATA WITH A WATCHER> and L</BUILDING YOUR
		1470	OWN COMPOSITE WATCHERS> idioms.
1285		1471
1286	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1472	=head2 WATCHER STATES
1287		1473
1288	Each watcher has, by default, a member C<void *data> that you can change	1474	There are various watcher states mentioned throughout this manual -
1289	and read at any time: libev will completely ignore it. This can be used	1475	active, pending and so on. In this section these states and the rules to
1290	to associate arbitrary data with your watcher. If you need more data and	1476	transition between them will be described in more detail - and while these
1291	don't want to allocate memory and store a pointer to it in that data	1477	rules might look complicated, they usually do "the right thing".
1292	member, you can also "subclass" the watcher type and provide your own
1293	data:
1294		1478
1295	struct my_io	1479	=over 4
1296	{
1297	ev_io io;
1298	int otherfd;
1299	void *somedata;
1300	struct whatever *mostinteresting;
1301	};
1302		1480
1303	...	1481	=item initialised
1304	struct my_io w;
1305	ev_io_init (&w.io, my_cb, fd, EV_READ);
1306		1482
1307	And since your callback will be called with a pointer to the watcher, you	1483	Before a watcher can be registered with the event loop it has to be
1308	can cast it back to your own type:	1484	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
		1485	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
1309		1486
1310	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)	1487	In this state it is simply some block of memory that is suitable for
1311	{	1488	use in an event loop. It can be moved around, freed, reused etc. at
1312	struct my_io *w = (struct my_io *)w_;	1489	will - as long as you either keep the memory contents intact, or call
1313	...	1490	C<ev_TYPE_init> again.
1314	}
1315		1491
1316	More interesting and less C-conformant ways of casting your callback type	1492	=item started/running/active
1317	instead have been omitted.
1318		1493
1319	Another common scenario is to use some data structure with multiple	1494	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
1320	embedded watchers:	1495	property of the event loop, and is actively waiting for events. While in
		1496	this state it cannot be accessed (except in a few documented ways), moved,
		1497	freed or anything else - the only legal thing is to keep a pointer to it,
		1498	and call libev functions on it that are documented to work on active watchers.
1321		1499
1322	struct my_biggy	1500	=item pending
1323	{
1324	int some_data;
1325	ev_timer t1;
1326	ev_timer t2;
1327	}
1328		1501
1329	In this case getting the pointer to C<my_biggy> is a bit more	1502	If a watcher is active and libev determines that an event it is interested
1330	complicated: Either you store the address of your C<my_biggy> struct	1503	in has occurred (such as a timer expiring), it will become pending. It will
1331	in the C<data> member of the watcher (for woozies), or you need to use	1504	stay in this pending state until either it is stopped or its callback is
1332	some pointer arithmetic using C<offsetof> inside your watchers (for real	1505	about to be invoked, so it is not normally pending inside the watcher
1333	programmers):	1506	callback.
1334		1507
1335	#include <stddef.h>	1508	The watcher might or might not be active while it is pending (for example,
		1509	an expired non-repeating timer can be pending but no longer active). If it
		1510	is stopped, it can be freely accessed (e.g. by calling C<ev_TYPE_set>),
		1511	but it is still property of the event loop at this time, so cannot be
		1512	moved, freed or reused. And if it is active the rules described in the
		1513	previous item still apply.
1336		1514
1337	static void	1515	It is also possible to feed an event on a watcher that is not active (e.g.
1338	t1_cb (EV_P_ ev_timer *w, int revents)	1516	via C<ev_feed_event>), in which case it becomes pending without being
1339	{	1517	active.
1340	struct my_biggy big = (struct my_biggy *)
1341	(((char *)w) - offsetof (struct my_biggy, t1));
1342	}
1343		1518
1344	static void	1519	=item stopped
1345	t2_cb (EV_P_ ev_timer *w, int revents)	1520
1346	{	1521	A watcher can be stopped implicitly by libev (in which case it might still
1347	struct my_biggy big = (struct my_biggy *)	1522	be pending), or explicitly by calling its C<ev_TYPE_stop> function. The
1348	(((char *)w) - offsetof (struct my_biggy, t2));	1523	latter will clear any pending state the watcher might be in, regardless
1349	}	1524	of whether it was active or not, so stopping a watcher explicitly before
		1525	freeing it is often a good idea.
		1526
		1527	While stopped (and not pending) the watcher is essentially in the
		1528	initialised state, that is, it can be reused, moved, modified in any way
		1529	you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
		1530	it again).
		1531
		1532	=back
1350		1533
1351	=head2 WATCHER PRIORITY MODELS	1534	=head2 WATCHER PRIORITY MODELS
1352		1535
1353	Many event loops support I<watcher priorities>, which are usually small	1536	Many event loops support I<watcher priorities>, which are usually small
1354	integers that influence the ordering of event callback invocation	1537	integers that influence the ordering of event callback invocation
1355	between watchers in some way, all else being equal.	1538	between watchers in some way, all else being equal.
1356		1539
1357	In libev, Watcher priorities can be set using C<ev_set_priority>. See its	1540	In libev, watcher priorities can be set using C<ev_set_priority>. See its
1358	description for the more technical details such as the actual priority	1541	description for the more technical details such as the actual priority
1359	range.	1542	range.
1360		1543
1361	There are two common ways how these these priorities are being interpreted	1544	There are two common ways how these these priorities are being interpreted
1362	by event loops:	1545	by event loops:
…		…
1481	In general you can register as many read and/or write event watchers per	1664	In general you can register as many read and/or write event watchers per
1482	fd as you want (as long as you don't confuse yourself). Setting all file	1665	fd as you want (as long as you don't confuse yourself). Setting all file
1483	descriptors to non-blocking mode is also usually a good idea (but not	1666	descriptors to non-blocking mode is also usually a good idea (but not
1484	required if you know what you are doing).	1667	required if you know what you are doing).
1485		1668
1486	If you cannot use non-blocking mode, then force the use of a
1487	known-to-be-good backend (at the time of this writing, this includes only
1488	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
1489	descriptors for which non-blocking operation makes no sense (such as
1490	files) - libev doesn't guarantee any specific behaviour in that case.
1491
1492	Another thing you have to watch out for is that it is quite easy to	1669	Another thing you have to watch out for is that it is quite easy to
1493	receive "spurious" readiness notifications, that is your callback might	1670	receive "spurious" readiness notifications, that is, your callback might
1494	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1671	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1495	because there is no data. Not only are some backends known to create a	1672	because there is no data. It is very easy to get into this situation even
1496	lot of those (for example Solaris ports), it is very easy to get into	1673	with a relatively standard program structure. Thus it is best to always
1497	this situation even with a relatively standard program structure. Thus	1674	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
1498	it is best to always use non-blocking I/O: An extra C<read>(2) returning
1499	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1675	preferable to a program hanging until some data arrives.
1500		1676
1501	If you cannot run the fd in non-blocking mode (for example you should	1677	If you cannot run the fd in non-blocking mode (for example you should
1502	not play around with an Xlib connection), then you have to separately	1678	not play around with an Xlib connection), then you have to separately
1503	re-test whether a file descriptor is really ready with a known-to-be good	1679	re-test whether a file descriptor is really ready with a known-to-be good
1504	interface such as poll (fortunately in our Xlib example, Xlib already	1680	interface such as poll (fortunately in the case of Xlib, it already does
1505	does this on its own, so its quite safe to use). Some people additionally	1681	this on its own, so its quite safe to use). Some people additionally
1506	use C<SIGALRM> and an interval timer, just to be sure you won't block	1682	use C<SIGALRM> and an interval timer, just to be sure you won't block
1507	indefinitely.	1683	indefinitely.
1508		1684
1509	But really, best use non-blocking mode.	1685	But really, best use non-blocking mode.
1510		1686
1511	=head3 The special problem of disappearing file descriptors	1687	=head3 The special problem of disappearing file descriptors
1512		1688
1513	Some backends (e.g. kqueue, epoll) need to be told about closing a file	1689	Some backends (e.g. kqueue, epoll, linuxaio) need to be told about closing
1514	descriptor (either due to calling C<close> explicitly or any other means,	1690	a file descriptor (either due to calling C<close> explicitly or any other
1515	such as C<dup2>). The reason is that you register interest in some file	1691	means, such as C<dup2>). The reason is that you register interest in some
1516	descriptor, but when it goes away, the operating system will silently drop	1692	file descriptor, but when it goes away, the operating system will silently
1517	this interest. If another file descriptor with the same number then is	1693	drop this interest. If another file descriptor with the same number then
1518	registered with libev, there is no efficient way to see that this is, in	1694	is registered with libev, there is no efficient way to see that this is,
1519	fact, a different file descriptor.	1695	in fact, a different file descriptor.
1520		1696
1521	To avoid having to explicitly tell libev about such cases, libev follows	1697	To avoid having to explicitly tell libev about such cases, libev follows
1522	the following policy: Each time C<ev_io_set> is being called, libev	1698	the following policy: Each time C<ev_io_set> is being called, libev
1523	will assume that this is potentially a new file descriptor, otherwise	1699	will assume that this is potentially a new file descriptor, otherwise
1524	it is assumed that the file descriptor stays the same. That means that	1700	it is assumed that the file descriptor stays the same. That means that
…		…
1538		1714
1539	There is no workaround possible except not registering events	1715	There is no workaround possible except not registering events
1540	for potentially C<dup ()>'ed file descriptors, or to resort to	1716	for potentially C<dup ()>'ed file descriptors, or to resort to
1541	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.	1717	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1542		1718
		1719	=head3 The special problem of files
		1720
		1721	Many people try to use C<select> (or libev) on file descriptors
		1722	representing files, and expect it to become ready when their program
		1723	doesn't block on disk accesses (which can take a long time on their own).
		1724
		1725	However, this cannot ever work in the "expected" way - you get a readiness
		1726	notification as soon as the kernel knows whether and how much data is
		1727	there, and in the case of open files, that's always the case, so you
		1728	always get a readiness notification instantly, and your read (or possibly
		1729	write) will still block on the disk I/O.
		1730
		1731	Another way to view it is that in the case of sockets, pipes, character
		1732	devices and so on, there is another party (the sender) that delivers data
		1733	on its own, but in the case of files, there is no such thing: the disk
		1734	will not send data on its own, simply because it doesn't know what you
		1735	wish to read - you would first have to request some data.
		1736
		1737	Since files are typically not-so-well supported by advanced notification
		1738	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1739	to files, even though you should not use it. The reason for this is
		1740	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1741	usually a tty, often a pipe, but also sometimes files or special devices
		1742	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1743	F</dev/urandom>), and even though the file might better be served with
		1744	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1745	it "just works" instead of freezing.
		1746
		1747	So avoid file descriptors pointing to files when you know it (e.g. use
		1748	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1749	when you rarely read from a file instead of from a socket, and want to
		1750	reuse the same code path.
		1751
1543	=head3 The special problem of fork	1752	=head3 The special problem of fork
1544		1753
1545	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit	1754	Some backends (epoll, kqueue, linuxaio, iouring) do not support C<fork ()>
1546	useless behaviour. Libev fully supports fork, but needs to be told about	1755	at all or exhibit useless behaviour. Libev fully supports fork, but needs
1547	it in the child.	1756	to be told about it in the child if you want to continue to use it in the
		1757	child.
1548		1758
1549	To support fork in your programs, you either have to call	1759	To support fork in your child processes, you have to call C<ev_loop_fork
1550	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,	1760	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
1551	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or	1761	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1552	C<EVBACKEND_POLL>.
1553		1762
1554	=head3 The special problem of SIGPIPE	1763	=head3 The special problem of SIGPIPE
1555		1764
1556	While not really specific to libev, it is easy to forget about C<SIGPIPE>:	1765	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1557	when writing to a pipe whose other end has been closed, your program gets	1766	when writing to a pipe whose other end has been closed, your program gets
…		…
1655	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1864	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1656	monotonic clock option helps a lot here).	1865	monotonic clock option helps a lot here).
1657		1866
1658	The callback is guaranteed to be invoked only I<after> its timeout has	1867	The callback is guaranteed to be invoked only I<after> its timeout has
1659	passed (not I<at>, so on systems with very low-resolution clocks this	1868	passed (not I<at>, so on systems with very low-resolution clocks this
1660	might introduce a small delay). If multiple timers become ready during the	1869	might introduce a small delay, see "the special problem of being too
		1870	early", below). If multiple timers become ready during the same loop
1661	same loop iteration then the ones with earlier time-out values are invoked	1871	iteration then the ones with earlier time-out values are invoked before
1662	before ones of the same priority with later time-out values (but this is	1872	ones of the same priority with later time-out values (but this is no
1663	no longer true when a callback calls C<ev_run> recursively).	1873	longer true when a callback calls C<ev_run> recursively).
1664		1874
1665	=head3 Be smart about timeouts	1875	=head3 Be smart about timeouts
1666		1876
1667	Many real-world problems involve some kind of timeout, usually for error	1877	Many real-world problems involve some kind of timeout, usually for error
1668	recovery. A typical example is an HTTP request - if the other side hangs,	1878	recovery. A typical example is an HTTP request - if the other side hangs,
…		…
1743		1953
1744	In this case, it would be more efficient to leave the C<ev_timer> alone,	1954	In this case, it would be more efficient to leave the C<ev_timer> alone,
1745	but remember the time of last activity, and check for a real timeout only	1955	but remember the time of last activity, and check for a real timeout only
1746	within the callback:	1956	within the callback:
1747		1957
		1958	ev_tstamp timeout = 60.;
1748	ev_tstamp last_activity; // time of last activity	1959	ev_tstamp last_activity; // time of last activity
		1960	ev_timer timer;
1749		1961
1750	static void	1962	static void
1751	callback (EV_P_ ev_timer *w, int revents)	1963	callback (EV_P_ ev_timer *w, int revents)
1752	{	1964	{
1753	ev_tstamp now = ev_now (EV_A);	1965	// calculate when the timeout would happen
1754	ev_tstamp timeout = last_activity + 60.;	1966	ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
1755		1967
1756	// if last_activity + 60. is older than now, we did time out	1968	// if negative, it means we the timeout already occurred
1757	if (timeout < now)	1969	if (after < 0.)
1758	{	1970	{
1759	// timeout occurred, take action	1971	// timeout occurred, take action
1760	}	1972	}
1761	else	1973	else
1762	{	1974	{
1763	// callback was invoked, but there was some activity, re-arm	1975	// callback was invoked, but there was some recent
1764	// the watcher to fire in last_activity + 60, which is	1976	// activity. simply restart the timer to time out
1765	// guaranteed to be in the future, so "again" is positive:	1977	// after "after" seconds, which is the earliest time
1766	w->repeat = timeout - now;	1978	// the timeout can occur.
		1979	ev_timer_set (w, after, 0.);
1767	ev_timer_again (EV_A_ w);	1980	ev_timer_start (EV_A_ w);
1768	}	1981	}
1769	}	1982	}
1770		1983
1771	To summarise the callback: first calculate the real timeout (defined	1984	To summarise the callback: first calculate in how many seconds the
1772	as "60 seconds after the last activity"), then check if that time has	1985	timeout will occur (by calculating the absolute time when it would occur,
1773	been reached, which means something I<did>, in fact, time out. Otherwise	1986	C<last_activity + timeout>, and subtracting the current time, C<ev_now
1774	the callback was invoked too early (C<timeout> is in the future), so	1987	(EV_A)> from that).
1775	re-schedule the timer to fire at that future time, to see if maybe we have
1776	a timeout then.
1777		1988
1778	Note how C<ev_timer_again> is used, taking advantage of the	1989	If this value is negative, then we are already past the timeout, i.e. we
1779	C<ev_timer_again> optimisation when the timer is already running.	1990	timed out, and need to do whatever is needed in this case.
		1991
		1992	Otherwise, we now the earliest time at which the timeout would trigger,
		1993	and simply start the timer with this timeout value.
		1994
		1995	In other words, each time the callback is invoked it will check whether
		1996	the timeout occurred. If not, it will simply reschedule itself to check
		1997	again at the earliest time it could time out. Rinse. Repeat.
1780		1998
1781	This scheme causes more callback invocations (about one every 60 seconds	1999	This scheme causes more callback invocations (about one every 60 seconds
1782	minus half the average time between activity), but virtually no calls to	2000	minus half the average time between activity), but virtually no calls to
1783	libev to change the timeout.	2001	libev to change the timeout.
1784		2002
1785	To start the timer, simply initialise the watcher and set C<last_activity>	2003	To start the machinery, simply initialise the watcher and set
1786	to the current time (meaning we just have some activity :), then call the	2004	C<last_activity> to the current time (meaning there was some activity just
1787	callback, which will "do the right thing" and start the timer:	2005	now), then call the callback, which will "do the right thing" and start
		2006	the timer:
1788		2007
		2008	last_activity = ev_now (EV_A);
1789	ev_init (timer, callback);	2009	ev_init (&timer, callback);
1790	last_activity = ev_now (loop);	2010	callback (EV_A_ &timer, 0);
1791	callback (loop, timer, EV_TIMER);
1792		2011
1793	And when there is some activity, simply store the current time in	2012	When there is some activity, simply store the current time in
1794	C<last_activity>, no libev calls at all:	2013	C<last_activity>, no libev calls at all:
1795		2014
		2015	if (activity detected)
1796	last_activity = ev_now (loop);	2016	last_activity = ev_now (EV_A);
		2017
		2018	When your timeout value changes, then the timeout can be changed by simply
		2019	providing a new value, stopping the timer and calling the callback, which
		2020	will again do the right thing (for example, time out immediately :).
		2021
		2022	timeout = new_value;
		2023	ev_timer_stop (EV_A_ &timer);
		2024	callback (EV_A_ &timer, 0);
1797		2025
1798	This technique is slightly more complex, but in most cases where the	2026	This technique is slightly more complex, but in most cases where the
1799	time-out is unlikely to be triggered, much more efficient.	2027	time-out is unlikely to be triggered, much more efficient.
1800
1801	Changing the timeout is trivial as well (if it isn't hard-coded in the
1802	callback :) - just change the timeout and invoke the callback, which will
1803	fix things for you.
1804		2028
1805	=item 4. Wee, just use a double-linked list for your timeouts.	2029	=item 4. Wee, just use a double-linked list for your timeouts.
1806		2030
1807	If there is not one request, but many thousands (millions...), all	2031	If there is not one request, but many thousands (millions...), all
1808	employing some kind of timeout with the same timeout value, then one can	2032	employing some kind of timeout with the same timeout value, then one can
…		…
1835	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is	2059	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
1836	rather complicated, but extremely efficient, something that really pays	2060	rather complicated, but extremely efficient, something that really pays
1837	off after the first million or so of active timers, i.e. it's usually	2061	off after the first million or so of active timers, i.e. it's usually
1838	overkill :)	2062	overkill :)
1839		2063
		2064	=head3 The special problem of being too early
		2065
		2066	If you ask a timer to call your callback after three seconds, then
		2067	you expect it to be invoked after three seconds - but of course, this
		2068	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		2069	guaranteed to any precision by libev - imagine somebody suspending the
		2070	process with a STOP signal for a few hours for example.
		2071
		2072	So, libev tries to invoke your callback as soon as possible I<after> the
		2073	delay has occurred, but cannot guarantee this.
		2074
		2075	A less obvious failure mode is calling your callback too early: many event
		2076	loops compare timestamps with a "elapsed delay >= requested delay", but
		2077	this can cause your callback to be invoked much earlier than you would
		2078	expect.
		2079
		2080	To see why, imagine a system with a clock that only offers full second
		2081	resolution (think windows if you can't come up with a broken enough OS
		2082	yourself). If you schedule a one-second timer at the time 500.9, then the
		2083	event loop will schedule your timeout to elapse at a system time of 500
		2084	(500.9 truncated to the resolution) + 1, or 501.
		2085
		2086	If an event library looks at the timeout 0.1s later, it will see "501 >=
		2087	501" and invoke the callback 0.1s after it was started, even though a
		2088	one-second delay was requested - this is being "too early", despite best
		2089	intentions.
		2090
		2091	This is the reason why libev will never invoke the callback if the elapsed
		2092	delay equals the requested delay, but only when the elapsed delay is
		2093	larger than the requested delay. In the example above, libev would only invoke
		2094	the callback at system time 502, or 1.1s after the timer was started.
		2095
		2096	So, while libev cannot guarantee that your callback will be invoked
		2097	exactly when requested, it I<can> and I<does> guarantee that the requested
		2098	delay has actually elapsed, or in other words, it always errs on the "too
		2099	late" side of things.
		2100
1840	=head3 The special problem of time updates	2101	=head3 The special problem of time updates
1841		2102
1842	Establishing the current time is a costly operation (it usually takes at	2103	Establishing the current time is a costly operation (it usually takes
1843	least two system calls): EV therefore updates its idea of the current	2104	at least one system call): EV therefore updates its idea of the current
1844	time only before and after C<ev_run> collects new events, which causes a	2105	time only before and after C<ev_run> collects new events, which causes a
1845	growing difference between C<ev_now ()> and C<ev_time ()> when handling	2106	growing difference between C<ev_now ()> and C<ev_time ()> when handling
1846	lots of events in one iteration.	2107	lots of events in one iteration.
1847		2108
1848	The relative timeouts are calculated relative to the C<ev_now ()>	2109	The relative timeouts are calculated relative to the C<ev_now ()>
1849	time. This is usually the right thing as this timestamp refers to the time	2110	time. This is usually the right thing as this timestamp refers to the time
1850	of the event triggering whatever timeout you are modifying/starting. If	2111	of the event triggering whatever timeout you are modifying/starting. If
1851	you suspect event processing to be delayed and you I<need> to base the	2112	you suspect event processing to be delayed and you I<need> to base the
1852	timeout on the current time, use something like this to adjust for this:	2113	timeout on the current time, use something like the following to adjust
		2114	for it:
1853		2115
1854	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2116	ev_timer_set (&timer, after + (ev_time () - ev_now ()), 0.);
1855		2117
1856	If the event loop is suspended for a long time, you can also force an	2118	If the event loop is suspended for a long time, you can also force an
1857	update of the time returned by C<ev_now ()> by calling C<ev_now_update	2119	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1858	()>.	2120	()>, although that will push the event time of all outstanding events
		2121	further into the future.
		2122
		2123	=head3 The special problem of unsynchronised clocks
		2124
		2125	Modern systems have a variety of clocks - libev itself uses the normal
		2126	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2127	jumps).
		2128
		2129	Neither of these clocks is synchronised with each other or any other clock
		2130	on the system, so C<ev_time ()> might return a considerably different time
		2131	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2132	a call to C<gettimeofday> might return a second count that is one higher
		2133	than a directly following call to C<time>.
		2134
		2135	The moral of this is to only compare libev-related timestamps with
		2136	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2137	a second or so.
		2138
		2139	One more problem arises due to this lack of synchronisation: if libev uses
		2140	the system monotonic clock and you compare timestamps from C<ev_time>
		2141	or C<ev_now> from when you started your timer and when your callback is
		2142	invoked, you will find that sometimes the callback is a bit "early".
		2143
		2144	This is because C<ev_timer>s work in real time, not wall clock time, so
		2145	libev makes sure your callback is not invoked before the delay happened,
		2146	I<measured according to the real time>, not the system clock.
		2147
		2148	If your timeouts are based on a physical timescale (e.g. "time out this
		2149	connection after 100 seconds") then this shouldn't bother you as it is
		2150	exactly the right behaviour.
		2151
		2152	If you want to compare wall clock/system timestamps to your timers, then
		2153	you need to use C<ev_periodic>s, as these are based on the wall clock
		2154	time, where your comparisons will always generate correct results.
1859		2155
1860	=head3 The special problems of suspended animation	2156	=head3 The special problems of suspended animation
1861		2157
1862	When you leave the server world it is quite customary to hit machines that	2158	When you leave the server world it is quite customary to hit machines that
1863	can suspend/hibernate - what happens to the clocks during such a suspend?	2159	can suspend/hibernate - what happens to the clocks during such a suspend?
…		…
1893		2189
1894	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	2190	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
1895		2191
1896	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	2192	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
1897		2193
1898	Configure the timer to trigger after C<after> seconds. If C<repeat>	2194	Configure the timer to trigger after C<after> seconds (fractional and
1899	is C<0.>, then it will automatically be stopped once the timeout is	2195	negative values are supported). If C<repeat> is C<0.>, then it will
1900	reached. If it is positive, then the timer will automatically be	2196	automatically be stopped once the timeout is reached. If it is positive,
1901	configured to trigger again C<repeat> seconds later, again, and again,	2197	then the timer will automatically be configured to trigger again C<repeat>
1902	until stopped manually.	2198	seconds later, again, and again, until stopped manually.
1903		2199
1904	The timer itself will do a best-effort at avoiding drift, that is, if	2200	The timer itself will do a best-effort at avoiding drift, that is, if
1905	you configure a timer to trigger every 10 seconds, then it will normally	2201	you configure a timer to trigger every 10 seconds, then it will normally
1906	trigger at exactly 10 second intervals. If, however, your program cannot	2202	trigger at exactly 10 second intervals. If, however, your program cannot
1907	keep up with the timer (because it takes longer than those 10 seconds to	2203	keep up with the timer (because it takes longer than those 10 seconds to
1908	do stuff) the timer will not fire more than once per event loop iteration.	2204	do stuff) the timer will not fire more than once per event loop iteration.
1909		2205
1910	=item ev_timer_again (loop, ev_timer *)	2206	=item ev_timer_again (loop, ev_timer *)
1911		2207
1912	This will act as if the timer timed out and restart it again if it is	2208	This will act as if the timer timed out, and restarts it again if it is
1913	repeating. The exact semantics are:	2209	repeating. It basically works like calling C<ev_timer_stop>, updating the
		2210	timeout to the C<repeat> value and calling C<ev_timer_start>.
1914		2211
		2212	The exact semantics are as in the following rules, all of which will be
		2213	applied to the watcher:
		2214
		2215	=over 4
		2216
1915	If the timer is pending, its pending status is cleared.	2217	=item If the timer is pending, the pending status is always cleared.
1916		2218
1917	If the timer is started but non-repeating, stop it (as if it timed out).	2219	=item If the timer is started but non-repeating, stop it (as if it timed
		2220	out, without invoking it).
1918		2221
1919	If the timer is repeating, either start it if necessary (with the	2222	=item If the timer is repeating, make the C<repeat> value the new timeout
1920	C<repeat> value), or reset the running timer to the C<repeat> value.	2223	and start the timer, if necessary.
1921		2224
		2225	=back
		2226
1922	This sounds a bit complicated, see L<Be smart about timeouts>, above, for a	2227	This sounds a bit complicated, see L</Be smart about timeouts>, above, for a
1923	usage example.	2228	usage example.
1924		2229
1925	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)	2230	=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
1926		2231
1927	Returns the remaining time until a timer fires. If the timer is active,	2232	Returns the remaining time until a timer fires. If the timer is active,
…		…
1980	Periodic watchers are also timers of a kind, but they are very versatile	2285	Periodic watchers are also timers of a kind, but they are very versatile
1981	(and unfortunately a bit complex).	2286	(and unfortunately a bit complex).
1982		2287
1983	Unlike C<ev_timer>, periodic watchers are not based on real time (or	2288	Unlike C<ev_timer>, periodic watchers are not based on real time (or
1984	relative time, the physical time that passes) but on wall clock time	2289	relative time, the physical time that passes) but on wall clock time
1985	(absolute time, the thing you can read on your calender or clock). The	2290	(absolute time, the thing you can read on your calendar or clock). The
1986	difference is that wall clock time can run faster or slower than real	2291	difference is that wall clock time can run faster or slower than real
1987	time, and time jumps are not uncommon (e.g. when you adjust your	2292	time, and time jumps are not uncommon (e.g. when you adjust your
1988	wrist-watch).	2293	wrist-watch).
1989		2294
1990	You can tell a periodic watcher to trigger after some specific point	2295	You can tell a periodic watcher to trigger after some specific point
…		…
1995	C<ev_timer>, which would still trigger roughly 10 seconds after starting	2300	C<ev_timer>, which would still trigger roughly 10 seconds after starting
1996	it, as it uses a relative timeout).	2301	it, as it uses a relative timeout).
1997		2302
1998	C<ev_periodic> watchers can also be used to implement vastly more complex	2303	C<ev_periodic> watchers can also be used to implement vastly more complex
1999	timers, such as triggering an event on each "midnight, local time", or	2304	timers, such as triggering an event on each "midnight, local time", or
2000	other complicated rules. This cannot be done with C<ev_timer> watchers, as	2305	other complicated rules. This cannot easily be done with C<ev_timer>
2001	those cannot react to time jumps.	2306	watchers, as those cannot react to time jumps.
2002		2307
2003	As with timers, the callback is guaranteed to be invoked only when the	2308	As with timers, the callback is guaranteed to be invoked only when the
2004	point in time where it is supposed to trigger has passed. If multiple	2309	point in time where it is supposed to trigger has passed. If multiple
2005	timers become ready during the same loop iteration then the ones with	2310	timers become ready during the same loop iteration then the ones with
2006	earlier time-out values are invoked before ones with later time-out values	2311	earlier time-out values are invoked before ones with later time-out values
…		…
2047		2352
2048	Another way to think about it (for the mathematically inclined) is that	2353	Another way to think about it (for the mathematically inclined) is that
2049	C<ev_periodic> will try to run the callback in this mode at the next possible	2354	C<ev_periodic> will try to run the callback in this mode at the next possible
2050	time where C<time = offset (mod interval)>, regardless of any time jumps.	2355	time where C<time = offset (mod interval)>, regardless of any time jumps.
2051		2356
2052	For numerical stability it is preferable that the C<offset> value is near	2357	The C<interval> I<MUST> be positive, and for numerical stability, the
2053	C<ev_now ()> (the current time), but there is no range requirement for	2358	interval value should be higher than C<1/8192> (which is around 100
2054	this value, and in fact is often specified as zero.	2359	microseconds) and C<offset> should be higher than C<0> and should have
		2360	at most a similar magnitude as the current time (say, within a factor of
		2361	ten). Typical values for offset are, in fact, C<0> or something between
		2362	C<0> and C<interval>, which is also the recommended range.
2055		2363
2056	Note also that there is an upper limit to how often a timer can fire (CPU	2364	Note also that there is an upper limit to how often a timer can fire (CPU
2057	speed for example), so if C<interval> is very small then timing stability	2365	speed for example), so if C<interval> is very small then timing stability
2058	will of course deteriorate. Libev itself tries to be exact to be about one	2366	will of course deteriorate. Libev itself tries to be exact to be about one
2059	millisecond (if the OS supports it and the machine is fast enough).	2367	millisecond (if the OS supports it and the machine is fast enough).
…		…
2089		2397
2090	NOTE: I<< This callback must always return a time that is higher than or	2398	NOTE: I<< This callback must always return a time that is higher than or
2091	equal to the passed C<now> value >>.	2399	equal to the passed C<now> value >>.
2092		2400
2093	This can be used to create very complex timers, such as a timer that	2401	This can be used to create very complex timers, such as a timer that
2094	triggers on "next midnight, local time". To do this, you would calculate the	2402	triggers on "next midnight, local time". To do this, you would calculate
2095	next midnight after C<now> and return the timestamp value for this. How	2403	the next midnight after C<now> and return the timestamp value for
2096	you do this is, again, up to you (but it is not trivial, which is the main	2404	this. Here is a (completely untested, no error checking) example on how to
2097	reason I omitted it as an example).	2405	do this:
		2406
		2407	#include <time.h>
		2408
		2409	static ev_tstamp
		2410	my_rescheduler (ev_periodic *w, ev_tstamp now)
		2411	{
		2412	time_t tnow = (time_t)now;
		2413	struct tm tm;
		2414	localtime_r (&tnow, &tm);
		2415
		2416	tm.tm_sec = tm.tm_min = tm.tm_hour = 0; // midnight current day
		2417	++tm.tm_mday; // midnight next day
		2418
		2419	return mktime (&tm);
		2420	}
		2421
		2422	Note: this code might run into trouble on days that have more then two
		2423	midnights (beginning and end).
2098		2424
2099	=back	2425	=back
2100		2426
2101	=item ev_periodic_again (loop, ev_periodic *)	2427	=item ev_periodic_again (loop, ev_periodic *)
2102		2428
…		…
2167		2493
2168	ev_periodic hourly_tick;	2494	ev_periodic hourly_tick;
2169	ev_periodic_init (&hourly_tick, clock_cb,	2495	ev_periodic_init (&hourly_tick, clock_cb,
2170	fmod (ev_now (loop), 3600.), 3600., 0);	2496	fmod (ev_now (loop), 3600.), 3600., 0);
2171	ev_periodic_start (loop, &hourly_tick);	2497	ev_periodic_start (loop, &hourly_tick);
2172		2498
2173		2499
2174	=head2 C<ev_signal> - signal me when a signal gets signalled!	2500	=head2 C<ev_signal> - signal me when a signal gets signalled!
2175		2501
2176	Signal watchers will trigger an event when the process receives a specific	2502	Signal watchers will trigger an event when the process receives a specific
2177	signal one or more times. Even though signals are very asynchronous, libev	2503	signal one or more times. Even though signals are very asynchronous, libev
2178	will try it's best to deliver signals synchronously, i.e. as part of the	2504	will try its best to deliver signals synchronously, i.e. as part of the
2179	normal event processing, like any other event.	2505	normal event processing, like any other event.
2180		2506
2181	If you want signals to be delivered truly asynchronously, just use	2507	If you want signals to be delivered truly asynchronously, just use
2182	C<sigaction> as you would do without libev and forget about sharing	2508	C<sigaction> as you would do without libev and forget about sharing
2183	the signal. You can even use C<ev_async> from a signal handler to	2509	the signal. You can even use C<ev_async> from a signal handler to
…		…
2187	only within the same loop, i.e. you can watch for C<SIGINT> in your	2513	only within the same loop, i.e. you can watch for C<SIGINT> in your
2188	default loop and for C<SIGIO> in another loop, but you cannot watch for	2514	default loop and for C<SIGIO> in another loop, but you cannot watch for
2189	C<SIGINT> in both the default loop and another loop at the same time. At	2515	C<SIGINT> in both the default loop and another loop at the same time. At
2190	the moment, C<SIGCHLD> is permanently tied to the default loop.	2516	the moment, C<SIGCHLD> is permanently tied to the default loop.
2191		2517
2192	When the first watcher gets started will libev actually register something	2518	Only after the first watcher for a signal is started will libev actually
2193	with the kernel (thus it coexists with your own signal handlers as long as	2519	register something with the kernel. It thus coexists with your own signal
2194	you don't register any with libev for the same signal).	2520	handlers as long as you don't register any with libev for the same signal.
2195		2521
2196	If possible and supported, libev will install its handlers with	2522	If possible and supported, libev will install its handlers with
2197	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should	2523	C<SA_RESTART> (or equivalent) behaviour enabled, so system calls should
2198	not be unduly interrupted. If you have a problem with system calls getting	2524	not be unduly interrupted. If you have a problem with system calls getting
2199	interrupted by signals you can block all signals in an C<ev_check> watcher	2525	interrupted by signals you can block all signals in an C<ev_check> watcher
…		…
2202	=head3 The special problem of inheritance over fork/execve/pthread_create	2528	=head3 The special problem of inheritance over fork/execve/pthread_create
2203		2529
2204	Both the signal mask (C<sigprocmask>) and the signal disposition	2530	Both the signal mask (C<sigprocmask>) and the signal disposition
2205	(C<sigaction>) are unspecified after starting a signal watcher (and after	2531	(C<sigaction>) are unspecified after starting a signal watcher (and after
2206	stopping it again), that is, libev might or might not block the signal,	2532	stopping it again), that is, libev might or might not block the signal,
2207	and might or might not set or restore the installed signal handler.	2533	and might or might not set or restore the installed signal handler (but
		2534	see C<EVFLAG_NOSIGMASK>).
2208		2535
2209	While this does not matter for the signal disposition (libev never	2536	While this does not matter for the signal disposition (libev never
2210	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on	2537	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
2211	C<execve>), this matters for the signal mask: many programs do not expect	2538	C<execve>), this matters for the signal mask: many programs do not expect
2212	certain signals to be blocked.	2539	certain signals to be blocked.
…		…
2225	I<has> to modify the signal mask, at least temporarily.	2552	I<has> to modify the signal mask, at least temporarily.
2226		2553
2227	So I can't stress this enough: I<If you do not reset your signal mask when	2554	So I can't stress this enough: I<If you do not reset your signal mask when
2228	you expect it to be empty, you have a race condition in your code>. This	2555	you expect it to be empty, you have a race condition in your code>. This
2229	is not a libev-specific thing, this is true for most event libraries.	2556	is not a libev-specific thing, this is true for most event libraries.
		2557
		2558	=head3 The special problem of threads signal handling
		2559
		2560	POSIX threads has problematic signal handling semantics, specifically,
		2561	a lot of functionality (sigfd, sigwait etc.) only really works if all
		2562	threads in a process block signals, which is hard to achieve.
		2563
		2564	When you want to use sigwait (or mix libev signal handling with your own
		2565	for the same signals), you can tackle this problem by globally blocking
		2566	all signals before creating any threads (or creating them with a fully set
		2567	sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
		2568	loops. Then designate one thread as "signal receiver thread" which handles
		2569	these signals. You can pass on any signals that libev might be interested
		2570	in by calling C<ev_feed_signal>.
2230		2571
2231	=head3 Watcher-Specific Functions and Data Members	2572	=head3 Watcher-Specific Functions and Data Members
2232		2573
2233	=over 4	2574	=over 4
2234		2575
…		…
2369		2710
2370	=head2 C<ev_stat> - did the file attributes just change?	2711	=head2 C<ev_stat> - did the file attributes just change?
2371		2712
2372	This watches a file system path for attribute changes. That is, it calls	2713	This watches a file system path for attribute changes. That is, it calls
2373	C<stat> on that path in regular intervals (or when the OS says it changed)	2714	C<stat> on that path in regular intervals (or when the OS says it changed)
2374	and sees if it changed compared to the last time, invoking the callback if	2715	and sees if it changed compared to the last time, invoking the callback
2375	it did.	2716	if it did. Starting the watcher C<stat>'s the file, so only changes that
		2717	happen after the watcher has been started will be reported.
2376		2718
2377	The path does not need to exist: changing from "path exists" to "path does	2719	The path does not need to exist: changing from "path exists" to "path does
2378	not exist" is a status change like any other. The condition "path does not	2720	not exist" is a status change like any other. The condition "path does not
2379	exist" (or more correctly "path cannot be stat'ed") is signified by the	2721	exist" (or more correctly "path cannot be stat'ed") is signified by the
2380	C<st_nlink> field being zero (which is otherwise always forced to be at	2722	C<st_nlink> field being zero (which is otherwise always forced to be at
…		…
2610	Apart from keeping your process non-blocking (which is a useful	2952	Apart from keeping your process non-blocking (which is a useful
2611	effect on its own sometimes), idle watchers are a good place to do	2953	effect on its own sometimes), idle watchers are a good place to do
2612	"pseudo-background processing", or delay processing stuff to after the	2954	"pseudo-background processing", or delay processing stuff to after the
2613	event loop has handled all outstanding events.	2955	event loop has handled all outstanding events.
2614		2956
		2957	=head3 Abusing an C<ev_idle> watcher for its side-effect
		2958
		2959	As long as there is at least one active idle watcher, libev will never
		2960	sleep unnecessarily. Or in other words, it will loop as fast as possible.
		2961	For this to work, the idle watcher doesn't need to be invoked at all - the
		2962	lowest priority will do.
		2963
		2964	This mode of operation can be useful together with an C<ev_check> watcher,
		2965	to do something on each event loop iteration - for example to balance load
		2966	between different connections.
		2967
		2968	See L</Abusing an ev_check watcher for its side-effect> for a longer
		2969	example.
		2970
2615	=head3 Watcher-Specific Functions and Data Members	2971	=head3 Watcher-Specific Functions and Data Members
2616		2972
2617	=over 4	2973	=over 4
2618		2974
2619	=item ev_idle_init (ev_idle *, callback)	2975	=item ev_idle_init (ev_idle *, callback)
…		…
2630	callback, free it. Also, use no error checking, as usual.	2986	callback, free it. Also, use no error checking, as usual.
2631		2987
2632	static void	2988	static void
2633	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)	2989	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
2634	{	2990	{
		2991	// stop the watcher
		2992	ev_idle_stop (loop, w);
		2993
		2994	// now we can free it
2635	free (w);	2995	free (w);
		2996
2636	// now do something you wanted to do when the program has	2997	// now do something you wanted to do when the program has
2637	// no longer anything immediate to do.	2998	// no longer anything immediate to do.
2638	}	2999	}
2639		3000
2640	ev_idle *idle_watcher = malloc (sizeof (ev_idle));	3001	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
…		…
2642	ev_idle_start (loop, idle_watcher);	3003	ev_idle_start (loop, idle_watcher);
2643		3004
2644		3005
2645	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	3006	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
2646		3007
2647	Prepare and check watchers are usually (but not always) used in pairs:	3008	Prepare and check watchers are often (but not always) used in pairs:
2648	prepare watchers get invoked before the process blocks and check watchers	3009	prepare watchers get invoked before the process blocks and check watchers
2649	afterwards.	3010	afterwards.
2650		3011
2651	You I<must not> call C<ev_run> or similar functions that enter	3012	You I<must not> call C<ev_run> (or similar functions that enter the
2652	the current event loop from either C<ev_prepare> or C<ev_check>	3013	current event loop) or C<ev_loop_fork> from either C<ev_prepare> or
2653	watchers. Other loops than the current one are fine, however. The	3014	C<ev_check> watchers. Other loops than the current one are fine,
2654	rationale behind this is that you do not need to check for recursion in	3015	however. The rationale behind this is that you do not need to check
2655	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	3016	for recursion in those watchers, i.e. the sequence will always be
2656	C<ev_check> so if you have one watcher of each kind they will always be	3017	C<ev_prepare>, blocking, C<ev_check> so if you have one watcher of each
2657	called in pairs bracketing the blocking call.	3018	kind they will always be called in pairs bracketing the blocking call.
2658		3019
2659	Their main purpose is to integrate other event mechanisms into libev and	3020	Their main purpose is to integrate other event mechanisms into libev and
2660	their use is somewhat advanced. They could be used, for example, to track	3021	their use is somewhat advanced. They could be used, for example, to track
2661	variable changes, implement your own watchers, integrate net-snmp or a	3022	variable changes, implement your own watchers, integrate net-snmp or a
2662	coroutine library and lots more. They are also occasionally useful if	3023	coroutine library and lots more. They are also occasionally useful if
…		…
2680	with priority higher than or equal to the event loop and one coroutine	3041	with priority higher than or equal to the event loop and one coroutine
2681	of lower priority, but only once, using idle watchers to keep the event	3042	of lower priority, but only once, using idle watchers to keep the event
2682	loop from blocking if lower-priority coroutines are active, thus mapping	3043	loop from blocking if lower-priority coroutines are active, thus mapping
2683	low-priority coroutines to idle/background tasks).	3044	low-priority coroutines to idle/background tasks).
2684		3045
2685	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	3046	When used for this purpose, it is recommended to give C<ev_check> watchers
2686	priority, to ensure that they are being run before any other watchers	3047	highest (C<EV_MAXPRI>) priority, to ensure that they are being run before
2687	after the poll (this doesn't matter for C<ev_prepare> watchers).	3048	any other watchers after the poll (this doesn't matter for C<ev_prepare>
		3049	watchers).
2688		3050
2689	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not	3051	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
2690	activate ("feed") events into libev. While libev fully supports this, they	3052	activate ("feed") events into libev. While libev fully supports this, they
2691	might get executed before other C<ev_check> watchers did their job. As	3053	might get executed before other C<ev_check> watchers did their job. As
2692	C<ev_check> watchers are often used to embed other (non-libev) event	3054	C<ev_check> watchers are often used to embed other (non-libev) event
2693	loops those other event loops might be in an unusable state until their	3055	loops those other event loops might be in an unusable state until their
2694	C<ev_check> watcher ran (always remind yourself to coexist peacefully with	3056	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
2695	others).	3057	others).
		3058
		3059	=head3 Abusing an C<ev_check> watcher for its side-effect
		3060
		3061	C<ev_check> (and less often also C<ev_prepare>) watchers can also be
		3062	useful because they are called once per event loop iteration. For
		3063	example, if you want to handle a large number of connections fairly, you
		3064	normally only do a bit of work for each active connection, and if there
		3065	is more work to do, you wait for the next event loop iteration, so other
		3066	connections have a chance of making progress.
		3067
		3068	Using an C<ev_check> watcher is almost enough: it will be called on the
		3069	next event loop iteration. However, that isn't as soon as possible -
		3070	without external events, your C<ev_check> watcher will not be invoked.
		3071
		3072	This is where C<ev_idle> watchers come in handy - all you need is a
		3073	single global idle watcher that is active as long as you have one active
		3074	C<ev_check> watcher. The C<ev_idle> watcher makes sure the event loop
		3075	will not sleep, and the C<ev_check> watcher makes sure a callback gets
		3076	invoked. Neither watcher alone can do that.
2696		3077
2697	=head3 Watcher-Specific Functions and Data Members	3078	=head3 Watcher-Specific Functions and Data Members
2698		3079
2699	=over 4	3080	=over 4
2700		3081
…		…
2901		3282
2902	=over 4	3283	=over 4
2903		3284
2904	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	3285	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
2905		3286
2906	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)	3287	=item ev_embed_set (ev_embed *, struct ev_loop *embedded_loop)
2907		3288
2908	Configures the watcher to embed the given loop, which must be	3289	Configures the watcher to embed the given loop, which must be
2909	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	3290	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
2910	invoked automatically, otherwise it is the responsibility of the callback	3291	invoked automatically, otherwise it is the responsibility of the callback
2911	to invoke it (it will continue to be called until the sweep has been done,	3292	to invoke it (it will continue to be called until the sweep has been done,
…		…
2932	used).	3313	used).
2933		3314
2934	struct ev_loop *loop_hi = ev_default_init (0);	3315	struct ev_loop *loop_hi = ev_default_init (0);
2935	struct ev_loop *loop_lo = 0;	3316	struct ev_loop *loop_lo = 0;
2936	ev_embed embed;	3317	ev_embed embed;
2937		3318
2938	// see if there is a chance of getting one that works	3319	// see if there is a chance of getting one that works
2939	// (remember that a flags value of 0 means autodetection)	3320	// (remember that a flags value of 0 means autodetection)
2940	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()	3321	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2941	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())	3322	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2942	: 0;	3323	: 0;
…		…
2956	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).	3337	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
2957		3338
2958	struct ev_loop *loop = ev_default_init (0);	3339	struct ev_loop *loop = ev_default_init (0);
2959	struct ev_loop *loop_socket = 0;	3340	struct ev_loop *loop_socket = 0;
2960	ev_embed embed;	3341	ev_embed embed;
2961		3342
2962	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)	3343	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2963	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))	3344	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2964	{	3345	{
2965	ev_embed_init (&embed, 0, loop_socket);	3346	ev_embed_init (&embed, 0, loop_socket);
2966	ev_embed_start (loop, &embed);	3347	ev_embed_start (loop, &embed);
…		…
2974		3355
2975	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	3356	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
2976		3357
2977	Fork watchers are called when a C<fork ()> was detected (usually because	3358	Fork watchers are called when a C<fork ()> was detected (usually because
2978	whoever is a good citizen cared to tell libev about it by calling	3359	whoever is a good citizen cared to tell libev about it by calling
2979	C<ev_default_fork> or C<ev_loop_fork>). The invocation is done before the	3360	C<ev_loop_fork>). The invocation is done before the event loop blocks next
2980	event loop blocks next and before C<ev_check> watchers are being called,	3361	and before C<ev_check> watchers are being called, and only in the child
2981	and only in the child after the fork. If whoever good citizen calling	3362	after the fork. If whoever good citizen calling C<ev_default_fork> cheats
2982	C<ev_default_fork> cheats and calls it in the wrong process, the fork	3363	and calls it in the wrong process, the fork handlers will be invoked, too,
2983	handlers will be invoked, too, of course.	3364	of course.
2984		3365
2985	=head3 The special problem of life after fork - how is it possible?	3366	=head3 The special problem of life after fork - how is it possible?
2986		3367
2987	Most uses of C<fork()> consist of forking, then some simple calls to set	3368	Most uses of C<fork ()> consist of forking, then some simple calls to set
2988	up/change the process environment, followed by a call to C<exec()>. This	3369	up/change the process environment, followed by a call to C<exec()>. This
2989	sequence should be handled by libev without any problems.	3370	sequence should be handled by libev without any problems.
2990		3371
2991	This changes when the application actually wants to do event handling	3372	This changes when the application actually wants to do event handling
2992	in the child, or both parent in child, in effect "continuing" after the	3373	in the child, or both parent in child, in effect "continuing" after the
…		…
3008	disadvantage of having to use multiple event loops (which do not support	3389	disadvantage of having to use multiple event loops (which do not support
3009	signal watchers).	3390	signal watchers).
3010		3391
3011	When this is not possible, or you want to use the default loop for	3392	When this is not possible, or you want to use the default loop for
3012	other reasons, then in the process that wants to start "fresh", call	3393	other reasons, then in the process that wants to start "fresh", call
3013	C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying	3394	C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
3014	the default loop will "orphan" (not stop) all registered watchers, so you	3395	Destroying the default loop will "orphan" (not stop) all registered
3015	have to be careful not to execute code that modifies those watchers. Note	3396	watchers, so you have to be careful not to execute code that modifies
3016	also that in that case, you have to re-register any signal watchers.	3397	those watchers. Note also that in that case, you have to re-register any
		3398	signal watchers.
3017		3399
3018	=head3 Watcher-Specific Functions and Data Members	3400	=head3 Watcher-Specific Functions and Data Members
3019		3401
3020	=over 4	3402	=over 4
3021		3403
3022	=item ev_fork_init (ev_signal *, callback)	3404	=item ev_fork_init (ev_fork *, callback)
3023		3405
3024	Initialises and configures the fork watcher - it has no parameters of any	3406	Initialises and configures the fork watcher - it has no parameters of any
3025	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,	3407	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
3026	believe me.	3408	really.
3027		3409
3028	=back	3410	=back
3029		3411
3030		3412
		3413	=head2 C<ev_cleanup> - even the best things end
		3414
		3415	Cleanup watchers are called just before the event loop is being destroyed
		3416	by a call to C<ev_loop_destroy>.
		3417
		3418	While there is no guarantee that the event loop gets destroyed, cleanup
		3419	watchers provide a convenient method to install cleanup hooks for your
		3420	program, worker threads and so on - you just to make sure to destroy the
		3421	loop when you want them to be invoked.
		3422
		3423	Cleanup watchers are invoked in the same way as any other watcher. Unlike
		3424	all other watchers, they do not keep a reference to the event loop (which
		3425	makes a lot of sense if you think about it). Like all other watchers, you
		3426	can call libev functions in the callback, except C<ev_cleanup_start>.
		3427
		3428	=head3 Watcher-Specific Functions and Data Members
		3429
		3430	=over 4
		3431
		3432	=item ev_cleanup_init (ev_cleanup *, callback)
		3433
		3434	Initialises and configures the cleanup watcher - it has no parameters of
		3435	any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
		3436	pointless, I assure you.
		3437
		3438	=back
		3439
		3440	Example: Register an atexit handler to destroy the default loop, so any
		3441	cleanup functions are called.
		3442
		3443	static void
		3444	program_exits (void)
		3445	{
		3446	ev_loop_destroy (EV_DEFAULT_UC);
		3447	}
		3448
		3449	...
		3450	atexit (program_exits);
		3451
		3452
3031	=head2 C<ev_async> - how to wake up an event loop	3453	=head2 C<ev_async> - how to wake up an event loop
3032		3454
3033	In general, you cannot use an C<ev_run> from multiple threads or other	3455	In general, you cannot use an C<ev_loop> from multiple threads or other
3034	asynchronous sources such as signal handlers (as opposed to multiple event	3456	asynchronous sources such as signal handlers (as opposed to multiple event
3035	loops - those are of course safe to use in different threads).	3457	loops - those are of course safe to use in different threads).
3036		3458
3037	Sometimes, however, you need to wake up an event loop you do not control,	3459	Sometimes, however, you need to wake up an event loop you do not control,
3038	for example because it belongs to another thread. This is what C<ev_async>	3460	for example because it belongs to another thread. This is what C<ev_async>
…		…
3040	it by calling C<ev_async_send>, which is thread- and signal safe.	3462	it by calling C<ev_async_send>, which is thread- and signal safe.
3041		3463
3042	This functionality is very similar to C<ev_signal> watchers, as signals,	3464	This functionality is very similar to C<ev_signal> watchers, as signals,
3043	too, are asynchronous in nature, and signals, too, will be compressed	3465	too, are asynchronous in nature, and signals, too, will be compressed
3044	(i.e. the number of callback invocations may be less than the number of	3466	(i.e. the number of callback invocations may be less than the number of
3045	C<ev_async_sent> calls).	3467	C<ev_async_send> calls). In fact, you could use signal watchers as a kind
3046		3468	of "global async watchers" by using a watcher on an otherwise unused
3047	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not	3469	signal, and C<ev_feed_signal> to signal this watcher from another thread,
3048	just the default loop.	3470	even without knowing which loop owns the signal.
3049		3471
3050	=head3 Queueing	3472	=head3 Queueing
3051		3473
3052	C<ev_async> does not support queueing of data in any way. The reason	3474	C<ev_async> does not support queueing of data in any way. The reason
3053	is that the author does not know of a simple (or any) algorithm for a	3475	is that the author does not know of a simple (or any) algorithm for a
…		…
3145	trust me.	3567	trust me.
3146		3568
3147	=item ev_async_send (loop, ev_async *)	3569	=item ev_async_send (loop, ev_async *)
3148		3570
3149	Sends/signals/activates the given C<ev_async> watcher, that is, feeds	3571	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
3150	an C<EV_ASYNC> event on the watcher into the event loop. Unlike	3572	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3573	returns.
		3574
3151	C<ev_feed_event>, this call is safe to do from other threads, signal or	3575	Unlike C<ev_feed_event>, this call is safe to do from other threads,
3152	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding	3576	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
3153	section below on what exactly this means).	3577	embedding section below on what exactly this means).
3154		3578
3155	Note that, as with other watchers in libev, multiple events might get	3579	Note that, as with other watchers in libev, multiple events might get
3156	compressed into a single callback invocation (another way to look at this	3580	compressed into a single callback invocation (another way to look at
3157	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,	3581	this is that C<ev_async> watchers are level-triggered: they are set on
3158	reset when the event loop detects that).	3582	C<ev_async_send>, reset when the event loop detects that).
3159		3583
3160	This call incurs the overhead of a system call only once per event loop	3584	This call incurs the overhead of at most one extra system call per event
3161	iteration, so while the overhead might be noticeable, it doesn't apply to	3585	loop iteration, if the event loop is blocked, and no syscall at all if
3162	repeated calls to C<ev_async_send> for the same event loop.	3586	the event loop (or your program) is processing events. That means that
		3587	repeated calls are basically free (there is no need to avoid calls for
		3588	performance reasons) and that the overhead becomes smaller (typically
		3589	zero) under load.
3163		3590
3164	=item bool = ev_async_pending (ev_async *)	3591	=item bool = ev_async_pending (ev_async *)
3165		3592
3166	Returns a non-zero value when C<ev_async_send> has been called on the	3593	Returns a non-zero value when C<ev_async_send> has been called on the
3167	watcher but the event has not yet been processed (or even noted) by the	3594	watcher but the event has not yet been processed (or even noted) by the
…		…
3184		3611
3185	There are some other functions of possible interest. Described. Here. Now.	3612	There are some other functions of possible interest. Described. Here. Now.
3186		3613
3187	=over 4	3614	=over 4
3188		3615
3189	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	3616	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback, arg)
3190		3617
3191	This function combines a simple timer and an I/O watcher, calls your	3618	This function combines a simple timer and an I/O watcher, calls your
3192	callback on whichever event happens first and automatically stops both	3619	callback on whichever event happens first and automatically stops both
3193	watchers. This is useful if you want to wait for a single event on an fd	3620	watchers. This is useful if you want to wait for a single event on an fd
3194	or timeout without having to allocate/configure/start/stop/free one or	3621	or timeout without having to allocate/configure/start/stop/free one or
…		…
3222	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	3649	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
3223		3650
3224	=item ev_feed_fd_event (loop, int fd, int revents)	3651	=item ev_feed_fd_event (loop, int fd, int revents)
3225		3652
3226	Feed an event on the given fd, as if a file descriptor backend detected	3653	Feed an event on the given fd, as if a file descriptor backend detected
3227	the given events it.	3654	the given events.
3228		3655
3229	=item ev_feed_signal_event (loop, int signum)	3656	=item ev_feed_signal_event (loop, int signum)
3230		3657
3231	Feed an event as if the given signal occurred (C<loop> must be the default	3658	Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
3232	loop!).	3659	which is async-safe.
3233		3660
3234	=back	3661	=back
		3662
		3663
		3664	=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
		3665
		3666	This section explains some common idioms that are not immediately
		3667	obvious. Note that examples are sprinkled over the whole manual, and this
		3668	section only contains stuff that wouldn't fit anywhere else.
		3669
		3670	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
		3671
		3672	Each watcher has, by default, a C<void *data> member that you can read
		3673	or modify at any time: libev will completely ignore it. This can be used
		3674	to associate arbitrary data with your watcher. If you need more data and
		3675	don't want to allocate memory separately and store a pointer to it in that
		3676	data member, you can also "subclass" the watcher type and provide your own
		3677	data:
		3678
		3679	struct my_io
		3680	{
		3681	ev_io io;
		3682	int otherfd;
		3683	void *somedata;
		3684	struct whatever *mostinteresting;
		3685	};
		3686
		3687	...
		3688	struct my_io w;
		3689	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3690
		3691	And since your callback will be called with a pointer to the watcher, you
		3692	can cast it back to your own type:
		3693
		3694	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
		3695	{
		3696	struct my_io *w = (struct my_io *)w_;
		3697	...
		3698	}
		3699
		3700	More interesting and less C-conformant ways of casting your callback
		3701	function type instead have been omitted.
		3702
		3703	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3704
		3705	Another common scenario is to use some data structure with multiple
		3706	embedded watchers, in effect creating your own watcher that combines
		3707	multiple libev event sources into one "super-watcher":
		3708
		3709	struct my_biggy
		3710	{
		3711	int some_data;
		3712	ev_timer t1;
		3713	ev_timer t2;
		3714	}
		3715
		3716	In this case getting the pointer to C<my_biggy> is a bit more
		3717	complicated: Either you store the address of your C<my_biggy> struct in
		3718	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3719	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3720	real programmers):
		3721
		3722	#include <stddef.h>
		3723
		3724	static void
		3725	t1_cb (EV_P_ ev_timer *w, int revents)
		3726	{
		3727	struct my_biggy big = (struct my_biggy *)
		3728	(((char *)w) - offsetof (struct my_biggy, t1));
		3729	}
		3730
		3731	static void
		3732	t2_cb (EV_P_ ev_timer *w, int revents)
		3733	{
		3734	struct my_biggy big = (struct my_biggy *)
		3735	(((char *)w) - offsetof (struct my_biggy, t2));
		3736	}
		3737
		3738	=head2 AVOIDING FINISHING BEFORE RETURNING
		3739
		3740	Often you have structures like this in event-based programs:
		3741
		3742	callback ()
		3743	{
		3744	free (request);
		3745	}
		3746
		3747	request = start_new_request (..., callback);
		3748
		3749	The intent is to start some "lengthy" operation. The C<request> could be
		3750	used to cancel the operation, or do other things with it.
		3751
		3752	It's not uncommon to have code paths in C<start_new_request> that
		3753	immediately invoke the callback, for example, to report errors. Or you add
		3754	some caching layer that finds that it can skip the lengthy aspects of the
		3755	operation and simply invoke the callback with the result.
		3756
		3757	The problem here is that this will happen I<before> C<start_new_request>
		3758	has returned, so C<request> is not set.
		3759
		3760	Even if you pass the request by some safer means to the callback, you
		3761	might want to do something to the request after starting it, such as
		3762	canceling it, which probably isn't working so well when the callback has
		3763	already been invoked.
		3764
		3765	A common way around all these issues is to make sure that
		3766	C<start_new_request> I<always> returns before the callback is invoked. If
		3767	C<start_new_request> immediately knows the result, it can artificially
		3768	delay invoking the callback by using a C<prepare> or C<idle> watcher for
		3769	example, or more sneakily, by reusing an existing (stopped) watcher and
		3770	pushing it into the pending queue:
		3771
		3772	ev_set_cb (watcher, callback);
		3773	ev_feed_event (EV_A_ watcher, 0);
		3774
		3775	This way, C<start_new_request> can safely return before the callback is
		3776	invoked, while not delaying callback invocation too much.
		3777
		3778	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
		3779
		3780	Often (especially in GUI toolkits) there are places where you have
		3781	I<modal> interaction, which is most easily implemented by recursively
		3782	invoking C<ev_run>.
		3783
		3784	This brings the problem of exiting - a callback might want to finish the
		3785	main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
		3786	a modal "Are you sure?" dialog is still waiting), or just the nested one
		3787	and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
		3788	other combination: In these cases, a simple C<ev_break> will not work.
		3789
		3790	The solution is to maintain "break this loop" variable for each C<ev_run>
		3791	invocation, and use a loop around C<ev_run> until the condition is
		3792	triggered, using C<EVRUN_ONCE>:
		3793
		3794	// main loop
		3795	int exit_main_loop = 0;
		3796
		3797	while (!exit_main_loop)
		3798	ev_run (EV_DEFAULT_ EVRUN_ONCE);
		3799
		3800	// in a modal watcher
		3801	int exit_nested_loop = 0;
		3802
		3803	while (!exit_nested_loop)
		3804	ev_run (EV_A_ EVRUN_ONCE);
		3805
		3806	To exit from any of these loops, just set the corresponding exit variable:
		3807
		3808	// exit modal loop
		3809	exit_nested_loop = 1;
		3810
		3811	// exit main program, after modal loop is finished
		3812	exit_main_loop = 1;
		3813
		3814	// exit both
		3815	exit_main_loop = exit_nested_loop = 1;
		3816
		3817	=head2 THREAD LOCKING EXAMPLE
		3818
		3819	Here is a fictitious example of how to run an event loop in a different
		3820	thread from where callbacks are being invoked and watchers are
		3821	created/added/removed.
		3822
		3823	For a real-world example, see the C<EV::Loop::Async> perl module,
		3824	which uses exactly this technique (which is suited for many high-level
		3825	languages).
		3826
		3827	The example uses a pthread mutex to protect the loop data, a condition
		3828	variable to wait for callback invocations, an async watcher to notify the
		3829	event loop thread and an unspecified mechanism to wake up the main thread.
		3830
		3831	First, you need to associate some data with the event loop:
		3832
		3833	typedef struct {
		3834	mutex_t lock; /* global loop lock */
		3835	ev_async async_w;
		3836	thread_t tid;
		3837	cond_t invoke_cv;
		3838	} userdata;
		3839
		3840	void prepare_loop (EV_P)
		3841	{
		3842	// for simplicity, we use a static userdata struct.
		3843	static userdata u;
		3844
		3845	ev_async_init (&u->async_w, async_cb);
		3846	ev_async_start (EV_A_ &u->async_w);
		3847
		3848	pthread_mutex_init (&u->lock, 0);
		3849	pthread_cond_init (&u->invoke_cv, 0);
		3850
		3851	// now associate this with the loop
		3852	ev_set_userdata (EV_A_ u);
		3853	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3854	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3855
		3856	// then create the thread running ev_run
		3857	pthread_create (&u->tid, 0, l_run, EV_A);
		3858	}
		3859
		3860	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3861	solely to wake up the event loop so it takes notice of any new watchers
		3862	that might have been added:
		3863
		3864	static void
		3865	async_cb (EV_P_ ev_async *w, int revents)
		3866	{
		3867	// just used for the side effects
		3868	}
		3869
		3870	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3871	protecting the loop data, respectively.
		3872
		3873	static void
		3874	l_release (EV_P)
		3875	{
		3876	userdata *u = ev_userdata (EV_A);
		3877	pthread_mutex_unlock (&u->lock);
		3878	}
		3879
		3880	static void
		3881	l_acquire (EV_P)
		3882	{
		3883	userdata *u = ev_userdata (EV_A);
		3884	pthread_mutex_lock (&u->lock);
		3885	}
		3886
		3887	The event loop thread first acquires the mutex, and then jumps straight
		3888	into C<ev_run>:
		3889
		3890	void *
		3891	l_run (void *thr_arg)
		3892	{
		3893	struct ev_loop *loop = (struct ev_loop *)thr_arg;
		3894
		3895	l_acquire (EV_A);
		3896	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3897	ev_run (EV_A_ 0);
		3898	l_release (EV_A);
		3899
		3900	return 0;
		3901	}
		3902
		3903	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3904	signal the main thread via some unspecified mechanism (signals? pipe
		3905	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3906	have been called (in a while loop because a) spurious wakeups are possible
		3907	and b) skipping inter-thread-communication when there are no pending
		3908	watchers is very beneficial):
		3909
		3910	static void
		3911	l_invoke (EV_P)
		3912	{
		3913	userdata *u = ev_userdata (EV_A);
		3914
		3915	while (ev_pending_count (EV_A))
		3916	{
		3917	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3918	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3919	}
		3920	}
		3921
		3922	Now, whenever the main thread gets told to invoke pending watchers, it
		3923	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3924	thread to continue:
		3925
		3926	static void
		3927	real_invoke_pending (EV_P)
		3928	{
		3929	userdata *u = ev_userdata (EV_A);
		3930
		3931	pthread_mutex_lock (&u->lock);
		3932	ev_invoke_pending (EV_A);
		3933	pthread_cond_signal (&u->invoke_cv);
		3934	pthread_mutex_unlock (&u->lock);
		3935	}
		3936
		3937	Whenever you want to start/stop a watcher or do other modifications to an
		3938	event loop, you will now have to lock:
		3939
		3940	ev_timer timeout_watcher;
		3941	userdata *u = ev_userdata (EV_A);
		3942
		3943	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3944
		3945	pthread_mutex_lock (&u->lock);
		3946	ev_timer_start (EV_A_ &timeout_watcher);
		3947	ev_async_send (EV_A_ &u->async_w);
		3948	pthread_mutex_unlock (&u->lock);
		3949
		3950	Note that sending the C<ev_async> watcher is required because otherwise
		3951	an event loop currently blocking in the kernel will have no knowledge
		3952	about the newly added timer. By waking up the loop it will pick up any new
		3953	watchers in the next event loop iteration.
		3954
		3955	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3956
		3957	While the overhead of a callback that e.g. schedules a thread is small, it
		3958	is still an overhead. If you embed libev, and your main usage is with some
		3959	kind of threads or coroutines, you might want to customise libev so that
		3960	doesn't need callbacks anymore.
		3961
		3962	Imagine you have coroutines that you can switch to using a function
		3963	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3964	and that due to some magic, the currently active coroutine is stored in a
		3965	global called C<current_coro>. Then you can build your own "wait for libev
		3966	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3967	the differing C<;> conventions):
		3968
		3969	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3970	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3971
		3972	That means instead of having a C callback function, you store the
		3973	coroutine to switch to in each watcher, and instead of having libev call
		3974	your callback, you instead have it switch to that coroutine.
		3975
		3976	A coroutine might now wait for an event with a function called
		3977	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		3978	matter when, or whether the watcher is active or not when this function is
		3979	called):
		3980
		3981	void
		3982	wait_for_event (ev_watcher *w)
		3983	{
		3984	ev_set_cb (w, current_coro);
		3985	switch_to (libev_coro);
		3986	}
		3987
		3988	That basically suspends the coroutine inside C<wait_for_event> and
		3989	continues the libev coroutine, which, when appropriate, switches back to
		3990	this or any other coroutine.
		3991
		3992	You can do similar tricks if you have, say, threads with an event queue -
		3993	instead of storing a coroutine, you store the queue object and instead of
		3994	switching to a coroutine, you push the watcher onto the queue and notify
		3995	any waiters.
		3996
		3997	To embed libev, see L</EMBEDDING>, but in short, it's easiest to create two
		3998	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		3999
		4000	// my_ev.h
		4001	#define EV_CB_DECLARE(type) struct my_coro *cb;
		4002	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		4003	#include "../libev/ev.h"
		4004
		4005	// my_ev.c
		4006	#define EV_H "my_ev.h"
		4007	#include "../libev/ev.c"
		4008
		4009	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		4010	F<my_ev.c> into your project. When properly specifying include paths, you
		4011	can even use F<ev.h> as header file name directly.
3235		4012
3236		4013
3237	=head1 LIBEVENT EMULATION	4014	=head1 LIBEVENT EMULATION
3238		4015
3239	Libev offers a compatibility emulation layer for libevent. It cannot	4016	Libev offers a compatibility emulation layer for libevent. It cannot
3240	emulate the internals of libevent, so here are some usage hints:	4017	emulate the internals of libevent, so here are some usage hints:
3241		4018
3242	=over 4	4019	=over 4
		4020
		4021	=item * Only the libevent-1.4.1-beta API is being emulated.
		4022
		4023	This was the newest libevent version available when libev was implemented,
		4024	and is still mostly unchanged in 2010.
3243		4025
3244	=item * Use it by including <event.h>, as usual.	4026	=item * Use it by including <event.h>, as usual.
3245		4027
3246	=item * The following members are fully supported: ev_base, ev_callback,	4028	=item * The following members are fully supported: ev_base, ev_callback,
3247	ev_arg, ev_fd, ev_res, ev_events.	4029	ev_arg, ev_fd, ev_res, ev_events.
…		…
3253	=item * Priorities are not currently supported. Initialising priorities	4035	=item * Priorities are not currently supported. Initialising priorities
3254	will fail and all watchers will have the same priority, even though there	4036	will fail and all watchers will have the same priority, even though there
3255	is an ev_pri field.	4037	is an ev_pri field.
3256		4038
3257	=item * In libevent, the last base created gets the signals, in libev, the	4039	=item * In libevent, the last base created gets the signals, in libev, the
3258	first base created (== the default loop) gets the signals.	4040	base that registered the signal gets the signals.
3259		4041
3260	=item * Other members are not supported.	4042	=item * Other members are not supported.
3261		4043
3262	=item * The libev emulation is I<not> ABI compatible to libevent, you need	4044	=item * The libev emulation is I<not> ABI compatible to libevent, you need
3263	to use the libev header file and library.	4045	to use the libev header file and library.
3264		4046
3265	=back	4047	=back
3266		4048
3267	=head1 C++ SUPPORT	4049	=head1 C++ SUPPORT
		4050
		4051	=head2 C API
		4052
		4053	The normal C API should work fine when used from C++: both ev.h and the
		4054	libev sources can be compiled as C++. Therefore, code that uses the C API
		4055	will work fine.
		4056
		4057	Proper exception specifications might have to be added to callbacks passed
		4058	to libev: exceptions may be thrown only from watcher callbacks, all other
		4059	callbacks (allocator, syserr, loop acquire/release and periodic reschedule
		4060	callbacks) must not throw exceptions, and might need a C<noexcept>
		4061	specification. If you have code that needs to be compiled as both C and
		4062	C++ you can use the C<EV_NOEXCEPT> macro for this:
		4063
		4064	static void
		4065	fatal_error (const char *msg) EV_NOEXCEPT
		4066	{
		4067	perror (msg);
		4068	abort ();
		4069	}
		4070
		4071	...
		4072	ev_set_syserr_cb (fatal_error);
		4073
		4074	The only API functions that can currently throw exceptions are C<ev_run>,
		4075	C<ev_invoke>, C<ev_invoke_pending> and C<ev_loop_destroy> (the latter
		4076	because it runs cleanup watchers).
		4077
		4078	Throwing exceptions in watcher callbacks is only supported if libev itself
		4079	is compiled with a C++ compiler or your C and C++ environments allow
		4080	throwing exceptions through C libraries (most do).
		4081
		4082	=head2 C++ API
3268		4083
3269	Libev comes with some simplistic wrapper classes for C++ that mainly allow	4084	Libev comes with some simplistic wrapper classes for C++ that mainly allow
3270	you to use some convenience methods to start/stop watchers and also change	4085	you to use some convenience methods to start/stop watchers and also change
3271	the callback model to a model using method callbacks on objects.	4086	the callback model to a model using method callbacks on objects.
3272		4087
3273	To use it,	4088	To use it,
3274		4089
3275	#include <ev++.h>	4090	#include <ev++.h>
3276		4091
3277	This automatically includes F<ev.h> and puts all of its definitions (many	4092	This automatically includes F<ev.h> and puts all of its definitions (many
3278	of them macros) into the global namespace. All C++ specific things are	4093	of them macros) into the global namespace. All C++ specific things are
3279	put into the C<ev> namespace. It should support all the same embedding	4094	put into the C<ev> namespace. It should support all the same embedding
…		…
3282	Care has been taken to keep the overhead low. The only data member the C++	4097	Care has been taken to keep the overhead low. The only data member the C++
3283	classes add (compared to plain C-style watchers) is the event loop pointer	4098	classes add (compared to plain C-style watchers) is the event loop pointer
3284	that the watcher is associated with (or no additional members at all if	4099	that the watcher is associated with (or no additional members at all if
3285	you disable C<EV_MULTIPLICITY> when embedding libev).	4100	you disable C<EV_MULTIPLICITY> when embedding libev).
3286		4101
3287	Currently, functions, and static and non-static member functions can be	4102	Currently, functions, static and non-static member functions and classes
3288	used as callbacks. Other types should be easy to add as long as they only	4103	with C<operator ()> can be used as callbacks. Other types should be easy
3289	need one additional pointer for context. If you need support for other	4104	to add as long as they only need one additional pointer for context. If
3290	types of functors please contact the author (preferably after implementing	4105	you need support for other types of functors please contact the author
3291	it).	4106	(preferably after implementing it).
		4107
		4108	For all this to work, your C++ compiler either has to use the same calling
		4109	conventions as your C compiler (for static member functions), or you have
		4110	to embed libev and compile libev itself as C++.
3292		4111
3293	Here is a list of things available in the C<ev> namespace:	4112	Here is a list of things available in the C<ev> namespace:
3294		4113
3295	=over 4	4114	=over 4
3296		4115
…		…
3306	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.	4125	=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
3307		4126
3308	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of	4127	For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
3309	the same name in the C<ev> namespace, with the exception of C<ev_signal>	4128	the same name in the C<ev> namespace, with the exception of C<ev_signal>
3310	which is called C<ev::sig> to avoid clashes with the C<signal> macro	4129	which is called C<ev::sig> to avoid clashes with the C<signal> macro
3311	defines by many implementations.	4130	defined by many implementations.
3312		4131
3313	All of those classes have these methods:	4132	All of those classes have these methods:
3314		4133
3315	=over 4	4134	=over 4
3316		4135
…		…
3378	void operator() (ev::io &w, int revents)	4197	void operator() (ev::io &w, int revents)
3379	{	4198	{
3380	...	4199	...
3381	}	4200	}
3382	}	4201	}
3383		4202
3384	myfunctor f;	4203	myfunctor f;
3385		4204
3386	ev::io w;	4205	ev::io w;
3387	w.set (&f);	4206	w.set (&f);
3388		4207
…		…
3406	Associates a different C<struct ev_loop> with this watcher. You can only	4225	Associates a different C<struct ev_loop> with this watcher. You can only
3407	do this when the watcher is inactive (and not pending either).	4226	do this when the watcher is inactive (and not pending either).
3408		4227
3409	=item w->set ([arguments])	4228	=item w->set ([arguments])
3410		4229
3411	Basically the same as C<ev_TYPE_set>, with the same arguments. Either this	4230	Basically the same as C<ev_TYPE_set> (except for C<ev::embed> watchers>),
3412	method or a suitable start method must be called at least once. Unlike the	4231	with the same arguments. Either this method or a suitable start method
3413	C counterpart, an active watcher gets automatically stopped and restarted	4232	must be called at least once. Unlike the C counterpart, an active watcher
3414	when reconfiguring it with this method.	4233	gets automatically stopped and restarted when reconfiguring it with this
		4234	method.
		4235
		4236	For C<ev::embed> watchers this method is called C<set_embed>, to avoid
		4237	clashing with the C<set (loop)> method.
3415		4238
3416	=item w->start ()	4239	=item w->start ()
3417		4240
3418	Starts the watcher. Note that there is no C<loop> argument, as the	4241	Starts the watcher. Note that there is no C<loop> argument, as the
3419	constructor already stores the event loop.	4242	constructor already stores the event loop.
…		…
3449	watchers in the constructor.	4272	watchers in the constructor.
3450		4273
3451	class myclass	4274	class myclass
3452	{	4275	{
3453	ev::io io ; void io_cb (ev::io &w, int revents);	4276	ev::io io ; void io_cb (ev::io &w, int revents);
3454	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);	4277	ev::io io2 ; void io2_cb (ev::io &w, int revents);
3455	ev::idle idle; void idle_cb (ev::idle &w, int revents);	4278	ev::idle idle; void idle_cb (ev::idle &w, int revents);
3456		4279
3457	myclass (int fd)	4280	myclass (int fd)
3458	{	4281	{
3459	io .set <myclass, &myclass::io_cb > (this);	4282	io .set <myclass, &myclass::io_cb > (this);
…		…
3510	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.	4333	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
3511		4334
3512	=item D	4335	=item D
3513		4336
3514	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to	4337	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
3515	be found at L<http://proj.llucax.com.ar/wiki/evd>.	4338	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
3516		4339
3517	=item Ocaml	4340	=item Ocaml
3518		4341
3519	Erkki Seppala has written Ocaml bindings for libev, to be found at	4342	Erkki Seppala has written Ocaml bindings for libev, to be found at
3520	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.	4343	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
…		…
3523		4346
3524	Brian Maher has written a partial interface to libev for lua (at the	4347	Brian Maher has written a partial interface to libev for lua (at the
3525	time of this writing, only C<ev_io> and C<ev_timer>), to be found at	4348	time of this writing, only C<ev_io> and C<ev_timer>), to be found at
3526	L<http://github.com/brimworks/lua-ev>.	4349	L<http://github.com/brimworks/lua-ev>.
3527		4350
		4351	=item Javascript
		4352
		4353	Node.js (L<http://nodejs.org>) uses libev as the underlying event library.
		4354
		4355	=item Others
		4356
		4357	There are others, and I stopped counting.
		4358
3528	=back	4359	=back
3529		4360
3530		4361
3531	=head1 MACRO MAGIC	4362	=head1 MACRO MAGIC
3532		4363
…		…
3568	suitable for use with C<EV_A>.	4399	suitable for use with C<EV_A>.
3569		4400
3570	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	4401	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
3571		4402
3572	Similar to the other two macros, this gives you the value of the default	4403	Similar to the other two macros, this gives you the value of the default
3573	loop, if multiple loops are supported ("ev loop default").	4404	loop, if multiple loops are supported ("ev loop default"). The default loop
		4405	will be initialised if it isn't already initialised.
		4406
		4407	For non-multiplicity builds, these macros do nothing, so you always have
		4408	to initialise the loop somewhere.
3574		4409
3575	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>	4410	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
3576		4411
3577	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the	4412	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
3578	default loop has been initialised (C<UC> == unchecked). Their behaviour	4413	default loop has been initialised (C<UC> == unchecked). Their behaviour
…		…
3645	ev_vars.h	4480	ev_vars.h
3646	ev_wrap.h	4481	ev_wrap.h
3647		4482
3648	ev_win32.c required on win32 platforms only	4483	ev_win32.c required on win32 platforms only
3649		4484
3650	ev_select.c only when select backend is enabled (which is enabled by default)	4485	ev_select.c only when select backend is enabled
3651	ev_poll.c only when poll backend is enabled (disabled by default)	4486	ev_poll.c only when poll backend is enabled
3652	ev_epoll.c only when the epoll backend is enabled (disabled by default)	4487	ev_epoll.c only when the epoll backend is enabled
		4488	ev_linuxaio.c only when the linux aio backend is enabled
		4489	ev_iouring.c only when the linux io_uring backend is enabled
3653	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	4490	ev_kqueue.c only when the kqueue backend is enabled
3654	ev_port.c only when the solaris port backend is enabled (disabled by default)	4491	ev_port.c only when the solaris port backend is enabled
3655		4492
3656	F<ev.c> includes the backend files directly when enabled, so you only need	4493	F<ev.c> includes the backend files directly when enabled, so you only need
3657	to compile this single file.	4494	to compile this single file.
3658		4495
3659	=head3 LIBEVENT COMPATIBILITY API	4496	=head3 LIBEVENT COMPATIBILITY API
…		…
3723	supported). It will also not define any of the structs usually found in	4560	supported). It will also not define any of the structs usually found in
3724	F<event.h> that are not directly supported by the libev core alone.	4561	F<event.h> that are not directly supported by the libev core alone.
3725		4562
3726	In standalone mode, libev will still try to automatically deduce the	4563	In standalone mode, libev will still try to automatically deduce the
3727	configuration, but has to be more conservative.	4564	configuration, but has to be more conservative.
		4565
		4566	=item EV_USE_FLOOR
		4567
		4568	If defined to be C<1>, libev will use the C<floor ()> function for its
		4569	periodic reschedule calculations, otherwise libev will fall back on a
		4570	portable (slower) implementation. If you enable this, you usually have to
		4571	link against libm or something equivalent. Enabling this when the C<floor>
		4572	function is not available will fail, so the safe default is to not enable
		4573	this.
3728		4574
3729	=item EV_USE_MONOTONIC	4575	=item EV_USE_MONOTONIC
3730		4576
3731	If defined to be C<1>, libev will try to detect the availability of the	4577	If defined to be C<1>, libev will try to detect the availability of the
3732	monotonic clock option at both compile time and runtime. Otherwise no	4578	monotonic clock option at both compile time and runtime. Otherwise no
…		…
3818	If programs implement their own fd to handle mapping on win32, then this	4664	If programs implement their own fd to handle mapping on win32, then this
3819	macro can be used to override the C<close> function, useful to unregister	4665	macro can be used to override the C<close> function, useful to unregister
3820	file descriptors again. Note that the replacement function has to close	4666	file descriptors again. Note that the replacement function has to close
3821	the underlying OS handle.	4667	the underlying OS handle.
3822		4668
		4669	=item EV_USE_WSASOCKET
		4670
		4671	If defined to be C<1>, libev will use C<WSASocket> to create its internal
		4672	communication socket, which works better in some environments. Otherwise,
		4673	the normal C<socket> function will be used, which works better in other
		4674	environments.
		4675
3823	=item EV_USE_POLL	4676	=item EV_USE_POLL
3824		4677
3825	If defined to be C<1>, libev will compile in support for the C<poll>(2)	4678	If defined to be C<1>, libev will compile in support for the C<poll>(2)
3826	backend. Otherwise it will be enabled on non-win32 platforms. It	4679	backend. Otherwise it will be enabled on non-win32 platforms. It
3827	takes precedence over select.	4680	takes precedence over select.
…		…
3831	If defined to be C<1>, libev will compile in support for the Linux	4684	If defined to be C<1>, libev will compile in support for the Linux
3832	C<epoll>(7) backend. Its availability will be detected at runtime,	4685	C<epoll>(7) backend. Its availability will be detected at runtime,
3833	otherwise another method will be used as fallback. This is the preferred	4686	otherwise another method will be used as fallback. This is the preferred
3834	backend for GNU/Linux systems. If undefined, it will be enabled if the	4687	backend for GNU/Linux systems. If undefined, it will be enabled if the
3835	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.	4688	headers indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
		4689
		4690	=item EV_USE_LINUXAIO
		4691
		4692	If defined to be C<1>, libev will compile in support for the Linux aio
		4693	backend (C<EV_USE_EPOLL> must also be enabled). If undefined, it will be
		4694	enabled on linux, otherwise disabled.
		4695
		4696	=item EV_USE_IOURING
		4697
		4698	If defined to be C<1>, libev will compile in support for the Linux
		4699	io_uring backend (C<EV_USE_EPOLL> must also be enabled). Due to it's
		4700	current limitations it has to be requested explicitly. If undefined, it
		4701	will be enabled on linux, otherwise disabled.
3836		4702
3837	=item EV_USE_KQUEUE	4703	=item EV_USE_KQUEUE
3838		4704
3839	If defined to be C<1>, libev will compile in support for the BSD style	4705	If defined to be C<1>, libev will compile in support for the BSD style
3840	C<kqueue>(2) backend. Its actual availability will be detected at runtime,	4706	C<kqueue>(2) backend. Its actual availability will be detected at runtime,
…		…
3862	If defined to be C<1>, libev will compile in support for the Linux inotify	4728	If defined to be C<1>, libev will compile in support for the Linux inotify
3863	interface to speed up C<ev_stat> watchers. Its actual availability will	4729	interface to speed up C<ev_stat> watchers. Its actual availability will
3864	be detected at runtime. If undefined, it will be enabled if the headers	4730	be detected at runtime. If undefined, it will be enabled if the headers
3865	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.	4731	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
3866		4732
		4733	=item EV_NO_SMP
		4734
		4735	If defined to be C<1>, libev will assume that memory is always coherent
		4736	between threads, that is, threads can be used, but threads never run on
		4737	different cpus (or different cpu cores). This reduces dependencies
		4738	and makes libev faster.
		4739
		4740	=item EV_NO_THREADS
		4741
		4742	If defined to be C<1>, libev will assume that it will never be called from
		4743	different threads (that includes signal handlers), which is a stronger
		4744	assumption than C<EV_NO_SMP>, above. This reduces dependencies and makes
		4745	libev faster.
		4746
3867	=item EV_ATOMIC_T	4747	=item EV_ATOMIC_T
3868		4748
3869	Libev requires an integer type (suitable for storing C<0> or C<1>) whose	4749	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
3870	access is atomic with respect to other threads or signal contexts. No such	4750	access is atomic with respect to other threads or signal contexts. No
3871	type is easily found in the C language, so you can provide your own type	4751	such type is easily found in the C language, so you can provide your own
3872	that you know is safe for your purposes. It is used both for signal handler "locking"	4752	type that you know is safe for your purposes. It is used both for signal
3873	as well as for signal and thread safety in C<ev_async> watchers.	4753	handler "locking" as well as for signal and thread safety in C<ev_async>
		4754	watchers.
3874		4755
3875	In the absence of this define, libev will use C<sig_atomic_t volatile>	4756	In the absence of this define, libev will use C<sig_atomic_t volatile>
3876	(from F<signal.h>), which is usually good enough on most platforms.	4757	(from F<signal.h>), which is usually good enough on most platforms.
3877		4758
3878	=item EV_H (h)	4759	=item EV_H (h)
…		…
3905	will have the C<struct ev_loop *> as first argument, and you can create	4786	will have the C<struct ev_loop *> as first argument, and you can create
3906	additional independent event loops. Otherwise there will be no support	4787	additional independent event loops. Otherwise there will be no support
3907	for multiple event loops and there is no first event loop pointer	4788	for multiple event loops and there is no first event loop pointer
3908	argument. Instead, all functions act on the single default loop.	4789	argument. Instead, all functions act on the single default loop.
3909		4790
		4791	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4792	default loop when multiplicity is switched off - you always have to
		4793	initialise the loop manually in this case.
		4794
3910	=item EV_MINPRI	4795	=item EV_MINPRI
3911		4796
3912	=item EV_MAXPRI	4797	=item EV_MAXPRI
3913		4798
3914	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to	4799	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
…		…
3950	#define EV_USE_POLL 1	4835	#define EV_USE_POLL 1
3951	#define EV_CHILD_ENABLE 1	4836	#define EV_CHILD_ENABLE 1
3952	#define EV_ASYNC_ENABLE 1	4837	#define EV_ASYNC_ENABLE 1
3953		4838
3954	The actual value is a bitset, it can be a combination of the following	4839	The actual value is a bitset, it can be a combination of the following
3955	values:	4840	values (by default, all of these are enabled):
3956		4841
3957	=over 4	4842	=over 4
3958		4843
3959	=item C<1> - faster/larger code	4844	=item C<1> - faster/larger code
3960		4845
…		…
3964	code size by roughly 30% on amd64).	4849	code size by roughly 30% on amd64).
3965		4850
3966	When optimising for size, use of compiler flags such as C<-Os> with	4851	When optimising for size, use of compiler flags such as C<-Os> with
3967	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of	4852	gcc is recommended, as well as C<-DNDEBUG>, as libev contains a number of
3968	assertions.	4853	assertions.
		4854
		4855	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4856	(e.g. gcc with C<-Os>).
3969		4857
3970	=item C<2> - faster/larger data structures	4858	=item C<2> - faster/larger data structures
3971		4859
3972	Replaces the small 2-heap for timer management by a faster 4-heap, larger	4860	Replaces the small 2-heap for timer management by a faster 4-heap, larger
3973	hash table sizes and so on. This will usually further increase code size	4861	hash table sizes and so on. This will usually further increase code size
3974	and can additionally have an effect on the size of data structures at	4862	and can additionally have an effect on the size of data structures at
3975	runtime.	4863	runtime.
3976		4864
		4865	The default is off when C<__OPTIMIZE_SIZE__> is defined by your compiler
		4866	(e.g. gcc with C<-Os>).
		4867
3977	=item C<4> - full API configuration	4868	=item C<4> - full API configuration
3978		4869
3979	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and	4870	This enables priorities (sets C<EV_MAXPRI>=2 and C<EV_MINPRI>=-2), and
3980	enables multiplicity (C<EV_MULTIPLICITY>=1).	4871	enables multiplicity (C<EV_MULTIPLICITY>=1).
3981		4872
…		…
4011		4902
4012	With an intelligent-enough linker (gcc+binutils are intelligent enough	4903	With an intelligent-enough linker (gcc+binutils are intelligent enough
4013	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by	4904	when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
4014	your program might be left out as well - a binary starting a timer and an	4905	your program might be left out as well - a binary starting a timer and an
4015	I/O watcher then might come out at only 5Kb.	4906	I/O watcher then might come out at only 5Kb.
		4907
		4908	=item EV_API_STATIC
		4909
		4910	If this symbol is defined (by default it is not), then all identifiers
		4911	will have static linkage. This means that libev will not export any
		4912	identifiers, and you cannot link against libev anymore. This can be useful
		4913	when you embed libev, only want to use libev functions in a single file,
		4914	and do not want its identifiers to be visible.
		4915
		4916	To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
		4917	wants to use libev.
		4918
		4919	This option only works when libev is compiled with a C compiler, as C++
		4920	doesn't support the required declaration syntax.
4016		4921
4017	=item EV_AVOID_STDIO	4922	=item EV_AVOID_STDIO
4018		4923
4019	If this is set to C<1> at compiletime, then libev will avoid using stdio	4924	If this is set to C<1> at compiletime, then libev will avoid using stdio
4020	functions (printf, scanf, perror etc.). This will increase the code size	4925	functions (printf, scanf, perror etc.). This will increase the code size
…		…
4078	in. If set to C<1>, then verification code will be compiled in, but not	4983	in. If set to C<1>, then verification code will be compiled in, but not
4079	called. If set to C<2>, then the internal verification code will be	4984	called. If set to C<2>, then the internal verification code will be
4080	called once per loop, which can slow down libev. If set to C<3>, then the	4985	called once per loop, which can slow down libev. If set to C<3>, then the
4081	verification code will be called very frequently, which will slow down	4986	verification code will be called very frequently, which will slow down
4082	libev considerably.	4987	libev considerably.
		4988
		4989	Verification errors are reported via C's C<assert> mechanism, so if you
		4990	disable that (e.g. by defining C<NDEBUG>) then no errors will be reported.
4083		4991
4084	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it	4992	The default is C<1>, unless C<EV_FEATURES> overrides it, in which case it
4085	will be C<0>.	4993	will be C<0>.
4086		4994
4087	=item EV_COMMON	4995	=item EV_COMMON
…		…
4164	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	5072	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
4165		5073
4166	#include "ev_cpp.h"	5074	#include "ev_cpp.h"
4167	#include "ev.c"	5075	#include "ev.c"
4168		5076
4169	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES	5077	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
4170		5078
4171	=head2 THREADS AND COROUTINES	5079	=head2 THREADS AND COROUTINES
4172		5080
4173	=head3 THREADS	5081	=head3 THREADS
4174		5082
…		…
4225	default loop and triggering an C<ev_async> watcher from the default loop	5133	default loop and triggering an C<ev_async> watcher from the default loop
4226	watcher callback into the event loop interested in the signal.	5134	watcher callback into the event loop interested in the signal.
4227		5135
4228	=back	5136	=back
4229		5137
4230	=head4 THREAD LOCKING EXAMPLE	5138	See also L</THREAD LOCKING EXAMPLE>.
4231
4232	Here is a fictitious example of how to run an event loop in a different
4233	thread than where callbacks are being invoked and watchers are
4234	created/added/removed.
4235
4236	For a real-world example, see the C<EV::Loop::Async> perl module,
4237	which uses exactly this technique (which is suited for many high-level
4238	languages).
4239
4240	The example uses a pthread mutex to protect the loop data, a condition
4241	variable to wait for callback invocations, an async watcher to notify the
4242	event loop thread and an unspecified mechanism to wake up the main thread.
4243
4244	First, you need to associate some data with the event loop:
4245
4246	typedef struct {
4247	mutex_t lock; /* global loop lock */
4248	ev_async async_w;
4249	thread_t tid;
4250	cond_t invoke_cv;
4251	} userdata;
4252
4253	void prepare_loop (EV_P)
4254	{
4255	// for simplicity, we use a static userdata struct.
4256	static userdata u;
4257
4258	ev_async_init (&u->async_w, async_cb);
4259	ev_async_start (EV_A_ &u->async_w);
4260
4261	pthread_mutex_init (&u->lock, 0);
4262	pthread_cond_init (&u->invoke_cv, 0);
4263
4264	// now associate this with the loop
4265	ev_set_userdata (EV_A_ u);
4266	ev_set_invoke_pending_cb (EV_A_ l_invoke);
4267	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
4268
4269	// then create the thread running ev_loop
4270	pthread_create (&u->tid, 0, l_run, EV_A);
4271	}
4272
4273	The callback for the C<ev_async> watcher does nothing: the watcher is used
4274	solely to wake up the event loop so it takes notice of any new watchers
4275	that might have been added:
4276
4277	static void
4278	async_cb (EV_P_ ev_async *w, int revents)
4279	{
4280	// just used for the side effects
4281	}
4282
4283	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
4284	protecting the loop data, respectively.
4285
4286	static void
4287	l_release (EV_P)
4288	{
4289	userdata *u = ev_userdata (EV_A);
4290	pthread_mutex_unlock (&u->lock);
4291	}
4292
4293	static void
4294	l_acquire (EV_P)
4295	{
4296	userdata *u = ev_userdata (EV_A);
4297	pthread_mutex_lock (&u->lock);
4298	}
4299
4300	The event loop thread first acquires the mutex, and then jumps straight
4301	into C<ev_run>:
4302
4303	void *
4304	l_run (void *thr_arg)
4305	{
4306	struct ev_loop *loop = (struct ev_loop *)thr_arg;
4307
4308	l_acquire (EV_A);
4309	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
4310	ev_run (EV_A_ 0);
4311	l_release (EV_A);
4312
4313	return 0;
4314	}
4315
4316	Instead of invoking all pending watchers, the C<l_invoke> callback will
4317	signal the main thread via some unspecified mechanism (signals? pipe
4318	writes? C<Async::Interrupt>?) and then waits until all pending watchers
4319	have been called (in a while loop because a) spurious wakeups are possible
4320	and b) skipping inter-thread-communication when there are no pending
4321	watchers is very beneficial):
4322
4323	static void
4324	l_invoke (EV_P)
4325	{
4326	userdata *u = ev_userdata (EV_A);
4327
4328	while (ev_pending_count (EV_A))
4329	{
4330	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
4331	pthread_cond_wait (&u->invoke_cv, &u->lock);
4332	}
4333	}
4334
4335	Now, whenever the main thread gets told to invoke pending watchers, it
4336	will grab the lock, call C<ev_invoke_pending> and then signal the loop
4337	thread to continue:
4338
4339	static void
4340	real_invoke_pending (EV_P)
4341	{
4342	userdata *u = ev_userdata (EV_A);
4343
4344	pthread_mutex_lock (&u->lock);
4345	ev_invoke_pending (EV_A);
4346	pthread_cond_signal (&u->invoke_cv);
4347	pthread_mutex_unlock (&u->lock);
4348	}
4349
4350	Whenever you want to start/stop a watcher or do other modifications to an
4351	event loop, you will now have to lock:
4352
4353	ev_timer timeout_watcher;
4354	userdata *u = ev_userdata (EV_A);
4355
4356	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
4357
4358	pthread_mutex_lock (&u->lock);
4359	ev_timer_start (EV_A_ &timeout_watcher);
4360	ev_async_send (EV_A_ &u->async_w);
4361	pthread_mutex_unlock (&u->lock);
4362
4363	Note that sending the C<ev_async> watcher is required because otherwise
4364	an event loop currently blocking in the kernel will have no knowledge
4365	about the newly added timer. By waking up the loop it will pick up any new
4366	watchers in the next event loop iteration.
4367		5139
4368	=head3 COROUTINES	5140	=head3 COROUTINES
4369		5141
4370	Libev is very accommodating to coroutines ("cooperative threads"):	5142	Libev is very accommodating to coroutines ("cooperative threads"):
4371	libev fully supports nesting calls to its functions from different	5143	libev fully supports nesting calls to its functions from different
…		…
4467	=head3 C<kqueue> is buggy	5239	=head3 C<kqueue> is buggy
4468		5240
4469	The kqueue syscall is broken in all known versions - most versions support	5241	The kqueue syscall is broken in all known versions - most versions support
4470	only sockets, many support pipes.	5242	only sockets, many support pipes.
4471		5243
4472	Libev tries to work around this by not using C<kqueue> by default on	5244	Libev tries to work around this by not using C<kqueue> by default on this
4473	this rotten platform, but of course you can still ask for it when creating	5245	rotten platform, but of course you can still ask for it when creating a
4474	a loop.	5246	loop - embedding a socket-only kqueue loop into a select-based one is
		5247	probably going to work well.
4475		5248
4476	=head3 C<poll> is buggy	5249	=head3 C<poll> is buggy
4477		5250
4478	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>	5251	Instead of fixing C<kqueue>, Apple replaced their (working) C<poll>
4479	implementation by something calling C<kqueue> internally around the 10.5.6	5252	implementation by something calling C<kqueue> internally around the 10.5.6
…		…
4498		5271
4499	=head3 C<errno> reentrancy	5272	=head3 C<errno> reentrancy
4500		5273
4501	The default compile environment on Solaris is unfortunately so	5274	The default compile environment on Solaris is unfortunately so
4502	thread-unsafe that you can't even use components/libraries compiled	5275	thread-unsafe that you can't even use components/libraries compiled
4503	without C<-D_REENTRANT> (as long as they use C<errno>), which, of course,	5276	without C<-D_REENTRANT> in a threaded program, which, of course, isn't
4504	isn't defined by default.	5277	defined by default. A valid, if stupid, implementation choice.
4505		5278
4506	If you want to use libev in threaded environments you have to make sure	5279	If you want to use libev in threaded environments you have to make sure
4507	it's compiled with C<_REENTRANT> defined.	5280	it's compiled with C<_REENTRANT> defined.
4508		5281
4509	=head3 Event port backend	5282	=head3 Event port backend
4510		5283
4511	The scalable event interface for Solaris is called "event ports". Unfortunately,	5284	The scalable event interface for Solaris is called "event
4512	this mechanism is very buggy. If you run into high CPU usage, your program	5285	ports". Unfortunately, this mechanism is very buggy in all major
		5286	releases. If you run into high CPU usage, your program freezes or you get
4513	freezes or you get a large number of spurious wakeups, make sure you have	5287	a large number of spurious wakeups, make sure you have all the relevant
4514	all the relevant and latest kernel patches applied. No, I don't know which	5288	and latest kernel patches applied. No, I don't know which ones, but there
4515	ones, but there are multiple ones.	5289	are multiple ones to apply, and afterwards, event ports actually work
		5290	great.
4516		5291
4517	If you can't get it to work, you can try running the program by setting	5292	If you can't get it to work, you can try running the program by setting
4518	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and	5293	the environment variable C<LIBEV_FLAGS=3> to only allow C<poll> and
4519	C<select> backends.	5294	C<select> backends.
4520		5295
4521	=head2 AIX POLL BUG	5296	=head2 AIX POLL BUG
4522		5297
4523	AIX unfortunately has a broken C<poll.h> header. Libev works around	5298	AIX unfortunately has a broken C<poll.h> header. Libev works around
4524	this by trying to avoid the poll backend altogether (i.e. it's not even	5299	this by trying to avoid the poll backend altogether (i.e. it's not even
4525	compiled in), which normally isn't a big problem as C<select> works fine	5300	compiled in), which normally isn't a big problem as C<select> works fine
4526	with large bitsets, and AIX is dead anyway.	5301	with large bitsets on AIX, and AIX is dead anyway.
4527		5302
4528	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS	5303	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
4529		5304
4530	=head3 General issues	5305	=head3 General issues
4531		5306
…		…
4533	requires, and its I/O model is fundamentally incompatible with the POSIX	5308	requires, and its I/O model is fundamentally incompatible with the POSIX
4534	model. Libev still offers limited functionality on this platform in	5309	model. Libev still offers limited functionality on this platform in
4535	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	5310	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
4536	descriptors. This only applies when using Win32 natively, not when using	5311	descriptors. This only applies when using Win32 natively, not when using
4537	e.g. cygwin. Actually, it only applies to the microsofts own compilers,	5312	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
4538	as every compielr comes with a slightly differently broken/incompatible	5313	as every compiler comes with a slightly differently broken/incompatible
4539	environment.	5314	environment.
4540		5315
4541	Lifting these limitations would basically require the full	5316	Lifting these limitations would basically require the full
4542	re-implementation of the I/O system. If you are into this kind of thing,	5317	re-implementation of the I/O system. If you are into this kind of thing,
4543	then note that glib does exactly that for you in a very portable way (note	5318	then note that glib does exactly that for you in a very portable way (note
…		…
4637	structure (guaranteed by POSIX but not by ISO C for example), but it also	5412	structure (guaranteed by POSIX but not by ISO C for example), but it also
4638	assumes that the same (machine) code can be used to call any watcher	5413	assumes that the same (machine) code can be used to call any watcher
4639	callback: The watcher callbacks have different type signatures, but libev	5414	callback: The watcher callbacks have different type signatures, but libev
4640	calls them using an C<ev_watcher *> internally.	5415	calls them using an C<ev_watcher *> internally.
4641		5416
		5417	=item null pointers and integer zero are represented by 0 bytes
		5418
		5419	Libev uses C<memset> to initialise structs and arrays to C<0> bytes, and
		5420	relies on this setting pointers and integers to null.
		5421
		5422	=item pointer accesses must be thread-atomic
		5423
		5424	Accessing a pointer value must be atomic, it must both be readable and
		5425	writable in one piece - this is the case on all current architectures.
		5426
4642	=item C<sig_atomic_t volatile> must be thread-atomic as well	5427	=item C<sig_atomic_t volatile> must be thread-atomic as well
4643		5428
4644	The type C<sig_atomic_t volatile> (or whatever is defined as	5429	The type C<sig_atomic_t volatile> (or whatever is defined as
4645	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different	5430	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
4646	threads. This is not part of the specification for C<sig_atomic_t>, but is	5431	threads. This is not part of the specification for C<sig_atomic_t>, but is
…		…
4654	thread" or will block signals process-wide, both behaviours would	5439	thread" or will block signals process-wide, both behaviours would
4655	be compatible with libev. Interaction between C<sigprocmask> and	5440	be compatible with libev. Interaction between C<sigprocmask> and
4656	C<pthread_sigmask> could complicate things, however.	5441	C<pthread_sigmask> could complicate things, however.
4657		5442
4658	The most portable way to handle signals is to block signals in all threads	5443	The most portable way to handle signals is to block signals in all threads
4659	except the initial one, and run the default loop in the initial thread as	5444	except the initial one, and run the signal handling loop in the initial
4660	well.	5445	thread as well.
4661		5446
4662	=item C<long> must be large enough for common memory allocation sizes	5447	=item C<long> must be large enough for common memory allocation sizes
4663		5448
4664	To improve portability and simplify its API, libev uses C<long> internally	5449	To improve portability and simplify its API, libev uses C<long> internally
4665	instead of C<size_t> when allocating its data structures. On non-POSIX	5450	instead of C<size_t> when allocating its data structures. On non-POSIX
…		…
4671		5456
4672	The type C<double> is used to represent timestamps. It is required to	5457	The type C<double> is used to represent timestamps. It is required to
4673	have at least 51 bits of mantissa (and 9 bits of exponent), which is	5458	have at least 51 bits of mantissa (and 9 bits of exponent), which is
4674	good enough for at least into the year 4000 with millisecond accuracy	5459	good enough for at least into the year 4000 with millisecond accuracy
4675	(the design goal for libev). This requirement is overfulfilled by	5460	(the design goal for libev). This requirement is overfulfilled by
4676	implementations using IEEE 754, which is basically all existing ones. With	5461	implementations using IEEE 754, which is basically all existing ones.
		5462
4677	IEEE 754 doubles, you get microsecond accuracy until at least 2200.	5463	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5464	year 2255 (and millisecond accuracy till the year 287396 - by then, libev
		5465	is either obsolete or somebody patched it to use C<long double> or
		5466	something like that, just kidding).
4678		5467
4679	=back	5468	=back
4680		5469
4681	If you know of other additional requirements drop me a note.	5470	If you know of other additional requirements drop me a note.
4682		5471
…		…
4744	=item Processing ev_async_send: O(number_of_async_watchers)	5533	=item Processing ev_async_send: O(number_of_async_watchers)
4745		5534
4746	=item Processing signals: O(max_signal_number)	5535	=item Processing signals: O(max_signal_number)
4747		5536
4748	Sending involves a system call I<iff> there were no other C<ev_async_send>	5537	Sending involves a system call I<iff> there were no other C<ev_async_send>
4749	calls in the current loop iteration. Checking for async and signal events	5538	calls in the current loop iteration and the loop is currently
		5539	blocked. Checking for async and signal events involves iterating over all
4750	involves iterating over all running async watchers or all signal numbers.	5540	running async watchers or all signal numbers.
4751		5541
4752	=back	5542	=back
4753		5543
4754		5544
4755	=head1 PORTING FROM LIBEV 3.X TO 4.X	5545	=head1 PORTING FROM LIBEV 3.X TO 4.X
4756		5546
4757	The major version 4 introduced some minor incompatible changes to the API.	5547	The major version 4 introduced some incompatible changes to the API.
4758		5548
4759	At the moment, the C<ev.h> header file tries to implement superficial	5549	At the moment, the C<ev.h> header file provides compatibility definitions
4760	compatibility, so most programs should still compile. Those might be	5550	for all changes, so most programs should still compile. The compatibility
4761	removed in later versions of libev, so better update early than late.	5551	layer might be removed in later versions of libev, so better update to the
		5552	new API early than late.
4762		5553
4763	=over 4	5554	=over 4
		5555
		5556	=item C<EV_COMPAT3> backwards compatibility mechanism
		5557
		5558	The backward compatibility mechanism can be controlled by
		5559	C<EV_COMPAT3>. See L</"PREPROCESSOR SYMBOLS/MACROS"> in the L</EMBEDDING>
		5560	section.
		5561
		5562	=item C<ev_default_destroy> and C<ev_default_fork> have been removed
		5563
		5564	These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
		5565
		5566	ev_loop_destroy (EV_DEFAULT_UC);
		5567	ev_loop_fork (EV_DEFAULT);
4764		5568
4765	=item function/symbol renames	5569	=item function/symbol renames
4766		5570
4767	A number of functions and symbols have been renamed:	5571	A number of functions and symbols have been renamed:
4768		5572
…		…
4787	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme	5591	ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
4788	as all other watcher types. Note that C<ev_loop_fork> is still called	5592	as all other watcher types. Note that C<ev_loop_fork> is still called
4789	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>	5593	C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
4790	typedef.	5594	typedef.
4791		5595
4792	=item C<EV_COMPAT3> backwards compatibility mechanism
4793
4794	The backward compatibility mechanism can be controlled by
4795	C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
4796	section.
4797
4798	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>	5596	=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
4799		5597
4800	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different	5598	The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
4801	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile	5599	mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
4802	and work, but the library code will of course be larger.	5600	and work, but the library code will of course be larger.
…		…
4808		5606
4809	=over 4	5607	=over 4
4810		5608
4811	=item active	5609	=item active
4812		5610
4813	A watcher is active as long as it has been started (has been attached to	5611	A watcher is active as long as it has been started and not yet stopped.
4814	an event loop) but not yet stopped (disassociated from the event loop).	5612	See L</WATCHER STATES> for details.
4815		5613
4816	=item application	5614	=item application
4817		5615
4818	In this document, an application is whatever is using libev.	5616	In this document, an application is whatever is using libev.
		5617
		5618	=item backend
		5619
		5620	The part of the code dealing with the operating system interfaces.
4819		5621
4820	=item callback	5622	=item callback
4821		5623
4822	The address of a function that is called when some event has been	5624	The address of a function that is called when some event has been
4823	detected. Callbacks are being passed the event loop, the watcher that	5625	detected. Callbacks are being passed the event loop, the watcher that
4824	received the event, and the actual event bitset.	5626	received the event, and the actual event bitset.
4825		5627
4826	=item callback invocation	5628	=item callback/watcher invocation
4827		5629
4828	The act of calling the callback associated with a watcher.	5630	The act of calling the callback associated with a watcher.
4829		5631
4830	=item event	5632	=item event
4831		5633
…		…
4850	The model used to describe how an event loop handles and processes	5652	The model used to describe how an event loop handles and processes
4851	watchers and events.	5653	watchers and events.
4852		5654
4853	=item pending	5655	=item pending
4854		5656
4855	A watcher is pending as soon as the corresponding event has been detected,	5657	A watcher is pending as soon as the corresponding event has been
4856	and stops being pending as soon as the watcher will be invoked or its	5658	detected. See L</WATCHER STATES> for details.
4857	pending status is explicitly cleared by the application.
4858
4859	A watcher can be pending, but not active. Stopping a watcher also clears
4860	its pending status.
4861		5659
4862	=item real time	5660	=item real time
4863		5661
4864	The physical time that is observed. It is apparently strictly monotonic :)	5662	The physical time that is observed. It is apparently strictly monotonic :)
4865		5663
4866	=item wall-clock time	5664	=item wall-clock time
4867		5665
4868	The time and date as shown on clocks. Unlike real time, it can actually	5666	The time and date as shown on clocks. Unlike real time, it can actually
4869	be wrong and jump forwards and backwards, e.g. when the you adjust your	5667	be wrong and jump forwards and backwards, e.g. when you adjust your
4870	clock.	5668	clock.
4871		5669
4872	=item watcher	5670	=item watcher
4873		5671
4874	A data structure that describes interest in certain events. Watchers need	5672	A data structure that describes interest in certain events. Watchers need
4875	to be started (attached to an event loop) before they can receive events.	5673	to be started (attached to an event loop) before they can receive events.
4876		5674
4877	=item watcher invocation
4878
4879	The act of calling the callback associated with a watcher.
4880
4881	=back	5675	=back
4882		5676
4883	=head1 AUTHOR	5677	=head1 AUTHOR
4884		5678
4885	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson.	5679	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
		5680	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
4886		5681

Diff Legend

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