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

Comparing libev/ev.pod (file contents):
Revision 1.94 by root, Fri Dec 21 04:38:45 2007 UTC vs.
Revision 1.115 by root, Mon Dec 31 01:32:59 2007 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
…		…
115		115
116	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
117	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
118	you actually want to know.	118	you actually want to know.
119		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
120	=item int ev_version_major ()	126	=item int ev_version_major ()
121		127
122	=item int ev_version_minor ()	128	=item int ev_version_minor ()
123		129
124	You can find out the major and minor ABI version numbers of the library	130	You can find out the major and minor ABI version numbers of the library
…		…
300	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	306	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
301		307
302	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
303	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,
304	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
305	using this backend. It doesn't scale too well (O(highest_fd)), but its usually	311	using this backend. It doesn't scale too well (O(highest_fd)), but its
306	the fastest backend for a low number of fds.	312	usually the fastest backend for a low number of (low-numbered :) fds.
		313
		314	To get good performance out of this backend you need a high amount of
		315	parallelity (most of the file descriptors should be busy). If you are
		316	writing a server, you should C<accept ()> in a loop to accept as many
		317	connections as possible during one iteration. You might also want to have
		318	a look at C<ev_set_io_collect_interval ()> to increase the amount of
		319	readyness notifications you get per iteration.
307		320
308	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	321	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
309		322
310	And this is your standard poll(2) backend. It's more complicated than	323	And this is your standard poll(2) backend. It's more complicated
311	select, but handles sparse fds better and has no artificial limit on the	324	than select, but handles sparse fds better and has no artificial
312	number of fds you can use (except it will slow down considerably with a	325	limit on the number of fds you can use (except it will slow down
313	lot of inactive fds). It scales similarly to select, i.e. O(total_fds).	326	considerably with a lot of inactive fds). It scales similarly to select,
		327	i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
		328	performance tips.
314		329
315	=item C<EVBACKEND_EPOLL> (value 4, Linux)	330	=item C<EVBACKEND_EPOLL> (value 4, Linux)
316		331
317	For few fds, this backend is a bit little slower than poll and select,	332	For few fds, this backend is a bit little slower than poll and select,
318	but it scales phenomenally better. While poll and select usually scale	333	but it scales phenomenally better. While poll and select usually scale
319	like O(total_fds) where n is the total number of fds (or the highest fd),	334	like O(total_fds) where n is the total number of fds (or the highest fd),
320	epoll scales either O(1) or O(active_fds). The epoll design has a number	335	epoll scales either O(1) or O(active_fds). The epoll design has a number
321	of shortcomings, such as silently dropping events in some hard-to-detect	336	of shortcomings, such as silently dropping events in some hard-to-detect
322	cases and rewuiring a syscall per fd change, no fork support and bad	337	cases and rewiring a syscall per fd change, no fork support and bad
323	support for dup:	338	support for dup.
324		339
325	While stopping, setting and starting an I/O watcher in the same iteration	340	While stopping, setting and starting an I/O watcher in the same iteration
326	will result in some caching, there is still a syscall per such incident	341	will result in some caching, there is still a syscall per such incident
327	(because the fd could point to a different file description now), so its	342	(because the fd could point to a different file description now), so its
328	best to avoid that. Also, C<dup ()>'ed file descriptors might not work	343	best to avoid that. Also, C<dup ()>'ed file descriptors might not work
…		…
330		345
331	Please note that epoll sometimes generates spurious notifications, so you	346	Please note that epoll sometimes generates spurious notifications, so you
332	need to use non-blocking I/O or other means to avoid blocking when no data	347	need to use non-blocking I/O or other means to avoid blocking when no data
333	(or space) is available.	348	(or space) is available.
334		349
		350	Best performance from this backend is achieved by not unregistering all
		351	watchers for a file descriptor until it has been closed, if possible, i.e.
		352	keep at least one watcher active per fd at all times.
		353
		354	While nominally embeddeble in other event loops, this feature is broken in
		355	all kernel versions tested so far.
		356
335	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)	357	=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
336		358
337	Kqueue deserves special mention, as at the time of this writing, it	359	Kqueue deserves special mention, as at the time of this writing, it
338	was broken on I<all> BSDs (usually it doesn't work with anything but	360	was broken on all BSDs except NetBSD (usually it doesn't work reliably
339	sockets and pipes, except on Darwin, where of course it's completely	361	with anything but sockets and pipes, except on Darwin, where of course
340	useless. On NetBSD, it seems to work for all the FD types I tested, so it
341	is used by default there). For this reason it's not being "autodetected"	362	it's completely useless). For this reason it's not being "autodetected"
342	unless you explicitly specify it explicitly in the flags (i.e. using	363	unless you explicitly specify it explicitly in the flags (i.e. using
343	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)	364	C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
344	system like NetBSD.	365	system like NetBSD.
345		366
		367	You still can embed kqueue into a normal poll or select backend and use it
		368	only for sockets (after having made sure that sockets work with kqueue on
		369	the target platform). See C<ev_embed> watchers for more info.
		370
346	It scales in the same way as the epoll backend, but the interface to the	371	It scales in the same way as the epoll backend, but the interface to the
347	kernel is more efficient (which says nothing about its actual speed,	372	kernel is more efficient (which says nothing about its actual speed, of
348	of course). While stopping, setting and starting an I/O watcher does	373	course). While stopping, setting and starting an I/O watcher does never
349	never cause an extra syscall as with epoll, it still adds up to two event	374	cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
350	changes per incident, support for C<fork ()> is very bad and it drops fds	375	two event changes per incident, support for C<fork ()> is very bad and it
351	silently in similarly hard-to-detetc cases.	376	drops fds silently in similarly hard-to-detect cases.
		377
		378	This backend usually performs well under most conditions.
		379
		380	While nominally embeddable in other event loops, this doesn't work
		381	everywhere, so you might need to test for this. And since it is broken
		382	almost everywhere, you should only use it when you have a lot of sockets
		383	(for which it usually works), by embedding it into another event loop
		384	(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for
		385	sockets.
352		386
353	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)	387	=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
354		388
355	This is not implemented yet (and might never be).	389	This is not implemented yet (and might never be, unless you send me an
		390	implementation). According to reports, C</dev/poll> only supports sockets
		391	and is not embeddable, which would limit the usefulness of this backend
		392	immensely.
356		393
357	=item C<EVBACKEND_PORT> (value 32, Solaris 10)	394	=item C<EVBACKEND_PORT> (value 32, Solaris 10)
358		395
359	This uses the Solaris 10 event port mechanism. As with everything on Solaris,	396	This uses the Solaris 10 event port mechanism. As with everything on Solaris,
360	it's really slow, but it still scales very well (O(active_fds)).	397	it's really slow, but it still scales very well (O(active_fds)).
361		398
362	Please note that solaris event ports can deliver a lot of spurious	399	Please note that solaris event ports can deliver a lot of spurious
363	notifications, so you need to use non-blocking I/O or other means to avoid	400	notifications, so you need to use non-blocking I/O or other means to avoid
364	blocking when no data (or space) is available.	401	blocking when no data (or space) is available.
365		402
		403	While this backend scales well, it requires one system call per active
		404	file descriptor per loop iteration. For small and medium numbers of file
		405	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
		406	might perform better.
		407
366	=item C<EVBACKEND_ALL>	408	=item C<EVBACKEND_ALL>
367		409
368	Try all backends (even potentially broken ones that wouldn't be tried	410	Try all backends (even potentially broken ones that wouldn't be tried
369	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as	411	with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
370	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.	412	C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
		413
		414	It is definitely not recommended to use this flag.
371		415
372	=back	416	=back
373		417
374	If one or more of these are ored into the flags value, then only these	418	If one or more of these are ored into the flags value, then only these
375	backends will be tried (in the reverse order as given here). If none are	419	backends will be tried (in the reverse order as given here). If none are
…		…
507	usually a better approach for this kind of thing.	551	usually a better approach for this kind of thing.
508		552
509	Here are the gory details of what C<ev_loop> does:	553	Here are the gory details of what C<ev_loop> does:
510		554
511	- Before the first iteration, call any pending watchers.	555	- Before the first iteration, call any pending watchers.
512	* If there are no active watchers (reference count is zero), return.	556	* If EVFLAG_FORKCHECK was used, check for a fork.
513	- Queue all prepare watchers and then call all outstanding watchers.	557	- If a fork was detected, queue and call all fork watchers.
		558	- Queue and call all prepare watchers.
514	- If we have been forked, recreate the kernel state.	559	- If we have been forked, recreate the kernel state.
515	- Update the kernel state with all outstanding changes.	560	- Update the kernel state with all outstanding changes.
516	- Update the "event loop time".	561	- Update the "event loop time".
517	- Calculate for how long to block.	562	- Calculate for how long to sleep or block, if at all
		563	(active idle watchers, EVLOOP_NONBLOCK or not having
		564	any active watchers at all will result in not sleeping).
		565	- Sleep if the I/O and timer collect interval say so.
518	- Block the process, waiting for any events.	566	- Block the process, waiting for any events.
519	- Queue all outstanding I/O (fd) events.	567	- Queue all outstanding I/O (fd) events.
520	- Update the "event loop time" and do time jump handling.	568	- Update the "event loop time" and do time jump handling.
521	- Queue all outstanding timers.	569	- Queue all outstanding timers.
522	- Queue all outstanding periodics.	570	- Queue all outstanding periodics.
523	- If no events are pending now, queue all idle watchers.	571	- If no events are pending now, queue all idle watchers.
524	- Queue all check watchers.	572	- Queue all check watchers.
525	- Call all queued watchers in reverse order (i.e. check watchers first).	573	- Call all queued watchers in reverse order (i.e. check watchers first).
526	Signals and child watchers are implemented as I/O watchers, and will	574	Signals and child watchers are implemented as I/O watchers, and will
527	be handled here by queueing them when their watcher gets executed.	575	be handled here by queueing them when their watcher gets executed.
528	- If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK	576	- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
529	were used, return, otherwise continue with step *.	577	were used, or there are no active watchers, return, otherwise
		578	continue with step *.
530		579
531	Example: Queue some jobs and then loop until no events are outsanding	580	Example: Queue some jobs and then loop until no events are outstanding
532	anymore.	581	anymore.
533		582
534	... queue jobs here, make sure they register event watchers as long	583	... queue jobs here, make sure they register event watchers as long
535	... as they still have work to do (even an idle watcher will do..)	584	... as they still have work to do (even an idle watcher will do..)
536	ev_loop (my_loop, 0);	585	ev_loop (my_loop, 0);
…		…
540		589
541	Can be used to make a call to C<ev_loop> return early (but only after it	590	Can be used to make a call to C<ev_loop> return early (but only after it
542	has processed all outstanding events). The C<how> argument must be either	591	has processed all outstanding events). The C<how> argument must be either
543	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or	592	C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
544	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.	593	C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
		594
		595	This "unloop state" will be cleared when entering C<ev_loop> again.
545		596
546	=item ev_ref (loop)	597	=item ev_ref (loop)
547		598
548	=item ev_unref (loop)	599	=item ev_unref (loop)
549		600
…		…
569	Example: For some weird reason, unregister the above signal handler again.	620	Example: For some weird reason, unregister the above signal handler again.
570		621
571	ev_ref (loop);	622	ev_ref (loop);
572	ev_signal_stop (loop, &exitsig);	623	ev_signal_stop (loop, &exitsig);
573		624
		625	=item ev_set_io_collect_interval (loop, ev_tstamp interval)
		626
		627	=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
		628
		629	These advanced functions influence the time that libev will spend waiting
		630	for events. Both are by default C<0>, meaning that libev will try to
		631	invoke timer/periodic callbacks and I/O callbacks with minimum latency.
		632
		633	Setting these to a higher value (the C<interval> I<must> be >= C<0>)
		634	allows libev to delay invocation of I/O and timer/periodic callbacks to
		635	increase efficiency of loop iterations.
		636
		637	The background is that sometimes your program runs just fast enough to
		638	handle one (or very few) event(s) per loop iteration. While this makes
		639	the program responsive, it also wastes a lot of CPU time to poll for new
		640	events, especially with backends like C<select ()> which have a high
		641	overhead for the actual polling but can deliver many events at once.
		642
		643	By setting a higher I<io collect interval> you allow libev to spend more
		644	time collecting I/O events, so you can handle more events per iteration,
		645	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
		646	C<ev_timer>) will be not affected. Setting this to a non-null value will
		647	introduce an additional C<ev_sleep ()> call into most loop iterations.
		648
		649	Likewise, by setting a higher I<timeout collect interval> you allow libev
		650	to spend more time collecting timeouts, at the expense of increased
		651	latency (the watcher callback will be called later). C<ev_io> watchers
		652	will not be affected. Setting this to a non-null value will not introduce
		653	any overhead in libev.
		654
		655	Many (busy) programs can usually benefit by setting the io collect
		656	interval to a value near C<0.1> or so, which is often enough for
		657	interactive servers (of course not for games), likewise for timeouts. It
		658	usually doesn't make much sense to set it to a lower value than C<0.01>,
		659	as this approsaches the timing granularity of most systems.
		660
574	=back	661	=back
575		662
576		663
577	=head1 ANATOMY OF A WATCHER	664	=head1 ANATOMY OF A WATCHER
578		665
…		…
903	In general you can register as many read and/or write event watchers per	990	In general you can register as many read and/or write event watchers per
904	fd as you want (as long as you don't confuse yourself). Setting all file	991	fd as you want (as long as you don't confuse yourself). Setting all file
905	descriptors to non-blocking mode is also usually a good idea (but not	992	descriptors to non-blocking mode is also usually a good idea (but not
906	required if you know what you are doing).	993	required if you know what you are doing).
907		994
908	You have to be careful with dup'ed file descriptors, though. Some backends
909	(the linux epoll backend is a notable example) cannot handle dup'ed file
910	descriptors correctly if you register interest in two or more fds pointing
911	to the same underlying file/socket/etc. description (that is, they share
912	the same underlying "file open").
913
914	If you must do this, then force the use of a known-to-be-good backend	995	If you must do this, then force the use of a known-to-be-good backend
915	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	996	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
916	C<EVBACKEND_POLL>).	997	C<EVBACKEND_POLL>).
917		998
918	Another thing you have to watch out for is that it is quite easy to	999	Another thing you have to watch out for is that it is quite easy to
…		…
949		1030
950	This is how one would do it normally anyway, the important point is that	1031	This is how one would do it normally anyway, the important point is that
951	the libev application should not optimise around libev but should leave	1032	the libev application should not optimise around libev but should leave
952	optimisations to libev.	1033	optimisations to libev.
953		1034
954	=head3 Ths special problem of dup'ed file descriptors	1035	=head3 The special problem of dup'ed file descriptors
955		1036
956	Some backends (e.g. epoll), cannot register events for file descriptors,	1037	Some backends (e.g. epoll), cannot register events for file descriptors,
957	but only events for the underlying file descriptions. That menas when you	1038	but only events for the underlying file descriptions. That means when you
958	have C<dup ()>'ed file descriptors and register events for them, only one	1039	have C<dup ()>'ed file descriptors or weirder constellations, and register
959	file descriptor might actually receive events.	1040	events for them, only one file descriptor might actually receive events.
960		1041
961	There is no workaorund possible except not registering events	1042	There is no workaround possible except not registering events
962	for potentially C<dup ()>'ed file descriptors or to resort to	1043	for potentially C<dup ()>'ed file descriptors, or to resort to
963	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.	1044	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
964		1045
965	=head3 The special problem of fork	1046	=head3 The special problem of fork
966		1047
967	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit	1048	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
…		…
993	=item int events [read-only]	1074	=item int events [read-only]
994		1075
995	The events being watched.	1076	The events being watched.
996		1077
997	=back	1078	=back
		1079
		1080	=head3 Examples
998		1081
999	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well	1082	Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1000	readable, but only once. Since it is likely line-buffered, you could	1083	readable, but only once. Since it is likely line-buffered, you could
1001	attempt to read a whole line in the callback.	1084	attempt to read a whole line in the callback.
1002		1085
…		…
1100	or C<ev_timer_again> is called and determines the next timeout (if any),	1183	or C<ev_timer_again> is called and determines the next timeout (if any),
1101	which is also when any modifications are taken into account.	1184	which is also when any modifications are taken into account.
1102		1185
1103	=back	1186	=back
1104		1187
		1188	=head3 Examples
		1189
1105	Example: Create a timer that fires after 60 seconds.	1190	Example: Create a timer that fires after 60 seconds.
1106		1191
1107	static void	1192	static void
1108	one_minute_cb (struct ev_loop loop, struct ev_timer w, int revents)	1193	one_minute_cb (struct ev_loop loop, struct ev_timer w, int revents)
1109	{	1194	{
…		…
1266	When active, contains the absolute time that the watcher is supposed to	1351	When active, contains the absolute time that the watcher is supposed to
1267	trigger next.	1352	trigger next.
1268		1353
1269	=back	1354	=back
1270		1355
		1356	=head3 Examples
		1357
1271	Example: Call a callback every hour, or, more precisely, whenever the	1358	Example: Call a callback every hour, or, more precisely, whenever the
1272	system clock is divisible by 3600. The callback invocation times have	1359	system clock is divisible by 3600. The callback invocation times have
1273	potentially a lot of jittering, but good long-term stability.	1360	potentially a lot of jittering, but good long-term stability.
1274		1361
1275	static void	1362	static void
…		…
1366		1453
1367	The process exit/trace status caused by C<rpid> (see your systems	1454	The process exit/trace status caused by C<rpid> (see your systems
1368	C<waitpid> and C<sys/wait.h> documentation for details).	1455	C<waitpid> and C<sys/wait.h> documentation for details).
1369		1456
1370	=back	1457	=back
		1458
		1459	=head3 Examples
1371		1460
1372	Example: Try to exit cleanly on SIGINT and SIGTERM.	1461	Example: Try to exit cleanly on SIGINT and SIGTERM.
1373		1462
1374	static void	1463	static void
1375	sigint_cb (struct ev_loop loop, struct ev_signal w, int revents)	1464	sigint_cb (struct ev_loop loop, struct ev_signal w, int revents)
…		…
1416	semantics of C<ev_stat> watchers, which means that libev sometimes needs	1505	semantics of C<ev_stat> watchers, which means that libev sometimes needs
1417	to fall back to regular polling again even with inotify, but changes are	1506	to fall back to regular polling again even with inotify, but changes are
1418	usually detected immediately, and if the file exists there will be no	1507	usually detected immediately, and if the file exists there will be no
1419	polling.	1508	polling.
1420		1509
		1510	=head3 Inotify
		1511
		1512	When C<inotify (7)> support has been compiled into libev (generally only
		1513	available on Linux) and present at runtime, it will be used to speed up
		1514	change detection where possible. The inotify descriptor will be created lazily
		1515	when the first C<ev_stat> watcher is being started.
		1516
		1517	Inotify presense does not change the semantics of C<ev_stat> watchers
		1518	except that changes might be detected earlier, and in some cases, to avoid
		1519	making regular C<stat> calls. Even in the presense of inotify support
		1520	there are many cases where libev has to resort to regular C<stat> polling.
		1521
		1522	(There is no support for kqueue, as apparently it cannot be used to
		1523	implement this functionality, due to the requirement of having a file
		1524	descriptor open on the object at all times).
		1525
		1526	=head3 The special problem of stat time resolution
		1527
		1528	The C<stat ()> syscall only supports full-second resolution portably, and
		1529	even on systems where the resolution is higher, many filesystems still
		1530	only support whole seconds.
		1531
		1532	That means that, if the time is the only thing that changes, you might
		1533	miss updates: on the first update, C<ev_stat> detects a change and calls
		1534	your callback, which does something. When there is another update within
		1535	the same second, C<ev_stat> will be unable to detect it.
		1536
		1537	The solution to this is to delay acting on a change for a second (or till
		1538	the next second boundary), using a roughly one-second delay C<ev_timer>
		1539	(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
		1540	is added to work around small timing inconsistencies of some operating
		1541	systems.
		1542
1421	=head3 Watcher-Specific Functions and Data Members	1543	=head3 Watcher-Specific Functions and Data Members
1422		1544
1423	=over 4	1545	=over 4
1424		1546
1425	=item ev_stat_init (ev_stat , callback, const char path, ev_tstamp interval)	1547	=item ev_stat_init (ev_stat , callback, const char path, ev_tstamp interval)
…		…
1462	=item const char *path [read-only]	1584	=item const char *path [read-only]
1463		1585
1464	The filesystem path that is being watched.	1586	The filesystem path that is being watched.
1465		1587
1466	=back	1588	=back
		1589
		1590	=head3 Examples
1467		1591
1468	Example: Watch C</etc/passwd> for attribute changes.	1592	Example: Watch C</etc/passwd> for attribute changes.
1469		1593
1470	static void	1594	static void
1471	passwd_cb (struct ev_loop loop, ev_stat w, int revents)	1595	passwd_cb (struct ev_loop loop, ev_stat w, int revents)
…		…
1484	}	1608	}
1485		1609
1486	...	1610	...
1487	ev_stat passwd;	1611	ev_stat passwd;
1488		1612
1489	ev_stat_init (&passwd, passwd_cb, "/etc/passwd");	1613	ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1490	ev_stat_start (loop, &passwd);	1614	ev_stat_start (loop, &passwd);
		1615
		1616	Example: Like above, but additionally use a one-second delay so we do not
		1617	miss updates (however, frequent updates will delay processing, too, so
		1618	one might do the work both on C<ev_stat> callback invocation I<and> on
		1619	C<ev_timer> callback invocation).
		1620
		1621	static ev_stat passwd;
		1622	static ev_timer timer;
		1623
		1624	static void
		1625	timer_cb (EV_P_ ev_timer *w, int revents)
		1626	{
		1627	ev_timer_stop (EV_A_ w);
		1628
		1629	/* now it's one second after the most recent passwd change */
		1630	}
		1631
		1632	static void
		1633	stat_cb (EV_P_ ev_stat *w, int revents)
		1634	{
		1635	/* reset the one-second timer */
		1636	ev_timer_again (EV_A_ &timer);
		1637	}
		1638
		1639	...
		1640	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
		1641	ev_stat_start (loop, &passwd);
		1642	ev_timer_init (&timer, timer_cb, 0., 1.01);
1491		1643
1492		1644
1493	=head2 C<ev_idle> - when you've got nothing better to do...	1645	=head2 C<ev_idle> - when you've got nothing better to do...
1494		1646
1495	Idle watchers trigger events when no other events of the same or higher	1647	Idle watchers trigger events when no other events of the same or higher
…		…
1520	Initialises and configures the idle watcher - it has no parameters of any	1672	Initialises and configures the idle watcher - it has no parameters of any
1521	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,	1673	kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1522	believe me.	1674	believe me.
1523		1675
1524	=back	1676	=back
		1677
		1678	=head3 Examples
1525		1679
1526	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the	1680	Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1527	callback, free it. Also, use no error checking, as usual.	1681	callback, free it. Also, use no error checking, as usual.
1528		1682
1529	static void	1683	static void
…		…
1581		1735
1582	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	1736	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1583	priority, to ensure that they are being run before any other watchers	1737	priority, to ensure that they are being run before any other watchers
1584	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	1738	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
1585	too) should not activate ("feed") events into libev. While libev fully	1739	too) should not activate ("feed") events into libev. While libev fully
1586	supports this, they will be called before other C<ev_check> watchers did	1740	supports this, they will be called before other C<ev_check> watchers
1587	their job. As C<ev_check> watchers are often used to embed other event	1741	did their job. As C<ev_check> watchers are often used to embed other
1588	loops those other event loops might be in an unusable state until their	1742	(non-libev) event loops those other event loops might be in an unusable
1589	C<ev_check> watcher ran (always remind yourself to coexist peacefully with	1743	state until their C<ev_check> watcher ran (always remind yourself to
1590	others).	1744	coexist peacefully with others).
1591		1745
1592	=head3 Watcher-Specific Functions and Data Members	1746	=head3 Watcher-Specific Functions and Data Members
1593		1747
1594	=over 4	1748	=over 4
1595		1749
…		…
1600	Initialises and configures the prepare or check watcher - they have no	1754	Initialises and configures the prepare or check watcher - they have no
1601	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>	1755	parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1602	macros, but using them is utterly, utterly and completely pointless.	1756	macros, but using them is utterly, utterly and completely pointless.
1603		1757
1604	=back	1758	=back
		1759
		1760	=head3 Examples
1605		1761
1606	There are a number of principal ways to embed other event loops or modules	1762	There are a number of principal ways to embed other event loops or modules
1607	into libev. Here are some ideas on how to include libadns into libev	1763	into libev. Here are some ideas on how to include libadns into libev
1608	(there is a Perl module named C<EV::ADNS> that does this, which you could	1764	(there is a Perl module named C<EV::ADNS> that does this, which you could
1609	use for an actually working example. Another Perl module named C<EV::Glib>	1765	use for an actually working example. Another Perl module named C<EV::Glib>
…		…
1734	=head2 C<ev_embed> - when one backend isn't enough...	1890	=head2 C<ev_embed> - when one backend isn't enough...
1735		1891
1736	This is a rather advanced watcher type that lets you embed one event loop	1892	This is a rather advanced watcher type that lets you embed one event loop
1737	into another (currently only C<ev_io> events are supported in the embedded	1893	into another (currently only C<ev_io> events are supported in the embedded
1738	loop, other types of watchers might be handled in a delayed or incorrect	1894	loop, other types of watchers might be handled in a delayed or incorrect
1739	fashion and must not be used). (See portability notes, below).	1895	fashion and must not be used).
1740		1896
1741	There are primarily two reasons you would want that: work around bugs and	1897	There are primarily two reasons you would want that: work around bugs and
1742	prioritise I/O.	1898	prioritise I/O.
1743		1899
1744	As an example for a bug workaround, the kqueue backend might only support	1900	As an example for a bug workaround, the kqueue backend might only support
…		…
1778	portable one.	1934	portable one.
1779		1935
1780	So when you want to use this feature you will always have to be prepared	1936	So when you want to use this feature you will always have to be prepared
1781	that you cannot get an embeddable loop. The recommended way to get around	1937	that you cannot get an embeddable loop. The recommended way to get around
1782	this is to have a separate variables for your embeddable loop, try to	1938	this is to have a separate variables for your embeddable loop, try to
1783	create it, and if that fails, use the normal loop for everything:	1939	create it, and if that fails, use the normal loop for everything.
		1940
		1941	=head3 Watcher-Specific Functions and Data Members
		1942
		1943	=over 4
		1944
		1945	=item ev_embed_init (ev_embed , callback, struct ev_loop embedded_loop)
		1946
		1947	=item ev_embed_set (ev_embed , callback, struct ev_loop embedded_loop)
		1948
		1949	Configures the watcher to embed the given loop, which must be
		1950	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
		1951	invoked automatically, otherwise it is the responsibility of the callback
		1952	to invoke it (it will continue to be called until the sweep has been done,
		1953	if you do not want thta, you need to temporarily stop the embed watcher).
		1954
		1955	=item ev_embed_sweep (loop, ev_embed *)
		1956
		1957	Make a single, non-blocking sweep over the embedded loop. This works
		1958	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
		1959	apropriate way for embedded loops.
		1960
		1961	=item struct ev_loop *other [read-only]
		1962
		1963	The embedded event loop.
		1964
		1965	=back
		1966
		1967	=head3 Examples
		1968
		1969	Example: Try to get an embeddable event loop and embed it into the default
		1970	event loop. If that is not possible, use the default loop. The default
		1971	loop is stored in C<loop_hi>, while the mebeddable loop is stored in
		1972	C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
		1973	used).
1784		1974
1785	struct ev_loop *loop_hi = ev_default_init (0);	1975	struct ev_loop *loop_hi = ev_default_init (0);
1786	struct ev_loop *loop_lo = 0;	1976	struct ev_loop *loop_lo = 0;
1787	struct ev_embed embed;	1977	struct ev_embed embed;
1788		1978
…		…
1799	ev_embed_start (loop_hi, &embed);	1989	ev_embed_start (loop_hi, &embed);
1800	}	1990	}
1801	else	1991	else
1802	loop_lo = loop_hi;	1992	loop_lo = loop_hi;
1803		1993
1804	=head2 Portability notes	1994	Example: Check if kqueue is available but not recommended and create
		1995	a kqueue backend for use with sockets (which usually work with any
		1996	kqueue implementation). Store the kqueue/socket-only event loop in
		1997	C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1805		1998
1806	Kqueue is nominally embeddable, but this is broken on all BSDs that I	1999	struct ev_loop *loop = ev_default_init (0);
1807	tried, in various ways. Usually the embedded event loop will simply never	2000	struct ev_loop *loop_socket = 0;
1808	receive events, sometimes it will only trigger a few times, sometimes in a	2001	struct ev_embed embed;
1809	loop. Epoll is also nominally embeddable, but many Linux kernel versions	2002
1810	will always eport the epoll fd as ready, even when no events are pending.	2003	if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
		2004	if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
		2005	{
		2006	ev_embed_init (&embed, 0, loop_socket);
		2007	ev_embed_start (loop, &embed);
		2008	}
1811		2009
1812	While libev allows embedding these backends (they are contained in	2010	if (!loop_socket)
1813	C<ev_embeddable_backends ()>), take extreme care that it will actually	2011	loop_socket = loop;
1814	work.
1815		2012
1816	When in doubt, create a dynamic event loop forced to use sockets (this	2013	// now use loop_socket for all sockets, and loop for everything else
1817	usually works) and possibly another thread and a pipe or so to report to
1818	your main event loop.
1819
1820	=head3 Watcher-Specific Functions and Data Members
1821
1822	=over 4
1823
1824	=item ev_embed_init (ev_embed , callback, struct ev_loop embedded_loop)
1825
1826	=item ev_embed_set (ev_embed , callback, struct ev_loop embedded_loop)
1827
1828	Configures the watcher to embed the given loop, which must be
1829	embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1830	invoked automatically, otherwise it is the responsibility of the callback
1831	to invoke it (it will continue to be called until the sweep has been done,
1832	if you do not want thta, you need to temporarily stop the embed watcher).
1833
1834	=item ev_embed_sweep (loop, ev_embed *)
1835
1836	Make a single, non-blocking sweep over the embedded loop. This works
1837	similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1838	apropriate way for embedded loops.
1839
1840	=item struct ev_loop *other [read-only]
1841
1842	The embedded event loop.
1843
1844	=back
1845		2014
1846		2015
1847	=head2 C<ev_fork> - the audacity to resume the event loop after a fork	2016	=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1848		2017
1849	Fork watchers are called when a C<fork ()> was detected (usually because	2018	Fork watchers are called when a C<fork ()> was detected (usually because
…		…
2297	runtime if successful). Otherwise no use of the realtime clock option will	2466	runtime if successful). Otherwise no use of the realtime clock option will
2298	be attempted. This effectively replaces C<gettimeofday> by C<clock_get	2467	be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2299	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the	2468	(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2300	note about libraries in the description of C<EV_USE_MONOTONIC>, though.	2469	note about libraries in the description of C<EV_USE_MONOTONIC>, though.
2301		2470
		2471	=item EV_USE_NANOSLEEP
		2472
		2473	If defined to be C<1>, libev will assume that C<nanosleep ()> is available
		2474	and will use it for delays. Otherwise it will use C<select ()>.
		2475
2302	=item EV_USE_SELECT	2476	=item EV_USE_SELECT
2303		2477
2304	If undefined or defined to be C<1>, libev will compile in support for the	2478	If undefined or defined to be C<1>, libev will compile in support for the
2305	C<select>(2) backend. No attempt at autodetection will be done: if no	2479	C<select>(2) backend. No attempt at autodetection will be done: if no
2306	other method takes over, select will be it. Otherwise the select backend	2480	other method takes over, select will be it. Otherwise the select backend
…		…
2323	wants osf handles on win32 (this is the case when the select to	2497	wants osf handles on win32 (this is the case when the select to
2324	be used is the winsock select). This means that it will call	2498	be used is the winsock select). This means that it will call
2325	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,	2499	C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2326	it is assumed that all these functions actually work on fds, even	2500	it is assumed that all these functions actually work on fds, even
2327	on win32. Should not be defined on non-win32 platforms.	2501	on win32. Should not be defined on non-win32 platforms.
		2502
		2503	=item EV_FD_TO_WIN32_HANDLE
		2504
		2505	If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
		2506	file descriptors to socket handles. When not defining this symbol (the
		2507	default), then libev will call C<_get_osfhandle>, which is usually
		2508	correct. In some cases, programs use their own file descriptor management,
		2509	in which case they can provide this function to map fds to socket handles.
2328		2510
2329	=item EV_USE_POLL	2511	=item EV_USE_POLL
2330		2512
2331	If defined to be C<1>, libev will compile in support for the C<poll>(2)	2513	If defined to be C<1>, libev will compile in support for the C<poll>(2)
2332	backend. Otherwise it will be enabled on non-win32 platforms. It	2514	backend. Otherwise it will be enabled on non-win32 platforms. It
…		…
2369	be detected at runtime.	2551	be detected at runtime.
2370		2552
2371	=item EV_H	2553	=item EV_H
2372		2554
2373	The name of the F<ev.h> header file used to include it. The default if	2555	The name of the F<ev.h> header file used to include it. The default if
2374	undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This	2556	undefined is C<"ev.h"> in F<event.h> and F<ev.c>. This can be used to
2375	can be used to virtually rename the F<ev.h> header file in case of conflicts.	2557	virtually rename the F<ev.h> header file in case of conflicts.
2376		2558
2377	=item EV_CONFIG_H	2559	=item EV_CONFIG_H
2378		2560
2379	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override	2561	If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2380	F<ev.c>'s idea of where to find the F<config.h> file, similarly to	2562	F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2381	C<EV_H>, above.	2563	C<EV_H>, above.
2382		2564
2383	=item EV_EVENT_H	2565	=item EV_EVENT_H
2384		2566
2385	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea	2567	Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2386	of how the F<event.h> header can be found.	2568	of how the F<event.h> header can be found, the dfeault is C<"event.h">.
2387		2569
2388	=item EV_PROTOTYPES	2570	=item EV_PROTOTYPES
2389		2571
2390	If defined to be C<0>, then F<ev.h> will not define any function	2572	If defined to be C<0>, then F<ev.h> will not define any function
2391	prototypes, but still define all the structs and other symbols. This is	2573	prototypes, but still define all the structs and other symbols. This is
…		…
2457	than enough. If you need to manage thousands of children you might want to	2639	than enough. If you need to manage thousands of children you might want to
2458	increase this value (I<must> be a power of two).	2640	increase this value (I<must> be a power of two).
2459		2641
2460	=item EV_INOTIFY_HASHSIZE	2642	=item EV_INOTIFY_HASHSIZE
2461		2643
2462	C<ev_staz> watchers use a small hash table to distribute workload by	2644	C<ev_stat> watchers use a small hash table to distribute workload by
2463	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	2645	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2464	usually more than enough. If you need to manage thousands of C<ev_stat>	2646	usually more than enough. If you need to manage thousands of C<ev_stat>
2465	watchers you might want to increase this value (I<must> be a power of	2647	watchers you might want to increase this value (I<must> be a power of
2466	two).	2648	two).
2467		2649
…		…
2563		2745
2564	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	2746	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2565		2747
2566	This means that, when you have a watcher that triggers in one hour and	2748	This means that, when you have a watcher that triggers in one hour and
2567	there are 100 watchers that would trigger before that then inserting will	2749	there are 100 watchers that would trigger before that then inserting will
2568	have to skip those 100 watchers.	2750	have to skip roughly seven (C<ld 100>) of these watchers.
2569		2751
2570	=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)	2752	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2571		2753
2572	That means that for changing a timer costs less than removing/adding them	2754	That means that changing a timer costs less than removing/adding them
2573	as only the relative motion in the event queue has to be paid for.	2755	as only the relative motion in the event queue has to be paid for.
2574		2756
2575	=item Starting io/check/prepare/idle/signal/child watchers: O(1)	2757	=item Starting io/check/prepare/idle/signal/child watchers: O(1)
2576		2758
2577	These just add the watcher into an array or at the head of a list.	2759	These just add the watcher into an array or at the head of a list.
		2760
2578	=item Stopping check/prepare/idle watchers: O(1)	2761	=item Stopping check/prepare/idle watchers: O(1)
2579		2762
2580	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	2763	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2581		2764
2582	These watchers are stored in lists then need to be walked to find the	2765	These watchers are stored in lists then need to be walked to find the
2583	correct watcher to remove. The lists are usually short (you don't usually	2766	correct watcher to remove. The lists are usually short (you don't usually
2584	have many watchers waiting for the same fd or signal).	2767	have many watchers waiting for the same fd or signal).
2585		2768
2586	=item Finding the next timer per loop iteration: O(1)	2769	=item Finding the next timer in each loop iteration: O(1)
		2770
		2771	By virtue of using a binary heap, the next timer is always found at the
		2772	beginning of the storage array.
2587		2773
2588	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	2774	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2589		2775
2590	A change means an I/O watcher gets started or stopped, which requires	2776	A change means an I/O watcher gets started or stopped, which requires
2591	libev to recalculate its status (and possibly tell the kernel).	2777	libev to recalculate its status (and possibly tell the kernel, depending
		2778	on backend and wether C<ev_io_set> was used).
2592		2779
2593	=item Activating one watcher: O(1)	2780	=item Activating one watcher (putting it into the pending state): O(1)
2594		2781
2595	=item Priority handling: O(number_of_priorities)	2782	=item Priority handling: O(number_of_priorities)
2596		2783
2597	Priorities are implemented by allocating some space for each	2784	Priorities are implemented by allocating some space for each
2598	priority. When doing priority-based operations, libev usually has to	2785	priority. When doing priority-based operations, libev usually has to
2599	linearly search all the priorities.	2786	linearly search all the priorities, but starting/stopping and activating
		2787	watchers becomes O(1) w.r.t. prioritiy handling.
2600		2788
2601	=back	2789	=back
2602		2790
2603		2791
		2792	=head1 Win32 platform limitations and workarounds
		2793
		2794	Win32 doesn't support any of the standards (e.g. POSIX) that libev
		2795	requires, and its I/O model is fundamentally incompatible with the POSIX
		2796	model. Libev still offers limited functionality on this platform in
		2797	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
		2798	descriptors. This only applies when using Win32 natively, not when using
		2799	e.g. cygwin.
		2800
		2801	There is no supported compilation method available on windows except
		2802	embedding it into other applications.
		2803
		2804	Due to the many, low, and arbitrary limits on the win32 platform and the
		2805	abysmal performance of winsockets, using a large number of sockets is not
		2806	recommended (and not reasonable). If your program needs to use more than
		2807	a hundred or so sockets, then likely it needs to use a totally different
		2808	implementation for windows, as libev offers the POSIX model, which cannot
		2809	be implemented efficiently on windows (microsoft monopoly games).
		2810
		2811	=over 4
		2812
		2813	=item The winsocket select function
		2814
		2815	The winsocket C<select> function doesn't follow POSIX in that it requires
		2816	socket I<handles> and not socket I<file descriptors>. This makes select
		2817	very inefficient, and also requires a mapping from file descriptors
		2818	to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
		2819	C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
		2820	symbols for more info.
		2821
		2822	The configuration for a "naked" win32 using the microsoft runtime
		2823	libraries and raw winsocket select is:
		2824
		2825	#define EV_USE_SELECT 1
		2826	#define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
		2827
		2828	Note that winsockets handling of fd sets is O(n), so you can easily get a
		2829	complexity in the O(n²) range when using win32.
		2830
		2831	=item Limited number of file descriptors
		2832
		2833	Windows has numerous arbitrary (and low) limits on things. Early versions
		2834	of winsocket's select only supported waiting for a max. of C<64> handles
		2835	(probably owning to the fact that all windows kernels can only wait for
		2836	C<64> things at the same time internally; microsoft recommends spawning a
		2837	chain of threads and wait for 63 handles and the previous thread in each).
		2838
		2839	Newer versions support more handles, but you need to define C<FD_SETSIZE>
		2840	to some high number (e.g. C<2048>) before compiling the winsocket select
		2841	call (which might be in libev or elsewhere, for example, perl does its own
		2842	select emulation on windows).
		2843
		2844	Another limit is the number of file descriptors in the microsoft runtime
		2845	libraries, which by default is C<64> (there must be a hidden I<64> fetish
		2846	or something like this inside microsoft). You can increase this by calling
		2847	C<_setmaxstdio>, which can increase this limit to C<2048> (another
		2848	arbitrary limit), but is broken in many versions of the microsoft runtime
		2849	libraries.
		2850
		2851	This might get you to about C<512> or C<2048> sockets (depending on
		2852	windows version and/or the phase of the moon). To get more, you need to
		2853	wrap all I/O functions and provide your own fd management, but the cost of
		2854	calling select (O(n²)) will likely make this unworkable.
		2855
		2856	=back
		2857
		2858
2604	=head1 AUTHOR	2859	=head1 AUTHOR
2605		2860
2606	Marc Lehmann <libev@schmorp.de>.	2861	Marc Lehmann <libev@schmorp.de>.
2607		2862

Diff Legend

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

Comparing libev/ev.pod (file contents): Revision 1.94 by root, Fri Dec 21 04:38:45 2007 UTC vs. Revision 1.115 by root, Mon Dec 31 01:32:59 2007 UTC

Diff Legend

Comparing libev/ev.pod (file contents):
Revision 1.94 by root, Fri Dec 21 04:38:45 2007 UTC vs.
Revision 1.115 by root, Mon Dec 31 01:32:59 2007 UTC