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

Comparing libev/ev.pod (file contents):
Revision 1.352 by root, Mon Jan 10 14:30:15 2011 UTC vs.
Revision 1.381 by root, Sat Aug 13 17:41:14 2011 UTC

…		…
58	ev_timer_start (loop, &timeout_watcher);	58	ev_timer_start (loop, &timeout_watcher);
59		59
60	// now wait for events to arrive	60	// now wait for events to arrive
61	ev_run (loop, 0);	61	ev_run (loop, 0);
62		62
63	// unloop was called, so exit	63	// break was called, so exit
64	return 0;	64	return 0;
65	}	65	}
66		66
67	=head1 ABOUT THIS DOCUMENT	67	=head1 ABOUT THIS DOCUMENT
68		68
…		…
178	you actually want to know. Also interesting is the combination of	178	you actually want to know. Also interesting is the combination of
179	C<ev_update_now> and C<ev_now>.	179	C<ev_update_now> and C<ev_now>.
180		180
181	=item ev_sleep (ev_tstamp interval)	181	=item ev_sleep (ev_tstamp interval)
182		182
183	Sleep for the given interval: The current thread will be blocked until	183	Sleep for the given interval: The current thread will be blocked
184	either it is interrupted or the given time interval has passed. Basically	184	until either it is interrupted or the given time interval has
		185	passed (approximately - it might return a bit earlier even if not
		186	interrupted). Returns immediately if C<< interval <= 0 >>.
		187
185	this is a sub-second-resolution C<sleep ()>.	188	Basically this is a sub-second-resolution C<sleep ()>.
		189
		190	The range of the C<interval> is limited - libev only guarantees to work
		191	with sleep times of up to one day (C<< interval <= 86400 >>).
186		192
187	=item int ev_version_major ()	193	=item int ev_version_major ()
188		194
189	=item int ev_version_minor ()	195	=item int ev_version_minor ()
190		196
…		…
435	example) that can't properly initialise their signal masks.	441	example) that can't properly initialise their signal masks.
436		442
437	=item C<EVFLAG_NOSIGMASK>	443	=item C<EVFLAG_NOSIGMASK>
438		444
439	When this flag is specified, then libev will avoid to modify the signal	445	When this flag is specified, then libev will avoid to modify the signal
440	mask. Specifically, this means you ahve to make sure signals are unblocked	446	mask. Specifically, this means you have to make sure signals are unblocked
441	when you want to receive them.	447	when you want to receive them.
442		448
443	This behaviour is useful when you want to do your own signal handling, or	449	This behaviour is useful when you want to do your own signal handling, or
444	want to handle signals only in specific threads and want to avoid libev	450	want to handle signals only in specific threads and want to avoid libev
445	unblocking the signals.	451	unblocking the signals.
		452
		453	It's also required by POSIX in a threaded program, as libev calls
		454	C<sigprocmask>, whose behaviour is officially unspecified.
446		455
447	This flag's behaviour will become the default in future versions of libev.	456	This flag's behaviour will become the default in future versions of libev.
448		457
449	=item C<EVBACKEND_SELECT> (value 1, portable select backend)	458	=item C<EVBACKEND_SELECT> (value 1, portable select backend)
450		459
…		…
480	=item C<EVBACKEND_EPOLL> (value 4, Linux)	489	=item C<EVBACKEND_EPOLL> (value 4, Linux)
481		490
482	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9	491	Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
483	kernels).	492	kernels).
484		493
485	For few fds, this backend is a bit little slower than poll and select,	494	For few fds, this backend is a bit little slower than poll and select, but
486	but it scales phenomenally better. While poll and select usually scale	495	it scales phenomenally better. While poll and select usually scale like
487	like O(total_fds) where n is the total number of fds (or the highest fd),	496	O(total_fds) where total_fds is the total number of fds (or the highest
488	epoll scales either O(1) or O(active_fds).	497	fd), epoll scales either O(1) or O(active_fds).
489		498
490	The epoll mechanism deserves honorable mention as the most misdesigned	499	The epoll mechanism deserves honorable mention as the most misdesigned
491	of the more advanced event mechanisms: mere annoyances include silently	500	of the more advanced event mechanisms: mere annoyances include silently
492	dropping file descriptors, requiring a system call per change per file	501	dropping file descriptors, requiring a system call per change per file
493	descriptor (and unnecessary guessing of parameters), problems with dup,	502	descriptor (and unnecessary guessing of parameters), problems with dup,
…		…
496	0.1ms) and so on. The biggest issue is fork races, however - if a program	505	0.1ms) and so on. The biggest issue is fork races, however - if a program
497	forks then I<both> parent and child process have to recreate the epoll	506	forks then I<both> parent and child process have to recreate the epoll
498	set, which can take considerable time (one syscall per file descriptor)	507	set, which can take considerable time (one syscall per file descriptor)
499	and is of course hard to detect.	508	and is of course hard to detect.
500		509
501	Epoll is also notoriously buggy - embedding epoll fds I<should> work, but	510	Epoll is also notoriously buggy - embedding epoll fds I<should> work,
502	of course I<doesn't>, and epoll just loves to report events for totally	511	but of course I<doesn't>, and epoll just loves to report events for
503	I<different> file descriptors (even already closed ones, so one cannot	512	totally I<different> file descriptors (even already closed ones, so
504	even remove them from the set) than registered in the set (especially	513	one cannot even remove them from the set) than registered in the set
505	on SMP systems). Libev tries to counter these spurious notifications by	514	(especially on SMP systems). Libev tries to counter these spurious
506	employing an additional generation counter and comparing that against the	515	notifications by employing an additional generation counter and comparing
507	events to filter out spurious ones, recreating the set when required. Last	516	that against the events to filter out spurious ones, recreating the set
		517	when required. Epoll also erroneously rounds down timeouts, but gives you
		518	no way to know when and by how much, so sometimes you have to busy-wait
		519	because epoll returns immediately despite a nonzero timeout. And last
508	not least, it also refuses to work with some file descriptors which work	520	not least, it also refuses to work with some file descriptors which work
509	perfectly fine with C<select> (files, many character devices...).	521	perfectly fine with C<select> (files, many character devices...).
510		522
511	Epoll is truly the train wreck analog among event poll mechanisms.	523	Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
		524	cobbled together in a hurry, no thought to design or interaction with
		525	others. Oh, the pain, will it ever stop...
512		526
513	While stopping, setting and starting an I/O watcher in the same iteration	527	While stopping, setting and starting an I/O watcher in the same iteration
514	will result in some caching, there is still a system call per such	528	will result in some caching, there is still a system call per such
515	incident (because the same I<file descriptor> could point to a different	529	incident (because the same I<file descriptor> could point to a different
516	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed	530	I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
…		…
594	among the OS-specific backends (I vastly prefer correctness over speed	608	among the OS-specific backends (I vastly prefer correctness over speed
595	hacks).	609	hacks).
596		610
597	On the negative side, the interface is I<bizarre> - so bizarre that	611	On the negative side, the interface is I<bizarre> - so bizarre that
598	even sun itself gets it wrong in their code examples: The event polling	612	even sun itself gets it wrong in their code examples: The event polling
599	function sometimes returning events to the caller even though an error	613	function sometimes returns events to the caller even though an error
600	occured, but with no indication whether it has done so or not (yes, it's	614	occurred, but with no indication whether it has done so or not (yes, it's
601	even documented that way) - deadly for edge-triggered interfaces where	615	even documented that way) - deadly for edge-triggered interfaces where you
602	you absolutely have to know whether an event occured or not because you	616	absolutely have to know whether an event occurred or not because you have
603	have to re-arm the watcher.	617	to re-arm the watcher.
604		618
605	Fortunately libev seems to be able to work around these idiocies.	619	Fortunately libev seems to be able to work around these idiocies.
606		620
607	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as	621	This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
608	C<EVBACKEND_POLL>.	622	C<EVBACKEND_POLL>.
…		…
820	This is useful if you are waiting for some external event in conjunction	834	This is useful if you are waiting for some external event in conjunction
821	with something not expressible using other libev watchers (i.e. "roll your	835	with something not expressible using other libev watchers (i.e. "roll your
822	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is	836	own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
823	usually a better approach for this kind of thing.	837	usually a better approach for this kind of thing.
824		838
825	Here are the gory details of what C<ev_run> does:	839	Here are the gory details of what C<ev_run> does (this is for your
		840	understanding, not a guarantee that things will work exactly like this in
		841	future versions):
826		842
827	- Increment loop depth.	843	- Increment loop depth.
828	- Reset the ev_break status.	844	- Reset the ev_break status.
829	- Before the first iteration, call any pending watchers.	845	- Before the first iteration, call any pending watchers.
830	LOOP:	846	LOOP:
…		…
863	anymore.	879	anymore.
864		880
865	... queue jobs here, make sure they register event watchers as long	881	... queue jobs here, make sure they register event watchers as long
866	... as they still have work to do (even an idle watcher will do..)	882	... as they still have work to do (even an idle watcher will do..)
867	ev_run (my_loop, 0);	883	ev_run (my_loop, 0);
868	... jobs done or somebody called unloop. yeah!	884	... jobs done or somebody called break. yeah!
869		885
870	=item ev_break (loop, how)	886	=item ev_break (loop, how)
871		887
872	Can be used to make a call to C<ev_run> return early (but only after it	888	Can be used to make a call to C<ev_run> return early (but only after it
873	has processed all outstanding events). The C<how> argument must be either	889	has processed all outstanding events). The C<how> argument must be either
…		…
936	overhead for the actual polling but can deliver many events at once.	952	overhead for the actual polling but can deliver many events at once.
937		953
938	By setting a higher I<io collect interval> you allow libev to spend more	954	By setting a higher I<io collect interval> you allow libev to spend more
939	time collecting I/O events, so you can handle more events per iteration,	955	time collecting I/O events, so you can handle more events per iteration,
940	at the cost of increasing latency. Timeouts (both C<ev_periodic> and	956	at the cost of increasing latency. Timeouts (both C<ev_periodic> and
941	C<ev_timer>) will be not affected. Setting this to a non-null value will	957	C<ev_timer>) will not be affected. Setting this to a non-null value will
942	introduce an additional C<ev_sleep ()> call into most loop iterations. The	958	introduce an additional C<ev_sleep ()> call into most loop iterations. The
943	sleep time ensures that libev will not poll for I/O events more often then	959	sleep time ensures that libev will not poll for I/O events more often then
944	once per this interval, on average.	960	once per this interval, on average (as long as the host time resolution is
		961	good enough).
945		962
946	Likewise, by setting a higher I<timeout collect interval> you allow libev	963	Likewise, by setting a higher I<timeout collect interval> you allow libev
947	to spend more time collecting timeouts, at the expense of increased	964	to spend more time collecting timeouts, at the expense of increased
948	latency/jitter/inexactness (the watcher callback will be called	965	latency/jitter/inexactness (the watcher callback will be called
949	later). C<ev_io> watchers will not be affected. Setting this to a non-null	966	later). C<ev_io> watchers will not be affected. Setting this to a non-null
…		…
1355	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related	1372	See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1356	functions that do not need a watcher.	1373	functions that do not need a watcher.
1357		1374
1358	=back	1375	=back
1359		1376
1360	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER	1377	See also the L<ASSOCIATING CUSTOM DATA WITH A WATCHER> and L<BUILDING YOUR
1361		1378	OWN COMPOSITE WATCHERS> idioms.
1362	Each watcher has, by default, a member C<void *data> that you can change
1363	and read at any time: libev will completely ignore it. This can be used
1364	to associate arbitrary data with your watcher. If you need more data and
1365	don't want to allocate memory and store a pointer to it in that data
1366	member, you can also "subclass" the watcher type and provide your own
1367	data:
1368
1369	struct my_io
1370	{
1371	ev_io io;
1372	int otherfd;
1373	void *somedata;
1374	struct whatever *mostinteresting;
1375	};
1376
1377	...
1378	struct my_io w;
1379	ev_io_init (&w.io, my_cb, fd, EV_READ);
1380
1381	And since your callback will be called with a pointer to the watcher, you
1382	can cast it back to your own type:
1383
1384	static void my_cb (struct ev_loop loop, ev_io w_, int revents)
1385	{
1386	struct my_io w = (struct my_io )w_;
1387	...
1388	}
1389
1390	More interesting and less C-conformant ways of casting your callback type
1391	instead have been omitted.
1392
1393	Another common scenario is to use some data structure with multiple
1394	embedded watchers:
1395
1396	struct my_biggy
1397	{
1398	int some_data;
1399	ev_timer t1;
1400	ev_timer t2;
1401	}
1402
1403	In this case getting the pointer to C<my_biggy> is a bit more
1404	complicated: Either you store the address of your C<my_biggy> struct
1405	in the C<data> member of the watcher (for woozies), or you need to use
1406	some pointer arithmetic using C<offsetof> inside your watchers (for real
1407	programmers):
1408
1409	#include <stddef.h>
1410
1411	static void
1412	t1_cb (EV_P_ ev_timer *w, int revents)
1413	{
1414	struct my_biggy big = (struct my_biggy *)
1415	(((char *)w) - offsetof (struct my_biggy, t1));
1416	}
1417
1418	static void
1419	t2_cb (EV_P_ ev_timer *w, int revents)
1420	{
1421	struct my_biggy big = (struct my_biggy *)
1422	(((char *)w) - offsetof (struct my_biggy, t2));
1423	}
1424		1379
1425	=head2 WATCHER STATES	1380	=head2 WATCHER STATES
1426		1381
1427	There are various watcher states mentioned throughout this manual -	1382	There are various watcher states mentioned throughout this manual -
1428	active, pending and so on. In this section these states and the rules to	1383	active, pending and so on. In this section these states and the rules to
…		…
1431		1386
1432	=over 4	1387	=over 4
1433		1388
1434	=item initialiased	1389	=item initialiased
1435		1390
1436	Before a watcher can be registered with the event looop it has to be	1391	Before a watcher can be registered with the event loop it has to be
1437	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to	1392	initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
1438	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.	1393	C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
1439		1394
1440	In this state it is simply some block of memory that is suitable for use	1395	In this state it is simply some block of memory that is suitable for
1441	in an event loop. It can be moved around, freed, reused etc. at will.	1396	use in an event loop. It can be moved around, freed, reused etc. at
		1397	will - as long as you either keep the memory contents intact, or call
		1398	C<ev_TYPE_init> again.
1442		1399
1443	=item started/running/active	1400	=item started/running/active
1444		1401
1445	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes	1402	Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
1446	property of the event loop, and is actively waiting for events. While in	1403	property of the event loop, and is actively waiting for events. While in
…		…
1474	latter will clear any pending state the watcher might be in, regardless	1431	latter will clear any pending state the watcher might be in, regardless
1475	of whether it was active or not, so stopping a watcher explicitly before	1432	of whether it was active or not, so stopping a watcher explicitly before
1476	freeing it is often a good idea.	1433	freeing it is often a good idea.
1477		1434
1478	While stopped (and not pending) the watcher is essentially in the	1435	While stopped (and not pending) the watcher is essentially in the
1479	initialised state, that is it can be reused, moved, modified in any way	1436	initialised state, that is, it can be reused, moved, modified in any way
1480	you wish.	1437	you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
		1438	it again).
1481		1439
1482	=back	1440	=back
1483		1441
1484	=head2 WATCHER PRIORITY MODELS	1442	=head2 WATCHER PRIORITY MODELS
1485		1443
…		…
1614	In general you can register as many read and/or write event watchers per	1572	In general you can register as many read and/or write event watchers per
1615	fd as you want (as long as you don't confuse yourself). Setting all file	1573	fd as you want (as long as you don't confuse yourself). Setting all file
1616	descriptors to non-blocking mode is also usually a good idea (but not	1574	descriptors to non-blocking mode is also usually a good idea (but not
1617	required if you know what you are doing).	1575	required if you know what you are doing).
1618		1576
1619	If you cannot use non-blocking mode, then force the use of a
1620	known-to-be-good backend (at the time of this writing, this includes only
1621	C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
1622	descriptors for which non-blocking operation makes no sense (such as
1623	files) - libev doesn't guarantee any specific behaviour in that case.
1624
1625	Another thing you have to watch out for is that it is quite easy to	1577	Another thing you have to watch out for is that it is quite easy to
1626	receive "spurious" readiness notifications, that is your callback might	1578	receive "spurious" readiness notifications, that is, your callback might
1627	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1579	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1628	because there is no data. Not only are some backends known to create a	1580	because there is no data. It is very easy to get into this situation even
1629	lot of those (for example Solaris ports), it is very easy to get into	1581	with a relatively standard program structure. Thus it is best to always
1630	this situation even with a relatively standard program structure. Thus	1582	use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
1631	it is best to always use non-blocking I/O: An extra C<read>(2) returning
1632	C<EAGAIN> is far preferable to a program hanging until some data arrives.	1583	preferable to a program hanging until some data arrives.
1633		1584
1634	If you cannot run the fd in non-blocking mode (for example you should	1585	If you cannot run the fd in non-blocking mode (for example you should
1635	not play around with an Xlib connection), then you have to separately	1586	not play around with an Xlib connection), then you have to separately
1636	re-test whether a file descriptor is really ready with a known-to-be good	1587	re-test whether a file descriptor is really ready with a known-to-be good
1637	interface such as poll (fortunately in our Xlib example, Xlib already	1588	interface such as poll (fortunately in the case of Xlib, it already does
1638	does this on its own, so its quite safe to use). Some people additionally	1589	this on its own, so its quite safe to use). Some people additionally
1639	use C<SIGALRM> and an interval timer, just to be sure you won't block	1590	use C<SIGALRM> and an interval timer, just to be sure you won't block
1640	indefinitely.	1591	indefinitely.
1641		1592
1642	But really, best use non-blocking mode.	1593	But really, best use non-blocking mode.
1643		1594
…		…
1671		1622
1672	There is no workaround possible except not registering events	1623	There is no workaround possible except not registering events
1673	for potentially C<dup ()>'ed file descriptors, or to resort to	1624	for potentially C<dup ()>'ed file descriptors, or to resort to
1674	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.	1625	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1675		1626
		1627	=head3 The special problem of files
		1628
		1629	Many people try to use C<select> (or libev) on file descriptors
		1630	representing files, and expect it to become ready when their program
		1631	doesn't block on disk accesses (which can take a long time on their own).
		1632
		1633	However, this cannot ever work in the "expected" way - you get a readiness
		1634	notification as soon as the kernel knows whether and how much data is
		1635	there, and in the case of open files, that's always the case, so you
		1636	always get a readiness notification instantly, and your read (or possibly
		1637	write) will still block on the disk I/O.
		1638
		1639	Another way to view it is that in the case of sockets, pipes, character
		1640	devices and so on, there is another party (the sender) that delivers data
		1641	on its own, but in the case of files, there is no such thing: the disk
		1642	will not send data on its own, simply because it doesn't know what you
		1643	wish to read - you would first have to request some data.
		1644
		1645	Since files are typically not-so-well supported by advanced notification
		1646	mechanism, libev tries hard to emulate POSIX behaviour with respect
		1647	to files, even though you should not use it. The reason for this is
		1648	convenience: sometimes you want to watch STDIN or STDOUT, which is
		1649	usually a tty, often a pipe, but also sometimes files or special devices
		1650	(for example, C<epoll> on Linux works with F</dev/random> but not with
		1651	F</dev/urandom>), and even though the file might better be served with
		1652	asynchronous I/O instead of with non-blocking I/O, it is still useful when
		1653	it "just works" instead of freezing.
		1654
		1655	So avoid file descriptors pointing to files when you know it (e.g. use
		1656	libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
		1657	when you rarely read from a file instead of from a socket, and want to
		1658	reuse the same code path.
		1659
1676	=head3 The special problem of fork	1660	=head3 The special problem of fork
1677		1661
1678	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit	1662	Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1679	useless behaviour. Libev fully supports fork, but needs to be told about	1663	useless behaviour. Libev fully supports fork, but needs to be told about
1680	it in the child.	1664	it in the child if you want to continue to use it in the child.
1681		1665
1682	To support fork in your programs, you either have to call	1666	To support fork in your child processes, you have to call C<ev_loop_fork
1683	C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,	1667	()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
1684	enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or	1668	C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1685	C<EVBACKEND_POLL>.
1686		1669
1687	=head3 The special problem of SIGPIPE	1670	=head3 The special problem of SIGPIPE
1688		1671
1689	While not really specific to libev, it is easy to forget about C<SIGPIPE>:	1672	While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1690	when writing to a pipe whose other end has been closed, your program gets	1673	when writing to a pipe whose other end has been closed, your program gets
…		…
1788	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1771	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1789	monotonic clock option helps a lot here).	1772	monotonic clock option helps a lot here).
1790		1773
1791	The callback is guaranteed to be invoked only I<after> its timeout has	1774	The callback is guaranteed to be invoked only I<after> its timeout has
1792	passed (not I<at>, so on systems with very low-resolution clocks this	1775	passed (not I<at>, so on systems with very low-resolution clocks this
1793	might introduce a small delay). If multiple timers become ready during the	1776	might introduce a small delay, see "the special problem of being too
		1777	early", below). If multiple timers become ready during the same loop
1794	same loop iteration then the ones with earlier time-out values are invoked	1778	iteration then the ones with earlier time-out values are invoked before
1795	before ones of the same priority with later time-out values (but this is	1779	ones of the same priority with later time-out values (but this is no
1796	no longer true when a callback calls C<ev_run> recursively).	1780	longer true when a callback calls C<ev_run> recursively).
1797		1781
1798	=head3 Be smart about timeouts	1782	=head3 Be smart about timeouts
1799		1783
1800	Many real-world problems involve some kind of timeout, usually for error	1784	Many real-world problems involve some kind of timeout, usually for error
1801	recovery. A typical example is an HTTP request - if the other side hangs,	1785	recovery. A typical example is an HTTP request - if the other side hangs,
…		…
1968	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is	1952	Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
1969	rather complicated, but extremely efficient, something that really pays	1953	rather complicated, but extremely efficient, something that really pays
1970	off after the first million or so of active timers, i.e. it's usually	1954	off after the first million or so of active timers, i.e. it's usually
1971	overkill :)	1955	overkill :)
1972		1956
		1957	=head3 The special problem of being too early
		1958
		1959	If you ask a timer to call your callback after three seconds, then
		1960	you expect it to be invoked after three seconds - but of course, this
		1961	cannot be guaranteed to infinite precision. Less obviously, it cannot be
		1962	guaranteed to any precision by libev - imagine somebody suspending the
		1963	process a STOP signal for a few hours for example.
		1964
		1965	So, libev tries to invoke your callback as soon as possible I<after> the
		1966	delay has occured, but cannot guarantee this.
		1967
		1968	A less obvious failure mode is calling your callback too early: many event
		1969	loops compare timestamps with a "elapsed delay >= requested delay", but
		1970	this can cause your callback to be invoked much earlier than you would
		1971	expect.
		1972
		1973	To see why, imagine a system with a clock that only offers full second
		1974	resolution (think windows if you can't come up with a broken enough OS
		1975	yourself). If you schedule a one-second timer at the time 500.9, then the
		1976	event loop will schedule your timeout to elapse at a system time of 500
		1977	(500.9 truncated to the resolution) + 1, or 501.
		1978
		1979	If an event library looks at the timeout 0.1s later, it will see "501 >=
		1980	501" and invoke the callback 0.1s after it was started, even though a
		1981	one-second delay was requested - this is being "too early", despite best
		1982	intentions.
		1983
		1984	This is the reason why libev will never invoke the callback if the elapsed
		1985	delay equals the requested delay, but only when the elapsed delay is
		1986	larger than the requested delay. In the example above, libev would only invoke
		1987	the callback at system time 502, or 1.1s after the timer was started.
		1988
		1989	So, while libev cannot guarantee that your callback will be invoked
		1990	exactly when requested, it I<can> and I<does> guarantee that the requested
		1991	delay has actually elapsed, or in other words, it always errs on the "too
		1992	late" side of things.
		1993
1973	=head3 The special problem of time updates	1994	=head3 The special problem of time updates
1974		1995
1975	Establishing the current time is a costly operation (it usually takes at	1996	Establishing the current time is a costly operation (it usually takes at
1976	least two system calls): EV therefore updates its idea of the current	1997	least two system calls): EV therefore updates its idea of the current
1977	time only before and after C<ev_run> collects new events, which causes a	1998	time only before and after C<ev_run> collects new events, which causes a
…		…
1987	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	2008	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1988		2009
1989	If the event loop is suspended for a long time, you can also force an	2010	If the event loop is suspended for a long time, you can also force an
1990	update of the time returned by C<ev_now ()> by calling C<ev_now_update	2011	update of the time returned by C<ev_now ()> by calling C<ev_now_update
1991	()>.	2012	()>.
		2013
		2014	=head3 The special problem of unsychronised clocks
		2015
		2016	Modern systems have a variety of clocks - libev itself uses the normal
		2017	"wall clock" clock and, if available, the monotonic clock (to avoid time
		2018	jumps).
		2019
		2020	Neither of these clocks is synchronised with each other or any other clock
		2021	on the system, so C<ev_time ()> might return a considerably different time
		2022	than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
		2023	a call to C<gettimeofday> might return a second count that is one higher
		2024	than a directly following call to C<time>.
		2025
		2026	The moral of this is to only compare libev-related timestamps with
		2027	C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
		2028	a seocnd or so.
		2029
		2030	One more problem arises due to this lack of synchronisation: if libev uses
		2031	the system monotonic clock and you compare timestamps from C<ev_time>
		2032	or C<ev_now> from when you started your timer and when your callback is
		2033	invoked, you will find that sometimes the callback is a bit "early".
		2034
		2035	This is because C<ev_timer>s work in real time, not wall clock time, so
		2036	libev makes sure your callback is not invoked before the delay happened,
		2037	I<measured according to the real time>, not the system clock.
		2038
		2039	If your timeouts are based on a physical timescale (e.g. "time out this
		2040	connection after 100 seconds") then this shouldn't bother you as it is
		2041	exactly the right behaviour.
		2042
		2043	If you want to compare wall clock/system timestamps to your timers, then
		2044	you need to use C<ev_periodic>s, as these are based on the wall clock
		2045	time, where your comparisons will always generate correct results.
1992		2046
1993	=head3 The special problems of suspended animation	2047	=head3 The special problems of suspended animation
1994		2048
1995	When you leave the server world it is quite customary to hit machines that	2049	When you leave the server world it is quite customary to hit machines that
1996	can suspend/hibernate - what happens to the clocks during such a suspend?	2050	can suspend/hibernate - what happens to the clocks during such a suspend?
…		…
2040	keep up with the timer (because it takes longer than those 10 seconds to	2094	keep up with the timer (because it takes longer than those 10 seconds to
2041	do stuff) the timer will not fire more than once per event loop iteration.	2095	do stuff) the timer will not fire more than once per event loop iteration.
2042		2096
2043	=item ev_timer_again (loop, ev_timer *)	2097	=item ev_timer_again (loop, ev_timer *)
2044		2098
2045	This will act as if the timer timed out and restart it again if it is	2099	This will act as if the timer timed out and restarts it again if it is
2046	repeating. The exact semantics are:	2100	repeating. The exact semantics are:
2047		2101
2048	If the timer is pending, its pending status is cleared.	2102	If the timer is pending, its pending status is cleared.
2049		2103
2050	If the timer is started but non-repeating, stop it (as if it timed out).	2104	If the timer is started but non-repeating, stop it (as if it timed out).
…		…
2180		2234
2181	Another way to think about it (for the mathematically inclined) is that	2235	Another way to think about it (for the mathematically inclined) is that
2182	C<ev_periodic> will try to run the callback in this mode at the next possible	2236	C<ev_periodic> will try to run the callback in this mode at the next possible
2183	time where C<time = offset (mod interval)>, regardless of any time jumps.	2237	time where C<time = offset (mod interval)>, regardless of any time jumps.
2184		2238
2185	For numerical stability it is preferable that the C<offset> value is near	2239	The C<interval> I<MUST> be positive, and for numerical stability, the
2186	C<ev_now ()> (the current time), but there is no range requirement for	2240	interval value should be higher than C<1/8192> (which is around 100
2187	this value, and in fact is often specified as zero.	2241	microseconds) and C<offset> should be higher than C<0> and should have
		2242	at most a similar magnitude as the current time (say, within a factor of
		2243	ten). Typical values for offset are, in fact, C<0> or something between
		2244	C<0> and C<interval>, which is also the recommended range.
2188		2245
2189	Note also that there is an upper limit to how often a timer can fire (CPU	2246	Note also that there is an upper limit to how often a timer can fire (CPU
2190	speed for example), so if C<interval> is very small then timing stability	2247	speed for example), so if C<interval> is very small then timing stability
2191	will of course deteriorate. Libev itself tries to be exact to be about one	2248	will of course deteriorate. Libev itself tries to be exact to be about one
2192	millisecond (if the OS supports it and the machine is fast enough).	2249	millisecond (if the OS supports it and the machine is fast enough).
…		…
2335	=head3 The special problem of inheritance over fork/execve/pthread_create	2392	=head3 The special problem of inheritance over fork/execve/pthread_create
2336		2393
2337	Both the signal mask (C<sigprocmask>) and the signal disposition	2394	Both the signal mask (C<sigprocmask>) and the signal disposition
2338	(C<sigaction>) are unspecified after starting a signal watcher (and after	2395	(C<sigaction>) are unspecified after starting a signal watcher (and after
2339	stopping it again), that is, libev might or might not block the signal,	2396	stopping it again), that is, libev might or might not block the signal,
2340	and might or might not set or restore the installed signal handler.	2397	and might or might not set or restore the installed signal handler (but
		2398	see C<EVFLAG_NOSIGMASK>).
2341		2399
2342	While this does not matter for the signal disposition (libev never	2400	While this does not matter for the signal disposition (libev never
2343	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on	2401	sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
2344	C<execve>), this matters for the signal mask: many programs do not expect	2402	C<execve>), this matters for the signal mask: many programs do not expect
2345	certain signals to be blocked.	2403	certain signals to be blocked.
…		…
3216	atexit (program_exits);	3274	atexit (program_exits);
3217		3275
3218		3276
3219	=head2 C<ev_async> - how to wake up an event loop	3277	=head2 C<ev_async> - how to wake up an event loop
3220		3278
3221	In general, you cannot use an C<ev_run> from multiple threads or other	3279	In general, you cannot use an C<ev_loop> from multiple threads or other
3222	asynchronous sources such as signal handlers (as opposed to multiple event	3280	asynchronous sources such as signal handlers (as opposed to multiple event
3223	loops - those are of course safe to use in different threads).	3281	loops - those are of course safe to use in different threads).
3224		3282
3225	Sometimes, however, you need to wake up an event loop you do not control,	3283	Sometimes, however, you need to wake up an event loop you do not control,
3226	for example because it belongs to another thread. This is what C<ev_async>	3284	for example because it belongs to another thread. This is what C<ev_async>
…		…
3233	C<ev_async_sent> calls). In fact, you could use signal watchers as a kind	3291	C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
3234	of "global async watchers" by using a watcher on an otherwise unused	3292	of "global async watchers" by using a watcher on an otherwise unused
3235	signal, and C<ev_feed_signal> to signal this watcher from another thread,	3293	signal, and C<ev_feed_signal> to signal this watcher from another thread,
3236	even without knowing which loop owns the signal.	3294	even without knowing which loop owns the signal.
3237		3295
3238	Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
3239	just the default loop.
3240
3241	=head3 Queueing	3296	=head3 Queueing
3242		3297
3243	C<ev_async> does not support queueing of data in any way. The reason	3298	C<ev_async> does not support queueing of data in any way. The reason
3244	is that the author does not know of a simple (or any) algorithm for a	3299	is that the author does not know of a simple (or any) algorithm for a
3245	multiple-writer-single-reader queue that works in all cases and doesn't	3300	multiple-writer-single-reader queue that works in all cases and doesn't
…		…
3336	trust me.	3391	trust me.
3337		3392
3338	=item ev_async_send (loop, ev_async *)	3393	=item ev_async_send (loop, ev_async *)
3339		3394
3340	Sends/signals/activates the given C<ev_async> watcher, that is, feeds	3395	Sends/signals/activates the given C<ev_async> watcher, that is, feeds
3341	an C<EV_ASYNC> event on the watcher into the event loop. Unlike	3396	an C<EV_ASYNC> event on the watcher into the event loop, and instantly
		3397	returns.
		3398
3342	C<ev_feed_event>, this call is safe to do from other threads, signal or	3399	Unlike C<ev_feed_event>, this call is safe to do from other threads,
3343	similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding	3400	signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
3344	section below on what exactly this means).	3401	embedding section below on what exactly this means).
3345		3402
3346	Note that, as with other watchers in libev, multiple events might get	3403	Note that, as with other watchers in libev, multiple events might get
3347	compressed into a single callback invocation (another way to look at this	3404	compressed into a single callback invocation (another way to look at
3348	is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>,	3405	this is that C<ev_async> watchers are level-triggered: they are set on
3349	reset when the event loop detects that).	3406	C<ev_async_send>, reset when the event loop detects that).
3350		3407
3351	This call incurs the overhead of a system call only once per event loop	3408	This call incurs the overhead of at most one extra system call per event
3352	iteration, so while the overhead might be noticeable, it doesn't apply to	3409	loop iteration, if the event loop is blocked, and no syscall at all if
3353	repeated calls to C<ev_async_send> for the same event loop.	3410	the event loop (or your program) is processing events. That means that
		3411	repeated calls are basically free (there is no need to avoid calls for
		3412	performance reasons) and that the overhead becomes smaller (typically
		3413	zero) under load.
3354		3414
3355	=item bool = ev_async_pending (ev_async *)	3415	=item bool = ev_async_pending (ev_async *)
3356		3416
3357	Returns a non-zero value when C<ev_async_send> has been called on the	3417	Returns a non-zero value when C<ev_async_send> has been called on the
3358	watcher but the event has not yet been processed (or even noted) by the	3418	watcher but the event has not yet been processed (or even noted) by the
…		…
3429		3489
3430	This section explains some common idioms that are not immediately	3490	This section explains some common idioms that are not immediately
3431	obvious. Note that examples are sprinkled over the whole manual, and this	3491	obvious. Note that examples are sprinkled over the whole manual, and this
3432	section only contains stuff that wouldn't fit anywhere else.	3492	section only contains stuff that wouldn't fit anywhere else.
3433		3493
3434	=over 4	3494	=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
3435		3495
3436	=item Model/nested event loop invocations and exit conditions.	3496	Each watcher has, by default, a C<void *data> member that you can read
		3497	or modify at any time: libev will completely ignore it. This can be used
		3498	to associate arbitrary data with your watcher. If you need more data and
		3499	don't want to allocate memory separately and store a pointer to it in that
		3500	data member, you can also "subclass" the watcher type and provide your own
		3501	data:
		3502
		3503	struct my_io
		3504	{
		3505	ev_io io;
		3506	int otherfd;
		3507	void *somedata;
		3508	struct whatever *mostinteresting;
		3509	};
		3510
		3511	...
		3512	struct my_io w;
		3513	ev_io_init (&w.io, my_cb, fd, EV_READ);
		3514
		3515	And since your callback will be called with a pointer to the watcher, you
		3516	can cast it back to your own type:
		3517
		3518	static void my_cb (struct ev_loop loop, ev_io w_, int revents)
		3519	{
		3520	struct my_io w = (struct my_io )w_;
		3521	...
		3522	}
		3523
		3524	More interesting and less C-conformant ways of casting your callback
		3525	function type instead have been omitted.
		3526
		3527	=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
		3528
		3529	Another common scenario is to use some data structure with multiple
		3530	embedded watchers, in effect creating your own watcher that combines
		3531	multiple libev event sources into one "super-watcher":
		3532
		3533	struct my_biggy
		3534	{
		3535	int some_data;
		3536	ev_timer t1;
		3537	ev_timer t2;
		3538	}
		3539
		3540	In this case getting the pointer to C<my_biggy> is a bit more
		3541	complicated: Either you store the address of your C<my_biggy> struct in
		3542	the C<data> member of the watcher (for woozies or C++ coders), or you need
		3543	to use some pointer arithmetic using C<offsetof> inside your watchers (for
		3544	real programmers):
		3545
		3546	#include <stddef.h>
		3547
		3548	static void
		3549	t1_cb (EV_P_ ev_timer *w, int revents)
		3550	{
		3551	struct my_biggy big = (struct my_biggy *)
		3552	(((char *)w) - offsetof (struct my_biggy, t1));
		3553	}
		3554
		3555	static void
		3556	t2_cb (EV_P_ ev_timer *w, int revents)
		3557	{
		3558	struct my_biggy big = (struct my_biggy *)
		3559	(((char *)w) - offsetof (struct my_biggy, t2));
		3560	}
		3561
		3562	=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
3437		3563
3438	Often (especially in GUI toolkits) there are places where you have	3564	Often (especially in GUI toolkits) there are places where you have
3439	I<modal> interaction, which is most easily implemented by recursively	3565	I<modal> interaction, which is most easily implemented by recursively
3440	invoking C<ev_run>.	3566	invoking C<ev_run>.
3441		3567
…		…
3470	exit_main_loop = 1;	3596	exit_main_loop = 1;
3471		3597
3472	// exit both	3598	// exit both
3473	exit_main_loop = exit_nested_loop = 1;	3599	exit_main_loop = exit_nested_loop = 1;
3474		3600
3475	=back	3601	=head2 THREAD LOCKING EXAMPLE
		3602
		3603	Here is a fictitious example of how to run an event loop in a different
		3604	thread from where callbacks are being invoked and watchers are
		3605	created/added/removed.
		3606
		3607	For a real-world example, see the C<EV::Loop::Async> perl module,
		3608	which uses exactly this technique (which is suited for many high-level
		3609	languages).
		3610
		3611	The example uses a pthread mutex to protect the loop data, a condition
		3612	variable to wait for callback invocations, an async watcher to notify the
		3613	event loop thread and an unspecified mechanism to wake up the main thread.
		3614
		3615	First, you need to associate some data with the event loop:
		3616
		3617	typedef struct {
		3618	mutex_t lock; /* global loop lock */
		3619	ev_async async_w;
		3620	thread_t tid;
		3621	cond_t invoke_cv;
		3622	} userdata;
		3623
		3624	void prepare_loop (EV_P)
		3625	{
		3626	// for simplicity, we use a static userdata struct.
		3627	static userdata u;
		3628
		3629	ev_async_init (&u->async_w, async_cb);
		3630	ev_async_start (EV_A_ &u->async_w);
		3631
		3632	pthread_mutex_init (&u->lock, 0);
		3633	pthread_cond_init (&u->invoke_cv, 0);
		3634
		3635	// now associate this with the loop
		3636	ev_set_userdata (EV_A_ u);
		3637	ev_set_invoke_pending_cb (EV_A_ l_invoke);
		3638	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
		3639
		3640	// then create the thread running ev_run
		3641	pthread_create (&u->tid, 0, l_run, EV_A);
		3642	}
		3643
		3644	The callback for the C<ev_async> watcher does nothing: the watcher is used
		3645	solely to wake up the event loop so it takes notice of any new watchers
		3646	that might have been added:
		3647
		3648	static void
		3649	async_cb (EV_P_ ev_async *w, int revents)
		3650	{
		3651	// just used for the side effects
		3652	}
		3653
		3654	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
		3655	protecting the loop data, respectively.
		3656
		3657	static void
		3658	l_release (EV_P)
		3659	{
		3660	userdata *u = ev_userdata (EV_A);
		3661	pthread_mutex_unlock (&u->lock);
		3662	}
		3663
		3664	static void
		3665	l_acquire (EV_P)
		3666	{
		3667	userdata *u = ev_userdata (EV_A);
		3668	pthread_mutex_lock (&u->lock);
		3669	}
		3670
		3671	The event loop thread first acquires the mutex, and then jumps straight
		3672	into C<ev_run>:
		3673
		3674	void *
		3675	l_run (void *thr_arg)
		3676	{
		3677	struct ev_loop loop = (struct ev_loop )thr_arg;
		3678
		3679	l_acquire (EV_A);
		3680	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
		3681	ev_run (EV_A_ 0);
		3682	l_release (EV_A);
		3683
		3684	return 0;
		3685	}
		3686
		3687	Instead of invoking all pending watchers, the C<l_invoke> callback will
		3688	signal the main thread via some unspecified mechanism (signals? pipe
		3689	writes? C<Async::Interrupt>?) and then waits until all pending watchers
		3690	have been called (in a while loop because a) spurious wakeups are possible
		3691	and b) skipping inter-thread-communication when there are no pending
		3692	watchers is very beneficial):
		3693
		3694	static void
		3695	l_invoke (EV_P)
		3696	{
		3697	userdata *u = ev_userdata (EV_A);
		3698
		3699	while (ev_pending_count (EV_A))
		3700	{
		3701	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
		3702	pthread_cond_wait (&u->invoke_cv, &u->lock);
		3703	}
		3704	}
		3705
		3706	Now, whenever the main thread gets told to invoke pending watchers, it
		3707	will grab the lock, call C<ev_invoke_pending> and then signal the loop
		3708	thread to continue:
		3709
		3710	static void
		3711	real_invoke_pending (EV_P)
		3712	{
		3713	userdata *u = ev_userdata (EV_A);
		3714
		3715	pthread_mutex_lock (&u->lock);
		3716	ev_invoke_pending (EV_A);
		3717	pthread_cond_signal (&u->invoke_cv);
		3718	pthread_mutex_unlock (&u->lock);
		3719	}
		3720
		3721	Whenever you want to start/stop a watcher or do other modifications to an
		3722	event loop, you will now have to lock:
		3723
		3724	ev_timer timeout_watcher;
		3725	userdata *u = ev_userdata (EV_A);
		3726
		3727	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
		3728
		3729	pthread_mutex_lock (&u->lock);
		3730	ev_timer_start (EV_A_ &timeout_watcher);
		3731	ev_async_send (EV_A_ &u->async_w);
		3732	pthread_mutex_unlock (&u->lock);
		3733
		3734	Note that sending the C<ev_async> watcher is required because otherwise
		3735	an event loop currently blocking in the kernel will have no knowledge
		3736	about the newly added timer. By waking up the loop it will pick up any new
		3737	watchers in the next event loop iteration.
		3738
		3739	=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
		3740
		3741	While the overhead of a callback that e.g. schedules a thread is small, it
		3742	is still an overhead. If you embed libev, and your main usage is with some
		3743	kind of threads or coroutines, you might want to customise libev so that
		3744	doesn't need callbacks anymore.
		3745
		3746	Imagine you have coroutines that you can switch to using a function
		3747	C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
		3748	and that due to some magic, the currently active coroutine is stored in a
		3749	global called C<current_coro>. Then you can build your own "wait for libev
		3750	event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
		3751	the differing C<;> conventions):
		3752
		3753	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3754	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
		3755
		3756	That means instead of having a C callback function, you store the
		3757	coroutine to switch to in each watcher, and instead of having libev call
		3758	your callback, you instead have it switch to that coroutine.
		3759
		3760	A coroutine might now wait for an event with a function called
		3761	C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
		3762	matter when, or whether the watcher is active or not when this function is
		3763	called):
		3764
		3765	void
		3766	wait_for_event (ev_watcher *w)
		3767	{
		3768	ev_cb_set (w) = current_coro;
		3769	switch_to (libev_coro);
		3770	}
		3771
		3772	That basically suspends the coroutine inside C<wait_for_event> and
		3773	continues the libev coroutine, which, when appropriate, switches back to
		3774	this or any other coroutine. I am sure if you sue this your own :)
		3775
		3776	You can do similar tricks if you have, say, threads with an event queue -
		3777	instead of storing a coroutine, you store the queue object and instead of
		3778	switching to a coroutine, you push the watcher onto the queue and notify
		3779	any waiters.
		3780
		3781	To embed libev, see L<EMBEDDING>, but in short, it's easiest to create two
		3782	files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
		3783
		3784	// my_ev.h
		3785	#define EV_CB_DECLARE(type) struct my_coro *cb;
		3786	#define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb);
		3787	#include "../libev/ev.h"
		3788
		3789	// my_ev.c
		3790	#define EV_H "my_ev.h"
		3791	#include "../libev/ev.c"
		3792
		3793	And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
		3794	F<my_ev.c> into your project. When properly specifying include paths, you
		3795	can even use F<ev.h> as header file name directly.
3476		3796
3477		3797
3478	=head1 LIBEVENT EMULATION	3798	=head1 LIBEVENT EMULATION
3479		3799
3480	Libev offers a compatibility emulation layer for libevent. It cannot	3800	Libev offers a compatibility emulation layer for libevent. It cannot
…		…
3695	watchers in the constructor.	4015	watchers in the constructor.
3696		4016
3697	class myclass	4017	class myclass
3698	{	4018	{
3699	ev::io io ; void io_cb (ev::io &w, int revents);	4019	ev::io io ; void io_cb (ev::io &w, int revents);
3700	ev::io2 io2 ; void io2_cb (ev::io &w, int revents);	4020	ev::io io2 ; void io2_cb (ev::io &w, int revents);
3701	ev::idle idle; void idle_cb (ev::idle &w, int revents);	4021	ev::idle idle; void idle_cb (ev::idle &w, int revents);
3702		4022
3703	myclass (int fd)	4023	myclass (int fd)
3704	{	4024	{
3705	io .set <myclass, &myclass::io_cb > (this);	4025	io .set <myclass, &myclass::io_cb > (this);
…		…
3756	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.	4076	L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
3757		4077
3758	=item D	4078	=item D
3759		4079
3760	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to	4080	Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
3761	be found at L<http://proj.llucax.com.ar/wiki/evd>.	4081	be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
3762		4082
3763	=item Ocaml	4083	=item Ocaml
3764		4084
3765	Erkki Seppala has written Ocaml bindings for libev, to be found at	4085	Erkki Seppala has written Ocaml bindings for libev, to be found at
3766	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.	4086	L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
…		…
3814	suitable for use with C<EV_A>.	4134	suitable for use with C<EV_A>.
3815		4135
3816	=item C<EV_DEFAULT>, C<EV_DEFAULT_>	4136	=item C<EV_DEFAULT>, C<EV_DEFAULT_>
3817		4137
3818	Similar to the other two macros, this gives you the value of the default	4138	Similar to the other two macros, this gives you the value of the default
3819	loop, if multiple loops are supported ("ev loop default").	4139	loop, if multiple loops are supported ("ev loop default"). The default loop
		4140	will be initialised if it isn't already initialised.
		4141
		4142	For non-multiplicity builds, these macros do nothing, so you always have
		4143	to initialise the loop somewhere.
3820		4144
3821	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>	4145	=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
3822		4146
3823	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the	4147	Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
3824	default loop has been initialised (C<UC> == unchecked). Their behaviour	4148	default loop has been initialised (C<UC> == unchecked). Their behaviour
…		…
3969	supported). It will also not define any of the structs usually found in	4293	supported). It will also not define any of the structs usually found in
3970	F<event.h> that are not directly supported by the libev core alone.	4294	F<event.h> that are not directly supported by the libev core alone.
3971		4295
3972	In standalone mode, libev will still try to automatically deduce the	4296	In standalone mode, libev will still try to automatically deduce the
3973	configuration, but has to be more conservative.	4297	configuration, but has to be more conservative.
		4298
		4299	=item EV_USE_FLOOR
		4300
		4301	If defined to be C<1>, libev will use the C<floor ()> function for its
		4302	periodic reschedule calculations, otherwise libev will fall back on a
		4303	portable (slower) implementation. If you enable this, you usually have to
		4304	link against libm or something equivalent. Enabling this when the C<floor>
		4305	function is not available will fail, so the safe default is to not enable
		4306	this.
3974		4307
3975	=item EV_USE_MONOTONIC	4308	=item EV_USE_MONOTONIC
3976		4309
3977	If defined to be C<1>, libev will try to detect the availability of the	4310	If defined to be C<1>, libev will try to detect the availability of the
3978	monotonic clock option at both compile time and runtime. Otherwise no	4311	monotonic clock option at both compile time and runtime. Otherwise no
…		…
4111	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.	4444	indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
4112		4445
4113	=item EV_ATOMIC_T	4446	=item EV_ATOMIC_T
4114		4447
4115	Libev requires an integer type (suitable for storing C<0> or C<1>) whose	4448	Libev requires an integer type (suitable for storing C<0> or C<1>) whose
4116	access is atomic with respect to other threads or signal contexts. No such	4449	access is atomic and serialised with respect to other threads or signal
4117	type is easily found in the C language, so you can provide your own type	4450	contexts. No such type is easily found in the C language, so you can
4118	that you know is safe for your purposes. It is used both for signal handler "locking"	4451	provide your own type that you know is safe for your purposes. It is used
4119	as well as for signal and thread safety in C<ev_async> watchers.	4452	both for signal handler "locking" as well as for signal and thread safety
		4453	in C<ev_async> watchers.
4120		4454
4121	In the absence of this define, libev will use C<sig_atomic_t volatile>	4455	In the absence of this define, libev will use C<sig_atomic_t volatile>
4122	(from F<signal.h>), which is usually good enough on most platforms.	4456	(from F<signal.h>), which is usually good enough on most platforms,
		4457	although strictly speaking using a type that also implies a memory fence
		4458	is required.
4123		4459
4124	=item EV_H (h)	4460	=item EV_H (h)
4125		4461
4126	The name of the F<ev.h> header file used to include it. The default if	4462	The name of the F<ev.h> header file used to include it. The default if
4127	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be	4463	undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
…		…
4150	If undefined or defined to C<1>, then all event-loop-specific functions	4486	If undefined or defined to C<1>, then all event-loop-specific functions
4151	will have the C<struct ev_loop *> as first argument, and you can create	4487	will have the C<struct ev_loop *> as first argument, and you can create
4152	additional independent event loops. Otherwise there will be no support	4488	additional independent event loops. Otherwise there will be no support
4153	for multiple event loops and there is no first event loop pointer	4489	for multiple event loops and there is no first event loop pointer
4154	argument. Instead, all functions act on the single default loop.	4490	argument. Instead, all functions act on the single default loop.
		4491
		4492	Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
		4493	default loop when multiplicity is switched off - you always have to
		4494	initialise the loop manually in this case.
4155		4495
4156	=item EV_MINPRI	4496	=item EV_MINPRI
4157		4497
4158	=item EV_MAXPRI	4498	=item EV_MAXPRI
4159		4499
…		…
4410	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	4750	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
4411		4751
4412	#include "ev_cpp.h"	4752	#include "ev_cpp.h"
4413	#include "ev.c"	4753	#include "ev.c"
4414		4754
4415	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES	4755	=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
4416		4756
4417	=head2 THREADS AND COROUTINES	4757	=head2 THREADS AND COROUTINES
4418		4758
4419	=head3 THREADS	4759	=head3 THREADS
4420		4760
…		…
4471	default loop and triggering an C<ev_async> watcher from the default loop	4811	default loop and triggering an C<ev_async> watcher from the default loop
4472	watcher callback into the event loop interested in the signal.	4812	watcher callback into the event loop interested in the signal.
4473		4813
4474	=back	4814	=back
4475		4815
4476	=head4 THREAD LOCKING EXAMPLE	4816	See also L<THREAD LOCKING EXAMPLE>.
4477
4478	Here is a fictitious example of how to run an event loop in a different
4479	thread than where callbacks are being invoked and watchers are
4480	created/added/removed.
4481
4482	For a real-world example, see the C<EV::Loop::Async> perl module,
4483	which uses exactly this technique (which is suited for many high-level
4484	languages).
4485
4486	The example uses a pthread mutex to protect the loop data, a condition
4487	variable to wait for callback invocations, an async watcher to notify the
4488	event loop thread and an unspecified mechanism to wake up the main thread.
4489
4490	First, you need to associate some data with the event loop:
4491
4492	typedef struct {
4493	mutex_t lock; /* global loop lock */
4494	ev_async async_w;
4495	thread_t tid;
4496	cond_t invoke_cv;
4497	} userdata;
4498
4499	void prepare_loop (EV_P)
4500	{
4501	// for simplicity, we use a static userdata struct.
4502	static userdata u;
4503
4504	ev_async_init (&u->async_w, async_cb);
4505	ev_async_start (EV_A_ &u->async_w);
4506
4507	pthread_mutex_init (&u->lock, 0);
4508	pthread_cond_init (&u->invoke_cv, 0);
4509
4510	// now associate this with the loop
4511	ev_set_userdata (EV_A_ u);
4512	ev_set_invoke_pending_cb (EV_A_ l_invoke);
4513	ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
4514
4515	// then create the thread running ev_loop
4516	pthread_create (&u->tid, 0, l_run, EV_A);
4517	}
4518
4519	The callback for the C<ev_async> watcher does nothing: the watcher is used
4520	solely to wake up the event loop so it takes notice of any new watchers
4521	that might have been added:
4522
4523	static void
4524	async_cb (EV_P_ ev_async *w, int revents)
4525	{
4526	// just used for the side effects
4527	}
4528
4529	The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
4530	protecting the loop data, respectively.
4531
4532	static void
4533	l_release (EV_P)
4534	{
4535	userdata *u = ev_userdata (EV_A);
4536	pthread_mutex_unlock (&u->lock);
4537	}
4538
4539	static void
4540	l_acquire (EV_P)
4541	{
4542	userdata *u = ev_userdata (EV_A);
4543	pthread_mutex_lock (&u->lock);
4544	}
4545
4546	The event loop thread first acquires the mutex, and then jumps straight
4547	into C<ev_run>:
4548
4549	void *
4550	l_run (void *thr_arg)
4551	{
4552	struct ev_loop loop = (struct ev_loop )thr_arg;
4553
4554	l_acquire (EV_A);
4555	pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
4556	ev_run (EV_A_ 0);
4557	l_release (EV_A);
4558
4559	return 0;
4560	}
4561
4562	Instead of invoking all pending watchers, the C<l_invoke> callback will
4563	signal the main thread via some unspecified mechanism (signals? pipe
4564	writes? C<Async::Interrupt>?) and then waits until all pending watchers
4565	have been called (in a while loop because a) spurious wakeups are possible
4566	and b) skipping inter-thread-communication when there are no pending
4567	watchers is very beneficial):
4568
4569	static void
4570	l_invoke (EV_P)
4571	{
4572	userdata *u = ev_userdata (EV_A);
4573
4574	while (ev_pending_count (EV_A))
4575	{
4576	wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
4577	pthread_cond_wait (&u->invoke_cv, &u->lock);
4578	}
4579	}
4580
4581	Now, whenever the main thread gets told to invoke pending watchers, it
4582	will grab the lock, call C<ev_invoke_pending> and then signal the loop
4583	thread to continue:
4584
4585	static void
4586	real_invoke_pending (EV_P)
4587	{
4588	userdata *u = ev_userdata (EV_A);
4589
4590	pthread_mutex_lock (&u->lock);
4591	ev_invoke_pending (EV_A);
4592	pthread_cond_signal (&u->invoke_cv);
4593	pthread_mutex_unlock (&u->lock);
4594	}
4595
4596	Whenever you want to start/stop a watcher or do other modifications to an
4597	event loop, you will now have to lock:
4598
4599	ev_timer timeout_watcher;
4600	userdata *u = ev_userdata (EV_A);
4601
4602	ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
4603
4604	pthread_mutex_lock (&u->lock);
4605	ev_timer_start (EV_A_ &timeout_watcher);
4606	ev_async_send (EV_A_ &u->async_w);
4607	pthread_mutex_unlock (&u->lock);
4608
4609	Note that sending the C<ev_async> watcher is required because otherwise
4610	an event loop currently blocking in the kernel will have no knowledge
4611	about the newly added timer. By waking up the loop it will pick up any new
4612	watchers in the next event loop iteration.
4613		4817
4614	=head3 COROUTINES	4818	=head3 COROUTINES
4615		4819
4616	Libev is very accommodating to coroutines ("cooperative threads"):	4820	Libev is very accommodating to coroutines ("cooperative threads"):
4617	libev fully supports nesting calls to its functions from different	4821	libev fully supports nesting calls to its functions from different
…		…
4782	requires, and its I/O model is fundamentally incompatible with the POSIX	4986	requires, and its I/O model is fundamentally incompatible with the POSIX
4783	model. Libev still offers limited functionality on this platform in	4987	model. Libev still offers limited functionality on this platform in
4784	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	4988	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
4785	descriptors. This only applies when using Win32 natively, not when using	4989	descriptors. This only applies when using Win32 natively, not when using
4786	e.g. cygwin. Actually, it only applies to the microsofts own compilers,	4990	e.g. cygwin. Actually, it only applies to the microsofts own compilers,
4787	as every compielr comes with a slightly differently broken/incompatible	4991	as every compiler comes with a slightly differently broken/incompatible
4788	environment.	4992	environment.
4789		4993
4790	Lifting these limitations would basically require the full	4994	Lifting these limitations would basically require the full
4791	re-implementation of the I/O system. If you are into this kind of thing,	4995	re-implementation of the I/O system. If you are into this kind of thing,
4792	then note that glib does exactly that for you in a very portable way (note	4996	then note that glib does exactly that for you in a very portable way (note
…		…
4925		5129
4926	The type C<double> is used to represent timestamps. It is required to	5130	The type C<double> is used to represent timestamps. It is required to
4927	have at least 51 bits of mantissa (and 9 bits of exponent), which is	5131	have at least 51 bits of mantissa (and 9 bits of exponent), which is
4928	good enough for at least into the year 4000 with millisecond accuracy	5132	good enough for at least into the year 4000 with millisecond accuracy
4929	(the design goal for libev). This requirement is overfulfilled by	5133	(the design goal for libev). This requirement is overfulfilled by
4930	implementations using IEEE 754, which is basically all existing ones. With	5134	implementations using IEEE 754, which is basically all existing ones.
		5135
4931	IEEE 754 doubles, you get microsecond accuracy until at least 2200.	5136	With IEEE 754 doubles, you get microsecond accuracy until at least the
		5137	year 2255 (and millisecond accuray till the year 287396 - by then, libev
		5138	is either obsolete or somebody patched it to use C<long double> or
		5139	something like that, just kidding).
4932		5140
4933	=back	5141	=back
4934		5142
4935	If you know of other additional requirements drop me a note.	5143	If you know of other additional requirements drop me a note.
4936		5144
…		…
4998	=item Processing ev_async_send: O(number_of_async_watchers)	5206	=item Processing ev_async_send: O(number_of_async_watchers)
4999		5207
5000	=item Processing signals: O(max_signal_number)	5208	=item Processing signals: O(max_signal_number)
5001		5209
5002	Sending involves a system call I<iff> there were no other C<ev_async_send>	5210	Sending involves a system call I<iff> there were no other C<ev_async_send>
5003	calls in the current loop iteration. Checking for async and signal events	5211	calls in the current loop iteration and the loop is currently
		5212	blocked. Checking for async and signal events involves iterating over all
5004	involves iterating over all running async watchers or all signal numbers.	5213	running async watchers or all signal numbers.
5005		5214
5006	=back	5215	=back
5007		5216
5008		5217
5009	=head1 PORTING FROM LIBEV 3.X TO 4.X	5218	=head1 PORTING FROM LIBEV 3.X TO 4.X
…		…
5126	The physical time that is observed. It is apparently strictly monotonic :)	5335	The physical time that is observed. It is apparently strictly monotonic :)
5127		5336
5128	=item wall-clock time	5337	=item wall-clock time
5129		5338
5130	The time and date as shown on clocks. Unlike real time, it can actually	5339	The time and date as shown on clocks. Unlike real time, it can actually
5131	be wrong and jump forwards and backwards, e.g. when the you adjust your	5340	be wrong and jump forwards and backwards, e.g. when you adjust your
5132	clock.	5341	clock.
5133		5342
5134	=item watcher	5343	=item watcher
5135		5344
5136	A data structure that describes interest in certain events. Watchers need	5345	A data structure that describes interest in certain events. Watchers need
…		…
5139	=back	5348	=back
5140		5349
5141	=head1 AUTHOR	5350	=head1 AUTHOR
5142		5351
5143	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael	5352	Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
5144	Magnusson and Emanuele Giaquinta.	5353	Magnusson and Emanuele Giaquinta, and minor corrections by many others.
5145		5354

Diff Legend

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

Comparing libev/ev.pod (file contents): Revision 1.352 by root, Mon Jan 10 14:30:15 2011 UTC vs. Revision 1.381 by root, Sat Aug 13 17:41:14 2011 UTC

Diff Legend

Comparing libev/ev.pod (file contents):
Revision 1.352 by root, Mon Jan 10 14:30:15 2011 UTC vs.
Revision 1.381 by root, Sat Aug 13 17:41:14 2011 UTC