[ViewVC] Diff of: cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.54 by root, Tue Nov 27 20:26:51 2007 UTC vs.
Revision 1.97 by root, Sat Dec 22 05:48:02 2007 UTC

…		…
48	return 0;	48	return 0;
49	}	49	}
50		50
51	=head1 DESCRIPTION	51	=head1 DESCRIPTION
52		52
		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
		55	time: L<http://cvs.schmorp.de/libev/ev.html>.
		56
53	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
54	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
55	these event sources and provide your program with events.	59	these event sources and provide your program with events.
56		60
57	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
58	(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
59	communicate events via a callback mechanism.	63	communicate events via a callback mechanism.
…		…
63	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
64	watcher.	68	watcher.
65		69
66	=head1 FEATURES	70	=head1 FEATURES
67		71
68	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
69	bsd-specific C<kqueue> and the solaris-specific event port mechanisms	73	BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
70	for file descriptor events (C<ev_io>), relative timers (C<ev_timer>),	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
71	absolute timers with customised rescheduling (C<ev_periodic>), synchronous	76	with customised rescheduling (C<ev_periodic>), synchronous signals
72	signals (C<ev_signal>), process status change events (C<ev_child>), and	77	(C<ev_signal>), process status change events (C<ev_child>), and event
73	event watchers dealing with the event loop mechanism itself (C<ev_idle>,	78	watchers dealing with the event loop mechanism itself (C<ev_idle>,
74	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as	79	C<ev_embed>, C<ev_prepare> and C<ev_check> watchers) as well as
75	file watchers (C<ev_stat>) and even limited support for fork events	80	file watchers (C<ev_stat>) and even limited support for fork events
76	(C<ev_fork>).	81	(C<ev_fork>).
77		82
78	It also is quite fast (see this	83	It also is quite fast (see this
…		…
93	Libev represents time as a single floating point number, representing the	98	Libev represents time as a single floating point number, representing the
94	(fractional) number of seconds since the (POSIX) epoch (somewhere near	99	(fractional) number of seconds since the (POSIX) epoch (somewhere near
95	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
96	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
97	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
98	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.
99		106
100	=head1 GLOBAL FUNCTIONS	107	=head1 GLOBAL FUNCTIONS
101		108
102	These functions can be called anytime, even before initialising the	109	These functions can be called anytime, even before initialising the
103	library in any way.	110	library in any way.
…		…
108		115
109	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
110	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
111	you actually want to know.	118	you actually want to know.
112		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
113	=item int ev_version_major ()	126	=item int ev_version_major ()
114		127
115	=item int ev_version_minor ()	128	=item int ev_version_minor ()
116		129
117	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
118	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
119	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
120	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
121	version of the library your program was compiled against.	134	version of the library your program was compiled against.
122		135
		136	These version numbers refer to the ABI version of the library, not the
		137	release version.
		138
123	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,
124	as this indicates an incompatible change. Minor versions are usually	140	as this indicates an incompatible change. Minor versions are usually
125	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
126	not a problem.	142	not a problem.
127		143
128	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
129	version.	145	version.
…		…
162	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for	178	C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
163	recommended ones.	179	recommended ones.
164		180
165	See the description of C<ev_embed> watchers for more info.	181	See the description of C<ev_embed> watchers for more info.
166		182
167	=item ev_set_allocator (void (cb)(void *ptr, size_t size))	183	=item ev_set_allocator (void (cb)(void *ptr, long size))
168		184
169	Sets the allocation function to use (the prototype and semantics are	185	Sets the allocation function to use (the prototype is similar - the
170	identical to the realloc C function). It is used to allocate and free	186	semantics is identical - to the realloc C function). It is used to
171	memory (no surprises here). If it returns zero when memory needs to be	187	allocate and free memory (no surprises here). If it returns zero when
172	allocated, the library might abort or take some potentially destructive	188	memory needs to be allocated, the library might abort or take some
173	action. The default is your system realloc function.	189	potentially destructive action. The default is your system realloc
		190	function.
174		191
175	You could override this function in high-availability programs to, say,	192	You could override this function in high-availability programs to, say,
176	free some memory if it cannot allocate memory, to use a special allocator,	193	free some memory if it cannot allocate memory, to use a special allocator,
177	or even to sleep a while and retry until some memory is available.	194	or even to sleep a while and retry until some memory is available.
178		195
…		…
264	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will	281	C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
265	override the flags completely if it is found in the environment. This is	282	override the flags completely if it is found in the environment. This is
266	useful to try out specific backends to test their performance, or to work	283	useful to try out specific backends to test their performance, or to work
267	around bugs.	284	around bugs.
268		285
		286	=item C<EVFLAG_FORKCHECK>
		287
		288	Instead of calling C<ev_default_fork> or C<ev_loop_fork> manually after
		289	a fork, you can also make libev check for a fork in each iteration by
		290	enabling this flag.
		291
		292	This works by calling C<getpid ()> on every iteration of the loop,
		293	and thus this might slow down your event loop if you do a lot of loop
		294	iterations and little real work, but is usually not noticeable (on my
		295	Linux system for example, C<getpid> is actually a simple 5-insn sequence
		296	without a syscall and thus I<very> fast, but my Linux system also has
		297	C<pthread_atfork> which is even faster).
		298
		299	The big advantage of this flag is that you can forget about fork (and
		300	forget about forgetting to tell libev about forking) when you use this
		301	flag.
		302
		303	This flag setting cannot be overriden or specified in the C<LIBEV_FLAGS>
		304	environment variable.
		305
269	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	306	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
270		307
271	This is your standard select(2) backend. Not I<completely> standard, as	308	This is your standard select(2) backend. Not I<completely> standard, as
272	libev tries to roll its own fd_set with no limits on the number of fds,	309	libev tries to roll its own fd_set with no limits on the number of fds,
273	but if that fails, expect a fairly low limit on the number of fds when	310	but if that fails, expect a fairly low limit on the number of fds when
…		…
282	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	319	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).
283		320
284	=item C<EVBACKEND_EPOLL> (value 4, Linux)	321	=item C<EVBACKEND_EPOLL> (value 4, Linux)
285		322
286	For few fds, this backend is a bit little slower than poll and select,	323	For few fds, this backend is a bit little slower than poll and select,
287	but it scales phenomenally better. While poll and select usually scale like	324	but it scales phenomenally better. While poll and select usually scale
288	O(total_fds) where n is the total number of fds (or the highest fd), epoll scales	325	like O(total_fds) where n is the total number of fds (or the highest fd),
289	either O(1) or O(active_fds).	326	epoll scales either O(1) or O(active_fds). The epoll design has a number
		327	of shortcomings, such as silently dropping events in some hard-to-detect
		328	cases and rewiring a syscall per fd change, no fork support and bad
		329	support for dup:
290		330
291	While stopping and starting an I/O watcher in the same iteration will	331	While stopping, setting and starting an I/O watcher in the same iteration
292	result in some caching, there is still a syscall per such incident	332	will result in some caching, there is still a syscall per such incident
293	(because the fd could point to a different file description now), so its	333	(because the fd could point to a different file description now), so its
294	best to avoid that. Also, dup()ed file descriptors might not work very	334	best to avoid that. Also, C<dup ()>'ed file descriptors might not work
295	well if you register events for both fds.	335	very well if you register events for both fds.
296		336
297	Please note that epoll sometimes generates spurious notifications, so you	337	Please note that epoll sometimes generates spurious notifications, so you
298	need to use non-blocking I/O or other means to avoid blocking when no data	338	need to use non-blocking I/O or other means to avoid blocking when no data
299	(or space) is available.	339	(or space) is available.
300		340
301	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	341	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
302		342
303	Kqueue deserves special mention, as at the time of this writing, it	343	Kqueue deserves special mention, as at the time of this writing, it
304	was broken on all BSDs except NetBSD (usually it doesn't work with	344	was broken on I<all> BSDs (usually it doesn't work with anything but
305	anything but sockets and pipes, except on Darwin, where of course its	345	sockets and pipes, except on Darwin, where of course it's completely
		346	useless. On NetBSD, it seems to work for all the FD types I tested, so it
306	completely useless). For this reason its not being "autodetected"	347	is used by default there). For this reason it's not being "autodetected"
307	unless you explicitly specify it explicitly in the flags (i.e. using	348	unless you explicitly specify it explicitly in the flags (i.e. using
308	C<EVBACKEND_KQUEUE>).	349	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
		350	system like NetBSD.
309		351
310	It scales in the same way as the epoll backend, but the interface to the	352	It scales in the same way as the epoll backend, but the interface to the
311	kernel is more efficient (which says nothing about its actual speed, of	353	kernel is more efficient (which says nothing about its actual speed,
312	course). While starting and stopping an I/O watcher does not cause an	354	of course). While stopping, setting and starting an I/O watcher does
313	extra syscall as with epoll, it still adds up to four event changes per	355	never cause an extra syscall as with epoll, it still adds up to two event
314	incident, so its best to avoid that.	356	changes per incident, support for C<fork ()> is very bad and it drops fds
		357	silently in similarly hard-to-detetc cases.
315		358
316	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	359	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
317		360
318	This is not implemented yet (and might never be).	361	This is not implemented yet (and might never be).
319		362
320	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	363	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
321		364
322	This uses the Solaris 10 port mechanism. As with everything on Solaris,	365	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
323	it's really slow, but it still scales very well (O(active_fds)).	366	it's really slow, but it still scales very well (O(active_fds)).
324		367
325	Please note that solaris ports can result in a lot of spurious	368	Please note that solaris event ports can deliver a lot of spurious
326	notifications, so you need to use non-blocking I/O or other means to avoid	369	notifications, so you need to use non-blocking I/O or other means to avoid
327	blocking when no data (or space) is available.	370	blocking when no data (or space) is available.
328		371
329	=item C<EVBACKEND_ALL>	372	=item C<EVBACKEND_ALL>
330		373
…		…
373	Destroys the default loop again (frees all memory and kernel state	416	Destroys the default loop again (frees all memory and kernel state
374	etc.). None of the active event watchers will be stopped in the normal	417	etc.). None of the active event watchers will be stopped in the normal
375	sense, so e.g. C<ev_is_active> might still return true. It is your	418	sense, so e.g. C<ev_is_active> might still return true. It is your
376	responsibility to either stop all watchers cleanly yoursef I<before>	419	responsibility to either stop all watchers cleanly yoursef I<before>
377	calling this function, or cope with the fact afterwards (which is usually	420	calling this function, or cope with the fact afterwards (which is usually
378	the easiest thing, youc na just ignore the watchers and/or C<free ()> them	421	the easiest thing, you can just ignore the watchers and/or C<free ()> them
379	for example).	422	for example).
		423
		424	Note that certain global state, such as signal state, will not be freed by
		425	this function, and related watchers (such as signal and child watchers)
		426	would need to be stopped manually.
		427
		428	In general it is not advisable to call this function except in the
		429	rare occasion where you really need to free e.g. the signal handling
		430	pipe fds. If you need dynamically allocated loops it is better to use
		431	C<ev_loop_new> and C<ev_loop_destroy>).
380		432
381	=item ev_loop_destroy (loop)	433	=item ev_loop_destroy (loop)
382		434
383	Like C<ev_default_destroy>, but destroys an event loop created by an	435	Like C<ev_default_destroy>, but destroys an event loop created by an
384	earlier call to C<ev_loop_new>.	436	earlier call to C<ev_loop_new>.
…		…
408		460
409	Like C<ev_default_fork>, but acts on an event loop created by	461	Like C<ev_default_fork>, but acts on an event loop created by
410	C<ev_loop_new>. Yes, you have to call this on every allocated event loop	462	C<ev_loop_new>. Yes, you have to call this on every allocated event loop
411	after fork, and how you do this is entirely your own problem.	463	after fork, and how you do this is entirely your own problem.
412		464
		465	=item unsigned int ev_loop_count (loop)
		466
		467	Returns the count of loop iterations for the loop, which is identical to
		468	the number of times libev did poll for new events. It starts at C<0> and
		469	happily wraps around with enough iterations.
		470
		471	This value can sometimes be useful as a generation counter of sorts (it
		472	"ticks" the number of loop iterations), as it roughly corresponds with
		473	C<ev_prepare> and C<ev_check> calls.
		474
413	=item unsigned int ev_backend (loop)	475	=item unsigned int ev_backend (loop)
414		476
415	Returns one of the C<EVBACKEND_*> flags indicating the event backend in	477	Returns one of the C<EVBACKEND_*> flags indicating the event backend in
416	use.	478	use.
417		479
…		…
419		481
420	Returns the current "event loop time", which is the time the event loop	482	Returns the current "event loop time", which is the time the event loop
421	received events and started processing them. This timestamp does not	483	received events and started processing them. This timestamp does not
422	change as long as callbacks are being processed, and this is also the base	484	change as long as callbacks are being processed, and this is also the base
423	time used for relative timers. You can treat it as the timestamp of the	485	time used for relative timers. You can treat it as the timestamp of the
424	event occuring (or more correctly, libev finding out about it).	486	event occurring (or more correctly, libev finding out about it).
425		487
426	=item ev_loop (loop, int flags)	488	=item ev_loop (loop, int flags)
427		489
428	Finally, this is it, the event handler. This function usually is called	490	Finally, this is it, the event handler. This function usually is called
429	after you initialised all your watchers and you want to start handling	491	after you initialised all your watchers and you want to start handling
…		…
450	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is	512	libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is
451	usually a better approach for this kind of thing.	513	usually a better approach for this kind of thing.
452		514
453	Here are the gory details of what C<ev_loop> does:	515	Here are the gory details of what C<ev_loop> does:
454		516
		517	- Before the first iteration, call any pending watchers.
455	* If there are no active watchers (reference count is zero), return.	518	* If there are no active watchers (reference count is zero), return.
456	- Queue prepare watchers and then call all outstanding watchers.	519	- Queue all prepare watchers and then call all outstanding watchers.
457	- If we have been forked, recreate the kernel state.	520	- If we have been forked, recreate the kernel state.
458	- Update the kernel state with all outstanding changes.	521	- Update the kernel state with all outstanding changes.
459	- Update the "event loop time".	522	- Update the "event loop time".
460	- Calculate for how long to block.	523	- Calculate for how long to block.
461	- Block the process, waiting for any events.	524	- Block the process, waiting for any events.
…		…
512	Example: For some weird reason, unregister the above signal handler again.	575	Example: For some weird reason, unregister the above signal handler again.
513		576
514	ev_ref (loop);	577	ev_ref (loop);
515	ev_signal_stop (loop, &exitsig);	578	ev_signal_stop (loop, &exitsig);
516		579
		580	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		581
		582	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		583
		584	These advanced functions influence the time that libev will spend waiting
		585	for events. Both are by default C<0>, meaning that libev will try to
		586	invoke timer/periodic callbacks and I/O callbacks with minimum latency.
		587
		588	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		589	allows libev to delay invocation of I/O and timer/periodic callbacks to
		590	increase efficiency of loop iterations.
		591
		592	The background is that sometimes your program runs just fast enough to
		593	handle one (or very few) event(s) per loop iteration. While this makes
		594	the program responsive, it also wastes a lot of CPU time to poll for new
		595	events, especially with backends like C<select ()> which have a high
		596	overhead for the actual polling but can deliver many events at once.
		597
		598	By setting a higher I<io collect interval> you allow libev to spend more
		599	time collecting I/O events, so you can handle more events per iteration,
		600	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		601	C<ev_timer>) will be not affected.
		602
		603	Likewise, by setting a higher I<timeout collect interval> you allow libev
		604	to spend more time collecting timeouts, at the expense of increased
		605	latency (the watcher callback will be called later). C<ev_io> watchers
		606	will not be affected.
		607
		608	Many programs can usually benefit by setting the io collect interval to
		609	a value near C<0.1> or so, which is often enough for interactive servers
		610	(of course not for games), likewise for timeouts. It usually doesn't make
		611	much sense to set it to a lower value than C<0.01>, as this approsaches
		612	the timing granularity of most systems.
		613
517	=back	614	=back
518		615
519		616
520	=head1 ANATOMY OF A WATCHER	617	=head1 ANATOMY OF A WATCHER
521		618
…		…
700	=item bool ev_is_pending (ev_TYPE *watcher)	797	=item bool ev_is_pending (ev_TYPE *watcher)
701		798
702	Returns a true value iff the watcher is pending, (i.e. it has outstanding	799	Returns a true value iff the watcher is pending, (i.e. it has outstanding
703	events but its callback has not yet been invoked). As long as a watcher	800	events but its callback has not yet been invoked). As long as a watcher
704	is pending (but not active) you must not call an init function on it (but	801	is pending (but not active) you must not call an init function on it (but
705	C<ev_TYPE_set> is safe) and you must make sure the watcher is available to	802	C<ev_TYPE_set> is safe), you must not change its priority, and you must
706	libev (e.g. you cnanot C<free ()> it).	803	make sure the watcher is available to libev (e.g. you cannot C<free ()>
		804	it).
707		805
708	=item callback = ev_cb (ev_TYPE *watcher)	806	=item callback ev_cb (ev_TYPE *watcher)
709		807
710	Returns the callback currently set on the watcher.	808	Returns the callback currently set on the watcher.
711		809
712	=item ev_cb_set (ev_TYPE *watcher, callback)	810	=item ev_cb_set (ev_TYPE *watcher, callback)
713		811
714	Change the callback. You can change the callback at virtually any time	812	Change the callback. You can change the callback at virtually any time
715	(modulo threads).	813	(modulo threads).
		814
		815	=item ev_set_priority (ev_TYPE *watcher, priority)
		816
		817	=item int ev_priority (ev_TYPE *watcher)
		818
		819	Set and query the priority of the watcher. The priority is a small
		820	integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
		821	(default: C<-2>). Pending watchers with higher priority will be invoked
		822	before watchers with lower priority, but priority will not keep watchers
		823	from being executed (except for C<ev_idle> watchers).
		824
		825	This means that priorities are I<only> used for ordering callback
		826	invocation after new events have been received. This is useful, for
		827	example, to reduce latency after idling, or more often, to bind two
		828	watchers on the same event and make sure one is called first.
		829
		830	If you need to suppress invocation when higher priority events are pending
		831	you need to look at C<ev_idle> watchers, which provide this functionality.
		832
		833	You I<must not> change the priority of a watcher as long as it is active or
		834	pending.
		835
		836	The default priority used by watchers when no priority has been set is
		837	always C<0>, which is supposed to not be too high and not be too low :).
		838
		839	Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
		840	fine, as long as you do not mind that the priority value you query might
		841	or might not have been adjusted to be within valid range.
		842
		843	=item ev_invoke (loop, ev_TYPE *watcher, int revents)
		844
		845	Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
		846	C<loop> nor C<revents> need to be valid as long as the watcher callback
		847	can deal with that fact.
		848
		849	=item int ev_clear_pending (loop, ev_TYPE *watcher)
		850
		851	If the watcher is pending, this function returns clears its pending status
		852	and returns its C<revents> bitset (as if its callback was invoked). If the
		853	watcher isn't pending it does nothing and returns C<0>.
716		854
717	=back	855	=back
718		856
719		857
720	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	858	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
…		…
741	{	879	{
742	struct my_io w = (struct my_io )w_;	880	struct my_io w = (struct my_io )w_;
743	...	881	...
744	}	882	}
745		883
746	More interesting and less C-conformant ways of catsing your callback type	884	More interesting and less C-conformant ways of casting your callback type
747	have been omitted....	885	instead have been omitted.
		886
		887	Another common scenario is having some data structure with multiple
		888	watchers:
		889
		890	struct my_biggy
		891	{
		892	int some_data;
		893	ev_timer t1;
		894	ev_timer t2;
		895	}
		896
		897	In this case getting the pointer to C<my_biggy> is a bit more complicated,
		898	you need to use C<offsetof>:
		899
		900	#include <stddef.h>
		901
		902	static void
		903	t1_cb (EV_P_ struct ev_timer *w, int revents)
		904	{
		905	struct my_biggy big = (struct my_biggy *
		906	(((char *)w) - offsetof (struct my_biggy, t1));
		907	}
		908
		909	static void
		910	t2_cb (EV_P_ struct ev_timer *w, int revents)
		911	{
		912	struct my_biggy big = (struct my_biggy *
		913	(((char *)w) - offsetof (struct my_biggy, t2));
		914	}
748		915
749		916
750	=head1 WATCHER TYPES	917	=head1 WATCHER TYPES
751		918
752	This section describes each watcher in detail, but will not repeat	919	This section describes each watcher in detail, but will not repeat
…		…
797	it is best to always use non-blocking I/O: An extra C<read>(2) returning	964	it is best to always use non-blocking I/O: An extra C<read>(2) returning
798	C<EAGAIN> is far preferable to a program hanging until some data arrives.	965	C<EAGAIN> is far preferable to a program hanging until some data arrives.
799		966
800	If you cannot run the fd in non-blocking mode (for example you should not	967	If you cannot run the fd in non-blocking mode (for example you should not
801	play around with an Xlib connection), then you have to seperately re-test	968	play around with an Xlib connection), then you have to seperately re-test
802	wether a file descriptor is really ready with a known-to-be good interface	969	whether a file descriptor is really ready with a known-to-be good interface
803	such as poll (fortunately in our Xlib example, Xlib already does this on	970	such as poll (fortunately in our Xlib example, Xlib already does this on
804	its own, so its quite safe to use).	971	its own, so its quite safe to use).
		972
		973	=head3 The special problem of disappearing file descriptors
		974
		975	Some backends (e.g. kqueue, epoll) need to be told about closing a file
		976	descriptor (either by calling C<close> explicitly or by any other means,
		977	such as C<dup>). The reason is that you register interest in some file
		978	descriptor, but when it goes away, the operating system will silently drop
		979	this interest. If another file descriptor with the same number then is
		980	registered with libev, there is no efficient way to see that this is, in
		981	fact, a different file descriptor.
		982
		983	To avoid having to explicitly tell libev about such cases, libev follows
		984	the following policy: Each time C<ev_io_set> is being called, libev
		985	will assume that this is potentially a new file descriptor, otherwise
		986	it is assumed that the file descriptor stays the same. That means that
		987	you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
		988	descriptor even if the file descriptor number itself did not change.
		989
		990	This is how one would do it normally anyway, the important point is that
		991	the libev application should not optimise around libev but should leave
		992	optimisations to libev.
		993
		994	=head3 The special problem of dup'ed file descriptors
		995
		996	Some backends (e.g. epoll), cannot register events for file descriptors,
		997	but only events for the underlying file descriptions. That menas when you
		998	have C<dup ()>'ed file descriptors and register events for them, only one
		999	file descriptor might actually receive events.
		1000
		1001	There is no workaorund possible except not registering events
		1002	for potentially C<dup ()>'ed file descriptors or to resort to
		1003	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
		1004
		1005	=head3 The special problem of fork
		1006
		1007	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
		1008	useless behaviour. Libev fully supports fork, but needs to be told about
		1009	it in the child.
		1010
		1011	To support fork in your programs, you either have to call
		1012	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
		1013	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
		1014	C<EVBACKEND_POLL>.
		1015
		1016
		1017	=head3 Watcher-Specific Functions
805		1018
806	=over 4	1019	=over 4
807		1020
808	=item ev_io_init (ev_io *, callback, int fd, int events)	1021	=item ev_io_init (ev_io *, callback, int fd, int events)
809		1022
…		…
863		1076
864	The callback is guarenteed to be invoked only when its timeout has passed,	1077	The callback is guarenteed to be invoked only when its timeout has passed,
865	but if multiple timers become ready during the same loop iteration then	1078	but if multiple timers become ready during the same loop iteration then
866	order of execution is undefined.	1079	order of execution is undefined.
867		1080
		1081	=head3 Watcher-Specific Functions and Data Members
		1082
868	=over 4	1083	=over 4
869		1084
870	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1085	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
871		1086
872	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1087	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
…		…
885	=item ev_timer_again (loop)	1100	=item ev_timer_again (loop)
886		1101
887	This will act as if the timer timed out and restart it again if it is	1102	This will act as if the timer timed out and restart it again if it is
888	repeating. The exact semantics are:	1103	repeating. The exact semantics are:
889		1104
		1105	If the timer is pending, its pending status is cleared.
		1106
890	If the timer is started but nonrepeating, stop it.	1107	If the timer is started but nonrepeating, stop it (as if it timed out).
891		1108
892	If the timer is repeating, either start it if necessary (with the repeat	1109	If the timer is repeating, either start it if necessary (with the
893	value), or reset the running timer to the repeat value.	1110	C<repeat> value), or reset the running timer to the C<repeat> value.
894		1111
895	This sounds a bit complicated, but here is a useful and typical	1112	This sounds a bit complicated, but here is a useful and typical
896	example: Imagine you have a tcp connection and you want a so-called	1113	example: Imagine you have a tcp connection and you want a so-called idle
897	idle timeout, that is, you want to be called when there have been,	1114	timeout, that is, you want to be called when there have been, say, 60
898	say, 60 seconds of inactivity on the socket. The easiest way to do	1115	seconds of inactivity on the socket. The easiest way to do this is to
899	this is to configure an C<ev_timer> with C<after>=C<repeat>=C<60> and calling	1116	configure an C<ev_timer> with a C<repeat> value of C<60> and then call
900	C<ev_timer_again> each time you successfully read or write some data. If	1117	C<ev_timer_again> each time you successfully read or write some data. If
901	you go into an idle state where you do not expect data to travel on the	1118	you go into an idle state where you do not expect data to travel on the
902	socket, you can stop the timer, and again will automatically restart it if	1119	socket, you can C<ev_timer_stop> the timer, and C<ev_timer_again> will
903	need be.	1120	automatically restart it if need be.
904		1121
905	You can also ignore the C<after> value and C<ev_timer_start> altogether	1122	That means you can ignore the C<after> value and C<ev_timer_start>
906	and only ever use the C<repeat> value:	1123	altogether and only ever use the C<repeat> value and C<ev_timer_again>:
907		1124
908	ev_timer_init (timer, callback, 0., 5.);	1125	ev_timer_init (timer, callback, 0., 5.);
909	ev_timer_again (loop, timer);	1126	ev_timer_again (loop, timer);
910	...	1127	...
911	timer->again = 17.;	1128	timer->again = 17.;
912	ev_timer_again (loop, timer);	1129	ev_timer_again (loop, timer);
913	...	1130	...
914	timer->again = 10.;	1131	timer->again = 10.;
915	ev_timer_again (loop, timer);	1132	ev_timer_again (loop, timer);
916		1133
917	This is more efficient then stopping/starting the timer eahc time you want	1134	This is more slightly efficient then stopping/starting the timer each time
918	to modify its timeout value.	1135	you want to modify its timeout value.
919		1136
920	=item ev_tstamp repeat [read-write]	1137	=item ev_tstamp repeat [read-write]
921		1138
922	The current C<repeat> value. Will be used each time the watcher times out	1139	The current C<repeat> value. Will be used each time the watcher times out
923	or C<ev_timer_again> is called and determines the next timeout (if any),	1140	or C<ev_timer_again> is called and determines the next timeout (if any),
…		…
965	but on wallclock time (absolute time). You can tell a periodic watcher	1182	but on wallclock time (absolute time). You can tell a periodic watcher
966	to trigger "at" some specific point in time. For example, if you tell a	1183	to trigger "at" some specific point in time. For example, if you tell a
967	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	1184	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()
968	+ 10.>) and then reset your system clock to the last year, then it will	1185	+ 10.>) and then reset your system clock to the last year, then it will
969	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1186	take a year to trigger the event (unlike an C<ev_timer>, which would trigger
970	roughly 10 seconds later and of course not if you reset your system time	1187	roughly 10 seconds later).
971	again).
972		1188
973	They can also be used to implement vastly more complex timers, such as	1189	They can also be used to implement vastly more complex timers, such as
974	triggering an event on eahc midnight, local time.	1190	triggering an event on each midnight, local time or other, complicated,
		1191	rules.
975		1192
976	As with timers, the callback is guarenteed to be invoked only when the	1193	As with timers, the callback is guarenteed to be invoked only when the
977	time (C<at>) has been passed, but if multiple periodic timers become ready	1194	time (C<at>) has been passed, but if multiple periodic timers become ready
978	during the same loop iteration then order of execution is undefined.	1195	during the same loop iteration then order of execution is undefined.
979		1196
		1197	=head3 Watcher-Specific Functions and Data Members
		1198
980	=over 4	1199	=over 4
981		1200
982	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)	1201	=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
983		1202
984	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)	1203	=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
…		…
986	Lots of arguments, lets sort it out... There are basically three modes of	1205	Lots of arguments, lets sort it out... There are basically three modes of
987	operation, and we will explain them from simplest to complex:	1206	operation, and we will explain them from simplest to complex:
988		1207
989	=over 4	1208	=over 4
990		1209
991	=item * absolute timer (interval = reschedule_cb = 0)	1210	=item * absolute timer (at = time, interval = reschedule_cb = 0)
992		1211
993	In this configuration the watcher triggers an event at the wallclock time	1212	In this configuration the watcher triggers an event at the wallclock time
994	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1213	C<at> and doesn't repeat. It will not adjust when a time jump occurs,
995	that is, if it is to be run at January 1st 2011 then it will run when the	1214	that is, if it is to be run at January 1st 2011 then it will run when the
996	system time reaches or surpasses this time.	1215	system time reaches or surpasses this time.
997		1216
998	=item * non-repeating interval timer (interval > 0, reschedule_cb = 0)	1217	=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
999		1218
1000	In this mode the watcher will always be scheduled to time out at the next	1219	In this mode the watcher will always be scheduled to time out at the next
1001	C<at + N * interval> time (for some integer N) and then repeat, regardless	1220	C<at + N * interval> time (for some integer N, which can also be negative)
1002	of any time jumps.	1221	and then repeat, regardless of any time jumps.
1003		1222
1004	This can be used to create timers that do not drift with respect to system	1223	This can be used to create timers that do not drift with respect to system
1005	time:	1224	time:
1006		1225
1007	ev_periodic_set (&periodic, 0., 3600., 0);	1226	ev_periodic_set (&periodic, 0., 3600., 0);
…		…
1013		1232
1014	Another way to think about it (for the mathematically inclined) is that	1233	Another way to think about it (for the mathematically inclined) is that
1015	C<ev_periodic> will try to run the callback in this mode at the next possible	1234	C<ev_periodic> will try to run the callback in this mode at the next possible
1016	time where C<time = at (mod interval)>, regardless of any time jumps.	1235	time where C<time = at (mod interval)>, regardless of any time jumps.
1017		1236
		1237	For numerical stability it is preferable that the C<at> value is near
		1238	C<ev_now ()> (the current time), but there is no range requirement for
		1239	this value.
		1240
1018	=item * manual reschedule mode (reschedule_cb = callback)	1241	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1019		1242
1020	In this mode the values for C<interval> and C<at> are both being	1243	In this mode the values for C<interval> and C<at> are both being
1021	ignored. Instead, each time the periodic watcher gets scheduled, the	1244	ignored. Instead, each time the periodic watcher gets scheduled, the
1022	reschedule callback will be called with the watcher as first, and the	1245	reschedule callback will be called with the watcher as first, and the
1023	current time as second argument.	1246	current time as second argument.
1024		1247
1025	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1248	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
1026	ever, or make any event loop modifications>. If you need to stop it,	1249	ever, or make any event loop modifications>. If you need to stop it,
1027	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by	1250	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by
1028	starting a prepare watcher).	1251	starting an C<ev_prepare> watcher, which is legal).
1029		1252
1030	Its prototype is C<ev_tstamp (reschedule_cb)(struct ev_periodic w,	1253	Its prototype is C<ev_tstamp (reschedule_cb)(struct ev_periodic w,
1031	ev_tstamp now)>, e.g.:	1254	ev_tstamp now)>, e.g.:
1032		1255
1033	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1256	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
…		…
1056	Simply stops and restarts the periodic watcher again. This is only useful	1279	Simply stops and restarts the periodic watcher again. This is only useful
1057	when you changed some parameters or the reschedule callback would return	1280	when you changed some parameters or the reschedule callback would return
1058	a different time than the last time it was called (e.g. in a crond like	1281	a different time than the last time it was called (e.g. in a crond like
1059	program when the crontabs have changed).	1282	program when the crontabs have changed).
1060		1283
		1284	=item ev_tstamp offset [read-write]
		1285
		1286	When repeating, this contains the offset value, otherwise this is the
		1287	absolute point in time (the C<at> value passed to C<ev_periodic_set>).
		1288
		1289	Can be modified any time, but changes only take effect when the periodic
		1290	timer fires or C<ev_periodic_again> is being called.
		1291
1061	=item ev_tstamp interval [read-write]	1292	=item ev_tstamp interval [read-write]
1062		1293
1063	The current interval value. Can be modified any time, but changes only	1294	The current interval value. Can be modified any time, but changes only
1064	take effect when the periodic timer fires or C<ev_periodic_again> is being	1295	take effect when the periodic timer fires or C<ev_periodic_again> is being
1065	called.	1296	called.
…		…
1067	=item ev_tstamp (reschedule_cb)(struct ev_periodic w, ev_tstamp now) [read-write]	1298	=item ev_tstamp (reschedule_cb)(struct ev_periodic w, ev_tstamp now) [read-write]
1068		1299
1069	The current reschedule callback, or C<0>, if this functionality is	1300	The current reschedule callback, or C<0>, if this functionality is
1070	switched off. Can be changed any time, but changes only take effect when	1301	switched off. Can be changed any time, but changes only take effect when
1071	the periodic timer fires or C<ev_periodic_again> is being called.	1302	the periodic timer fires or C<ev_periodic_again> is being called.
		1303
		1304	=item ev_tstamp at [read-only]
		1305
		1306	When active, contains the absolute time that the watcher is supposed to
		1307	trigger next.
1072		1308
1073	=back	1309	=back
1074		1310
1075	Example: Call a callback every hour, or, more precisely, whenever the	1311	Example: Call a callback every hour, or, more precisely, whenever the
1076	system clock is divisible by 3600. The callback invocation times have	1312	system clock is divisible by 3600. The callback invocation times have
…		…
1118	with the kernel (thus it coexists with your own signal handlers as long	1354	with the kernel (thus it coexists with your own signal handlers as long
1119	as you don't register any with libev). Similarly, when the last signal	1355	as you don't register any with libev). Similarly, when the last signal
1120	watcher for a signal is stopped libev will reset the signal handler to	1356	watcher for a signal is stopped libev will reset the signal handler to
1121	SIG_DFL (regardless of what it was set to before).	1357	SIG_DFL (regardless of what it was set to before).
1122		1358
		1359	=head3 Watcher-Specific Functions and Data Members
		1360
1123	=over 4	1361	=over 4
1124		1362
1125	=item ev_signal_init (ev_signal *, callback, int signum)	1363	=item ev_signal_init (ev_signal *, callback, int signum)
1126		1364
1127	=item ev_signal_set (ev_signal *, int signum)	1365	=item ev_signal_set (ev_signal *, int signum)
…		…
1138		1376
1139	=head2 C<ev_child> - watch out for process status changes	1377	=head2 C<ev_child> - watch out for process status changes
1140		1378
1141	Child watchers trigger when your process receives a SIGCHLD in response to	1379	Child watchers trigger when your process receives a SIGCHLD in response to
1142	some child status changes (most typically when a child of yours dies).	1380	some child status changes (most typically when a child of yours dies).
		1381
		1382	=head3 Watcher-Specific Functions and Data Members
1143		1383
1144	=over 4	1384	=over 4
1145		1385
1146	=item ev_child_init (ev_child *, callback, int pid)	1386	=item ev_child_init (ev_child *, callback, int pid)
1147		1387
…		…
1192	not exist" is a status change like any other. The condition "path does	1432	not exist" is a status change like any other. The condition "path does
1193	not exist" is signified by the C<st_nlink> field being zero (which is	1433	not exist" is signified by the C<st_nlink> field being zero (which is
1194	otherwise always forced to be at least one) and all the other fields of	1434	otherwise always forced to be at least one) and all the other fields of
1195	the stat buffer having unspecified contents.	1435	the stat buffer having unspecified contents.
1196		1436
		1437	The path I<should> be absolute and I<must not> end in a slash. If it is
		1438	relative and your working directory changes, the behaviour is undefined.
		1439
1197	Since there is no standard to do this, the portable implementation simply	1440	Since there is no standard to do this, the portable implementation simply
1198	calls C<stat (2)> regulalry on the path to see if it changed somehow. You	1441	calls C<stat (2)> regularly on the path to see if it changed somehow. You
1199	can specify a recommended polling interval for this case. If you specify	1442	can specify a recommended polling interval for this case. If you specify
1200	a polling interval of C<0> (highly recommended!) then a I<suitable,	1443	a polling interval of C<0> (highly recommended!) then a I<suitable,
1201	unspecified default> value will be used (which you can expect to be around	1444	unspecified default> value will be used (which you can expect to be around
1202	five seconds, although this might change dynamically). Libev will also	1445	five seconds, although this might change dynamically). Libev will also
1203	impose a minimum interval which is currently around C<0.1>, but thats	1446	impose a minimum interval which is currently around C<0.1>, but thats
…		…
1205		1448
1206	This watcher type is not meant for massive numbers of stat watchers,	1449	This watcher type is not meant for massive numbers of stat watchers,
1207	as even with OS-supported change notifications, this can be	1450	as even with OS-supported change notifications, this can be
1208	resource-intensive.	1451	resource-intensive.
1209		1452
1210	At the time of this writing, no specific OS backends are implemented, but	1453	At the time of this writing, only the Linux inotify interface is
1211	if demand increases, at least a kqueue and inotify backend will be added.	1454	implemented (implementing kqueue support is left as an exercise for the
		1455	reader). Inotify will be used to give hints only and should not change the
		1456	semantics of C<ev_stat> watchers, which means that libev sometimes needs
		1457	to fall back to regular polling again even with inotify, but changes are
		1458	usually detected immediately, and if the file exists there will be no
		1459	polling.
		1460
		1461	=head3 Watcher-Specific Functions and Data Members
1212		1462
1213	=over 4	1463	=over 4
1214		1464
1215	=item ev_stat_init (ev_stat , callback, const char path, ev_tstamp interval)	1465	=item ev_stat_init (ev_stat , callback, const char path, ev_tstamp interval)
1216		1466
…		…
1280	ev_stat_start (loop, &passwd);	1530	ev_stat_start (loop, &passwd);
1281		1531
1282		1532
1283	=head2 C<ev_idle> - when you've got nothing better to do...	1533	=head2 C<ev_idle> - when you've got nothing better to do...
1284		1534
1285	Idle watchers trigger events when there are no other events are pending	1535	Idle watchers trigger events when no other events of the same or higher
1286	(prepare, check and other idle watchers do not count). That is, as long	1536	priority are pending (prepare, check and other idle watchers do not
1287	as your process is busy handling sockets or timeouts (or even signals,	1537	count).
1288	imagine) it will not be triggered. But when your process is idle all idle	1538
1289	watchers are being called again and again, once per event loop iteration -	1539	That is, as long as your process is busy handling sockets or timeouts
		1540	(or even signals, imagine) of the same or higher priority it will not be
		1541	triggered. But when your process is idle (or only lower-priority watchers
		1542	are pending), the idle watchers are being called once per event loop
1290	until stopped, that is, or your process receives more events and becomes	1543	iteration - until stopped, that is, or your process receives more events
1291	busy.	1544	and becomes busy again with higher priority stuff.
1292		1545
1293	The most noteworthy effect is that as long as any idle watchers are	1546	The most noteworthy effect is that as long as any idle watchers are
1294	active, the process will not block when waiting for new events.	1547	active, the process will not block when waiting for new events.
1295		1548
1296	Apart from keeping your process non-blocking (which is a useful	1549	Apart from keeping your process non-blocking (which is a useful
1297	effect on its own sometimes), idle watchers are a good place to do	1550	effect on its own sometimes), idle watchers are a good place to do
1298	"pseudo-background processing", or delay processing stuff to after the	1551	"pseudo-background processing", or delay processing stuff to after the
1299	event loop has handled all outstanding events.	1552	event loop has handled all outstanding events.
		1553
		1554	=head3 Watcher-Specific Functions and Data Members
1300		1555
1301	=over 4	1556	=over 4
1302		1557
1303	=item ev_idle_init (ev_signal *, callback)	1558	=item ev_idle_init (ev_signal *, callback)
1304		1559
…		…
1362	with priority higher than or equal to the event loop and one coroutine	1617	with priority higher than or equal to the event loop and one coroutine
1363	of lower priority, but only once, using idle watchers to keep the event	1618	of lower priority, but only once, using idle watchers to keep the event
1364	loop from blocking if lower-priority coroutines are active, thus mapping	1619	loop from blocking if lower-priority coroutines are active, thus mapping
1365	low-priority coroutines to idle/background tasks).	1620	low-priority coroutines to idle/background tasks).
1366		1621
		1622	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
		1623	priority, to ensure that they are being run before any other watchers
		1624	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
		1625	too) should not activate ("feed") events into libev. While libev fully
		1626	supports this, they will be called before other C<ev_check> watchers did
		1627	their job. As C<ev_check> watchers are often used to embed other event
		1628	loops those other event loops might be in an unusable state until their
		1629	C<ev_check> watcher ran (always remind yourself to coexist peacefully with
		1630	others).
		1631
		1632	=head3 Watcher-Specific Functions and Data Members
		1633
1367	=over 4	1634	=over 4
1368		1635
1369	=item ev_prepare_init (ev_prepare *, callback)	1636	=item ev_prepare_init (ev_prepare *, callback)
1370		1637
1371	=item ev_check_init (ev_check *, callback)	1638	=item ev_check_init (ev_check *, callback)
…		…
1374	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	1641	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1375	macros, but using them is utterly, utterly and completely pointless.	1642	macros, but using them is utterly, utterly and completely pointless.
1376		1643
1377	=back	1644	=back
1378		1645
1379	Example: To include a library such as adns, you would add IO watchers	1646	There are a number of principal ways to embed other event loops or modules
1380	and a timeout watcher in a prepare handler, as required by libadns, and	1647	into libev. Here are some ideas on how to include libadns into libev
		1648	(there is a Perl module named C<EV::ADNS> that does this, which you could
		1649	use for an actually working example. Another Perl module named C<EV::Glib>
		1650	embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV
		1651	into the Glib event loop).
		1652
		1653	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1381	in a check watcher, destroy them and call into libadns. What follows is	1654	and in a check watcher, destroy them and call into libadns. What follows
1382	pseudo-code only of course:	1655	is pseudo-code only of course. This requires you to either use a low
		1656	priority for the check watcher or use C<ev_clear_pending> explicitly, as
		1657	the callbacks for the IO/timeout watchers might not have been called yet.
1383		1658
1384	static ev_io iow [nfd];	1659	static ev_io iow [nfd];
1385	static ev_timer tw;	1660	static ev_timer tw;
1386		1661
1387	static void	1662	static void
1388	io_cb (ev_loop loop, ev_io w, int revents)	1663	io_cb (ev_loop loop, ev_io w, int revents)
1389	{	1664	{
1390	// set the relevant poll flags
1391	// could also call adns_processreadable etc. here
1392	struct pollfd fd = (struct pollfd )w->data;
1393	if (revents & EV_READ ) fd->revents \|= fd->events & POLLIN;
1394	if (revents & EV_WRITE) fd->revents \|= fd->events & POLLOUT;
1395	}	1665	}
1396		1666
1397	// create io watchers for each fd and a timer before blocking	1667	// create io watchers for each fd and a timer before blocking
1398	static void	1668	static void
1399	adns_prepare_cb (ev_loop loop, ev_prepare w, int revents)	1669	adns_prepare_cb (ev_loop loop, ev_prepare w, int revents)
1400	{	1670	{
1401	int timeout = 3600000;truct pollfd fds [nfd];	1671	int timeout = 3600000;
		1672	struct pollfd fds [nfd];
1402	// actual code will need to loop here and realloc etc.	1673	// actual code will need to loop here and realloc etc.
1403	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));	1674	adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1404		1675
1405	/* the callback is illegal, but won't be called as we stop during check */	1676	/* the callback is illegal, but won't be called as we stop during check */
1406	ev_timer_init (&tw, 0, timeout * 1e-3);	1677	ev_timer_init (&tw, 0, timeout * 1e-3);
1407	ev_timer_start (loop, &tw);	1678	ev_timer_start (loop, &tw);
1408		1679
1409	// create on ev_io per pollfd	1680	// create one ev_io per pollfd
1410	for (int i = 0; i < nfd; ++i)	1681	for (int i = 0; i < nfd; ++i)
1411	{	1682	{
1412	ev_io_init (iow + i, io_cb, fds [i].fd,	1683	ev_io_init (iow + i, io_cb, fds [i].fd,
1413	((fds [i].events & POLLIN ? EV_READ : 0)	1684	((fds [i].events & POLLIN ? EV_READ : 0)
1414	\| (fds [i].events & POLLOUT ? EV_WRITE : 0)));	1685	\| (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1415		1686
1416	fds [i].revents = 0;	1687	fds [i].revents = 0;
1417	iow [i].data = fds + i;
1418	ev_io_start (loop, iow + i);	1688	ev_io_start (loop, iow + i);
1419	}	1689	}
1420	}	1690	}
1421		1691
1422	// stop all watchers after blocking	1692	// stop all watchers after blocking
…		…
1424	adns_check_cb (ev_loop loop, ev_check w, int revents)	1694	adns_check_cb (ev_loop loop, ev_check w, int revents)
1425	{	1695	{
1426	ev_timer_stop (loop, &tw);	1696	ev_timer_stop (loop, &tw);
1427		1697
1428	for (int i = 0; i < nfd; ++i)	1698	for (int i = 0; i < nfd; ++i)
		1699	{
		1700	// set the relevant poll flags
		1701	// could also call adns_processreadable etc. here
		1702	struct pollfd *fd = fds + i;
		1703	int revents = ev_clear_pending (iow + i);
		1704	if (revents & EV_READ ) fd->revents \|= fd->events & POLLIN;
		1705	if (revents & EV_WRITE) fd->revents \|= fd->events & POLLOUT;
		1706
		1707	// now stop the watcher
1429	ev_io_stop (loop, iow + i);	1708	ev_io_stop (loop, iow + i);
		1709	}
1430		1710
1431	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));	1711	adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
		1712	}
		1713
		1714	Method 2: This would be just like method 1, but you run C<adns_afterpoll>
		1715	in the prepare watcher and would dispose of the check watcher.
		1716
		1717	Method 3: If the module to be embedded supports explicit event
		1718	notification (adns does), you can also make use of the actual watcher
		1719	callbacks, and only destroy/create the watchers in the prepare watcher.
		1720
		1721	static void
		1722	timer_cb (EV_P_ ev_timer *w, int revents)
		1723	{
		1724	adns_state ads = (adns_state)w->data;
		1725	update_now (EV_A);
		1726
		1727	adns_processtimeouts (ads, &tv_now);
		1728	}
		1729
		1730	static void
		1731	io_cb (EV_P_ ev_io *w, int revents)
		1732	{
		1733	adns_state ads = (adns_state)w->data;
		1734	update_now (EV_A);
		1735
		1736	if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
		1737	if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
		1738	}
		1739
		1740	// do not ever call adns_afterpoll
		1741
		1742	Method 4: Do not use a prepare or check watcher because the module you
		1743	want to embed is too inflexible to support it. Instead, youc na override
		1744	their poll function. The drawback with this solution is that the main
		1745	loop is now no longer controllable by EV. The C<Glib::EV> module does
		1746	this.
		1747
		1748	static gint
		1749	event_poll_func (GPollFD *fds, guint nfds, gint timeout)
		1750	{
		1751	int got_events = 0;
		1752
		1753	for (n = 0; n < nfds; ++n)
		1754	// create/start io watcher that sets the relevant bits in fds[n] and increment got_events
		1755
		1756	if (timeout >= 0)
		1757	// create/start timer
		1758
		1759	// poll
		1760	ev_loop (EV_A_ 0);
		1761
		1762	// stop timer again
		1763	if (timeout >= 0)
		1764	ev_timer_stop (EV_A_ &to);
		1765
		1766	// stop io watchers again - their callbacks should have set
		1767	for (n = 0; n < nfds; ++n)
		1768	ev_io_stop (EV_A_ iow [n]);
		1769
		1770	return got_events;
1432	}	1771	}
1433		1772
1434		1773
1435	=head2 C<ev_embed> - when one backend isn't enough...	1774	=head2 C<ev_embed> - when one backend isn't enough...
1436		1775
1437	This is a rather advanced watcher type that lets you embed one event loop	1776	This is a rather advanced watcher type that lets you embed one event loop
1438	into another (currently only C<ev_io> events are supported in the embedded	1777	into another (currently only C<ev_io> events are supported in the embedded
1439	loop, other types of watchers might be handled in a delayed or incorrect	1778	loop, other types of watchers might be handled in a delayed or incorrect
1440	fashion and must not be used).	1779	fashion and must not be used). (See portability notes, below).
1441		1780
1442	There are primarily two reasons you would want that: work around bugs and	1781	There are primarily two reasons you would want that: work around bugs and
1443	prioritise I/O.	1782	prioritise I/O.
1444		1783
1445	As an example for a bug workaround, the kqueue backend might only support	1784	As an example for a bug workaround, the kqueue backend might only support
…		…
1500	ev_embed_start (loop_hi, &embed);	1839	ev_embed_start (loop_hi, &embed);
1501	}	1840	}
1502	else	1841	else
1503	loop_lo = loop_hi;	1842	loop_lo = loop_hi;
1504		1843
		1844	=head2 Portability notes
		1845
		1846	Kqueue is nominally embeddable, but this is broken on all BSDs that I
		1847	tried, in various ways. Usually the embedded event loop will simply never
		1848	receive events, sometimes it will only trigger a few times, sometimes in a
		1849	loop. Epoll is also nominally embeddable, but many Linux kernel versions
		1850	will always eport the epoll fd as ready, even when no events are pending.
		1851
		1852	While libev allows embedding these backends (they are contained in
		1853	C<ev_embeddable_backends ()>), take extreme care that it will actually
		1854	work.
		1855
		1856	When in doubt, create a dynamic event loop forced to use sockets (this
		1857	usually works) and possibly another thread and a pipe or so to report to
		1858	your main event loop.
		1859
		1860	=head3 Watcher-Specific Functions and Data Members
		1861
1505	=over 4	1862	=over 4
1506		1863
1507	=item ev_embed_init (ev_embed , callback, struct ev_loop embedded_loop)	1864	=item ev_embed_init (ev_embed , callback, struct ev_loop embedded_loop)
1508		1865
1509	=item ev_embed_set (ev_embed , callback, struct ev_loop embedded_loop)	1866	=item ev_embed_set (ev_embed , callback, struct ev_loop embedded_loop)
…		…
1518		1875
1519	Make a single, non-blocking sweep over the embedded loop. This works	1876	Make a single, non-blocking sweep over the embedded loop. This works
1520	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most	1877	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1521	apropriate way for embedded loops.	1878	apropriate way for embedded loops.
1522		1879
1523	=item struct ev_loop *loop [read-only]	1880	=item struct ev_loop *other [read-only]
1524		1881
1525	The embedded event loop.	1882	The embedded event loop.
1526		1883
1527	=back	1884	=back
1528		1885
…		…
1535	event loop blocks next and before C<ev_check> watchers are being called,	1892	event loop blocks next and before C<ev_check> watchers are being called,
1536	and only in the child after the fork. If whoever good citizen calling	1893	and only in the child after the fork. If whoever good citizen calling
1537	C<ev_default_fork> cheats and calls it in the wrong process, the fork	1894	C<ev_default_fork> cheats and calls it in the wrong process, the fork
1538	handlers will be invoked, too, of course.	1895	handlers will be invoked, too, of course.
1539		1896
		1897	=head3 Watcher-Specific Functions and Data Members
		1898
1540	=over 4	1899	=over 4
1541		1900
1542	=item ev_fork_init (ev_signal *, callback)	1901	=item ev_fork_init (ev_signal *, callback)
1543		1902
1544	Initialises and configures the fork watcher - it has no parameters of any	1903	Initialises and configures the fork watcher - it has no parameters of any
…		…
1640		1999
1641	To use it,	2000	To use it,
1642		2001
1643	#include <ev++.h>	2002	#include <ev++.h>
1644		2003
1645	(it is not installed by default). This automatically includes F<ev.h>	2004	This automatically includes F<ev.h> and puts all of its definitions (many
1646	and puts all of its definitions (many of them macros) into the global	2005	of them macros) into the global namespace. All C++ specific things are
1647	namespace. All C++ specific things are put into the C<ev> namespace.	2006	put into the C<ev> namespace. It should support all the same embedding
		2007	options as F<ev.h>, most notably C<EV_MULTIPLICITY>.
1648		2008
1649	It should support all the same embedding options as F<ev.h>, most notably	2009	Care has been taken to keep the overhead low. The only data member the C++
1650	C<EV_MULTIPLICITY>.	2010	classes add (compared to plain C-style watchers) is the event loop pointer
		2011	that the watcher is associated with (or no additional members at all if
		2012	you disable C<EV_MULTIPLICITY> when embedding libev).
		2013
		2014	Currently, functions, and static and non-static member functions can be
		2015	used as callbacks. Other types should be easy to add as long as they only
		2016	need one additional pointer for context. If you need support for other
		2017	types of functors please contact the author (preferably after implementing
		2018	it).
1651		2019
1652	Here is a list of things available in the C<ev> namespace:	2020	Here is a list of things available in the C<ev> namespace:
1653		2021
1654	=over 4	2022	=over 4
1655		2023
…		…
1671		2039
1672	All of those classes have these methods:	2040	All of those classes have these methods:
1673		2041
1674	=over 4	2042	=over 4
1675		2043
1676	=item ev::TYPE::TYPE (object , object::method )	2044	=item ev::TYPE::TYPE ()
1677		2045
1678	=item ev::TYPE::TYPE (object , object::method , struct ev_loop *)	2046	=item ev::TYPE::TYPE (struct ev_loop *)
1679		2047
1680	=item ev::TYPE::~TYPE	2048	=item ev::TYPE::~TYPE
1681		2049
1682	The constructor takes a pointer to an object and a method pointer to	2050	The constructor (optionally) takes an event loop to associate the watcher
1683	the event handler callback to call in this class. The constructor calls	2051	with. If it is omitted, it will use C<EV_DEFAULT>.
1684	C<ev_init> for you, which means you have to call the C<set> method	2052
1685	before starting it. If you do not specify a loop then the constructor	2053	The constructor calls C<ev_init> for you, which means you have to call the
1686	automatically associates the default loop with this watcher.	2054	C<set> method before starting it.
		2055
		2056	It will not set a callback, however: You have to call the templated C<set>
		2057	method to set a callback before you can start the watcher.
		2058
		2059	(The reason why you have to use a method is a limitation in C++ which does
		2060	not allow explicit template arguments for constructors).
1687		2061
1688	The destructor automatically stops the watcher if it is active.	2062	The destructor automatically stops the watcher if it is active.
		2063
		2064	=item w->set<class, &class::method> (object *)
		2065
		2066	This method sets the callback method to call. The method has to have a
		2067	signature of C<void (*)(ev_TYPE &, int)>, it receives the watcher as
		2068	first argument and the C<revents> as second. The object must be given as
		2069	parameter and is stored in the C<data> member of the watcher.
		2070
		2071	This method synthesizes efficient thunking code to call your method from
		2072	the C callback that libev requires. If your compiler can inline your
		2073	callback (i.e. it is visible to it at the place of the C<set> call and
		2074	your compiler is good :), then the method will be fully inlined into the
		2075	thunking function, making it as fast as a direct C callback.
		2076
		2077	Example: simple class declaration and watcher initialisation
		2078
		2079	struct myclass
		2080	{
		2081	void io_cb (ev::io &w, int revents) { }
		2082	}
		2083
		2084	myclass obj;
		2085	ev::io iow;
		2086	iow.set <myclass, &myclass::io_cb> (&obj);
		2087
		2088	=item w->set<function> (void *data = 0)
		2089
		2090	Also sets a callback, but uses a static method or plain function as
		2091	callback. The optional C<data> argument will be stored in the watcher's
		2092	C<data> member and is free for you to use.
		2093
		2094	The prototype of the C<function> must be C<void (*)(ev::TYPE &w, int)>.
		2095
		2096	See the method-C<set> above for more details.
		2097
		2098	Example:
		2099
		2100	static void io_cb (ev::io &w, int revents) { }
		2101	iow.set <io_cb> ();
1689		2102
1690	=item w->set (struct ev_loop *)	2103	=item w->set (struct ev_loop *)
1691		2104
1692	Associates a different C<struct ev_loop> with this watcher. You can only	2105	Associates a different C<struct ev_loop> with this watcher. You can only
1693	do this when the watcher is inactive (and not pending either).	2106	do this when the watcher is inactive (and not pending either).
1694		2107
1695	=item w->set ([args])	2108	=item w->set ([args])
1696		2109
1697	Basically the same as C<ev_TYPE_set>, with the same args. Must be	2110	Basically the same as C<ev_TYPE_set>, with the same args. Must be
1698	called at least once. Unlike the C counterpart, an active watcher gets	2111	called at least once. Unlike the C counterpart, an active watcher gets
1699	automatically stopped and restarted.	2112	automatically stopped and restarted when reconfiguring it with this
		2113	method.
1700		2114
1701	=item w->start ()	2115	=item w->start ()
1702		2116
1703	Starts the watcher. Note that there is no C<loop> argument as the	2117	Starts the watcher. Note that there is no C<loop> argument, as the
1704	constructor already takes the loop.	2118	constructor already stores the event loop.
1705		2119
1706	=item w->stop ()	2120	=item w->stop ()
1707		2121
1708	Stops the watcher if it is active. Again, no C<loop> argument.	2122	Stops the watcher if it is active. Again, no C<loop> argument.
1709		2123
1710	=item w->again () C<ev::timer>, C<ev::periodic> only	2124	=item w->again () (C<ev::timer>, C<ev::periodic> only)
1711		2125
1712	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding	2126	For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
1713	C<ev_TYPE_again> function.	2127	C<ev_TYPE_again> function.
1714		2128
1715	=item w->sweep () C<ev::embed> only	2129	=item w->sweep () (C<ev::embed> only)
1716		2130
1717	Invokes C<ev_embed_sweep>.	2131	Invokes C<ev_embed_sweep>.
1718		2132
1719	=item w->update () C<ev::stat> only	2133	=item w->update () (C<ev::stat> only)
1720		2134
1721	Invokes C<ev_stat_stat>.	2135	Invokes C<ev_stat_stat>.
1722		2136
1723	=back	2137	=back
1724		2138
…		…
1734		2148
1735	myclass ();	2149	myclass ();
1736	}	2150	}
1737		2151
1738	myclass::myclass (int fd)	2152	myclass::myclass (int fd)
1739	: io (this, &myclass::io_cb),
1740	idle (this, &myclass::idle_cb)
1741	{	2153	{
		2154	io .set <myclass, &myclass::io_cb > (this);
		2155	idle.set <myclass, &myclass::idle_cb> (this);
		2156
1742	io.start (fd, ev::READ);	2157	io.start (fd, ev::READ);
1743	}	2158	}
1744		2159
1745		2160
1746	=head1 MACRO MAGIC	2161	=head1 MACRO MAGIC
1747		2162
1748	Libev can be compiled with a variety of options, the most fundemantal is	2163	Libev can be compiled with a variety of options, the most fundamantal
1749	C<EV_MULTIPLICITY>. This option determines wether (most) functions and	2164	of which is C<EV_MULTIPLICITY>. This option determines whether (most)
1750	callbacks have an initial C<struct ev_loop *> argument.	2165	functions and callbacks have an initial C<struct ev_loop *> argument.
1751		2166
1752	To make it easier to write programs that cope with either variant, the	2167	To make it easier to write programs that cope with either variant, the
1753	following macros are defined:	2168	following macros are defined:
1754		2169
1755	=over 4	2170	=over 4
…		…
1787	Similar to the other two macros, this gives you the value of the default	2202	Similar to the other two macros, this gives you the value of the default
1788	loop, if multiple loops are supported ("ev loop default").	2203	loop, if multiple loops are supported ("ev loop default").
1789		2204
1790	=back	2205	=back
1791		2206
1792	Example: Declare and initialise a check watcher, working regardless of	2207	Example: Declare and initialise a check watcher, utilising the above
1793	wether multiple loops are supported or not.	2208	macros so it will work regardless of whether multiple loops are supported
		2209	or not.
1794		2210
1795	static void	2211	static void
1796	check_cb (EV_P_ ev_timer *w, int revents)	2212	check_cb (EV_P_ ev_timer *w, int revents)
1797	{	2213	{
1798	ev_check_stop (EV_A_ w);	2214	ev_check_stop (EV_A_ w);
…		…
1801	ev_check check;	2217	ev_check check;
1802	ev_check_init (&check, check_cb);	2218	ev_check_init (&check, check_cb);
1803	ev_check_start (EV_DEFAULT_ &check);	2219	ev_check_start (EV_DEFAULT_ &check);
1804	ev_loop (EV_DEFAULT_ 0);	2220	ev_loop (EV_DEFAULT_ 0);
1805		2221
1806
1807	=head1 EMBEDDING	2222	=head1 EMBEDDING
1808		2223
1809	Libev can (and often is) directly embedded into host	2224	Libev can (and often is) directly embedded into host
1810	applications. Examples of applications that embed it include the Deliantra	2225	applications. Examples of applications that embed it include the Deliantra
1811	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)	2226	Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
1812	and rxvt-unicode.	2227	and rxvt-unicode.
1813		2228
1814	The goal is to enable you to just copy the neecssary files into your	2229	The goal is to enable you to just copy the necessary files into your
1815	source directory without having to change even a single line in them, so	2230	source directory without having to change even a single line in them, so
1816	you can easily upgrade by simply copying (or having a checked-out copy of	2231	you can easily upgrade by simply copying (or having a checked-out copy of
1817	libev somewhere in your source tree).	2232	libev somewhere in your source tree).
1818		2233
1819	=head2 FILESETS	2234	=head2 FILESETS
…		…
1850	ev_vars.h	2265	ev_vars.h
1851	ev_wrap.h	2266	ev_wrap.h
1852		2267
1853	ev_win32.c required on win32 platforms only	2268	ev_win32.c required on win32 platforms only
1854		2269
1855	ev_select.c only when select backend is enabled (which is by default)	2270	ev_select.c only when select backend is enabled (which is enabled by default)
1856	ev_poll.c only when poll backend is enabled (disabled by default)	2271	ev_poll.c only when poll backend is enabled (disabled by default)
1857	ev_epoll.c only when the epoll backend is enabled (disabled by default)	2272	ev_epoll.c only when the epoll backend is enabled (disabled by default)
1858	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)	2273	ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
1859	ev_port.c only when the solaris port backend is enabled (disabled by default)	2274	ev_port.c only when the solaris port backend is enabled (disabled by default)
1860		2275
…		…
1909		2324
1910	If defined to be C<1>, libev will try to detect the availability of the	2325	If defined to be C<1>, libev will try to detect the availability of the
1911	monotonic clock option at both compiletime and runtime. Otherwise no use	2326	monotonic clock option at both compiletime and runtime. Otherwise no use
1912	of the monotonic clock option will be attempted. If you enable this, you	2327	of the monotonic clock option will be attempted. If you enable this, you
1913	usually have to link against librt or something similar. Enabling it when	2328	usually have to link against librt or something similar. Enabling it when
1914	the functionality isn't available is safe, though, althoguh you have	2329	the functionality isn't available is safe, though, although you have
1915	to make sure you link against any libraries where the C<clock_gettime>	2330	to make sure you link against any libraries where the C<clock_gettime>
1916	function is hiding in (often F<-lrt>).	2331	function is hiding in (often F<-lrt>).
1917		2332
1918	=item EV_USE_REALTIME	2333	=item EV_USE_REALTIME
1919		2334
1920	If defined to be C<1>, libev will try to detect the availability of the	2335	If defined to be C<1>, libev will try to detect the availability of the
1921	realtime clock option at compiletime (and assume its availability at	2336	realtime clock option at compiletime (and assume its availability at
1922	runtime if successful). Otherwise no use of the realtime clock option will	2337	runtime if successful). Otherwise no use of the realtime clock option will
1923	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	2338	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
1924	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries	2339	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
1925	in the description of C<EV_USE_MONOTONIC>, though.	2340	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
		2341
		2342	=item EV_USE_NANOSLEEP
		2343
		2344	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		2345	and will use it for delays. Otherwise it will use C<select ()>.
1926		2346
1927	=item EV_USE_SELECT	2347	=item EV_USE_SELECT
1928		2348
1929	If undefined or defined to be C<1>, libev will compile in support for the	2349	If undefined or defined to be C<1>, libev will compile in support for the
1930	C<select>(2) backend. No attempt at autodetection will be done: if no	2350	C<select>(2) backend. No attempt at autodetection will be done: if no
…		…
1985		2405
1986	=item EV_USE_DEVPOLL	2406	=item EV_USE_DEVPOLL
1987		2407
1988	reserved for future expansion, works like the USE symbols above.	2408	reserved for future expansion, works like the USE symbols above.
1989		2409
		2410	=item EV_USE_INOTIFY
		2411
		2412	If defined to be C<1>, libev will compile in support for the Linux inotify
		2413	interface to speed up C<ev_stat> watchers. Its actual availability will
		2414	be detected at runtime.
		2415
1990	=item EV_H	2416	=item EV_H
1991		2417
1992	The name of the F<ev.h> header file used to include it. The default if	2418	The name of the F<ev.h> header file used to include it. The default if
1993	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	2419	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This
1994	can be used to virtually rename the F<ev.h> header file in case of conflicts.	2420	can be used to virtually rename the F<ev.h> header file in case of conflicts.
…		…
2017	will have the C<struct ev_loop *> as first argument, and you can create	2443	will have the C<struct ev_loop *> as first argument, and you can create
2018	additional independent event loops. Otherwise there will be no support	2444	additional independent event loops. Otherwise there will be no support
2019	for multiple event loops and there is no first event loop pointer	2445	for multiple event loops and there is no first event loop pointer
2020	argument. Instead, all functions act on the single default loop.	2446	argument. Instead, all functions act on the single default loop.
2021		2447
		2448	=item EV_MINPRI
		2449
		2450	=item EV_MAXPRI
		2451
		2452	The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
		2453	C<EV_MAXPRI>, but otherwise there are no non-obvious limitations. You can
		2454	provide for more priorities by overriding those symbols (usually defined
		2455	to be C<-2> and C<2>, respectively).
		2456
		2457	When doing priority-based operations, libev usually has to linearly search
		2458	all the priorities, so having many of them (hundreds) uses a lot of space
		2459	and time, so using the defaults of five priorities (-2 .. +2) is usually
		2460	fine.
		2461
		2462	If your embedding app does not need any priorities, defining these both to
		2463	C<0> will save some memory and cpu.
		2464
2022	=item EV_PERIODIC_ENABLE	2465	=item EV_PERIODIC_ENABLE
2023		2466
2024	If undefined or defined to be C<1>, then periodic timers are supported. If	2467	If undefined or defined to be C<1>, then periodic timers are supported. If
2025	defined to be C<0>, then they are not. Disabling them saves a few kB of	2468	defined to be C<0>, then they are not. Disabling them saves a few kB of
2026	code.	2469	code.
2027		2470
		2471	=item EV_IDLE_ENABLE
		2472
		2473	If undefined or defined to be C<1>, then idle watchers are supported. If
		2474	defined to be C<0>, then they are not. Disabling them saves a few kB of
		2475	code.
		2476
2028	=item EV_EMBED_ENABLE	2477	=item EV_EMBED_ENABLE
2029		2478
2030	If undefined or defined to be C<1>, then embed watchers are supported. If	2479	If undefined or defined to be C<1>, then embed watchers are supported. If
2031	defined to be C<0>, then they are not.	2480	defined to be C<0>, then they are not.
2032		2481
…		…
2049	=item EV_PID_HASHSIZE	2498	=item EV_PID_HASHSIZE
2050		2499
2051	C<ev_child> watchers use a small hash table to distribute workload by	2500	C<ev_child> watchers use a small hash table to distribute workload by
2052	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	2501	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
2053	than enough. If you need to manage thousands of children you might want to	2502	than enough. If you need to manage thousands of children you might want to
2054	increase this value.	2503	increase this value (I<must> be a power of two).
		2504
		2505	=item EV_INOTIFY_HASHSIZE
		2506
		2507	C<ev_staz> watchers use a small hash table to distribute workload by
		2508	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
		2509	usually more than enough. If you need to manage thousands of C<ev_stat>
		2510	watchers you might want to increase this value (I<must> be a power of
		2511	two).
2055		2512
2056	=item EV_COMMON	2513	=item EV_COMMON
2057		2514
2058	By default, all watchers have a C<void *data> member. By redefining	2515	By default, all watchers have a C<void *data> member. By redefining
2059	this macro to a something else you can include more and other types of	2516	this macro to a something else you can include more and other types of
…		…
2072		2529
2073	=item ev_set_cb (ev, cb)	2530	=item ev_set_cb (ev, cb)
2074		2531
2075	Can be used to change the callback member declaration in each watcher,	2532	Can be used to change the callback member declaration in each watcher,
2076	and the way callbacks are invoked and set. Must expand to a struct member	2533	and the way callbacks are invoked and set. Must expand to a struct member
2077	definition and a statement, respectively. See the F<ev.v> header file for	2534	definition and a statement, respectively. See the F<ev.h> header file for
2078	their default definitions. One possible use for overriding these is to	2535	their default definitions. One possible use for overriding these is to
2079	avoid the C<struct ev_loop *> as first argument in all cases, or to use	2536	avoid the C<struct ev_loop *> as first argument in all cases, or to use
2080	method calls instead of plain function calls in C++.	2537	method calls instead of plain function calls in C++.
		2538
		2539	=head2 EXPORTED API SYMBOLS
		2540
		2541	If you need to re-export the API (e.g. via a dll) and you need a list of
		2542	exported symbols, you can use the provided F<Symbol.*> files which list
		2543	all public symbols, one per line:
		2544
		2545	Symbols.ev for libev proper
		2546	Symbols.event for the libevent emulation
		2547
		2548	This can also be used to rename all public symbols to avoid clashes with
		2549	multiple versions of libev linked together (which is obviously bad in
		2550	itself, but sometimes it is inconvinient to avoid this).
		2551
		2552	A sed command like this will create wrapper C<#define>'s that you need to
		2553	include before including F<ev.h>:
		2554
		2555	<Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
		2556
		2557	This would create a file F<wrap.h> which essentially looks like this:
		2558
		2559	#define ev_backend myprefix_ev_backend
		2560	#define ev_check_start myprefix_ev_check_start
		2561	#define ev_check_stop myprefix_ev_check_stop
		2562	...
2081		2563
2082	=head2 EXAMPLES	2564	=head2 EXAMPLES
2083		2565
2084	For a real-world example of a program the includes libev	2566	For a real-world example of a program the includes libev
2085	verbatim, you can have a look at the EV perl module	2567	verbatim, you can have a look at the EV perl module
…		…
2088	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file	2570	interface) and F<EV.xs> (implementation) files. Only the F<EV.xs> file
2089	will be compiled. It is pretty complex because it provides its own header	2571	will be compiled. It is pretty complex because it provides its own header
2090	file.	2572	file.
2091		2573
2092	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file	2574	The usage in rxvt-unicode is simpler. It has a F<ev_cpp.h> header file
2093	that everybody includes and which overrides some autoconf choices:	2575	that everybody includes and which overrides some configure choices:
2094		2576
		2577	#define EV_MINIMAL 1
2095	#define EV_USE_POLL 0	2578	#define EV_USE_POLL 0
2096	#define EV_MULTIPLICITY 0	2579	#define EV_MULTIPLICITY 0
2097	#define EV_PERIODICS 0	2580	#define EV_PERIODIC_ENABLE 0
		2581	#define EV_STAT_ENABLE 0
		2582	#define EV_FORK_ENABLE 0
2098	#define EV_CONFIG_H <config.h>	2583	#define EV_CONFIG_H <config.h>
		2584	#define EV_MINPRI 0
		2585	#define EV_MAXPRI 0
2099		2586
2100	#include "ev++.h"	2587	#include "ev++.h"
2101		2588
2102	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	2589	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
2103		2590
…		…
2109		2596
2110	In this section the complexities of (many of) the algorithms used inside	2597	In this section the complexities of (many of) the algorithms used inside
2111	libev will be explained. For complexity discussions about backends see the	2598	libev will be explained. For complexity discussions about backends see the
2112	documentation for C<ev_default_init>.	2599	documentation for C<ev_default_init>.
2113		2600
		2601	All of the following are about amortised time: If an array needs to be
		2602	extended, libev needs to realloc and move the whole array, but this
		2603	happens asymptotically never with higher number of elements, so O(1) might
		2604	mean it might do a lengthy realloc operation in rare cases, but on average
		2605	it is much faster and asymptotically approaches constant time.
		2606
2114	=over 4	2607	=over 4
2115		2608
2116	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	2609	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2117		2610
		2611	This means that, when you have a watcher that triggers in one hour and
		2612	there are 100 watchers that would trigger before that then inserting will
		2613	have to skip those 100 watchers.
		2614
2118	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	2615	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)
2119		2616
		2617	That means that for changing a timer costs less than removing/adding them
		2618	as only the relative motion in the event queue has to be paid for.
		2619
2120	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	2620	=item Starting io/check/prepare/idle/signal/child watchers: O(1)
2121		2621
		2622	These just add the watcher into an array or at the head of a list.
2122	=item Stopping check/prepare/idle watchers: O(1)	2623	=item Stopping check/prepare/idle watchers: O(1)
2123		2624
2124	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % 16))	2625	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
		2626
		2627	These watchers are stored in lists then need to be walked to find the
		2628	correct watcher to remove. The lists are usually short (you don't usually
		2629	have many watchers waiting for the same fd or signal).
2125		2630
2126	=item Finding the next timer per loop iteration: O(1)	2631	=item Finding the next timer per loop iteration: O(1)
2127		2632
2128	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	2633	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2129		2634
		2635	A change means an I/O watcher gets started or stopped, which requires
		2636	libev to recalculate its status (and possibly tell the kernel).
		2637
2130	=item Activating one watcher: O(1)	2638	=item Activating one watcher: O(1)
2131		2639
		2640	=item Priority handling: O(number_of_priorities)
		2641
		2642	Priorities are implemented by allocating some space for each
		2643	priority. When doing priority-based operations, libev usually has to
		2644	linearly search all the priorities.
		2645
2132	=back	2646	=back
2133		2647
2134		2648
2135	=head1 AUTHOR	2649	=head1 AUTHOR
2136		2650

Diff Legend

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

Comparing libev/ev.pod (file contents): Revision 1.54 by root, Tue Nov 27 20:26:51 2007 UTC vs. Revision 1.97 by root, Sat Dec 22 05:48:02 2007 UTC

Diff Legend

Comparing libev/ev.pod (file contents):
Revision 1.54 by root, Tue Nov 27 20:26:51 2007 UTC vs.
Revision 1.97 by root, Sat Dec 22 05:48:02 2007 UTC