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

Comparing libev/ev.pod (file contents):
Revision 1.97 by root, Sat Dec 22 05:48:02 2007 UTC vs.
Revision 1.134 by root, Sat Mar 8 07:04:56 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;
…		…
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
…		…
260	flags. If that is troubling you, check C<ev_backend ()> afterwards).	260	flags. If that is troubling you, check C<ev_backend ()> afterwards).
261		261
262	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
263	function.	263	function.
264		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
265	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
266	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>).
267		274
268	The following flags are supported:	275	The following flags are supported:
269		276
…		…
306	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	313	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
307		314
308	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
309	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,
310	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
311	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
312	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.
313		327
314	=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)
315		329
316	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
317	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
318	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
319	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.
320		336
321	=item C<EVBACKEND_EPOLL> (value 4, Linux)	337	=item C<EVBACKEND_EPOLL> (value 4, Linux)
322		338
323	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,
324	but it scales phenomenally better. While poll and select usually scale	340	but it scales phenomenally better. While poll and select usually scale
325	like O(total_fds) where n is the total number of fds (or the highest fd),	341	like O(total_fds) where n is the total number of fds (or the highest fd),
326	epoll scales either O(1) or O(active_fds). The epoll design has a number	342	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	343	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	344	cases and rewiring a syscall per fd change, no fork support and bad
329	support for dup:	345	support for dup.
330		346
331	While stopping, setting and starting an I/O watcher in the same iteration	347	While stopping, setting and starting an I/O watcher in the same iteration
332	will 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
333	(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
334	best to avoid that. Also, C<dup ()>'ed file descriptors might not work	350	best to avoid that. Also, C<dup ()>'ed file descriptors might not work
…		…
336		352
337	Please note that epoll sometimes generates spurious notifications, so you	353	Please note that epoll sometimes generates spurious notifications, so you
338	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
339	(or space) is available.	355	(or space) is available.
340		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
341	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	364	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
342		365
343	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
344	was broken on I<all> BSDs (usually it doesn't work with anything but	367	was broken on all BSDs except NetBSD (usually it doesn't work reliably
345	sockets and pipes, except on Darwin, where of course it's completely	368	with anything but sockets and pipes, except on Darwin, where of course
346	useless. On NetBSD, it seems to work for all the FD types I tested, so it
347	is used by default there). For this reason it's not being "autodetected"	369	it's completely useless). For this reason it's not being "autodetected"
348	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
349	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	371	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
350	system like NetBSD.	372	system like NetBSD.
351		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.
		377
352	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
353	kernel is more efficient (which says nothing about its actual speed,	379	kernel is more efficient (which says nothing about its actual speed, of
354	of course). While stopping, setting and starting an I/O watcher does	380	course). While stopping, setting and starting an I/O watcher does never
355	never cause an extra syscall as with epoll, it still adds up to two event	381	cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
356	changes per incident, support for C<fork ()> is very bad and it drops fds	382	two event changes per incident, support for C<fork ()> is very bad and it
357	silently in similarly hard-to-detetc cases.	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.
358		393
359	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	394	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
360		395
361	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.
362		400
363	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	401	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
364		402
365	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	403	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
366	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)).
367		405
368	Please note that solaris event ports can deliver a lot of spurious	406	Please note that solaris event ports can deliver a lot of spurious
369	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
370	blocking when no data (or space) is available.	408	blocking when no data (or space) is available.
371		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.
		418
372	=item C<EVBACKEND_ALL>	419	=item C<EVBACKEND_ALL>
373		420
374	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
375	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
376	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	423	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
377		424
		425	It is definitely not recommended to use this flag.
		426
378	=back	427	=back
379		428
380	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
381	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
382	specified, most compiled-in backend will be tried, usually in reverse	431	specified, all backends in C<ev_recommended_backends ()> will be tried.
383	order of their flag values :)
384		432
385	The most typical usage is like this:	433	The most typical usage is like this:
386		434
387	if (!ev_default_loop (0))	435	if (!ev_default_loop (0))
388	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");	436	fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
…		…
435	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
436	earlier call to C<ev_loop_new>.	484	earlier call to C<ev_loop_new>.
437		485
438	=item ev_default_fork ()	486	=item ev_default_fork ()
439		487
		488	This function sets a flag that causes subsequent C<ev_loop> iterations
440	This function reinitialises the kernel state for backends that have	489	to reinitialise the kernel state for backends that have one. Despite the
441	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
442	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
443	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.
444		494
445	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
446	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
447	fork+exec, you don't have to call it.	497	you just fork+exec, you don't have to call it at all.
448		498
449	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
450	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
451	quite nicely into a call to C<pthread_atfork>:	501	quite nicely into a call to C<pthread_atfork>:
452		502
453	pthread_atfork (0, 0, ev_default_fork);	503	pthread_atfork (0, 0, ev_default_fork);
454		504
455	At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use
456	without calling this function, so if you force one of those backends you
457	do not need to care.
458
459	=item ev_loop_fork (loop)	505	=item ev_loop_fork (loop)
460		506
461	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
462	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
463	after fork, and how you do this is entirely your own problem.	509	after fork, and how you do this is entirely your own problem.
		510
		511	=item int ev_is_default_loop (loop)
		512
		513	Returns true when the given loop actually is the default loop, false otherwise.
464		514
465	=item unsigned int ev_loop_count (loop)	515	=item unsigned int ev_loop_count (loop)
466		516
467	Returns the count of loop iterations for the loop, which is identical to	517	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	518	the number of times libev did poll for new events. It starts at C<0> and
…		…
513	usually a better approach for this kind of thing.	563	usually a better approach for this kind of thing.
514		564
515	Here are the gory details of what C<ev_loop> does:	565	Here are the gory details of what C<ev_loop> does:
516		566
517	- Before the first iteration, call any pending watchers.	567	- Before the first iteration, call any pending watchers.
518	* If there are no active watchers (reference count is zero), return.	568	* If EVFLAG_FORKCHECK was used, check for a fork.
519	- Queue all prepare watchers and then call all outstanding watchers.	569	- If a fork was detected, queue and call all fork watchers.
		570	- Queue and call all prepare watchers.
520	- If we have been forked, recreate the kernel state.	571	- If we have been forked, recreate the kernel state.
521	- Update the kernel state with all outstanding changes.	572	- Update the kernel state with all outstanding changes.
522	- Update the "event loop time".	573	- Update the "event loop time".
523	- Calculate for how long to block.	574	- Calculate for how long to sleep or block, if at all
		575	(active idle watchers, EVLOOP_NONBLOCK or not having
		576	any active watchers at all will result in not sleeping).
		577	- Sleep if the I/O and timer collect interval say so.
524	- Block the process, waiting for any events.	578	- Block the process, waiting for any events.
525	- Queue all outstanding I/O (fd) events.	579	- Queue all outstanding I/O (fd) events.
526	- Update the "event loop time" and do time jump handling.	580	- Update the "event loop time" and do time jump handling.
527	- Queue all outstanding timers.	581	- Queue all outstanding timers.
528	- Queue all outstanding periodics.	582	- Queue all outstanding periodics.
529	- If no events are pending now, queue all idle watchers.	583	- If no events are pending now, queue all idle watchers.
530	- Queue all check watchers.	584	- Queue all check watchers.
531	- Call all queued watchers in reverse order (i.e. check watchers first).	585	- Call all queued watchers in reverse order (i.e. check watchers first).
532	Signals and child watchers are implemented as I/O watchers, and will	586	Signals and child watchers are implemented as I/O watchers, and will
533	be handled here by queueing them when their watcher gets executed.	587	be handled here by queueing them when their watcher gets executed.
534	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	588	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
535	were used, return, otherwise continue with step *.	589	were used, or there are no active watchers, return, otherwise
		590	continue with step *.
536		591
537	Example: Queue some jobs and then loop until no events are outsanding	592	Example: Queue some jobs and then loop until no events are outstanding
538	anymore.	593	anymore.
539		594
540	... queue jobs here, make sure they register event watchers as long	595	... queue jobs here, make sure they register event watchers as long
541	... as they still have work to do (even an idle watcher will do..)	596	... as they still have work to do (even an idle watcher will do..)
542	ev_loop (my_loop, 0);	597	ev_loop (my_loop, 0);
…		…
546		601
547	Can be used to make a call to C<ev_loop> return early (but only after it	602	Can be used to make a call to C<ev_loop> return early (but only after it
548	has processed all outstanding events). The C<how> argument must be either	603	has processed all outstanding events). The C<how> argument must be either
549	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	604	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
550	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	605	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
		606
		607	This "unloop state" will be cleared when entering C<ev_loop> again.
551		608
552	=item ev_ref (loop)	609	=item ev_ref (loop)
553		610
554	=item ev_unref (loop)	611	=item ev_unref (loop)
555		612
…		…
560	returning, ev_unref() after starting, and ev_ref() before stopping it. For	617	returning, ev_unref() after starting, and ev_ref() before stopping it. For
561	example, libev itself uses this for its internal signal pipe: It is not	618	example, libev itself uses this for its internal signal pipe: It is not
562	visible to the libev user and should not keep C<ev_loop> from exiting if	619	visible to the libev user and should not keep C<ev_loop> from exiting if
563	no event watchers registered by it are active. It is also an excellent	620	no event watchers registered by it are active. It is also an excellent
564	way to do this for generic recurring timers or from within third-party	621	way to do this for generic recurring timers or from within third-party
565	libraries. Just remember to I<unref after start> and I<ref before stop>.	622	libraries. Just remember to I<unref after start> and I<ref before stop>
		623	(but only if the watcher wasn't active before, or was active before,
		624	respectively).
566		625
567	Example: Create a signal watcher, but keep it from keeping C<ev_loop>	626	Example: Create a signal watcher, but keep it from keeping C<ev_loop>
568	running when nothing else is active.	627	running when nothing else is active.
569		628
570	struct ev_signal exitsig;	629	struct ev_signal exitsig;
…		…
596	overhead for the actual polling but can deliver many events at once.	655	overhead for the actual polling but can deliver many events at once.
597		656
598	By setting a higher I<io collect interval> you allow libev to spend more	657	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,	658	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	659	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
601	C<ev_timer>) will be not affected.	660	C<ev_timer>) will be not affected. Setting this to a non-null value will
		661	introduce an additional C<ev_sleep ()> call into most loop iterations.
602		662
603	Likewise, by setting a higher I<timeout collect interval> you allow libev	663	Likewise, by setting a higher I<timeout collect interval> you allow libev
604	to spend more time collecting timeouts, at the expense of increased	664	to spend more time collecting timeouts, at the expense of increased
605	latency (the watcher callback will be called later). C<ev_io> watchers	665	latency (the watcher callback will be called later). C<ev_io> watchers
606	will not be affected.	666	will not be affected. Setting this to a non-null value will not introduce
		667	any overhead in libev.
607		668
608	Many programs can usually benefit by setting the io collect interval to	669	Many (busy) programs can usually benefit by setting the io collect
609	a value near C<0.1> or so, which is often enough for interactive servers	670	interval to a value near C<0.1> or so, which is often enough for
610	(of course not for games), likewise for timeouts. It usually doesn't make	671	interactive servers (of course not for games), likewise for timeouts. It
611	much sense to set it to a lower value than C<0.01>, as this approsaches	672	usually doesn't make much sense to set it to a lower value than C<0.01>,
612	the timing granularity of most systems.	673	as this approsaches the timing granularity of most systems.
613		674
614	=back	675	=back
615		676
616		677
617	=head1 ANATOMY OF A WATCHER	678	=head1 ANATOMY OF A WATCHER
…		…
716		777
717	=item C<EV_FORK>	778	=item C<EV_FORK>
718		779
719	The event loop has been resumed in the child process after fork (see	780	The event loop has been resumed in the child process after fork (see
720	C<ev_fork>).	781	C<ev_fork>).
		782
		783	=item C<EV_ASYNC>
		784
		785	The given async watcher has been asynchronously notified (see C<ev_async>).
721		786
722	=item C<EV_ERROR>	787	=item C<EV_ERROR>
723		788
724	An unspecified error has occured, the watcher has been stopped. This might	789	An unspecified error has occured, the watcher has been stopped. This might
725	happen because the watcher could not be properly started because libev	790	happen because the watcher could not be properly started because libev
…		…
943	In general you can register as many read and/or write event watchers per	1008	In general you can register as many read and/or write event watchers per
944	fd as you want (as long as you don't confuse yourself). Setting all file	1009	fd as you want (as long as you don't confuse yourself). Setting all file
945	descriptors to non-blocking mode is also usually a good idea (but not	1010	descriptors to non-blocking mode is also usually a good idea (but not
946	required if you know what you are doing).	1011	required if you know what you are doing).
947		1012
948	You have to be careful with dup'ed file descriptors, though. Some backends
949	(the linux epoll backend is a notable example) cannot handle dup'ed file
950	descriptors correctly if you register interest in two or more fds pointing
951	to the same underlying file/socket/etc. description (that is, they share
952	the same underlying "file open").
953
954	If you must do this, then force the use of a known-to-be-good backend	1013	If you must do this, then force the use of a known-to-be-good backend
955	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	1014	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
956	C<EVBACKEND_POLL>).	1015	C<EVBACKEND_POLL>).
957		1016
958	Another thing you have to watch out for is that it is quite easy to	1017	Another thing you have to watch out for is that it is quite easy to
…		…
992	optimisations to libev.	1051	optimisations to libev.
993		1052
994	=head3 The special problem of dup'ed file descriptors	1053	=head3 The special problem of dup'ed file descriptors
995		1054
996	Some backends (e.g. epoll), cannot register events for file descriptors,	1055	Some backends (e.g. epoll), cannot register events for file descriptors,
997	but only events for the underlying file descriptions. That menas when you	1056	but only events for the underlying file descriptions. That means when you
998	have C<dup ()>'ed file descriptors and register events for them, only one	1057	have C<dup ()>'ed file descriptors or weirder constellations, and register
999	file descriptor might actually receive events.	1058	events for them, only one file descriptor might actually receive events.
1000		1059
1001	There is no workaorund possible except not registering events	1060	There is no workaround possible except not registering events
1002	for potentially C<dup ()>'ed file descriptors or to resort to	1061	for potentially C<dup ()>'ed file descriptors, or to resort to
1003	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.	1062	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1004		1063
1005	=head3 The special problem of fork	1064	=head3 The special problem of fork
1006		1065
1007	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit	1066	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
…		…
1033	=item int events [read-only]	1092	=item int events [read-only]
1034		1093
1035	The events being watched.	1094	The events being watched.
1036		1095
1037	=back	1096	=back
		1097
		1098	=head3 Examples
1038		1099
1039	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1100	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1040	readable, but only once. Since it is likely line-buffered, you could	1101	readable, but only once. Since it is likely line-buffered, you could
1041	attempt to read a whole line in the callback.	1102	attempt to read a whole line in the callback.
1042		1103
…		…
1095	configure a timer to trigger every 10 seconds, then it will trigger at	1156	configure a timer to trigger every 10 seconds, then it will trigger at
1096	exactly 10 second intervals. If, however, your program cannot keep up with	1157	exactly 10 second intervals. If, however, your program cannot keep up with
1097	the timer (because it takes longer than those 10 seconds to do stuff) the	1158	the timer (because it takes longer than those 10 seconds to do stuff) the
1098	timer will not fire more than once per event loop iteration.	1159	timer will not fire more than once per event loop iteration.
1099		1160
1100	=item ev_timer_again (loop)	1161	=item ev_timer_again (loop, ev_timer *)
1101		1162
1102	This will act as if the timer timed out and restart it again if it is	1163	This will act as if the timer timed out and restart it again if it is
1103	repeating. The exact semantics are:	1164	repeating. The exact semantics are:
1104		1165
1105	If the timer is pending, its pending status is cleared.	1166	If the timer is pending, its pending status is cleared.
…		…
1140	or C<ev_timer_again> is called and determines the next timeout (if any),	1201	or C<ev_timer_again> is called and determines the next timeout (if any),
1141	which is also when any modifications are taken into account.	1202	which is also when any modifications are taken into account.
1142		1203
1143	=back	1204	=back
1144		1205
		1206	=head3 Examples
		1207
1145	Example: Create a timer that fires after 60 seconds.	1208	Example: Create a timer that fires after 60 seconds.
1146		1209
1147	static void	1210	static void
1148	one_minute_cb (struct ev_loop loop, struct ev_timer w, int revents)	1211	one_minute_cb (struct ev_loop loop, struct ev_timer w, int revents)
1149	{	1212	{
…		…
1212	In this configuration the watcher triggers an event at the wallclock time	1275	In this configuration the watcher triggers an event at the wallclock time
1213	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1276	C<at> and doesn't repeat. It will not adjust when a time jump occurs,
1214	that is, if it is to be run at January 1st 2011 then it will run when the	1277	that is, if it is to be run at January 1st 2011 then it will run when the
1215	system time reaches or surpasses this time.	1278	system time reaches or surpasses this time.
1216		1279
1217	=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	1280	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1218		1281
1219	In this mode the watcher will always be scheduled to time out at the next	1282	In this mode the watcher will always be scheduled to time out at the next
1220	C<at + N * interval> time (for some integer N, which can also be negative)	1283	C<at + N * interval> time (for some integer N, which can also be negative)
1221	and then repeat, regardless of any time jumps.	1284	and then repeat, regardless of any time jumps.
1222		1285
…		…
1305		1368
1306	When active, contains the absolute time that the watcher is supposed to	1369	When active, contains the absolute time that the watcher is supposed to
1307	trigger next.	1370	trigger next.
1308		1371
1309	=back	1372	=back
		1373
		1374	=head3 Examples
1310		1375
1311	Example: Call a callback every hour, or, more precisely, whenever the	1376	Example: Call a callback every hour, or, more precisely, whenever the
1312	system clock is divisible by 3600. The callback invocation times have	1377	system clock is divisible by 3600. The callback invocation times have
1313	potentially a lot of jittering, but good long-term stability.	1378	potentially a lot of jittering, but good long-term stability.
1314		1379
…		…
1371		1436
1372	The signal the watcher watches out for.	1437	The signal the watcher watches out for.
1373		1438
1374	=back	1439	=back
1375		1440
		1441	=head3 Examples
		1442
		1443	Example: Try to exit cleanly on SIGINT and SIGTERM.
		1444
		1445	static void
		1446	sigint_cb (struct ev_loop loop, struct ev_signal w, int revents)
		1447	{
		1448	ev_unloop (loop, EVUNLOOP_ALL);
		1449	}
		1450
		1451	struct ev_signal signal_watcher;
		1452	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
		1453	ev_signal_start (loop, &sigint_cb);
		1454
1376		1455
1377	=head2 C<ev_child> - watch out for process status changes	1456	=head2 C<ev_child> - watch out for process status changes
1378		1457
1379	Child watchers trigger when your process receives a SIGCHLD in response to	1458	Child watchers trigger when your process receives a SIGCHLD in response to
1380	some child status changes (most typically when a child of yours dies).	1459	some child status changes (most typically when a child of yours dies). It
		1460	is permissible to install a child watcher I<after> the child has been
		1461	forked (which implies it might have already exited), as long as the event
		1462	loop isn't entered (or is continued from a watcher).
		1463
		1464	Only the default event loop is capable of handling signals, and therefore
		1465	you can only rgeister child watchers in the default event loop.
		1466
		1467	=head3 Process Interaction
		1468
		1469	Libev grabs C<SIGCHLD> as soon as the default event loop is
		1470	initialised. This is necessary to guarantee proper behaviour even if
		1471	the first child watcher is started after the child exits. The occurance
		1472	of C<SIGCHLD> is recorded asynchronously, but child reaping is done
		1473	synchronously as part of the event loop processing. Libev always reaps all
		1474	children, even ones not watched.
		1475
		1476	=head3 Overriding the Built-In Processing
		1477
		1478	Libev offers no special support for overriding the built-in child
		1479	processing, but if your application collides with libev's default child
		1480	handler, you can override it easily by installing your own handler for
		1481	C<SIGCHLD> after initialising the default loop, and making sure the
		1482	default loop never gets destroyed. You are encouraged, however, to use an
		1483	event-based approach to child reaping and thus use libev's support for
		1484	that, so other libev users can use C<ev_child> watchers freely.
1381		1485
1382	=head3 Watcher-Specific Functions and Data Members	1486	=head3 Watcher-Specific Functions and Data Members
1383		1487
1384	=over 4	1488	=over 4
1385		1489
1386	=item ev_child_init (ev_child *, callback, int pid)	1490	=item ev_child_init (ev_child *, callback, int pid, int trace)
1387		1491
1388	=item ev_child_set (ev_child *, int pid)	1492	=item ev_child_set (ev_child *, int pid, int trace)
1389		1493
1390	Configures the watcher to wait for status changes of process C<pid> (or	1494	Configures the watcher to wait for status changes of process C<pid> (or
1391	I<any> process if C<pid> is specified as C<0>). The callback can look	1495	I<any> process if C<pid> is specified as C<0>). The callback can look
1392	at the C<rstatus> member of the C<ev_child> watcher structure to see	1496	at the C<rstatus> member of the C<ev_child> watcher structure to see
1393	the status word (use the macros from C<sys/wait.h> and see your systems	1497	the status word (use the macros from C<sys/wait.h> and see your systems
1394	C<waitpid> documentation). The C<rpid> member contains the pid of the	1498	C<waitpid> documentation). The C<rpid> member contains the pid of the
1395	process causing the status change.	1499	process causing the status change. C<trace> must be either C<0> (only
		1500	activate the watcher when the process terminates) or C<1> (additionally
		1501	activate the watcher when the process is stopped or continued).
1396		1502
1397	=item int pid [read-only]	1503	=item int pid [read-only]
1398		1504
1399	The process id this watcher watches out for, or C<0>, meaning any process id.	1505	The process id this watcher watches out for, or C<0>, meaning any process id.
1400		1506
…		…
1407	The process exit/trace status caused by C<rpid> (see your systems	1513	The process exit/trace status caused by C<rpid> (see your systems
1408	C<waitpid> and C<sys/wait.h> documentation for details).	1514	C<waitpid> and C<sys/wait.h> documentation for details).
1409		1515
1410	=back	1516	=back
1411		1517
1412	Example: Try to exit cleanly on SIGINT and SIGTERM.	1518	=head3 Examples
		1519
		1520	Example: C<fork()> a new process and install a child handler to wait for
		1521	its completion.
		1522
		1523	ev_child cw;
1413		1524
1414	static void	1525	static void
1415	sigint_cb (struct ev_loop loop, struct ev_signal w, int revents)	1526	child_cb (EV_P_ struct ev_child *w, int revents)
1416	{	1527	{
1417	ev_unloop (loop, EVUNLOOP_ALL);	1528	ev_child_stop (EV_A_ w);
		1529	printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1418	}	1530	}
1419		1531
1420	struct ev_signal signal_watcher;	1532	pid_t pid = fork ();
1421	ev_signal_init (&signal_watcher, sigint_cb, SIGINT);	1533
1422	ev_signal_start (loop, &sigint_cb);	1534	if (pid < 0)
		1535	// error
		1536	else if (pid == 0)
		1537	{
		1538	// the forked child executes here
		1539	exit (1);
		1540	}
		1541	else
		1542	{
		1543	ev_child_init (&cw, child_cb, pid, 0);
		1544	ev_child_start (EV_DEFAULT_ &cw);
		1545	}
1423		1546
1424		1547
1425	=head2 C<ev_stat> - did the file attributes just change?	1548	=head2 C<ev_stat> - did the file attributes just change?
1426		1549
1427	This watches a filesystem path for attribute changes. That is, it calls	1550	This watches a filesystem path for attribute changes. That is, it calls
…		…
1456	semantics of C<ev_stat> watchers, which means that libev sometimes needs	1579	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	1580	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	1581	usually detected immediately, and if the file exists there will be no
1459	polling.	1582	polling.
1460		1583
		1584	=head3 Inotify
		1585
		1586	When C<inotify (7)> support has been compiled into libev (generally only
		1587	available on Linux) and present at runtime, it will be used to speed up
		1588	change detection where possible. The inotify descriptor will be created lazily
		1589	when the first C<ev_stat> watcher is being started.
		1590
		1591	Inotify presense does not change the semantics of C<ev_stat> watchers
		1592	except that changes might be detected earlier, and in some cases, to avoid
		1593	making regular C<stat> calls. Even in the presense of inotify support
		1594	there are many cases where libev has to resort to regular C<stat> polling.
		1595
		1596	(There is no support for kqueue, as apparently it cannot be used to
		1597	implement this functionality, due to the requirement of having a file
		1598	descriptor open on the object at all times).
		1599
		1600	=head3 The special problem of stat time resolution
		1601
		1602	The C<stat ()> syscall only supports full-second resolution portably, and
		1603	even on systems where the resolution is higher, many filesystems still
		1604	only support whole seconds.
		1605
		1606	That means that, if the time is the only thing that changes, you might
		1607	miss updates: on the first update, C<ev_stat> detects a change and calls
		1608	your callback, which does something. When there is another update within
		1609	the same second, C<ev_stat> will be unable to detect it.
		1610
		1611	The solution to this is to delay acting on a change for a second (or till
		1612	the next second boundary), using a roughly one-second delay C<ev_timer>
		1613	(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
		1614	is added to work around small timing inconsistencies of some operating
		1615	systems.
		1616
1461	=head3 Watcher-Specific Functions and Data Members	1617	=head3 Watcher-Specific Functions and Data Members
1462		1618
1463	=over 4	1619	=over 4
1464		1620
1465	=item ev_stat_init (ev_stat , callback, const char path, ev_tstamp interval)	1621	=item ev_stat_init (ev_stat , callback, const char path, ev_tstamp interval)
…		…
1474		1630
1475	The callback will be receive C<EV_STAT> when a change was detected,	1631	The callback will be receive C<EV_STAT> when a change was detected,
1476	relative to the attributes at the time the watcher was started (or the	1632	relative to the attributes at the time the watcher was started (or the
1477	last change was detected).	1633	last change was detected).
1478		1634
1479	=item ev_stat_stat (ev_stat *)	1635	=item ev_stat_stat (loop, ev_stat *)
1480		1636
1481	Updates the stat buffer immediately with new values. If you change the	1637	Updates the stat buffer immediately with new values. If you change the
1482	watched path in your callback, you could call this fucntion to avoid	1638	watched path in your callback, you could call this fucntion to avoid
1483	detecting this change (while introducing a race condition). Can also be	1639	detecting this change (while introducing a race condition). Can also be
1484	useful simply to find out the new values.	1640	useful simply to find out the new values.
…		…
1502	=item const char *path [read-only]	1658	=item const char *path [read-only]
1503		1659
1504	The filesystem path that is being watched.	1660	The filesystem path that is being watched.
1505		1661
1506	=back	1662	=back
		1663
		1664	=head3 Examples
1507		1665
1508	Example: Watch C</etc/passwd> for attribute changes.	1666	Example: Watch C</etc/passwd> for attribute changes.
1509		1667
1510	static void	1668	static void
1511	passwd_cb (struct ev_loop loop, ev_stat w, int revents)	1669	passwd_cb (struct ev_loop loop, ev_stat w, int revents)
…		…
1524	}	1682	}
1525		1683
1526	...	1684	...
1527	ev_stat passwd;	1685	ev_stat passwd;
1528		1686
1529	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	1687	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1530	ev_stat_start (loop, &passwd);	1688	ev_stat_start (loop, &passwd);
		1689
		1690	Example: Like above, but additionally use a one-second delay so we do not
		1691	miss updates (however, frequent updates will delay processing, too, so
		1692	one might do the work both on C<ev_stat> callback invocation I<and> on
		1693	C<ev_timer> callback invocation).
		1694
		1695	static ev_stat passwd;
		1696	static ev_timer timer;
		1697
		1698	static void
		1699	timer_cb (EV_P_ ev_timer *w, int revents)
		1700	{
		1701	ev_timer_stop (EV_A_ w);
		1702
		1703	/* now it's one second after the most recent passwd change */
		1704	}
		1705
		1706	static void
		1707	stat_cb (EV_P_ ev_stat *w, int revents)
		1708	{
		1709	/* reset the one-second timer */
		1710	ev_timer_again (EV_A_ &timer);
		1711	}
		1712
		1713	...
		1714	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		1715	ev_stat_start (loop, &passwd);
		1716	ev_timer_init (&timer, timer_cb, 0., 1.01);
1531		1717
1532		1718
1533	=head2 C<ev_idle> - when you've got nothing better to do...	1719	=head2 C<ev_idle> - when you've got nothing better to do...
1534		1720
1535	Idle watchers trigger events when no other events of the same or higher	1721	Idle watchers trigger events when no other events of the same or higher
…		…
1561	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	1747	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1562	believe me.	1748	believe me.
1563		1749
1564	=back	1750	=back
1565		1751
		1752	=head3 Examples
		1753
1566	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	1754	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1567	callback, free it. Also, use no error checking, as usual.	1755	callback, free it. Also, use no error checking, as usual.
1568		1756
1569	static void	1757	static void
1570	idle_cb (struct ev_loop loop, struct ev_idle w, int revents)	1758	idle_cb (struct ev_loop loop, struct ev_idle w, int revents)
1571	{	1759	{
1572	free (w);	1760	free (w);
1573	// now do something you wanted to do when the program has	1761	// now do something you wanted to do when the program has
1574	// no longer asnything immediate to do.	1762	// no longer anything immediate to do.
1575	}	1763	}
1576		1764
1577	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));	1765	struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1578	ev_idle_init (idle_watcher, idle_cb);	1766	ev_idle_init (idle_watcher, idle_cb);
1579	ev_idle_start (loop, idle_cb);	1767	ev_idle_start (loop, idle_cb);
…		…
1621		1809
1622	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	1810	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	1811	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,	1812	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	1813	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	1814	supports this, they will be called before other C<ev_check> watchers
1627	their job. As C<ev_check> watchers are often used to embed other event	1815	did their job. As C<ev_check> watchers are often used to embed other
1628	loops those other event loops might be in an unusable state until their	1816	(non-libev) event loops those other event loops might be in an unusable
1629	C<ev_check> watcher ran (always remind yourself to coexist peacefully with	1817	state until their C<ev_check> watcher ran (always remind yourself to
1630	others).	1818	coexist peacefully with others).
1631		1819
1632	=head3 Watcher-Specific Functions and Data Members	1820	=head3 Watcher-Specific Functions and Data Members
1633		1821
1634	=over 4	1822	=over 4
1635		1823
…		…
1640	Initialises and configures the prepare or check watcher - they have no	1828	Initialises and configures the prepare or check watcher - they have no
1641	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	1829	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1642	macros, but using them is utterly, utterly and completely pointless.	1830	macros, but using them is utterly, utterly and completely pointless.
1643		1831
1644	=back	1832	=back
		1833
		1834	=head3 Examples
1645		1835
1646	There are a number of principal ways to embed other event loops or modules	1836	There are a number of principal ways to embed other event loops or modules
1647	into libev. Here are some ideas on how to include libadns into libev	1837	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	1838	(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>	1839	use for an actually working example. Another Perl module named C<EV::Glib>
…		…
1774	=head2 C<ev_embed> - when one backend isn't enough...	1964	=head2 C<ev_embed> - when one backend isn't enough...
1775		1965
1776	This is a rather advanced watcher type that lets you embed one event loop	1966	This is a rather advanced watcher type that lets you embed one event loop
1777	into another (currently only C<ev_io> events are supported in the embedded	1967	into another (currently only C<ev_io> events are supported in the embedded
1778	loop, other types of watchers might be handled in a delayed or incorrect	1968	loop, other types of watchers might be handled in a delayed or incorrect
1779	fashion and must not be used). (See portability notes, below).	1969	fashion and must not be used).
1780		1970
1781	There are primarily two reasons you would want that: work around bugs and	1971	There are primarily two reasons you would want that: work around bugs and
1782	prioritise I/O.	1972	prioritise I/O.
1783		1973
1784	As an example for a bug workaround, the kqueue backend might only support	1974	As an example for a bug workaround, the kqueue backend might only support
…		…
1818	portable one.	2008	portable one.
1819		2009
1820	So when you want to use this feature you will always have to be prepared	2010	So when you want to use this feature you will always have to be prepared
1821	that you cannot get an embeddable loop. The recommended way to get around	2011	that you cannot get an embeddable loop. The recommended way to get around
1822	this is to have a separate variables for your embeddable loop, try to	2012	this is to have a separate variables for your embeddable loop, try to
1823	create it, and if that fails, use the normal loop for everything:	2013	create it, and if that fails, use the normal loop for everything.
		2014
		2015	=head3 Watcher-Specific Functions and Data Members
		2016
		2017	=over 4
		2018
		2019	=item ev_embed_init (ev_embed , callback, struct ev_loop embedded_loop)
		2020
		2021	=item ev_embed_set (ev_embed , callback, struct ev_loop embedded_loop)
		2022
		2023	Configures the watcher to embed the given loop, which must be
		2024	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		2025	invoked automatically, otherwise it is the responsibility of the callback
		2026	to invoke it (it will continue to be called until the sweep has been done,
		2027	if you do not want thta, you need to temporarily stop the embed watcher).
		2028
		2029	=item ev_embed_sweep (loop, ev_embed *)
		2030
		2031	Make a single, non-blocking sweep over the embedded loop. This works
		2032	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		2033	apropriate way for embedded loops.
		2034
		2035	=item struct ev_loop *other [read-only]
		2036
		2037	The embedded event loop.
		2038
		2039	=back
		2040
		2041	=head3 Examples
		2042
		2043	Example: Try to get an embeddable event loop and embed it into the default
		2044	event loop. If that is not possible, use the default loop. The default
		2045	loop is stored in C<loop_hi>, while the mebeddable loop is stored in
		2046	C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
		2047	used).
1824		2048
1825	struct ev_loop *loop_hi = ev_default_init (0);	2049	struct ev_loop *loop_hi = ev_default_init (0);
1826	struct ev_loop *loop_lo = 0;	2050	struct ev_loop *loop_lo = 0;
1827	struct ev_embed embed;	2051	struct ev_embed embed;
1828		2052
…		…
1839	ev_embed_start (loop_hi, &embed);	2063	ev_embed_start (loop_hi, &embed);
1840	}	2064	}
1841	else	2065	else
1842	loop_lo = loop_hi;	2066	loop_lo = loop_hi;
1843		2067
1844	=head2 Portability notes	2068	Example: Check if kqueue is available but not recommended and create
		2069	a kqueue backend for use with sockets (which usually work with any
		2070	kqueue implementation). Store the kqueue/socket-only event loop in
		2071	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1845		2072
1846	Kqueue is nominally embeddable, but this is broken on all BSDs that I	2073	struct ev_loop *loop = ev_default_init (0);
1847	tried, in various ways. Usually the embedded event loop will simply never	2074	struct ev_loop *loop_socket = 0;
1848	receive events, sometimes it will only trigger a few times, sometimes in a	2075	struct ev_embed embed;
1849	loop. Epoll is also nominally embeddable, but many Linux kernel versions	2076
1850	will always eport the epoll fd as ready, even when no events are pending.	2077	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2078	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2079	{
		2080	ev_embed_init (&embed, 0, loop_socket);
		2081	ev_embed_start (loop, &embed);
		2082	}
1851		2083
1852	While libev allows embedding these backends (they are contained in	2084	if (!loop_socket)
1853	C<ev_embeddable_backends ()>), take extreme care that it will actually	2085	loop_socket = loop;
1854	work.
1855		2086
1856	When in doubt, create a dynamic event loop forced to use sockets (this	2087	// now use loop_socket for all sockets, and loop for everything else
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
1862	=over 4
1863
1864	=item ev_embed_init (ev_embed , callback, struct ev_loop embedded_loop)
1865
1866	=item ev_embed_set (ev_embed , callback, struct ev_loop embedded_loop)
1867
1868	Configures the watcher to embed the given loop, which must be
1869	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1870	invoked automatically, otherwise it is the responsibility of the callback
1871	to invoke it (it will continue to be called until the sweep has been done,
1872	if you do not want thta, you need to temporarily stop the embed watcher).
1873
1874	=item ev_embed_sweep (loop, ev_embed *)
1875
1876	Make a single, non-blocking sweep over the embedded loop. This works
1877	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1878	apropriate way for embedded loops.
1879
1880	=item struct ev_loop *other [read-only]
1881
1882	The embedded event loop.
1883
1884	=back
1885		2088
1886		2089
1887	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	2090	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1888		2091
1889	Fork watchers are called when a C<fork ()> was detected (usually because	2092	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
1905	believe me.	2108	believe me.
1906		2109
1907	=back	2110	=back
1908		2111
1909		2112
		2113	=head2 C<ev_async> - how to wake up another event loop
		2114
		2115	In general, you cannot use an C<ev_loop> from multiple threads or other
		2116	asynchronous sources such as signal handlers (as opposed to multiple event
		2117	loops - those are of course safe to use in different threads).
		2118
		2119	Sometimes, however, you need to wake up another event loop you do not
		2120	control, for example because it belongs to another thread. This is what
		2121	C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
		2122	can signal it by calling C<ev_async_send>, which is thread- and signal
		2123	safe.
		2124
		2125	This functionality is very similar to C<ev_signal> watchers, as signals,
		2126	too, are asynchronous in nature, and signals, too, will be compressed
		2127	(i.e. the number of callback invocations may be less than the number of
		2128	C<ev_async_sent> calls).
		2129
		2130	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
		2131	just the default loop.
		2132
		2133	=head3 Queueing
		2134
		2135	C<ev_async> does not support queueing of data in any way. The reason
		2136	is that the author does not know of a simple (or any) algorithm for a
		2137	multiple-writer-single-reader queue that works in all cases and doesn't
		2138	need elaborate support such as pthreads.
		2139
		2140	That means that if you want to queue data, you have to provide your own
		2141	queue. But at least I can tell you would implement locking around your
		2142	queue:
		2143
		2144	=over 4
		2145
		2146	=item queueing from a signal handler context
		2147
		2148	To implement race-free queueing, you simply add to the queue in the signal
		2149	handler but you block the signal handler in the watcher callback. Here is an example that does that for
		2150	some fictitiuous SIGUSR1 handler:
		2151
		2152	static ev_async mysig;
		2153
		2154	static void
		2155	sigusr1_handler (void)
		2156	{
		2157	sometype data;
		2158
		2159	// no locking etc.
		2160	queue_put (data);
		2161	ev_async_send (EV_DEFAULT_ &mysig);
		2162	}
		2163
		2164	static void
		2165	mysig_cb (EV_P_ ev_async *w, int revents)
		2166	{
		2167	sometype data;
		2168	sigset_t block, prev;
		2169
		2170	sigemptyset (&block);
		2171	sigaddset (&block, SIGUSR1);
		2172	sigprocmask (SIG_BLOCK, &block, &prev);
		2173
		2174	while (queue_get (&data))
		2175	process (data);
		2176
		2177	if (sigismember (&prev, SIGUSR1)
		2178	sigprocmask (SIG_UNBLOCK, &block, 0);
		2179	}
		2180
		2181	(Note: pthreads in theory requires you to use C<pthread_setmask>
		2182	instead of C<sigprocmask> when you use threads, but libev doesn't do it
		2183	either...).
		2184
		2185	=item queueing from a thread context
		2186
		2187	The strategy for threads is different, as you cannot (easily) block
		2188	threads but you can easily preempt them, so to queue safely you need to
		2189	employ a traditional mutex lock, such as in this pthread example:
		2190
		2191	static ev_async mysig;
		2192	static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
		2193
		2194	static void
		2195	otherthread (void)
		2196	{
		2197	// only need to lock the actual queueing operation
		2198	pthread_mutex_lock (&mymutex);
		2199	queue_put (data);
		2200	pthread_mutex_unlock (&mymutex);
		2201
		2202	ev_async_send (EV_DEFAULT_ &mysig);
		2203	}
		2204
		2205	static void
		2206	mysig_cb (EV_P_ ev_async *w, int revents)
		2207	{
		2208	pthread_mutex_lock (&mymutex);
		2209
		2210	while (queue_get (&data))
		2211	process (data);
		2212
		2213	pthread_mutex_unlock (&mymutex);
		2214	}
		2215
		2216	=back
		2217
		2218
		2219	=head3 Watcher-Specific Functions and Data Members
		2220
		2221	=over 4
		2222
		2223	=item ev_async_init (ev_async *, callback)
		2224
		2225	Initialises and configures the async watcher - it has no parameters of any
		2226	kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,
		2227	believe me.
		2228
		2229	=item ev_async_send (loop, ev_async *)
		2230
		2231	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
		2232	an C<EV_ASYNC> event on the watcher into the event loop. Unlike
		2233	C<ev_feed_event>, this call is safe to do in other threads, signal or
		2234	similar contexts (see the dicusssion of C<EV_ATOMIC_T> in the embedding
		2235	section below on what exactly this means).
		2236
		2237	This call incurs the overhead of a syscall only once per loop iteration,
		2238	so while the overhead might be noticable, it doesn't apply to repeated
		2239	calls to C<ev_async_send>.
		2240
		2241	=back
		2242
		2243
1910	=head1 OTHER FUNCTIONS	2244	=head1 OTHER FUNCTIONS
1911		2245
1912	There are some other functions of possible interest. Described. Here. Now.	2246	There are some other functions of possible interest. Described. Here. Now.
1913		2247
1914	=over 4	2248	=over 4
…		…
2141	Example: Define a class with an IO and idle watcher, start one of them in	2475	Example: Define a class with an IO and idle watcher, start one of them in
2142	the constructor.	2476	the constructor.
2143		2477
2144	class myclass	2478	class myclass
2145	{	2479	{
2146	ev_io io; void io_cb (ev::io &w, int revents);	2480	ev::io io; void io_cb (ev::io &w, int revents);
2147	ev_idle idle void idle_cb (ev::idle &w, int revents);	2481	ev:idle idle void idle_cb (ev::idle &w, int revents);
2148		2482
2149	myclass ();	2483	myclass (int fd)
2150	}
2151
2152	myclass::myclass (int fd)
2153	{	2484	{
2154	io .set <myclass, &myclass::io_cb > (this);	2485	io .set <myclass, &myclass::io_cb > (this);
2155	idle.set <myclass, &myclass::idle_cb> (this);	2486	idle.set <myclass, &myclass::idle_cb> (this);
2156		2487
2157	io.start (fd, ev::READ);	2488	io.start (fd, ev::READ);
		2489	}
2158	}	2490	};
2159		2491
2160		2492
2161	=head1 MACRO MAGIC	2493	=head1 MACRO MAGIC
2162		2494
2163	Libev can be compiled with a variety of options, the most fundamantal	2495	Libev can be compiled with a variety of options, the most fundamantal
…		…
2368	wants osf handles on win32 (this is the case when the select to	2700	wants osf handles on win32 (this is the case when the select to
2369	be used is the winsock select). This means that it will call	2701	be used is the winsock select). This means that it will call
2370	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	2702	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2371	it is assumed that all these functions actually work on fds, even	2703	it is assumed that all these functions actually work on fds, even
2372	on win32. Should not be defined on non-win32 platforms.	2704	on win32. Should not be defined on non-win32 platforms.
		2705
		2706	=item EV_FD_TO_WIN32_HANDLE
		2707
		2708	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		2709	file descriptors to socket handles. When not defining this symbol (the
		2710	default), then libev will call C<_get_osfhandle>, which is usually
		2711	correct. In some cases, programs use their own file descriptor management,
		2712	in which case they can provide this function to map fds to socket handles.
2373		2713
2374	=item EV_USE_POLL	2714	=item EV_USE_POLL
2375		2715
2376	If defined to be C<1>, libev will compile in support for the C<poll>(2)	2716	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2377	backend. Otherwise it will be enabled on non-win32 platforms. It	2717	backend. Otherwise it will be enabled on non-win32 platforms. It
…		…
2411		2751
2412	If defined to be C<1>, libev will compile in support for the Linux inotify	2752	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	2753	interface to speed up C<ev_stat> watchers. Its actual availability will
2414	be detected at runtime.	2754	be detected at runtime.
2415		2755
		2756	=item EV_ATOMIC_T
		2757
		2758	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
		2759	access is atomic with respect to other threads or signal contexts. No such
		2760	type is easily found in the C language, so you can provide your own type
		2761	that you know is safe for your purposes. It is used both for signal handler "locking"
		2762	as well as for signal and thread safety in C<ev_async> watchers.
		2763
		2764	In the absense of this define, libev will use C<sig_atomic_t volatile>
		2765	(from F<signal.h>), which is usually good enough on most platforms.
		2766
2416	=item EV_H	2767	=item EV_H
2417		2768
2418	The name of the F<ev.h> header file used to include it. The default if	2769	The name of the F<ev.h> header file used to include it. The default if
2419	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	2770	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2420	can be used to virtually rename the F<ev.h> header file in case of conflicts.	2771	used to virtually rename the F<ev.h> header file in case of conflicts.
2421		2772
2422	=item EV_CONFIG_H	2773	=item EV_CONFIG_H
2423		2774
2424	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	2775	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2425	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	2776	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2426	C<EV_H>, above.	2777	C<EV_H>, above.
2427		2778
2428	=item EV_EVENT_H	2779	=item EV_EVENT_H
2429		2780
2430	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	2781	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2431	of how the F<event.h> header can be found.	2782	of how the F<event.h> header can be found, the default is C<"event.h">.
2432		2783
2433	=item EV_PROTOTYPES	2784	=item EV_PROTOTYPES
2434		2785
2435	If defined to be C<0>, then F<ev.h> will not define any function	2786	If defined to be C<0>, then F<ev.h> will not define any function
2436	prototypes, but still define all the structs and other symbols. This is	2787	prototypes, but still define all the structs and other symbols. This is
…		…
2487	=item EV_FORK_ENABLE	2838	=item EV_FORK_ENABLE
2488		2839
2489	If undefined or defined to be C<1>, then fork watchers are supported. If	2840	If undefined or defined to be C<1>, then fork watchers are supported. If
2490	defined to be C<0>, then they are not.	2841	defined to be C<0>, then they are not.
2491		2842
		2843	=item EV_ASYNC_ENABLE
		2844
		2845	If undefined or defined to be C<1>, then async watchers are supported. If
		2846	defined to be C<0>, then they are not.
		2847
2492	=item EV_MINIMAL	2848	=item EV_MINIMAL
2493		2849
2494	If you need to shave off some kilobytes of code at the expense of some	2850	If you need to shave off some kilobytes of code at the expense of some
2495	speed, define this symbol to C<1>. Currently only used for gcc to override	2851	speed, define this symbol to C<1>. Currently only used for gcc to override
2496	some inlining decisions, saves roughly 30% codesize of amd64.	2852	some inlining decisions, saves roughly 30% codesize of amd64.
…		…
2502	than enough. If you need to manage thousands of children you might want to	2858	than enough. If you need to manage thousands of children you might want to
2503	increase this value (I<must> be a power of two).	2859	increase this value (I<must> be a power of two).
2504		2860
2505	=item EV_INOTIFY_HASHSIZE	2861	=item EV_INOTIFY_HASHSIZE
2506		2862
2507	C<ev_staz> watchers use a small hash table to distribute workload by	2863	C<ev_stat> 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>),	2864	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>	2865	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	2866	watchers you might want to increase this value (I<must> be a power of
2511	two).	2867	two).
2512		2868
…		…
2608		2964
2609	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	2965	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2610		2966
2611	This means that, when you have a watcher that triggers in one hour and	2967	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	2968	there are 100 watchers that would trigger before that then inserting will
2613	have to skip those 100 watchers.	2969	have to skip roughly seven (C<ld 100>) of these watchers.
2614		2970
2615	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	2971	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2616		2972
2617	That means that for changing a timer costs less than removing/adding them	2973	That means that changing a timer costs less than removing/adding them
2618	as only the relative motion in the event queue has to be paid for.	2974	as only the relative motion in the event queue has to be paid for.
2619		2975
2620	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	2976	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2621		2977
2622	These just add the watcher into an array or at the head of a list.	2978	These just add the watcher into an array or at the head of a list.
		2979
2623	=item Stopping check/prepare/idle watchers: O(1)	2980	=item Stopping check/prepare/idle/fork/async watchers: O(1)
2624		2981
2625	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	2982	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2626		2983
2627	These watchers are stored in lists then need to be walked to find the	2984	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	2985	correct watcher to remove. The lists are usually short (you don't usually
2629	have many watchers waiting for the same fd or signal).	2986	have many watchers waiting for the same fd or signal).
2630		2987
2631	=item Finding the next timer per loop iteration: O(1)	2988	=item Finding the next timer in each loop iteration: O(1)
		2989
		2990	By virtue of using a binary heap, the next timer is always found at the
		2991	beginning of the storage array.
2632		2992
2633	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	2993	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2634		2994
2635	A change means an I/O watcher gets started or stopped, which requires	2995	A change means an I/O watcher gets started or stopped, which requires
2636	libev to recalculate its status (and possibly tell the kernel).	2996	libev to recalculate its status (and possibly tell the kernel, depending
		2997	on backend and wether C<ev_io_set> was used).
2637		2998
2638	=item Activating one watcher: O(1)	2999	=item Activating one watcher (putting it into the pending state): O(1)
2639		3000
2640	=item Priority handling: O(number_of_priorities)	3001	=item Priority handling: O(number_of_priorities)
2641		3002
2642	Priorities are implemented by allocating some space for each	3003	Priorities are implemented by allocating some space for each
2643	priority. When doing priority-based operations, libev usually has to	3004	priority. When doing priority-based operations, libev usually has to
2644	linearly search all the priorities.	3005	linearly search all the priorities, but starting/stopping and activating
		3006	watchers becomes O(1) w.r.t. priority handling.
		3007
		3008	=item Sending an ev_async: O(1)
		3009
		3010	=item Processing ev_async_send: O(number_of_async_watchers)
		3011
		3012	=item Processing signals: O(max_signal_number)
		3013
		3014	Sending involves a syscall I<iff> there were no other C<ev_async_send>
		3015	calls in the current loop iteration. Checking for async and signal events
		3016	involves iterating over all running async watchers or all signal numbers.
2645		3017
2646	=back	3018	=back
2647		3019
2648		3020
		3021	=head1 Win32 platform limitations and workarounds
		3022
		3023	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		3024	requires, and its I/O model is fundamentally incompatible with the POSIX
		3025	model. Libev still offers limited functionality on this platform in
		3026	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		3027	descriptors. This only applies when using Win32 natively, not when using
		3028	e.g. cygwin.
		3029
		3030	There is no supported compilation method available on windows except
		3031	embedding it into other applications.
		3032
		3033	Due to the many, low, and arbitrary limits on the win32 platform and the
		3034	abysmal performance of winsockets, using a large number of sockets is not
		3035	recommended (and not reasonable). If your program needs to use more than
		3036	a hundred or so sockets, then likely it needs to use a totally different
		3037	implementation for windows, as libev offers the POSIX model, which cannot
		3038	be implemented efficiently on windows (microsoft monopoly games).
		3039
		3040	=over 4
		3041
		3042	=item The winsocket select function
		3043
		3044	The winsocket C<select> function doesn't follow POSIX in that it requires
		3045	socket I<handles> and not socket I<file descriptors>. This makes select
		3046	very inefficient, and also requires a mapping from file descriptors
		3047	to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
		3048	C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
		3049	symbols for more info.
		3050
		3051	The configuration for a "naked" win32 using the microsoft runtime
		3052	libraries and raw winsocket select is:
		3053
		3054	#define EV_USE_SELECT 1
		3055	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		3056
		3057	Note that winsockets handling of fd sets is O(n), so you can easily get a
		3058	complexity in the O(n²) range when using win32.
		3059
		3060	=item Limited number of file descriptors
		3061
		3062	Windows has numerous arbitrary (and low) limits on things. Early versions
		3063	of winsocket's select only supported waiting for a max. of C<64> handles
		3064	(probably owning to the fact that all windows kernels can only wait for
		3065	C<64> things at the same time internally; microsoft recommends spawning a
		3066	chain of threads and wait for 63 handles and the previous thread in each).
		3067
		3068	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		3069	to some high number (e.g. C<2048>) before compiling the winsocket select
		3070	call (which might be in libev or elsewhere, for example, perl does its own
		3071	select emulation on windows).
		3072
		3073	Another limit is the number of file descriptors in the microsoft runtime
		3074	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		3075	or something like this inside microsoft). You can increase this by calling
		3076	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		3077	arbitrary limit), but is broken in many versions of the microsoft runtime
		3078	libraries.
		3079
		3080	This might get you to about C<512> or C<2048> sockets (depending on
		3081	windows version and/or the phase of the moon). To get more, you need to
		3082	wrap all I/O functions and provide your own fd management, but the cost of
		3083	calling select (O(n²)) will likely make this unworkable.
		3084
		3085	=back
		3086
		3087
2649	=head1 AUTHOR	3088	=head1 AUTHOR
2650		3089
2651	Marc Lehmann <libev@schmorp.de>.	3090	Marc Lehmann <libev@schmorp.de>.
2652		3091

Diff Legend

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

Comparing libev/ev.pod (file contents): Revision 1.97 by root, Sat Dec 22 05:48:02 2007 UTC vs. Revision 1.134 by root, Sat Mar 8 07:04:56 2008 UTC

Diff Legend

Comparing libev/ev.pod (file contents):
Revision 1.97 by root, Sat Dec 22 05:48:02 2007 UTC vs.
Revision 1.134 by root, Sat Mar 8 07:04:56 2008 UTC