ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.159 by root, Thu May 22 02:44:57 2008 UTC vs.
Revision 1.200 by root, Thu Oct 23 07:33:45 2008 UTC

…		…
2		2
3	libev - a high performance full-featured event loop written in C	3	libev - a high performance full-featured event loop written in C
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	#include <ev.h>	7	#include <ev.h>
8		8
9	=head2 EXAMPLE PROGRAM	9	=head2 EXAMPLE PROGRAM
10		10
11	// a single header file is required	11	// a single header file is required
12	#include <ev.h>	12	#include <ev.h>
13		13
14	// every watcher type has its own typedef'd struct	14	// every watcher type has its own typedef'd struct
15	// with the name ev_<type>	15	// with the name ev_TYPE
16	ev_io stdin_watcher;	16	ev_io stdin_watcher;
17	ev_timer timeout_watcher;	17	ev_timer timeout_watcher;
18		18
19	// all watcher callbacks have a similar signature	19	// all watcher callbacks have a similar signature
20	// this callback is called when data is readable on stdin	20	// this callback is called when data is readable on stdin
21	static void	21	static void
22	stdin_cb (EV_P_ struct ev_io *w, int revents)	22	stdin_cb (EV_P_ ev_io *w, int revents)
23	{	23	{
24	puts ("stdin ready");	24	puts ("stdin ready");
25	// for one-shot events, one must manually stop the watcher	25	// for one-shot events, one must manually stop the watcher
26	// with its corresponding stop function.	26	// with its corresponding stop function.
27	ev_io_stop (EV_A_ w);	27	ev_io_stop (EV_A_ w);
28		28
29	// this causes all nested ev_loop's to stop iterating	29	// this causes all nested ev_loop's to stop iterating
30	ev_unloop (EV_A_ EVUNLOOP_ALL);	30	ev_unloop (EV_A_ EVUNLOOP_ALL);
31	}	31	}
32		32
33	// another callback, this time for a time-out	33	// another callback, this time for a time-out
34	static void	34	static void
35	timeout_cb (EV_P_ struct ev_timer *w, int revents)	35	timeout_cb (EV_P_ ev_timer *w, int revents)
36	{	36	{
37	puts ("timeout");	37	puts ("timeout");
38	// this causes the innermost ev_loop to stop iterating	38	// this causes the innermost ev_loop to stop iterating
39	ev_unloop (EV_A_ EVUNLOOP_ONE);	39	ev_unloop (EV_A_ EVUNLOOP_ONE);
40	}	40	}
41		41
42	int	42	int
43	main (void)	43	main (void)
44	{	44	{
45	// use the default event loop unless you have special needs	45	// use the default event loop unless you have special needs
46	struct ev_loop *loop = ev_default_loop (0);	46	ev_loop *loop = ev_default_loop (0);
47		47
48	// initialise an io watcher, then start it	48	// initialise an io watcher, then start it
49	// this one will watch for stdin to become readable	49	// this one will watch for stdin to become readable
50	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);	50	ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
51	ev_io_start (loop, &stdin_watcher);	51	ev_io_start (loop, &stdin_watcher);
52		52
53	// initialise a timer watcher, then start it	53	// initialise a timer watcher, then start it
54	// simple non-repeating 5.5 second timeout	54	// simple non-repeating 5.5 second timeout
55	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);	55	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
56	ev_timer_start (loop, &timeout_watcher);	56	ev_timer_start (loop, &timeout_watcher);
57		57
58	// now wait for events to arrive	58	// now wait for events to arrive
59	ev_loop (loop, 0);	59	ev_loop (loop, 0);
60		60
61	// unloop was called, so exit	61	// unloop was called, so exit
62	return 0;	62	return 0;
63	}	63	}
64		64
65	=head1 DESCRIPTION	65	=head1 DESCRIPTION
66		66
67	The newest version of this document is also available as an html-formatted	67	The newest version of this document is also available as an html-formatted
68	web page you might find easier to navigate when reading it for the first	68	web page you might find easier to navigate when reading it for the first
…		…
103	Libev is very configurable. In this manual the default (and most common)	103	Libev is very configurable. In this manual the default (and most common)
104	configuration will be described, which supports multiple event loops. For	104	configuration will be described, which supports multiple event loops. For
105	more info about various configuration options please have a look at	105	more info about various configuration options please have a look at
106	B<EMBED> section in this manual. If libev was configured without support	106	B<EMBED> section in this manual. If libev was configured without support
107	for multiple event loops, then all functions taking an initial argument of	107	for multiple event loops, then all functions taking an initial argument of
108	name C<loop> (which is always of type C<struct ev_loop *>) will not have	108	name C<loop> (which is always of type C<ev_loop *>) will not have
109	this argument.	109	this argument.
110		110
111	=head2 TIME REPRESENTATION	111	=head2 TIME REPRESENTATION
112		112
113	Libev represents time as a single floating point number, representing the	113	Libev represents time as a single floating point number, representing the
114	(fractional) number of seconds since the (POSIX) epoch (somewhere near	114	(fractional) number of seconds since the (POSIX) epoch (somewhere near
115	the beginning of 1970, details are complicated, don't ask). This type is	115	the beginning of 1970, details are complicated, don't ask). This type is
116	called C<ev_tstamp>, which is what you should use too. It usually aliases	116	called C<ev_tstamp>, which is what you should use too. It usually aliases
117	to the C<double> type in C, and when you need to do any calculations on	117	to the C<double> type in C, and when you need to do any calculations on
118	it, you should treat it as some floatingpoint value. Unlike the name	118	it, you should treat it as some floating point value. Unlike the name
119	component C<stamp> might indicate, it is also used for time differences	119	component C<stamp> might indicate, it is also used for time differences
120	throughout libev.	120	throughout libev.
		121
		122	=head1 ERROR HANDLING
		123
		124	Libev knows three classes of errors: operating system errors, usage errors
		125	and internal errors (bugs).
		126
		127	When libev catches an operating system error it cannot handle (for example
		128	a system call indicating a condition libev cannot fix), it calls the callback
		129	set via C<ev_set_syserr_cb>, which is supposed to fix the problem or
		130	abort. The default is to print a diagnostic message and to call C<abort
		131	()>.
		132
		133	When libev detects a usage error such as a negative timer interval, then
		134	it will print a diagnostic message and abort (via the C<assert> mechanism,
		135	so C<NDEBUG> will disable this checking): these are programming errors in
		136	the libev caller and need to be fixed there.
		137
		138	Libev also has a few internal error-checking C<assert>ions, and also has
		139	extensive consistency checking code. These do not trigger under normal
		140	circumstances, as they indicate either a bug in libev or worse.
		141
121		142
122	=head1 GLOBAL FUNCTIONS	143	=head1 GLOBAL FUNCTIONS
123		144
124	These functions can be called anytime, even before initialising the	145	These functions can be called anytime, even before initialising the
125	library in any way.	146	library in any way.
…		…
134		155
135	=item ev_sleep (ev_tstamp interval)	156	=item ev_sleep (ev_tstamp interval)
136		157
137	Sleep for the given interval: The current thread will be blocked until	158	Sleep for the given interval: The current thread will be blocked until
138	either it is interrupted or the given time interval has passed. Basically	159	either it is interrupted or the given time interval has passed. Basically
139	this is a subsecond-resolution C<sleep ()>.	160	this is a sub-second-resolution C<sleep ()>.
140		161
141	=item int ev_version_major ()	162	=item int ev_version_major ()
142		163
143	=item int ev_version_minor ()	164	=item int ev_version_minor ()
144		165
…		…
157	not a problem.	178	not a problem.
158		179
159	Example: Make sure we haven't accidentally been linked against the wrong	180	Example: Make sure we haven't accidentally been linked against the wrong
160	version.	181	version.
161		182
162	assert (("libev version mismatch",	183	assert (("libev version mismatch",
163	ev_version_major () == EV_VERSION_MAJOR	184	ev_version_major () == EV_VERSION_MAJOR
164	&& ev_version_minor () >= EV_VERSION_MINOR));	185	&& ev_version_minor () >= EV_VERSION_MINOR));
165		186
166	=item unsigned int ev_supported_backends ()	187	=item unsigned int ev_supported_backends ()
167		188
168	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>	189	Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
169	value) compiled into this binary of libev (independent of their	190	value) compiled into this binary of libev (independent of their
…		…
171	a description of the set values.	192	a description of the set values.
172		193
173	Example: make sure we have the epoll method, because yeah this is cool and	194	Example: make sure we have the epoll method, because yeah this is cool and
174	a must have and can we have a torrent of it please!!!11	195	a must have and can we have a torrent of it please!!!11
175		196
176	assert (("sorry, no epoll, no sex",	197	assert (("sorry, no epoll, no sex",
177	ev_supported_backends () & EVBACKEND_EPOLL));	198	ev_supported_backends () & EVBACKEND_EPOLL));
178		199
179	=item unsigned int ev_recommended_backends ()	200	=item unsigned int ev_recommended_backends ()
180		201
181	Return the set of all backends compiled into this binary of libev and also	202	Return the set of all backends compiled into this binary of libev and also
182	recommended for this platform. This set is often smaller than the one	203	recommended for this platform. This set is often smaller than the one
183	returned by C<ev_supported_backends>, as for example kqueue is broken on	204	returned by C<ev_supported_backends>, as for example kqueue is broken on
184	most BSDs and will not be autodetected unless you explicitly request it	205	most BSDs and will not be auto-detected unless you explicitly request it
185	(assuming you know what you are doing). This is the set of backends that	206	(assuming you know what you are doing). This is the set of backends that
186	libev will probe for if you specify no backends explicitly.	207	libev will probe for if you specify no backends explicitly.
187		208
188	=item unsigned int ev_embeddable_backends ()	209	=item unsigned int ev_embeddable_backends ()
189		210
…		…
193	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	214	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
194	recommended ones.	215	recommended ones.
195		216
196	See the description of C<ev_embed> watchers for more info.	217	See the description of C<ev_embed> watchers for more info.
197		218
198	=item ev_set_allocator (void *(*cb)(void *ptr, long size))	219	=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT]
199		220
200	Sets the allocation function to use (the prototype is similar - the	221	Sets the allocation function to use (the prototype is similar - the
201	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is	222	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
202	used to allocate and free memory (no surprises here). If it returns zero	223	used to allocate and free memory (no surprises here). If it returns zero
203	when memory needs to be allocated (C<size != 0>), the library might abort	224	when memory needs to be allocated (C<size != 0>), the library might abort
…		…
229	}	250	}
230		251
231	...	252	...
232	ev_set_allocator (persistent_realloc);	253	ev_set_allocator (persistent_realloc);
233		254
234	=item ev_set_syserr_cb (void (*cb)(const char *msg));	255	=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT]
235		256
236	Set the callback function to call on a retryable syscall error (such	257	Set the callback function to call on a retryable system call error (such
237	as failed select, poll, epoll_wait). The message is a printable string	258	as failed select, poll, epoll_wait). The message is a printable string
238	indicating the system call or subsystem causing the problem. If this	259	indicating the system call or subsystem causing the problem. If this
239	callback is set, then libev will expect it to remedy the sitution, no	260	callback is set, then libev will expect it to remedy the situation, no
240	matter what, when it returns. That is, libev will generally retry the	261	matter what, when it returns. That is, libev will generally retry the
241	requested operation, or, if the condition doesn't go away, do bad stuff	262	requested operation, or, if the condition doesn't go away, do bad stuff
242	(such as abort).	263	(such as abort).
243		264
244	Example: This is basically the same thing that libev does internally, too.	265	Example: This is basically the same thing that libev does internally, too.
…		…
255		276
256	=back	277	=back
257		278
258	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP	279	=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
259		280
260	An event loop is described by a C<struct ev_loop *>. The library knows two	281	An event loop is described by a C<struct ev_loop *> (the C<struct>
261	types of such loops, the I<default> loop, which supports signals and child	282	is I<not> optional in this case, as there is also an C<ev_loop>
262	events, and dynamically created loops which do not.	283	I<function>).
		284
		285	The library knows two types of such loops, the I<default> loop, which
		286	supports signals and child events, and dynamically created loops which do
		287	not.
263		288
264	=over 4	289	=over 4
265		290
266	=item struct ev_loop *ev_default_loop (unsigned int flags)	291	=item struct ev_loop *ev_default_loop (unsigned int flags)
267		292
…		…
277	from multiple threads, you have to lock (note also that this is unlikely,	302	from multiple threads, you have to lock (note also that this is unlikely,
278	as loops cannot bes hared easily between threads anyway).	303	as loops cannot bes hared easily between threads anyway).
279		304
280	The default loop is the only loop that can handle C<ev_signal> and	305	The default loop is the only loop that can handle C<ev_signal> and
281	C<ev_child> watchers, and to do this, it always registers a handler	306	C<ev_child> watchers, and to do this, it always registers a handler
282	for C<SIGCHLD>. If this is a problem for your app you can either	307	for C<SIGCHLD>. If this is a problem for your application you can either
283	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you	308	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
284	can simply overwrite the C<SIGCHLD> signal handler I<after> calling	309	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
285	C<ev_default_init>.	310	C<ev_default_init>.
286		311
287	The flags argument can be used to specify special behaviour or specific	312	The flags argument can be used to specify special behaviour or specific
…		…
296	The default flags value. Use this if you have no clue (it's the right	321	The default flags value. Use this if you have no clue (it's the right
297	thing, believe me).	322	thing, believe me).
298		323
299	=item C<EVFLAG_NOENV>	324	=item C<EVFLAG_NOENV>
300		325
301	If this flag bit is ored into the flag value (or the program runs setuid	326	If this flag bit is or'ed into the flag value (or the program runs setuid
302	or setgid) then libev will I<not> look at the environment variable	327	or setgid) then libev will I<not> look at the environment variable
303	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	328	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
304	override the flags completely if it is found in the environment. This is	329	override the flags completely if it is found in the environment. This is
305	useful to try out specific backends to test their performance, or to work	330	useful to try out specific backends to test their performance, or to work
306	around bugs.	331	around bugs.
…		…
313		338
314	This works by calling C<getpid ()> on every iteration of the loop,	339	This works by calling C<getpid ()> on every iteration of the loop,
315	and thus this might slow down your event loop if you do a lot of loop	340	and thus this might slow down your event loop if you do a lot of loop
316	iterations and little real work, but is usually not noticeable (on my	341	iterations and little real work, but is usually not noticeable (on my
317	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence	342	GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
318	without a syscall and thus I<very> fast, but my GNU/Linux system also has	343	without a system call and thus I<very> fast, but my GNU/Linux system also has
319	C<pthread_atfork> which is even faster).	344	C<pthread_atfork> which is even faster).
320		345
321	The big advantage of this flag is that you can forget about fork (and	346	The big advantage of this flag is that you can forget about fork (and
322	forget about forgetting to tell libev about forking) when you use this	347	forget about forgetting to tell libev about forking) when you use this
323	flag.	348	flag.
324		349
325	This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS>	350	This flag setting cannot be overridden or specified in the C<LIBEV_FLAGS>
326	environment variable.	351	environment variable.
327		352
328	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	353	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
329		354
330	This is your standard select(2) backend. Not I<completely> standard, as	355	This is your standard select(2) backend. Not I<completely> standard, as
…		…
332	but if that fails, expect a fairly low limit on the number of fds when	357	but if that fails, expect a fairly low limit on the number of fds when
333	using this backend. It doesn't scale too well (O(highest_fd)), but its	358	using this backend. It doesn't scale too well (O(highest_fd)), but its
334	usually the fastest backend for a low number of (low-numbered :) fds.	359	usually the fastest backend for a low number of (low-numbered :) fds.
335		360
336	To get good performance out of this backend you need a high amount of	361	To get good performance out of this backend you need a high amount of
337	parallelity (most of the file descriptors should be busy). If you are	362	parallelism (most of the file descriptors should be busy). If you are
338	writing a server, you should C<accept ()> in a loop to accept as many	363	writing a server, you should C<accept ()> in a loop to accept as many
339	connections as possible during one iteration. You might also want to have	364	connections as possible during one iteration. You might also want to have
340	a look at C<ev_set_io_collect_interval ()> to increase the amount of	365	a look at C<ev_set_io_collect_interval ()> to increase the amount of
341	readiness notifications you get per iteration.	366	readiness notifications you get per iteration.
		367
		368	This backend maps C<EV_READ> to the C<readfds> set and C<EV_WRITE> to the
		369	C<writefds> set (and to work around Microsoft Windows bugs, also onto the
		370	C<exceptfds> set on that platform).
342		371
343	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	372	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
344		373
345	And this is your standard poll(2) backend. It's more complicated	374	And this is your standard poll(2) backend. It's more complicated
346	than select, but handles sparse fds better and has no artificial	375	than select, but handles sparse fds better and has no artificial
347	limit on the number of fds you can use (except it will slow down	376	limit on the number of fds you can use (except it will slow down
348	considerably with a lot of inactive fds). It scales similarly to select,	377	considerably with a lot of inactive fds). It scales similarly to select,
349	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for	378	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
350	performance tips.	379	performance tips.
351		380
		381	This backend maps C<EV_READ> to C<POLLIN | POLLERR | POLLHUP>, and
		382	C<EV_WRITE> to C<POLLOUT | POLLERR | POLLHUP>.
		383
352	=item C<EVBACKEND_EPOLL> (value 4, Linux)	384	=item C<EVBACKEND_EPOLL> (value 4, Linux)
353		385
354	For few fds, this backend is a bit little slower than poll and select,	386	For few fds, this backend is a bit little slower than poll and select,
355	but it scales phenomenally better. While poll and select usually scale	387	but it scales phenomenally better. While poll and select usually scale
356	like O(total_fds) where n is the total number of fds (or the highest fd),	388	like O(total_fds) where n is the total number of fds (or the highest fd),
357	epoll scales either O(1) or O(active_fds). The epoll design has a number	389	epoll scales either O(1) or O(active_fds). The epoll design has a number
358	of shortcomings, such as silently dropping events in some hard-to-detect	390	of shortcomings, such as silently dropping events in some hard-to-detect
359	cases and requiring a syscall per fd change, no fork support and bad	391	cases and requiring a system call per fd change, no fork support and bad
360	support for dup.	392	support for dup.
361		393
362	While stopping, setting and starting an I/O watcher in the same iteration	394	While stopping, setting and starting an I/O watcher in the same iteration
363	will result in some caching, there is still a syscall per such incident	395	will result in some caching, there is still a system call per such incident
364	(because the fd could point to a different file description now), so its	396	(because the fd could point to a different file description now), so its
365	best to avoid that. Also, C<dup ()>'ed file descriptors might not work	397	best to avoid that. Also, C<dup ()>'ed file descriptors might not work
366	very well if you register events for both fds.	398	very well if you register events for both fds.
367		399
368	Please note that epoll sometimes generates spurious notifications, so you	400	Please note that epoll sometimes generates spurious notifications, so you
369	need to use non-blocking I/O or other means to avoid blocking when no data	401	need to use non-blocking I/O or other means to avoid blocking when no data
370	(or space) is available.	402	(or space) is available.
371		403
372	Best performance from this backend is achieved by not unregistering all	404	Best performance from this backend is achieved by not unregistering all
373	watchers for a file descriptor until it has been closed, if possible, i.e.	405	watchers for a file descriptor until it has been closed, if possible,
374	keep at least one watcher active per fd at all times.	406	i.e. keep at least one watcher active per fd at all times. Stopping and
		407	starting a watcher (without re-setting it) also usually doesn't cause
		408	extra overhead.
375		409
376	While nominally embeddeble in other event loops, this feature is broken in	410	While nominally embeddable in other event loops, this feature is broken in
377	all kernel versions tested so far.	411	all kernel versions tested so far.
378		412
		413	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		414	C<EVBACKEND_POLL>.
		415
379	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	416	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
380		417
381	Kqueue deserves special mention, as at the time of this writing, it	418	Kqueue deserves special mention, as at the time of this writing, it was
382	was broken on all BSDs except NetBSD (usually it doesn't work reliably	419	broken on all BSDs except NetBSD (usually it doesn't work reliably with
383	with anything but sockets and pipes, except on Darwin, where of course	420	anything but sockets and pipes, except on Darwin, where of course it's
384	it's completely useless). For this reason it's not being "autodetected"	421	completely useless). For this reason it's not being "auto-detected" unless
385	unless you explicitly specify it explicitly in the flags (i.e. using	422	you explicitly specify it in the flags (i.e. using C<EVBACKEND_KQUEUE>) or
386	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	423	libev was compiled on a known-to-be-good (-enough) system like NetBSD.
387	system like NetBSD.
388		424
389	You still can embed kqueue into a normal poll or select backend and use it	425	You still can embed kqueue into a normal poll or select backend and use it
390	only for sockets (after having made sure that sockets work with kqueue on	426	only for sockets (after having made sure that sockets work with kqueue on
391	the target platform). See C<ev_embed> watchers for more info.	427	the target platform). See C<ev_embed> watchers for more info.
392		428
393	It scales in the same way as the epoll backend, but the interface to the	429	It scales in the same way as the epoll backend, but the interface to the
394	kernel is more efficient (which says nothing about its actual speed, of	430	kernel is more efficient (which says nothing about its actual speed, of
395	course). While stopping, setting and starting an I/O watcher does never	431	course). While stopping, setting and starting an I/O watcher does never
396	cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to	432	cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
397	two event changes per incident, support for C<fork ()> is very bad and it	433	two event changes per incident. Support for C<fork ()> is very bad and it
398	drops fds silently in similarly hard-to-detect cases.	434	drops fds silently in similarly hard-to-detect cases.
399		435
400	This backend usually performs well under most conditions.	436	This backend usually performs well under most conditions.
401		437
402	While nominally embeddable in other event loops, this doesn't work	438	While nominally embeddable in other event loops, this doesn't work
403	everywhere, so you might need to test for this. And since it is broken	439	everywhere, so you might need to test for this. And since it is broken
404	almost everywhere, you should only use it when you have a lot of sockets	440	almost everywhere, you should only use it when you have a lot of sockets
405	(for which it usually works), by embedding it into another event loop	441	(for which it usually works), by embedding it into another event loop
406	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for	442	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and, did I mention it,
407	sockets.	443	using it only for sockets.
		444
		445	This backend maps C<EV_READ> into an C<EVFILT_READ> kevent with
		446	C<NOTE_EOF>, and C<EV_WRITE> into an C<EVFILT_WRITE> kevent with
		447	C<NOTE_EOF>.
408		448
409	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	449	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
410		450
411	This is not implemented yet (and might never be, unless you send me an	451	This is not implemented yet (and might never be, unless you send me an
412	implementation). According to reports, C</dev/poll> only supports sockets	452	implementation). According to reports, C</dev/poll> only supports sockets
…		…
416	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	456	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
417		457
418	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	458	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
419	it's really slow, but it still scales very well (O(active_fds)).	459	it's really slow, but it still scales very well (O(active_fds)).
420		460
421	Please note that solaris event ports can deliver a lot of spurious	461	Please note that Solaris event ports can deliver a lot of spurious
422	notifications, so you need to use non-blocking I/O or other means to avoid	462	notifications, so you need to use non-blocking I/O or other means to avoid
423	blocking when no data (or space) is available.	463	blocking when no data (or space) is available.
424		464
425	While this backend scales well, it requires one system call per active	465	While this backend scales well, it requires one system call per active
426	file descriptor per loop iteration. For small and medium numbers of file	466	file descriptor per loop iteration. For small and medium numbers of file
427	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend	467	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
428	might perform better.	468	might perform better.
429		469
430	On the positive side, ignoring the spurious readiness notifications, this	470	On the positive side, with the exception of the spurious readiness
431	backend actually performed to specification in all tests and is fully	471	notifications, this backend actually performed fully to specification
432	embeddable, which is a rare feat among the OS-specific backends.	472	in all tests and is fully embeddable, which is a rare feat among the
		473	OS-specific backends.
		474
		475	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
		476	C<EVBACKEND_POLL>.
433		477
434	=item C<EVBACKEND_ALL>	478	=item C<EVBACKEND_ALL>
435		479
436	Try all backends (even potentially broken ones that wouldn't be tried	480	Try all backends (even potentially broken ones that wouldn't be tried
437	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	481	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
…		…
439		483
440	It is definitely not recommended to use this flag.	484	It is definitely not recommended to use this flag.
441		485
442	=back	486	=back
443		487
444	If one or more of these are ored into the flags value, then only these	488	If one or more of these are or'ed into the flags value, then only these
445	backends will be tried (in the reverse order as listed here). If none are	489	backends will be tried (in the reverse order as listed here). If none are
446	specified, all backends in C<ev_recommended_backends ()> will be tried.	490	specified, all backends in C<ev_recommended_backends ()> will be tried.
447		491
448	The most typical usage is like this:	492	Example: This is the most typical usage.
449		493
450	if (!ev_default_loop (0))	494	if (!ev_default_loop (0))
451	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");	495	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
452		496
453	Restrict libev to the select and poll backends, and do not allow	497	Example: Restrict libev to the select and poll backends, and do not allow
454	environment settings to be taken into account:	498	environment settings to be taken into account:
455		499
456	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);	500	ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
457		501
458	Use whatever libev has to offer, but make sure that kqueue is used if	502	Example: Use whatever libev has to offer, but make sure that kqueue is
459	available (warning, breaks stuff, best use only with your own private	503	used if available (warning, breaks stuff, best use only with your own
460	event loop and only if you know the OS supports your types of fds):	504	private event loop and only if you know the OS supports your types of
		505	fds):
461		506
462	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);	507	ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
463		508
464	=item struct ev_loop *ev_loop_new (unsigned int flags)	509	=item struct ev_loop *ev_loop_new (unsigned int flags)
465		510
466	Similar to C<ev_default_loop>, but always creates a new event loop that is	511	Similar to C<ev_default_loop>, but always creates a new event loop that is
467	always distinct from the default loop. Unlike the default loop, it cannot	512	always distinct from the default loop. Unlike the default loop, it cannot
…		…
472	libev with threads is indeed to create one loop per thread, and using the	517	libev with threads is indeed to create one loop per thread, and using the
473	default loop in the "main" or "initial" thread.	518	default loop in the "main" or "initial" thread.
474		519
475	Example: Try to create a event loop that uses epoll and nothing else.	520	Example: Try to create a event loop that uses epoll and nothing else.
476		521
477	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);	522	struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
478	if (!epoller)	523	if (!epoller)
479	fatal ("no epoll found here, maybe it hides under your chair");	524	fatal ("no epoll found here, maybe it hides under your chair");
480		525
481	=item ev_default_destroy ()	526	=item ev_default_destroy ()
482		527
483	Destroys the default loop again (frees all memory and kernel state	528	Destroys the default loop again (frees all memory and kernel state
484	etc.). None of the active event watchers will be stopped in the normal	529	etc.). None of the active event watchers will be stopped in the normal
485	sense, so e.g. C<ev_is_active> might still return true. It is your	530	sense, so e.g. C<ev_is_active> might still return true. It is your
486	responsibility to either stop all watchers cleanly yoursef I<before>	531	responsibility to either stop all watchers cleanly yourself I<before>
487	calling this function, or cope with the fact afterwards (which is usually	532	calling this function, or cope with the fact afterwards (which is usually
488	the easiest thing, you can just ignore the watchers and/or C<free ()> them	533	the easiest thing, you can just ignore the watchers and/or C<free ()> them
489	for example).	534	for example).
490		535
491	Note that certain global state, such as signal state, will not be freed by	536	Note that certain global state, such as signal state, will not be freed by
…		…
523		568
524	=item ev_loop_fork (loop)	569	=item ev_loop_fork (loop)
525		570
526	Like C<ev_default_fork>, but acts on an event loop created by	571	Like C<ev_default_fork>, but acts on an event loop created by
527	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	572	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
528	after fork, and how you do this is entirely your own problem.	573	after fork that you want to re-use in the child, and how you do this is
		574	entirely your own problem.
529		575
530	=item int ev_is_default_loop (loop)	576	=item int ev_is_default_loop (loop)
531		577
532	Returns true when the given loop actually is the default loop, false otherwise.	578	Returns true when the given loop is, in fact, the default loop, and false
		579	otherwise.
533		580
534	=item unsigned int ev_loop_count (loop)	581	=item unsigned int ev_loop_count (loop)
535		582
536	Returns the count of loop iterations for the loop, which is identical to	583	Returns the count of loop iterations for the loop, which is identical to
537	the number of times libev did poll for new events. It starts at C<0> and	584	the number of times libev did poll for new events. It starts at C<0> and
…		…
552	received events and started processing them. This timestamp does not	599	received events and started processing them. This timestamp does not
553	change as long as callbacks are being processed, and this is also the base	600	change as long as callbacks are being processed, and this is also the base
554	time used for relative timers. You can treat it as the timestamp of the	601	time used for relative timers. You can treat it as the timestamp of the
555	event occurring (or more correctly, libev finding out about it).	602	event occurring (or more correctly, libev finding out about it).
556		603
		604	=item ev_now_update (loop)
		605
		606	Establishes the current time by querying the kernel, updating the time
		607	returned by C<ev_now ()> in the progress. This is a costly operation and
		608	is usually done automatically within C<ev_loop ()>.
		609
		610	This function is rarely useful, but when some event callback runs for a
		611	very long time without entering the event loop, updating libev's idea of
		612	the current time is a good idea.
		613
		614	See also "The special problem of time updates" in the C<ev_timer> section.
		615
557	=item ev_loop (loop, int flags)	616	=item ev_loop (loop, int flags)
558		617
559	Finally, this is it, the event handler. This function usually is called	618	Finally, this is it, the event handler. This function usually is called
560	after you initialised all your watchers and you want to start handling	619	after you initialised all your watchers and you want to start handling
561	events.	620	events.
…		…
563	If the flags argument is specified as C<0>, it will not return until	622	If the flags argument is specified as C<0>, it will not return until
564	either no event watchers are active anymore or C<ev_unloop> was called.	623	either no event watchers are active anymore or C<ev_unloop> was called.
565		624
566	Please note that an explicit C<ev_unloop> is usually better than	625	Please note that an explicit C<ev_unloop> is usually better than
567	relying on all watchers to be stopped when deciding when a program has	626	relying on all watchers to be stopped when deciding when a program has
568	finished (especially in interactive programs), but having a program that	627	finished (especially in interactive programs), but having a program
569	automatically loops as long as it has to and no longer by virtue of	628	that automatically loops as long as it has to and no longer by virtue
570	relying on its watchers stopping correctly is a thing of beauty.	629	of relying on its watchers stopping correctly, that is truly a thing of
		630	beauty.
571		631
572	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle	632	A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
573	those events and any outstanding ones, but will not block your process in	633	those events and any already outstanding ones, but will not block your
574	case there are no events and will return after one iteration of the loop.	634	process in case there are no events and will return after one iteration of
		635	the loop.
575		636
576	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if	637	A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
577	neccessary) and will handle those and any outstanding ones. It will block	638	necessary) and will handle those and any already outstanding ones. It
578	your process until at least one new event arrives, and will return after	639	will block your process until at least one new event arrives (which could
579	one iteration of the loop. This is useful if you are waiting for some	640	be an event internal to libev itself, so there is no guarentee that a
580	external event in conjunction with something not expressible using other	641	user-registered callback will be called), and will return after one
		642	iteration of the loop.
		643
		644	This is useful if you are waiting for some external event in conjunction
		645	with something not expressible using other libev watchers (i.e. "roll your
581	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	646	own C<ev_loop>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
582	usually a better approach for this kind of thing.	647	usually a better approach for this kind of thing.
583		648
584	Here are the gory details of what C<ev_loop> does:	649	Here are the gory details of what C<ev_loop> does:
585		650
586	- Before the first iteration, call any pending watchers.	651	- Before the first iteration, call any pending watchers.
587	* If EVFLAG_FORKCHECK was used, check for a fork.	652	* If EVFLAG_FORKCHECK was used, check for a fork.
588	- If a fork was detected, queue and call all fork watchers.	653	- If a fork was detected (by any means), queue and call all fork watchers.
589	- Queue and call all prepare watchers.	654	- Queue and call all prepare watchers.
590	- If we have been forked, recreate the kernel state.	655	- If we have been forked, detach and recreate the kernel state
		656	as to not disturb the other process.
591	- Update the kernel state with all outstanding changes.	657	- Update the kernel state with all outstanding changes.
592	- Update the "event loop time".	658	- Update the "event loop time" (ev_now ()).
593	- Calculate for how long to sleep or block, if at all	659	- Calculate for how long to sleep or block, if at all
594	(active idle watchers, EVLOOP_NONBLOCK or not having	660	(active idle watchers, EVLOOP_NONBLOCK or not having
595	any active watchers at all will result in not sleeping).	661	any active watchers at all will result in not sleeping).
596	- Sleep if the I/O and timer collect interval say so.	662	- Sleep if the I/O and timer collect interval say so.
597	- Block the process, waiting for any events.	663	- Block the process, waiting for any events.
598	- Queue all outstanding I/O (fd) events.	664	- Queue all outstanding I/O (fd) events.
599	- Update the "event loop time" and do time jump handling.	665	- Update the "event loop time" (ev_now ()), and do time jump adjustments.
600	- Queue all outstanding timers.	666	- Queue all expired timers.
601	- Queue all outstanding periodics.	667	- Queue all expired periodics.
602	- If no events are pending now, queue all idle watchers.	668	- Unless any events are pending now, queue all idle watchers.
603	- Queue all check watchers.	669	- Queue all check watchers.
604	- Call all queued watchers in reverse order (i.e. check watchers first).	670	- Call all queued watchers in reverse order (i.e. check watchers first).
605	Signals and child watchers are implemented as I/O watchers, and will	671	Signals and child watchers are implemented as I/O watchers, and will
606	be handled here by queueing them when their watcher gets executed.	672	be handled here by queueing them when their watcher gets executed.
607	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	673	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
…		…
612	anymore.	678	anymore.
613		679
614	... queue jobs here, make sure they register event watchers as long	680	... queue jobs here, make sure they register event watchers as long
615	... as they still have work to do (even an idle watcher will do..)	681	... as they still have work to do (even an idle watcher will do..)
616	ev_loop (my_loop, 0);	682	ev_loop (my_loop, 0);
617	... jobs done. yeah!	683	... jobs done or somebody called unloop. yeah!
618		684
619	=item ev_unloop (loop, how)	685	=item ev_unloop (loop, how)
620		686
621	Can be used to make a call to C<ev_loop> return early (but only after it	687	Can be used to make a call to C<ev_loop> return early (but only after it
622	has processed all outstanding events). The C<how> argument must be either	688	has processed all outstanding events). The C<how> argument must be either
623	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	689	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
624	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	690	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
625		691
626	This "unloop state" will be cleared when entering C<ev_loop> again.	692	This "unloop state" will be cleared when entering C<ev_loop> again.
627		693
		694	It is safe to call C<ev_unloop> from otuside any C<ev_loop> calls.
		695
628	=item ev_ref (loop)	696	=item ev_ref (loop)
629		697
630	=item ev_unref (loop)	698	=item ev_unref (loop)
631		699
632	Ref/unref can be used to add or remove a reference count on the event	700	Ref/unref can be used to add or remove a reference count on the event
633	loop: Every watcher keeps one reference, and as long as the reference	701	loop: Every watcher keeps one reference, and as long as the reference
634	count is nonzero, C<ev_loop> will not return on its own. If you have	702	count is nonzero, C<ev_loop> will not return on its own.
		703
635	a watcher you never unregister that should not keep C<ev_loop> from	704	If you have a watcher you never unregister that should not keep C<ev_loop>
636	returning, ev_unref() after starting, and ev_ref() before stopping it. For	705	from returning, call ev_unref() after starting, and ev_ref() before
		706	stopping it.
		707
637	example, libev itself uses this for its internal signal pipe: It is not	708	As an example, libev itself uses this for its internal signal pipe: It is
638	visible to the libev user and should not keep C<ev_loop> from exiting if	709	not visible to the libev user and should not keep C<ev_loop> from exiting
639	no event watchers registered by it are active. It is also an excellent	710	if no event watchers registered by it are active. It is also an excellent
640	way to do this for generic recurring timers or from within third-party	711	way to do this for generic recurring timers or from within third-party
641	libraries. Just remember to I<unref after start> and I<ref before stop>	712	libraries. Just remember to I<unref after start> and I<ref before stop>
642	(but only if the watcher wasn't active before, or was active before,	713	(but only if the watcher wasn't active before, or was active before,
643	respectively).	714	respectively).
644		715
645	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	716	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
646	running when nothing else is active.	717	running when nothing else is active.
647		718
648	struct ev_signal exitsig;	719	ev_signal exitsig;
649	ev_signal_init (&exitsig, sig_cb, SIGINT);	720	ev_signal_init (&exitsig, sig_cb, SIGINT);
650	ev_signal_start (loop, &exitsig);	721	ev_signal_start (loop, &exitsig);
651	evf_unref (loop);	722	evf_unref (loop);
652		723
653	Example: For some weird reason, unregister the above signal handler again.	724	Example: For some weird reason, unregister the above signal handler again.
654		725
655	ev_ref (loop);	726	ev_ref (loop);
656	ev_signal_stop (loop, &exitsig);	727	ev_signal_stop (loop, &exitsig);
657		728
658	=item ev_set_io_collect_interval (loop, ev_tstamp interval)	729	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
659		730
660	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)	731	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
661		732
662	These advanced functions influence the time that libev will spend waiting	733	These advanced functions influence the time that libev will spend waiting
663	for events. Both are by default C<0>, meaning that libev will try to	734	for events. Both time intervals are by default C<0>, meaning that libev
664	invoke timer/periodic callbacks and I/O callbacks with minimum latency.	735	will try to invoke timer/periodic callbacks and I/O callbacks with minimum
		736	latency.
665		737
666	Setting these to a higher value (the C<interval> I<must> be >= C<0>)	738	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
667	allows libev to delay invocation of I/O and timer/periodic callbacks to	739	allows libev to delay invocation of I/O and timer/periodic callbacks
668	increase efficiency of loop iterations.	740	to increase efficiency of loop iterations (or to increase power-saving
		741	opportunities).
669		742
670	The background is that sometimes your program runs just fast enough to	743	The idea is that sometimes your program runs just fast enough to handle
671	handle one (or very few) event(s) per loop iteration. While this makes	744	one (or very few) event(s) per loop iteration. While this makes the
672	the program responsive, it also wastes a lot of CPU time to poll for new	745	program responsive, it also wastes a lot of CPU time to poll for new
673	events, especially with backends like C<select ()> which have a high	746	events, especially with backends like C<select ()> which have a high
674	overhead for the actual polling but can deliver many events at once.	747	overhead for the actual polling but can deliver many events at once.
675		748
676	By setting a higher I<io collect interval> you allow libev to spend more	749	By setting a higher I<io collect interval> you allow libev to spend more
677	time collecting I/O events, so you can handle more events per iteration,	750	time collecting I/O events, so you can handle more events per iteration,
…		…
679	C<ev_timer>) will be not affected. Setting this to a non-null value will	752	C<ev_timer>) will be not affected. Setting this to a non-null value will
680	introduce an additional C<ev_sleep ()> call into most loop iterations.	753	introduce an additional C<ev_sleep ()> call into most loop iterations.
681		754
682	Likewise, by setting a higher I<timeout collect interval> you allow libev	755	Likewise, by setting a higher I<timeout collect interval> you allow libev
683	to spend more time collecting timeouts, at the expense of increased	756	to spend more time collecting timeouts, at the expense of increased
684	latency (the watcher callback will be called later). C<ev_io> watchers	757	latency/jitter/inexactness (the watcher callback will be called
685	will not be affected. Setting this to a non-null value will not introduce	758	later). C<ev_io> watchers will not be affected. Setting this to a non-null
686	any overhead in libev.	759	value will not introduce any overhead in libev.
687		760
688	Many (busy) programs can usually benefit by setting the io collect	761	Many (busy) programs can usually benefit by setting the I/O collect
689	interval to a value near C<0.1> or so, which is often enough for	762	interval to a value near C<0.1> or so, which is often enough for
690	interactive servers (of course not for games), likewise for timeouts. It	763	interactive servers (of course not for games), likewise for timeouts. It
691	usually doesn't make much sense to set it to a lower value than C<0.01>,	764	usually doesn't make much sense to set it to a lower value than C<0.01>,
692	as this approsaches the timing granularity of most systems.	765	as this approaches the timing granularity of most systems.
		766
		767	Setting the I<timeout collect interval> can improve the opportunity for
		768	saving power, as the program will "bundle" timer callback invocations that
		769	are "near" in time together, by delaying some, thus reducing the number of
		770	times the process sleeps and wakes up again. Another useful technique to
		771	reduce iterations/wake-ups is to use C<ev_periodic> watchers and make sure
		772	they fire on, say, one-second boundaries only.
693		773
694	=item ev_loop_verify (loop)	774	=item ev_loop_verify (loop)
695		775
696	This function only does something when C<EV_VERIFY> support has been	776	This function only does something when C<EV_VERIFY> support has been
697	compiled in. It tries to go through all internal structures and checks	777	compiled in, which is the default for non-minimal builds. It tries to go
698	them for validity. If anything is found to be inconsistent, it will print	778	through all internal structures and checks them for validity. If anything
699	an error message to standard error and call C<abort ()>.	779	is found to be inconsistent, it will print an error message to standard
		780	error and call C<abort ()>.
700		781
701	This can be used to catch bugs inside libev itself: under normal	782	This can be used to catch bugs inside libev itself: under normal
702	circumstances, this function will never abort as of course libev keeps its	783	circumstances, this function will never abort as of course libev keeps its
703	data structures consistent.	784	data structures consistent.
704		785
705	=back	786	=back
706		787
707		788
708	=head1 ANATOMY OF A WATCHER	789	=head1 ANATOMY OF A WATCHER
709		790
		791	In the following description, uppercase C<TYPE> in names stands for the
		792	watcher type, e.g. C<ev_TYPE_start> can mean C<ev_timer_start> for timer
		793	watchers and C<ev_io_start> for I/O watchers.
		794
710	A watcher is a structure that you create and register to record your	795	A watcher is a structure that you create and register to record your
711	interest in some event. For instance, if you want to wait for STDIN to	796	interest in some event. For instance, if you want to wait for STDIN to
712	become readable, you would create an C<ev_io> watcher for that:	797	become readable, you would create an C<ev_io> watcher for that:
713		798
714	static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)	799	static void my_cb (struct ev_loop *loop, ev_io *w, int revents)
715	{	800	{
716	ev_io_stop (w);	801	ev_io_stop (w);
717	ev_unloop (loop, EVUNLOOP_ALL);	802	ev_unloop (loop, EVUNLOOP_ALL);
718	}	803	}
719		804
720	struct ev_loop *loop = ev_default_loop (0);	805	struct ev_loop *loop = ev_default_loop (0);
		806
721	struct ev_io stdin_watcher;	807	ev_io stdin_watcher;
		808
722	ev_init (&stdin_watcher, my_cb);	809	ev_init (&stdin_watcher, my_cb);
723	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);	810	ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
724	ev_io_start (loop, &stdin_watcher);	811	ev_io_start (loop, &stdin_watcher);
		812
725	ev_loop (loop, 0);	813	ev_loop (loop, 0);
726		814
727	As you can see, you are responsible for allocating the memory for your	815	As you can see, you are responsible for allocating the memory for your
728	watcher structures (and it is usually a bad idea to do this on the stack,	816	watcher structures (and it is I<usually> a bad idea to do this on the
729	although this can sometimes be quite valid).	817	stack).
		818
		819	Each watcher has an associated watcher structure (called C<struct ev_TYPE>
		820	or simply C<ev_TYPE>, as typedefs are provided for all watcher structs).
730		821
731	Each watcher structure must be initialised by a call to C<ev_init	822	Each watcher structure must be initialised by a call to C<ev_init
732	(watcher *, callback)>, which expects a callback to be provided. This	823	(watcher *, callback)>, which expects a callback to be provided. This
733	callback gets invoked each time the event occurs (or, in the case of io	824	callback gets invoked each time the event occurs (or, in the case of I/O
734	watchers, each time the event loop detects that the file descriptor given	825	watchers, each time the event loop detects that the file descriptor given
735	is readable and/or writable).	826	is readable and/or writable).
736		827
737	Each watcher type has its own C<< ev_<type>_set (watcher *, ...) >> macro	828	Each watcher type further has its own C<< ev_TYPE_set (watcher *, ...) >>
738	with arguments specific to this watcher type. There is also a macro	829	macro to configure it, with arguments specific to the watcher type. There
739	to combine initialisation and setting in one call: C<< ev_<type>_init	830	is also a macro to combine initialisation and setting in one call: C<<
740	(watcher *, callback, ...) >>.	831	ev_TYPE_init (watcher *, callback, ...) >>.
741		832
742	To make the watcher actually watch out for events, you have to start it	833	To make the watcher actually watch out for events, you have to start it
743	with a watcher-specific start function (C<< ev_<type>_start (loop, watcher	834	with a watcher-specific start function (C<< ev_TYPE_start (loop, watcher
744	*) >>), and you can stop watching for events at any time by calling the	835	*) >>), and you can stop watching for events at any time by calling the
745	corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.	836	corresponding stop function (C<< ev_TYPE_stop (loop, watcher *) >>.
746		837
747	As long as your watcher is active (has been started but not stopped) you	838	As long as your watcher is active (has been started but not stopped) you
748	must not touch the values stored in it. Most specifically you must never	839	must not touch the values stored in it. Most specifically you must never
749	reinitialise it or call its C<set> macro.	840	reinitialise it or call its C<ev_TYPE_set> macro.
750		841
751	Each and every callback receives the event loop pointer as first, the	842	Each and every callback receives the event loop pointer as first, the
752	registered watcher structure as second, and a bitset of received events as	843	registered watcher structure as second, and a bitset of received events as
753	third argument.	844	third argument.
754		845
…		…
814		905
815	The given async watcher has been asynchronously notified (see C<ev_async>).	906	The given async watcher has been asynchronously notified (see C<ev_async>).
816		907
817	=item C<EV_ERROR>	908	=item C<EV_ERROR>
818		909
819	An unspecified error has occured, the watcher has been stopped. This might	910	An unspecified error has occurred, the watcher has been stopped. This might
820	happen because the watcher could not be properly started because libev	911	happen because the watcher could not be properly started because libev
821	ran out of memory, a file descriptor was found to be closed or any other	912	ran out of memory, a file descriptor was found to be closed or any other
		913	problem. Libev considers these application bugs.
		914
822	problem. You best act on it by reporting the problem and somehow coping	915	You best act on it by reporting the problem and somehow coping with the
823	with the watcher being stopped.	916	watcher being stopped. Note that well-written programs should not receive
		917	an error ever, so when your watcher receives it, this usually indicates a
		918	bug in your program.
824		919
825	Libev will usually signal a few "dummy" events together with an error,	920	Libev will usually signal a few "dummy" events together with an error, for
826	for example it might indicate that a fd is readable or writable, and if	921	example it might indicate that a fd is readable or writable, and if your
827	your callbacks is well-written it can just attempt the operation and cope	922	callbacks is well-written it can just attempt the operation and cope with
828	with the error from read() or write(). This will not work in multithreaded	923	the error from read() or write(). This will not work in multi-threaded
829	programs, though, so beware.	924	programs, though, as the fd could already be closed and reused for another
		925	thing, so beware.
830		926
831	=back	927	=back
832		928
833	=head2 GENERIC WATCHER FUNCTIONS	929	=head2 GENERIC WATCHER FUNCTIONS
834
835	In the following description, C<TYPE> stands for the watcher type,
836	e.g. C<timer> for C<ev_timer> watchers and C<io> for C<ev_io> watchers.
837		930
838	=over 4	931	=over 4
839		932
840	=item C<ev_init> (ev_TYPE *watcher, callback)	933	=item C<ev_init> (ev_TYPE *watcher, callback)
841		934
…		…
847	which rolls both calls into one.	940	which rolls both calls into one.
848		941
849	You can reinitialise a watcher at any time as long as it has been stopped	942	You can reinitialise a watcher at any time as long as it has been stopped
850	(or never started) and there are no pending events outstanding.	943	(or never started) and there are no pending events outstanding.
851		944
852	The callback is always of type C<void (*)(ev_loop *loop, ev_TYPE *watcher,	945	The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
853	int revents)>.	946	int revents)>.
		947
		948	Example: Initialise an C<ev_io> watcher in two steps.
		949
		950	ev_io w;
		951	ev_init (&w, my_cb);
		952	ev_io_set (&w, STDIN_FILENO, EV_READ);
854		953
855	=item C<ev_TYPE_set> (ev_TYPE *, [args])	954	=item C<ev_TYPE_set> (ev_TYPE *, [args])
856		955
857	This macro initialises the type-specific parts of a watcher. You need to	956	This macro initialises the type-specific parts of a watcher. You need to
858	call C<ev_init> at least once before you call this macro, but you can	957	call C<ev_init> at least once before you call this macro, but you can
…		…
861	difference to the C<ev_init> macro).	960	difference to the C<ev_init> macro).
862		961
863	Although some watcher types do not have type-specific arguments	962	Although some watcher types do not have type-specific arguments
864	(e.g. C<ev_prepare>) you still need to call its C<set> macro.	963	(e.g. C<ev_prepare>) you still need to call its C<set> macro.
865		964
		965	See C<ev_init>, above, for an example.
		966
866	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])	967	=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
867		968
868	This convinience macro rolls both C<ev_init> and C<ev_TYPE_set> macro	969	This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
869	calls into a single call. This is the most convinient method to initialise	970	calls into a single call. This is the most convenient method to initialise
870	a watcher. The same limitations apply, of course.	971	a watcher. The same limitations apply, of course.
		972
		973	Example: Initialise and set an C<ev_io> watcher in one step.
		974
		975	ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
871		976
872	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)	977	=item C<ev_TYPE_start> (loop *, ev_TYPE *watcher)
873		978
874	Starts (activates) the given watcher. Only active watchers will receive	979	Starts (activates) the given watcher. Only active watchers will receive
875	events. If the watcher is already active nothing will happen.	980	events. If the watcher is already active nothing will happen.
876		981
		982	Example: Start the C<ev_io> watcher that is being abused as example in this
		983	whole section.
		984
		985	ev_io_start (EV_DEFAULT_UC, &w);
		986
877	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)	987	=item C<ev_TYPE_stop> (loop *, ev_TYPE *watcher)
878		988
879	Stops the given watcher again (if active) and clears the pending	989	Stops the given watcher if active, and clears the pending status (whether
		990	the watcher was active or not).
		991
880	status. It is possible that stopped watchers are pending (for example,	992	It is possible that stopped watchers are pending - for example,
881	non-repeating timers are being stopped when they become pending), but	993	non-repeating timers are being stopped when they become pending - but
882	C<ev_TYPE_stop> ensures that the watcher is neither active nor pending. If	994	calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
883	you want to free or reuse the memory used by the watcher it is therefore a	995	pending. If you want to free or reuse the memory used by the watcher it is
884	good idea to always call its C<ev_TYPE_stop> function.	996	therefore a good idea to always call its C<ev_TYPE_stop> function.
885		997
886	=item bool ev_is_active (ev_TYPE *watcher)	998	=item bool ev_is_active (ev_TYPE *watcher)
887		999
888	Returns a true value iff the watcher is active (i.e. it has been started	1000	Returns a true value iff the watcher is active (i.e. it has been started
889	and not yet been stopped). As long as a watcher is active you must not modify	1001	and not yet been stopped). As long as a watcher is active you must not modify
…		…
937		1049
938	=item ev_invoke (loop, ev_TYPE *watcher, int revents)	1050	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
939		1051
940	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither	1052	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
941	C<loop> nor C<revents> need to be valid as long as the watcher callback	1053	C<loop> nor C<revents> need to be valid as long as the watcher callback
942	can deal with that fact.	1054	can deal with that fact, as both are simply passed through to the
		1055	callback.
943		1056
944	=item int ev_clear_pending (loop, ev_TYPE *watcher)	1057	=item int ev_clear_pending (loop, ev_TYPE *watcher)
945		1058
946	If the watcher is pending, this function returns clears its pending status	1059	If the watcher is pending, this function clears its pending status and
947	and returns its C<revents> bitset (as if its callback was invoked). If the	1060	returns its C<revents> bitset (as if its callback was invoked). If the
948	watcher isn't pending it does nothing and returns C<0>.	1061	watcher isn't pending it does nothing and returns C<0>.
949		1062
		1063	Sometimes it can be useful to "poll" a watcher instead of waiting for its
		1064	callback to be invoked, which can be accomplished with this function.
		1065
950	=back	1066	=back
951		1067
952		1068
953	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1069	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
954		1070
955	Each watcher has, by default, a member C<void *data> that you can change	1071	Each watcher has, by default, a member C<void *data> that you can change
956	and read at any time, libev will completely ignore it. This can be used	1072	and read at any time: libev will completely ignore it. This can be used
957	to associate arbitrary data with your watcher. If you need more data and	1073	to associate arbitrary data with your watcher. If you need more data and
958	don't want to allocate memory and store a pointer to it in that data	1074	don't want to allocate memory and store a pointer to it in that data
959	member, you can also "subclass" the watcher type and provide your own	1075	member, you can also "subclass" the watcher type and provide your own
960	data:	1076	data:
961		1077
962	struct my_io	1078	struct my_io
963	{	1079	{
964	struct ev_io io;	1080	ev_io io;
965	int otherfd;	1081	int otherfd;
966	void *somedata;	1082	void *somedata;
967	struct whatever *mostinteresting;	1083	struct whatever *mostinteresting;
968	}	1084	};
		1085
		1086	...
		1087	struct my_io w;
		1088	ev_io_init (&w.io, my_cb, fd, EV_READ);
969		1089
970	And since your callback will be called with a pointer to the watcher, you	1090	And since your callback will be called with a pointer to the watcher, you
971	can cast it back to your own type:	1091	can cast it back to your own type:
972		1092
973	static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)	1093	static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
974	{	1094	{
975	struct my_io *w = (struct my_io *)w_;	1095	struct my_io *w = (struct my_io *)w_;
976	...	1096	...
977	}	1097	}
978		1098
979	More interesting and less C-conformant ways of casting your callback type	1099	More interesting and less C-conformant ways of casting your callback type
980	instead have been omitted.	1100	instead have been omitted.
981		1101
982	Another common scenario is having some data structure with multiple	1102	Another common scenario is to use some data structure with multiple
983	watchers:	1103	embedded watchers:
984		1104
985	struct my_biggy	1105	struct my_biggy
986	{	1106	{
987	int some_data;	1107	int some_data;
988	ev_timer t1;	1108	ev_timer t1;
989	ev_timer t2;	1109	ev_timer t2;
990	}	1110	}
991		1111
992	In this case getting the pointer to C<my_biggy> is a bit more complicated,	1112	In this case getting the pointer to C<my_biggy> is a bit more
993	you need to use C<offsetof>:	1113	complicated: Either you store the address of your C<my_biggy> struct
		1114	in the C<data> member of the watcher (for woozies), or you need to use
		1115	some pointer arithmetic using C<offsetof> inside your watchers (for real
		1116	programmers):
994		1117
995	#include <stddef.h>	1118	#include <stddef.h>
996		1119
997	static void	1120	static void
998	t1_cb (EV_P_ struct ev_timer *w, int revents)	1121	t1_cb (EV_P_ ev_timer *w, int revents)
999	{	1122	{
1000	struct my_biggy big = (struct my_biggy *	1123	struct my_biggy big = (struct my_biggy *
1001	(((char *)w) - offsetof (struct my_biggy, t1));	1124	(((char *)w) - offsetof (struct my_biggy, t1));
1002	}	1125	}
1003		1126
1004	static void	1127	static void
1005	t2_cb (EV_P_ struct ev_timer *w, int revents)	1128	t2_cb (EV_P_ ev_timer *w, int revents)
1006	{	1129	{
1007	struct my_biggy big = (struct my_biggy *	1130	struct my_biggy big = (struct my_biggy *
1008	(((char *)w) - offsetof (struct my_biggy, t2));	1131	(((char *)w) - offsetof (struct my_biggy, t2));
1009	}	1132	}
1010		1133
1011		1134
1012	=head1 WATCHER TYPES	1135	=head1 WATCHER TYPES
1013		1136
1014	This section describes each watcher in detail, but will not repeat	1137	This section describes each watcher in detail, but will not repeat
…		…
1038	In general you can register as many read and/or write event watchers per	1161	In general you can register as many read and/or write event watchers per
1039	fd as you want (as long as you don't confuse yourself). Setting all file	1162	fd as you want (as long as you don't confuse yourself). Setting all file
1040	descriptors to non-blocking mode is also usually a good idea (but not	1163	descriptors to non-blocking mode is also usually a good idea (but not
1041	required if you know what you are doing).	1164	required if you know what you are doing).
1042		1165
1043	If you must do this, then force the use of a known-to-be-good backend	1166	If you cannot use non-blocking mode, then force the use of a
1044	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	1167	known-to-be-good backend (at the time of this writing, this includes only
1045	C<EVBACKEND_POLL>).	1168	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>).
1046		1169
1047	Another thing you have to watch out for is that it is quite easy to	1170	Another thing you have to watch out for is that it is quite easy to
1048	receive "spurious" readiness notifications, that is your callback might	1171	receive "spurious" readiness notifications, that is your callback might
1049	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1172	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1050	because there is no data. Not only are some backends known to create a	1173	because there is no data. Not only are some backends known to create a
1051	lot of those (for example solaris ports), it is very easy to get into	1174	lot of those (for example Solaris ports), it is very easy to get into
1052	this situation even with a relatively standard program structure. Thus	1175	this situation even with a relatively standard program structure. Thus
1053	it is best to always use non-blocking I/O: An extra C<read>(2) returning	1176	it is best to always use non-blocking I/O: An extra C<read>(2) returning
1054	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1177	C<EAGAIN> is far preferable to a program hanging until some data arrives.
1055		1178
1056	If you cannot run the fd in non-blocking mode (for example you should not	1179	If you cannot run the fd in non-blocking mode (for example you should
1057	play around with an Xlib connection), then you have to seperately re-test	1180	not play around with an Xlib connection), then you have to separately
1058	whether a file descriptor is really ready with a known-to-be good interface	1181	re-test whether a file descriptor is really ready with a known-to-be good
1059	such as poll (fortunately in our Xlib example, Xlib already does this on	1182	interface such as poll (fortunately in our Xlib example, Xlib already
1060	its own, so its quite safe to use).	1183	does this on its own, so its quite safe to use). Some people additionally
		1184	use C<SIGALRM> and an interval timer, just to be sure you won't block
		1185	indefinitely.
		1186
		1187	But really, best use non-blocking mode.
1061		1188
1062	=head3 The special problem of disappearing file descriptors	1189	=head3 The special problem of disappearing file descriptors
1063		1190
1064	Some backends (e.g. kqueue, epoll) need to be told about closing a file	1191	Some backends (e.g. kqueue, epoll) need to be told about closing a file
1065	descriptor (either by calling C<close> explicitly or by any other means,	1192	descriptor (either due to calling C<close> explicitly or any other means,
1066	such as C<dup>). The reason is that you register interest in some file	1193	such as C<dup2>). The reason is that you register interest in some file
1067	descriptor, but when it goes away, the operating system will silently drop	1194	descriptor, but when it goes away, the operating system will silently drop
1068	this interest. If another file descriptor with the same number then is	1195	this interest. If another file descriptor with the same number then is
1069	registered with libev, there is no efficient way to see that this is, in	1196	registered with libev, there is no efficient way to see that this is, in
1070	fact, a different file descriptor.	1197	fact, a different file descriptor.
1071		1198
…		…
1102	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or	1229	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
1103	C<EVBACKEND_POLL>.	1230	C<EVBACKEND_POLL>.
1104		1231
1105	=head3 The special problem of SIGPIPE	1232	=head3 The special problem of SIGPIPE
1106		1233
1107	While not really specific to libev, it is easy to forget about SIGPIPE:	1234	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1108	when reading from a pipe whose other end has been closed, your program	1235	when writing to a pipe whose other end has been closed, your program gets
1109	gets send a SIGPIPE, which, by default, aborts your program. For most	1236	sent a SIGPIPE, which, by default, aborts your program. For most programs
1110	programs this is sensible behaviour, for daemons, this is usually	1237	this is sensible behaviour, for daemons, this is usually undesirable.
1111	undesirable.
1112		1238
1113	So when you encounter spurious, unexplained daemon exits, make sure you	1239	So when you encounter spurious, unexplained daemon exits, make sure you
1114	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon	1240	ignore SIGPIPE (and maybe make sure you log the exit status of your daemon
1115	somewhere, as that would have given you a big clue).	1241	somewhere, as that would have given you a big clue).
1116		1242
…		…
1122	=item ev_io_init (ev_io *, callback, int fd, int events)	1248	=item ev_io_init (ev_io *, callback, int fd, int events)
1123		1249
1124	=item ev_io_set (ev_io *, int fd, int events)	1250	=item ev_io_set (ev_io *, int fd, int events)
1125		1251
1126	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to	1252	Configures an C<ev_io> watcher. The C<fd> is the file descriptor to
1127	rceeive events for and events is either C<EV_READ>, C<EV_WRITE> or	1253	receive events for and C<events> is either C<EV_READ>, C<EV_WRITE> or
1128	C<EV_READ | EV_WRITE> to receive the given events.	1254	C<EV_READ | EV_WRITE>, to express the desire to receive the given events.
1129		1255
1130	=item int fd [read-only]	1256	=item int fd [read-only]
1131		1257
1132	The file descriptor being watched.	1258	The file descriptor being watched.
1133		1259
…		…
1141		1267
1142	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1268	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1143	readable, but only once. Since it is likely line-buffered, you could	1269	readable, but only once. Since it is likely line-buffered, you could
1144	attempt to read a whole line in the callback.	1270	attempt to read a whole line in the callback.
1145		1271
1146	static void	1272	static void
1147	stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1273	stdin_readable_cb (struct ev_loop *loop, ev_io *w, int revents)
1148	{	1274	{
1149	ev_io_stop (loop, w);	1275	ev_io_stop (loop, w);
1150	.. read from stdin here (or from w->fd) and haqndle any I/O errors	1276	.. read from stdin here (or from w->fd) and handle any I/O errors
1151	}	1277	}
1152		1278
1153	...	1279	...
1154	struct ev_loop *loop = ev_default_init (0);	1280	struct ev_loop *loop = ev_default_init (0);
1155	struct ev_io stdin_readable;	1281	ev_io stdin_readable;
1156	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);	1282	ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1157	ev_io_start (loop, &stdin_readable);	1283	ev_io_start (loop, &stdin_readable);
1158	ev_loop (loop, 0);	1284	ev_loop (loop, 0);
1159		1285
1160		1286
1161	=head2 C<ev_timer> - relative and optionally repeating timeouts	1287	=head2 C<ev_timer> - relative and optionally repeating timeouts
1162		1288
1163	Timer watchers are simple relative timers that generate an event after a	1289	Timer watchers are simple relative timers that generate an event after a
1164	given time, and optionally repeating in regular intervals after that.	1290	given time, and optionally repeating in regular intervals after that.
1165		1291
1166	The timers are based on real time, that is, if you register an event that	1292	The timers are based on real time, that is, if you register an event that
1167	times out after an hour and you reset your system clock to january last	1293	times out after an hour and you reset your system clock to January last
1168	year, it will still time out after (roughly) and hour. "Roughly" because	1294	year, it will still time out after (roughly) one hour. "Roughly" because
1169	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1295	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1170	monotonic clock option helps a lot here).	1296	monotonic clock option helps a lot here).
		1297
		1298	The callback is guaranteed to be invoked only I<after> its timeout has
		1299	passed, but if multiple timers become ready during the same loop iteration
		1300	then order of execution is undefined.
		1301
		1302	=head3 Be smart about timeouts
		1303
		1304	Many real-world problems involve some kind of timeout, usually for error
		1305	recovery. A typical example is an HTTP request - if the other side hangs,
		1306	you want to raise some error after a while.
		1307
		1308	What follows are some ways to handle this problem, from obvious and
		1309	inefficient to smart and efficient.
		1310
		1311	In the following, a 60 second activity timeout is assumed - a timeout that
		1312	gets reset to 60 seconds each time there is activity (e.g. each time some
		1313	data or other life sign was received).
		1314
		1315	=over 4
		1316
		1317	=item 1. Use a timer and stop, reinitialise and start it on activity.
		1318
		1319	This is the most obvious, but not the most simple way: In the beginning,
		1320	start the watcher:
		1321
		1322	ev_timer_init (timer, callback, 60., 0.);
		1323	ev_timer_start (loop, timer);
		1324
		1325	Then, each time there is some activity, C<ev_timer_stop> it, initialise it
		1326	and start it again:
		1327
		1328	ev_timer_stop (loop, timer);
		1329	ev_timer_set (timer, 60., 0.);
		1330	ev_timer_start (loop, timer);
		1331
		1332	This is relatively simple to implement, but means that each time there is
		1333	some activity, libev will first have to remove the timer from its internal
		1334	data structure and then add it again. Libev tries to be fast, but it's
		1335	still not a constant-time operation.
		1336
		1337	=item 2. Use a timer and re-start it with C<ev_timer_again> inactivity.
		1338
		1339	This is the easiest way, and involves using C<ev_timer_again> instead of
		1340	C<ev_timer_start>.
		1341
		1342	To implement this, configure an C<ev_timer> with a C<repeat> value
		1343	of C<60> and then call C<ev_timer_again> at start and each time you
		1344	successfully read or write some data. If you go into an idle state where
		1345	you do not expect data to travel on the socket, you can C<ev_timer_stop>
		1346	the timer, and C<ev_timer_again> will automatically restart it if need be.
		1347
		1348	That means you can ignore both the C<ev_timer_start> function and the
		1349	C<after> argument to C<ev_timer_set>, and only ever use the C<repeat>
		1350	member and C<ev_timer_again>.
		1351
		1352	At start:
		1353
		1354	ev_timer_init (timer, callback);
		1355	timer->repeat = 60.;
		1356	ev_timer_again (loop, timer);
		1357
		1358	Each time there is some activity:
		1359
		1360	ev_timer_again (loop, timer);
		1361
		1362	It is even possible to change the time-out on the fly, regardless of
		1363	whether the watcher is active or not:
		1364
		1365	timer->repeat = 30.;
		1366	ev_timer_again (loop, timer);
		1367
		1368	This is slightly more efficient then stopping/starting the timer each time
		1369	you want to modify its timeout value, as libev does not have to completely
		1370	remove and re-insert the timer from/into its internal data structure.
		1371
		1372	It is, however, even simpler than the "obvious" way to do it.
		1373
		1374	=item 3. Let the timer time out, but then re-arm it as required.
		1375
		1376	This method is more tricky, but usually most efficient: Most timeouts are
		1377	relatively long compared to the intervals between other activity - in
		1378	our example, within 60 seconds, there are usually many I/O events with
		1379	associated activity resets.
		1380
		1381	In this case, it would be more efficient to leave the C<ev_timer> alone,
		1382	but remember the time of last activity, and check for a real timeout only
		1383	within the callback:
		1384
		1385	ev_tstamp last_activity; // time of last activity
		1386
		1387	static void
		1388	callback (EV_P_ ev_timer *w, int revents)
		1389	{
		1390	ev_tstamp now = ev_now (EV_A);
		1391	ev_tstamp timeout = last_activity + 60.;
		1392
		1393	// if last_activity + 60. is older than now, we did time out
		1394	if (timeout < now)
		1395	{
		1396	// timeout occured, take action
		1397	}
		1398	else
		1399	{
		1400	// callback was invoked, but there was some activity, re-arm
		1401	// the watcher to fire in last_activity + 60, which is
		1402	// guaranteed to be in the future, so "again" is positive:
		1403	w->again = timeout - now;
		1404	ev_timer_again (EV_A_ w);
		1405	}
		1406	}
		1407
		1408	To summarise the callback: first calculate the real timeout (defined
		1409	as "60 seconds after the last activity"), then check if that time has
		1410	been reached, which means something I<did>, in fact, time out. Otherwise
		1411	the callback was invoked too early (C<timeout> is in the future), so
		1412	re-schedule the timer to fire at that future time, to see if maybe we have
		1413	a timeout then.
		1414
		1415	Note how C<ev_timer_again> is used, taking advantage of the
		1416	C<ev_timer_again> optimisation when the timer is already running.
		1417
		1418	This scheme causes more callback invocations (about one every 60 seconds
		1419	minus half the average time between activity), but virtually no calls to
		1420	libev to change the timeout.
		1421
		1422	To start the timer, simply initialise the watcher and set C<last_activity>
		1423	to the current time (meaning we just have some activity :), then call the
		1424	callback, which will "do the right thing" and start the timer:
		1425
		1426	ev_timer_init (timer, callback);
		1427	last_activity = ev_now (loop);
		1428	callback (loop, timer, EV_TIMEOUT);
		1429
		1430	And when there is some activity, simply store the current time in
		1431	C<last_activity>, no libev calls at all:
		1432
		1433	last_actiivty = ev_now (loop);
		1434
		1435	This technique is slightly more complex, but in most cases where the
		1436	time-out is unlikely to be triggered, much more efficient.
		1437
		1438	Changing the timeout is trivial as well (if it isn't hard-coded in the
		1439	callback :) - just change the timeout and invoke the callback, which will
		1440	fix things for you.
		1441
		1442	=item 4. Wee, just use a double-linked list for your timeouts.
		1443
		1444	If there is not one request, but many thousands (millions...), all
		1445	employing some kind of timeout with the same timeout value, then one can
		1446	do even better:
		1447
		1448	When starting the timeout, calculate the timeout value and put the timeout
		1449	at the I<end> of the list.
		1450
		1451	Then use an C<ev_timer> to fire when the timeout at the I<beginning> of
		1452	the list is expected to fire (for example, using the technique #3).
		1453
		1454	When there is some activity, remove the timer from the list, recalculate
		1455	the timeout, append it to the end of the list again, and make sure to
		1456	update the C<ev_timer> if it was taken from the beginning of the list.
		1457
		1458	This way, one can manage an unlimited number of timeouts in O(1) time for
		1459	starting, stopping and updating the timers, at the expense of a major
		1460	complication, and having to use a constant timeout. The constant timeout
		1461	ensures that the list stays sorted.
		1462
		1463	=back
		1464
		1465	So which method the best?
		1466
		1467	Method #2 is a simple no-brain-required solution that is adequate in most
		1468	situations. Method #3 requires a bit more thinking, but handles many cases
		1469	better, and isn't very complicated either. In most case, choosing either
		1470	one is fine, with #3 being better in typical situations.
		1471
		1472	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
		1473	rather complicated, but extremely efficient, something that really pays
		1474	off after the first million or so of active timers, i.e. it's usually
		1475	overkill :)
		1476
		1477	=head3 The special problem of time updates
		1478
		1479	Establishing the current time is a costly operation (it usually takes at
		1480	least two system calls): EV therefore updates its idea of the current
		1481	time only before and after C<ev_loop> collects new events, which causes a
		1482	growing difference between C<ev_now ()> and C<ev_time ()> when handling
		1483	lots of events in one iteration.
1171		1484
1172	The relative timeouts are calculated relative to the C<ev_now ()>	1485	The relative timeouts are calculated relative to the C<ev_now ()>
1173	time. This is usually the right thing as this timestamp refers to the time	1486	time. This is usually the right thing as this timestamp refers to the time
1174	of the event triggering whatever timeout you are modifying/starting. If	1487	of the event triggering whatever timeout you are modifying/starting. If
1175	you suspect event processing to be delayed and you I<need> to base the timeout	1488	you suspect event processing to be delayed and you I<need> to base the
1176	on the current time, use something like this to adjust for this:	1489	timeout on the current time, use something like this to adjust for this:
1177		1490
1178	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1491	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1179		1492
1180	The callback is guarenteed to be invoked only after its timeout has passed,	1493	If the event loop is suspended for a long time, you can also force an
1181	but if multiple timers become ready during the same loop iteration then	1494	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1182	order of execution is undefined.	1495	()>.
1183		1496
1184	=head3 Watcher-Specific Functions and Data Members	1497	=head3 Watcher-Specific Functions and Data Members
1185		1498
1186	=over 4	1499	=over 4
1187		1500
…		…
1206	This will act as if the timer timed out and restart it again if it is	1519	This will act as if the timer timed out and restart it again if it is
1207	repeating. The exact semantics are:	1520	repeating. The exact semantics are:
1208		1521
1209	If the timer is pending, its pending status is cleared.	1522	If the timer is pending, its pending status is cleared.
1210		1523
1211	If the timer is started but nonrepeating, stop it (as if it timed out).	1524	If the timer is started but non-repeating, stop it (as if it timed out).
1212		1525
1213	If the timer is repeating, either start it if necessary (with the	1526	If the timer is repeating, either start it if necessary (with the
1214	C<repeat> value), or reset the running timer to the C<repeat> value.	1527	C<repeat> value), or reset the running timer to the C<repeat> value.
1215		1528
1216	This sounds a bit complicated, but here is a useful and typical	1529	This sounds a bit complicated, see "Be smart about timeouts", above, for a
1217	example: Imagine you have a tcp connection and you want a so-called idle	1530	usage example.
1218	timeout, that is, you want to be called when there have been, say, 60
1219	seconds of inactivity on the socket. The easiest way to do this is to
1220	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
1221	C<ev_timer_again> each time you successfully read or write some data. If
1222	you go into an idle state where you do not expect data to travel on the
1223	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
1224	automatically restart it if need be.
1225
1226	That means you can ignore the C<after> value and C<ev_timer_start>
1227	altogether and only ever use the C<repeat> value and C<ev_timer_again>:
1228
1229	ev_timer_init (timer, callback, 0., 5.);
1230	ev_timer_again (loop, timer);
1231	...
1232	timer->again = 17.;
1233	ev_timer_again (loop, timer);
1234	...
1235	timer->again = 10.;
1236	ev_timer_again (loop, timer);
1237
1238	This is more slightly efficient then stopping/starting the timer each time
1239	you want to modify its timeout value.
1240		1531
1241	=item ev_tstamp repeat [read-write]	1532	=item ev_tstamp repeat [read-write]
1242		1533
1243	The current C<repeat> value. Will be used each time the watcher times out	1534	The current C<repeat> value. Will be used each time the watcher times out
1244	or C<ev_timer_again> is called and determines the next timeout (if any),	1535	or C<ev_timer_again> is called, and determines the next timeout (if any),
1245	which is also when any modifications are taken into account.	1536	which is also when any modifications are taken into account.
1246		1537
1247	=back	1538	=back
1248		1539
1249	=head3 Examples	1540	=head3 Examples
1250		1541
1251	Example: Create a timer that fires after 60 seconds.	1542	Example: Create a timer that fires after 60 seconds.
1252		1543
1253	static void	1544	static void
1254	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1545	one_minute_cb (struct ev_loop *loop, ev_timer *w, int revents)
1255	{	1546	{
1256	.. one minute over, w is actually stopped right here	1547	.. one minute over, w is actually stopped right here
1257	}	1548	}
1258		1549
1259	struct ev_timer mytimer;	1550	ev_timer mytimer;
1260	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);	1551	ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1261	ev_timer_start (loop, &mytimer);	1552	ev_timer_start (loop, &mytimer);
1262		1553
1263	Example: Create a timeout timer that times out after 10 seconds of	1554	Example: Create a timeout timer that times out after 10 seconds of
1264	inactivity.	1555	inactivity.
1265		1556
1266	static void	1557	static void
1267	timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1558	timeout_cb (struct ev_loop *loop, ev_timer *w, int revents)
1268	{	1559	{
1269	.. ten seconds without any activity	1560	.. ten seconds without any activity
1270	}	1561	}
1271		1562
1272	struct ev_timer mytimer;	1563	ev_timer mytimer;
1273	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */	1564	ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1274	ev_timer_again (&mytimer); /* start timer */	1565	ev_timer_again (&mytimer); /* start timer */
1275	ev_loop (loop, 0);	1566	ev_loop (loop, 0);
1276		1567
1277	// and in some piece of code that gets executed on any "activity":	1568	// and in some piece of code that gets executed on any "activity":
1278	// reset the timeout to start ticking again at 10 seconds	1569	// reset the timeout to start ticking again at 10 seconds
1279	ev_timer_again (&mytimer);	1570	ev_timer_again (&mytimer);
1280		1571
1281		1572
1282	=head2 C<ev_periodic> - to cron or not to cron?	1573	=head2 C<ev_periodic> - to cron or not to cron?
1283		1574
1284	Periodic watchers are also timers of a kind, but they are very versatile	1575	Periodic watchers are also timers of a kind, but they are very versatile
1285	(and unfortunately a bit complex).	1576	(and unfortunately a bit complex).
1286		1577
1287	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1578	Unlike C<ev_timer>'s, they are not based on real time (or relative time)
1288	but on wallclock time (absolute time). You can tell a periodic watcher	1579	but on wall clock time (absolute time). You can tell a periodic watcher
1289	to trigger after some specific point in time. For example, if you tell a	1580	to trigger after some specific point in time. For example, if you tell a
1290	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	1581	periodic watcher to trigger in 10 seconds (by specifying e.g. C<ev_now ()
1291	+ 10.>, that is, an absolute time not a delay) and then reset your system	1582	+ 10.>, that is, an absolute time not a delay) and then reset your system
1292	clock to january of the previous year, then it will take more than year	1583	clock to January of the previous year, then it will take more than year
1293	to trigger the event (unlike an C<ev_timer>, which would still trigger	1584	to trigger the event (unlike an C<ev_timer>, which would still trigger
1294	roughly 10 seconds later as it uses a relative timeout).	1585	roughly 10 seconds later as it uses a relative timeout).
1295		1586
1296	C<ev_periodic>s can also be used to implement vastly more complex timers,	1587	C<ev_periodic>s can also be used to implement vastly more complex timers,
1297	such as triggering an event on each "midnight, local time", or other	1588	such as triggering an event on each "midnight, local time", or other
1298	complicated, rules.	1589	complicated rules.
1299		1590
1300	As with timers, the callback is guarenteed to be invoked only when the	1591	As with timers, the callback is guaranteed to be invoked only when the
1301	time (C<at>) has passed, but if multiple periodic timers become ready	1592	time (C<at>) has passed, but if multiple periodic timers become ready
1302	during the same loop iteration then order of execution is undefined.	1593	during the same loop iteration, then order of execution is undefined.
1303		1594
1304	=head3 Watcher-Specific Functions and Data Members	1595	=head3 Watcher-Specific Functions and Data Members
1305		1596
1306	=over 4	1597	=over 4
1307		1598
1308	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1599	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
1309		1600
1310	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1601	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
1311		1602
1312	Lots of arguments, lets sort it out... There are basically three modes of	1603	Lots of arguments, lets sort it out... There are basically three modes of
1313	operation, and we will explain them from simplest to complex:	1604	operation, and we will explain them from simplest to most complex:
1314		1605
1315	=over 4	1606	=over 4
1316		1607
1317	=item * absolute timer (at = time, interval = reschedule_cb = 0)	1608	=item * absolute timer (at = time, interval = reschedule_cb = 0)
1318		1609
1319	In this configuration the watcher triggers an event after the wallclock	1610	In this configuration the watcher triggers an event after the wall clock
1320	time C<at> has passed and doesn't repeat. It will not adjust when a time	1611	time C<at> has passed. It will not repeat and will not adjust when a time
1321	jump occurs, that is, if it is to be run at January 1st 2011 then it will	1612	jump occurs, that is, if it is to be run at January 1st 2011 then it will
1322	run when the system time reaches or surpasses this time.	1613	only run when the system clock reaches or surpasses this time.
1323		1614
1324	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	1615	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1325		1616
1326	In this mode the watcher will always be scheduled to time out at the next	1617	In this mode the watcher will always be scheduled to time out at the next
1327	C<at + N * interval> time (for some integer N, which can also be negative)	1618	C<at + N * interval> time (for some integer N, which can also be negative)
1328	and then repeat, regardless of any time jumps.	1619	and then repeat, regardless of any time jumps.
1329		1620
1330	This can be used to create timers that do not drift with respect to system	1621	This can be used to create timers that do not drift with respect to the
1331	time, for example, here is a C<ev_periodic> that triggers each hour, on	1622	system clock, for example, here is a C<ev_periodic> that triggers each
1332	the hour:	1623	hour, on the hour:
1333		1624
1334	ev_periodic_set (&periodic, 0., 3600., 0);	1625	ev_periodic_set (&periodic, 0., 3600., 0);
1335		1626
1336	This doesn't mean there will always be 3600 seconds in between triggers,	1627	This doesn't mean there will always be 3600 seconds in between triggers,
1337	but only that the the callback will be called when the system time shows a	1628	but only that the callback will be called when the system time shows a
1338	full hour (UTC), or more correctly, when the system time is evenly divisible	1629	full hour (UTC), or more correctly, when the system time is evenly divisible
1339	by 3600.	1630	by 3600.
1340		1631
1341	Another way to think about it (for the mathematically inclined) is that	1632	Another way to think about it (for the mathematically inclined) is that
1342	C<ev_periodic> will try to run the callback in this mode at the next possible	1633	C<ev_periodic> will try to run the callback in this mode at the next possible
…		…
1344		1635
1345	For numerical stability it is preferable that the C<at> value is near	1636	For numerical stability it is preferable that the C<at> value is near
1346	C<ev_now ()> (the current time), but there is no range requirement for	1637	C<ev_now ()> (the current time), but there is no range requirement for
1347	this value, and in fact is often specified as zero.	1638	this value, and in fact is often specified as zero.
1348		1639
1349	Note also that there is an upper limit to how often a timer can fire (cpu	1640	Note also that there is an upper limit to how often a timer can fire (CPU
1350	speed for example), so if C<interval> is very small then timing stability	1641	speed for example), so if C<interval> is very small then timing stability
1351	will of course detoriate. Libev itself tries to be exact to be about one	1642	will of course deteriorate. Libev itself tries to be exact to be about one
1352	millisecond (if the OS supports it and the machine is fast enough).	1643	millisecond (if the OS supports it and the machine is fast enough).
1353		1644
1354	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)	1645	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1355		1646
1356	In this mode the values for C<interval> and C<at> are both being	1647	In this mode the values for C<interval> and C<at> are both being
…		…
1363		1654
1364	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop	1655	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
1365	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the	1656	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
1366	only event loop modification you are allowed to do).	1657	only event loop modification you are allowed to do).
1367		1658
1368	The callback prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic	1659	The callback prototype is C<ev_tstamp (*reschedule_cb)(ev_periodic
1369	*w, ev_tstamp now)>, e.g.:	1660	*w, ev_tstamp now)>, e.g.:
1370		1661
		1662	static ev_tstamp
1371	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1663	my_rescheduler (ev_periodic *w, ev_tstamp now)
1372	{	1664	{
1373	return now + 60.;	1665	return now + 60.;
1374	}	1666	}
1375		1667
1376	It must return the next time to trigger, based on the passed time value	1668	It must return the next time to trigger, based on the passed time value
…		…
1413		1705
1414	The current interval value. Can be modified any time, but changes only	1706	The current interval value. Can be modified any time, but changes only
1415	take effect when the periodic timer fires or C<ev_periodic_again> is being	1707	take effect when the periodic timer fires or C<ev_periodic_again> is being
1416	called.	1708	called.
1417		1709
1418	=item ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]	1710	=item ev_tstamp (*reschedule_cb)(ev_periodic *w, ev_tstamp now) [read-write]
1419		1711
1420	The current reschedule callback, or C<0>, if this functionality is	1712	The current reschedule callback, or C<0>, if this functionality is
1421	switched off. Can be changed any time, but changes only take effect when	1713	switched off. Can be changed any time, but changes only take effect when
1422	the periodic timer fires or C<ev_periodic_again> is being called.	1714	the periodic timer fires or C<ev_periodic_again> is being called.
1423		1715
1424	=back	1716	=back
1425		1717
1426	=head3 Examples	1718	=head3 Examples
1427		1719
1428	Example: Call a callback every hour, or, more precisely, whenever the	1720	Example: Call a callback every hour, or, more precisely, whenever the
1429	system clock is divisible by 3600. The callback invocation times have	1721	system time is divisible by 3600. The callback invocation times have
1430	potentially a lot of jittering, but good long-term stability.	1722	potentially a lot of jitter, but good long-term stability.
1431		1723
1432	static void	1724	static void
1433	clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)	1725	clock_cb (struct ev_loop *loop, ev_io *w, int revents)
1434	{	1726	{
1435	... its now a full hour (UTC, or TAI or whatever your clock follows)	1727	... its now a full hour (UTC, or TAI or whatever your clock follows)
1436	}	1728	}
1437		1729
1438	struct ev_periodic hourly_tick;	1730	ev_periodic hourly_tick;
1439	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);	1731	ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1440	ev_periodic_start (loop, &hourly_tick);	1732	ev_periodic_start (loop, &hourly_tick);
1441		1733
1442	Example: The same as above, but use a reschedule callback to do it:	1734	Example: The same as above, but use a reschedule callback to do it:
1443		1735
1444	#include <math.h>	1736	#include <math.h>
1445		1737
1446	static ev_tstamp	1738	static ev_tstamp
1447	my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)	1739	my_scheduler_cb (ev_periodic *w, ev_tstamp now)
1448	{	1740	{
1449	return fmod (now, 3600.) + 3600.;	1741	return now + (3600. - fmod (now, 3600.));
1450	}	1742	}
1451		1743
1452	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);	1744	ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1453		1745
1454	Example: Call a callback every hour, starting now:	1746	Example: Call a callback every hour, starting now:
1455		1747
1456	struct ev_periodic hourly_tick;	1748	ev_periodic hourly_tick;
1457	ev_periodic_init (&hourly_tick, clock_cb,	1749	ev_periodic_init (&hourly_tick, clock_cb,
1458	fmod (ev_now (loop), 3600.), 3600., 0);	1750	fmod (ev_now (loop), 3600.), 3600., 0);
1459	ev_periodic_start (loop, &hourly_tick);	1751	ev_periodic_start (loop, &hourly_tick);
1460		1752
1461		1753
1462	=head2 C<ev_signal> - signal me when a signal gets signalled!	1754	=head2 C<ev_signal> - signal me when a signal gets signalled!
1463		1755
1464	Signal watchers will trigger an event when the process receives a specific	1756	Signal watchers will trigger an event when the process receives a specific
1465	signal one or more times. Even though signals are very asynchronous, libev	1757	signal one or more times. Even though signals are very asynchronous, libev
1466	will try it's best to deliver signals synchronously, i.e. as part of the	1758	will try it's best to deliver signals synchronously, i.e. as part of the
1467	normal event processing, like any other event.	1759	normal event processing, like any other event.
1468		1760
		1761	If you want signals asynchronously, just use C<sigaction> as you would
		1762	do without libev and forget about sharing the signal. You can even use
		1763	C<ev_async> from a signal handler to synchronously wake up an event loop.
		1764
1469	You can configure as many watchers as you like per signal. Only when the	1765	You can configure as many watchers as you like per signal. Only when the
1470	first watcher gets started will libev actually register a signal watcher	1766	first watcher gets started will libev actually register a signal handler
1471	with the kernel (thus it coexists with your own signal handlers as long	1767	with the kernel (thus it coexists with your own signal handlers as long as
1472	as you don't register any with libev). Similarly, when the last signal	1768	you don't register any with libev for the same signal). Similarly, when
1473	watcher for a signal is stopped libev will reset the signal handler to	1769	the last signal watcher for a signal is stopped, libev will reset the
1474	SIG_DFL (regardless of what it was set to before).	1770	signal handler to SIG_DFL (regardless of what it was set to before).
1475		1771
1476	If possible and supported, libev will install its handlers with	1772	If possible and supported, libev will install its handlers with
1477	C<SA_RESTART> behaviour enabled, so syscalls should not be unduly	1773	C<SA_RESTART> behaviour enabled, so system calls should not be unduly
1478	interrupted. If you have a problem with syscalls getting interrupted by	1774	interrupted. If you have a problem with system calls getting interrupted by
1479	signals you can block all signals in an C<ev_check> watcher and unblock	1775	signals you can block all signals in an C<ev_check> watcher and unblock
1480	them in an C<ev_prepare> watcher.	1776	them in an C<ev_prepare> watcher.
1481		1777
1482	=head3 Watcher-Specific Functions and Data Members	1778	=head3 Watcher-Specific Functions and Data Members
1483		1779
…		…
1496		1792
1497	=back	1793	=back
1498		1794
1499	=head3 Examples	1795	=head3 Examples
1500		1796
1501	Example: Try to exit cleanly on SIGINT and SIGTERM.	1797	Example: Try to exit cleanly on SIGINT.
1502		1798
1503	static void	1799	static void
1504	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	1800	sigint_cb (struct ev_loop *loop, ev_signal *w, int revents)
1505	{	1801	{
1506	ev_unloop (loop, EVUNLOOP_ALL);	1802	ev_unloop (loop, EVUNLOOP_ALL);
1507	}	1803	}
1508		1804
1509	struct ev_signal signal_watcher;	1805	ev_signal signal_watcher;
1510	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	1806	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1511	ev_signal_start (loop, &sigint_cb);	1807	ev_signal_start (loop, &signal_watcher);
1512		1808
1513		1809
1514	=head2 C<ev_child> - watch out for process status changes	1810	=head2 C<ev_child> - watch out for process status changes
1515		1811
1516	Child watchers trigger when your process receives a SIGCHLD in response to	1812	Child watchers trigger when your process receives a SIGCHLD in response to
1517	some child status changes (most typically when a child of yours dies). It	1813	some child status changes (most typically when a child of yours dies or
1518	is permissible to install a child watcher I<after> the child has been	1814	exits). It is permissible to install a child watcher I<after> the child
1519	forked (which implies it might have already exited), as long as the event	1815	has been forked (which implies it might have already exited), as long
1520	loop isn't entered (or is continued from a watcher).	1816	as the event loop isn't entered (or is continued from a watcher), i.e.,
		1817	forking and then immediately registering a watcher for the child is fine,
		1818	but forking and registering a watcher a few event loop iterations later is
		1819	not.
1521		1820
1522	Only the default event loop is capable of handling signals, and therefore	1821	Only the default event loop is capable of handling signals, and therefore
1523	you can only rgeister child watchers in the default event loop.	1822	you can only register child watchers in the default event loop.
1524		1823
1525	=head3 Process Interaction	1824	=head3 Process Interaction
1526		1825
1527	Libev grabs C<SIGCHLD> as soon as the default event loop is	1826	Libev grabs C<SIGCHLD> as soon as the default event loop is
1528	initialised. This is necessary to guarantee proper behaviour even if	1827	initialised. This is necessary to guarantee proper behaviour even if
1529	the first child watcher is started after the child exits. The occurance	1828	the first child watcher is started after the child exits. The occurrence
1530	of C<SIGCHLD> is recorded asynchronously, but child reaping is done	1829	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
1531	synchronously as part of the event loop processing. Libev always reaps all	1830	synchronously as part of the event loop processing. Libev always reaps all
1532	children, even ones not watched.	1831	children, even ones not watched.
1533		1832
1534	=head3 Overriding the Built-In Processing	1833	=head3 Overriding the Built-In Processing
…		…
1538	handler, you can override it easily by installing your own handler for	1837	handler, you can override it easily by installing your own handler for
1539	C<SIGCHLD> after initialising the default loop, and making sure the	1838	C<SIGCHLD> after initialising the default loop, and making sure the
1540	default loop never gets destroyed. You are encouraged, however, to use an	1839	default loop never gets destroyed. You are encouraged, however, to use an
1541	event-based approach to child reaping and thus use libev's support for	1840	event-based approach to child reaping and thus use libev's support for
1542	that, so other libev users can use C<ev_child> watchers freely.	1841	that, so other libev users can use C<ev_child> watchers freely.
		1842
		1843	=head3 Stopping the Child Watcher
		1844
		1845	Currently, the child watcher never gets stopped, even when the
		1846	child terminates, so normally one needs to stop the watcher in the
		1847	callback. Future versions of libev might stop the watcher automatically
		1848	when a child exit is detected.
1543		1849
1544	=head3 Watcher-Specific Functions and Data Members	1850	=head3 Watcher-Specific Functions and Data Members
1545		1851
1546	=over 4	1852	=over 4
1547		1853
…		…
1576	=head3 Examples	1882	=head3 Examples
1577		1883
1578	Example: C<fork()> a new process and install a child handler to wait for	1884	Example: C<fork()> a new process and install a child handler to wait for
1579	its completion.	1885	its completion.
1580		1886
1581	ev_child cw;	1887	ev_child cw;
1582		1888
1583	static void	1889	static void
1584	child_cb (EV_P_ struct ev_child *w, int revents)	1890	child_cb (EV_P_ ev_child *w, int revents)
1585	{	1891	{
1586	ev_child_stop (EV_A_ w);	1892	ev_child_stop (EV_A_ w);
1587	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);	1893	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1588	}	1894	}
1589		1895
1590	pid_t pid = fork ();	1896	pid_t pid = fork ();
1591		1897
1592	if (pid < 0)	1898	if (pid < 0)
1593	// error	1899	// error
1594	else if (pid == 0)	1900	else if (pid == 0)
1595	{	1901	{
1596	// the forked child executes here	1902	// the forked child executes here
1597	exit (1);	1903	exit (1);
1598	}	1904	}
1599	else	1905	else
1600	{	1906	{
1601	ev_child_init (&cw, child_cb, pid, 0);	1907	ev_child_init (&cw, child_cb, pid, 0);
1602	ev_child_start (EV_DEFAULT_ &cw);	1908	ev_child_start (EV_DEFAULT_ &cw);
1603	}	1909	}
1604		1910
1605		1911
1606	=head2 C<ev_stat> - did the file attributes just change?	1912	=head2 C<ev_stat> - did the file attributes just change?
1607		1913
1608	This watches a filesystem path for attribute changes. That is, it calls	1914	This watches a file system path for attribute changes. That is, it calls
1609	C<stat> regularly (or when the OS says it changed) and sees if it changed	1915	C<stat> regularly (or when the OS says it changed) and sees if it changed
1610	compared to the last time, invoking the callback if it did.	1916	compared to the last time, invoking the callback if it did.
1611		1917
1612	The path does not need to exist: changing from "path exists" to "path does	1918	The path does not need to exist: changing from "path exists" to "path does
1613	not exist" is a status change like any other. The condition "path does	1919	not exist" is a status change like any other. The condition "path does
…		…
1616	the stat buffer having unspecified contents.	1922	the stat buffer having unspecified contents.
1617		1923
1618	The path I<should> be absolute and I<must not> end in a slash. If it is	1924	The path I<should> be absolute and I<must not> end in a slash. If it is
1619	relative and your working directory changes, the behaviour is undefined.	1925	relative and your working directory changes, the behaviour is undefined.
1620		1926
1621	Since there is no standard to do this, the portable implementation simply	1927	Since there is no standard kernel interface to do this, the portable
1622	calls C<stat (2)> regularly on the path to see if it changed somehow. You	1928	implementation simply calls C<stat (2)> regularly on the path to see if
1623	can specify a recommended polling interval for this case. If you specify	1929	it changed somehow. You can specify a recommended polling interval for
1624	a polling interval of C<0> (highly recommended!) then a I<suitable,	1930	this case. If you specify a polling interval of C<0> (highly recommended!)
1625	unspecified default> value will be used (which you can expect to be around	1931	then a I<suitable, unspecified default> value will be used (which
1626	five seconds, although this might change dynamically). Libev will also	1932	you can expect to be around five seconds, although this might change
1627	impose a minimum interval which is currently around C<0.1>, but thats	1933	dynamically). Libev will also impose a minimum interval which is currently
1628	usually overkill.	1934	around C<0.1>, but thats usually overkill.
1629		1935
1630	This watcher type is not meant for massive numbers of stat watchers,	1936	This watcher type is not meant for massive numbers of stat watchers,
1631	as even with OS-supported change notifications, this can be	1937	as even with OS-supported change notifications, this can be
1632	resource-intensive.	1938	resource-intensive.
1633		1939
1634	At the time of this writing, only the Linux inotify interface is	1940	At the time of this writing, the only OS-specific interface implemented
1635	implemented (implementing kqueue support is left as an exercise for the	1941	is the Linux inotify interface (implementing kqueue support is left as
1636	reader, note, however, that the author sees no way of implementing ev_stat	1942	an exercise for the reader. Note, however, that the author sees no way
1637	semantics with kqueue). Inotify will be used to give hints only and should	1943	of implementing C<ev_stat> semantics with kqueue).
1638	not change the semantics of C<ev_stat> watchers, which means that libev
1639	sometimes needs to fall back to regular polling again even with inotify,
1640	but changes are usually detected immediately, and if the file exists there
1641	will be no polling.
1642		1944
1643	=head3 ABI Issues (Largefile Support)	1945	=head3 ABI Issues (Largefile Support)
1644		1946
1645	Libev by default (unless the user overrides this) uses the default	1947	Libev by default (unless the user overrides this) uses the default
1646	compilation environment, which means that on systems with optionally	1948	compilation environment, which means that on systems with large file
1647	disabled large file support, you get the 32 bit version of the stat	1949	support disabled by default, you get the 32 bit version of the stat
1648	structure. When using the library from programs that change the ABI to	1950	structure. When using the library from programs that change the ABI to
1649	use 64 bit file offsets the programs will fail. In that case you have to	1951	use 64 bit file offsets the programs will fail. In that case you have to
1650	compile libev with the same flags to get binary compatibility. This is	1952	compile libev with the same flags to get binary compatibility. This is
1651	obviously the case with any flags that change the ABI, but the problem is	1953	obviously the case with any flags that change the ABI, but the problem is
1652	most noticably with ev_stat and largefile support.	1954	most noticeably disabled with ev_stat and large file support.
1653		1955
1654	=head3 Inotify	1956	The solution for this is to lobby your distribution maker to make large
		1957	file interfaces available by default (as e.g. FreeBSD does) and not
		1958	optional. Libev cannot simply switch on large file support because it has
		1959	to exchange stat structures with application programs compiled using the
		1960	default compilation environment.
1655		1961
		1962	=head3 Inotify and Kqueue
		1963
1656	When C<inotify (7)> support has been compiled into libev (generally only	1964	When C<inotify (7)> support has been compiled into libev (generally
		1965	only available with Linux 2.6.25 or above due to bugs in earlier
1657	available on Linux) and present at runtime, it will be used to speed up	1966	implementations) and present at runtime, it will be used to speed up
1658	change detection where possible. The inotify descriptor will be created lazily	1967	change detection where possible. The inotify descriptor will be created
1659	when the first C<ev_stat> watcher is being started.	1968	lazily when the first C<ev_stat> watcher is being started.
1660		1969
1661	Inotify presence does not change the semantics of C<ev_stat> watchers	1970	Inotify presence does not change the semantics of C<ev_stat> watchers
1662	except that changes might be detected earlier, and in some cases, to avoid	1971	except that changes might be detected earlier, and in some cases, to avoid
1663	making regular C<stat> calls. Even in the presence of inotify support	1972	making regular C<stat> calls. Even in the presence of inotify support
1664	there are many cases where libev has to resort to regular C<stat> polling.	1973	there are many cases where libev has to resort to regular C<stat> polling,
		1974	but as long as the path exists, libev usually gets away without polling.
1665		1975
1666	(There is no support for kqueue, as apparently it cannot be used to	1976	There is no support for kqueue, as apparently it cannot be used to
1667	implement this functionality, due to the requirement of having a file	1977	implement this functionality, due to the requirement of having a file
1668	descriptor open on the object at all times).	1978	descriptor open on the object at all times, and detecting renames, unlinks
		1979	etc. is difficult.
1669		1980
1670	=head3 The special problem of stat time resolution	1981	=head3 The special problem of stat time resolution
1671		1982
1672	The C<stat ()> syscall only supports full-second resolution portably, and	1983	The C<stat ()> system call only supports full-second resolution portably, and
1673	even on systems where the resolution is higher, many filesystems still	1984	even on systems where the resolution is higher, most file systems still
1674	only support whole seconds.	1985	only support whole seconds.
1675		1986
1676	That means that, if the time is the only thing that changes, you can	1987	That means that, if the time is the only thing that changes, you can
1677	easily miss updates: on the first update, C<ev_stat> detects a change and	1988	easily miss updates: on the first update, C<ev_stat> detects a change and
1678	calls your callback, which does something. When there is another update	1989	calls your callback, which does something. When there is another update
1679	within the same second, C<ev_stat> will be unable to detect it as the stat	1990	within the same second, C<ev_stat> will be unable to detect unless the
1680	data does not change.	1991	stat data does change in other ways (e.g. file size).
1681		1992
1682	The solution to this is to delay acting on a change for slightly more	1993	The solution to this is to delay acting on a change for slightly more
1683	than a second (or till slightly after the next full second boundary), using	1994	than a second (or till slightly after the next full second boundary), using
1684	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);	1995	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
1685	ev_timer_again (loop, w)>).	1996	ev_timer_again (loop, w)>).
…		…
1705	C<path>. The C<interval> is a hint on how quickly a change is expected to	2016	C<path>. The C<interval> is a hint on how quickly a change is expected to
1706	be detected and should normally be specified as C<0> to let libev choose	2017	be detected and should normally be specified as C<0> to let libev choose
1707	a suitable value. The memory pointed to by C<path> must point to the same	2018	a suitable value. The memory pointed to by C<path> must point to the same
1708	path for as long as the watcher is active.	2019	path for as long as the watcher is active.
1709		2020
1710	The callback will receive C<EV_STAT> when a change was detected, relative	2021	The callback will receive an C<EV_STAT> event when a change was detected,
1711	to the attributes at the time the watcher was started (or the last change	2022	relative to the attributes at the time the watcher was started (or the
1712	was detected).	2023	last change was detected).
1713		2024
1714	=item ev_stat_stat (loop, ev_stat *)	2025	=item ev_stat_stat (loop, ev_stat *)
1715		2026
1716	Updates the stat buffer immediately with new values. If you change the	2027	Updates the stat buffer immediately with new values. If you change the
1717	watched path in your callback, you could call this function to avoid	2028	watched path in your callback, you could call this function to avoid
…		…
1738		2049
1739	The specified interval.	2050	The specified interval.
1740		2051
1741	=item const char *path [read-only]	2052	=item const char *path [read-only]
1742		2053
1743	The filesystem path that is being watched.	2054	The file system path that is being watched.
1744		2055
1745	=back	2056	=back
1746		2057
1747	=head3 Examples	2058	=head3 Examples
1748		2059
1749	Example: Watch C</etc/passwd> for attribute changes.	2060	Example: Watch C</etc/passwd> for attribute changes.
1750		2061
1751	static void	2062	static void
1752	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	2063	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1753	{	2064	{
1754	/* /etc/passwd changed in some way */	2065	/* /etc/passwd changed in some way */
1755	if (w->attr.st_nlink)	2066	if (w->attr.st_nlink)
1756	{	2067	{
1757	printf ("passwd current size %ld\n", (long)w->attr.st_size);	2068	printf ("passwd current size %ld\n", (long)w->attr.st_size);
1758	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);	2069	printf ("passwd current atime %ld\n", (long)w->attr.st_mtime);
1759	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);	2070	printf ("passwd current mtime %ld\n", (long)w->attr.st_mtime);
1760	}	2071	}
1761	else	2072	else
1762	/* you shalt not abuse printf for puts */	2073	/* you shalt not abuse printf for puts */
1763	puts ("wow, /etc/passwd is not there, expect problems. "	2074	puts ("wow, /etc/passwd is not there, expect problems. "
1764	"if this is windows, they already arrived\n");	2075	"if this is windows, they already arrived\n");
1765	}	2076	}
1766		2077
1767	...	2078	...
1768	ev_stat passwd;	2079	ev_stat passwd;
1769		2080
1770	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);	2081	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1771	ev_stat_start (loop, &passwd);	2082	ev_stat_start (loop, &passwd);
1772		2083
1773	Example: Like above, but additionally use a one-second delay so we do not	2084	Example: Like above, but additionally use a one-second delay so we do not
1774	miss updates (however, frequent updates will delay processing, too, so	2085	miss updates (however, frequent updates will delay processing, too, so
1775	one might do the work both on C<ev_stat> callback invocation I<and> on	2086	one might do the work both on C<ev_stat> callback invocation I<and> on
1776	C<ev_timer> callback invocation).	2087	C<ev_timer> callback invocation).
1777		2088
1778	static ev_stat passwd;	2089	static ev_stat passwd;
1779	static ev_timer timer;	2090	static ev_timer timer;
1780		2091
1781	static void	2092	static void
1782	timer_cb (EV_P_ ev_timer *w, int revents)	2093	timer_cb (EV_P_ ev_timer *w, int revents)
1783	{	2094	{
1784	ev_timer_stop (EV_A_ w);	2095	ev_timer_stop (EV_A_ w);
1785		2096
1786	/* now it's one second after the most recent passwd change */	2097	/* now it's one second after the most recent passwd change */
1787	}	2098	}
1788		2099
1789	static void	2100	static void
1790	stat_cb (EV_P_ ev_stat *w, int revents)	2101	stat_cb (EV_P_ ev_stat *w, int revents)
1791	{	2102	{
1792	/* reset the one-second timer */	2103	/* reset the one-second timer */
1793	ev_timer_again (EV_A_ &timer);	2104	ev_timer_again (EV_A_ &timer);
1794	}	2105	}
1795		2106
1796	...	2107	...
1797	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);	2108	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1798	ev_stat_start (loop, &passwd);	2109	ev_stat_start (loop, &passwd);
1799	ev_timer_init (&timer, timer_cb, 0., 1.02);	2110	ev_timer_init (&timer, timer_cb, 0., 1.02);
1800		2111
1801		2112
1802	=head2 C<ev_idle> - when you've got nothing better to do...	2113	=head2 C<ev_idle> - when you've got nothing better to do...
1803		2114
1804	Idle watchers trigger events when no other events of the same or higher	2115	Idle watchers trigger events when no other events of the same or higher
1805	priority are pending (prepare, check and other idle watchers do not	2116	priority are pending (prepare, check and other idle watchers do not count
1806	count).	2117	as receiving "events").
1807		2118
1808	That is, as long as your process is busy handling sockets or timeouts	2119	That is, as long as your process is busy handling sockets or timeouts
1809	(or even signals, imagine) of the same or higher priority it will not be	2120	(or even signals, imagine) of the same or higher priority it will not be
1810	triggered. But when your process is idle (or only lower-priority watchers	2121	triggered. But when your process is idle (or only lower-priority watchers
1811	are pending), the idle watchers are being called once per event loop	2122	are pending), the idle watchers are being called once per event loop
…		…
1835	=head3 Examples	2146	=head3 Examples
1836		2147
1837	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	2148	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1838	callback, free it. Also, use no error checking, as usual.	2149	callback, free it. Also, use no error checking, as usual.
1839		2150
1840	static void	2151	static void
1841	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	2152	idle_cb (struct ev_loop *loop, ev_idle *w, int revents)
1842	{	2153	{
1843	free (w);	2154	free (w);
1844	// now do something you wanted to do when the program has	2155	// now do something you wanted to do when the program has
1845	// no longer anything immediate to do.	2156	// no longer anything immediate to do.
1846	}	2157	}
1847		2158
1848	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	2159	ev_idle *idle_watcher = malloc (sizeof (ev_idle));
1849	ev_idle_init (idle_watcher, idle_cb);	2160	ev_idle_init (idle_watcher, idle_cb);
1850	ev_idle_start (loop, idle_cb);	2161	ev_idle_start (loop, idle_cb);
1851		2162
1852		2163
1853	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!	2164	=head2 C<ev_prepare> and C<ev_check> - customise your event loop!
1854		2165
1855	Prepare and check watchers are usually (but not always) used in tandem:	2166	Prepare and check watchers are usually (but not always) used in pairs:
1856	prepare watchers get invoked before the process blocks and check watchers	2167	prepare watchers get invoked before the process blocks and check watchers
1857	afterwards.	2168	afterwards.
1858		2169
1859	You I<must not> call C<ev_loop> or similar functions that enter	2170	You I<must not> call C<ev_loop> or similar functions that enter
1860	the current event loop from either C<ev_prepare> or C<ev_check>	2171	the current event loop from either C<ev_prepare> or C<ev_check>
…		…
1863	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,	2174	those watchers, i.e. the sequence will always be C<ev_prepare>, blocking,
1864	C<ev_check> so if you have one watcher of each kind they will always be	2175	C<ev_check> so if you have one watcher of each kind they will always be
1865	called in pairs bracketing the blocking call.	2176	called in pairs bracketing the blocking call.
1866		2177
1867	Their main purpose is to integrate other event mechanisms into libev and	2178	Their main purpose is to integrate other event mechanisms into libev and
1868	their use is somewhat advanced. This could be used, for example, to track	2179	their use is somewhat advanced. They could be used, for example, to track
1869	variable changes, implement your own watchers, integrate net-snmp or a	2180	variable changes, implement your own watchers, integrate net-snmp or a
1870	coroutine library and lots more. They are also occasionally useful if	2181	coroutine library and lots more. They are also occasionally useful if
1871	you cache some data and want to flush it before blocking (for example,	2182	you cache some data and want to flush it before blocking (for example,
1872	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>	2183	in X programs you might want to do an C<XFlush ()> in an C<ev_prepare>
1873	watcher).	2184	watcher).
1874		2185
1875	This is done by examining in each prepare call which file descriptors need	2186	This is done by examining in each prepare call which file descriptors
1876	to be watched by the other library, registering C<ev_io> watchers for	2187	need to be watched by the other library, registering C<ev_io> watchers
1877	them and starting an C<ev_timer> watcher for any timeouts (many libraries	2188	for them and starting an C<ev_timer> watcher for any timeouts (many
1878	provide just this functionality). Then, in the check watcher you check for	2189	libraries provide exactly this functionality). Then, in the check watcher,
1879	any events that occured (by checking the pending status of all watchers	2190	you check for any events that occurred (by checking the pending status
1880	and stopping them) and call back into the library. The I/O and timer	2191	of all watchers and stopping them) and call back into the library. The
1881	callbacks will never actually be called (but must be valid nevertheless,	2192	I/O and timer callbacks will never actually be called (but must be valid
1882	because you never know, you know?).	2193	nevertheless, because you never know, you know?).
1883		2194
1884	As another example, the Perl Coro module uses these hooks to integrate	2195	As another example, the Perl Coro module uses these hooks to integrate
1885	coroutines into libev programs, by yielding to other active coroutines	2196	coroutines into libev programs, by yielding to other active coroutines
1886	during each prepare and only letting the process block if no coroutines	2197	during each prepare and only letting the process block if no coroutines
1887	are ready to run (it's actually more complicated: it only runs coroutines	2198	are ready to run (it's actually more complicated: it only runs coroutines
…		…
1890	loop from blocking if lower-priority coroutines are active, thus mapping	2201	loop from blocking if lower-priority coroutines are active, thus mapping
1891	low-priority coroutines to idle/background tasks).	2202	low-priority coroutines to idle/background tasks).
1892		2203
1893	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	2204	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1894	priority, to ensure that they are being run before any other watchers	2205	priority, to ensure that they are being run before any other watchers
		2206	after the poll (this doesn't matter for C<ev_prepare> watchers).
		2207
1895	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	2208	Also, C<ev_check> watchers (and C<ev_prepare> watchers, too) should not
1896	too) should not activate ("feed") events into libev. While libev fully	2209	activate ("feed") events into libev. While libev fully supports this, they
1897	supports this, they might get executed before other C<ev_check> watchers	2210	might get executed before other C<ev_check> watchers did their job. As
1898	did their job. As C<ev_check> watchers are often used to embed other	2211	C<ev_check> watchers are often used to embed other (non-libev) event
1899	(non-libev) event loops those other event loops might be in an unusable	2212	loops those other event loops might be in an unusable state until their
1900	state until their C<ev_check> watcher ran (always remind yourself to	2213	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
1901	coexist peacefully with others).	2214	others).
1902		2215
1903	=head3 Watcher-Specific Functions and Data Members	2216	=head3 Watcher-Specific Functions and Data Members
1904		2217
1905	=over 4	2218	=over 4
1906		2219
…		…
1908		2221
1909	=item ev_check_init (ev_check *, callback)	2222	=item ev_check_init (ev_check *, callback)
1910		2223
1911	Initialises and configures the prepare or check watcher - they have no	2224	Initialises and configures the prepare or check watcher - they have no
1912	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	2225	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1913	macros, but using them is utterly, utterly and completely pointless.	2226	macros, but using them is utterly, utterly, utterly and completely
		2227	pointless.
1914		2228
1915	=back	2229	=back
1916		2230
1917	=head3 Examples	2231	=head3 Examples
1918		2232
…		…
1927	and in a check watcher, destroy them and call into libadns. What follows	2241	and in a check watcher, destroy them and call into libadns. What follows
1928	is pseudo-code only of course. This requires you to either use a low	2242	is pseudo-code only of course. This requires you to either use a low
1929	priority for the check watcher or use C<ev_clear_pending> explicitly, as	2243	priority for the check watcher or use C<ev_clear_pending> explicitly, as
1930	the callbacks for the IO/timeout watchers might not have been called yet.	2244	the callbacks for the IO/timeout watchers might not have been called yet.
1931		2245
1932	static ev_io iow [nfd];	2246	static ev_io iow [nfd];
1933	static ev_timer tw;	2247	static ev_timer tw;
1934		2248
1935	static void	2249	static void
1936	io_cb (ev_loop *loop, ev_io *w, int revents)	2250	io_cb (struct ev_loop *loop, ev_io *w, int revents)
1937	{	2251	{
1938	}	2252	}
1939		2253
1940	// create io watchers for each fd and a timer before blocking	2254	// create io watchers for each fd and a timer before blocking
1941	static void	2255	static void
1942	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	2256	adns_prepare_cb (struct ev_loop *loop, ev_prepare *w, int revents)
1943	{	2257	{
1944	int timeout = 3600000;	2258	int timeout = 3600000;
1945	struct pollfd fds [nfd];	2259	struct pollfd fds [nfd];
1946	// actual code will need to loop here and realloc etc.	2260	// actual code will need to loop here and realloc etc.
1947	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	2261	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1948		2262
1949	/* the callback is illegal, but won't be called as we stop during check */	2263	/* the callback is illegal, but won't be called as we stop during check */
1950	ev_timer_init (&tw, 0, timeout * 1e-3);	2264	ev_timer_init (&tw, 0, timeout * 1e-3);
1951	ev_timer_start (loop, &tw);	2265	ev_timer_start (loop, &tw);
1952		2266
1953	// create one ev_io per pollfd	2267	// create one ev_io per pollfd
1954	for (int i = 0; i < nfd; ++i)	2268	for (int i = 0; i < nfd; ++i)
1955	{	2269	{
1956	ev_io_init (iow + i, io_cb, fds [i].fd,	2270	ev_io_init (iow + i, io_cb, fds [i].fd,
1957	((fds [i].events & POLLIN ? EV_READ : 0)	2271	((fds [i].events & POLLIN ? EV_READ : 0)
1958	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	2272	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1959		2273
1960	fds [i].revents = 0;	2274	fds [i].revents = 0;
1961	ev_io_start (loop, iow + i);	2275	ev_io_start (loop, iow + i);
1962	}	2276	}
1963	}	2277	}
1964		2278
1965	// stop all watchers after blocking	2279	// stop all watchers after blocking
1966	static void	2280	static void
1967	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	2281	adns_check_cb (struct ev_loop *loop, ev_check *w, int revents)
1968	{	2282	{
1969	ev_timer_stop (loop, &tw);	2283	ev_timer_stop (loop, &tw);
1970		2284
1971	for (int i = 0; i < nfd; ++i)	2285	for (int i = 0; i < nfd; ++i)
1972	{	2286	{
1973	// set the relevant poll flags	2287	// set the relevant poll flags
1974	// could also call adns_processreadable etc. here	2288	// could also call adns_processreadable etc. here
1975	struct pollfd *fd = fds + i;	2289	struct pollfd *fd = fds + i;
1976	int revents = ev_clear_pending (iow + i);	2290	int revents = ev_clear_pending (iow + i);
1977	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;	2291	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1978	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;	2292	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1979		2293
1980	// now stop the watcher	2294	// now stop the watcher
1981	ev_io_stop (loop, iow + i);	2295	ev_io_stop (loop, iow + i);
1982	}	2296	}
1983		2297
1984	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	2298	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1985	}	2299	}
1986		2300
1987	Method 2: This would be just like method 1, but you run C<adns_afterpoll>	2301	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
1988	in the prepare watcher and would dispose of the check watcher.	2302	in the prepare watcher and would dispose of the check watcher.
1989		2303
1990	Method 3: If the module to be embedded supports explicit event	2304	Method 3: If the module to be embedded supports explicit event
1991	notification (adns does), you can also make use of the actual watcher	2305	notification (libadns does), you can also make use of the actual watcher
1992	callbacks, and only destroy/create the watchers in the prepare watcher.	2306	callbacks, and only destroy/create the watchers in the prepare watcher.
1993		2307
1994	static void	2308	static void
1995	timer_cb (EV_P_ ev_timer *w, int revents)	2309	timer_cb (EV_P_ ev_timer *w, int revents)
1996	{	2310	{
1997	adns_state ads = (adns_state)w->data;	2311	adns_state ads = (adns_state)w->data;
1998	update_now (EV_A);	2312	update_now (EV_A);
1999		2313
2000	adns_processtimeouts (ads, &tv_now);	2314	adns_processtimeouts (ads, &tv_now);
2001	}	2315	}
2002		2316
2003	static void	2317	static void
2004	io_cb (EV_P_ ev_io *w, int revents)	2318	io_cb (EV_P_ ev_io *w, int revents)
2005	{	2319	{
2006	adns_state ads = (adns_state)w->data;	2320	adns_state ads = (adns_state)w->data;
2007	update_now (EV_A);	2321	update_now (EV_A);
2008		2322
2009	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);	2323	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
2010	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);	2324	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
2011	}	2325	}
2012		2326
2013	// do not ever call adns_afterpoll	2327	// do not ever call adns_afterpoll
2014		2328
2015	Method 4: Do not use a prepare or check watcher because the module you	2329	Method 4: Do not use a prepare or check watcher because the module you
2016	want to embed is too inflexible to support it. Instead, youc na override	2330	want to embed is not flexible enough to support it. Instead, you can
2017	their poll function. The drawback with this solution is that the main	2331	override their poll function. The drawback with this solution is that the
2018	loop is now no longer controllable by EV. The C<Glib::EV> module does	2332	main loop is now no longer controllable by EV. The C<Glib::EV> module uses
2019	this.	2333	this approach, effectively embedding EV as a client into the horrible
		2334	libglib event loop.
2020		2335
2021	static gint	2336	static gint
2022	event_poll_func (GPollFD *fds, guint nfds, gint timeout)	2337	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
2023	{	2338	{
2024	int got_events = 0;	2339	int got_events = 0;
2025		2340
2026	for (n = 0; n < nfds; ++n)	2341	for (n = 0; n < nfds; ++n)
2027	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events	2342	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
2028		2343
2029	if (timeout >= 0)	2344	if (timeout >= 0)
2030	// create/start timer	2345	// create/start timer
2031		2346
2032	// poll	2347	// poll
2033	ev_loop (EV_A_ 0);	2348	ev_loop (EV_A_ 0);
2034		2349
2035	// stop timer again	2350	// stop timer again
2036	if (timeout >= 0)	2351	if (timeout >= 0)
2037	ev_timer_stop (EV_A_ &to);	2352	ev_timer_stop (EV_A_ &to);
2038		2353
2039	// stop io watchers again - their callbacks should have set	2354	// stop io watchers again - their callbacks should have set
2040	for (n = 0; n < nfds; ++n)	2355	for (n = 0; n < nfds; ++n)
2041	ev_io_stop (EV_A_ iow [n]);	2356	ev_io_stop (EV_A_ iow [n]);
2042		2357
2043	return got_events;	2358	return got_events;
2044	}	2359	}
2045		2360
2046		2361
2047	=head2 C<ev_embed> - when one backend isn't enough...	2362	=head2 C<ev_embed> - when one backend isn't enough...
2048		2363
2049	This is a rather advanced watcher type that lets you embed one event loop	2364	This is a rather advanced watcher type that lets you embed one event loop
…		…
2055	prioritise I/O.	2370	prioritise I/O.
2056		2371
2057	As an example for a bug workaround, the kqueue backend might only support	2372	As an example for a bug workaround, the kqueue backend might only support
2058	sockets on some platform, so it is unusable as generic backend, but you	2373	sockets on some platform, so it is unusable as generic backend, but you
2059	still want to make use of it because you have many sockets and it scales	2374	still want to make use of it because you have many sockets and it scales
2060	so nicely. In this case, you would create a kqueue-based loop and embed it	2375	so nicely. In this case, you would create a kqueue-based loop and embed
2061	into your default loop (which might use e.g. poll). Overall operation will	2376	it into your default loop (which might use e.g. poll). Overall operation
2062	be a bit slower because first libev has to poll and then call kevent, but	2377	will be a bit slower because first libev has to call C<poll> and then
2063	at least you can use both at what they are best.	2378	C<kevent>, but at least you can use both mechanisms for what they are
		2379	best: C<kqueue> for scalable sockets and C<poll> if you want it to work :)
2064		2380
2065	As for prioritising I/O: rarely you have the case where some fds have	2381	As for prioritising I/O: under rare circumstances you have the case where
2066	to be watched and handled very quickly (with low latency), and even	2382	some fds have to be watched and handled very quickly (with low latency),
2067	priorities and idle watchers might have too much overhead. In this case	2383	and even priorities and idle watchers might have too much overhead. In
2068	you would put all the high priority stuff in one loop and all the rest in	2384	this case you would put all the high priority stuff in one loop and all
2069	a second one, and embed the second one in the first.	2385	the rest in a second one, and embed the second one in the first.
2070		2386
2071	As long as the watcher is active, the callback will be invoked every time	2387	As long as the watcher is active, the callback will be invoked every time
2072	there might be events pending in the embedded loop. The callback must then	2388	there might be events pending in the embedded loop. The callback must then
2073	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke	2389	call C<ev_embed_sweep (mainloop, watcher)> to make a single sweep and invoke
2074	their callbacks (you could also start an idle watcher to give the embedded	2390	their callbacks (you could also start an idle watcher to give the embedded
…		…
2082	interested in that.	2398	interested in that.
2083		2399
2084	Also, there have not currently been made special provisions for forking:	2400	Also, there have not currently been made special provisions for forking:
2085	when you fork, you not only have to call C<ev_loop_fork> on both loops,	2401	when you fork, you not only have to call C<ev_loop_fork> on both loops,
2086	but you will also have to stop and restart any C<ev_embed> watchers	2402	but you will also have to stop and restart any C<ev_embed> watchers
2087	yourself.	2403	yourself - but you can use a fork watcher to handle this automatically,
		2404	and future versions of libev might do just that.
2088		2405
2089	Unfortunately, not all backends are embeddable, only the ones returned by	2406	Unfortunately, not all backends are embeddable: only the ones returned by
2090	C<ev_embeddable_backends> are, which, unfortunately, does not include any	2407	C<ev_embeddable_backends> are, which, unfortunately, does not include any
2091	portable one.	2408	portable one.
2092		2409
2093	So when you want to use this feature you will always have to be prepared	2410	So when you want to use this feature you will always have to be prepared
2094	that you cannot get an embeddable loop. The recommended way to get around	2411	that you cannot get an embeddable loop. The recommended way to get around
2095	this is to have a separate variables for your embeddable loop, try to	2412	this is to have a separate variables for your embeddable loop, try to
2096	create it, and if that fails, use the normal loop for everything.	2413	create it, and if that fails, use the normal loop for everything.
2097		2414
		2415	=head3 C<ev_embed> and fork
		2416
		2417	While the C<ev_embed> watcher is running, forks in the embedding loop will
		2418	automatically be applied to the embedded loop as well, so no special
		2419	fork handling is required in that case. When the watcher is not running,
		2420	however, it is still the task of the libev user to call C<ev_loop_fork ()>
		2421	as applicable.
		2422
2098	=head3 Watcher-Specific Functions and Data Members	2423	=head3 Watcher-Specific Functions and Data Members
2099		2424
2100	=over 4	2425	=over 4
2101		2426
2102	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	2427	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
…		…
2105		2430
2106	Configures the watcher to embed the given loop, which must be	2431	Configures the watcher to embed the given loop, which must be
2107	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be	2432	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
2108	invoked automatically, otherwise it is the responsibility of the callback	2433	invoked automatically, otherwise it is the responsibility of the callback
2109	to invoke it (it will continue to be called until the sweep has been done,	2434	to invoke it (it will continue to be called until the sweep has been done,
2110	if you do not want thta, you need to temporarily stop the embed watcher).	2435	if you do not want that, you need to temporarily stop the embed watcher).
2111		2436
2112	=item ev_embed_sweep (loop, ev_embed *)	2437	=item ev_embed_sweep (loop, ev_embed *)
2113		2438
2114	Make a single, non-blocking sweep over the embedded loop. This works	2439	Make a single, non-blocking sweep over the embedded loop. This works
2115	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	2440	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
2116	apropriate way for embedded loops.	2441	appropriate way for embedded loops.
2117		2442
2118	=item struct ev_loop *other [read-only]	2443	=item struct ev_loop *other [read-only]
2119		2444
2120	The embedded event loop.	2445	The embedded event loop.
2121		2446
…		…
2123		2448
2124	=head3 Examples	2449	=head3 Examples
2125		2450
2126	Example: Try to get an embeddable event loop and embed it into the default	2451	Example: Try to get an embeddable event loop and embed it into the default
2127	event loop. If that is not possible, use the default loop. The default	2452	event loop. If that is not possible, use the default loop. The default
2128	loop is stored in C<loop_hi>, while the mebeddable loop is stored in	2453	loop is stored in C<loop_hi>, while the embeddable loop is stored in
2129	C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be	2454	C<loop_lo> (which is C<loop_hi> in the case no embeddable loop can be
2130	used).	2455	used).
2131		2456
2132	struct ev_loop *loop_hi = ev_default_init (0);	2457	struct ev_loop *loop_hi = ev_default_init (0);
2133	struct ev_loop *loop_lo = 0;	2458	struct ev_loop *loop_lo = 0;
2134	struct ev_embed embed;	2459	ev_embed embed;
2135		2460
2136	// see if there is a chance of getting one that works	2461	// see if there is a chance of getting one that works
2137	// (remember that a flags value of 0 means autodetection)	2462	// (remember that a flags value of 0 means autodetection)
2138	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()	2463	loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2139	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())	2464	? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2140	: 0;	2465	: 0;
2141		2466
2142	// if we got one, then embed it, otherwise default to loop_hi	2467	// if we got one, then embed it, otherwise default to loop_hi
2143	if (loop_lo)	2468	if (loop_lo)
2144	{	2469	{
2145	ev_embed_init (&embed, 0, loop_lo);	2470	ev_embed_init (&embed, 0, loop_lo);
2146	ev_embed_start (loop_hi, &embed);	2471	ev_embed_start (loop_hi, &embed);
2147	}	2472	}
2148	else	2473	else
2149	loop_lo = loop_hi;	2474	loop_lo = loop_hi;
2150		2475
2151	Example: Check if kqueue is available but not recommended and create	2476	Example: Check if kqueue is available but not recommended and create
2152	a kqueue backend for use with sockets (which usually work with any	2477	a kqueue backend for use with sockets (which usually work with any
2153	kqueue implementation). Store the kqueue/socket-only event loop in	2478	kqueue implementation). Store the kqueue/socket-only event loop in
2154	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).	2479	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
2155		2480
2156	struct ev_loop *loop = ev_default_init (0);	2481	struct ev_loop *loop = ev_default_init (0);
2157	struct ev_loop *loop_socket = 0;	2482	struct ev_loop *loop_socket = 0;
2158	struct ev_embed embed;	2483	ev_embed embed;
2159		2484
2160	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)	2485	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2161	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))	2486	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2162	{	2487	{
2163	ev_embed_init (&embed, 0, loop_socket);	2488	ev_embed_init (&embed, 0, loop_socket);
2164	ev_embed_start (loop, &embed);	2489	ev_embed_start (loop, &embed);
2165	}	2490	}
2166		2491
2167	if (!loop_socket)	2492	if (!loop_socket)
2168	loop_socket = loop;	2493	loop_socket = loop;
2169		2494
2170	// now use loop_socket for all sockets, and loop for everything else	2495	// now use loop_socket for all sockets, and loop for everything else
2171		2496
2172		2497
2173	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	2498	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
2174		2499
2175	Fork watchers are called when a C<fork ()> was detected (usually because	2500	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
2219	is that the author does not know of a simple (or any) algorithm for a	2544	is that the author does not know of a simple (or any) algorithm for a
2220	multiple-writer-single-reader queue that works in all cases and doesn't	2545	multiple-writer-single-reader queue that works in all cases and doesn't
2221	need elaborate support such as pthreads.	2546	need elaborate support such as pthreads.
2222		2547
2223	That means that if you want to queue data, you have to provide your own	2548	That means that if you want to queue data, you have to provide your own
2224	queue. But at least I can tell you would implement locking around your	2549	queue. But at least I can tell you how to implement locking around your
2225	queue:	2550	queue:
2226		2551
2227	=over 4	2552	=over 4
2228		2553
2229	=item queueing from a signal handler context	2554	=item queueing from a signal handler context
2230		2555
2231	To implement race-free queueing, you simply add to the queue in the signal	2556	To implement race-free queueing, you simply add to the queue in the signal
2232	handler but you block the signal handler in the watcher callback. Here is an example that does that for	2557	handler but you block the signal handler in the watcher callback. Here is
2233	some fictitiuous SIGUSR1 handler:	2558	an example that does that for some fictitious SIGUSR1 handler:
2234		2559
2235	static ev_async mysig;	2560	static ev_async mysig;
2236		2561
2237	static void	2562	static void
2238	sigusr1_handler (void)	2563	sigusr1_handler (void)
…		…
2305		2630
2306	=item ev_async_init (ev_async *, callback)	2631	=item ev_async_init (ev_async *, callback)
2307		2632
2308	Initialises and configures the async watcher - it has no parameters of any	2633	Initialises and configures the async watcher - it has no parameters of any
2309	kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,	2634	kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,
2310	believe me.	2635	trust me.
2311		2636
2312	=item ev_async_send (loop, ev_async *)	2637	=item ev_async_send (loop, ev_async *)
2313		2638
2314	Sends/signals/activates the given C<ev_async> watcher, that is, feeds	2639	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
2315	an C<EV_ASYNC> event on the watcher into the event loop. Unlike	2640	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
2316	C<ev_feed_event>, this call is safe to do in other threads, signal or	2641	C<ev_feed_event>, this call is safe to do from other threads, signal or
2317	similar contexts (see the dicusssion of C<EV_ATOMIC_T> in the embedding	2642	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding
2318	section below on what exactly this means).	2643	section below on what exactly this means).
2319		2644
2320	This call incurs the overhead of a syscall only once per loop iteration,	2645	This call incurs the overhead of a system call only once per loop iteration,
2321	so while the overhead might be noticable, it doesn't apply to repeated	2646	so while the overhead might be noticeable, it doesn't apply to repeated
2322	calls to C<ev_async_send>.	2647	calls to C<ev_async_send>.
2323		2648
2324	=item bool = ev_async_pending (ev_async *)	2649	=item bool = ev_async_pending (ev_async *)
2325		2650
2326	Returns a non-zero value when C<ev_async_send> has been called on the	2651	Returns a non-zero value when C<ev_async_send> has been called on the
…		…
2328	event loop.	2653	event loop.
2329		2654
2330	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When	2655	C<ev_async_send> sets a flag in the watcher and wakes up the loop. When
2331	the loop iterates next and checks for the watcher to have become active,	2656	the loop iterates next and checks for the watcher to have become active,
2332	it will reset the flag again. C<ev_async_pending> can be used to very	2657	it will reset the flag again. C<ev_async_pending> can be used to very
2333	quickly check wether invoking the loop might be a good idea.	2658	quickly check whether invoking the loop might be a good idea.
2334		2659
2335	Not that this does I<not> check wether the watcher itself is pending, only	2660	Not that this does I<not> check whether the watcher itself is pending, only
2336	wether it has been requested to make this watcher pending.	2661	whether it has been requested to make this watcher pending.
2337		2662
2338	=back	2663	=back
2339		2664
2340		2665
2341	=head1 OTHER FUNCTIONS	2666	=head1 OTHER FUNCTIONS
…		…
2345	=over 4	2670	=over 4
2346		2671
2347	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2672	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
2348		2673
2349	This function combines a simple timer and an I/O watcher, calls your	2674	This function combines a simple timer and an I/O watcher, calls your
2350	callback on whichever event happens first and automatically stop both	2675	callback on whichever event happens first and automatically stops both
2351	watchers. This is useful if you want to wait for a single event on an fd	2676	watchers. This is useful if you want to wait for a single event on an fd
2352	or timeout without having to allocate/configure/start/stop/free one or	2677	or timeout without having to allocate/configure/start/stop/free one or
2353	more watchers yourself.	2678	more watchers yourself.
2354		2679
2355	If C<fd> is less than 0, then no I/O watcher will be started and events	2680	If C<fd> is less than 0, then no I/O watcher will be started and the
2356	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	2681	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
2357	C<events> set will be craeted and started.	2682	the given C<fd> and C<events> set will be created and started.
2358		2683
2359	If C<timeout> is less than 0, then no timeout watcher will be	2684	If C<timeout> is less than 0, then no timeout watcher will be
2360	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	2685	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
2361	repeat = 0) will be started. While C<0> is a valid timeout, it is of	2686	repeat = 0) will be started. C<0> is a valid timeout.
2362	dubious value.
2363		2687
2364	The callback has the type C<void (*cb)(int revents, void *arg)> and gets	2688	The callback has the type C<void (*cb)(int revents, void *arg)> and gets
2365	passed an C<revents> set like normal event callbacks (a combination of	2689	passed an C<revents> set like normal event callbacks (a combination of
2366	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	2690	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
2367	value passed to C<ev_once>:	2691	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2692	a timeout and an io event at the same time - you probably should give io
		2693	events precedence.
2368		2694
		2695	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
		2696
2369	static void stdin_ready (int revents, void *arg)	2697	static void stdin_ready (int revents, void *arg)
2370	{	2698	{
2371	if (revents & EV_TIMEOUT)
2372	/* doh, nothing entered */;
2373	else if (revents & EV_READ)	2699	if (revents & EV_READ)
2374	/* stdin might have data for us, joy! */;	2700	/* stdin might have data for us, joy! */;
		2701	else if (revents & EV_TIMEOUT)
		2702	/* doh, nothing entered */;
2375	}	2703	}
2376		2704
2377	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	2705	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
2378		2706
2379	=item ev_feed_event (ev_loop *, watcher *, int revents)	2707	=item ev_feed_event (struct ev_loop *, watcher *, int revents)
2380		2708
2381	Feeds the given event set into the event loop, as if the specified event	2709	Feeds the given event set into the event loop, as if the specified event
2382	had happened for the specified watcher (which must be a pointer to an	2710	had happened for the specified watcher (which must be a pointer to an
2383	initialised but not necessarily started event watcher).	2711	initialised but not necessarily started event watcher).
2384		2712
2385	=item ev_feed_fd_event (ev_loop *, int fd, int revents)	2713	=item ev_feed_fd_event (struct ev_loop *, int fd, int revents)
2386		2714
2387	Feed an event on the given fd, as if a file descriptor backend detected	2715	Feed an event on the given fd, as if a file descriptor backend detected
2388	the given events it.	2716	the given events it.
2389		2717
2390	=item ev_feed_signal_event (ev_loop *loop, int signum)	2718	=item ev_feed_signal_event (struct ev_loop *loop, int signum)
2391		2719
2392	Feed an event as if the given signal occured (C<loop> must be the default	2720	Feed an event as if the given signal occurred (C<loop> must be the default
2393	loop!).	2721	loop!).
2394		2722
2395	=back	2723	=back
2396		2724
2397		2725
…		…
2426	=back	2754	=back
2427		2755
2428	=head1 C++ SUPPORT	2756	=head1 C++ SUPPORT
2429		2757
2430	Libev comes with some simplistic wrapper classes for C++ that mainly allow	2758	Libev comes with some simplistic wrapper classes for C++ that mainly allow
2431	you to use some convinience methods to start/stop watchers and also change	2759	you to use some convenience methods to start/stop watchers and also change
2432	the callback model to a model using method callbacks on objects.	2760	the callback model to a model using method callbacks on objects.
2433		2761
2434	To use it,	2762	To use it,
2435		2763
2436	#include <ev++.h>	2764	#include <ev++.h>
2437		2765
2438	This automatically includes F<ev.h> and puts all of its definitions (many	2766	This automatically includes F<ev.h> and puts all of its definitions (many
2439	of them macros) into the global namespace. All C++ specific things are	2767	of them macros) into the global namespace. All C++ specific things are
2440	put into the C<ev> namespace. It should support all the same embedding	2768	put into the C<ev> namespace. It should support all the same embedding
2441	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.	2769	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
…		…
2508	your compiler is good :), then the method will be fully inlined into the	2836	your compiler is good :), then the method will be fully inlined into the
2509	thunking function, making it as fast as a direct C callback.	2837	thunking function, making it as fast as a direct C callback.
2510		2838
2511	Example: simple class declaration and watcher initialisation	2839	Example: simple class declaration and watcher initialisation
2512		2840
2513	struct myclass	2841	struct myclass
2514	{	2842	{
2515	void io_cb (ev::io &w, int revents) { }	2843	void io_cb (ev::io &w, int revents) { }
2516	}	2844	}
2517		2845
2518	myclass obj;	2846	myclass obj;
2519	ev::io iow;	2847	ev::io iow;
2520	iow.set <myclass, &myclass::io_cb> (&obj);	2848	iow.set <myclass, &myclass::io_cb> (&obj);
2521		2849
2522	=item w->set<function> (void *data = 0)	2850	=item w->set<function> (void *data = 0)
2523		2851
2524	Also sets a callback, but uses a static method or plain function as	2852	Also sets a callback, but uses a static method or plain function as
2525	callback. The optional C<data> argument will be stored in the watcher's	2853	callback. The optional C<data> argument will be stored in the watcher's
…		…
2527		2855
2528	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.	2856	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
2529		2857
2530	See the method-C<set> above for more details.	2858	See the method-C<set> above for more details.
2531		2859
2532	Example:	2860	Example: Use a plain function as callback.
2533		2861
2534	static void io_cb (ev::io &w, int revents) { }	2862	static void io_cb (ev::io &w, int revents) { }
2535	iow.set <io_cb> ();	2863	iow.set <io_cb> ();
2536		2864
2537	=item w->set (struct ev_loop *)	2865	=item w->set (struct ev_loop *)
2538		2866
2539	Associates a different C<struct ev_loop> with this watcher. You can only	2867	Associates a different C<struct ev_loop> with this watcher. You can only
2540	do this when the watcher is inactive (and not pending either).	2868	do this when the watcher is inactive (and not pending either).
2541		2869
2542	=item w->set ([args])	2870	=item w->set ([arguments])
2543		2871
2544	Basically the same as C<ev_TYPE_set>, with the same args. Must be	2872	Basically the same as C<ev_TYPE_set>, with the same arguments. Must be
2545	called at least once. Unlike the C counterpart, an active watcher gets	2873	called at least once. Unlike the C counterpart, an active watcher gets
2546	automatically stopped and restarted when reconfiguring it with this	2874	automatically stopped and restarted when reconfiguring it with this
2547	method.	2875	method.
2548		2876
2549	=item w->start ()	2877	=item w->start ()
…		…
2573	=back	2901	=back
2574		2902
2575	Example: Define a class with an IO and idle watcher, start one of them in	2903	Example: Define a class with an IO and idle watcher, start one of them in
2576	the constructor.	2904	the constructor.
2577		2905
2578	class myclass	2906	class myclass
2579	{	2907	{
2580	ev::io io; void io_cb (ev::io &w, int revents);	2908	ev::io io ; void io_cb (ev::io &w, int revents);
2581	ev:idle idle void idle_cb (ev::idle &w, int revents);	2909	ev::idle idle; void idle_cb (ev::idle &w, int revents);
2582		2910
2583	myclass (int fd)	2911	myclass (int fd)
2584	{	2912	{
2585	io .set <myclass, &myclass::io_cb > (this);	2913	io .set <myclass, &myclass::io_cb > (this);
2586	idle.set <myclass, &myclass::idle_cb> (this);	2914	idle.set <myclass, &myclass::idle_cb> (this);
2587		2915
2588	io.start (fd, ev::READ);	2916	io.start (fd, ev::READ);
2589	}	2917	}
2590	};	2918	};
2591		2919
2592		2920
2593	=head1 OTHER LANGUAGE BINDINGS	2921	=head1 OTHER LANGUAGE BINDINGS
2594		2922
2595	Libev does not offer other language bindings itself, but bindings for a	2923	Libev does not offer other language bindings itself, but bindings for a
2596	numbe rof languages exist in the form of third-party packages. If you know	2924	number of languages exist in the form of third-party packages. If you know
2597	any interesting language binding in addition to the ones listed here, drop	2925	any interesting language binding in addition to the ones listed here, drop
2598	me a note.	2926	me a note.
2599		2927
2600	=over 4	2928	=over 4
2601		2929
2602	=item Perl	2930	=item Perl
2603		2931
2604	The EV module implements the full libev API and is actually used to test	2932	The EV module implements the full libev API and is actually used to test
2605	libev. EV is developed together with libev. Apart from the EV core module,	2933	libev. EV is developed together with libev. Apart from the EV core module,
2606	there are additional modules that implement libev-compatible interfaces	2934	there are additional modules that implement libev-compatible interfaces
2607	to C<libadns> (C<EV::ADNS>), C<Net::SNMP> (C<Net::SNMP::EV>) and the	2935	to C<libadns> (C<EV::ADNS>, but C<AnyEvent::DNS> is preferred nowadays),
2608	C<libglib> event core (C<Glib::EV> and C<EV::Glib>).	2936	C<Net::SNMP> (C<Net::SNMP::EV>) and the C<libglib> event core (C<Glib::EV>
		2937	and C<EV::Glib>).
2609		2938
2610	It can be found and installed via CPAN, its homepage is found at	2939	It can be found and installed via CPAN, its homepage is at
2611	L<http://software.schmorp.de/pkg/EV>.	2940	L<http://software.schmorp.de/pkg/EV>.
2612		2941
		2942	=item Python
		2943
		2944	Python bindings can be found at L<http://code.google.com/p/pyev/>. It
		2945	seems to be quite complete and well-documented. Note, however, that the
		2946	patch they require for libev is outright dangerous as it breaks the ABI
		2947	for everybody else, and therefore, should never be applied in an installed
		2948	libev (if python requires an incompatible ABI then it needs to embed
		2949	libev).
		2950
2613	=item Ruby	2951	=item Ruby
2614		2952
2615	Tony Arcieri has written a ruby extension that offers access to a subset	2953	Tony Arcieri has written a ruby extension that offers access to a subset
2616	of the libev API and adds filehandle abstractions, asynchronous DNS and	2954	of the libev API and adds file handle abstractions, asynchronous DNS and
2617	more on top of it. It can be found via gem servers. Its homepage is at	2955	more on top of it. It can be found via gem servers. Its homepage is at
2618	L<http://rev.rubyforge.org/>.	2956	L<http://rev.rubyforge.org/>.
2619		2957
2620	=item D	2958	=item D
2621		2959
2622	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to	2960	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
2623	be found at L<http://git.llucax.com.ar/?p=software/ev.d.git;a=summary>.	2961	be found at L<http://proj.llucax.com.ar/wiki/evd>.
2624		2962
2625	=back	2963	=back
2626		2964
2627		2965
2628	=head1 MACRO MAGIC	2966	=head1 MACRO MAGIC
2629		2967
2630	Libev can be compiled with a variety of options, the most fundamantal	2968	Libev can be compiled with a variety of options, the most fundamental
2631	of which is C<EV_MULTIPLICITY>. This option determines whether (most)	2969	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
2632	functions and callbacks have an initial C<struct ev_loop *> argument.	2970	functions and callbacks have an initial C<struct ev_loop *> argument.
2633		2971
2634	To make it easier to write programs that cope with either variant, the	2972	To make it easier to write programs that cope with either variant, the
2635	following macros are defined:	2973	following macros are defined:
…		…
2640		2978
2641	This provides the loop I<argument> for functions, if one is required ("ev	2979	This provides the loop I<argument> for functions, if one is required ("ev
2642	loop argument"). The C<EV_A> form is used when this is the sole argument,	2980	loop argument"). The C<EV_A> form is used when this is the sole argument,
2643	C<EV_A_> is used when other arguments are following. Example:	2981	C<EV_A_> is used when other arguments are following. Example:
2644		2982
2645	ev_unref (EV_A);	2983	ev_unref (EV_A);
2646	ev_timer_add (EV_A_ watcher);	2984	ev_timer_add (EV_A_ watcher);
2647	ev_loop (EV_A_ 0);	2985	ev_loop (EV_A_ 0);
2648		2986
2649	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,	2987	It assumes the variable C<loop> of type C<struct ev_loop *> is in scope,
2650	which is often provided by the following macro.	2988	which is often provided by the following macro.
2651		2989
2652	=item C<EV_P>, C<EV_P_>	2990	=item C<EV_P>, C<EV_P_>
2653		2991
2654	This provides the loop I<parameter> for functions, if one is required ("ev	2992	This provides the loop I<parameter> for functions, if one is required ("ev
2655	loop parameter"). The C<EV_P> form is used when this is the sole parameter,	2993	loop parameter"). The C<EV_P> form is used when this is the sole parameter,
2656	C<EV_P_> is used when other parameters are following. Example:	2994	C<EV_P_> is used when other parameters are following. Example:
2657		2995
2658	// this is how ev_unref is being declared	2996	// this is how ev_unref is being declared
2659	static void ev_unref (EV_P);	2997	static void ev_unref (EV_P);
2660		2998
2661	// this is how you can declare your typical callback	2999	// this is how you can declare your typical callback
2662	static void cb (EV_P_ ev_timer *w, int revents)	3000	static void cb (EV_P_ ev_timer *w, int revents)
2663		3001
2664	It declares a parameter C<loop> of type C<struct ev_loop *>, quite	3002	It declares a parameter C<loop> of type C<struct ev_loop *>, quite
2665	suitable for use with C<EV_A>.	3003	suitable for use with C<EV_A>.
2666		3004
2667	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	3005	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
…		…
2683		3021
2684	Example: Declare and initialise a check watcher, utilising the above	3022	Example: Declare and initialise a check watcher, utilising the above
2685	macros so it will work regardless of whether multiple loops are supported	3023	macros so it will work regardless of whether multiple loops are supported
2686	or not.	3024	or not.
2687		3025
2688	static void	3026	static void
2689	check_cb (EV_P_ ev_timer *w, int revents)	3027	check_cb (EV_P_ ev_timer *w, int revents)
2690	{	3028	{
2691	ev_check_stop (EV_A_ w);	3029	ev_check_stop (EV_A_ w);
2692	}	3030	}
2693		3031
2694	ev_check check;	3032	ev_check check;
2695	ev_check_init (&check, check_cb);	3033	ev_check_init (&check, check_cb);
2696	ev_check_start (EV_DEFAULT_ &check);	3034	ev_check_start (EV_DEFAULT_ &check);
2697	ev_loop (EV_DEFAULT_ 0);	3035	ev_loop (EV_DEFAULT_ 0);
2698		3036
2699	=head1 EMBEDDING	3037	=head1 EMBEDDING
2700		3038
2701	Libev can (and often is) directly embedded into host	3039	Libev can (and often is) directly embedded into host
2702	applications. Examples of applications that embed it include the Deliantra	3040	applications. Examples of applications that embed it include the Deliantra
…		…
2709	libev somewhere in your source tree).	3047	libev somewhere in your source tree).
2710		3048
2711	=head2 FILESETS	3049	=head2 FILESETS
2712		3050
2713	Depending on what features you need you need to include one or more sets of files	3051	Depending on what features you need you need to include one or more sets of files
2714	in your app.	3052	in your application.
2715		3053
2716	=head3 CORE EVENT LOOP	3054	=head3 CORE EVENT LOOP
2717		3055
2718	To include only the libev core (all the C<ev_*> functions), with manual	3056	To include only the libev core (all the C<ev_*> functions), with manual
2719	configuration (no autoconf):	3057	configuration (no autoconf):
2720		3058
2721	#define EV_STANDALONE 1	3059	#define EV_STANDALONE 1
2722	#include "ev.c"	3060	#include "ev.c"
2723		3061
2724	This will automatically include F<ev.h>, too, and should be done in a	3062	This will automatically include F<ev.h>, too, and should be done in a
2725	single C source file only to provide the function implementations. To use	3063	single C source file only to provide the function implementations. To use
2726	it, do the same for F<ev.h> in all files wishing to use this API (best	3064	it, do the same for F<ev.h> in all files wishing to use this API (best
2727	done by writing a wrapper around F<ev.h> that you can include instead and	3065	done by writing a wrapper around F<ev.h> that you can include instead and
2728	where you can put other configuration options):	3066	where you can put other configuration options):
2729		3067
2730	#define EV_STANDALONE 1	3068	#define EV_STANDALONE 1
2731	#include "ev.h"	3069	#include "ev.h"
2732		3070
2733	Both header files and implementation files can be compiled with a C++	3071	Both header files and implementation files can be compiled with a C++
2734	compiler (at least, thats a stated goal, and breakage will be treated	3072	compiler (at least, thats a stated goal, and breakage will be treated
2735	as a bug).	3073	as a bug).
2736		3074
2737	You need the following files in your source tree, or in a directory	3075	You need the following files in your source tree, or in a directory
2738	in your include path (e.g. in libev/ when using -Ilibev):	3076	in your include path (e.g. in libev/ when using -Ilibev):
2739		3077
2740	ev.h	3078	ev.h
2741	ev.c	3079	ev.c
2742	ev_vars.h	3080	ev_vars.h
2743	ev_wrap.h	3081	ev_wrap.h
2744		3082
2745	ev_win32.c required on win32 platforms only	3083	ev_win32.c required on win32 platforms only
2746		3084
2747	ev_select.c only when select backend is enabled (which is enabled by default)	3085	ev_select.c only when select backend is enabled (which is enabled by default)
2748	ev_poll.c only when poll backend is enabled (disabled by default)	3086	ev_poll.c only when poll backend is enabled (disabled by default)
2749	ev_epoll.c only when the epoll backend is enabled (disabled by default)	3087	ev_epoll.c only when the epoll backend is enabled (disabled by default)
2750	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	3088	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2751	ev_port.c only when the solaris port backend is enabled (disabled by default)	3089	ev_port.c only when the solaris port backend is enabled (disabled by default)
2752		3090
2753	F<ev.c> includes the backend files directly when enabled, so you only need	3091	F<ev.c> includes the backend files directly when enabled, so you only need
2754	to compile this single file.	3092	to compile this single file.
2755		3093
2756	=head3 LIBEVENT COMPATIBILITY API	3094	=head3 LIBEVENT COMPATIBILITY API
2757		3095
2758	To include the libevent compatibility API, also include:	3096	To include the libevent compatibility API, also include:
2759		3097
2760	#include "event.c"	3098	#include "event.c"
2761		3099
2762	in the file including F<ev.c>, and:	3100	in the file including F<ev.c>, and:
2763		3101
2764	#include "event.h"	3102	#include "event.h"
2765		3103
2766	in the files that want to use the libevent API. This also includes F<ev.h>.	3104	in the files that want to use the libevent API. This also includes F<ev.h>.
2767		3105
2768	You need the following additional files for this:	3106	You need the following additional files for this:
2769		3107
2770	event.h	3108	event.h
2771	event.c	3109	event.c
2772		3110
2773	=head3 AUTOCONF SUPPORT	3111	=head3 AUTOCONF SUPPORT
2774		3112
2775	Instead of using C<EV_STANDALONE=1> and providing your config in	3113	Instead of using C<EV_STANDALONE=1> and providing your configuration in
2776	whatever way you want, you can also C<m4_include([libev.m4])> in your	3114	whatever way you want, you can also C<m4_include([libev.m4])> in your
2777	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then	3115	F<configure.ac> and leave C<EV_STANDALONE> undefined. F<ev.c> will then
2778	include F<config.h> and configure itself accordingly.	3116	include F<config.h> and configure itself accordingly.
2779		3117
2780	For this of course you need the m4 file:	3118	For this of course you need the m4 file:
2781		3119
2782	libev.m4	3120	libev.m4
2783		3121
2784	=head2 PREPROCESSOR SYMBOLS/MACROS	3122	=head2 PREPROCESSOR SYMBOLS/MACROS
2785		3123
2786	Libev can be configured via a variety of preprocessor symbols you have to	3124	Libev can be configured via a variety of preprocessor symbols you have to
2787	define before including any of its files. The default in the absense of	3125	define before including any of its files. The default in the absence of
2788	autoconf is noted for every option.	3126	autoconf is documented for every option.
2789		3127
2790	=over 4	3128	=over 4
2791		3129
2792	=item EV_STANDALONE	3130	=item EV_STANDALONE
2793		3131
…		…
2798	F<event.h> that are not directly supported by the libev core alone.	3136	F<event.h> that are not directly supported by the libev core alone.
2799		3137
2800	=item EV_USE_MONOTONIC	3138	=item EV_USE_MONOTONIC
2801		3139
2802	If defined to be C<1>, libev will try to detect the availability of the	3140	If defined to be C<1>, libev will try to detect the availability of the
2803	monotonic clock option at both compiletime and runtime. Otherwise no use	3141	monotonic clock option at both compile time and runtime. Otherwise no use
2804	of the monotonic clock option will be attempted. If you enable this, you	3142	of the monotonic clock option will be attempted. If you enable this, you
2805	usually have to link against librt or something similar. Enabling it when	3143	usually have to link against librt or something similar. Enabling it when
2806	the functionality isn't available is safe, though, although you have	3144	the functionality isn't available is safe, though, although you have
2807	to make sure you link against any libraries where the C<clock_gettime>	3145	to make sure you link against any libraries where the C<clock_gettime>
2808	function is hiding in (often F<-lrt>).	3146	function is hiding in (often F<-lrt>).
2809		3147
2810	=item EV_USE_REALTIME	3148	=item EV_USE_REALTIME
2811		3149
2812	If defined to be C<1>, libev will try to detect the availability of the	3150	If defined to be C<1>, libev will try to detect the availability of the
2813	realtime clock option at compiletime (and assume its availability at	3151	real-time clock option at compile time (and assume its availability at
2814	runtime if successful). Otherwise no use of the realtime clock option will	3152	runtime if successful). Otherwise no use of the real-time clock option will
2815	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	3153	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2816	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the	3154	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2817	note about libraries in the description of C<EV_USE_MONOTONIC>, though.	3155	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
2818		3156
2819	=item EV_USE_NANOSLEEP	3157	=item EV_USE_NANOSLEEP
…		…
2830	2.7 or newer, otherwise disabled.	3168	2.7 or newer, otherwise disabled.
2831		3169
2832	=item EV_USE_SELECT	3170	=item EV_USE_SELECT
2833		3171
2834	If undefined or defined to be C<1>, libev will compile in support for the	3172	If undefined or defined to be C<1>, libev will compile in support for the
2835	C<select>(2) backend. No attempt at autodetection will be done: if no	3173	C<select>(2) backend. No attempt at auto-detection will be done: if no
2836	other method takes over, select will be it. Otherwise the select backend	3174	other method takes over, select will be it. Otherwise the select backend
2837	will not be compiled in.	3175	will not be compiled in.
2838		3176
2839	=item EV_SELECT_USE_FD_SET	3177	=item EV_SELECT_USE_FD_SET
2840		3178
2841	If defined to C<1>, then the select backend will use the system C<fd_set>	3179	If defined to C<1>, then the select backend will use the system C<fd_set>
2842	structure. This is useful if libev doesn't compile due to a missing	3180	structure. This is useful if libev doesn't compile due to a missing
2843	C<NFDBITS> or C<fd_mask> definition or it misguesses the bitset layout on	3181	C<NFDBITS> or C<fd_mask> definition or it mis-guesses the bitset layout on
2844	exotic systems. This usually limits the range of file descriptors to some	3182	exotic systems. This usually limits the range of file descriptors to some
2845	low limit such as 1024 or might have other limitations (winsocket only	3183	low limit such as 1024 or might have other limitations (winsocket only
2846	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might	3184	allows 64 sockets). The C<FD_SETSIZE> macro, set before compilation, might
2847	influence the size of the C<fd_set> used.	3185	influence the size of the C<fd_set> used.
2848		3186
…		…
2897	otherwise another method will be used as fallback. This is the preferred	3235	otherwise another method will be used as fallback. This is the preferred
2898	backend for Solaris 10 systems.	3236	backend for Solaris 10 systems.
2899		3237
2900	=item EV_USE_DEVPOLL	3238	=item EV_USE_DEVPOLL
2901		3239
2902	reserved for future expansion, works like the USE symbols above.	3240	Reserved for future expansion, works like the USE symbols above.
2903		3241
2904	=item EV_USE_INOTIFY	3242	=item EV_USE_INOTIFY
2905		3243
2906	If defined to be C<1>, libev will compile in support for the Linux inotify	3244	If defined to be C<1>, libev will compile in support for the Linux inotify
2907	interface to speed up C<ev_stat> watchers. Its actual availability will	3245	interface to speed up C<ev_stat> watchers. Its actual availability will
…		…
2914	access is atomic with respect to other threads or signal contexts. No such	3252	access is atomic with respect to other threads or signal contexts. No such
2915	type is easily found in the C language, so you can provide your own type	3253	type is easily found in the C language, so you can provide your own type
2916	that you know is safe for your purposes. It is used both for signal handler "locking"	3254	that you know is safe for your purposes. It is used both for signal handler "locking"
2917	as well as for signal and thread safety in C<ev_async> watchers.	3255	as well as for signal and thread safety in C<ev_async> watchers.
2918		3256
2919	In the absense of this define, libev will use C<sig_atomic_t volatile>	3257	In the absence of this define, libev will use C<sig_atomic_t volatile>
2920	(from F<signal.h>), which is usually good enough on most platforms.	3258	(from F<signal.h>), which is usually good enough on most platforms.
2921		3259
2922	=item EV_H	3260	=item EV_H
2923		3261
2924	The name of the F<ev.h> header file used to include it. The default if	3262	The name of the F<ev.h> header file used to include it. The default if
…		…
2963	When doing priority-based operations, libev usually has to linearly search	3301	When doing priority-based operations, libev usually has to linearly search
2964	all the priorities, so having many of them (hundreds) uses a lot of space	3302	all the priorities, so having many of them (hundreds) uses a lot of space
2965	and time, so using the defaults of five priorities (-2 .. +2) is usually	3303	and time, so using the defaults of five priorities (-2 .. +2) is usually
2966	fine.	3304	fine.
2967		3305
2968	If your embedding app does not need any priorities, defining these both to	3306	If your embedding application does not need any priorities, defining these
2969	C<0> will save some memory and cpu.	3307	both to C<0> will save some memory and CPU.
2970		3308
2971	=item EV_PERIODIC_ENABLE	3309	=item EV_PERIODIC_ENABLE
2972		3310
2973	If undefined or defined to be C<1>, then periodic timers are supported. If	3311	If undefined or defined to be C<1>, then periodic timers are supported. If
2974	defined to be C<0>, then they are not. Disabling them saves a few kB of	3312	defined to be C<0>, then they are not. Disabling them saves a few kB of
…		…
2981	code.	3319	code.
2982		3320
2983	=item EV_EMBED_ENABLE	3321	=item EV_EMBED_ENABLE
2984		3322
2985	If undefined or defined to be C<1>, then embed watchers are supported. If	3323	If undefined or defined to be C<1>, then embed watchers are supported. If
2986	defined to be C<0>, then they are not.	3324	defined to be C<0>, then they are not. Embed watchers rely on most other
		3325	watcher types, which therefore must not be disabled.
2987		3326
2988	=item EV_STAT_ENABLE	3327	=item EV_STAT_ENABLE
2989		3328
2990	If undefined or defined to be C<1>, then stat watchers are supported. If	3329	If undefined or defined to be C<1>, then stat watchers are supported. If
2991	defined to be C<0>, then they are not.	3330	defined to be C<0>, then they are not.
…		…
3002		3341
3003	=item EV_MINIMAL	3342	=item EV_MINIMAL
3004		3343
3005	If you need to shave off some kilobytes of code at the expense of some	3344	If you need to shave off some kilobytes of code at the expense of some
3006	speed, define this symbol to C<1>. Currently this is used to override some	3345	speed, define this symbol to C<1>. Currently this is used to override some
3007	inlining decisions, saves roughly 30% codesize of amd64. It also selects a	3346	inlining decisions, saves roughly 30% code size on amd64. It also selects a
3008	much smaller 2-heap for timer management over the default 4-heap.	3347	much smaller 2-heap for timer management over the default 4-heap.
3009		3348
3010	=item EV_PID_HASHSIZE	3349	=item EV_PID_HASHSIZE
3011		3350
3012	C<ev_child> watchers use a small hash table to distribute workload by	3351	C<ev_child> watchers use a small hash table to distribute workload by
…		…
3023	two).	3362	two).
3024		3363
3025	=item EV_USE_4HEAP	3364	=item EV_USE_4HEAP
3026		3365
3027	Heaps are not very cache-efficient. To improve the cache-efficiency of the	3366	Heaps are not very cache-efficient. To improve the cache-efficiency of the
3028	timer and periodics heap, libev uses a 4-heap when this symbol is defined	3367	timer and periodics heaps, libev uses a 4-heap when this symbol is defined
3029	to C<1>. The 4-heap uses more complicated (longer) code but has	3368	to C<1>. The 4-heap uses more complicated (longer) code but has noticeably
3030	noticably faster performance with many (thousands) of watchers.	3369	faster performance with many (thousands) of watchers.
3031		3370
3032	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>	3371	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
3033	(disabled).	3372	(disabled).
3034		3373
3035	=item EV_HEAP_CACHE_AT	3374	=item EV_HEAP_CACHE_AT
3036		3375
3037	Heaps are not very cache-efficient. To improve the cache-efficiency of the	3376	Heaps are not very cache-efficient. To improve the cache-efficiency of the
3038	timer and periodics heap, libev can cache the timestamp (I<at>) within	3377	timer and periodics heaps, libev can cache the timestamp (I<at>) within
3039	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),	3378	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
3040	which uses 8-12 bytes more per watcher and a few hundred bytes more code,	3379	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
3041	but avoids random read accesses on heap changes. This improves performance	3380	but avoids random read accesses on heap changes. This improves performance
3042	noticably with with many (hundreds) of watchers.	3381	noticeably with many (hundreds) of watchers.
3043		3382
3044	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>	3383	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
3045	(disabled).	3384	(disabled).
3046		3385
3047	=item EV_VERIFY	3386	=item EV_VERIFY
…		…
3053	called once per loop, which can slow down libev. If set to C<3>, then the	3392	called once per loop, which can slow down libev. If set to C<3>, then the
3054	verification code will be called very frequently, which will slow down	3393	verification code will be called very frequently, which will slow down
3055	libev considerably.	3394	libev considerably.
3056		3395
3057	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be	3396	The default is C<1>, unless C<EV_MINIMAL> is set, in which case it will be
3058	C<0.>	3397	C<0>.
3059		3398
3060	=item EV_COMMON	3399	=item EV_COMMON
3061		3400
3062	By default, all watchers have a C<void *data> member. By redefining	3401	By default, all watchers have a C<void *data> member. By redefining
3063	this macro to a something else you can include more and other types of	3402	this macro to a something else you can include more and other types of
3064	members. You have to define it each time you include one of the files,	3403	members. You have to define it each time you include one of the files,
3065	though, and it must be identical each time.	3404	though, and it must be identical each time.
3066		3405
3067	For example, the perl EV module uses something like this:	3406	For example, the perl EV module uses something like this:
3068		3407
3069	#define EV_COMMON \	3408	#define EV_COMMON \
3070	SV *self; /* contains this struct */ \	3409	SV *self; /* contains this struct */ \
3071	SV *cb_sv, *fh /* note no trailing ";" */	3410	SV *cb_sv, *fh /* note no trailing ";" */
3072		3411
3073	=item EV_CB_DECLARE (type)	3412	=item EV_CB_DECLARE (type)
3074		3413
3075	=item EV_CB_INVOKE (watcher, revents)	3414	=item EV_CB_INVOKE (watcher, revents)
3076		3415
…		…
3081	definition and a statement, respectively. See the F<ev.h> header file for	3420	definition and a statement, respectively. See the F<ev.h> header file for
3082	their default definitions. One possible use for overriding these is to	3421	their default definitions. One possible use for overriding these is to
3083	avoid the C<struct ev_loop *> as first argument in all cases, or to use	3422	avoid the C<struct ev_loop *> as first argument in all cases, or to use
3084	method calls instead of plain function calls in C++.	3423	method calls instead of plain function calls in C++.
3085		3424
		3425	=back
		3426
3086	=head2 EXPORTED API SYMBOLS	3427	=head2 EXPORTED API SYMBOLS
3087		3428
3088	If you need to re-export the API (e.g. via a dll) and you need a list of	3429	If you need to re-export the API (e.g. via a DLL) and you need a list of
3089	exported symbols, you can use the provided F<Symbol.*> files which list	3430	exported symbols, you can use the provided F<Symbol.*> files which list
3090	all public symbols, one per line:	3431	all public symbols, one per line:
3091		3432
3092	Symbols.ev for libev proper	3433	Symbols.ev for libev proper
3093	Symbols.event for the libevent emulation	3434	Symbols.event for the libevent emulation
3094		3435
3095	This can also be used to rename all public symbols to avoid clashes with	3436	This can also be used to rename all public symbols to avoid clashes with
3096	multiple versions of libev linked together (which is obviously bad in	3437	multiple versions of libev linked together (which is obviously bad in
3097	itself, but sometimes it is inconvinient to avoid this).	3438	itself, but sometimes it is inconvenient to avoid this).
3098		3439
3099	A sed command like this will create wrapper C<#define>'s that you need to	3440	A sed command like this will create wrapper C<#define>'s that you need to
3100	include before including F<ev.h>:	3441	include before including F<ev.h>:
3101		3442
3102	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h	3443	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
…		…
3119	file.	3460	file.
3120		3461
3121	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	3462	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
3122	that everybody includes and which overrides some configure choices:	3463	that everybody includes and which overrides some configure choices:
3123		3464
3124	#define EV_MINIMAL 1	3465	#define EV_MINIMAL 1
3125	#define EV_USE_POLL 0	3466	#define EV_USE_POLL 0
3126	#define EV_MULTIPLICITY 0	3467	#define EV_MULTIPLICITY 0
3127	#define EV_PERIODIC_ENABLE 0	3468	#define EV_PERIODIC_ENABLE 0
3128	#define EV_STAT_ENABLE 0	3469	#define EV_STAT_ENABLE 0
3129	#define EV_FORK_ENABLE 0	3470	#define EV_FORK_ENABLE 0
3130	#define EV_CONFIG_H <config.h>	3471	#define EV_CONFIG_H <config.h>
3131	#define EV_MINPRI 0	3472	#define EV_MINPRI 0
3132	#define EV_MAXPRI 0	3473	#define EV_MAXPRI 0
3133		3474
3134	#include "ev++.h"	3475	#include "ev++.h"
3135		3476
3136	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	3477	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
3137		3478
3138	#include "ev_cpp.h"	3479	#include "ev_cpp.h"
3139	#include "ev.c"	3480	#include "ev.c"
3140		3481
		3482	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
3141		3483
3142	=head1 THREADS AND COROUTINES	3484	=head2 THREADS AND COROUTINES
3143		3485
3144	=head2 THREADS	3486	=head3 THREADS
3145		3487
3146	Libev itself is completely threadsafe, but it uses no locking. This	3488	All libev functions are reentrant and thread-safe unless explicitly
		3489	documented otherwise, but libev implements no locking itself. This means
3147	means that you can use as many loops as you want in parallel, as long as	3490	that you can use as many loops as you want in parallel, as long as there
3148	only one thread ever calls into one libev function with the same loop	3491	are no concurrent calls into any libev function with the same loop
3149	parameter.	3492	parameter (C<ev_default_*> calls have an implicit default loop parameter,
		3493	of course): libev guarantees that different event loops share no data
		3494	structures that need any locking.
3150		3495
3151	Or put differently: calls with different loop parameters can be done in	3496	Or to put it differently: calls with different loop parameters can be done
3152	parallel from multiple threads, calls with the same loop parameter must be	3497	concurrently from multiple threads, calls with the same loop parameter
3153	done serially (but can be done from different threads, as long as only one	3498	must be done serially (but can be done from different threads, as long as
3154	thread ever is inside a call at any point in time, e.g. by using a mutex	3499	only one thread ever is inside a call at any point in time, e.g. by using
3155	per loop).	3500	a mutex per loop).
3156		3501
3157	If you want to know which design is best for your problem, then I cannot	3502	Specifically to support threads (and signal handlers), libev implements
		3503	so-called C<ev_async> watchers, which allow some limited form of
		3504	concurrency on the same event loop, namely waking it up "from the
		3505	outside".
		3506
		3507	If you want to know which design (one loop, locking, or multiple loops
		3508	without or something else still) is best for your problem, then I cannot
3158	help you but by giving some generic advice:	3509	help you, but here is some generic advice:
3159		3510
3160	=over 4	3511	=over 4
3161		3512
3162	=item * most applications have a main thread: use the default libev loop	3513	=item * most applications have a main thread: use the default libev loop
3163	in that thread, or create a seperate thread running only the default loop.	3514	in that thread, or create a separate thread running only the default loop.
3164		3515
3165	This helps integrating other libraries or software modules that use libev	3516	This helps integrating other libraries or software modules that use libev
3166	themselves and don't care/know about threading.	3517	themselves and don't care/know about threading.
3167		3518
3168	=item * one loop per thread is usually a good model.	3519	=item * one loop per thread is usually a good model.
3169		3520
3170	Doing this is almost never wrong, sometimes a better-performance model	3521	Doing this is almost never wrong, sometimes a better-performance model
3171	exists, but it is always a good start.	3522	exists, but it is always a good start.
3172		3523
3173	=item * other models exist, such as the leader/follower pattern, where one	3524	=item * other models exist, such as the leader/follower pattern, where one
3174	loop is handed through multiple threads in a kind of round-robbin fashion.	3525	loop is handed through multiple threads in a kind of round-robin fashion.
3175		3526
3176	Chosing a model is hard - look around, learn, know that usually you cna do	3527	Choosing a model is hard - look around, learn, know that usually you can do
3177	better than you currently do :-)	3528	better than you currently do :-)
3178		3529
3179	=item * often you need to talk to some other thread which blocks in the	3530	=item * often you need to talk to some other thread which blocks in the
		3531	event loop.
		3532
3180	event loop - C<ev_async> watchers can be used to wake them up from other	3533	C<ev_async> watchers can be used to wake them up from other threads safely
3181	threads safely (or from signal contexts...).	3534	(or from signal contexts...).
		3535
		3536	An example use would be to communicate signals or other events that only
		3537	work in the default loop by registering the signal watcher with the
		3538	default loop and triggering an C<ev_async> watcher from the default loop
		3539	watcher callback into the event loop interested in the signal.
3182		3540
3183	=back	3541	=back
3184		3542
3185	=head2 COROUTINES	3543	=head3 COROUTINES
3186		3544
3187	Libev is much more accomodating to coroutines ("cooperative threads"):	3545	Libev is very accommodating to coroutines ("cooperative threads"):
3188	libev fully supports nesting calls to it's functions from different	3546	libev fully supports nesting calls to its functions from different
3189	coroutines (e.g. you can call C<ev_loop> on the same loop from two	3547	coroutines (e.g. you can call C<ev_loop> on the same loop from two
3190	different coroutines and switch freely between both coroutines running the	3548	different coroutines, and switch freely between both coroutines running the
3191	loop, as long as you don't confuse yourself). The only exception is that	3549	loop, as long as you don't confuse yourself). The only exception is that
3192	you must not do this from C<ev_periodic> reschedule callbacks.	3550	you must not do this from C<ev_periodic> reschedule callbacks.
3193		3551
3194	Care has been invested into making sure that libev does not keep local	3552	Care has been taken to ensure that libev does not keep local state inside
3195	state inside C<ev_loop>, and other calls do not usually allow coroutine	3553	C<ev_loop>, and other calls do not usually allow for coroutine switches as
3196	switches.	3554	they do not clal any callbacks.
3197		3555
		3556	=head2 COMPILER WARNINGS
3198		3557
3199	=head1 COMPLEXITIES	3558	Depending on your compiler and compiler settings, you might get no or a
		3559	lot of warnings when compiling libev code. Some people are apparently
		3560	scared by this.
3200		3561
3201	In this section the complexities of (many of) the algorithms used inside	3562	However, these are unavoidable for many reasons. For one, each compiler
3202	libev will be explained. For complexity discussions about backends see the	3563	has different warnings, and each user has different tastes regarding
3203	documentation for C<ev_default_init>.	3564	warning options. "Warn-free" code therefore cannot be a goal except when
		3565	targeting a specific compiler and compiler-version.
3204		3566
3205	All of the following are about amortised time: If an array needs to be	3567	Another reason is that some compiler warnings require elaborate
3206	extended, libev needs to realloc and move the whole array, but this	3568	workarounds, or other changes to the code that make it less clear and less
3207	happens asymptotically never with higher number of elements, so O(1) might	3569	maintainable.
3208	mean it might do a lengthy realloc operation in rare cases, but on average
3209	it is much faster and asymptotically approaches constant time.
3210		3570
3211	=over 4	3571	And of course, some compiler warnings are just plain stupid, or simply
		3572	wrong (because they don't actually warn about the condition their message
		3573	seems to warn about). For example, certain older gcc versions had some
		3574	warnings that resulted an extreme number of false positives. These have
		3575	been fixed, but some people still insist on making code warn-free with
		3576	such buggy versions.
3212		3577
3213	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	3578	While libev is written to generate as few warnings as possible,
		3579	"warn-free" code is not a goal, and it is recommended not to build libev
		3580	with any compiler warnings enabled unless you are prepared to cope with
		3581	them (e.g. by ignoring them). Remember that warnings are just that:
		3582	warnings, not errors, or proof of bugs.
3214		3583
3215	This means that, when you have a watcher that triggers in one hour and
3216	there are 100 watchers that would trigger before that then inserting will
3217	have to skip roughly seven (C<ld 100>) of these watchers.
3218		3584
3219	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)	3585	=head2 VALGRIND
3220		3586
3221	That means that changing a timer costs less than removing/adding them	3587	Valgrind has a special section here because it is a popular tool that is
3222	as only the relative motion in the event queue has to be paid for.	3588	highly useful. Unfortunately, valgrind reports are very hard to interpret.
3223		3589
3224	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)	3590	If you think you found a bug (memory leak, uninitialised data access etc.)
		3591	in libev, then check twice: If valgrind reports something like:
3225		3592
3226	These just add the watcher into an array or at the head of a list.	3593	==2274== definitely lost: 0 bytes in 0 blocks.
		3594	==2274== possibly lost: 0 bytes in 0 blocks.
		3595	==2274== still reachable: 256 bytes in 1 blocks.
3227		3596
3228	=item Stopping check/prepare/idle/fork/async watchers: O(1)	3597	Then there is no memory leak, just as memory accounted to global variables
		3598	is not a memleak - the memory is still being refernced, and didn't leak.
3229		3599
3230	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	3600	Similarly, under some circumstances, valgrind might report kernel bugs
		3601	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3602	although an acceptable workaround has been found here), or it might be
		3603	confused.
3231		3604
3232	These watchers are stored in lists then need to be walked to find the	3605	Keep in mind that valgrind is a very good tool, but only a tool. Don't
3233	correct watcher to remove. The lists are usually short (you don't usually	3606	make it into some kind of religion.
3234	have many watchers waiting for the same fd or signal).
3235		3607
3236	=item Finding the next timer in each loop iteration: O(1)	3608	If you are unsure about something, feel free to contact the mailing list
		3609	with the full valgrind report and an explanation on why you think this
		3610	is a bug in libev (best check the archives, too :). However, don't be
		3611	annoyed when you get a brisk "this is no bug" answer and take the chance
		3612	of learning how to interpret valgrind properly.
3237		3613
3238	By virtue of using a binary or 4-heap, the next timer is always found at a	3614	If you need, for some reason, empty reports from valgrind for your project
3239	fixed position in the storage array.	3615	I suggest using suppression lists.
3240		3616
3241	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
3242		3617
3243	A change means an I/O watcher gets started or stopped, which requires	3618	=head1 PORTABILITY NOTES
3244	libev to recalculate its status (and possibly tell the kernel, depending
3245	on backend and wether C<ev_io_set> was used).
3246		3619
3247	=item Activating one watcher (putting it into the pending state): O(1)	3620	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
3248
3249	=item Priority handling: O(number_of_priorities)
3250
3251	Priorities are implemented by allocating some space for each
3252	priority. When doing priority-based operations, libev usually has to
3253	linearly search all the priorities, but starting/stopping and activating
3254	watchers becomes O(1) w.r.t. priority handling.
3255
3256	=item Sending an ev_async: O(1)
3257
3258	=item Processing ev_async_send: O(number_of_async_watchers)
3259
3260	=item Processing signals: O(max_signal_number)
3261
3262	Sending involves a syscall I<iff> there were no other C<ev_async_send>
3263	calls in the current loop iteration. Checking for async and signal events
3264	involves iterating over all running async watchers or all signal numbers.
3265
3266	=back
3267
3268
3269	=head1 Win32 platform limitations and workarounds
3270		3621
3271	Win32 doesn't support any of the standards (e.g. POSIX) that libev	3622	Win32 doesn't support any of the standards (e.g. POSIX) that libev
3272	requires, and its I/O model is fundamentally incompatible with the POSIX	3623	requires, and its I/O model is fundamentally incompatible with the POSIX
3273	model. Libev still offers limited functionality on this platform in	3624	model. Libev still offers limited functionality on this platform in
3274	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	3625	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
…		…
3281	way (note also that glib is the slowest event library known to man).	3632	way (note also that glib is the slowest event library known to man).
3282		3633
3283	There is no supported compilation method available on windows except	3634	There is no supported compilation method available on windows except
3284	embedding it into other applications.	3635	embedding it into other applications.
3285		3636
		3637	Not a libev limitation but worth mentioning: windows apparently doesn't
		3638	accept large writes: instead of resulting in a partial write, windows will
		3639	either accept everything or return C<ENOBUFS> if the buffer is too large,
		3640	so make sure you only write small amounts into your sockets (less than a
		3641	megabyte seems safe, but this apparently depends on the amount of memory
		3642	available).
		3643
3286	Due to the many, low, and arbitrary limits on the win32 platform and	3644	Due to the many, low, and arbitrary limits on the win32 platform and
3287	the abysmal performance of winsockets, using a large number of sockets	3645	the abysmal performance of winsockets, using a large number of sockets
3288	is not recommended (and not reasonable). If your program needs to use	3646	is not recommended (and not reasonable). If your program needs to use
3289	more than a hundred or so sockets, then likely it needs to use a totally	3647	more than a hundred or so sockets, then likely it needs to use a totally
3290	different implementation for windows, as libev offers the POSIX readiness	3648	different implementation for windows, as libev offers the POSIX readiness
3291	notification model, which cannot be implemented efficiently on windows	3649	notification model, which cannot be implemented efficiently on windows
3292	(microsoft monopoly games).	3650	(Microsoft monopoly games).
		3651
		3652	A typical way to use libev under windows is to embed it (see the embedding
		3653	section for details) and use the following F<evwrap.h> header file instead
		3654	of F<ev.h>:
		3655
		3656	#define EV_STANDALONE /* keeps ev from requiring config.h */
		3657	#define EV_SELECT_IS_WINSOCKET 1 /* configure libev for windows select */
		3658
		3659	#include "ev.h"
		3660
		3661	And compile the following F<evwrap.c> file into your project (make sure
		3662	you do I<not> compile the F<ev.c> or any other embedded source files!):
		3663
		3664	#include "evwrap.h"
		3665	#include "ev.c"
3293		3666
3294	=over 4	3667	=over 4
3295		3668
3296	=item The winsocket select function	3669	=item The winsocket select function
3297		3670
3298	The winsocket C<select> function doesn't follow POSIX in that it requires	3671	The winsocket C<select> function doesn't follow POSIX in that it
3299	socket I<handles> and not socket I<file descriptors>. This makes select	3672	requires socket I<handles> and not socket I<file descriptors> (it is
3300	very inefficient, and also requires a mapping from file descriptors	3673	also extremely buggy). This makes select very inefficient, and also
3301	to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,	3674	requires a mapping from file descriptors to socket handles (the Microsoft
3302	C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor	3675	C runtime provides the function C<_open_osfhandle> for this). See the
3303	symbols for more info.	3676	discussion of the C<EV_SELECT_USE_FD_SET>, C<EV_SELECT_IS_WINSOCKET> and
		3677	C<EV_FD_TO_WIN32_HANDLE> preprocessor symbols for more info.
3304		3678
3305	The configuration for a "naked" win32 using the microsoft runtime	3679	The configuration for a "naked" win32 using the Microsoft runtime
3306	libraries and raw winsocket select is:	3680	libraries and raw winsocket select is:
3307		3681
3308	#define EV_USE_SELECT 1	3682	#define EV_USE_SELECT 1
3309	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */	3683	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
3310		3684
3311	Note that winsockets handling of fd sets is O(n), so you can easily get a	3685	Note that winsockets handling of fd sets is O(n), so you can easily get a
3312	complexity in the O(n²) range when using win32.	3686	complexity in the O(n²) range when using win32.
3313		3687
3314	=item Limited number of file descriptors	3688	=item Limited number of file descriptors
3315		3689
3316	Windows has numerous arbitrary (and low) limits on things.	3690	Windows has numerous arbitrary (and low) limits on things.
3317		3691
3318	Early versions of winsocket's select only supported waiting for a maximum	3692	Early versions of winsocket's select only supported waiting for a maximum
3319	of C<64> handles (probably owning to the fact that all windows kernels	3693	of C<64> handles (probably owning to the fact that all windows kernels
3320	can only wait for C<64> things at the same time internally; microsoft	3694	can only wait for C<64> things at the same time internally; Microsoft
3321	recommends spawning a chain of threads and wait for 63 handles and the	3695	recommends spawning a chain of threads and wait for 63 handles and the
3322	previous thread in each. Great).	3696	previous thread in each. Great).
3323		3697
3324	Newer versions support more handles, but you need to define C<FD_SETSIZE>	3698	Newer versions support more handles, but you need to define C<FD_SETSIZE>
3325	to some high number (e.g. C<2048>) before compiling the winsocket select	3699	to some high number (e.g. C<2048>) before compiling the winsocket select
3326	call (which might be in libev or elsewhere, for example, perl does its own	3700	call (which might be in libev or elsewhere, for example, perl does its own
3327	select emulation on windows).	3701	select emulation on windows).
3328		3702
3329	Another limit is the number of file descriptors in the microsoft runtime	3703	Another limit is the number of file descriptors in the Microsoft runtime
3330	libraries, which by default is C<64> (there must be a hidden I<64> fetish	3704	libraries, which by default is C<64> (there must be a hidden I<64> fetish
3331	or something like this inside microsoft). You can increase this by calling	3705	or something like this inside Microsoft). You can increase this by calling
3332	C<_setmaxstdio>, which can increase this limit to C<2048> (another	3706	C<_setmaxstdio>, which can increase this limit to C<2048> (another
3333	arbitrary limit), but is broken in many versions of the microsoft runtime	3707	arbitrary limit), but is broken in many versions of the Microsoft runtime
3334	libraries.	3708	libraries.
3335		3709
3336	This might get you to about C<512> or C<2048> sockets (depending on	3710	This might get you to about C<512> or C<2048> sockets (depending on
3337	windows version and/or the phase of the moon). To get more, you need to	3711	windows version and/or the phase of the moon). To get more, you need to
3338	wrap all I/O functions and provide your own fd management, but the cost of	3712	wrap all I/O functions and provide your own fd management, but the cost of
3339	calling select (O(n²)) will likely make this unworkable.	3713	calling select (O(n²)) will likely make this unworkable.
3340		3714
3341	=back	3715	=back
3342		3716
3343
3344	=head1 PORTABILITY REQUIREMENTS	3717	=head2 PORTABILITY REQUIREMENTS
3345		3718
3346	In addition to a working ISO-C implementation, libev relies on a few	3719	In addition to a working ISO-C implementation and of course the
3347	additional extensions:	3720	backend-specific APIs, libev relies on a few additional extensions:
3348		3721
3349	=over 4	3722	=over 4
3350		3723
		3724	=item C<void (*)(ev_watcher_type *, int revents)> must have compatible
		3725	calling conventions regardless of C<ev_watcher_type *>.
		3726
		3727	Libev assumes not only that all watcher pointers have the same internal
		3728	structure (guaranteed by POSIX but not by ISO C for example), but it also
		3729	assumes that the same (machine) code can be used to call any watcher
		3730	callback: The watcher callbacks have different type signatures, but libev
		3731	calls them using an C<ev_watcher *> internally.
		3732
3351	=item C<sig_atomic_t volatile> must be thread-atomic as well	3733	=item C<sig_atomic_t volatile> must be thread-atomic as well
3352		3734
3353	The type C<sig_atomic_t volatile> (or whatever is defined as	3735	The type C<sig_atomic_t volatile> (or whatever is defined as
3354	C<EV_ATOMIC_T>) must be atomic w.r.t. accesses from different	3736	C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
3355	threads. This is not part of the specification for C<sig_atomic_t>, but is	3737	threads. This is not part of the specification for C<sig_atomic_t>, but is
3356	believed to be sufficiently portable.	3738	believed to be sufficiently portable.
3357		3739
3358	=item C<sigprocmask> must work in a threaded environment	3740	=item C<sigprocmask> must work in a threaded environment
3359		3741
…		…
3368	except the initial one, and run the default loop in the initial thread as	3750	except the initial one, and run the default loop in the initial thread as
3369	well.	3751	well.
3370		3752
3371	=item C<long> must be large enough for common memory allocation sizes	3753	=item C<long> must be large enough for common memory allocation sizes
3372		3754
3373	To improve portability and simplify using libev, libev uses C<long>	3755	To improve portability and simplify its API, libev uses C<long> internally
3374	internally instead of C<size_t> when allocating its data structures. On	3756	instead of C<size_t> when allocating its data structures. On non-POSIX
3375	non-POSIX systems (Microsoft...) this might be unexpectedly low, but	3757	systems (Microsoft...) this might be unexpectedly low, but is still at
3376	is still at least 31 bits everywhere, which is enough for hundreds of	3758	least 31 bits everywhere, which is enough for hundreds of millions of
3377	millions of watchers.	3759	watchers.
3378		3760
3379	=item C<double> must hold a time value in seconds with enough accuracy	3761	=item C<double> must hold a time value in seconds with enough accuracy
3380		3762
3381	The type C<double> is used to represent timestamps. It is required to	3763	The type C<double> is used to represent timestamps. It is required to
3382	have at least 51 bits of mantissa (and 9 bits of exponent), which is good	3764	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
…		…
3386	=back	3768	=back
3387		3769
3388	If you know of other additional requirements drop me a note.	3770	If you know of other additional requirements drop me a note.
3389		3771
3390		3772
3391	=head1 VALGRIND	3773	=head1 ALGORITHMIC COMPLEXITIES
3392		3774
3393	Valgrind has a special section here because it is a popular tool that is	3775	In this section the complexities of (many of) the algorithms used inside
3394	highly useful, but valgrind reports are very hard to interpret.	3776	libev will be documented. For complexity discussions about backends see
		3777	the documentation for C<ev_default_init>.
3395		3778
3396	If you think you found a bug (memory leak, uninitialised data access etc.)	3779	All of the following are about amortised time: If an array needs to be
3397	in libev, then check twice: If valgrind reports something like:	3780	extended, libev needs to realloc and move the whole array, but this
		3781	happens asymptotically rarer with higher number of elements, so O(1) might
		3782	mean that libev does a lengthy realloc operation in rare cases, but on
		3783	average it is much faster and asymptotically approaches constant time.
3398		3784
3399	==2274== definitely lost: 0 bytes in 0 blocks.	3785	=over 4
3400	==2274== possibly lost: 0 bytes in 0 blocks.
3401	==2274== still reachable: 256 bytes in 1 blocks.
3402		3786
3403	then there is no memory leak. Similarly, under some circumstances,	3787	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
3404	valgrind might report kernel bugs as if it were a bug in libev, or it
3405	might be confused (it is a very good tool, but only a tool).
3406		3788
3407	If you are unsure about something, feel free to contact the mailing list	3789	This means that, when you have a watcher that triggers in one hour and
3408	with the full valgrind report and an explanation on why you think this is	3790	there are 100 watchers that would trigger before that, then inserting will
3409	a bug in libev. However, don't be annoyed when you get a brisk "this is	3791	have to skip roughly seven (C<ld 100>) of these watchers.
3410	no bug" answer and take the chance of learning how to interpret valgrind
3411	properly.
3412		3792
3413	If you need, for some reason, empty reports from valgrind for your project	3793	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
3414	I suggest using suppression lists.	3794
		3795	That means that changing a timer costs less than removing/adding them,
		3796	as only the relative motion in the event queue has to be paid for.
		3797
		3798	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
		3799
		3800	These just add the watcher into an array or at the head of a list.
		3801
		3802	=item Stopping check/prepare/idle/fork/async watchers: O(1)
		3803
		3804	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		3805
		3806	These watchers are stored in lists, so they need to be walked to find the
		3807	correct watcher to remove. The lists are usually short (you don't usually
		3808	have many watchers waiting for the same fd or signal: one is typical, two
		3809	is rare).
		3810
		3811	=item Finding the next timer in each loop iteration: O(1)
		3812
		3813	By virtue of using a binary or 4-heap, the next timer is always found at a
		3814	fixed position in the storage array.
		3815
		3816	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		3817
		3818	A change means an I/O watcher gets started or stopped, which requires
		3819	libev to recalculate its status (and possibly tell the kernel, depending
		3820	on backend and whether C<ev_io_set> was used).
		3821
		3822	=item Activating one watcher (putting it into the pending state): O(1)
		3823
		3824	=item Priority handling: O(number_of_priorities)
		3825
		3826	Priorities are implemented by allocating some space for each
		3827	priority. When doing priority-based operations, libev usually has to
		3828	linearly search all the priorities, but starting/stopping and activating
		3829	watchers becomes O(1) with respect to priority handling.
		3830
		3831	=item Sending an ev_async: O(1)
		3832
		3833	=item Processing ev_async_send: O(number_of_async_watchers)
		3834
		3835	=item Processing signals: O(max_signal_number)
		3836
		3837	Sending involves a system call I<iff> there were no other C<ev_async_send>
		3838	calls in the current loop iteration. Checking for async and signal events
		3839	involves iterating over all running async watchers or all signal numbers.
		3840
		3841	=back
3415		3842
3416		3843
3417	=head1 AUTHOR	3844	=head1 AUTHOR
3418		3845
3419	Marc Lehmann <libev@schmorp.de>.	3846	Marc Lehmann <libev@schmorp.de>.

Diff Legend

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