ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.70 by root, Fri Dec 7 19:23:48 2007 UTC vs.
Revision 1.130 by root, Wed Feb 6 18:34:24 2008 UTC

…		…
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	#include <ev.h>	7	#include <ev.h>
8		8
9	=head1 EXAMPLE PROGRAM	9	=head2 EXAMPLE PROGRAM
10		10
11	#include <ev.h>	11	#include <ev.h>
12		12
13	ev_io stdin_watcher;	13	ev_io stdin_watcher;
14	ev_timer timeout_watcher;	14	ev_timer timeout_watcher;
…		…
53	The newest version of this document is also available as a html-formatted	53	The newest version of this document is also available as a html-formatted
54	web page you might find easier to navigate when reading it for the first	54	web page you might find easier to navigate when reading it for the first
55	time: L<http://cvs.schmorp.de/libev/ev.html>.	55	time: L<http://cvs.schmorp.de/libev/ev.html>.
56		56
57	Libev is an event loop: you register interest in certain events (such as a	57	Libev is an event loop: you register interest in certain events (such as a
58	file descriptor being readable or a timeout occuring), and it will manage	58	file descriptor being readable or a timeout occurring), and it will manage
59	these event sources and provide your program with events.	59	these event sources and provide your program with events.
60		60
61	To do this, it must take more or less complete control over your process	61	To do this, it must take more or less complete control over your process
62	(or thread) by executing the I<event loop> handler, and will then	62	(or thread) by executing the I<event loop> handler, and will then
63	communicate events via a callback mechanism.	63	communicate events via a callback mechanism.
…		…
65	You register interest in certain events by registering so-called I<event	65	You register interest in certain events by registering so-called I<event
66	watchers>, which are relatively small C structures you initialise with the	66	watchers>, which are relatively small C structures you initialise with the
67	details of the event, and then hand it over to libev by I<starting> the	67	details of the event, and then hand it over to libev by I<starting> the
68	watcher.	68	watcher.
69		69
70	=head1 FEATURES	70	=head2 FEATURES
71		71
72	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the	72	Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms	73	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74	for file descriptor events (C<ev_io>), the Linux C<inotify> interface	74	for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers	75	(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
…		…
82		82
83	It also is quite fast (see this	83	It also is quite fast (see this
84	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent	84	L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85	for example).	85	for example).
86		86
87	=head1 CONVENTIONS	87	=head2 CONVENTIONS
88		88
89	Libev is very configurable. In this manual the default configuration will	89	Libev is very configurable. In this manual the default configuration will
90	be described, which supports multiple event loops. For more info about	90	be described, which supports multiple event loops. For more info about
91	various configuration options please have a look at B<EMBED> section in	91	various configuration options please have a look at B<EMBED> section in
92	this manual. If libev was configured without support for multiple event	92	this manual. If libev was configured without support for multiple event
93	loops, then all functions taking an initial argument of name C<loop>	93	loops, then all functions taking an initial argument of name C<loop>
94	(which is always of type C<struct ev_loop *>) will not have this argument.	94	(which is always of type C<struct ev_loop *>) will not have this argument.
95		95
96	=head1 TIME REPRESENTATION	96	=head2 TIME REPRESENTATION
97		97
98	Libev represents time as a single floating point number, representing the	98	Libev represents time as a single floating point number, representing the
99	(fractional) number of seconds since the (POSIX) epoch (somewhere near	99	(fractional) number of seconds since the (POSIX) epoch (somewhere near
100	the beginning of 1970, details are complicated, don't ask). This type is	100	the beginning of 1970, details are complicated, don't ask). This type is
101	called C<ev_tstamp>, which is what you should use too. It usually aliases	101	called C<ev_tstamp>, which is what you should use too. It usually aliases
102	to the C<double> type in C, and when you need to do any calculations on	102	to the C<double> type in C, and when you need to do any calculations on
103	it, you should treat it as such.	103	it, you should treat it as some floatingpoint value. Unlike the name
		104	component C<stamp> might indicate, it is also used for time differences
		105	throughout libev.
104		106
105	=head1 GLOBAL FUNCTIONS	107	=head1 GLOBAL FUNCTIONS
106		108
107	These functions can be called anytime, even before initialising the	109	These functions can be called anytime, even before initialising the
108	library in any way.	110	library in any way.
…		…
113		115
114	Returns the current time as libev would use it. Please note that the	116	Returns the current time as libev would use it. Please note that the
115	C<ev_now> function is usually faster and also often returns the timestamp	117	C<ev_now> function is usually faster and also often returns the timestamp
116	you actually want to know.	118	you actually want to know.
117		119
		120	=item ev_sleep (ev_tstamp interval)
		121
		122	Sleep for the given interval: The current thread will be blocked until
		123	either it is interrupted or the given time interval has passed. Basically
		124	this is a subsecond-resolution C<sleep ()>.
		125
118	=item int ev_version_major ()	126	=item int ev_version_major ()
119		127
120	=item int ev_version_minor ()	128	=item int ev_version_minor ()
121		129
122	You can find out the major and minor version numbers of the library	130	You can find out the major and minor ABI version numbers of the library
123	you linked against by calling the functions C<ev_version_major> and	131	you linked against by calling the functions C<ev_version_major> and
124	C<ev_version_minor>. If you want, you can compare against the global	132	C<ev_version_minor>. If you want, you can compare against the global
125	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the	133	symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
126	version of the library your program was compiled against.	134	version of the library your program was compiled against.
127		135
		136	These version numbers refer to the ABI version of the library, not the
		137	release version.
		138
128	Usually, it's a good idea to terminate if the major versions mismatch,	139	Usually, it's a good idea to terminate if the major versions mismatch,
129	as this indicates an incompatible change. Minor versions are usually	140	as this indicates an incompatible change. Minor versions are usually
130	compatible to older versions, so a larger minor version alone is usually	141	compatible to older versions, so a larger minor version alone is usually
131	not a problem.	142	not a problem.
132		143
133	Example: Make sure we haven't accidentally been linked against the wrong	144	Example: Make sure we haven't accidentally been linked against the wrong
134	version.	145	version.
…		…
249	flags. If that is troubling you, check C<ev_backend ()> afterwards).	260	flags. If that is troubling you, check C<ev_backend ()> afterwards).
250		261
251	If you don't know what event loop to use, use the one returned from this	262	If you don't know what event loop to use, use the one returned from this
252	function.	263	function.
253		264
		265	The default loop is the only loop that can handle C<ev_signal> and
		266	C<ev_child> watchers, and to do this, it always registers a handler
		267	for C<SIGCHLD>. If this is a problem for your app you can either
		268	create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
		269	can simply overwrite the C<SIGCHLD> signal handler I<after> calling
		270	C<ev_default_init>.
		271
254	The flags argument can be used to specify special behaviour or specific	272	The flags argument can be used to specify special behaviour or specific
255	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).	273	backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
256		274
257	The following flags are supported:	275	The following flags are supported:
258		276
…		…
295	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	313	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
296		314
297	This is your standard select(2) backend. Not I<completely> standard, as	315	This is your standard select(2) backend. Not I<completely> standard, as
298	libev tries to roll its own fd_set with no limits on the number of fds,	316	libev tries to roll its own fd_set with no limits on the number of fds,
299	but if that fails, expect a fairly low limit on the number of fds when	317	but if that fails, expect a fairly low limit on the number of fds when
300	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	318	using this backend. It doesn't scale too well (O(highest_fd)), but its
301	the fastest backend for a low number of fds.	319	usually the fastest backend for a low number of (low-numbered :) fds.
		320
		321	To get good performance out of this backend you need a high amount of
		322	parallelity (most of the file descriptors should be busy). If you are
		323	writing a server, you should C<accept ()> in a loop to accept as many
		324	connections as possible during one iteration. You might also want to have
		325	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		326	readyness notifications you get per iteration.
302		327
303	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	328	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
304		329
305	And this is your standard poll(2) backend. It's more complicated than	330	And this is your standard poll(2) backend. It's more complicated
306	select, but handles sparse fds better and has no artificial limit on the	331	than select, but handles sparse fds better and has no artificial
307	number of fds you can use (except it will slow down considerably with a	332	limit on the number of fds you can use (except it will slow down
308	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	333	considerably with a lot of inactive fds). It scales similarly to select,
		334	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		335	performance tips.
309		336
310	=item C<EVBACKEND_EPOLL> (value 4, Linux)	337	=item C<EVBACKEND_EPOLL> (value 4, Linux)
311		338
312	For few fds, this backend is a bit little slower than poll and select,	339	For few fds, this backend is a bit little slower than poll and select,
313	but it scales phenomenally better. While poll and select usually scale like	340	but it scales phenomenally better. While poll and select usually scale
314	O(total_fds) where n is the total number of fds (or the highest fd), epoll scales	341	like O(total_fds) where n is the total number of fds (or the highest fd),
315	either O(1) or O(active_fds).	342	epoll scales either O(1) or O(active_fds). The epoll design has a number
		343	of shortcomings, such as silently dropping events in some hard-to-detect
		344	cases and rewiring a syscall per fd change, no fork support and bad
		345	support for dup.
316		346
317	While stopping and starting an I/O watcher in the same iteration will	347	While stopping, setting and starting an I/O watcher in the same iteration
318	result in some caching, there is still a syscall per such incident	348	will result in some caching, there is still a syscall per such incident
319	(because the fd could point to a different file description now), so its	349	(because the fd could point to a different file description now), so its
320	best to avoid that. Also, dup()ed file descriptors might not work very	350	best to avoid that. Also, C<dup ()>'ed file descriptors might not work
321	well if you register events for both fds.	351	very well if you register events for both fds.
322		352
323	Please note that epoll sometimes generates spurious notifications, so you	353	Please note that epoll sometimes generates spurious notifications, so you
324	need to use non-blocking I/O or other means to avoid blocking when no data	354	need to use non-blocking I/O or other means to avoid blocking when no data
325	(or space) is available.	355	(or space) is available.
326		356
		357	Best performance from this backend is achieved by not unregistering all
		358	watchers for a file descriptor until it has been closed, if possible, i.e.
		359	keep at least one watcher active per fd at all times.
		360
		361	While nominally embeddeble in other event loops, this feature is broken in
		362	all kernel versions tested so far.
		363
327	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	364	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
328		365
329	Kqueue deserves special mention, as at the time of this writing, it	366	Kqueue deserves special mention, as at the time of this writing, it
330	was broken on all BSDs except NetBSD (usually it doesn't work with	367	was broken on all BSDs except NetBSD (usually it doesn't work reliably
331	anything but sockets and pipes, except on Darwin, where of course its	368	with anything but sockets and pipes, except on Darwin, where of course
332	completely useless). For this reason its not being "autodetected"	369	it's completely useless). For this reason it's not being "autodetected"
333	unless you explicitly specify it explicitly in the flags (i.e. using	370	unless you explicitly specify it explicitly in the flags (i.e. using
334	C<EVBACKEND_KQUEUE>).	371	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		372	system like NetBSD.
		373
		374	You still can embed kqueue into a normal poll or select backend and use it
		375	only for sockets (after having made sure that sockets work with kqueue on
		376	the target platform). See C<ev_embed> watchers for more info.
335		377
336	It scales in the same way as the epoll backend, but the interface to the	378	It scales in the same way as the epoll backend, but the interface to the
337	kernel is more efficient (which says nothing about its actual speed, of	379	kernel is more efficient (which says nothing about its actual speed, of
338	course). While starting and stopping an I/O watcher does not cause an	380	course). While stopping, setting and starting an I/O watcher does never
339	extra syscall as with epoll, it still adds up to four event changes per	381	cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
340	incident, so its best to avoid that.	382	two event changes per incident, support for C<fork ()> is very bad and it
		383	drops fds silently in similarly hard-to-detect cases.
		384
		385	This backend usually performs well under most conditions.
		386
		387	While nominally embeddable in other event loops, this doesn't work
		388	everywhere, so you might need to test for this. And since it is broken
		389	almost everywhere, you should only use it when you have a lot of sockets
		390	(for which it usually works), by embedding it into another event loop
		391	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for
		392	sockets.
341		393
342	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	394	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
343		395
344	This is not implemented yet (and might never be).	396	This is not implemented yet (and might never be, unless you send me an
		397	implementation). According to reports, C</dev/poll> only supports sockets
		398	and is not embeddable, which would limit the usefulness of this backend
		399	immensely.
345		400
346	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	401	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
347		402
348	This uses the Solaris 10 port mechanism. As with everything on Solaris,	403	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
349	it's really slow, but it still scales very well (O(active_fds)).	404	it's really slow, but it still scales very well (O(active_fds)).
350		405
351	Please note that solaris ports can result in a lot of spurious	406	Please note that solaris event ports can deliver a lot of spurious
352	notifications, so you need to use non-blocking I/O or other means to avoid	407	notifications, so you need to use non-blocking I/O or other means to avoid
353	blocking when no data (or space) is available.	408	blocking when no data (or space) is available.
		409
		410	While this backend scales well, it requires one system call per active
		411	file descriptor per loop iteration. For small and medium numbers of file
		412	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		413	might perform better.
		414
		415	On the positive side, ignoring the spurious readyness notifications, this
		416	backend actually performed to specification in all tests and is fully
		417	embeddable, which is a rare feat among the OS-specific backends.
354		418
355	=item C<EVBACKEND_ALL>	419	=item C<EVBACKEND_ALL>
356		420
357	Try all backends (even potentially broken ones that wouldn't be tried	421	Try all backends (even potentially broken ones that wouldn't be tried
358	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	422	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
359	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	423	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
360		424
		425	It is definitely not recommended to use this flag.
		426
361	=back	427	=back
362		428
363	If one or more of these are ored into the flags value, then only these	429	If one or more of these are ored into the flags value, then only these
364	backends will be tried (in the reverse order as given here). If none are	430	backends will be tried (in the reverse order as listed here). If none are
365	specified, most compiled-in backend will be tried, usually in reverse	431	specified, all backends in C<ev_recommended_backends ()> will be tried.
366	order of their flag values :)
367		432
368	The most typical usage is like this:	433	The most typical usage is like this:
369		434
370	if (!ev_default_loop (0))	435	if (!ev_default_loop (0))
371	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");	436	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
…		…
399	Destroys the default loop again (frees all memory and kernel state	464	Destroys the default loop again (frees all memory and kernel state
400	etc.). None of the active event watchers will be stopped in the normal	465	etc.). None of the active event watchers will be stopped in the normal
401	sense, so e.g. C<ev_is_active> might still return true. It is your	466	sense, so e.g. C<ev_is_active> might still return true. It is your
402	responsibility to either stop all watchers cleanly yoursef I<before>	467	responsibility to either stop all watchers cleanly yoursef I<before>
403	calling this function, or cope with the fact afterwards (which is usually	468	calling this function, or cope with the fact afterwards (which is usually
404	the easiest thing, youc na just ignore the watchers and/or C<free ()> them	469	the easiest thing, you can just ignore the watchers and/or C<free ()> them
405	for example).	470	for example).
		471
		472	Note that certain global state, such as signal state, will not be freed by
		473	this function, and related watchers (such as signal and child watchers)
		474	would need to be stopped manually.
		475
		476	In general it is not advisable to call this function except in the
		477	rare occasion where you really need to free e.g. the signal handling
		478	pipe fds. If you need dynamically allocated loops it is better to use
		479	C<ev_loop_new> and C<ev_loop_destroy>).
406		480
407	=item ev_loop_destroy (loop)	481	=item ev_loop_destroy (loop)
408		482
409	Like C<ev_default_destroy>, but destroys an event loop created by an	483	Like C<ev_default_destroy>, but destroys an event loop created by an
410	earlier call to C<ev_loop_new>.	484	earlier call to C<ev_loop_new>.
411		485
412	=item ev_default_fork ()	486	=item ev_default_fork ()
413		487
		488	This function sets a flag that causes subsequent C<ev_loop> iterations
414	This function reinitialises the kernel state for backends that have	489	to reinitialise the kernel state for backends that have one. Despite the
415	one. Despite the name, you can call it anytime, but it makes most sense	490	name, you can call it anytime, but it makes most sense after forking, in
416	after forking, in either the parent or child process (or both, but that	491	the child process (or both child and parent, but that again makes little
417	again makes little sense).	492	sense). You I<must> call it in the child before using any of the libev
		493	functions, and it will only take effect at the next C<ev_loop> iteration.
418		494
419	You I<must> call this function in the child process after forking if and	495	On the other hand, you only need to call this function in the child
420	only if you want to use the event library in both processes. If you just	496	process if and only if you want to use the event library in the child. If
421	fork+exec, you don't have to call it.	497	you just fork+exec, you don't have to call it at all.
422		498
423	The function itself is quite fast and it's usually not a problem to call	499	The function itself is quite fast and it's usually not a problem to call
424	it just in case after a fork. To make this easy, the function will fit in	500	it just in case after a fork. To make this easy, the function will fit in
425	quite nicely into a call to C<pthread_atfork>:	501	quite nicely into a call to C<pthread_atfork>:
426		502
427	pthread_atfork (0, 0, ev_default_fork);	503	pthread_atfork (0, 0, ev_default_fork);
428
429	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use
430	without calling this function, so if you force one of those backends you
431	do not need to care.
432		504
433	=item ev_loop_fork (loop)	505	=item ev_loop_fork (loop)
434		506
435	Like C<ev_default_fork>, but acts on an event loop created by	507	Like C<ev_default_fork>, but acts on an event loop created by
436	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	508	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
…		…
455		527
456	Returns the current "event loop time", which is the time the event loop	528	Returns the current "event loop time", which is the time the event loop
457	received events and started processing them. This timestamp does not	529	received events and started processing them. This timestamp does not
458	change as long as callbacks are being processed, and this is also the base	530	change as long as callbacks are being processed, and this is also the base
459	time used for relative timers. You can treat it as the timestamp of the	531	time used for relative timers. You can treat it as the timestamp of the
460	event occuring (or more correctly, libev finding out about it).	532	event occurring (or more correctly, libev finding out about it).
461		533
462	=item ev_loop (loop, int flags)	534	=item ev_loop (loop, int flags)
463		535
464	Finally, this is it, the event handler. This function usually is called	536	Finally, this is it, the event handler. This function usually is called
465	after you initialised all your watchers and you want to start handling	537	after you initialised all your watchers and you want to start handling
…		…
486	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	558	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is
487	usually a better approach for this kind of thing.	559	usually a better approach for this kind of thing.
488		560
489	Here are the gory details of what C<ev_loop> does:	561	Here are the gory details of what C<ev_loop> does:
490		562
491	* If there are no active watchers (reference count is zero), return.	563	- Before the first iteration, call any pending watchers.
492	- Queue prepare watchers and then call all outstanding watchers.	564	* If EVFLAG_FORKCHECK was used, check for a fork.
		565	- If a fork was detected, queue and call all fork watchers.
		566	- Queue and call all prepare watchers.
493	- If we have been forked, recreate the kernel state.	567	- If we have been forked, recreate the kernel state.
494	- Update the kernel state with all outstanding changes.	568	- Update the kernel state with all outstanding changes.
495	- Update the "event loop time".	569	- Update the "event loop time".
496	- Calculate for how long to block.	570	- Calculate for how long to sleep or block, if at all
		571	(active idle watchers, EVLOOP_NONBLOCK or not having
		572	any active watchers at all will result in not sleeping).
		573	- Sleep if the I/O and timer collect interval say so.
497	- Block the process, waiting for any events.	574	- Block the process, waiting for any events.
498	- Queue all outstanding I/O (fd) events.	575	- Queue all outstanding I/O (fd) events.
499	- Update the "event loop time" and do time jump handling.	576	- Update the "event loop time" and do time jump handling.
500	- Queue all outstanding timers.	577	- Queue all outstanding timers.
501	- Queue all outstanding periodics.	578	- Queue all outstanding periodics.
502	- If no events are pending now, queue all idle watchers.	579	- If no events are pending now, queue all idle watchers.
503	- Queue all check watchers.	580	- Queue all check watchers.
504	- Call all queued watchers in reverse order (i.e. check watchers first).	581	- Call all queued watchers in reverse order (i.e. check watchers first).
505	Signals and child watchers are implemented as I/O watchers, and will	582	Signals and child watchers are implemented as I/O watchers, and will
506	be handled here by queueing them when their watcher gets executed.	583	be handled here by queueing them when their watcher gets executed.
507	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	584	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
508	were used, return, otherwise continue with step *.	585	were used, or there are no active watchers, return, otherwise
		586	continue with step *.
509		587
510	Example: Queue some jobs and then loop until no events are outsanding	588	Example: Queue some jobs and then loop until no events are outstanding
511	anymore.	589	anymore.
512		590
513	... queue jobs here, make sure they register event watchers as long	591	... queue jobs here, make sure they register event watchers as long
514	... as they still have work to do (even an idle watcher will do..)	592	... as they still have work to do (even an idle watcher will do..)
515	ev_loop (my_loop, 0);	593	ev_loop (my_loop, 0);
…		…
519		597
520	Can be used to make a call to C<ev_loop> return early (but only after it	598	Can be used to make a call to C<ev_loop> return early (but only after it
521	has processed all outstanding events). The C<how> argument must be either	599	has processed all outstanding events). The C<how> argument must be either
522	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	600	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
523	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	601	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
		602
		603	This "unloop state" will be cleared when entering C<ev_loop> again.
524		604
525	=item ev_ref (loop)	605	=item ev_ref (loop)
526		606
527	=item ev_unref (loop)	607	=item ev_unref (loop)
528		608
…		…
533	returning, ev_unref() after starting, and ev_ref() before stopping it. For	613	returning, ev_unref() after starting, and ev_ref() before stopping it. For
534	example, libev itself uses this for its internal signal pipe: It is not	614	example, libev itself uses this for its internal signal pipe: It is not
535	visible to the libev user and should not keep C<ev_loop> from exiting if	615	visible to the libev user and should not keep C<ev_loop> from exiting if
536	no event watchers registered by it are active. It is also an excellent	616	no event watchers registered by it are active. It is also an excellent
537	way to do this for generic recurring timers or from within third-party	617	way to do this for generic recurring timers or from within third-party
538	libraries. Just remember to I<unref after start> and I<ref before stop>.	618	libraries. Just remember to I<unref after start> and I<ref before stop>
		619	(but only if the watcher wasn't active before, or was active before,
		620	respectively).
539		621
540	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	622	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
541	running when nothing else is active.	623	running when nothing else is active.
542		624
543	struct ev_signal exitsig;	625	struct ev_signal exitsig;
…		…
547		629
548	Example: For some weird reason, unregister the above signal handler again.	630	Example: For some weird reason, unregister the above signal handler again.
549		631
550	ev_ref (loop);	632	ev_ref (loop);
551	ev_signal_stop (loop, &exitsig);	633	ev_signal_stop (loop, &exitsig);
		634
		635	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		636
		637	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		638
		639	These advanced functions influence the time that libev will spend waiting
		640	for events. Both are by default C<0>, meaning that libev will try to
		641	invoke timer/periodic callbacks and I/O callbacks with minimum latency.
		642
		643	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		644	allows libev to delay invocation of I/O and timer/periodic callbacks to
		645	increase efficiency of loop iterations.
		646
		647	The background is that sometimes your program runs just fast enough to
		648	handle one (or very few) event(s) per loop iteration. While this makes
		649	the program responsive, it also wastes a lot of CPU time to poll for new
		650	events, especially with backends like C<select ()> which have a high
		651	overhead for the actual polling but can deliver many events at once.
		652
		653	By setting a higher I<io collect interval> you allow libev to spend more
		654	time collecting I/O events, so you can handle more events per iteration,
		655	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		656	C<ev_timer>) will be not affected. Setting this to a non-null value will
		657	introduce an additional C<ev_sleep ()> call into most loop iterations.
		658
		659	Likewise, by setting a higher I<timeout collect interval> you allow libev
		660	to spend more time collecting timeouts, at the expense of increased
		661	latency (the watcher callback will be called later). C<ev_io> watchers
		662	will not be affected. Setting this to a non-null value will not introduce
		663	any overhead in libev.
		664
		665	Many (busy) programs can usually benefit by setting the io collect
		666	interval to a value near C<0.1> or so, which is often enough for
		667	interactive servers (of course not for games), likewise for timeouts. It
		668	usually doesn't make much sense to set it to a lower value than C<0.01>,
		669	as this approsaches the timing granularity of most systems.
552		670
553	=back	671	=back
554		672
555		673
556	=head1 ANATOMY OF A WATCHER	674	=head1 ANATOMY OF A WATCHER
…		…
656	=item C<EV_FORK>	774	=item C<EV_FORK>
657		775
658	The event loop has been resumed in the child process after fork (see	776	The event loop has been resumed in the child process after fork (see
659	C<ev_fork>).	777	C<ev_fork>).
660		778
		779	=item C<EV_ASYNC>
		780
		781	The given async watcher has been asynchronously notified (see C<ev_async>).
		782
661	=item C<EV_ERROR>	783	=item C<EV_ERROR>
662		784
663	An unspecified error has occured, the watcher has been stopped. This might	785	An unspecified error has occured, the watcher has been stopped. This might
664	happen because the watcher could not be properly started because libev	786	happen because the watcher could not be properly started because libev
665	ran out of memory, a file descriptor was found to be closed or any other	787	ran out of memory, a file descriptor was found to be closed or any other
…		…
736	=item bool ev_is_pending (ev_TYPE *watcher)	858	=item bool ev_is_pending (ev_TYPE *watcher)
737		859
738	Returns a true value iff the watcher is pending, (i.e. it has outstanding	860	Returns a true value iff the watcher is pending, (i.e. it has outstanding
739	events but its callback has not yet been invoked). As long as a watcher	861	events but its callback has not yet been invoked). As long as a watcher
740	is pending (but not active) you must not call an init function on it (but	862	is pending (but not active) you must not call an init function on it (but
741	C<ev_TYPE_set> is safe) and you must make sure the watcher is available to	863	C<ev_TYPE_set> is safe), you must not change its priority, and you must
742	libev (e.g. you cnanot C<free ()> it).	864	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		865	it).
743		866
744	=item callback ev_cb (ev_TYPE *watcher)	867	=item callback ev_cb (ev_TYPE *watcher)
745		868
746	Returns the callback currently set on the watcher.	869	Returns the callback currently set on the watcher.
747		870
…		…
766	watchers on the same event and make sure one is called first.	889	watchers on the same event and make sure one is called first.
767		890
768	If you need to suppress invocation when higher priority events are pending	891	If you need to suppress invocation when higher priority events are pending
769	you need to look at C<ev_idle> watchers, which provide this functionality.	892	you need to look at C<ev_idle> watchers, which provide this functionality.
770		893
		894	You I<must not> change the priority of a watcher as long as it is active or
		895	pending.
		896
771	The default priority used by watchers when no priority has been set is	897	The default priority used by watchers when no priority has been set is
772	always C<0>, which is supposed to not be too high and not be too low :).	898	always C<0>, which is supposed to not be too high and not be too low :).
773		899
774	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is	900	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
775	fine, as long as you do not mind that the priority value you query might	901	fine, as long as you do not mind that the priority value you query might
776	or might not have been adjusted to be within valid range.	902	or might not have been adjusted to be within valid range.
		903
		904	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		905
		906	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		907	C<loop> nor C<revents> need to be valid as long as the watcher callback
		908	can deal with that fact.
		909
		910	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		911
		912	If the watcher is pending, this function returns clears its pending status
		913	and returns its C<revents> bitset (as if its callback was invoked). If the
		914	watcher isn't pending it does nothing and returns C<0>.
777		915
778	=back	916	=back
779		917
780		918
781	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	919	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
…		…
866	In general you can register as many read and/or write event watchers per	1004	In general you can register as many read and/or write event watchers per
867	fd as you want (as long as you don't confuse yourself). Setting all file	1005	fd as you want (as long as you don't confuse yourself). Setting all file
868	descriptors to non-blocking mode is also usually a good idea (but not	1006	descriptors to non-blocking mode is also usually a good idea (but not
869	required if you know what you are doing).	1007	required if you know what you are doing).
870		1008
871	You have to be careful with dup'ed file descriptors, though. Some backends
872	(the linux epoll backend is a notable example) cannot handle dup'ed file
873	descriptors correctly if you register interest in two or more fds pointing
874	to the same underlying file/socket/etc. description (that is, they share
875	the same underlying "file open").
876
877	If you must do this, then force the use of a known-to-be-good backend	1009	If you must do this, then force the use of a known-to-be-good backend
878	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	1010	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
879	C<EVBACKEND_POLL>).	1011	C<EVBACKEND_POLL>).
880		1012
881	Another thing you have to watch out for is that it is quite easy to	1013	Another thing you have to watch out for is that it is quite easy to
…		…
891	play around with an Xlib connection), then you have to seperately re-test	1023	play around with an Xlib connection), then you have to seperately re-test
892	whether a file descriptor is really ready with a known-to-be good interface	1024	whether a file descriptor is really ready with a known-to-be good interface
893	such as poll (fortunately in our Xlib example, Xlib already does this on	1025	such as poll (fortunately in our Xlib example, Xlib already does this on
894	its own, so its quite safe to use).	1026	its own, so its quite safe to use).
895		1027
		1028	=head3 The special problem of disappearing file descriptors
		1029
		1030	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		1031	descriptor (either by calling C<close> explicitly or by any other means,
		1032	such as C<dup>). The reason is that you register interest in some file
		1033	descriptor, but when it goes away, the operating system will silently drop
		1034	this interest. If another file descriptor with the same number then is
		1035	registered with libev, there is no efficient way to see that this is, in
		1036	fact, a different file descriptor.
		1037
		1038	To avoid having to explicitly tell libev about such cases, libev follows
		1039	the following policy: Each time C<ev_io_set> is being called, libev
		1040	will assume that this is potentially a new file descriptor, otherwise
		1041	it is assumed that the file descriptor stays the same. That means that
		1042	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		1043	descriptor even if the file descriptor number itself did not change.
		1044
		1045	This is how one would do it normally anyway, the important point is that
		1046	the libev application should not optimise around libev but should leave
		1047	optimisations to libev.
		1048
		1049	=head3 The special problem of dup'ed file descriptors
		1050
		1051	Some backends (e.g. epoll), cannot register events for file descriptors,
		1052	but only events for the underlying file descriptions. That means when you
		1053	have C<dup ()>'ed file descriptors or weirder constellations, and register
		1054	events for them, only one file descriptor might actually receive events.
		1055
		1056	There is no workaround possible except not registering events
		1057	for potentially C<dup ()>'ed file descriptors, or to resort to
		1058	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1059
		1060	=head3 The special problem of fork
		1061
		1062	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1063	useless behaviour. Libev fully supports fork, but needs to be told about
		1064	it in the child.
		1065
		1066	To support fork in your programs, you either have to call
		1067	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1068	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1069	C<EVBACKEND_POLL>.
		1070
		1071
		1072	=head3 Watcher-Specific Functions
		1073
896	=over 4	1074	=over 4
897		1075
898	=item ev_io_init (ev_io *, callback, int fd, int events)	1076	=item ev_io_init (ev_io *, callback, int fd, int events)
899		1077
900	=item ev_io_set (ev_io *, int fd, int events)	1078	=item ev_io_set (ev_io *, int fd, int events)
…		…
910	=item int events [read-only]	1088	=item int events [read-only]
911		1089
912	The events being watched.	1090	The events being watched.
913		1091
914	=back	1092	=back
		1093
		1094	=head3 Examples
915		1095
916	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1096	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
917	readable, but only once. Since it is likely line-buffered, you could	1097	readable, but only once. Since it is likely line-buffered, you could
918	attempt to read a whole line in the callback.	1098	attempt to read a whole line in the callback.
919		1099
…		…
952	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1132	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
953		1133
954	The callback is guarenteed to be invoked only when its timeout has passed,	1134	The callback is guarenteed to be invoked only when its timeout has passed,
955	but if multiple timers become ready during the same loop iteration then	1135	but if multiple timers become ready during the same loop iteration then
956	order of execution is undefined.	1136	order of execution is undefined.
		1137
		1138	=head3 Watcher-Specific Functions and Data Members
957		1139
958	=over 4	1140	=over 4
959		1141
960	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1142	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
961		1143
…		…
1015	or C<ev_timer_again> is called and determines the next timeout (if any),	1197	or C<ev_timer_again> is called and determines the next timeout (if any),
1016	which is also when any modifications are taken into account.	1198	which is also when any modifications are taken into account.
1017		1199
1018	=back	1200	=back
1019		1201
		1202	=head3 Examples
		1203
1020	Example: Create a timer that fires after 60 seconds.	1204	Example: Create a timer that fires after 60 seconds.
1021		1205
1022	static void	1206	static void
1023	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)	1207	one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1024	{	1208	{
…		…
1057	but on wallclock time (absolute time). You can tell a periodic watcher	1241	but on wallclock time (absolute time). You can tell a periodic watcher
1058	to trigger "at" some specific point in time. For example, if you tell a	1242	to trigger "at" some specific point in time. For example, if you tell a
1059	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	1243	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()
1060	+ 10.>) and then reset your system clock to the last year, then it will	1244	+ 10.>) and then reset your system clock to the last year, then it will
1061	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1245	take a year to trigger the event (unlike an C<ev_timer>, which would trigger
1062	roughly 10 seconds later and of course not if you reset your system time	1246	roughly 10 seconds later).
1063	again).
1064		1247
1065	They can also be used to implement vastly more complex timers, such as	1248	They can also be used to implement vastly more complex timers, such as
1066	triggering an event on eahc midnight, local time.	1249	triggering an event on each midnight, local time or other, complicated,
		1250	rules.
1067		1251
1068	As with timers, the callback is guarenteed to be invoked only when the	1252	As with timers, the callback is guarenteed to be invoked only when the
1069	time (C<at>) has been passed, but if multiple periodic timers become ready	1253	time (C<at>) has been passed, but if multiple periodic timers become ready
1070	during the same loop iteration then order of execution is undefined.	1254	during the same loop iteration then order of execution is undefined.
1071		1255
		1256	=head3 Watcher-Specific Functions and Data Members
		1257
1072	=over 4	1258	=over 4
1073		1259
1074	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1260	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
1075		1261
1076	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1262	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
…		…
1078	Lots of arguments, lets sort it out... There are basically three modes of	1264	Lots of arguments, lets sort it out... There are basically three modes of
1079	operation, and we will explain them from simplest to complex:	1265	operation, and we will explain them from simplest to complex:
1080		1266
1081	=over 4	1267	=over 4
1082		1268
1083	=item * absolute timer (interval = reschedule_cb = 0)	1269	=item * absolute timer (at = time, interval = reschedule_cb = 0)
1084		1270
1085	In this configuration the watcher triggers an event at the wallclock time	1271	In this configuration the watcher triggers an event at the wallclock time
1086	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1272	C<at> and doesn't repeat. It will not adjust when a time jump occurs,
1087	that is, if it is to be run at January 1st 2011 then it will run when the	1273	that is, if it is to be run at January 1st 2011 then it will run when the
1088	system time reaches or surpasses this time.	1274	system time reaches or surpasses this time.
1089		1275
1090	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1276	=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1091		1277
1092	In this mode the watcher will always be scheduled to time out at the next	1278	In this mode the watcher will always be scheduled to time out at the next
1093	C<at + N * interval> time (for some integer N) and then repeat, regardless	1279	C<at + N * interval> time (for some integer N, which can also be negative)
1094	of any time jumps.	1280	and then repeat, regardless of any time jumps.
1095		1281
1096	This can be used to create timers that do not drift with respect to system	1282	This can be used to create timers that do not drift with respect to system
1097	time:	1283	time:
1098		1284
1099	ev_periodic_set (&periodic, 0., 3600., 0);	1285	ev_periodic_set (&periodic, 0., 3600., 0);
…		…
1105		1291
1106	Another way to think about it (for the mathematically inclined) is that	1292	Another way to think about it (for the mathematically inclined) is that
1107	C<ev_periodic> will try to run the callback in this mode at the next possible	1293	C<ev_periodic> will try to run the callback in this mode at the next possible
1108	time where C<time = at (mod interval)>, regardless of any time jumps.	1294	time where C<time = at (mod interval)>, regardless of any time jumps.
1109		1295
		1296	For numerical stability it is preferable that the C<at> value is near
		1297	C<ev_now ()> (the current time), but there is no range requirement for
		1298	this value.
		1299
1110	=item * manual reschedule mode (reschedule_cb = callback)	1300	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1111		1301
1112	In this mode the values for C<interval> and C<at> are both being	1302	In this mode the values for C<interval> and C<at> are both being
1113	ignored. Instead, each time the periodic watcher gets scheduled, the	1303	ignored. Instead, each time the periodic watcher gets scheduled, the
1114	reschedule callback will be called with the watcher as first, and the	1304	reschedule callback will be called with the watcher as first, and the
1115	current time as second argument.	1305	current time as second argument.
1116		1306
1117	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1307	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
1118	ever, or make any event loop modifications>. If you need to stop it,	1308	ever, or make any event loop modifications>. If you need to stop it,
1119	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	1309	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by
1120	starting a prepare watcher).	1310	starting an C<ev_prepare> watcher, which is legal).
1121		1311
1122	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,	1312	Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1123	ev_tstamp now)>, e.g.:	1313	ev_tstamp now)>, e.g.:
1124		1314
1125	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1315	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
…		…
1148	Simply stops and restarts the periodic watcher again. This is only useful	1338	Simply stops and restarts the periodic watcher again. This is only useful
1149	when you changed some parameters or the reschedule callback would return	1339	when you changed some parameters or the reschedule callback would return
1150	a different time than the last time it was called (e.g. in a crond like	1340	a different time than the last time it was called (e.g. in a crond like
1151	program when the crontabs have changed).	1341	program when the crontabs have changed).
1152		1342
		1343	=item ev_tstamp offset [read-write]
		1344
		1345	When repeating, this contains the offset value, otherwise this is the
		1346	absolute point in time (the C<at> value passed to C<ev_periodic_set>).
		1347
		1348	Can be modified any time, but changes only take effect when the periodic
		1349	timer fires or C<ev_periodic_again> is being called.
		1350
1153	=item ev_tstamp interval [read-write]	1351	=item ev_tstamp interval [read-write]
1154		1352
1155	The current interval value. Can be modified any time, but changes only	1353	The current interval value. Can be modified any time, but changes only
1156	take effect when the periodic timer fires or C<ev_periodic_again> is being	1354	take effect when the periodic timer fires or C<ev_periodic_again> is being
1157	called.	1355	called.
…		…
1160		1358
1161	The current reschedule callback, or C<0>, if this functionality is	1359	The current reschedule callback, or C<0>, if this functionality is
1162	switched off. Can be changed any time, but changes only take effect when	1360	switched off. Can be changed any time, but changes only take effect when
1163	the periodic timer fires or C<ev_periodic_again> is being called.	1361	the periodic timer fires or C<ev_periodic_again> is being called.
1164		1362
		1363	=item ev_tstamp at [read-only]
		1364
		1365	When active, contains the absolute time that the watcher is supposed to
		1366	trigger next.
		1367
1165	=back	1368	=back
		1369
		1370	=head3 Examples
1166		1371
1167	Example: Call a callback every hour, or, more precisely, whenever the	1372	Example: Call a callback every hour, or, more precisely, whenever the
1168	system clock is divisible by 3600. The callback invocation times have	1373	system clock is divisible by 3600. The callback invocation times have
1169	potentially a lot of jittering, but good long-term stability.	1374	potentially a lot of jittering, but good long-term stability.
1170		1375
…		…
1210	with the kernel (thus it coexists with your own signal handlers as long	1415	with the kernel (thus it coexists with your own signal handlers as long
1211	as you don't register any with libev). Similarly, when the last signal	1416	as you don't register any with libev). Similarly, when the last signal
1212	watcher for a signal is stopped libev will reset the signal handler to	1417	watcher for a signal is stopped libev will reset the signal handler to
1213	SIG_DFL (regardless of what it was set to before).	1418	SIG_DFL (regardless of what it was set to before).
1214		1419
		1420	=head3 Watcher-Specific Functions and Data Members
		1421
1215	=over 4	1422	=over 4
1216		1423
1217	=item ev_signal_init (ev_signal *, callback, int signum)	1424	=item ev_signal_init (ev_signal *, callback, int signum)
1218		1425
1219	=item ev_signal_set (ev_signal *, int signum)	1426	=item ev_signal_set (ev_signal *, int signum)
…		…
1231	=head2 C<ev_child> - watch out for process status changes	1438	=head2 C<ev_child> - watch out for process status changes
1232		1439
1233	Child watchers trigger when your process receives a SIGCHLD in response to	1440	Child watchers trigger when your process receives a SIGCHLD in response to
1234	some child status changes (most typically when a child of yours dies).	1441	some child status changes (most typically when a child of yours dies).
1235		1442
		1443	=head3 Watcher-Specific Functions and Data Members
		1444
1236	=over 4	1445	=over 4
1237		1446
1238	=item ev_child_init (ev_child *, callback, int pid)	1447	=item ev_child_init (ev_child *, callback, int pid, int trace)
1239		1448
1240	=item ev_child_set (ev_child *, int pid)	1449	=item ev_child_set (ev_child *, int pid, int trace)
1241		1450
1242	Configures the watcher to wait for status changes of process C<pid> (or	1451	Configures the watcher to wait for status changes of process C<pid> (or
1243	I<any> process if C<pid> is specified as C<0>). The callback can look	1452	I<any> process if C<pid> is specified as C<0>). The callback can look
1244	at the C<rstatus> member of the C<ev_child> watcher structure to see	1453	at the C<rstatus> member of the C<ev_child> watcher structure to see
1245	the status word (use the macros from C<sys/wait.h> and see your systems	1454	the status word (use the macros from C<sys/wait.h> and see your systems
1246	C<waitpid> documentation). The C<rpid> member contains the pid of the	1455	C<waitpid> documentation). The C<rpid> member contains the pid of the
1247	process causing the status change.	1456	process causing the status change. C<trace> must be either C<0> (only
		1457	activate the watcher when the process terminates) or C<1> (additionally
		1458	activate the watcher when the process is stopped or continued).
1248		1459
1249	=item int pid [read-only]	1460	=item int pid [read-only]
1250		1461
1251	The process id this watcher watches out for, or C<0>, meaning any process id.	1462	The process id this watcher watches out for, or C<0>, meaning any process id.
1252		1463
…		…
1258		1469
1259	The process exit/trace status caused by C<rpid> (see your systems	1470	The process exit/trace status caused by C<rpid> (see your systems
1260	C<waitpid> and C<sys/wait.h> documentation for details).	1471	C<waitpid> and C<sys/wait.h> documentation for details).
1261		1472
1262	=back	1473	=back
		1474
		1475	=head3 Examples
1263		1476
1264	Example: Try to exit cleanly on SIGINT and SIGTERM.	1477	Example: Try to exit cleanly on SIGINT and SIGTERM.
1265		1478
1266	static void	1479	static void
1267	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)	1480	sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
…		…
1308	semantics of C<ev_stat> watchers, which means that libev sometimes needs	1521	semantics of C<ev_stat> watchers, which means that libev sometimes needs
1309	to fall back to regular polling again even with inotify, but changes are	1522	to fall back to regular polling again even with inotify, but changes are
1310	usually detected immediately, and if the file exists there will be no	1523	usually detected immediately, and if the file exists there will be no
1311	polling.	1524	polling.
1312		1525
		1526	=head3 Inotify
		1527
		1528	When C<inotify (7)> support has been compiled into libev (generally only
		1529	available on Linux) and present at runtime, it will be used to speed up
		1530	change detection where possible. The inotify descriptor will be created lazily
		1531	when the first C<ev_stat> watcher is being started.
		1532
		1533	Inotify presense does not change the semantics of C<ev_stat> watchers
		1534	except that changes might be detected earlier, and in some cases, to avoid
		1535	making regular C<stat> calls. Even in the presense of inotify support
		1536	there are many cases where libev has to resort to regular C<stat> polling.
		1537
		1538	(There is no support for kqueue, as apparently it cannot be used to
		1539	implement this functionality, due to the requirement of having a file
		1540	descriptor open on the object at all times).
		1541
		1542	=head3 The special problem of stat time resolution
		1543
		1544	The C<stat ()> syscall only supports full-second resolution portably, and
		1545	even on systems where the resolution is higher, many filesystems still
		1546	only support whole seconds.
		1547
		1548	That means that, if the time is the only thing that changes, you might
		1549	miss updates: on the first update, C<ev_stat> detects a change and calls
		1550	your callback, which does something. When there is another update within
		1551	the same second, C<ev_stat> will be unable to detect it.
		1552
		1553	The solution to this is to delay acting on a change for a second (or till
		1554	the next second boundary), using a roughly one-second delay C<ev_timer>
		1555	(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
		1556	is added to work around small timing inconsistencies of some operating
		1557	systems.
		1558
		1559	=head3 Watcher-Specific Functions and Data Members
		1560
1313	=over 4	1561	=over 4
1314		1562
1315	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)	1563	=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1316		1564
1317	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)	1565	=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
…		…
1352	=item const char *path [read-only]	1600	=item const char *path [read-only]
1353		1601
1354	The filesystem path that is being watched.	1602	The filesystem path that is being watched.
1355		1603
1356	=back	1604	=back
		1605
		1606	=head3 Examples
1357		1607
1358	Example: Watch C</etc/passwd> for attribute changes.	1608	Example: Watch C</etc/passwd> for attribute changes.
1359		1609
1360	static void	1610	static void
1361	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)	1611	passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
…		…
1374	}	1624	}
1375		1625
1376	...	1626	...
1377	ev_stat passwd;	1627	ev_stat passwd;
1378		1628
1379	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	1629	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1380	ev_stat_start (loop, &passwd);	1630	ev_stat_start (loop, &passwd);
		1631
		1632	Example: Like above, but additionally use a one-second delay so we do not
		1633	miss updates (however, frequent updates will delay processing, too, so
		1634	one might do the work both on C<ev_stat> callback invocation I<and> on
		1635	C<ev_timer> callback invocation).
		1636
		1637	static ev_stat passwd;
		1638	static ev_timer timer;
		1639
		1640	static void
		1641	timer_cb (EV_P_ ev_timer *w, int revents)
		1642	{
		1643	ev_timer_stop (EV_A_ w);
		1644
		1645	/* now it's one second after the most recent passwd change */
		1646	}
		1647
		1648	static void
		1649	stat_cb (EV_P_ ev_stat *w, int revents)
		1650	{
		1651	/* reset the one-second timer */
		1652	ev_timer_again (EV_A_ &timer);
		1653	}
		1654
		1655	...
		1656	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		1657	ev_stat_start (loop, &passwd);
		1658	ev_timer_init (&timer, timer_cb, 0., 1.01);
1381		1659
1382		1660
1383	=head2 C<ev_idle> - when you've got nothing better to do...	1661	=head2 C<ev_idle> - when you've got nothing better to do...
1384		1662
1385	Idle watchers trigger events when no other events of the same or higher	1663	Idle watchers trigger events when no other events of the same or higher
…		…
1399	Apart from keeping your process non-blocking (which is a useful	1677	Apart from keeping your process non-blocking (which is a useful
1400	effect on its own sometimes), idle watchers are a good place to do	1678	effect on its own sometimes), idle watchers are a good place to do
1401	"pseudo-background processing", or delay processing stuff to after the	1679	"pseudo-background processing", or delay processing stuff to after the
1402	event loop has handled all outstanding events.	1680	event loop has handled all outstanding events.
1403		1681
		1682	=head3 Watcher-Specific Functions and Data Members
		1683
1404	=over 4	1684	=over 4
1405		1685
1406	=item ev_idle_init (ev_signal *, callback)	1686	=item ev_idle_init (ev_signal *, callback)
1407		1687
1408	Initialises and configures the idle watcher - it has no parameters of any	1688	Initialises and configures the idle watcher - it has no parameters of any
1409	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	1689	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1410	believe me.	1690	believe me.
1411		1691
1412	=back	1692	=back
		1693
		1694	=head3 Examples
1413		1695
1414	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	1696	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1415	callback, free it. Also, use no error checking, as usual.	1697	callback, free it. Also, use no error checking, as usual.
1416		1698
1417	static void	1699	static void
1418	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)	1700	idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1419	{	1701	{
1420	free (w);	1702	free (w);
1421	// now do something you wanted to do when the program has	1703	// now do something you wanted to do when the program has
1422	// no longer asnything immediate to do.	1704	// no longer anything immediate to do.
1423	}	1705	}
1424		1706
1425	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	1707	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1426	ev_idle_init (idle_watcher, idle_cb);	1708	ev_idle_init (idle_watcher, idle_cb);
1427	ev_idle_start (loop, idle_cb);	1709	ev_idle_start (loop, idle_cb);
…		…
1465	with priority higher than or equal to the event loop and one coroutine	1747	with priority higher than or equal to the event loop and one coroutine
1466	of lower priority, but only once, using idle watchers to keep the event	1748	of lower priority, but only once, using idle watchers to keep the event
1467	loop from blocking if lower-priority coroutines are active, thus mapping	1749	loop from blocking if lower-priority coroutines are active, thus mapping
1468	low-priority coroutines to idle/background tasks).	1750	low-priority coroutines to idle/background tasks).
1469		1751
		1752	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		1753	priority, to ensure that they are being run before any other watchers
		1754	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
		1755	too) should not activate ("feed") events into libev. While libev fully
		1756	supports this, they will be called before other C<ev_check> watchers
		1757	did their job. As C<ev_check> watchers are often used to embed other
		1758	(non-libev) event loops those other event loops might be in an unusable
		1759	state until their C<ev_check> watcher ran (always remind yourself to
		1760	coexist peacefully with others).
		1761
		1762	=head3 Watcher-Specific Functions and Data Members
		1763
1470	=over 4	1764	=over 4
1471		1765
1472	=item ev_prepare_init (ev_prepare *, callback)	1766	=item ev_prepare_init (ev_prepare *, callback)
1473		1767
1474	=item ev_check_init (ev_check *, callback)	1768	=item ev_check_init (ev_check *, callback)
…		…
1477	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	1771	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1478	macros, but using them is utterly, utterly and completely pointless.	1772	macros, but using them is utterly, utterly and completely pointless.
1479		1773
1480	=back	1774	=back
1481		1775
1482	Example: To include a library such as adns, you would add IO watchers	1776	=head3 Examples
1483	and a timeout watcher in a prepare handler, as required by libadns, and	1777
		1778	There are a number of principal ways to embed other event loops or modules
		1779	into libev. Here are some ideas on how to include libadns into libev
		1780	(there is a Perl module named C<EV::ADNS> that does this, which you could
		1781	use for an actually working example. Another Perl module named C<EV::Glib>
		1782	embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV
		1783	into the Glib event loop).
		1784
		1785	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1484	in a check watcher, destroy them and call into libadns. What follows is	1786	and in a check watcher, destroy them and call into libadns. What follows
1485	pseudo-code only of course:	1787	is pseudo-code only of course. This requires you to either use a low
		1788	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		1789	the callbacks for the IO/timeout watchers might not have been called yet.
1486		1790
1487	static ev_io iow [nfd];	1791	static ev_io iow [nfd];
1488	static ev_timer tw;	1792	static ev_timer tw;
1489		1793
1490	static void	1794	static void
1491	io_cb (ev_loop *loop, ev_io *w, int revents)	1795	io_cb (ev_loop *loop, ev_io *w, int revents)
1492	{	1796	{
1493	// set the relevant poll flags
1494	// could also call adns_processreadable etc. here
1495	struct pollfd *fd = (struct pollfd *)w->data;
1496	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1497	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1498	}	1797	}
1499		1798
1500	// create io watchers for each fd and a timer before blocking	1799	// create io watchers for each fd and a timer before blocking
1501	static void	1800	static void
1502	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)	1801	adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
…		…
1508		1807
1509	/* the callback is illegal, but won't be called as we stop during check */	1808	/* the callback is illegal, but won't be called as we stop during check */
1510	ev_timer_init (&tw, 0, timeout * 1e-3);	1809	ev_timer_init (&tw, 0, timeout * 1e-3);
1511	ev_timer_start (loop, &tw);	1810	ev_timer_start (loop, &tw);
1512		1811
1513	// create on ev_io per pollfd	1812	// create one ev_io per pollfd
1514	for (int i = 0; i < nfd; ++i)	1813	for (int i = 0; i < nfd; ++i)
1515	{	1814	{
1516	ev_io_init (iow + i, io_cb, fds [i].fd,	1815	ev_io_init (iow + i, io_cb, fds [i].fd,
1517	((fds [i].events & POLLIN ? EV_READ : 0)	1816	((fds [i].events & POLLIN ? EV_READ : 0)
1518	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	1817	| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1519		1818
1520	fds [i].revents = 0;	1819	fds [i].revents = 0;
1521	iow [i].data = fds + i;
1522	ev_io_start (loop, iow + i);	1820	ev_io_start (loop, iow + i);
1523	}	1821	}
1524	}	1822	}
1525		1823
1526	// stop all watchers after blocking	1824	// stop all watchers after blocking
…		…
1528	adns_check_cb (ev_loop *loop, ev_check *w, int revents)	1826	adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1529	{	1827	{
1530	ev_timer_stop (loop, &tw);	1828	ev_timer_stop (loop, &tw);
1531		1829
1532	for (int i = 0; i < nfd; ++i)	1830	for (int i = 0; i < nfd; ++i)
		1831	{
		1832	// set the relevant poll flags
		1833	// could also call adns_processreadable etc. here
		1834	struct pollfd *fd = fds + i;
		1835	int revents = ev_clear_pending (iow + i);
		1836	if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
		1837	if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
		1838
		1839	// now stop the watcher
1533	ev_io_stop (loop, iow + i);	1840	ev_io_stop (loop, iow + i);
		1841	}
1534		1842
1535	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	1843	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		1844	}
		1845
		1846	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		1847	in the prepare watcher and would dispose of the check watcher.
		1848
		1849	Method 3: If the module to be embedded supports explicit event
		1850	notification (adns does), you can also make use of the actual watcher
		1851	callbacks, and only destroy/create the watchers in the prepare watcher.
		1852
		1853	static void
		1854	timer_cb (EV_P_ ev_timer *w, int revents)
		1855	{
		1856	adns_state ads = (adns_state)w->data;
		1857	update_now (EV_A);
		1858
		1859	adns_processtimeouts (ads, &tv_now);
		1860	}
		1861
		1862	static void
		1863	io_cb (EV_P_ ev_io *w, int revents)
		1864	{
		1865	adns_state ads = (adns_state)w->data;
		1866	update_now (EV_A);
		1867
		1868	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		1869	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		1870	}
		1871
		1872	// do not ever call adns_afterpoll
		1873
		1874	Method 4: Do not use a prepare or check watcher because the module you
		1875	want to embed is too inflexible to support it. Instead, youc na override
		1876	their poll function. The drawback with this solution is that the main
		1877	loop is now no longer controllable by EV. The C<Glib::EV> module does
		1878	this.
		1879
		1880	static gint
		1881	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		1882	{
		1883	int got_events = 0;
		1884
		1885	for (n = 0; n < nfds; ++n)
		1886	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		1887
		1888	if (timeout >= 0)
		1889	// create/start timer
		1890
		1891	// poll
		1892	ev_loop (EV_A_ 0);
		1893
		1894	// stop timer again
		1895	if (timeout >= 0)
		1896	ev_timer_stop (EV_A_ &to);
		1897
		1898	// stop io watchers again - their callbacks should have set
		1899	for (n = 0; n < nfds; ++n)
		1900	ev_io_stop (EV_A_ iow [n]);
		1901
		1902	return got_events;
1536	}	1903	}
1537		1904
1538		1905
1539	=head2 C<ev_embed> - when one backend isn't enough...	1906	=head2 C<ev_embed> - when one backend isn't enough...
1540		1907
…		…
1583	portable one.	1950	portable one.
1584		1951
1585	So when you want to use this feature you will always have to be prepared	1952	So when you want to use this feature you will always have to be prepared
1586	that you cannot get an embeddable loop. The recommended way to get around	1953	that you cannot get an embeddable loop. The recommended way to get around
1587	this is to have a separate variables for your embeddable loop, try to	1954	this is to have a separate variables for your embeddable loop, try to
1588	create it, and if that fails, use the normal loop for everything:	1955	create it, and if that fails, use the normal loop for everything.
		1956
		1957	=head3 Watcher-Specific Functions and Data Members
		1958
		1959	=over 4
		1960
		1961	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
		1962
		1963	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
		1964
		1965	Configures the watcher to embed the given loop, which must be
		1966	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		1967	invoked automatically, otherwise it is the responsibility of the callback
		1968	to invoke it (it will continue to be called until the sweep has been done,
		1969	if you do not want thta, you need to temporarily stop the embed watcher).
		1970
		1971	=item ev_embed_sweep (loop, ev_embed *)
		1972
		1973	Make a single, non-blocking sweep over the embedded loop. This works
		1974	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		1975	apropriate way for embedded loops.
		1976
		1977	=item struct ev_loop *other [read-only]
		1978
		1979	The embedded event loop.
		1980
		1981	=back
		1982
		1983	=head3 Examples
		1984
		1985	Example: Try to get an embeddable event loop and embed it into the default
		1986	event loop. If that is not possible, use the default loop. The default
		1987	loop is stored in C<loop_hi>, while the mebeddable loop is stored in
		1988	C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
		1989	used).
1589		1990
1590	struct ev_loop *loop_hi = ev_default_init (0);	1991	struct ev_loop *loop_hi = ev_default_init (0);
1591	struct ev_loop *loop_lo = 0;	1992	struct ev_loop *loop_lo = 0;
1592	struct ev_embed embed;	1993	struct ev_embed embed;
1593		1994
…		…
1604	ev_embed_start (loop_hi, &embed);	2005	ev_embed_start (loop_hi, &embed);
1605	}	2006	}
1606	else	2007	else
1607	loop_lo = loop_hi;	2008	loop_lo = loop_hi;
1608		2009
1609	=over 4	2010	Example: Check if kqueue is available but not recommended and create
		2011	a kqueue backend for use with sockets (which usually work with any
		2012	kqueue implementation). Store the kqueue/socket-only event loop in
		2013	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1610		2014
1611	=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)	2015	struct ev_loop *loop = ev_default_init (0);
		2016	struct ev_loop *loop_socket = 0;
		2017	struct ev_embed embed;
		2018
		2019	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2020	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2021	{
		2022	ev_embed_init (&embed, 0, loop_socket);
		2023	ev_embed_start (loop, &embed);
		2024	}
1612		2025
1613	=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)	2026	if (!loop_socket)
		2027	loop_socket = loop;
1614		2028
1615	Configures the watcher to embed the given loop, which must be	2029	// now use loop_socket for all sockets, and loop for everything else
1616	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1617	invoked automatically, otherwise it is the responsibility of the callback
1618	to invoke it (it will continue to be called until the sweep has been done,
1619	if you do not want thta, you need to temporarily stop the embed watcher).
1620
1621	=item ev_embed_sweep (loop, ev_embed *)
1622
1623	Make a single, non-blocking sweep over the embedded loop. This works
1624	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1625	apropriate way for embedded loops.
1626
1627	=item struct ev_loop *loop [read-only]
1628
1629	The embedded event loop.
1630
1631	=back
1632		2030
1633		2031
1634	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	2032	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1635		2033
1636	Fork watchers are called when a C<fork ()> was detected (usually because	2034	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
1639	event loop blocks next and before C<ev_check> watchers are being called,	2037	event loop blocks next and before C<ev_check> watchers are being called,
1640	and only in the child after the fork. If whoever good citizen calling	2038	and only in the child after the fork. If whoever good citizen calling
1641	C<ev_default_fork> cheats and calls it in the wrong process, the fork	2039	C<ev_default_fork> cheats and calls it in the wrong process, the fork
1642	handlers will be invoked, too, of course.	2040	handlers will be invoked, too, of course.
1643		2041
		2042	=head3 Watcher-Specific Functions and Data Members
		2043
1644	=over 4	2044	=over 4
1645		2045
1646	=item ev_fork_init (ev_signal *, callback)	2046	=item ev_fork_init (ev_signal *, callback)
1647		2047
1648	Initialises and configures the fork watcher - it has no parameters of any	2048	Initialises and configures the fork watcher - it has no parameters of any
1649	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,	2049	kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
1650	believe me.	2050	believe me.
		2051
		2052	=back
		2053
		2054
		2055	=head2 C<ev_async> - how to wake up another event loop
		2056
		2057	In general, you cannot use an C<ev_loop> from multiple threads or other
		2058	asynchronous sources such as signal handlers (as opposed to multiple event
		2059	loops - those are of course safe to use in different threads).
		2060
		2061	Sometimes, however, you need to wake up another event loop you do not
		2062	control, for example because it belongs to another thread. This is what
		2063	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2064	can signal it by calling C<ev_async_send>, which is thread- and signal
		2065	safe.
		2066
		2067	This functionality is very similar to C<ev_signal> watchers, as signals,
		2068	too, are asynchronous in nature, and signals, too, will be compressed
		2069	(i.e. the number of callback invocations may be less than the number of
		2070	C<ev_async_sent> calls).
		2071
		2072	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2073	just the default loop.
		2074
		2075	=head3 Queueing
		2076
		2077	C<ev_async> does not support queueing of data in any way. The reason
		2078	is that the author does not know of a simple (or any) algorithm for a
		2079	multiple-writer-single-reader queue that works in all cases and doesn't
		2080	need elaborate support such as pthreads.
		2081
		2082	That means that if you want to queue data, you have to provide your own
		2083	queue. But at least I can tell you would implement locking around your
		2084	queue:
		2085
		2086	=over 4
		2087
		2088	=item queueing from a signal handler context
		2089
		2090	To implement race-free queueing, you simply add to the queue in the signal
		2091	handler but you block the signal handler in the watcher callback. Here is an example that does that for
		2092	some fictitiuous SIGUSR1 handler:
		2093
		2094	static ev_async mysig;
		2095
		2096	static void
		2097	sigusr1_handler (void)
		2098	{
		2099	sometype data;
		2100
		2101	// no locking etc.
		2102	queue_put (data);
		2103	ev_async_send (DEFAULT_ &mysig);
		2104	}
		2105
		2106	static void
		2107	mysig_cb (EV_P_ ev_async *w, int revents)
		2108	{
		2109	sometype data;
		2110	sigset_t block, prev;
		2111
		2112	sigemptyset (&block);
		2113	sigaddset (&block, SIGUSR1);
		2114	sigprocmask (SIG_BLOCK, &block, &prev);
		2115
		2116	while (queue_get (&data))
		2117	process (data);
		2118
		2119	if (sigismember (&prev, SIGUSR1)
		2120	sigprocmask (SIG_UNBLOCK, &block, 0);
		2121	}
		2122
		2123	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2124	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2125	either...).
		2126
		2127	=item queueing from a thread context
		2128
		2129	The strategy for threads is different, as you cannot (easily) block
		2130	threads but you can easily preempt them, so to queue safely you need to
		2131	employ a traditional mutex lock, such as in this pthread example:
		2132
		2133	static ev_async mysig;
		2134	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2135
		2136	static void
		2137	otherthread (void)
		2138	{
		2139	// only need to lock the actual queueing operation
		2140	pthread_mutex_lock (&mymutex);
		2141	queue_put (data);
		2142	pthread_mutex_unlock (&mymutex);
		2143
		2144	ev_async_send (DEFAULT_ &mysig);
		2145	}
		2146
		2147	static void
		2148	mysig_cb (EV_P_ ev_async *w, int revents)
		2149	{
		2150	pthread_mutex_lock (&mymutex);
		2151
		2152	while (queue_get (&data))
		2153	process (data);
		2154
		2155	pthread_mutex_unlock (&mymutex);
		2156	}
		2157
		2158	=back
		2159
		2160
		2161	=head3 Watcher-Specific Functions and Data Members
		2162
		2163	=over 4
		2164
		2165	=item ev_async_init (ev_async *, callback)
		2166
		2167	Initialises and configures the async watcher - it has no parameters of any
		2168	kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,
		2169	believe me.
		2170
		2171	=item ev_async_send (loop, ev_async *)
		2172
		2173	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2174	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2175	C<ev_feed_event>, this call is safe to do in other threads, signal or
		2176	similar contexts (see the dicusssion of C<EV_ATOMIC_T> in the embedding
		2177	section below on what exactly this means).
		2178
		2179	This call incurs the overhead of a syscall only once per loop iteration,
		2180	so while the overhead might be noticable, it doesn't apply to repeated
		2181	calls to C<ev_async_send>.
1651		2182
1652	=back	2183	=back
1653		2184
1654		2185
1655	=head1 OTHER FUNCTIONS	2186	=head1 OTHER FUNCTIONS
…		…
1744		2275
1745	To use it,	2276	To use it,
1746		2277
1747	#include <ev++.h>	2278	#include <ev++.h>
1748		2279
1749	(it is not installed by default). This automatically includes F<ev.h>	2280	This automatically includes F<ev.h> and puts all of its definitions (many
1750	and puts all of its definitions (many of them macros) into the global	2281	of them macros) into the global namespace. All C++ specific things are
1751	namespace. All C++ specific things are put into the C<ev> namespace.	2282	put into the C<ev> namespace. It should support all the same embedding
		2283	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
1752		2284
1753	It should support all the same embedding options as F<ev.h>, most notably	2285	Care has been taken to keep the overhead low. The only data member the C++
1754	C<EV_MULTIPLICITY>.	2286	classes add (compared to plain C-style watchers) is the event loop pointer
		2287	that the watcher is associated with (or no additional members at all if
		2288	you disable C<EV_MULTIPLICITY> when embedding libev).
		2289
		2290	Currently, functions, and static and non-static member functions can be
		2291	used as callbacks. Other types should be easy to add as long as they only
		2292	need one additional pointer for context. If you need support for other
		2293	types of functors please contact the author (preferably after implementing
		2294	it).
1755		2295
1756	Here is a list of things available in the C<ev> namespace:	2296	Here is a list of things available in the C<ev> namespace:
1757		2297
1758	=over 4	2298	=over 4
1759		2299
…		…
1775		2315
1776	All of those classes have these methods:	2316	All of those classes have these methods:
1777		2317
1778	=over 4	2318	=over 4
1779		2319
1780	=item ev::TYPE::TYPE (object *, object::method *)	2320	=item ev::TYPE::TYPE ()
1781		2321
1782	=item ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)	2322	=item ev::TYPE::TYPE (struct ev_loop *)
1783		2323
1784	=item ev::TYPE::~TYPE	2324	=item ev::TYPE::~TYPE
1785		2325
1786	The constructor takes a pointer to an object and a method pointer to	2326	The constructor (optionally) takes an event loop to associate the watcher
1787	the event handler callback to call in this class. The constructor calls	2327	with. If it is omitted, it will use C<EV_DEFAULT>.
1788	C<ev_init> for you, which means you have to call the C<set> method	2328
1789	before starting it. If you do not specify a loop then the constructor	2329	The constructor calls C<ev_init> for you, which means you have to call the
1790	automatically associates the default loop with this watcher.	2330	C<set> method before starting it.
		2331
		2332	It will not set a callback, however: You have to call the templated C<set>
		2333	method to set a callback before you can start the watcher.
		2334
		2335	(The reason why you have to use a method is a limitation in C++ which does
		2336	not allow explicit template arguments for constructors).
1791		2337
1792	The destructor automatically stops the watcher if it is active.	2338	The destructor automatically stops the watcher if it is active.
		2339
		2340	=item w->set<class, &class::method> (object *)
		2341
		2342	This method sets the callback method to call. The method has to have a
		2343	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		2344	first argument and the C<revents> as second. The object must be given as
		2345	parameter and is stored in the C<data> member of the watcher.
		2346
		2347	This method synthesizes efficient thunking code to call your method from
		2348	the C callback that libev requires. If your compiler can inline your
		2349	callback (i.e. it is visible to it at the place of the C<set> call and
		2350	your compiler is good :), then the method will be fully inlined into the
		2351	thunking function, making it as fast as a direct C callback.
		2352
		2353	Example: simple class declaration and watcher initialisation
		2354
		2355	struct myclass
		2356	{
		2357	void io_cb (ev::io &w, int revents) { }
		2358	}
		2359
		2360	myclass obj;
		2361	ev::io iow;
		2362	iow.set <myclass, &myclass::io_cb> (&obj);
		2363
		2364	=item w->set<function> (void *data = 0)
		2365
		2366	Also sets a callback, but uses a static method or plain function as
		2367	callback. The optional C<data> argument will be stored in the watcher's
		2368	C<data> member and is free for you to use.
		2369
		2370	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		2371
		2372	See the method-C<set> above for more details.
		2373
		2374	Example:
		2375
		2376	static void io_cb (ev::io &w, int revents) { }
		2377	iow.set <io_cb> ();
1793		2378
1794	=item w->set (struct ev_loop *)	2379	=item w->set (struct ev_loop *)
1795		2380
1796	Associates a different C<struct ev_loop> with this watcher. You can only	2381	Associates a different C<struct ev_loop> with this watcher. You can only
1797	do this when the watcher is inactive (and not pending either).	2382	do this when the watcher is inactive (and not pending either).
1798		2383
1799	=item w->set ([args])	2384	=item w->set ([args])
1800		2385
1801	Basically the same as C<ev_TYPE_set>, with the same args. Must be	2386	Basically the same as C<ev_TYPE_set>, with the same args. Must be
1802	called at least once. Unlike the C counterpart, an active watcher gets	2387	called at least once. Unlike the C counterpart, an active watcher gets
1803	automatically stopped and restarted.	2388	automatically stopped and restarted when reconfiguring it with this
		2389	method.
1804		2390
1805	=item w->start ()	2391	=item w->start ()
1806		2392
1807	Starts the watcher. Note that there is no C<loop> argument as the	2393	Starts the watcher. Note that there is no C<loop> argument, as the
1808	constructor already takes the loop.	2394	constructor already stores the event loop.
1809		2395
1810	=item w->stop ()	2396	=item w->stop ()
1811		2397
1812	Stops the watcher if it is active. Again, no C<loop> argument.	2398	Stops the watcher if it is active. Again, no C<loop> argument.
1813		2399
1814	=item w->again () C<ev::timer>, C<ev::periodic> only	2400	=item w->again () (C<ev::timer>, C<ev::periodic> only)
1815		2401
1816	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding	2402	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
1817	C<ev_TYPE_again> function.	2403	C<ev_TYPE_again> function.
1818		2404
1819	=item w->sweep () C<ev::embed> only	2405	=item w->sweep () (C<ev::embed> only)
1820		2406
1821	Invokes C<ev_embed_sweep>.	2407	Invokes C<ev_embed_sweep>.
1822		2408
1823	=item w->update () C<ev::stat> only	2409	=item w->update () (C<ev::stat> only)
1824		2410
1825	Invokes C<ev_stat_stat>.	2411	Invokes C<ev_stat_stat>.
1826		2412
1827	=back	2413	=back
1828		2414
…		…
1831	Example: Define a class with an IO and idle watcher, start one of them in	2417	Example: Define a class with an IO and idle watcher, start one of them in
1832	the constructor.	2418	the constructor.
1833		2419
1834	class myclass	2420	class myclass
1835	{	2421	{
1836	ev_io io; void io_cb (ev::io &w, int revents);	2422	ev::io io; void io_cb (ev::io &w, int revents);
1837	ev_idle idle void idle_cb (ev::idle &w, int revents);	2423	ev:idle idle void idle_cb (ev::idle &w, int revents);
1838		2424
1839	myclass ();	2425	myclass (int fd)
1840	}
1841
1842	myclass::myclass (int fd)
1843	: io (this, &myclass::io_cb),
1844	idle (this, &myclass::idle_cb)
1845	{	2426	{
		2427	io .set <myclass, &myclass::io_cb > (this);
		2428	idle.set <myclass, &myclass::idle_cb> (this);
		2429
1846	io.start (fd, ev::READ);	2430	io.start (fd, ev::READ);
		2431	}
1847	}	2432	};
1848		2433
1849		2434
1850	=head1 MACRO MAGIC	2435	=head1 MACRO MAGIC
1851		2436
1852	Libev can be compiled with a variety of options, the most fundemantal is	2437	Libev can be compiled with a variety of options, the most fundamantal
1853	C<EV_MULTIPLICITY>. This option determines whether (most) functions and	2438	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
1854	callbacks have an initial C<struct ev_loop *> argument.	2439	functions and callbacks have an initial C<struct ev_loop *> argument.
1855		2440
1856	To make it easier to write programs that cope with either variant, the	2441	To make it easier to write programs that cope with either variant, the
1857	following macros are defined:	2442	following macros are defined:
1858		2443
1859	=over 4	2444	=over 4
…		…
1913	Libev can (and often is) directly embedded into host	2498	Libev can (and often is) directly embedded into host
1914	applications. Examples of applications that embed it include the Deliantra	2499	applications. Examples of applications that embed it include the Deliantra
1915	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)	2500	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1916	and rxvt-unicode.	2501	and rxvt-unicode.
1917		2502
1918	The goal is to enable you to just copy the neecssary files into your	2503	The goal is to enable you to just copy the necessary files into your
1919	source directory without having to change even a single line in them, so	2504	source directory without having to change even a single line in them, so
1920	you can easily upgrade by simply copying (or having a checked-out copy of	2505	you can easily upgrade by simply copying (or having a checked-out copy of
1921	libev somewhere in your source tree).	2506	libev somewhere in your source tree).
1922		2507
1923	=head2 FILESETS	2508	=head2 FILESETS
…		…
2013		2598
2014	If defined to be C<1>, libev will try to detect the availability of the	2599	If defined to be C<1>, libev will try to detect the availability of the
2015	monotonic clock option at both compiletime and runtime. Otherwise no use	2600	monotonic clock option at both compiletime and runtime. Otherwise no use
2016	of the monotonic clock option will be attempted. If you enable this, you	2601	of the monotonic clock option will be attempted. If you enable this, you
2017	usually have to link against librt or something similar. Enabling it when	2602	usually have to link against librt or something similar. Enabling it when
2018	the functionality isn't available is safe, though, althoguh you have	2603	the functionality isn't available is safe, though, although you have
2019	to make sure you link against any libraries where the C<clock_gettime>	2604	to make sure you link against any libraries where the C<clock_gettime>
2020	function is hiding in (often F<-lrt>).	2605	function is hiding in (often F<-lrt>).
2021		2606
2022	=item EV_USE_REALTIME	2607	=item EV_USE_REALTIME
2023		2608
2024	If defined to be C<1>, libev will try to detect the availability of the	2609	If defined to be C<1>, libev will try to detect the availability of the
2025	realtime clock option at compiletime (and assume its availability at	2610	realtime clock option at compiletime (and assume its availability at
2026	runtime if successful). Otherwise no use of the realtime clock option will	2611	runtime if successful). Otherwise no use of the realtime clock option will
2027	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	2612	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2028	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries	2613	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2029	in the description of C<EV_USE_MONOTONIC>, though.	2614	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
		2615
		2616	=item EV_USE_NANOSLEEP
		2617
		2618	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		2619	and will use it for delays. Otherwise it will use C<select ()>.
2030		2620
2031	=item EV_USE_SELECT	2621	=item EV_USE_SELECT
2032		2622
2033	If undefined or defined to be C<1>, libev will compile in support for the	2623	If undefined or defined to be C<1>, libev will compile in support for the
2034	C<select>(2) backend. No attempt at autodetection will be done: if no	2624	C<select>(2) backend. No attempt at autodetection will be done: if no
…		…
2052	wants osf handles on win32 (this is the case when the select to	2642	wants osf handles on win32 (this is the case when the select to
2053	be used is the winsock select). This means that it will call	2643	be used is the winsock select). This means that it will call
2054	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	2644	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2055	it is assumed that all these functions actually work on fds, even	2645	it is assumed that all these functions actually work on fds, even
2056	on win32. Should not be defined on non-win32 platforms.	2646	on win32. Should not be defined on non-win32 platforms.
		2647
		2648	=item EV_FD_TO_WIN32_HANDLE
		2649
		2650	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		2651	file descriptors to socket handles. When not defining this symbol (the
		2652	default), then libev will call C<_get_osfhandle>, which is usually
		2653	correct. In some cases, programs use their own file descriptor management,
		2654	in which case they can provide this function to map fds to socket handles.
2057		2655
2058	=item EV_USE_POLL	2656	=item EV_USE_POLL
2059		2657
2060	If defined to be C<1>, libev will compile in support for the C<poll>(2)	2658	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2061	backend. Otherwise it will be enabled on non-win32 platforms. It	2659	backend. Otherwise it will be enabled on non-win32 platforms. It
…		…
2095		2693
2096	If defined to be C<1>, libev will compile in support for the Linux inotify	2694	If defined to be C<1>, libev will compile in support for the Linux inotify
2097	interface to speed up C<ev_stat> watchers. Its actual availability will	2695	interface to speed up C<ev_stat> watchers. Its actual availability will
2098	be detected at runtime.	2696	be detected at runtime.
2099		2697
		2698	=item EV_ATOMIC_T
		2699
		2700	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		2701	access is atomic with respect to other threads or signal contexts. No such
		2702	type is easily found in the C language, so you can provide your own type
		2703	that you know is safe for your purposes. It is used both for signal handler "locking"
		2704	as well as for signal and thread safety in C<ev_async> watchers.
		2705
		2706	In the absense of this define, libev will use C<sig_atomic_t volatile>
		2707	(from F<signal.h>), which is usually good enough on most platforms.
		2708
2100	=item EV_H	2709	=item EV_H
2101		2710
2102	The name of the F<ev.h> header file used to include it. The default if	2711	The name of the F<ev.h> header file used to include it. The default if
2103	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	2712	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2104	can be used to virtually rename the F<ev.h> header file in case of conflicts.	2713	used to virtually rename the F<ev.h> header file in case of conflicts.
2105		2714
2106	=item EV_CONFIG_H	2715	=item EV_CONFIG_H
2107		2716
2108	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	2717	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2109	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	2718	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2110	C<EV_H>, above.	2719	C<EV_H>, above.
2111		2720
2112	=item EV_EVENT_H	2721	=item EV_EVENT_H
2113		2722
2114	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	2723	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2115	of how the F<event.h> header can be found.	2724	of how the F<event.h> header can be found, the default is C<"event.h">.
2116		2725
2117	=item EV_PROTOTYPES	2726	=item EV_PROTOTYPES
2118		2727
2119	If defined to be C<0>, then F<ev.h> will not define any function	2728	If defined to be C<0>, then F<ev.h> will not define any function
2120	prototypes, but still define all the structs and other symbols. This is	2729	prototypes, but still define all the structs and other symbols. This is
…		…
2171	=item EV_FORK_ENABLE	2780	=item EV_FORK_ENABLE
2172		2781
2173	If undefined or defined to be C<1>, then fork watchers are supported. If	2782	If undefined or defined to be C<1>, then fork watchers are supported. If
2174	defined to be C<0>, then they are not.	2783	defined to be C<0>, then they are not.
2175		2784
		2785	=item EV_ASYNC_ENABLE
		2786
		2787	If undefined or defined to be C<1>, then async watchers are supported. If
		2788	defined to be C<0>, then they are not.
		2789
2176	=item EV_MINIMAL	2790	=item EV_MINIMAL
2177		2791
2178	If you need to shave off some kilobytes of code at the expense of some	2792	If you need to shave off some kilobytes of code at the expense of some
2179	speed, define this symbol to C<1>. Currently only used for gcc to override	2793	speed, define this symbol to C<1>. Currently only used for gcc to override
2180	some inlining decisions, saves roughly 30% codesize of amd64.	2794	some inlining decisions, saves roughly 30% codesize of amd64.
…		…
2186	than enough. If you need to manage thousands of children you might want to	2800	than enough. If you need to manage thousands of children you might want to
2187	increase this value (I<must> be a power of two).	2801	increase this value (I<must> be a power of two).
2188		2802
2189	=item EV_INOTIFY_HASHSIZE	2803	=item EV_INOTIFY_HASHSIZE
2190		2804
2191	C<ev_staz> watchers use a small hash table to distribute workload by	2805	C<ev_stat> watchers use a small hash table to distribute workload by
2192	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	2806	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2193	usually more than enough. If you need to manage thousands of C<ev_stat>	2807	usually more than enough. If you need to manage thousands of C<ev_stat>
2194	watchers you might want to increase this value (I<must> be a power of	2808	watchers you might want to increase this value (I<must> be a power of
2195	two).	2809	two).
2196		2810
…		…
2213		2827
2214	=item ev_set_cb (ev, cb)	2828	=item ev_set_cb (ev, cb)
2215		2829
2216	Can be used to change the callback member declaration in each watcher,	2830	Can be used to change the callback member declaration in each watcher,
2217	and the way callbacks are invoked and set. Must expand to a struct member	2831	and the way callbacks are invoked and set. Must expand to a struct member
2218	definition and a statement, respectively. See the F<ev.v> header file for	2832	definition and a statement, respectively. See the F<ev.h> header file for
2219	their default definitions. One possible use for overriding these is to	2833	their default definitions. One possible use for overriding these is to
2220	avoid the C<struct ev_loop *> as first argument in all cases, or to use	2834	avoid the C<struct ev_loop *> as first argument in all cases, or to use
2221	method calls instead of plain function calls in C++.	2835	method calls instead of plain function calls in C++.
		2836
		2837	=head2 EXPORTED API SYMBOLS
		2838
		2839	If you need to re-export the API (e.g. via a dll) and you need a list of
		2840	exported symbols, you can use the provided F<Symbol.*> files which list
		2841	all public symbols, one per line:
		2842
		2843	Symbols.ev for libev proper
		2844	Symbols.event for the libevent emulation
		2845
		2846	This can also be used to rename all public symbols to avoid clashes with
		2847	multiple versions of libev linked together (which is obviously bad in
		2848	itself, but sometimes it is inconvinient to avoid this).
		2849
		2850	A sed command like this will create wrapper C<#define>'s that you need to
		2851	include before including F<ev.h>:
		2852
		2853	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		2854
		2855	This would create a file F<wrap.h> which essentially looks like this:
		2856
		2857	#define ev_backend myprefix_ev_backend
		2858	#define ev_check_start myprefix_ev_check_start
		2859	#define ev_check_stop myprefix_ev_check_stop
		2860	...
2222		2861
2223	=head2 EXAMPLES	2862	=head2 EXAMPLES
2224		2863
2225	For a real-world example of a program the includes libev	2864	For a real-world example of a program the includes libev
2226	verbatim, you can have a look at the EV perl module	2865	verbatim, you can have a look at the EV perl module
…		…
2267		2906
2268	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	2907	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2269		2908
2270	This means that, when you have a watcher that triggers in one hour and	2909	This means that, when you have a watcher that triggers in one hour and
2271	there are 100 watchers that would trigger before that then inserting will	2910	there are 100 watchers that would trigger before that then inserting will
2272	have to skip those 100 watchers.	2911	have to skip roughly seven (C<ld 100>) of these watchers.
2273		2912
2274	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	2913	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2275		2914
2276	That means that for changing a timer costs less than removing/adding them	2915	That means that changing a timer costs less than removing/adding them
2277	as only the relative motion in the event queue has to be paid for.	2916	as only the relative motion in the event queue has to be paid for.
2278		2917
2279	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	2918	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2280		2919
2281	These just add the watcher into an array or at the head of a list.	2920	These just add the watcher into an array or at the head of a list.
		2921
2282	=item Stopping check/prepare/idle watchers: O(1)	2922	=item Stopping check/prepare/idle/fork/async watchers: O(1)
2283		2923
2284	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	2924	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2285		2925
2286	These watchers are stored in lists then need to be walked to find the	2926	These watchers are stored in lists then need to be walked to find the
2287	correct watcher to remove. The lists are usually short (you don't usually	2927	correct watcher to remove. The lists are usually short (you don't usually
2288	have many watchers waiting for the same fd or signal).	2928	have many watchers waiting for the same fd or signal).
2289		2929
2290	=item Finding the next timer per loop iteration: O(1)	2930	=item Finding the next timer in each loop iteration: O(1)
		2931
		2932	By virtue of using a binary heap, the next timer is always found at the
		2933	beginning of the storage array.
2291		2934
2292	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	2935	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2293		2936
2294	A change means an I/O watcher gets started or stopped, which requires	2937	A change means an I/O watcher gets started or stopped, which requires
2295	libev to recalculate its status (and possibly tell the kernel).	2938	libev to recalculate its status (and possibly tell the kernel, depending
		2939	on backend and wether C<ev_io_set> was used).
2296		2940
2297	=item Activating one watcher: O(1)	2941	=item Activating one watcher (putting it into the pending state): O(1)
2298		2942
2299	=item Priority handling: O(number_of_priorities)	2943	=item Priority handling: O(number_of_priorities)
2300		2944
2301	Priorities are implemented by allocating some space for each	2945	Priorities are implemented by allocating some space for each
2302	priority. When doing priority-based operations, libev usually has to	2946	priority. When doing priority-based operations, libev usually has to
2303	linearly search all the priorities.	2947	linearly search all the priorities, but starting/stopping and activating
		2948	watchers becomes O(1) w.r.t. priority handling.
		2949
		2950	=item Sending an ev_async: O(1)
		2951
		2952	=item Processing ev_async_send: O(number_of_async_watchers)
		2953
		2954	=item Processing signals: O(max_signal_number)
		2955
		2956	Sending involves a syscall I<iff> there were no other C<ev_async_send>
		2957	calls in the current loop iteration. Checking for async and signal events
		2958	involves iterating over all running async watchers or all signal numbers.
2304		2959
2305	=back	2960	=back
2306		2961
2307		2962
		2963	=head1 Win32 platform limitations and workarounds
		2964
		2965	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		2966	requires, and its I/O model is fundamentally incompatible with the POSIX
		2967	model. Libev still offers limited functionality on this platform in
		2968	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		2969	descriptors. This only applies when using Win32 natively, not when using
		2970	e.g. cygwin.
		2971
		2972	There is no supported compilation method available on windows except
		2973	embedding it into other applications.
		2974
		2975	Due to the many, low, and arbitrary limits on the win32 platform and the
		2976	abysmal performance of winsockets, using a large number of sockets is not
		2977	recommended (and not reasonable). If your program needs to use more than
		2978	a hundred or so sockets, then likely it needs to use a totally different
		2979	implementation for windows, as libev offers the POSIX model, which cannot
		2980	be implemented efficiently on windows (microsoft monopoly games).
		2981
		2982	=over 4
		2983
		2984	=item The winsocket select function
		2985
		2986	The winsocket C<select> function doesn't follow POSIX in that it requires
		2987	socket I<handles> and not socket I<file descriptors>. This makes select
		2988	very inefficient, and also requires a mapping from file descriptors
		2989	to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
		2990	C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
		2991	symbols for more info.
		2992
		2993	The configuration for a "naked" win32 using the microsoft runtime
		2994	libraries and raw winsocket select is:
		2995
		2996	#define EV_USE_SELECT 1
		2997	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		2998
		2999	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3000	complexity in the O(n²) range when using win32.
		3001
		3002	=item Limited number of file descriptors
		3003
		3004	Windows has numerous arbitrary (and low) limits on things. Early versions
		3005	of winsocket's select only supported waiting for a max. of C<64> handles
		3006	(probably owning to the fact that all windows kernels can only wait for
		3007	C<64> things at the same time internally; microsoft recommends spawning a
		3008	chain of threads and wait for 63 handles and the previous thread in each).
		3009
		3010	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3011	to some high number (e.g. C<2048>) before compiling the winsocket select
		3012	call (which might be in libev or elsewhere, for example, perl does its own
		3013	select emulation on windows).
		3014
		3015	Another limit is the number of file descriptors in the microsoft runtime
		3016	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3017	or something like this inside microsoft). You can increase this by calling
		3018	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3019	arbitrary limit), but is broken in many versions of the microsoft runtime
		3020	libraries.
		3021
		3022	This might get you to about C<512> or C<2048> sockets (depending on
		3023	windows version and/or the phase of the moon). To get more, you need to
		3024	wrap all I/O functions and provide your own fd management, but the cost of
		3025	calling select (O(n²)) will likely make this unworkable.
		3026
		3027	=back
		3028
		3029
2308	=head1 AUTHOR	3030	=head1 AUTHOR
2309		3031
2310	Marc Lehmann <libev@schmorp.de>.	3032	Marc Lehmann <libev@schmorp.de>.
2311		3033

Diff Legend

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