ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.19 by root, Sun Dec 10 23:59:15 2006 UTC vs.
Revision 1.303 by root, Sat Dec 5 02:52:03 2009 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent - provide framework for multiple event loops	3	AnyEvent - the DBI of event loop programming
4		4
5	Event, Coro, Glib, Tk, Perl - various supported event loops	5	EV, Event, Glib, Tk, Perl, Event::Lib, Irssi, rxvt-unicode, IO::Async, Qt
		6	and POE are various supported event loops/environments.
6		7
7	=head1 SYNOPSIS	8	=head1 SYNOPSIS
8		9
9	use AnyEvent;	10	use AnyEvent;
10		11
		12	# file descriptor readable
11	my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub {	13	my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ... });
		14
		15	# one-shot or repeating timers
		16	my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
		17	my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...
		18
		19	print AnyEvent->now; # prints current event loop time
		20	print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
		21
		22	# POSIX signal
		23	my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
		24
		25	# child process exit
		26	my $w = AnyEvent->child (pid => $pid, cb => sub {
		27	my ($pid, $status) = @_;
12	...	28	...
13	});	29	});
14		30
15	my $w = AnyEvent->timer (after => $seconds, cb => sub {	31	# called when event loop idle (if applicable)
16	...	32	my $w = AnyEvent->idle (cb => sub { ... });
17	});
18		33
19	my $w = AnyEvent->condvar; # stores wether a condition was flagged	34	my $w = AnyEvent->condvar; # stores whether a condition was flagged
		35	$w->send; # wake up current and all future recv's
20	$w->wait; # enters "main loop" till $condvar gets ->broadcast	36	$w->recv; # enters "main loop" till $condvar gets ->send
21	$w->broadcast; # wake up current and all future wait's	37	# use a condvar in callback mode:
		38	$w->cb (sub { $_[0]->recv });
		39
		40	=head1 INTRODUCTION/TUTORIAL
		41
		42	This manpage is mainly a reference manual. If you are interested
		43	in a tutorial or some gentle introduction, have a look at the
		44	L<AnyEvent::Intro> manpage.
		45
		46	=head1 SUPPORT
		47
		48	There is a mailinglist for discussing all things AnyEvent, and an IRC
		49	channel, too.
		50
		51	See the AnyEvent project page at the B<Schmorpforge Ta-Sa Software
		52	Repository>, at L<http://anyevent.schmorp.de>, for more info.
		53
		54	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
		55
		56	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
		57	nowadays. So what is different about AnyEvent?
		58
		59	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
		60	policy> and AnyEvent is I<small and efficient>.
		61
		62	First and foremost, I<AnyEvent is not an event model> itself, it only
		63	interfaces to whatever event model the main program happens to use, in a
		64	pragmatic way. For event models and certain classes of immortals alike,
		65	the statement "there can only be one" is a bitter reality: In general,
		66	only one event loop can be active at the same time in a process. AnyEvent
		67	cannot change this, but it can hide the differences between those event
		68	loops.
		69
		70	The goal of AnyEvent is to offer module authors the ability to do event
		71	programming (waiting for I/O or timer events) without subscribing to a
		72	religion, a way of living, and most importantly: without forcing your
		73	module users into the same thing by forcing them to use the same event
		74	model you use.
		75
		76	For modules like POE or IO::Async (which is a total misnomer as it is
		77	actually doing all I/O I<synchronously>...), using them in your module is
		78	like joining a cult: After you joined, you are dependent on them and you
		79	cannot use anything else, as they are simply incompatible to everything
		80	that isn't them. What's worse, all the potential users of your
		81	module are I<also> forced to use the same event loop you use.
		82
		83	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
		84	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
		85	with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
		86	your module uses one of those, every user of your module has to use it,
		87	too. But if your module uses AnyEvent, it works transparently with all
		88	event models it supports (including stuff like IO::Async, as long as those
		89	use one of the supported event loops. It is trivial to add new event loops
		90	to AnyEvent, too, so it is future-proof).
		91
		92	In addition to being free of having to use I<the one and only true event
		93	model>, AnyEvent also is free of bloat and policy: with POE or similar
		94	modules, you get an enormous amount of code and strict rules you have to
		95	follow. AnyEvent, on the other hand, is lean and up to the point, by only
		96	offering the functionality that is necessary, in as thin as a wrapper as
		97	technically possible.
		98
		99	Of course, AnyEvent comes with a big (and fully optional!) toolbox
		100	of useful functionality, such as an asynchronous DNS resolver, 100%
		101	non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
		102	such as Windows) and lots of real-world knowledge and workarounds for
		103	platform bugs and differences.
		104
		105	Now, if you I<do want> lots of policy (this can arguably be somewhat
		106	useful) and you want to force your users to use the one and only event
		107	model, you should I<not> use this module.
22		108
23	=head1 DESCRIPTION	109	=head1 DESCRIPTION
24		110
25	L<AnyEvent> provides an identical interface to multiple event loops. This	111	L<AnyEvent> provides an identical interface to multiple event loops. This
26	allows module authors to utilise an event loop without forcing module	112	allows module authors to utilise an event loop without forcing module
27	users to use the same event loop (as only a single event loop can coexist	113	users to use the same event loop (as only a single event loop can coexist
28	peacefully at any one time).	114	peacefully at any one time).
29		115
30	The interface itself is vaguely similar but not identical to the Event	116	The interface itself is vaguely similar, but not identical to the L<Event>
31	module.	117	module.
32		118
33	On the first call of any method, the module tries to detect the currently	119	During the first call of any watcher-creation method, the module tries
34	loaded event loop by probing wether any of the following modules is	120	to detect the currently loaded event loop by probing whether one of the
35	loaded: L<Coro::Event>, L<Event>, L<Glib>, L<Tk>. The first one found is	121	following modules is already loaded: L<EV>,
36	used. If none is found, the module tries to load these modules in the	122	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,
37	order given. The first one that could be successfully loaded will be	123	L<POE>. The first one found is used. If none are found, the module tries
38	used. If still none could be found, AnyEvent will fall back to a pure-perl	124	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
39	event loop, which is also not very efficient.	125	adaptor should always succeed) in the order given. The first one that can
		126	be successfully loaded will be used. If, after this, still none could be
		127	found, AnyEvent will fall back to a pure-perl event loop, which is not
		128	very efficient, but should work everywhere.
40		129
41	Because AnyEvent first checks for modules that are already loaded, loading	130	Because AnyEvent first checks for modules that are already loaded, loading
42	an Event model explicitly before first using AnyEvent will likely make	131	an event model explicitly before first using AnyEvent will likely make
43	that model the default. For example:	132	that model the default. For example:
44		133
45	use Tk;	134	use Tk;
46	use AnyEvent;	135	use AnyEvent;
47		136
48	# .. AnyEvent will likely default to Tk	137	# .. AnyEvent will likely default to Tk
49		138
		139	The I<likely> means that, if any module loads another event model and
		140	starts using it, all bets are off. Maybe you should tell their authors to
		141	use AnyEvent so their modules work together with others seamlessly...
		142
50	The pure-perl implementation of AnyEvent is called	143	The pure-perl implementation of AnyEvent is called
51	C<AnyEvent::Impl::Perl>. Like other event modules you can load it	144	C<AnyEvent::Impl::Perl>. Like other event modules you can load it
52	explicitly.	145	explicitly and enjoy the high availability of that event loop :)
53		146
54	=head1 WATCHERS	147	=head1 WATCHERS
55		148
56	AnyEvent has the central concept of a I<watcher>, which is an object that	149	AnyEvent has the central concept of a I<watcher>, which is an object that
57	stores relevant data for each kind of event you are waiting for, such as	150	stores relevant data for each kind of event you are waiting for, such as
58	the callback to call, the filehandle to watch, etc.	151	the callback to call, the file handle to watch, etc.
59		152
60	These watchers are normal Perl objects with normal Perl lifetime. After	153	These watchers are normal Perl objects with normal Perl lifetime. After
61	creating a watcher it will immediately "watch" for events and invoke	154	creating a watcher it will immediately "watch" for events and invoke the
		155	callback when the event occurs (of course, only when the event model
		156	is in control).
		157
		158	Note that B<callbacks must not permanently change global variables>
		159	potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
		160	callbacks must not C<die> >>. The former is good programming practise in
		161	Perl and the latter stems from the fact that exception handling differs
		162	widely between event loops.
		163
62	the callback. To disable the watcher you have to destroy it (e.g. by	164	To disable the watcher you have to destroy it (e.g. by setting the
63	setting the variable that stores it to C<undef> or otherwise deleting all	165	variable you store it in to C<undef> or otherwise deleting all references
64	references to it).	166	to it).
65		167
66	All watchers are created by calling a method on the C<AnyEvent> class.	168	All watchers are created by calling a method on the C<AnyEvent> class.
67		169
		170	Many watchers either are used with "recursion" (repeating timers for
		171	example), or need to refer to their watcher object in other ways.
		172
		173	An any way to achieve that is this pattern:
		174
		175	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
		176	# you can use $w here, for example to undef it
		177	undef $w;
		178	});
		179
		180	Note that C<my $w; $w => combination. This is necessary because in Perl,
		181	my variables are only visible after the statement in which they are
		182	declared.
		183
68	=head2 IO WATCHERS	184	=head2 I/O WATCHERS
69		185
		186	$w = AnyEvent->io (
		187	fh => <filehandle_or_fileno>,
		188	poll => <"r" or "w">,
		189	cb => <callback>,
		190	);
		191
70	You can create I/O watcher by calling the C<< AnyEvent->io >> method with	192	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
71	the following mandatory arguments:	193	with the following mandatory key-value pairs as arguments:
72		194
73	C<fh> the Perl I<filehandle> (not filedescriptor) to watch for	195	C<fh> is the Perl I<file handle> (or a naked file descriptor) to watch
		196	for events (AnyEvent might or might not keep a reference to this file
		197	handle). Note that only file handles pointing to things for which
		198	non-blocking operation makes sense are allowed. This includes sockets,
		199	most character devices, pipes, fifos and so on, but not for example files
		200	or block devices.
		201
74	events. C<poll> must be a string that is either C<r> or C<w>, that creates	202	C<poll> must be a string that is either C<r> or C<w>, which creates a
75	a watcher waiting for "r"eadable or "w"ritable events. C<cb> teh callback	203	watcher waiting for "r"eadable or "w"ritable events, respectively.
76	to invoke everytime the filehandle becomes ready.
77		204
78	Only one io watcher per C<fh> and C<poll> combination is allowed (i.e. on	205	C<cb> is the callback to invoke each time the file handle becomes ready.
79	a socket you can have one r + one w, not any more (limitation comes from
80	Tk - if you are sure you are not using Tk this limitation is gone).
81		206
82	Filehandles will be kept alive, so as long as the watcher exists, the	207	Although the callback might get passed parameters, their value and
83	filehandle exists, too.	208	presence is undefined and you cannot rely on them. Portable AnyEvent
		209	callbacks cannot use arguments passed to I/O watcher callbacks.
84		210
85	Example:	211	The I/O watcher might use the underlying file descriptor or a copy of it.
		212	You must not close a file handle as long as any watcher is active on the
		213	underlying file descriptor.
86		214
		215	Some event loops issue spurious readyness notifications, so you should
		216	always use non-blocking calls when reading/writing from/to your file
		217	handles.
		218
87	# wait for readability of STDIN, then read a line and disable the watcher	219	Example: wait for readability of STDIN, then read a line and disable the
		220	watcher.
		221
88	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	222	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
89	chomp (my $input = <STDIN>);	223	chomp (my $input = <STDIN>);
90	warn "read: $input\n";	224	warn "read: $input\n";
91	undef $w;	225	undef $w;
92	});	226	});
93		227
94	=head2 TIME WATCHERS	228	=head2 TIME WATCHERS
95		229
		230	$w = AnyEvent->timer (after => <seconds>, cb => <callback>);
		231
		232	$w = AnyEvent->timer (
		233	after => <fractional_seconds>,
		234	interval => <fractional_seconds>,
		235	cb => <callback>,
		236	);
		237
96	You can create a time watcher by calling the C<< AnyEvent->timer >>	238	You can create a time watcher by calling the C<< AnyEvent->timer >>
97	method with the following mandatory arguments:	239	method with the following mandatory arguments:
98		240
99	C<after> after how many seconds (fractions are supported) should the timer	241	C<after> specifies after how many seconds (fractional values are
100	activate. C<cb> the callback to invoke.	242	supported) the callback should be invoked. C<cb> is the callback to invoke
		243	in that case.
101		244
102	The timer callback will be invoked at most once: if you want a repeating	245	Although the callback might get passed parameters, their value and
103	timer you have to create a new watcher (this is a limitation by both Tk	246	presence is undefined and you cannot rely on them. Portable AnyEvent
104	and Glib).	247	callbacks cannot use arguments passed to time watcher callbacks.
105		248
106	Example:	249	The callback will normally be invoked once only. If you specify another
		250	parameter, C<interval>, as a strictly positive number (> 0), then the
		251	callback will be invoked regularly at that interval (in fractional
		252	seconds) after the first invocation. If C<interval> is specified with a
		253	false value, then it is treated as if it were missing.
107		254
		255	The callback will be rescheduled before invoking the callback, but no
		256	attempt is done to avoid timer drift in most backends, so the interval is
		257	only approximate.
		258
108	# fire an event after 7.7 seconds	259	Example: fire an event after 7.7 seconds.
		260
109	my $w = AnyEvent->timer (after => 7.7, cb => sub {	261	my $w = AnyEvent->timer (after => 7.7, cb => sub {
110	warn "timeout\n";	262	warn "timeout\n";
111	});	263	});
112		264
113	# to cancel the timer:	265	# to cancel the timer:
114	undef $w	266	undef $w;
115		267
116	=head2 CONDITION WATCHERS	268	Example 2: fire an event after 0.5 seconds, then roughly every second.
117		269
118	Condition watchers can be created by calling the C<< AnyEvent->condvar >>	270	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
119	method without any arguments.	271	warn "timeout\n";
		272	};
120		273
121	A condition watcher watches for a condition - precisely that the C<<	274	=head3 TIMING ISSUES
122	->broadcast >> method has been called.
123		275
124	The watcher has only two methods:	276	There are two ways to handle timers: based on real time (relative, "fire
		277	in 10 seconds") and based on wallclock time (absolute, "fire at 12
		278	o'clock").
		279
		280	While most event loops expect timers to specified in a relative way, they
		281	use absolute time internally. This makes a difference when your clock
		282	"jumps", for example, when ntp decides to set your clock backwards from
		283	the wrong date of 2014-01-01 to 2008-01-01, a watcher that is supposed to
		284	fire "after" a second might actually take six years to finally fire.
		285
		286	AnyEvent cannot compensate for this. The only event loop that is conscious
		287	about these issues is L<EV>, which offers both relative (ev_timer, based
		288	on true relative time) and absolute (ev_periodic, based on wallclock time)
		289	timers.
		290
		291	AnyEvent always prefers relative timers, if available, matching the
		292	AnyEvent API.
		293
		294	AnyEvent has two additional methods that return the "current time":
125		295
126	=over 4	296	=over 4
127		297
128	=item $cv->wait	298	=item AnyEvent->time
129		299
130	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	300	This returns the "current wallclock time" as a fractional number of
131	called on c<$cv>, while servicing other watchers normally.	301	seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
		302	return, and the result is guaranteed to be compatible with those).
132		303
133	Not all event models support a blocking wait - some die in that case, so	304	It progresses independently of any event loop processing, i.e. each call
134	if you are using this from a module, never require a blocking wait, but	305	will check the system clock, which usually gets updated frequently.
135	let the caller decide wether the call will block or not (for example,
136	by coupling condition variables with some kind of request results and
137	supporting callbacks so the caller knows that getting the result will not
138	block, while still suppporting blockign waits if the caller so desires).
139		306
140	You can only wait once on a condition - additional calls will return	307	=item AnyEvent->now
141	immediately.
142		308
143	=item $cv->broadcast	309	This also returns the "current wallclock time", but unlike C<time>, above,
		310	this value might change only once per event loop iteration, depending on
		311	the event loop (most return the same time as C<time>, above). This is the
		312	time that AnyEvent's timers get scheduled against.
144		313
145	Flag the condition as ready - a running C<< ->wait >> and all further	314	I<In almost all cases (in all cases if you don't care), this is the
146	calls to C<wait> will return after this method has been called. If nobody	315	function to call when you want to know the current time.>
147	is waiting the broadcast will be remembered..
148		316
149	Example:	317	This function is also often faster then C<< AnyEvent->time >>, and
		318	thus the preferred method if you want some timestamp (for example,
		319	L<AnyEvent::Handle> uses this to update it's activity timeouts).
		320
		321	The rest of this section is only of relevance if you try to be very exact
		322	with your timing, you can skip it without bad conscience.
		323
		324	For a practical example of when these times differ, consider L<Event::Lib>
		325	and L<EV> and the following set-up:
		326
		327	The event loop is running and has just invoked one of your callback at
		328	time=500 (assume no other callbacks delay processing). In your callback,
		329	you wait a second by executing C<sleep 1> (blocking the process for a
		330	second) and then (at time=501) you create a relative timer that fires
		331	after three seconds.
		332
		333	With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
		334	both return C<501>, because that is the current time, and the timer will
		335	be scheduled to fire at time=504 (C<501> + C<3>).
		336
		337	With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
		338	time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
		339	last event processing phase started. With L<EV>, your timer gets scheduled
		340	to run at time=503 (C<500> + C<3>).
		341
		342	In one sense, L<Event::Lib> is more exact, as it uses the current time
		343	regardless of any delays introduced by event processing. However, most
		344	callbacks do not expect large delays in processing, so this causes a
		345	higher drift (and a lot more system calls to get the current time).
		346
		347	In another sense, L<EV> is more exact, as your timer will be scheduled at
		348	the same time, regardless of how long event processing actually took.
		349
		350	In either case, if you care (and in most cases, you don't), then you
		351	can get whatever behaviour you want with any event loop, by taking the
		352	difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
		353	account.
		354
		355	=item AnyEvent->now_update
		356
		357	Some event loops (such as L<EV> or L<AnyEvent::Impl::Perl>) cache
		358	the current time for each loop iteration (see the discussion of L<<
		359	AnyEvent->now >>, above).
		360
		361	When a callback runs for a long time (or when the process sleeps), then
		362	this "current" time will differ substantially from the real time, which
		363	might affect timers and time-outs.
		364
		365	When this is the case, you can call this method, which will update the
		366	event loop's idea of "current time".
		367
		368	A typical example would be a script in a web server (e.g. C<mod_perl>) -
		369	when mod_perl executes the script, then the event loop will have the wrong
		370	idea about the "current time" (being potentially far in the past, when the
		371	script ran the last time). In that case you should arrange a call to C<<
		372	AnyEvent->now_update >> each time the web server process wakes up again
		373	(e.g. at the start of your script, or in a handler).
		374
		375	Note that updating the time I<might> cause some events to be handled.
		376
		377	=back
		378
		379	=head2 SIGNAL WATCHERS
		380
		381	$w = AnyEvent->signal (signal => <uppercase_signal_name>, cb => <callback>);
		382
		383	You can watch for signals using a signal watcher, C<signal> is the signal
		384	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
		385	callback to be invoked whenever a signal occurs.
		386
		387	Although the callback might get passed parameters, their value and
		388	presence is undefined and you cannot rely on them. Portable AnyEvent
		389	callbacks cannot use arguments passed to signal watcher callbacks.
		390
		391	Multiple signal occurrences can be clumped together into one callback
		392	invocation, and callback invocation will be synchronous. Synchronous means
		393	that it might take a while until the signal gets handled by the process,
		394	but it is guaranteed not to interrupt any other callbacks.
		395
		396	The main advantage of using these watchers is that you can share a signal
		397	between multiple watchers, and AnyEvent will ensure that signals will not
		398	interrupt your program at bad times.
		399
		400	This watcher might use C<%SIG> (depending on the event loop used),
		401	so programs overwriting those signals directly will likely not work
		402	correctly.
		403
		404	Example: exit on SIGINT
		405
		406	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
		407
		408	=head3 Restart Behaviour
		409
		410	While restart behaviour is up to the event loop implementation, most will
		411	not restart syscalls (that includes L<Async::Interrupt> and AnyEvent's
		412	pure perl implementation).
		413
		414	=head3 Safe/Unsafe Signals
		415
		416	Perl signals can be either "safe" (synchronous to opcode handling) or
		417	"unsafe" (asynchronous) - the former might get delayed indefinitely, the
		418	latter might corrupt your memory.
		419
		420	AnyEvent signal handlers are, in addition, synchronous to the event loop,
		421	i.e. they will not interrupt your running perl program but will only be
		422	called as part of the normal event handling (just like timer, I/O etc.
		423	callbacks, too).
		424
		425	=head3 Signal Races, Delays and Workarounds
		426
		427	Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
		428	callbacks to signals in a generic way, which is a pity, as you cannot
		429	do race-free signal handling in perl, requiring C libraries for
		430	this. AnyEvent will try to do it's best, which means in some cases,
		431	signals will be delayed. The maximum time a signal might be delayed is
		432	specified in C<$AnyEvent::MAX_SIGNAL_LATENCY> (default: 10 seconds). This
		433	variable can be changed only before the first signal watcher is created,
		434	and should be left alone otherwise. This variable determines how often
		435	AnyEvent polls for signals (in case a wake-up was missed). Higher values
		436	will cause fewer spurious wake-ups, which is better for power and CPU
		437	saving.
		438
		439	All these problems can be avoided by installing the optional
		440	L<Async::Interrupt> module, which works with most event loops. It will not
		441	work with inherently broken event loops such as L<Event> or L<Event::Lib>
		442	(and not with L<POE> currently, as POE does it's own workaround with
		443	one-second latency). For those, you just have to suffer the delays.
		444
		445	=head2 CHILD PROCESS WATCHERS
		446
		447	$w = AnyEvent->child (pid => <process id>, cb => <callback>);
		448
		449	You can also watch on a child process exit and catch its exit status.
		450
		451	The child process is specified by the C<pid> argument (one some backends,
		452	using C<0> watches for any child process exit, on others this will
		453	croak). The watcher will be triggered only when the child process has
		454	finished and an exit status is available, not on any trace events
		455	(stopped/continued).
		456
		457	The callback will be called with the pid and exit status (as returned by
		458	waitpid), so unlike other watcher types, you I<can> rely on child watcher
		459	callback arguments.
		460
		461	This watcher type works by installing a signal handler for C<SIGCHLD>,
		462	and since it cannot be shared, nothing else should use SIGCHLD or reap
		463	random child processes (waiting for specific child processes, e.g. inside
		464	C<system>, is just fine).
		465
		466	There is a slight catch to child watchers, however: you usually start them
		467	I<after> the child process was created, and this means the process could
		468	have exited already (and no SIGCHLD will be sent anymore).
		469
		470	Not all event models handle this correctly (neither POE nor IO::Async do,
		471	see their AnyEvent::Impl manpages for details), but even for event models
		472	that I<do> handle this correctly, they usually need to be loaded before
		473	the process exits (i.e. before you fork in the first place). AnyEvent's
		474	pure perl event loop handles all cases correctly regardless of when you
		475	start the watcher.
		476
		477	This means you cannot create a child watcher as the very first
		478	thing in an AnyEvent program, you I<have> to create at least one
		479	watcher before you C<fork> the child (alternatively, you can call
		480	C<AnyEvent::detect>).
		481
		482	As most event loops do not support waiting for child events, they will be
		483	emulated by AnyEvent in most cases, in which the latency and race problems
		484	mentioned in the description of signal watchers apply.
		485
		486	Example: fork a process and wait for it
		487
		488	my $done = AnyEvent->condvar;
		489
		490	my $pid = fork or exit 5;
		491
		492	my $w = AnyEvent->child (
		493	pid => $pid,
		494	cb => sub {
		495	my ($pid, $status) = @_;
		496	warn "pid $pid exited with status $status";
		497	$done->send;
		498	},
		499	);
		500
		501	# do something else, then wait for process exit
		502	$done->recv;
		503
		504	=head2 IDLE WATCHERS
		505
		506	$w = AnyEvent->idle (cb => <callback>);
		507
		508	Sometimes there is a need to do something, but it is not so important
		509	to do it instantly, but only when there is nothing better to do. This
		510	"nothing better to do" is usually defined to be "no other events need
		511	attention by the event loop".
		512
		513	Idle watchers ideally get invoked when the event loop has nothing
		514	better to do, just before it would block the process to wait for new
		515	events. Instead of blocking, the idle watcher is invoked.
		516
		517	Most event loops unfortunately do not really support idle watchers (only
		518	EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
		519	will simply call the callback "from time to time".
		520
		521	Example: read lines from STDIN, but only process them when the
		522	program is otherwise idle:
		523
		524	my @lines; # read data
		525	my $idle_w;
		526	my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
		527	push @lines, scalar <STDIN>;
		528
		529	# start an idle watcher, if not already done
		530	$idle_w ||= AnyEvent->idle (cb => sub {
		531	# handle only one line, when there are lines left
		532	if (my $line = shift @lines) {
		533	print "handled when idle: $line";
		534	} else {
		535	# otherwise disable the idle watcher again
		536	undef $idle_w;
		537	}
		538	});
		539	});
		540
		541	=head2 CONDITION VARIABLES
		542
		543	$cv = AnyEvent->condvar;
		544
		545	$cv->send (<list>);
		546	my @res = $cv->recv;
		547
		548	If you are familiar with some event loops you will know that all of them
		549	require you to run some blocking "loop", "run" or similar function that
		550	will actively watch for new events and call your callbacks.
		551
		552	AnyEvent is slightly different: it expects somebody else to run the event
		553	loop and will only block when necessary (usually when told by the user).
		554
		555	The instrument to do that is called a "condition variable", so called
		556	because they represent a condition that must become true.
		557
		558	Now is probably a good time to look at the examples further below.
		559
		560	Condition variables can be created by calling the C<< AnyEvent->condvar
		561	>> method, usually without arguments. The only argument pair allowed is
		562	C<cb>, which specifies a callback to be called when the condition variable
		563	becomes true, with the condition variable as the first argument (but not
		564	the results).
		565
		566	After creation, the condition variable is "false" until it becomes "true"
		567	by calling the C<send> method (or calling the condition variable as if it
		568	were a callback, read about the caveats in the description for the C<<
		569	->send >> method).
		570
		571	Condition variables are similar to callbacks, except that you can
		572	optionally wait for them. They can also be called merge points - points
		573	in time where multiple outstanding events have been processed. And yet
		574	another way to call them is transactions - each condition variable can be
		575	used to represent a transaction, which finishes at some point and delivers
		576	a result. And yet some people know them as "futures" - a promise to
		577	compute/deliver something that you can wait for.
		578
		579	Condition variables are very useful to signal that something has finished,
		580	for example, if you write a module that does asynchronous http requests,
		581	then a condition variable would be the ideal candidate to signal the
		582	availability of results. The user can either act when the callback is
		583	called or can synchronously C<< ->recv >> for the results.
		584
		585	You can also use them to simulate traditional event loops - for example,
		586	you can block your main program until an event occurs - for example, you
		587	could C<< ->recv >> in your main program until the user clicks the Quit
		588	button of your app, which would C<< ->send >> the "quit" event.
		589
		590	Note that condition variables recurse into the event loop - if you have
		591	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
		592	lose. Therefore, condition variables are good to export to your caller, but
		593	you should avoid making a blocking wait yourself, at least in callbacks,
		594	as this asks for trouble.
		595
		596	Condition variables are represented by hash refs in perl, and the keys
		597	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		598	easy (it is often useful to build your own transaction class on top of
		599	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		600	it's C<new> method in your own C<new> method.
		601
		602	There are two "sides" to a condition variable - the "producer side" which
		603	eventually calls C<< -> send >>, and the "consumer side", which waits
		604	for the send to occur.
		605
		606	Example: wait for a timer.
150		607
151	# wait till the result is ready	608	# wait till the result is ready
152	my $result_ready = AnyEvent->condvar;	609	my $result_ready = AnyEvent->condvar;
153		610
154	# do something such as adding a timer	611	# do something such as adding a timer
155	# or socket watcher the calls $result_ready->broadcast	612	# or socket watcher the calls $result_ready->send
156	# when the "result" is ready.	613	# when the "result" is ready.
		614	# in this case, we simply use a timer:
		615	my $w = AnyEvent->timer (
		616	after => 1,
		617	cb => sub { $result_ready->send },
		618	);
157		619
		620	# this "blocks" (while handling events) till the callback
		621	# calls ->send
158	$result_ready->wait;	622	$result_ready->recv;
		623
		624	Example: wait for a timer, but take advantage of the fact that condition
		625	variables are also callable directly.
		626
		627	my $done = AnyEvent->condvar;
		628	my $delay = AnyEvent->timer (after => 5, cb => $done);
		629	$done->recv;
		630
		631	Example: Imagine an API that returns a condvar and doesn't support
		632	callbacks. This is how you make a synchronous call, for example from
		633	the main program:
		634
		635	use AnyEvent::CouchDB;
		636
		637	...
		638
		639	my @info = $couchdb->info->recv;
		640
		641	And this is how you would just set a callback to be called whenever the
		642	results are available:
		643
		644	$couchdb->info->cb (sub {
		645	my @info = $_[0]->recv;
		646	});
		647
		648	=head3 METHODS FOR PRODUCERS
		649
		650	These methods should only be used by the producing side, i.e. the
		651	code/module that eventually sends the signal. Note that it is also
		652	the producer side which creates the condvar in most cases, but it isn't
		653	uncommon for the consumer to create it as well.
		654
		655	=over 4
		656
		657	=item $cv->send (...)
		658
		659	Flag the condition as ready - a running C<< ->recv >> and all further
		660	calls to C<recv> will (eventually) return after this method has been
		661	called. If nobody is waiting the send will be remembered.
		662
		663	If a callback has been set on the condition variable, it is called
		664	immediately from within send.
		665
		666	Any arguments passed to the C<send> call will be returned by all
		667	future C<< ->recv >> calls.
		668
		669	Condition variables are overloaded so one can call them directly (as if
		670	they were a code reference). Calling them directly is the same as calling
		671	C<send>.
		672
		673	=item $cv->croak ($error)
		674
		675	Similar to send, but causes all call's to C<< ->recv >> to invoke
		676	C<Carp::croak> with the given error message/object/scalar.
		677
		678	This can be used to signal any errors to the condition variable
		679	user/consumer. Doing it this way instead of calling C<croak> directly
		680	delays the error detetcion, but has the overwhelmign advantage that it
		681	diagnoses the error at the place where the result is expected, and not
		682	deep in some event clalback without connection to the actual code causing
		683	the problem.
		684
		685	=item $cv->begin ([group callback])
		686
		687	=item $cv->end
		688
		689	These two methods can be used to combine many transactions/events into
		690	one. For example, a function that pings many hosts in parallel might want
		691	to use a condition variable for the whole process.
		692
		693	Every call to C<< ->begin >> will increment a counter, and every call to
		694	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		695	>>, the (last) callback passed to C<begin> will be executed, passing the
		696	condvar as first argument. That callback is I<supposed> to call C<< ->send
		697	>>, but that is not required. If no group callback was set, C<send> will
		698	be called without any arguments.
		699
		700	You can think of C<< $cv->send >> giving you an OR condition (one call
		701	sends), while C<< $cv->begin >> and C<< $cv->end >> giving you an AND
		702	condition (all C<begin> calls must be C<end>'ed before the condvar sends).
		703
		704	Let's start with a simple example: you have two I/O watchers (for example,
		705	STDOUT and STDERR for a program), and you want to wait for both streams to
		706	close before activating a condvar:
		707
		708	my $cv = AnyEvent->condvar;
		709
		710	$cv->begin; # first watcher
		711	my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
		712	defined sysread $fh1, my $buf, 4096
		713	or $cv->end;
		714	});
		715
		716	$cv->begin; # second watcher
		717	my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
		718	defined sysread $fh2, my $buf, 4096
		719	or $cv->end;
		720	});
		721
		722	$cv->recv;
		723
		724	This works because for every event source (EOF on file handle), there is
		725	one call to C<begin>, so the condvar waits for all calls to C<end> before
		726	sending.
		727
		728	The ping example mentioned above is slightly more complicated, as the
		729	there are results to be passwd back, and the number of tasks that are
		730	begung can potentially be zero:
		731
		732	my $cv = AnyEvent->condvar;
		733
		734	my %result;
		735	$cv->begin (sub { shift->send (\%result) });
		736
		737	for my $host (@list_of_hosts) {
		738	$cv->begin;
		739	ping_host_then_call_callback $host, sub {
		740	$result{$host} = ...;
		741	$cv->end;
		742	};
		743	}
		744
		745	$cv->end;
		746
		747	This code fragment supposedly pings a number of hosts and calls
		748	C<send> after results for all then have have been gathered - in any
		749	order. To achieve this, the code issues a call to C<begin> when it starts
		750	each ping request and calls C<end> when it has received some result for
		751	it. Since C<begin> and C<end> only maintain a counter, the order in which
		752	results arrive is not relevant.
		753
		754	There is an additional bracketing call to C<begin> and C<end> outside the
		755	loop, which serves two important purposes: first, it sets the callback
		756	to be called once the counter reaches C<0>, and second, it ensures that
		757	C<send> is called even when C<no> hosts are being pinged (the loop
		758	doesn't execute once).
		759
		760	This is the general pattern when you "fan out" into multiple (but
		761	potentially none) subrequests: use an outer C<begin>/C<end> pair to set
		762	the callback and ensure C<end> is called at least once, and then, for each
		763	subrequest you start, call C<begin> and for each subrequest you finish,
		764	call C<end>.
159		765
160	=back	766	=back
161		767
162	=head2 SIGNAL WATCHERS	768	=head3 METHODS FOR CONSUMERS
163		769
164	You can listen for signals using a signal watcher, C<signal> is the signal	770	These methods should only be used by the consuming side, i.e. the
165	I<name> without any C<SIG> prefix.	771	code awaits the condition.
166
167	These watchers might use C<%SIG>, so programs overwriting those signals
168	directly will likely not work correctly.
169
170	Example: exit on SIGINT
171
172	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
173
174	=head1 GLOBALS
175		772
176	=over 4	773	=over 4
177		774
		775	=item $cv->recv
		776
		777	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
		778	>> methods have been called on c<$cv>, while servicing other watchers
		779	normally.
		780
		781	You can only wait once on a condition - additional calls are valid but
		782	will return immediately.
		783
		784	If an error condition has been set by calling C<< ->croak >>, then this
		785	function will call C<croak>.
		786
		787	In list context, all parameters passed to C<send> will be returned,
		788	in scalar context only the first one will be returned.
		789
		790	Note that doing a blocking wait in a callback is not supported by any
		791	event loop, that is, recursive invocation of a blocking C<< ->recv
		792	>> is not allowed, and the C<recv> call will C<croak> if such a
		793	condition is detected. This condition can be slightly loosened by using
		794	L<Coro::AnyEvent>, which allows you to do a blocking C<< ->recv >> from
		795	any thread that doesn't run the event loop itself.
		796
		797	Not all event models support a blocking wait - some die in that case
		798	(programs might want to do that to stay interactive), so I<if you are
		799	using this from a module, never require a blocking wait>. Instead, let the
		800	caller decide whether the call will block or not (for example, by coupling
		801	condition variables with some kind of request results and supporting
		802	callbacks so the caller knows that getting the result will not block,
		803	while still supporting blocking waits if the caller so desires).
		804
		805	You can ensure that C<< -recv >> never blocks by setting a callback and
		806	only calling C<< ->recv >> from within that callback (or at a later
		807	time). This will work even when the event loop does not support blocking
		808	waits otherwise.
		809
		810	=item $bool = $cv->ready
		811
		812	Returns true when the condition is "true", i.e. whether C<send> or
		813	C<croak> have been called.
		814
		815	=item $cb = $cv->cb ($cb->($cv))
		816
		817	This is a mutator function that returns the callback set and optionally
		818	replaces it before doing so.
		819
		820	The callback will be called when the condition becomes (or already was)
		821	"true", i.e. when C<send> or C<croak> are called (or were called), with
		822	the only argument being the condition variable itself. Calling C<recv>
		823	inside the callback or at any later time is guaranteed not to block.
		824
		825	=back
		826
		827	=head1 SUPPORTED EVENT LOOPS/BACKENDS
		828
		829	The available backend classes are (every class has its own manpage):
		830
		831	=over 4
		832
		833	=item Backends that are autoprobed when no other event loop can be found.
		834
		835	EV is the preferred backend when no other event loop seems to be in
		836	use. If EV is not installed, then AnyEvent will fall back to its own
		837	pure-perl implementation, which is available everywhere as it comes with
		838	AnyEvent itself.
		839
		840	AnyEvent::Impl::EV based on EV (interface to libev, best choice).
		841	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
		842
		843	=item Backends that are transparently being picked up when they are used.
		844
		845	These will be used when they are currently loaded when the first watcher
		846	is created, in which case it is assumed that the application is using
		847	them. This means that AnyEvent will automatically pick the right backend
		848	when the main program loads an event module before anything starts to
		849	create watchers. Nothing special needs to be done by the main program.
		850
		851	AnyEvent::Impl::Event based on Event, very stable, few glitches.
		852	AnyEvent::Impl::Glib based on Glib, slow but very stable.
		853	AnyEvent::Impl::Tk based on Tk, very broken.
		854	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
		855	AnyEvent::Impl::POE based on POE, very slow, some limitations.
		856	AnyEvent::Impl::Irssi used when running within irssi.
		857
		858	=item Backends with special needs.
		859
		860	Qt requires the Qt::Application to be instantiated first, but will
		861	otherwise be picked up automatically. As long as the main program
		862	instantiates the application before any AnyEvent watchers are created,
		863	everything should just work.
		864
		865	AnyEvent::Impl::Qt based on Qt.
		866
		867	Support for IO::Async can only be partial, as it is too broken and
		868	architecturally limited to even support the AnyEvent API. It also
		869	is the only event loop that needs the loop to be set explicitly, so
		870	it can only be used by a main program knowing about AnyEvent. See
		871	L<AnyEvent::Impl::Async> for the gory details.
		872
		873	AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
		874
		875	=item Event loops that are indirectly supported via other backends.
		876
		877	Some event loops can be supported via other modules:
		878
		879	There is no direct support for WxWidgets (L<Wx>) or L<Prima>.
		880
		881	B<WxWidgets> has no support for watching file handles. However, you can
		882	use WxWidgets through the POE adaptor, as POE has a Wx backend that simply
		883	polls 20 times per second, which was considered to be too horrible to even
		884	consider for AnyEvent.
		885
		886	B<Prima> is not supported as nobody seems to be using it, but it has a POE
		887	backend, so it can be supported through POE.
		888
		889	AnyEvent knows about both L<Prima> and L<Wx>, however, and will try to
		890	load L<POE> when detecting them, in the hope that POE will pick them up,
		891	in which case everything will be automatic.
		892
		893	=back
		894
		895	=head1 GLOBAL VARIABLES AND FUNCTIONS
		896
		897	These are not normally required to use AnyEvent, but can be useful to
		898	write AnyEvent extension modules.
		899
		900	=over 4
		901
178	=item $AnyEvent::MODEL	902	=item $AnyEvent::MODEL
179		903
180	Contains C<undef> until the first watcher is being created. Then it	904	Contains C<undef> until the first watcher is being created, before the
		905	backend has been autodetected.
		906
181	contains the event model that is being used, which is the name of the	907	Afterwards it contains the event model that is being used, which is the
182	Perl class implementing the model. This class is usually one of the	908	name of the Perl class implementing the model. This class is usually one
183	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	909	of the C<AnyEvent::Impl:xxx> modules, but can be any other class in the
184	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	910	case AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode> it
185		911	will be C<urxvt::anyevent>).
186	The known classes so far are:
187
188	AnyEvent::Impl::Coro based on Coro::Event, best choise.
189	AnyEvent::Impl::Event based on Event, also best choice :)
190	AnyEvent::Impl::Glib based on Glib, second-best choice.
191	AnyEvent::Impl::Tk based on Tk, very bad choice.
192	AnyEvent::Impl::Perl pure-perl implementation, inefficient.
193		912
194	=item AnyEvent::detect	913	=item AnyEvent::detect
195		914
196	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model if	915	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
197	necessary. You should only call this function right before you would have	916	if necessary. You should only call this function right before you would
198	created an AnyEvent watcher anyway, that is, very late at runtime.	917	have created an AnyEvent watcher anyway, that is, as late as possible at
		918	runtime, and not e.g. while initialising of your module.
		919
		920	If you need to do some initialisation before AnyEvent watchers are
		921	created, use C<post_detect>.
		922
		923	=item $guard = AnyEvent::post_detect { BLOCK }
		924
		925	Arranges for the code block to be executed as soon as the event model is
		926	autodetected (or immediately if this has already happened).
		927
		928	The block will be executed I<after> the actual backend has been detected
		929	(C<$AnyEvent::MODEL> is set), but I<before> any watchers have been
		930	created, so it is possible to e.g. patch C<@AnyEvent::ISA> or do
		931	other initialisations - see the sources of L<AnyEvent::Strict> or
		932	L<AnyEvent::AIO> to see how this is used.
		933
		934	The most common usage is to create some global watchers, without forcing
		935	event module detection too early, for example, L<AnyEvent::AIO> creates
		936	and installs the global L<IO::AIO> watcher in a C<post_detect> block to
		937	avoid autodetecting the event module at load time.
		938
		939	If called in scalar or list context, then it creates and returns an object
		940	that automatically removes the callback again when it is destroyed (or
		941	C<undef> when the hook was immediately executed). See L<AnyEvent::AIO> for
		942	a case where this is useful.
		943
		944	Example: Create a watcher for the IO::AIO module and store it in
		945	C<$WATCHER>. Only do so after the event loop is initialised, though.
		946
		947	our WATCHER;
		948
		949	my $guard = AnyEvent::post_detect {
		950	$WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
		951	};
		952
		953	# the ||= is important in case post_detect immediately runs the block,
		954	# as to not clobber the newly-created watcher. assigning both watcher and
		955	# post_detect guard to the same variable has the advantage of users being
		956	# able to just C<undef $WATCHER> if the watcher causes them grief.
		957
		958	$WATCHER ||= $guard;
		959
		960	=item @AnyEvent::post_detect
		961
		962	If there are any code references in this array (you can C<push> to it
		963	before or after loading AnyEvent), then they will called directly after
		964	the event loop has been chosen.
		965
		966	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		967	if it is defined then the event loop has already been detected, and the
		968	array will be ignored.
		969
		970	Best use C<AnyEvent::post_detect { BLOCK }> when your application allows
		971	it,as it takes care of these details.
		972
		973	This variable is mainly useful for modules that can do something useful
		974	when AnyEvent is used and thus want to know when it is initialised, but do
		975	not need to even load it by default. This array provides the means to hook
		976	into AnyEvent passively, without loading it.
199		977
200	=back	978	=back
201		979
202	=head1 WHAT TO DO IN A MODULE	980	=head1 WHAT TO DO IN A MODULE
203		981
204	As a module author, you should "use AnyEvent" and call AnyEvent methods	982	As a module author, you should C<use AnyEvent> and call AnyEvent methods
205	freely, but you should not load a specific event module or rely on it.	983	freely, but you should not load a specific event module or rely on it.
206		984
207	Be careful when you create watchers in the module body - Anyevent will	985	Be careful when you create watchers in the module body - AnyEvent will
208	decide which event module to use as soon as the first method is called, so	986	decide which event module to use as soon as the first method is called, so
209	by calling AnyEvent in your module body you force the user of your module	987	by calling AnyEvent in your module body you force the user of your module
210	to load the event module first.	988	to load the event module first.
211		989
		990	Never call C<< ->recv >> on a condition variable unless you I<know> that
		991	the C<< ->send >> method has been called on it already. This is
		992	because it will stall the whole program, and the whole point of using
		993	events is to stay interactive.
		994
		995	It is fine, however, to call C<< ->recv >> when the user of your module
		996	requests it (i.e. if you create a http request object ad have a method
		997	called C<results> that returns the results, it should call C<< ->recv >>
		998	freely, as the user of your module knows what she is doing. always).
		999
212	=head1 WHAT TO DO IN THE MAIN PROGRAM	1000	=head1 WHAT TO DO IN THE MAIN PROGRAM
213		1001
214	There will always be a single main program - the only place that should	1002	There will always be a single main program - the only place that should
215	dictate which event model to use.	1003	dictate which event model to use.
216		1004
217	If it doesn't care, it can just "use AnyEvent" and use it itself, or not	1005	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
218	do anything special and let AnyEvent decide which implementation to chose.	1006	do anything special (it does not need to be event-based) and let AnyEvent
		1007	decide which implementation to chose if some module relies on it.
219		1008
220	If the main program relies on a specific event model (for example, in Gtk2	1009	If the main program relies on a specific event model - for example, in
221	programs you have to rely on either Glib or Glib::Event), you should load	1010	Gtk2 programs you have to rely on the Glib module - you should load the
222	it before loading AnyEvent or any module that uses it, generally, as early	1011	event module before loading AnyEvent or any module that uses it: generally
223	as possible. The reason is that modules might create watchers when they	1012	speaking, you should load it as early as possible. The reason is that
224	are loaded, and AnyEvent will decide on the event model to use as soon as	1013	modules might create watchers when they are loaded, and AnyEvent will
225	it creates watchers, and it might chose the wrong one unless you load the	1014	decide on the event model to use as soon as it creates watchers, and it
226	correct one yourself.	1015	might chose the wrong one unless you load the correct one yourself.
227		1016
228	You can chose to use a rather inefficient pure-perl implementation by	1017	You can chose to use a pure-perl implementation by loading the
229	loading the C<AnyEvent::Impl::Perl> module, but letting AnyEvent chose is	1018	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
230	generally better.	1019	everywhere, but letting AnyEvent chose the model is generally better.
		1020
		1021	=head2 MAINLOOP EMULATION
		1022
		1023	Sometimes (often for short test scripts, or even standalone programs who
		1024	only want to use AnyEvent), you do not want to run a specific event loop.
		1025
		1026	In that case, you can use a condition variable like this:
		1027
		1028	AnyEvent->condvar->recv;
		1029
		1030	This has the effect of entering the event loop and looping forever.
		1031
		1032	Note that usually your program has some exit condition, in which case
		1033	it is better to use the "traditional" approach of storing a condition
		1034	variable somewhere, waiting for it, and sending it when the program should
		1035	exit cleanly.
		1036
		1037
		1038	=head1 OTHER MODULES
		1039
		1040	The following is a non-exhaustive list of additional modules that use
		1041	AnyEvent as a client and can therefore be mixed easily with other AnyEvent
		1042	modules and other event loops in the same program. Some of the modules
		1043	come with AnyEvent, most are available via CPAN.
		1044
		1045	=over 4
		1046
		1047	=item L<AnyEvent::Util>
		1048
		1049	Contains various utility functions that replace often-used but blocking
		1050	functions such as C<inet_aton> by event-/callback-based versions.
		1051
		1052	=item L<AnyEvent::Socket>
		1053
		1054	Provides various utility functions for (internet protocol) sockets,
		1055	addresses and name resolution. Also functions to create non-blocking tcp
		1056	connections or tcp servers, with IPv6 and SRV record support and more.
		1057
		1058	=item L<AnyEvent::Handle>
		1059
		1060	Provide read and write buffers, manages watchers for reads and writes,
		1061	supports raw and formatted I/O, I/O queued and fully transparent and
		1062	non-blocking SSL/TLS (via L<AnyEvent::TLS>.
		1063
		1064	=item L<AnyEvent::DNS>
		1065
		1066	Provides rich asynchronous DNS resolver capabilities.
		1067
		1068	=item L<AnyEvent::HTTP>
		1069
		1070	A simple-to-use HTTP library that is capable of making a lot of concurrent
		1071	HTTP requests.
		1072
		1073	=item L<AnyEvent::HTTPD>
		1074
		1075	Provides a simple web application server framework.
		1076
		1077	=item L<AnyEvent::FastPing>
		1078
		1079	The fastest ping in the west.
		1080
		1081	=item L<AnyEvent::DBI>
		1082
		1083	Executes L<DBI> requests asynchronously in a proxy process.
		1084
		1085	=item L<AnyEvent::AIO>
		1086
		1087	Truly asynchronous I/O, should be in the toolbox of every event
		1088	programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
		1089	together.
		1090
		1091	=item L<AnyEvent::BDB>
		1092
		1093	Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
		1094	L<BDB> and AnyEvent together.
		1095
		1096	=item L<AnyEvent::GPSD>
		1097
		1098	A non-blocking interface to gpsd, a daemon delivering GPS information.
		1099
		1100	=item L<AnyEvent::IRC>
		1101
		1102	AnyEvent based IRC client module family (replacing the older Net::IRC3).
		1103
		1104	=item L<AnyEvent::XMPP>
		1105
		1106	AnyEvent based XMPP (Jabber protocol) module family (replacing the older
		1107	Net::XMPP2>.
		1108
		1109	=item L<AnyEvent::IGS>
		1110
		1111	A non-blocking interface to the Internet Go Server protocol (used by
		1112	L<App::IGS>).
		1113
		1114	=item L<Net::FCP>
		1115
		1116	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		1117	of AnyEvent.
		1118
		1119	=item L<Event::ExecFlow>
		1120
		1121	High level API for event-based execution flow control.
		1122
		1123	=item L<Coro>
		1124
		1125	Has special support for AnyEvent via L<Coro::AnyEvent>.
		1126
		1127	=back
231		1128
232	=cut	1129	=cut
233		1130
234	package AnyEvent;	1131	package AnyEvent;
235		1132
236	no warnings;	1133	# basically a tuned-down version of common::sense
237	use strict;	1134	sub common_sense {
		1135	# from common:.sense 1.0
		1136	${^WARNING_BITS} = "\xfc\x3f\x33\x00\x0f\xf3\xcf\xc0\xf3\xfc\x33\x03";
		1137	# use strict vars subs
		1138	$^H |= 0x00000600;
		1139	}
		1140
		1141	BEGIN { AnyEvent::common_sense }
		1142
238	use Carp;	1143	use Carp ();
239		1144
240	our $VERSION = '2.5';	1145	our $VERSION = '5.22';
241	our $MODEL;	1146	our $MODEL;
242		1147
243	our $AUTOLOAD;	1148	our $AUTOLOAD;
244	our @ISA;	1149	our @ISA;
245		1150
246	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
247
248	our @REGISTRY;	1151	our @REGISTRY;
249		1152
		1153	our $VERBOSE;
		1154
		1155	BEGIN {
		1156	eval "sub WIN32(){ " . (($^O =~ /mswin32/i)*1) ." }";
		1157	eval "sub TAINT(){ " . (${^TAINT}*1) . " }";
		1158
		1159	delete @ENV{grep /^PERL_ANYEVENT_/, keys %ENV}
		1160	if ${^TAINT};
		1161
		1162	$VERBOSE = $ENV{PERL_ANYEVENT_VERBOSE}*1;
		1163
		1164	}
		1165
		1166	our $MAX_SIGNAL_LATENCY = 10;
		1167
		1168	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		1169
		1170	{
		1171	my $idx;
		1172	$PROTOCOL{$_} = ++$idx
		1173	for reverse split /\s*,\s*/,
		1174	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		1175	}
		1176
250	my @models = (	1177	my @models = (
251	[Coro::Event:: => AnyEvent::Impl::Coro::],	1178	[EV:: => AnyEvent::Impl::EV:: , 1],
		1179	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl:: , 1],
		1180	# everything below here will not (normally) be autoprobed
		1181	# as the pureperl backend should work everywhere
		1182	# and is usually faster
252	[Event:: => AnyEvent::Impl::Event::],	1183	[Event:: => AnyEvent::Impl::Event::, 1],
253	[Glib:: => AnyEvent::Impl::Glib::],	1184	[Glib:: => AnyEvent::Impl::Glib:: , 1], # becomes extremely slow with many watchers
		1185	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
		1186	[Irssi:: => AnyEvent::Impl::Irssi::], # Irssi has a bogus "Event" package
		1187	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		1188	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
		1189	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
254	[Tk:: => AnyEvent::Impl::Tk::],	1190	[Wx:: => AnyEvent::Impl::POE::],
255	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	1191	[Prima:: => AnyEvent::Impl::POE::],
		1192	# IO::Async is just too broken - we would need workarounds for its
		1193	# byzantine signal and broken child handling, among others.
		1194	# IO::Async is rather hard to detect, as it doesn't have any
		1195	# obvious default class.
		1196	[IO::Async:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1197	[IO::Async::Loop:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1198	[IO::Async::Notifier:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1199	[AnyEvent::Impl::IOAsync:: => AnyEvent::Impl::IOAsync::], # requires special main program
256	);	1200	);
257		1201
258	our %method = map +($_ => 1), qw(io timer condvar broadcast wait signal one_event DESTROY);	1202	our %method = map +($_ => 1),
		1203	qw(io timer time now now_update signal child idle condvar one_event DESTROY);
		1204
		1205	our @post_detect;
		1206
		1207	sub post_detect(&) {
		1208	my ($cb) = @_;
		1209
		1210	if ($MODEL) {
		1211	$cb->();
		1212
		1213	undef
		1214	} else {
		1215	push @post_detect, $cb;
		1216
		1217	defined wantarray
		1218	? bless \$cb, "AnyEvent::Util::postdetect"
		1219	: ()
		1220	}
		1221	}
		1222
		1223	sub AnyEvent::Util::postdetect::DESTROY {
		1224	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		1225	}
259		1226
260	sub detect() {	1227	sub detect() {
261	unless ($MODEL) {	1228	unless ($MODEL) {
262	no strict 'refs';	1229	local $SIG{__DIE__};
		1230
		1231	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
		1232	my $model = "AnyEvent::Impl::$1";
		1233	if (eval "require $model") {
		1234	$MODEL = $model;
		1235	warn "AnyEvent: loaded model '$model' (forced by \$ENV{PERL_ANYEVENT_MODEL}), using it.\n" if $VERBOSE >= 2;
		1236	} else {
		1237	warn "AnyEvent: unable to load model '$model' (from \$ENV{PERL_ANYEVENT_MODEL}):\n$@" if $VERBOSE;
		1238	}
		1239	}
263		1240
264	# check for already loaded models	1241	# check for already loaded models
		1242	unless ($MODEL) {
265	for (@REGISTRY, @models) {	1243	for (@REGISTRY, @models) {
266	my ($package, $model) = @$_;	1244	my ($package, $model) = @$_;
267	if (${"$package\::VERSION"} > 0) {	1245	if (${"$package\::VERSION"} > 0) {
268	if (eval "require $model") {	1246	if (eval "require $model") {
269	$MODEL = $model;	1247	$MODEL = $model;
270	warn "AnyEvent: found model '$model', using it.\n" if $verbose > 1;	1248	warn "AnyEvent: autodetected model '$model', using it.\n" if $VERBOSE >= 2;
271	last;	1249	last;
		1250	}
272	}	1251	}
273	}	1252	}
		1253
		1254	unless ($MODEL) {
		1255	# try to autoload a model
		1256	for (@REGISTRY, @models) {
		1257	my ($package, $model, $autoload) = @$_;
		1258	if (
		1259	$autoload
		1260	and eval "require $package"
		1261	and ${"$package\::VERSION"} > 0
		1262	and eval "require $model"
		1263	) {
		1264	$MODEL = $model;
		1265	warn "AnyEvent: autoloaded model '$model', using it.\n" if $VERBOSE >= 2;
		1266	last;
		1267	}
		1268	}
		1269
		1270	$MODEL
		1271	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.\n";
		1272	}
274	}	1273	}
275		1274
276	unless ($MODEL) {	1275	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
277	# try to load a model
278
279	for (@REGISTRY, @models) {
280	my ($package, $model) = @$_;
281	if (eval "require $model") {
282	$MODEL = $model;
283	warn "AnyEvent: autoprobed and loaded model '$model', using it.\n" if $verbose > 1;
284	last;
285	}
286	}
287
288	$MODEL
289	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: Event (or Coro+Event), Glib or Tk.";
290	}
291		1276
292	unshift @ISA, $MODEL;	1277	unshift @ISA, $MODEL;
293	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	1278
		1279	require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
		1280
		1281	(shift @post_detect)->() while @post_detect;
294	}	1282	}
295		1283
296	$MODEL	1284	$MODEL
297	}	1285	}
298		1286
299	sub AUTOLOAD {	1287	sub AUTOLOAD {
300	(my $func = $AUTOLOAD) =~ s/.*://;	1288	(my $func = $AUTOLOAD) =~ s/.*://;
301		1289
302	$method{$func}	1290	$method{$func}
303	or croak "$func: not a valid method for AnyEvent objects";	1291	or Carp::croak "$func: not a valid method for AnyEvent objects";
304		1292
305	detect unless $MODEL;	1293	detect unless $MODEL;
306		1294
307	my $class = shift;	1295	my $class = shift;
308	$class->$func (@_);	1296	$class->$func (@_);
309	}	1297	}
310		1298
		1299	# utility function to dup a filehandle. this is used by many backends
		1300	# to support binding more than one watcher per filehandle (they usually
		1301	# allow only one watcher per fd, so we dup it to get a different one).
		1302	sub _dupfh($$;$$) {
		1303	my ($poll, $fh, $r, $w) = @_;
		1304
		1305	# cygwin requires the fh mode to be matching, unix doesn't
		1306	my ($rw, $mode) = $poll eq "r" ? ($r, "<&") : ($w, ">&");
		1307
		1308	open my $fh2, $mode, $fh
		1309	or die "AnyEvent->io: cannot dup() filehandle in mode '$poll': $!,";
		1310
		1311	# we assume CLOEXEC is already set by perl in all important cases
		1312
		1313	($fh2, $rw)
		1314	}
		1315
		1316	=head1 SIMPLIFIED AE API
		1317
		1318	Starting with version 5.0, AnyEvent officially supports a second, much
		1319	simpler, API that is designed to reduce the calling, typing and memory
		1320	overhead.
		1321
		1322	See the L<AE> manpage for details.
		1323
		1324	=cut
		1325
		1326	package AE;
		1327
		1328	our $VERSION = $AnyEvent::VERSION;
		1329
		1330	sub io($$$) {
		1331	AnyEvent->io (fh => $_[0], poll => $_[1] ? "w" : "r", cb => $_[2])
		1332	}
		1333
		1334	sub timer($$$) {
		1335	AnyEvent->timer (after => $_[0], interval => $_[1], cb => $_[2])
		1336	}
		1337
		1338	sub signal($$) {
		1339	AnyEvent->signal (signal => $_[0], cb => $_[1])
		1340	}
		1341
		1342	sub child($$) {
		1343	AnyEvent->child (pid => $_[0], cb => $_[1])
		1344	}
		1345
		1346	sub idle($) {
		1347	AnyEvent->idle (cb => $_[0])
		1348	}
		1349
		1350	sub cv(;&) {
		1351	AnyEvent->condvar (@_ ? (cb => $_[0]) : ())
		1352	}
		1353
		1354	sub now() {
		1355	AnyEvent->now
		1356	}
		1357
		1358	sub now_update() {
		1359	AnyEvent->now_update
		1360	}
		1361
		1362	sub time() {
		1363	AnyEvent->time
		1364	}
		1365
311	package AnyEvent::Base;	1366	package AnyEvent::Base;
312		1367
		1368	# default implementations for many methods
		1369
		1370	sub _time() {
		1371	# probe for availability of Time::HiRes
		1372	if (eval "use Time::HiRes (); Time::HiRes::time (); 1") {
		1373	warn "AnyEvent: using Time::HiRes for sub-second timing accuracy.\n" if $VERBOSE >= 8;
		1374	*_time = \&Time::HiRes::time;
		1375	# if (eval "use POSIX (); (POSIX::times())...
		1376	} else {
		1377	warn "AnyEvent: using built-in time(), WARNING, no sub-second resolution!\n" if $VERBOSE;
		1378	*_time = sub { time }; # epic fail
		1379	}
		1380
		1381	&_time
		1382	}
		1383
		1384	sub time { _time }
		1385	sub now { _time }
		1386	sub now_update { }
		1387
		1388	# default implementation for ->condvar
		1389
		1390	sub condvar {
		1391	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, "AnyEvent::CondVar"
		1392	}
		1393
313	# default implementation for signal	1394	# default implementation for ->signal
314		1395
315	our %SIG_CB;	1396	our $HAVE_ASYNC_INTERRUPT;
		1397
		1398	sub _have_async_interrupt() {
		1399	$HAVE_ASYNC_INTERRUPT = 1*(!$ENV{PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT}
		1400	&& eval "use Async::Interrupt 1.02 (); 1")
		1401	unless defined $HAVE_ASYNC_INTERRUPT;
		1402
		1403	$HAVE_ASYNC_INTERRUPT
		1404	}
		1405
		1406	our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
		1407	our (%SIG_ASY, %SIG_ASY_W);
		1408	our ($SIG_COUNT, $SIG_TW);
		1409
		1410	sub _signal_exec {
		1411	$HAVE_ASYNC_INTERRUPT
		1412	? $SIGPIPE_R->drain
		1413	: sysread $SIGPIPE_R, (my $dummy), 9;
		1414
		1415	while (%SIG_EV) {
		1416	for (keys %SIG_EV) {
		1417	delete $SIG_EV{$_};
		1418	$_->() for values %{ $SIG_CB{$_} || {} };
		1419	}
		1420	}
		1421	}
		1422
		1423	# install a dummy wakeup watcher to reduce signal catching latency
		1424	sub _sig_add() {
		1425	unless ($SIG_COUNT++) {
		1426	# try to align timer on a full-second boundary, if possible
		1427	my $NOW = AE::now;
		1428
		1429	$SIG_TW = AE::timer
		1430	$MAX_SIGNAL_LATENCY - ($NOW - int $NOW),
		1431	$MAX_SIGNAL_LATENCY,
		1432	sub { } # just for the PERL_ASYNC_CHECK
		1433	;
		1434	}
		1435	}
		1436
		1437	sub _sig_del {
		1438	undef $SIG_TW
		1439	unless --$SIG_COUNT;
		1440	}
		1441
		1442	our $_sig_name_init; $_sig_name_init = sub {
		1443	eval q{ # poor man's autoloading
		1444	undef $_sig_name_init;
		1445
		1446	if (_have_async_interrupt) {
		1447	*sig2num = \&Async::Interrupt::sig2num;
		1448	*sig2name = \&Async::Interrupt::sig2name;
		1449	} else {
		1450	require Config;
		1451
		1452	my %signame2num;
		1453	@signame2num{ split ' ', $Config::Config{sig_name} }
		1454	= split ' ', $Config::Config{sig_num};
		1455
		1456	my @signum2name;
		1457	@signum2name[values %signame2num] = keys %signame2num;
		1458
		1459	*sig2num = sub($) {
		1460	$_[0] > 0 ? shift : $signame2num{+shift}
		1461	};
		1462	*sig2name = sub ($) {
		1463	$_[0] > 0 ? $signum2name[+shift] : shift
		1464	};
		1465	}
		1466	};
		1467	die if $@;
		1468	};
		1469
		1470	sub sig2num ($) { &$_sig_name_init; &sig2num }
		1471	sub sig2name($) { &$_sig_name_init; &sig2name }
316		1472
317	sub signal {	1473	sub signal {
		1474	eval q{ # poor man's autoloading {}
		1475	# probe for availability of Async::Interrupt
		1476	if (_have_async_interrupt) {
		1477	warn "AnyEvent: using Async::Interrupt for race-free signal handling.\n" if $VERBOSE >= 8;
		1478
		1479	$SIGPIPE_R = new Async::Interrupt::EventPipe;
		1480	$SIG_IO = AE::io $SIGPIPE_R->fileno, 0, \&_signal_exec;
		1481
		1482	} else {
		1483	warn "AnyEvent: using emulated perl signal handling with latency timer.\n" if $VERBOSE >= 8;
		1484
		1485	require Fcntl;
		1486
		1487	if (AnyEvent::WIN32) {
		1488	require AnyEvent::Util;
		1489
		1490	($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
		1491	AnyEvent::Util::fh_nonblocking ($SIGPIPE_R, 1) if $SIGPIPE_R;
		1492	AnyEvent::Util::fh_nonblocking ($SIGPIPE_W, 1) if $SIGPIPE_W; # just in case
		1493	} else {
		1494	pipe $SIGPIPE_R, $SIGPIPE_W;
		1495	fcntl $SIGPIPE_R, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_R;
		1496	fcntl $SIGPIPE_W, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_W; # just in case
		1497
		1498	# not strictly required, as $^F is normally 2, but let's make sure...
		1499	fcntl $SIGPIPE_R, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
		1500	fcntl $SIGPIPE_W, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
		1501	}
		1502
		1503	$SIGPIPE_R
		1504	or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
		1505
		1506	$SIG_IO = AE::io $SIGPIPE_R, 0, \&_signal_exec;
		1507	}
		1508
		1509	*signal = sub {
		1510	my (undef, %arg) = @_;
		1511
		1512	my $signal = uc $arg{signal}
		1513	or Carp::croak "required option 'signal' is missing";
		1514
		1515	if ($HAVE_ASYNC_INTERRUPT) {
		1516	# async::interrupt
		1517
		1518	$signal = sig2num $signal;
		1519	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1520
		1521	$SIG_ASY{$signal} ||= new Async::Interrupt
		1522	cb => sub { undef $SIG_EV{$signal} },
		1523	signal => $signal,
		1524	pipe => [$SIGPIPE_R->filenos],
		1525	pipe_autodrain => 0,
		1526	;
		1527
		1528	} else {
		1529	# pure perl
		1530
		1531	# AE::Util has been loaded in signal
		1532	$signal = sig2name $signal;
		1533	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1534
		1535	$SIG{$signal} ||= sub {
		1536	local $!;
		1537	syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
		1538	undef $SIG_EV{$signal};
		1539	};
		1540
		1541	# can't do signal processing without introducing races in pure perl,
		1542	# so limit the signal latency.
		1543	_sig_add;
		1544	}
		1545
		1546	bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
		1547	};
		1548
		1549	*AnyEvent::Base::signal::DESTROY = sub {
		1550	my ($signal, $cb) = @{$_[0]};
		1551
		1552	_sig_del;
		1553
		1554	delete $SIG_CB{$signal}{$cb};
		1555
		1556	$HAVE_ASYNC_INTERRUPT
		1557	? delete $SIG_ASY{$signal}
		1558	: # delete doesn't work with older perls - they then
		1559	# print weird messages, or just unconditionally exit
		1560	# instead of getting the default action.
		1561	undef $SIG{$signal}
		1562	unless keys %{ $SIG_CB{$signal} };
		1563	};
		1564	};
		1565	die if $@;
		1566	&signal
		1567	}
		1568
		1569	# default implementation for ->child
		1570
		1571	our %PID_CB;
		1572	our $CHLD_W;
		1573	our $CHLD_DELAY_W;
		1574	our $WNOHANG;
		1575
		1576	sub _emit_childstatus($$) {
		1577	my (undef, $rpid, $rstatus) = @_;
		1578
		1579	$_->($rpid, $rstatus)
		1580	for values %{ $PID_CB{$rpid} || {} },
		1581	values %{ $PID_CB{0} || {} };
		1582	}
		1583
		1584	sub _sigchld {
		1585	my $pid;
		1586
		1587	AnyEvent->_emit_childstatus ($pid, $?)
		1588	while ($pid = waitpid -1, $WNOHANG) > 0;
		1589	}
		1590
		1591	sub child {
318	my (undef, %arg) = @_;	1592	my (undef, %arg) = @_;
319		1593
320	my $signal = uc $arg{signal}	1594	defined (my $pid = $arg{pid} + 0)
321	or Carp::croak "required option 'signal' is missing";	1595	or Carp::croak "required option 'pid' is missing";
322		1596
323	my $w = bless [$signal, $arg{cb}], "AnyEvent::Base::Signal";
324
325	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};	1597	$PID_CB{$pid}{$arg{cb}} = $arg{cb};
326	$SIG{$signal} ||= sub {	1598
327	$_->() for values %{ $SIG_CB{$signal} };	1599	# WNOHANG is almost cetrainly 1 everywhere
		1600	$WNOHANG ||= $^O =~ /^(?:openbsd|netbsd|linux|freebsd|cygwin|MSWin32)$/
		1601	? 1
		1602	: eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
		1603
		1604	unless ($CHLD_W) {
		1605	$CHLD_W = AE::signal CHLD => \&_sigchld;
		1606	# child could be a zombie already, so make at least one round
		1607	&_sigchld;
		1608	}
		1609
		1610	bless [$pid, $arg{cb}], "AnyEvent::Base::child"
		1611	}
		1612
		1613	sub AnyEvent::Base::child::DESTROY {
		1614	my ($pid, $cb) = @{$_[0]};
		1615
		1616	delete $PID_CB{$pid}{$cb};
		1617	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
		1618
		1619	undef $CHLD_W unless keys %PID_CB;
		1620	}
		1621
		1622	# idle emulation is done by simply using a timer, regardless
		1623	# of whether the process is idle or not, and not letting
		1624	# the callback use more than 50% of the time.
		1625	sub idle {
		1626	my (undef, %arg) = @_;
		1627
		1628	my ($cb, $w, $rcb) = $arg{cb};
		1629
		1630	$rcb = sub {
		1631	if ($cb) {
		1632	$w = _time;
		1633	&$cb;
		1634	$w = _time - $w;
		1635
		1636	# never use more then 50% of the time for the idle watcher,
		1637	# within some limits
		1638	$w = 0.0001 if $w < 0.0001;
		1639	$w = 5 if $w > 5;
		1640
		1641	$w = AE::timer $w, 0, $rcb;
		1642	} else {
		1643	# clean up...
		1644	undef $w;
		1645	undef $rcb;
		1646	}
328	};	1647	};
329		1648
330	$w	1649	$w = AE::timer 0.05, 0, $rcb;
331	}
332		1650
		1651	bless \\$cb, "AnyEvent::Base::idle"
		1652	}
		1653
333	sub AnyEvent::Base::Signal::DESTROY {	1654	sub AnyEvent::Base::idle::DESTROY {
334	my ($signal, $cb) = @{$_[0]};	1655	undef $${$_[0]};
335
336	delete $SIG_CB{$signal}{$cb};
337
338	$SIG{$signal} = 'DEFAULT' unless keys %{ $SIG_CB{$signal} };
339	}	1656	}
		1657
		1658	package AnyEvent::CondVar;
		1659
		1660	our @ISA = AnyEvent::CondVar::Base::;
		1661
		1662	package AnyEvent::CondVar::Base;
		1663
		1664	#use overload
		1665	# '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1666	# fallback => 1;
		1667
		1668	# save 300+ kilobytes by dirtily hardcoding overloading
		1669	${"AnyEvent::CondVar::Base::OVERLOAD"}{dummy}++; # Register with magic by touching.
		1670	*{'AnyEvent::CondVar::Base::()'} = sub { }; # "Make it findable via fetchmethod."
		1671	*{'AnyEvent::CondVar::Base::(&{}'} = sub { my $self = shift; sub { $self->send (@_) } }; # &{}
		1672	${'AnyEvent::CondVar::Base::()'} = 1; # fallback
		1673
		1674	our $WAITING;
		1675
		1676	sub _send {
		1677	# nop
		1678	}
		1679
		1680	sub send {
		1681	my $cv = shift;
		1682	$cv->{_ae_sent} = [@_];
		1683	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1684	$cv->_send;
		1685	}
		1686
		1687	sub croak {
		1688	$_[0]{_ae_croak} = $_[1];
		1689	$_[0]->send;
		1690	}
		1691
		1692	sub ready {
		1693	$_[0]{_ae_sent}
		1694	}
		1695
		1696	sub _wait {
		1697	$WAITING
		1698	and !$_[0]{_ae_sent}
		1699	and Carp::croak "AnyEvent::CondVar: recursive blocking wait detected";
		1700
		1701	local $WAITING = 1;
		1702	AnyEvent->one_event while !$_[0]{_ae_sent};
		1703	}
		1704
		1705	sub recv {
		1706	$_[0]->_wait;
		1707
		1708	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		1709	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		1710	}
		1711
		1712	sub cb {
		1713	my $cv = shift;
		1714
		1715	@_
		1716	and $cv->{_ae_cb} = shift
		1717	and $cv->{_ae_sent}
		1718	and (delete $cv->{_ae_cb})->($cv);
		1719
		1720	$cv->{_ae_cb}
		1721	}
		1722
		1723	sub begin {
		1724	++$_[0]{_ae_counter};
		1725	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		1726	}
		1727
		1728	sub end {
		1729	return if --$_[0]{_ae_counter};
		1730	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		1731	}
		1732
		1733	# undocumented/compatibility with pre-3.4
		1734	*broadcast = \&send;
		1735	*wait = \&_wait;
		1736
		1737	=head1 ERROR AND EXCEPTION HANDLING
		1738
		1739	In general, AnyEvent does not do any error handling - it relies on the
		1740	caller to do that if required. The L<AnyEvent::Strict> module (see also
		1741	the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
		1742	checking of all AnyEvent methods, however, which is highly useful during
		1743	development.
		1744
		1745	As for exception handling (i.e. runtime errors and exceptions thrown while
		1746	executing a callback), this is not only highly event-loop specific, but
		1747	also not in any way wrapped by this module, as this is the job of the main
		1748	program.
		1749
		1750	The pure perl event loop simply re-throws the exception (usually
		1751	within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
		1752	$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
		1753	so on.
		1754
		1755	=head1 ENVIRONMENT VARIABLES
		1756
		1757	The following environment variables are used by this module or its
		1758	submodules.
		1759
		1760	Note that AnyEvent will remove I<all> environment variables starting with
		1761	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		1762	enabled.
		1763
		1764	=over 4
		1765
		1766	=item C<PERL_ANYEVENT_VERBOSE>
		1767
		1768	By default, AnyEvent will be completely silent except in fatal
		1769	conditions. You can set this environment variable to make AnyEvent more
		1770	talkative.
		1771
		1772	When set to C<1> or higher, causes AnyEvent to warn about unexpected
		1773	conditions, such as not being able to load the event model specified by
		1774	C<PERL_ANYEVENT_MODEL>.
		1775
		1776	When set to C<2> or higher, cause AnyEvent to report to STDERR which event
		1777	model it chooses.
		1778
		1779	When set to C<8> or higher, then AnyEvent will report extra information on
		1780	which optional modules it loads and how it implements certain features.
		1781
		1782	=item C<PERL_ANYEVENT_STRICT>
		1783
		1784	AnyEvent does not do much argument checking by default, as thorough
		1785	argument checking is very costly. Setting this variable to a true value
		1786	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		1787	check the arguments passed to most method calls. If it finds any problems,
		1788	it will croak.
		1789
		1790	In other words, enables "strict" mode.
		1791
		1792	Unlike C<use strict> (or it's modern cousin, C<< use L<common::sense>
		1793	>>, it is definitely recommended to keep it off in production. Keeping
		1794	C<PERL_ANYEVENT_STRICT=1> in your environment while developing programs
		1795	can be very useful, however.
		1796
		1797	=item C<PERL_ANYEVENT_MODEL>
		1798
		1799	This can be used to specify the event model to be used by AnyEvent, before
		1800	auto detection and -probing kicks in. It must be a string consisting
		1801	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
		1802	and the resulting module name is loaded and if the load was successful,
		1803	used as event model. If it fails to load AnyEvent will proceed with
		1804	auto detection and -probing.
		1805
		1806	This functionality might change in future versions.
		1807
		1808	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
		1809	could start your program like this:
		1810
		1811	PERL_ANYEVENT_MODEL=Perl perl ...
		1812
		1813	=item C<PERL_ANYEVENT_PROTOCOLS>
		1814
		1815	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1816	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1817	of auto probing).
		1818
		1819	Must be set to a comma-separated list of protocols or address families,
		1820	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1821	used, and preference will be given to protocols mentioned earlier in the
		1822	list.
		1823
		1824	This variable can effectively be used for denial-of-service attacks
		1825	against local programs (e.g. when setuid), although the impact is likely
		1826	small, as the program has to handle conenction and other failures anyways.
		1827
		1828	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1829	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1830	- only support IPv4, never try to resolve or contact IPv6
		1831	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1832	IPv6, but prefer IPv6 over IPv4.
		1833
		1834	=item C<PERL_ANYEVENT_EDNS0>
		1835
		1836	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1837	for DNS. This extension is generally useful to reduce DNS traffic, but
		1838	some (broken) firewalls drop such DNS packets, which is why it is off by
		1839	default.
		1840
		1841	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1842	EDNS0 in its DNS requests.
		1843
		1844	=item C<PERL_ANYEVENT_MAX_FORKS>
		1845
		1846	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		1847	will create in parallel.
		1848
		1849	=item C<PERL_ANYEVENT_MAX_OUTSTANDING_DNS>
		1850
		1851	The default value for the C<max_outstanding> parameter for the default DNS
		1852	resolver - this is the maximum number of parallel DNS requests that are
		1853	sent to the DNS server.
		1854
		1855	=item C<PERL_ANYEVENT_RESOLV_CONF>
		1856
		1857	The file to use instead of F</etc/resolv.conf> (or OS-specific
		1858	configuration) in the default resolver. When set to the empty string, no
		1859	default config will be used.
		1860
		1861	=item C<PERL_ANYEVENT_CA_FILE>, C<PERL_ANYEVENT_CA_PATH>.
		1862
		1863	When neither C<ca_file> nor C<ca_path> was specified during
		1864	L<AnyEvent::TLS> context creation, and either of these environment
		1865	variables exist, they will be used to specify CA certificate locations
		1866	instead of a system-dependent default.
		1867
		1868	=item C<PERL_ANYEVENT_AVOID_GUARD> and C<PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT>
		1869
		1870	When these are set to C<1>, then the respective modules are not
		1871	loaded. Mostly good for testing AnyEvent itself.
		1872
		1873	=back
340		1874
341	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1875	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
		1876
		1877	This is an advanced topic that you do not normally need to use AnyEvent in
		1878	a module. This section is only of use to event loop authors who want to
		1879	provide AnyEvent compatibility.
342		1880
343	If you need to support another event library which isn't directly	1881	If you need to support another event library which isn't directly
344	supported by AnyEvent, you can supply your own interface to it by	1882	supported by AnyEvent, you can supply your own interface to it by
345	pushing, before the first watcher gets created, the package name of	1883	pushing, before the first watcher gets created, the package name of
346	the event module and the package name of the interface to use onto	1884	the event module and the package name of the interface to use onto
347	C<@AnyEvent::REGISTRY>. You can do that before and even without loading	1885	C<@AnyEvent::REGISTRY>. You can do that before and even without loading
348	AnyEvent.	1886	AnyEvent, so it is reasonably cheap.
349		1887
350	Example:	1888	Example:
351		1889
352	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];	1890	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];
353		1891
354	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>	1892	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>
355	package/class when it finds the C<urxvt> package/module is loaded. When	1893	package/class when it finds the C<urxvt> package/module is already loaded.
		1894
356	AnyEvent is loaded and asked to find a suitable event model, it will	1895	When AnyEvent is loaded and asked to find a suitable event model, it
357	first check for the presence of urxvt.	1896	will first check for the presence of urxvt by trying to C<use> the
		1897	C<urxvt::anyevent> module.
358		1898
359	The class should provide implementations for all watcher types (see	1899	The class should provide implementations for all watcher types. See
360	L<AnyEvent::Impl::Event> (source code), L<AnyEvent::Impl::Glib>	1900	L<AnyEvent::Impl::EV> (source code), L<AnyEvent::Impl::Glib> (Source code)
361	(Source code) and so on for actual examples, use C<perldoc -m	1901	and so on for actual examples. Use C<perldoc -m AnyEvent::Impl::Glib> to
362	AnyEvent::Impl::Glib> to see the sources).	1902	see the sources.
363		1903
		1904	If you don't provide C<signal> and C<child> watchers than AnyEvent will
		1905	provide suitable (hopefully) replacements.
		1906
364	The above isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)	1907	The above example isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)
365	uses the above line as-is. An interface isn't included in AnyEvent	1908	terminal emulator uses the above line as-is. An interface isn't included
366	because it doesn't make sense outside the embedded interpreter inside	1909	in AnyEvent because it doesn't make sense outside the embedded interpreter
367	I<rxvt-unicode>, and it is updated and maintained as part of the	1910	inside I<rxvt-unicode>, and it is updated and maintained as part of the
368	I<rxvt-unicode> distribution.	1911	I<rxvt-unicode> distribution.
369		1912
370	I<rxvt-unicode> also cheats a bit by not providing blocking access to	1913	I<rxvt-unicode> also cheats a bit by not providing blocking access to
371	condition variables: code blocking while waiting for a condition will	1914	condition variables: code blocking while waiting for a condition will
372	C<die>. This still works with most modules/usages, and blocking calls must	1915	C<die>. This still works with most modules/usages, and blocking calls must
373	not be in an interactive appliation, so it makes sense.	1916	not be done in an interactive application, so it makes sense.
374		1917
375	=head1 ENVIRONMENT VARIABLES
376
377	The following environment variables are used by this module:
378
379	C<PERL_ANYEVENT_VERBOSE> when set to C<2> or higher, reports which event
380	model gets used.
381
382	=head1 EXAMPLE	1918	=head1 EXAMPLE PROGRAM
383		1919
384	The following program uses an io watcher to read data from stdin, a timer	1920	The following program uses an I/O watcher to read data from STDIN, a timer
385	to display a message once per second, and a condvar to exit the program	1921	to display a message once per second, and a condition variable to quit the
386	when the user enters quit:	1922	program when the user enters quit:
387		1923
388	use AnyEvent;	1924	use AnyEvent;
389		1925
390	my $cv = AnyEvent->condvar;	1926	my $cv = AnyEvent->condvar;
391		1927
392	my $io_watcher = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	1928	my $io_watcher = AnyEvent->io (
		1929	fh => \*STDIN,
		1930	poll => 'r',
		1931	cb => sub {
393	warn "io event <$_[0]>\n"; # will always output <r>	1932	warn "io event <$_[0]>\n"; # will always output <r>
394	chomp (my $input = <STDIN>); # read a line	1933	chomp (my $input = <STDIN>); # read a line
395	warn "read: $input\n"; # output what has been read	1934	warn "read: $input\n"; # output what has been read
396	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1935	$cv->send if $input =~ /^q/i; # quit program if /^q/i
		1936	},
		1937	);
		1938
		1939	my $time_watcher = AnyEvent->timer (after => 1, interval => 1, cb => sub {
		1940	warn "timeout\n"; # print 'timeout' at most every second
397	});	1941	});
398		1942
399	my $time_watcher; # can only be used once
400
401	sub new_timer {
402	$timer = AnyEvent->timer (after => 1, cb => sub {
403	warn "timeout\n"; # print 'timeout' about every second
404	&new_timer; # and restart the time
405	});
406	}
407
408	new_timer; # create first timer
409
410	$cv->wait; # wait until user enters /^q/i	1943	$cv->recv; # wait until user enters /^q/i
411		1944
412	=head1 REAL-WORLD EXAMPLE	1945	=head1 REAL-WORLD EXAMPLE
413		1946
414	Consider the L<Net::FCP> module. It features (among others) the following	1947	Consider the L<Net::FCP> module. It features (among others) the following
415	API calls, which are to freenet what HTTP GET requests are to http:	1948	API calls, which are to freenet what HTTP GET requests are to http:
…		…
465	syswrite $txn->{fh}, $txn->{request}	1998	syswrite $txn->{fh}, $txn->{request}
466	or die "connection or write error";	1999	or die "connection or write error";
467	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	2000	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
468		2001
469	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	2002	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
470	result and signals any possible waiters that the request ahs finished:	2003	result and signals any possible waiters that the request has finished:
471		2004
472	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	2005	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
473		2006
474	if (end-of-file or data complete) {	2007	if (end-of-file or data complete) {
475	$txn->{result} = $txn->{buf};	2008	$txn->{result} = $txn->{buf};
476	$txn->{finished}->broadcast;	2009	$txn->{finished}->send;
477	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	2010	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
478	}	2011	}
479		2012
480	The C<result> method, finally, just waits for the finished signal (if the	2013	The C<result> method, finally, just waits for the finished signal (if the
481	request was already finished, it doesn't wait, of course, and returns the	2014	request was already finished, it doesn't wait, of course, and returns the
482	data:	2015	data:
483		2016
484	$txn->{finished}->wait;	2017	$txn->{finished}->recv;
485	return $txn->{result};	2018	return $txn->{result};
486		2019
487	The actual code goes further and collects all errors (C<die>s, exceptions)	2020	The actual code goes further and collects all errors (C<die>s, exceptions)
488	that occured during request processing. The C<result> method detects	2021	that occurred during request processing. The C<result> method detects
489	wether an exception as thrown (it is stored inside the $txn object)	2022	whether an exception as thrown (it is stored inside the $txn object)
490	and just throws the exception, which means connection errors and other	2023	and just throws the exception, which means connection errors and other
491	problems get reported tot he code that tries to use the result, not in a	2024	problems get reported tot he code that tries to use the result, not in a
492	random callback.	2025	random callback.
493		2026
494	All of this enables the following usage styles:	2027	All of this enables the following usage styles:
495		2028
496	1. Blocking:	2029	1. Blocking:
497		2030
498	my $data = $fcp->client_get ($url);	2031	my $data = $fcp->client_get ($url);
499		2032
500	2. Blocking, but parallelizing:	2033	2. Blocking, but running in parallel:
501		2034
502	my @datas = map $_->result,	2035	my @datas = map $_->result,
503	map $fcp->txn_client_get ($_),	2036	map $fcp->txn_client_get ($_),
504	@urls;	2037	@urls;
505		2038
506	Both blocking examples work without the module user having to know	2039	Both blocking examples work without the module user having to know
507	anything about events.	2040	anything about events.
508		2041
509	3a. Event-based in a main program, using any support Event module:	2042	3a. Event-based in a main program, using any supported event module:
510		2043
511	use Event;	2044	use EV;
512		2045
513	$fcp->txn_client_get ($url)->cb (sub {	2046	$fcp->txn_client_get ($url)->cb (sub {
514	my $txn = shift;	2047	my $txn = shift;
515	my $data = $txn->result;	2048	my $data = $txn->result;
516	...	2049	...
517	});	2050	});
518		2051
519	Event::loop;	2052	EV::loop;
520		2053
521	3b. The module user could use AnyEvent, too:	2054	3b. The module user could use AnyEvent, too:
522		2055
523	use AnyEvent;	2056	use AnyEvent;
524		2057
525	my $quit = AnyEvent->condvar;	2058	my $quit = AnyEvent->condvar;
526		2059
527	$fcp->txn_client_get ($url)->cb (sub {	2060	$fcp->txn_client_get ($url)->cb (sub {
528	...	2061	...
529	$quit->broadcast;	2062	$quit->send;
530	});	2063	});
531		2064
532	$quit->wait;	2065	$quit->recv;
		2066
		2067
		2068	=head1 BENCHMARKS
		2069
		2070	To give you an idea of the performance and overheads that AnyEvent adds
		2071	over the event loops themselves and to give you an impression of the speed
		2072	of various event loops I prepared some benchmarks.
		2073
		2074	=head2 BENCHMARKING ANYEVENT OVERHEAD
		2075
		2076	Here is a benchmark of various supported event models used natively and
		2077	through AnyEvent. The benchmark creates a lot of timers (with a zero
		2078	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
		2079	which it is), lets them fire exactly once and destroys them again.
		2080
		2081	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
		2082	distribution. It uses the L<AE> interface, which makes a real difference
		2083	for the EV and Perl backends only.
		2084
		2085	=head3 Explanation of the columns
		2086
		2087	I<watcher> is the number of event watchers created/destroyed. Since
		2088	different event models feature vastly different performances, each event
		2089	loop was given a number of watchers so that overall runtime is acceptable
		2090	and similar between tested event loop (and keep them from crashing): Glib
		2091	would probably take thousands of years if asked to process the same number
		2092	of watchers as EV in this benchmark.
		2093
		2094	I<bytes> is the number of bytes (as measured by the resident set size,
		2095	RSS) consumed by each watcher. This method of measuring captures both C
		2096	and Perl-based overheads.
		2097
		2098	I<create> is the time, in microseconds (millionths of seconds), that it
		2099	takes to create a single watcher. The callback is a closure shared between
		2100	all watchers, to avoid adding memory overhead. That means closure creation
		2101	and memory usage is not included in the figures.
		2102
		2103	I<invoke> is the time, in microseconds, used to invoke a simple
		2104	callback. The callback simply counts down a Perl variable and after it was
		2105	invoked "watcher" times, it would C<< ->send >> a condvar once to
		2106	signal the end of this phase.
		2107
		2108	I<destroy> is the time, in microseconds, that it takes to destroy a single
		2109	watcher.
		2110
		2111	=head3 Results
		2112
		2113	name watchers bytes create invoke destroy comment
		2114	EV/EV 100000 223 0.47 0.43 0.27 EV native interface
		2115	EV/Any 100000 223 0.48 0.42 0.26 EV + AnyEvent watchers
		2116	Coro::EV/Any 100000 223 0.47 0.42 0.26 coroutines + Coro::Signal
		2117	Perl/Any 100000 431 2.70 0.74 0.92 pure perl implementation
		2118	Event/Event 16000 516 31.16 31.84 0.82 Event native interface
		2119	Event/Any 16000 1203 42.61 34.79 1.80 Event + AnyEvent watchers
		2120	IOAsync/Any 16000 1911 41.92 27.45 16.81 via IO::Async::Loop::IO_Poll
		2121	IOAsync/Any 16000 1726 40.69 26.37 15.25 via IO::Async::Loop::Epoll
		2122	Glib/Any 16000 1118 89.00 12.57 51.17 quadratic behaviour
		2123	Tk/Any 2000 1346 20.96 10.75 8.00 SEGV with >> 2000 watchers
		2124	POE/Any 2000 6951 108.97 795.32 14.24 via POE::Loop::Event
		2125	POE/Any 2000 6648 94.79 774.40 575.51 via POE::Loop::Select
		2126
		2127	=head3 Discussion
		2128
		2129	The benchmark does I<not> measure scalability of the event loop very
		2130	well. For example, a select-based event loop (such as the pure perl one)
		2131	can never compete with an event loop that uses epoll when the number of
		2132	file descriptors grows high. In this benchmark, all events become ready at
		2133	the same time, so select/poll-based implementations get an unnatural speed
		2134	boost.
		2135
		2136	Also, note that the number of watchers usually has a nonlinear effect on
		2137	overall speed, that is, creating twice as many watchers doesn't take twice
		2138	the time - usually it takes longer. This puts event loops tested with a
		2139	higher number of watchers at a disadvantage.
		2140
		2141	To put the range of results into perspective, consider that on the
		2142	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		2143	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		2144	cycles with POE.
		2145
		2146	C<EV> is the sole leader regarding speed and memory use, which are both
		2147	maximal/minimal, respectively. When using the L<AE> API there is zero
		2148	overhead (when going through the AnyEvent API create is about 5-6 times
		2149	slower, with other times being equal, so still uses far less memory than
		2150	any other event loop and is still faster than Event natively).
		2151
		2152	The pure perl implementation is hit in a few sweet spots (both the
		2153	constant timeout and the use of a single fd hit optimisations in the perl
		2154	interpreter and the backend itself). Nevertheless this shows that it
		2155	adds very little overhead in itself. Like any select-based backend its
		2156	performance becomes really bad with lots of file descriptors (and few of
		2157	them active), of course, but this was not subject of this benchmark.
		2158
		2159	The C<Event> module has a relatively high setup and callback invocation
		2160	cost, but overall scores in on the third place.
		2161
		2162	C<IO::Async> performs admirably well, about on par with C<Event>, even
		2163	when using its pure perl backend.
		2164
		2165	C<Glib>'s memory usage is quite a bit higher, but it features a
		2166	faster callback invocation and overall ends up in the same class as
		2167	C<Event>. However, Glib scales extremely badly, doubling the number of
		2168	watchers increases the processing time by more than a factor of four,
		2169	making it completely unusable when using larger numbers of watchers
		2170	(note that only a single file descriptor was used in the benchmark, so
		2171	inefficiencies of C<poll> do not account for this).
		2172
		2173	The C<Tk> adaptor works relatively well. The fact that it crashes with
		2174	more than 2000 watchers is a big setback, however, as correctness takes
		2175	precedence over speed. Nevertheless, its performance is surprising, as the
		2176	file descriptor is dup()ed for each watcher. This shows that the dup()
		2177	employed by some adaptors is not a big performance issue (it does incur a
		2178	hidden memory cost inside the kernel which is not reflected in the figures
		2179	above).
		2180
		2181	C<POE>, regardless of underlying event loop (whether using its pure perl
		2182	select-based backend or the Event module, the POE-EV backend couldn't
		2183	be tested because it wasn't working) shows abysmal performance and
		2184	memory usage with AnyEvent: Watchers use almost 30 times as much memory
		2185	as EV watchers, and 10 times as much memory as Event (the high memory
		2186	requirements are caused by requiring a session for each watcher). Watcher
		2187	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		2188	implementation.
		2189
		2190	The design of the POE adaptor class in AnyEvent can not really account
		2191	for the performance issues, though, as session creation overhead is
		2192	small compared to execution of the state machine, which is coded pretty
		2193	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		2194	using multiple sessions is not a good approach, especially regarding
		2195	memory usage, even the author of POE could not come up with a faster
		2196	design).
		2197
		2198	=head3 Summary
		2199
		2200	=over 4
		2201
		2202	=item * Using EV through AnyEvent is faster than any other event loop
		2203	(even when used without AnyEvent), but most event loops have acceptable
		2204	performance with or without AnyEvent.
		2205
		2206	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
		2207	the actual event loop, only with extremely fast event loops such as EV
		2208	adds AnyEvent significant overhead.
		2209
		2210	=item * You should avoid POE like the plague if you want performance or
		2211	reasonable memory usage.
		2212
		2213	=back
		2214
		2215	=head2 BENCHMARKING THE LARGE SERVER CASE
		2216
		2217	This benchmark actually benchmarks the event loop itself. It works by
		2218	creating a number of "servers": each server consists of a socket pair, a
		2219	timeout watcher that gets reset on activity (but never fires), and an I/O
		2220	watcher waiting for input on one side of the socket. Each time the socket
		2221	watcher reads a byte it will write that byte to a random other "server".
		2222
		2223	The effect is that there will be a lot of I/O watchers, only part of which
		2224	are active at any one point (so there is a constant number of active
		2225	fds for each loop iteration, but which fds these are is random). The
		2226	timeout is reset each time something is read because that reflects how
		2227	most timeouts work (and puts extra pressure on the event loops).
		2228
		2229	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		2230	(1%) are active. This mirrors the activity of large servers with many
		2231	connections, most of which are idle at any one point in time.
		2232
		2233	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		2234	distribution. It uses the L<AE> interface, which makes a real difference
		2235	for the EV and Perl backends only.
		2236
		2237	=head3 Explanation of the columns
		2238
		2239	I<sockets> is the number of sockets, and twice the number of "servers" (as
		2240	each server has a read and write socket end).
		2241
		2242	I<create> is the time it takes to create a socket pair (which is
		2243	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		2244
		2245	I<request>, the most important value, is the time it takes to handle a
		2246	single "request", that is, reading the token from the pipe and forwarding
		2247	it to another server. This includes deleting the old timeout and creating
		2248	a new one that moves the timeout into the future.
		2249
		2250	=head3 Results
		2251
		2252	name sockets create request
		2253	EV 20000 62.66 7.99
		2254	Perl 20000 68.32 32.64
		2255	IOAsync 20000 174.06 101.15 epoll
		2256	IOAsync 20000 174.67 610.84 poll
		2257	Event 20000 202.69 242.91
		2258	Glib 20000 557.01 1689.52
		2259	POE 20000 341.54 12086.32 uses POE::Loop::Event
		2260
		2261	=head3 Discussion
		2262
		2263	This benchmark I<does> measure scalability and overall performance of the
		2264	particular event loop.
		2265
		2266	EV is again fastest. Since it is using epoll on my system, the setup time
		2267	is relatively high, though.
		2268
		2269	Perl surprisingly comes second. It is much faster than the C-based event
		2270	loops Event and Glib.
		2271
		2272	IO::Async performs very well when using its epoll backend, and still quite
		2273	good compared to Glib when using its pure perl backend.
		2274
		2275	Event suffers from high setup time as well (look at its code and you will
		2276	understand why). Callback invocation also has a high overhead compared to
		2277	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		2278	uses select or poll in basically all documented configurations.
		2279
		2280	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		2281	clearly fails to perform with many filehandles or in busy servers.
		2282
		2283	POE is still completely out of the picture, taking over 1000 times as long
		2284	as EV, and over 100 times as long as the Perl implementation, even though
		2285	it uses a C-based event loop in this case.
		2286
		2287	=head3 Summary
		2288
		2289	=over 4
		2290
		2291	=item * The pure perl implementation performs extremely well.
		2292
		2293	=item * Avoid Glib or POE in large projects where performance matters.
		2294
		2295	=back
		2296
		2297	=head2 BENCHMARKING SMALL SERVERS
		2298
		2299	While event loops should scale (and select-based ones do not...) even to
		2300	large servers, most programs we (or I :) actually write have only a few
		2301	I/O watchers.
		2302
		2303	In this benchmark, I use the same benchmark program as in the large server
		2304	case, but it uses only eight "servers", of which three are active at any
		2305	one time. This should reflect performance for a small server relatively
		2306	well.
		2307
		2308	The columns are identical to the previous table.
		2309
		2310	=head3 Results
		2311
		2312	name sockets create request
		2313	EV 16 20.00 6.54
		2314	Perl 16 25.75 12.62
		2315	Event 16 81.27 35.86
		2316	Glib 16 32.63 15.48
		2317	POE 16 261.87 276.28 uses POE::Loop::Event
		2318
		2319	=head3 Discussion
		2320
		2321	The benchmark tries to test the performance of a typical small
		2322	server. While knowing how various event loops perform is interesting, keep
		2323	in mind that their overhead in this case is usually not as important, due
		2324	to the small absolute number of watchers (that is, you need efficiency and
		2325	speed most when you have lots of watchers, not when you only have a few of
		2326	them).
		2327
		2328	EV is again fastest.
		2329
		2330	Perl again comes second. It is noticeably faster than the C-based event
		2331	loops Event and Glib, although the difference is too small to really
		2332	matter.
		2333
		2334	POE also performs much better in this case, but is is still far behind the
		2335	others.
		2336
		2337	=head3 Summary
		2338
		2339	=over 4
		2340
		2341	=item * C-based event loops perform very well with small number of
		2342	watchers, as the management overhead dominates.
		2343
		2344	=back
		2345
		2346	=head2 THE IO::Lambda BENCHMARK
		2347
		2348	Recently I was told about the benchmark in the IO::Lambda manpage, which
		2349	could be misinterpreted to make AnyEvent look bad. In fact, the benchmark
		2350	simply compares IO::Lambda with POE, and IO::Lambda looks better (which
		2351	shouldn't come as a surprise to anybody). As such, the benchmark is
		2352	fine, and mostly shows that the AnyEvent backend from IO::Lambda isn't
		2353	very optimal. But how would AnyEvent compare when used without the extra
		2354	baggage? To explore this, I wrote the equivalent benchmark for AnyEvent.
		2355
		2356	The benchmark itself creates an echo-server, and then, for 500 times,
		2357	connects to the echo server, sends a line, waits for the reply, and then
		2358	creates the next connection. This is a rather bad benchmark, as it doesn't
		2359	test the efficiency of the framework or much non-blocking I/O, but it is a
		2360	benchmark nevertheless.
		2361
		2362	name runtime
		2363	Lambda/select 0.330 sec
		2364	+ optimized 0.122 sec
		2365	Lambda/AnyEvent 0.327 sec
		2366	+ optimized 0.138 sec
		2367	Raw sockets/select 0.077 sec
		2368	POE/select, components 0.662 sec
		2369	POE/select, raw sockets 0.226 sec
		2370	POE/select, optimized 0.404 sec
		2371
		2372	AnyEvent/select/nb 0.085 sec
		2373	AnyEvent/EV/nb 0.068 sec
		2374	+state machine 0.134 sec
		2375
		2376	The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
		2377	benchmarks actually make blocking connects and use 100% blocking I/O,
		2378	defeating the purpose of an event-based solution. All of the newly
		2379	written AnyEvent benchmarks use 100% non-blocking connects (using
		2380	AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
		2381	resolver), so AnyEvent is at a disadvantage here, as non-blocking connects
		2382	generally require a lot more bookkeeping and event handling than blocking
		2383	connects (which involve a single syscall only).
		2384
		2385	The last AnyEvent benchmark additionally uses L<AnyEvent::Handle>, which
		2386	offers similar expressive power as POE and IO::Lambda, using conventional
		2387	Perl syntax. This means that both the echo server and the client are 100%
		2388	non-blocking, further placing it at a disadvantage.
		2389
		2390	As you can see, the AnyEvent + EV combination even beats the
		2391	hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
		2392	backend easily beats IO::Lambda and POE.
		2393
		2394	And even the 100% non-blocking version written using the high-level (and
		2395	slow :) L<AnyEvent::Handle> abstraction beats both POE and IO::Lambda
		2396	higher level ("unoptimised") abstractions by a large margin, even though
		2397	it does all of DNS, tcp-connect and socket I/O in a non-blocking way.
		2398
		2399	The two AnyEvent benchmarks programs can be found as F<eg/ae0.pl> and
		2400	F<eg/ae2.pl> in the AnyEvent distribution, the remaining benchmarks are
		2401	part of the IO::Lambda distribution and were used without any changes.
		2402
		2403
		2404	=head1 SIGNALS
		2405
		2406	AnyEvent currently installs handlers for these signals:
		2407
		2408	=over 4
		2409
		2410	=item SIGCHLD
		2411
		2412	A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
		2413	emulation for event loops that do not support them natively. Also, some
		2414	event loops install a similar handler.
		2415
		2416	Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, then
		2417	AnyEvent will reset it to default, to avoid losing child exit statuses.
		2418
		2419	=item SIGPIPE
		2420
		2421	A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
		2422	when AnyEvent gets loaded.
		2423
		2424	The rationale for this is that AnyEvent users usually do not really depend
		2425	on SIGPIPE delivery (which is purely an optimisation for shell use, or
		2426	badly-written programs), but C<SIGPIPE> can cause spurious and rare
		2427	program exits as a lot of people do not expect C<SIGPIPE> when writing to
		2428	some random socket.
		2429
		2430	The rationale for installing a no-op handler as opposed to ignoring it is
		2431	that this way, the handler will be restored to defaults on exec.
		2432
		2433	Feel free to install your own handler, or reset it to defaults.
		2434
		2435	=back
		2436
		2437	=cut
		2438
		2439	undef $SIG{CHLD}
		2440	if $SIG{CHLD} eq 'IGNORE';
		2441
		2442	$SIG{PIPE} = sub { }
		2443	unless defined $SIG{PIPE};
		2444
		2445	=head1 RECOMMENDED/OPTIONAL MODULES
		2446
		2447	One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
		2448	it's built-in modules) are required to use it.
		2449
		2450	That does not mean that AnyEvent won't take advantage of some additional
		2451	modules if they are installed.
		2452
		2453	This section explains which additional modules will be used, and how they
		2454	affect AnyEvent's operation.
		2455
		2456	=over 4
		2457
		2458	=item L<Async::Interrupt>
		2459
		2460	This slightly arcane module is used to implement fast signal handling: To
		2461	my knowledge, there is no way to do completely race-free and quick
		2462	signal handling in pure perl. To ensure that signals still get
		2463	delivered, AnyEvent will start an interval timer to wake up perl (and
		2464	catch the signals) with some delay (default is 10 seconds, look for
		2465	C<$AnyEvent::MAX_SIGNAL_LATENCY>).
		2466
		2467	If this module is available, then it will be used to implement signal
		2468	catching, which means that signals will not be delayed, and the event loop
		2469	will not be interrupted regularly, which is more efficient (and good for
		2470	battery life on laptops).
		2471
		2472	This affects not just the pure-perl event loop, but also other event loops
		2473	that have no signal handling on their own (e.g. Glib, Tk, Qt).
		2474
		2475	Some event loops (POE, Event, Event::Lib) offer signal watchers natively,
		2476	and either employ their own workarounds (POE) or use AnyEvent's workaround
		2477	(using C<$AnyEvent::MAX_SIGNAL_LATENCY>). Installing L<Async::Interrupt>
		2478	does nothing for those backends.
		2479
		2480	=item L<EV>
		2481
		2482	This module isn't really "optional", as it is simply one of the backend
		2483	event loops that AnyEvent can use. However, it is simply the best event
		2484	loop available in terms of features, speed and stability: It supports
		2485	the AnyEvent API optimally, implements all the watcher types in XS, does
		2486	automatic timer adjustments even when no monotonic clock is available,
		2487	can take avdantage of advanced kernel interfaces such as C<epoll> and
		2488	C<kqueue>, and is the fastest backend I<by far>. You can even embed
		2489	L<Glib>/L<Gtk2> in it (or vice versa, see L<EV::Glib> and L<Glib::EV>).
		2490
		2491	=item L<Guard>
		2492
		2493	The guard module, when used, will be used to implement
		2494	C<AnyEvent::Util::guard>. This speeds up guards considerably (and uses a
		2495	lot less memory), but otherwise doesn't affect guard operation much. It is
		2496	purely used for performance.
		2497
		2498	=item L<JSON> and L<JSON::XS>
		2499
		2500	One of these modules is required when you want to read or write JSON data
		2501	via L<AnyEvent::Handle>. It is also written in pure-perl, but can take
		2502	advantage of the ultra-high-speed L<JSON::XS> module when it is installed.
		2503
		2504	In fact, L<AnyEvent::Handle> will use L<JSON::XS> by default if it is
		2505	installed.
		2506
		2507	=item L<Net::SSLeay>
		2508
		2509	Implementing TLS/SSL in Perl is certainly interesting, but not very
		2510	worthwhile: If this module is installed, then L<AnyEvent::Handle> (with
		2511	the help of L<AnyEvent::TLS>), gains the ability to do TLS/SSL.
		2512
		2513	=item L<Time::HiRes>
		2514
		2515	This module is part of perl since release 5.008. It will be used when the
		2516	chosen event library does not come with a timing source on it's own. The
		2517	pure-perl event loop (L<AnyEvent::Impl::Perl>) will additionally use it to
		2518	try to use a monotonic clock for timing stability.
		2519
		2520	=back
		2521
		2522
		2523	=head1 FORK
		2524
		2525	Most event libraries are not fork-safe. The ones who are usually are
		2526	because they rely on inefficient but fork-safe C<select> or C<poll>
		2527	calls. Only L<EV> is fully fork-aware.
		2528
		2529	This means that, in general, you cannot fork and do event processing
		2530	in the child if a watcher was created before the fork (which in turn
		2531	initialises the event library).
		2532
		2533	If you have to fork, you must either do so I<before> creating your first
		2534	watcher OR you must not use AnyEvent at all in the child OR you must do
		2535	something completely out of the scope of AnyEvent.
		2536
		2537	The problem of doing event processing in the parent I<and> the child
		2538	is much more complicated: even for backends that I<are> fork-aware or
		2539	fork-safe, their behaviour is not usually what you want: fork clones all
		2540	watchers, that means all timers, I/O watchers etc. are active in both
		2541	parent and child, which is almost never what you want.
		2542
		2543
		2544	=head1 SECURITY CONSIDERATIONS
		2545
		2546	AnyEvent can be forced to load any event model via
		2547	$ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used to
		2548	execute arbitrary code or directly gain access, it can easily be used to
		2549	make the program hang or malfunction in subtle ways, as AnyEvent watchers
		2550	will not be active when the program uses a different event model than
		2551	specified in the variable.
		2552
		2553	You can make AnyEvent completely ignore this variable by deleting it
		2554	before the first watcher gets created, e.g. with a C<BEGIN> block:
		2555
		2556	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
		2557
		2558	use AnyEvent;
		2559
		2560	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		2561	be used to probe what backend is used and gain other information (which is
		2562	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		2563	$ENV{PERL_ANYEVENT_STRICT}.
		2564
		2565	Note that AnyEvent will remove I<all> environment variables starting with
		2566	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		2567	enabled.
		2568
		2569
		2570	=head1 BUGS
		2571
		2572	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		2573	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		2574	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		2575	memleaks, such as leaking on C<map> and C<grep> but it is usually not as
		2576	pronounced).
		2577
533		2578
534	=head1 SEE ALSO	2579	=head1 SEE ALSO
535		2580
536	Event modules: L<Coro::Event>, L<Coro>, L<Event>, L<Glib::Event>, L<Glib>.	2581	Utility functions: L<AnyEvent::Util>.
537		2582
538	Implementations: L<AnyEvent::Impl::Coro>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>.	2583	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
		2584	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
539		2585
540	Nontrivial usage example: L<Net::FCP>.	2586	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
		2587	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
		2588	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
		2589	L<AnyEvent::Impl::POE>, L<AnyEvent::Impl::IOAsync>, L<Anyevent::Impl::Irssi>.
541		2590
542	=head1	2591	Non-blocking file handles, sockets, TCP clients and
		2592	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>, L<AnyEvent::TLS>.
		2593
		2594	Asynchronous DNS: L<AnyEvent::DNS>.
		2595
		2596	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>,
		2597	L<Coro::Event>,
		2598
		2599	Nontrivial usage examples: L<AnyEvent::GPSD>, L<AnyEvent::XMPP>,
		2600	L<AnyEvent::HTTP>.
		2601
		2602
		2603	=head1 AUTHOR
		2604
		2605	Marc Lehmann <schmorp@schmorp.de>
		2606	http://home.schmorp.de/
543		2607
544	=cut	2608	=cut
545		2609
546	1	2610	1
547		2611

Diff Legend

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