ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.33 by root, Fri Nov 9 19:34:32 2007 UTC vs.
Revision 1.244 by root, Fri Jul 17 23:15:57 2009 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent - provide framework for multiple event loops	3	AnyEvent - provide framework for multiple event loops
4		4
5	Event, Coro, Glib, Tk, Perl - various supported event loops	5	EV, Event, Glib, Tk, Perl, Event::Lib, Qt and POE are various supported
		6	event loops.
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 WHY YOU SHOULD USE THIS MODULE (OR NOT)
		47
		48	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
		49	nowadays. So what is different about AnyEvent?
		50
		51	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
		52	policy> and AnyEvent is I<small and efficient>.
		53
		54	First and foremost, I<AnyEvent is not an event model> itself, it only
		55	interfaces to whatever event model the main program happens to use, in a
		56	pragmatic way. For event models and certain classes of immortals alike,
		57	the statement "there can only be one" is a bitter reality: In general,
		58	only one event loop can be active at the same time in a process. AnyEvent
		59	cannot change this, but it can hide the differences between those event
		60	loops.
		61
		62	The goal of AnyEvent is to offer module authors the ability to do event
		63	programming (waiting for I/O or timer events) without subscribing to a
		64	religion, a way of living, and most importantly: without forcing your
		65	module users into the same thing by forcing them to use the same event
		66	model you use.
		67
		68	For modules like POE or IO::Async (which is a total misnomer as it is
		69	actually doing all I/O I<synchronously>...), using them in your module is
		70	like joining a cult: After you joined, you are dependent on them and you
		71	cannot use anything else, as they are simply incompatible to everything
		72	that isn't them. What's worse, all the potential users of your
		73	module are I<also> forced to use the same event loop you use.
		74
		75	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
		76	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
		77	with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
		78	your module uses one of those, every user of your module has to use it,
		79	too. But if your module uses AnyEvent, it works transparently with all
		80	event models it supports (including stuff like IO::Async, as long as those
		81	use one of the supported event loops. It is trivial to add new event loops
		82	to AnyEvent, too, so it is future-proof).
		83
		84	In addition to being free of having to use I<the one and only true event
		85	model>, AnyEvent also is free of bloat and policy: with POE or similar
		86	modules, you get an enormous amount of code and strict rules you have to
		87	follow. AnyEvent, on the other hand, is lean and up to the point, by only
		88	offering the functionality that is necessary, in as thin as a wrapper as
		89	technically possible.
		90
		91	Of course, AnyEvent comes with a big (and fully optional!) toolbox
		92	of useful functionality, such as an asynchronous DNS resolver, 100%
		93	non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
		94	such as Windows) and lots of real-world knowledge and workarounds for
		95	platform bugs and differences.
		96
		97	Now, if you I<do want> lots of policy (this can arguably be somewhat
		98	useful) and you want to force your users to use the one and only event
		99	model, you should I<not> use this module.
22		100
23	=head1 DESCRIPTION	101	=head1 DESCRIPTION
24		102
25	L<AnyEvent> provides an identical interface to multiple event loops. This	103	L<AnyEvent> provides an identical interface to multiple event loops. This
26	allows module authors to utilise an event loop without forcing module	104	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	105	users to use the same event loop (as only a single event loop can coexist
28	peacefully at any one time).	106	peacefully at any one time).
29		107
30	The interface itself is vaguely similar but not identical to the Event	108	The interface itself is vaguely similar, but not identical to the L<Event>
31	module.	109	module.
32		110
33	On the first call of any method, the module tries to detect the currently	111	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	112	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	113	following modules is already loaded: L<EV>,
36	used. If none is found, the module tries to load these modules in the	114	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	115	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	116	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
39	event loop, which is also not very efficient.	117	adaptor should always succeed) in the order given. The first one that can
		118	be successfully loaded will be used. If, after this, still none could be
		119	found, AnyEvent will fall back to a pure-perl event loop, which is not
		120	very efficient, but should work everywhere.
40		121
41	Because AnyEvent first checks for modules that are already loaded, loading	122	Because AnyEvent first checks for modules that are already loaded, loading
42	an Event model explicitly before first using AnyEvent will likely make	123	an event model explicitly before first using AnyEvent will likely make
43	that model the default. For example:	124	that model the default. For example:
44		125
45	use Tk;	126	use Tk;
46	use AnyEvent;	127	use AnyEvent;
47		128
48	# .. AnyEvent will likely default to Tk	129	# .. AnyEvent will likely default to Tk
49		130
		131	The I<likely> means that, if any module loads another event model and
		132	starts using it, all bets are off. Maybe you should tell their authors to
		133	use AnyEvent so their modules work together with others seamlessly...
		134
50	The pure-perl implementation of AnyEvent is called	135	The pure-perl implementation of AnyEvent is called
51	C<AnyEvent::Impl::Perl>. Like other event modules you can load it	136	C<AnyEvent::Impl::Perl>. Like other event modules you can load it
52	explicitly.	137	explicitly and enjoy the high availability of that event loop :)
53		138
54	=head1 WATCHERS	139	=head1 WATCHERS
55		140
56	AnyEvent has the central concept of a I<watcher>, which is an object that	141	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	142	stores relevant data for each kind of event you are waiting for, such as
58	the callback to call, the filehandle to watch, etc.	143	the callback to call, the file handle to watch, etc.
59		144
60	These watchers are normal Perl objects with normal Perl lifetime. After	145	These watchers are normal Perl objects with normal Perl lifetime. After
61	creating a watcher it will immediately "watch" for events and invoke	146	creating a watcher it will immediately "watch" for events and invoke the
		147	callback when the event occurs (of course, only when the event model
		148	is in control).
		149
		150	Note that B<callbacks must not permanently change global variables>
		151	potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
		152	callbacks must not C<die> >>. The former is good programming practise in
		153	Perl and the latter stems from the fact that exception handling differs
		154	widely between event loops.
		155
62	the callback. To disable the watcher you have to destroy it (e.g. by	156	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	157	variable you store it in to C<undef> or otherwise deleting all references
64	references to it).	158	to it).
65		159
66	All watchers are created by calling a method on the C<AnyEvent> class.	160	All watchers are created by calling a method on the C<AnyEvent> class.
67		161
		162	Many watchers either are used with "recursion" (repeating timers for
		163	example), or need to refer to their watcher object in other ways.
		164
		165	An any way to achieve that is this pattern:
		166
		167	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
		168	# you can use $w here, for example to undef it
		169	undef $w;
		170	});
		171
		172	Note that C<my $w; $w => combination. This is necessary because in Perl,
		173	my variables are only visible after the statement in which they are
		174	declared.
		175
68	=head2 IO WATCHERS	176	=head2 I/O WATCHERS
69		177
70	You can create I/O watcher by calling the C<< AnyEvent->io >> method with	178	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
71	the following mandatory arguments:	179	with the following mandatory key-value pairs as arguments:
72		180
73	C<fh> the Perl I<filehandle> (not filedescriptor) to watch for	181	C<fh> is the Perl I<file handle> (or a naked file descriptor) to watch
		182	for events (AnyEvent might or might not keep a reference to this file
		183	handle). Note that only file handles pointing to things for which
		184	non-blocking operation makes sense are allowed. This includes sockets,
		185	most character devices, pipes, fifos and so on, but not for example files
		186	or block devices.
		187
74	events. C<poll> must be a string that is either C<r> or C<w>, that creates	188	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	189	watcher waiting for "r"eadable or "w"ritable events, respectively.
76	to invoke everytime the filehandle becomes ready.
77		190
78	Only one io watcher per C<fh> and C<poll> combination is allowed (i.e. on	191	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		192
82	Filehandles will be kept alive, so as long as the watcher exists, the	193	Although the callback might get passed parameters, their value and
83	filehandle exists, too.	194	presence is undefined and you cannot rely on them. Portable AnyEvent
		195	callbacks cannot use arguments passed to I/O watcher callbacks.
84		196
85	Example:	197	The I/O watcher might use the underlying file descriptor or a copy of it.
		198	You must not close a file handle as long as any watcher is active on the
		199	underlying file descriptor.
86		200
		201	Some event loops issue spurious readyness notifications, so you should
		202	always use non-blocking calls when reading/writing from/to your file
		203	handles.
		204
87	# wait for readability of STDIN, then read a line and disable the watcher	205	Example: wait for readability of STDIN, then read a line and disable the
		206	watcher.
		207
88	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	208	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
89	chomp (my $input = <STDIN>);	209	chomp (my $input = <STDIN>);
90	warn "read: $input\n";	210	warn "read: $input\n";
91	undef $w;	211	undef $w;
92	});	212	});
…		…
94	=head2 TIME WATCHERS	214	=head2 TIME WATCHERS
95		215
96	You can create a time watcher by calling the C<< AnyEvent->timer >>	216	You can create a time watcher by calling the C<< AnyEvent->timer >>
97	method with the following mandatory arguments:	217	method with the following mandatory arguments:
98		218
99	C<after> after how many seconds (fractions are supported) should the timer	219	C<after> specifies after how many seconds (fractional values are
100	activate. C<cb> the callback to invoke.	220	supported) the callback should be invoked. C<cb> is the callback to invoke
		221	in that case.
101		222
102	The timer callback will be invoked at most once: if you want a repeating	223	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	224	presence is undefined and you cannot rely on them. Portable AnyEvent
104	and Glib).	225	callbacks cannot use arguments passed to time watcher callbacks.
105		226
106	Example:	227	The callback will normally be invoked once only. If you specify another
		228	parameter, C<interval>, as a strictly positive number (> 0), then the
		229	callback will be invoked regularly at that interval (in fractional
		230	seconds) after the first invocation. If C<interval> is specified with a
		231	false value, then it is treated as if it were missing.
107		232
		233	The callback will be rescheduled before invoking the callback, but no
		234	attempt is done to avoid timer drift in most backends, so the interval is
		235	only approximate.
		236
108	# fire an event after 7.7 seconds	237	Example: fire an event after 7.7 seconds.
		238
109	my $w = AnyEvent->timer (after => 7.7, cb => sub {	239	my $w = AnyEvent->timer (after => 7.7, cb => sub {
110	warn "timeout\n";	240	warn "timeout\n";
111	});	241	});
112		242
113	# to cancel the timer:	243	# to cancel the timer:
114	undef $w	244	undef $w;
115		245
116	=head2 CONDITION WATCHERS	246	Example 2: fire an event after 0.5 seconds, then roughly every second.
117		247
118	Condition watchers can be created by calling the C<< AnyEvent->condvar >>	248	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
119	method without any arguments.	249	warn "timeout\n";
		250	};
120		251
121	A condition watcher watches for a condition - precisely that the C<<	252	=head3 TIMING ISSUES
122	->broadcast >> method has been called.
123		253
124	The watcher has only two methods:	254	There are two ways to handle timers: based on real time (relative, "fire
		255	in 10 seconds") and based on wallclock time (absolute, "fire at 12
		256	o'clock").
		257
		258	While most event loops expect timers to specified in a relative way, they
		259	use absolute time internally. This makes a difference when your clock
		260	"jumps", for example, when ntp decides to set your clock backwards from
		261	the wrong date of 2014-01-01 to 2008-01-01, a watcher that is supposed to
		262	fire "after" a second might actually take six years to finally fire.
		263
		264	AnyEvent cannot compensate for this. The only event loop that is conscious
		265	about these issues is L<EV>, which offers both relative (ev_timer, based
		266	on true relative time) and absolute (ev_periodic, based on wallclock time)
		267	timers.
		268
		269	AnyEvent always prefers relative timers, if available, matching the
		270	AnyEvent API.
		271
		272	AnyEvent has two additional methods that return the "current time":
125		273
126	=over 4	274	=over 4
127		275
128	=item $cv->wait	276	=item AnyEvent->time
129		277
130	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	278	This returns the "current wallclock time" as a fractional number of
131	called on c<$cv>, while servicing other watchers normally.	279	seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
		280	return, and the result is guaranteed to be compatible with those).
132		281
133	Not all event models support a blocking wait - some die in that case, so	282	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	283	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		284
140	You can only wait once on a condition - additional calls will return	285	=item AnyEvent->now
141	immediately.
142		286
143	=item $cv->broadcast	287	This also returns the "current wallclock time", but unlike C<time>, above,
		288	this value might change only once per event loop iteration, depending on
		289	the event loop (most return the same time as C<time>, above). This is the
		290	time that AnyEvent's timers get scheduled against.
144		291
145	Flag the condition as ready - a running C<< ->wait >> and all further	292	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	293	function to call when you want to know the current time.>
147	is waiting the broadcast will be remembered..
148		294
149	Example:	295	This function is also often faster then C<< AnyEvent->time >>, and
		296	thus the preferred method if you want some timestamp (for example,
		297	L<AnyEvent::Handle> uses this to update it's activity timeouts).
		298
		299	The rest of this section is only of relevance if you try to be very exact
		300	with your timing, you can skip it without bad conscience.
		301
		302	For a practical example of when these times differ, consider L<Event::Lib>
		303	and L<EV> and the following set-up:
		304
		305	The event loop is running and has just invoked one of your callback at
		306	time=500 (assume no other callbacks delay processing). In your callback,
		307	you wait a second by executing C<sleep 1> (blocking the process for a
		308	second) and then (at time=501) you create a relative timer that fires
		309	after three seconds.
		310
		311	With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
		312	both return C<501>, because that is the current time, and the timer will
		313	be scheduled to fire at time=504 (C<501> + C<3>).
		314
		315	With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
		316	time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
		317	last event processing phase started. With L<EV>, your timer gets scheduled
		318	to run at time=503 (C<500> + C<3>).
		319
		320	In one sense, L<Event::Lib> is more exact, as it uses the current time
		321	regardless of any delays introduced by event processing. However, most
		322	callbacks do not expect large delays in processing, so this causes a
		323	higher drift (and a lot more system calls to get the current time).
		324
		325	In another sense, L<EV> is more exact, as your timer will be scheduled at
		326	the same time, regardless of how long event processing actually took.
		327
		328	In either case, if you care (and in most cases, you don't), then you
		329	can get whatever behaviour you want with any event loop, by taking the
		330	difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
		331	account.
		332
		333	=item AnyEvent->now_update
		334
		335	Some event loops (such as L<EV> or L<AnyEvent::Impl::Perl>) cache
		336	the current time for each loop iteration (see the discussion of L<<
		337	AnyEvent->now >>, above).
		338
		339	When a callback runs for a long time (or when the process sleeps), then
		340	this "current" time will differ substantially from the real time, which
		341	might affect timers and time-outs.
		342
		343	When this is the case, you can call this method, which will update the
		344	event loop's idea of "current time".
		345
		346	Note that updating the time I<might> cause some events to be handled.
		347
		348	=back
		349
		350	=head2 SIGNAL WATCHERS
		351
		352	You can watch for signals using a signal watcher, C<signal> is the signal
		353	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
		354	callback to be invoked whenever a signal occurs.
		355
		356	Although the callback might get passed parameters, their value and
		357	presence is undefined and you cannot rely on them. Portable AnyEvent
		358	callbacks cannot use arguments passed to signal watcher callbacks.
		359
		360	Multiple signal occurrences can be clumped together into one callback
		361	invocation, and callback invocation will be synchronous. Synchronous means
		362	that it might take a while until the signal gets handled by the process,
		363	but it is guaranteed not to interrupt any other callbacks.
		364
		365	The main advantage of using these watchers is that you can share a signal
		366	between multiple watchers, and AnyEvent will ensure that signals will not
		367	interrupt your program at bad times.
		368
		369	This watcher might use C<%SIG> (depending on the event loop used),
		370	so programs overwriting those signals directly will likely not work
		371	correctly.
		372
		373	Also note that many event loops (e.g. Glib, Tk, Qt, IO::Async) do not
		374	support attaching callbacks to signals, which is a pity, as you cannot do
		375	race-free signal handling in perl. AnyEvent will try to do it's best, but
		376	in some cases, signals will be delayed. The maximum time a signal might
		377	be delayed is specified in C<$AnyEvent::MAX_SIGNAL_LATENCY> (default: 10
		378	seconds). This variable can be changed only before the first signal
		379	watcher is created, and should be left alone otherwise. Higher values
		380	will cause fewer spurious wake-ups, which is better for power and CPU
		381	saving. All these problems can be avoided by installing the optional
		382	L<Async::Interrupt> module.
		383
		384	Example: exit on SIGINT
		385
		386	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
		387
		388	=head2 CHILD PROCESS WATCHERS
		389
		390	You can also watch on a child process exit and catch its exit status.
		391
		392	The child process is specified by the C<pid> argument (if set to C<0>, it
		393	watches for any child process exit). The watcher will triggered only when
		394	the child process has finished and an exit status is available, not on
		395	any trace events (stopped/continued).
		396
		397	The callback will be called with the pid and exit status (as returned by
		398	waitpid), so unlike other watcher types, you I<can> rely on child watcher
		399	callback arguments.
		400
		401	This watcher type works by installing a signal handler for C<SIGCHLD>,
		402	and since it cannot be shared, nothing else should use SIGCHLD or reap
		403	random child processes (waiting for specific child processes, e.g. inside
		404	C<system>, is just fine).
		405
		406	There is a slight catch to child watchers, however: you usually start them
		407	I<after> the child process was created, and this means the process could
		408	have exited already (and no SIGCHLD will be sent anymore).
		409
		410	Not all event models handle this correctly (neither POE nor IO::Async do,
		411	see their AnyEvent::Impl manpages for details), but even for event models
		412	that I<do> handle this correctly, they usually need to be loaded before
		413	the process exits (i.e. before you fork in the first place). AnyEvent's
		414	pure perl event loop handles all cases correctly regardless of when you
		415	start the watcher.
		416
		417	This means you cannot create a child watcher as the very first
		418	thing in an AnyEvent program, you I<have> to create at least one
		419	watcher before you C<fork> the child (alternatively, you can call
		420	C<AnyEvent::detect>).
		421
		422	As most event loops do not support waiting for child events, they will be
		423	emulated by AnyEvent in most cases, in which the latency and race problems
		424	mentioned in the description of signal watchers apply.
		425
		426	Example: fork a process and wait for it
		427
		428	my $done = AnyEvent->condvar;
		429
		430	my $pid = fork or exit 5;
		431
		432	my $w = AnyEvent->child (
		433	pid => $pid,
		434	cb => sub {
		435	my ($pid, $status) = @_;
		436	warn "pid $pid exited with status $status";
		437	$done->send;
		438	},
		439	);
		440
		441	# do something else, then wait for process exit
		442	$done->recv;
		443
		444	=head2 IDLE WATCHERS
		445
		446	Sometimes there is a need to do something, but it is not so important
		447	to do it instantly, but only when there is nothing better to do. This
		448	"nothing better to do" is usually defined to be "no other events need
		449	attention by the event loop".
		450
		451	Idle watchers ideally get invoked when the event loop has nothing
		452	better to do, just before it would block the process to wait for new
		453	events. Instead of blocking, the idle watcher is invoked.
		454
		455	Most event loops unfortunately do not really support idle watchers (only
		456	EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
		457	will simply call the callback "from time to time".
		458
		459	Example: read lines from STDIN, but only process them when the
		460	program is otherwise idle:
		461
		462	my @lines; # read data
		463	my $idle_w;
		464	my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
		465	push @lines, scalar <STDIN>;
		466
		467	# start an idle watcher, if not already done
		468	$idle_w ||= AnyEvent->idle (cb => sub {
		469	# handle only one line, when there are lines left
		470	if (my $line = shift @lines) {
		471	print "handled when idle: $line";
		472	} else {
		473	# otherwise disable the idle watcher again
		474	undef $idle_w;
		475	}
		476	});
		477	});
		478
		479	=head2 CONDITION VARIABLES
		480
		481	If you are familiar with some event loops you will know that all of them
		482	require you to run some blocking "loop", "run" or similar function that
		483	will actively watch for new events and call your callbacks.
		484
		485	AnyEvent is slightly different: it expects somebody else to run the event
		486	loop and will only block when necessary (usually when told by the user).
		487
		488	The instrument to do that is called a "condition variable", so called
		489	because they represent a condition that must become true.
		490
		491	Now is probably a good time to look at the examples further below.
		492
		493	Condition variables can be created by calling the C<< AnyEvent->condvar
		494	>> method, usually without arguments. The only argument pair allowed is
		495	C<cb>, which specifies a callback to be called when the condition variable
		496	becomes true, with the condition variable as the first argument (but not
		497	the results).
		498
		499	After creation, the condition variable is "false" until it becomes "true"
		500	by calling the C<send> method (or calling the condition variable as if it
		501	were a callback, read about the caveats in the description for the C<<
		502	->send >> method).
		503
		504	Condition variables are similar to callbacks, except that you can
		505	optionally wait for them. They can also be called merge points - points
		506	in time where multiple outstanding events have been processed. And yet
		507	another way to call them is transactions - each condition variable can be
		508	used to represent a transaction, which finishes at some point and delivers
		509	a result.
		510
		511	Condition variables are very useful to signal that something has finished,
		512	for example, if you write a module that does asynchronous http requests,
		513	then a condition variable would be the ideal candidate to signal the
		514	availability of results. The user can either act when the callback is
		515	called or can synchronously C<< ->recv >> for the results.
		516
		517	You can also use them to simulate traditional event loops - for example,
		518	you can block your main program until an event occurs - for example, you
		519	could C<< ->recv >> in your main program until the user clicks the Quit
		520	button of your app, which would C<< ->send >> the "quit" event.
		521
		522	Note that condition variables recurse into the event loop - if you have
		523	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
		524	lose. Therefore, condition variables are good to export to your caller, but
		525	you should avoid making a blocking wait yourself, at least in callbacks,
		526	as this asks for trouble.
		527
		528	Condition variables are represented by hash refs in perl, and the keys
		529	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		530	easy (it is often useful to build your own transaction class on top of
		531	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		532	it's C<new> method in your own C<new> method.
		533
		534	There are two "sides" to a condition variable - the "producer side" which
		535	eventually calls C<< -> send >>, and the "consumer side", which waits
		536	for the send to occur.
		537
		538	Example: wait for a timer.
150		539
151	# wait till the result is ready	540	# wait till the result is ready
152	my $result_ready = AnyEvent->condvar;	541	my $result_ready = AnyEvent->condvar;
153		542
154	# do something such as adding a timer	543	# do something such as adding a timer
155	# or socket watcher the calls $result_ready->broadcast	544	# or socket watcher the calls $result_ready->send
156	# when the "result" is ready.	545	# when the "result" is ready.
		546	# in this case, we simply use a timer:
		547	my $w = AnyEvent->timer (
		548	after => 1,
		549	cb => sub { $result_ready->send },
		550	);
157		551
		552	# this "blocks" (while handling events) till the callback
		553	# calls -<send
158	$result_ready->wait;	554	$result_ready->recv;
		555
		556	Example: wait for a timer, but take advantage of the fact that condition
		557	variables are also callable directly.
		558
		559	my $done = AnyEvent->condvar;
		560	my $delay = AnyEvent->timer (after => 5, cb => $done);
		561	$done->recv;
		562
		563	Example: Imagine an API that returns a condvar and doesn't support
		564	callbacks. This is how you make a synchronous call, for example from
		565	the main program:
		566
		567	use AnyEvent::CouchDB;
		568
		569	...
		570
		571	my @info = $couchdb->info->recv;
		572
		573	And this is how you would just set a callback to be called whenever the
		574	results are available:
		575
		576	$couchdb->info->cb (sub {
		577	my @info = $_[0]->recv;
		578	});
		579
		580	=head3 METHODS FOR PRODUCERS
		581
		582	These methods should only be used by the producing side, i.e. the
		583	code/module that eventually sends the signal. Note that it is also
		584	the producer side which creates the condvar in most cases, but it isn't
		585	uncommon for the consumer to create it as well.
		586
		587	=over 4
		588
		589	=item $cv->send (...)
		590
		591	Flag the condition as ready - a running C<< ->recv >> and all further
		592	calls to C<recv> will (eventually) return after this method has been
		593	called. If nobody is waiting the send will be remembered.
		594
		595	If a callback has been set on the condition variable, it is called
		596	immediately from within send.
		597
		598	Any arguments passed to the C<send> call will be returned by all
		599	future C<< ->recv >> calls.
		600
		601	Condition variables are overloaded so one can call them directly (as if
		602	they were a code reference). Calling them directly is the same as calling
		603	C<send>.
		604
		605	=item $cv->croak ($error)
		606
		607	Similar to send, but causes all call's to C<< ->recv >> to invoke
		608	C<Carp::croak> with the given error message/object/scalar.
		609
		610	This can be used to signal any errors to the condition variable
		611	user/consumer. Doing it this way instead of calling C<croak> directly
		612	delays the error detetcion, but has the overwhelmign advantage that it
		613	diagnoses the error at the place where the result is expected, and not
		614	deep in some event clalback without connection to the actual code causing
		615	the problem.
		616
		617	=item $cv->begin ([group callback])
		618
		619	=item $cv->end
		620
		621	These two methods can be used to combine many transactions/events into
		622	one. For example, a function that pings many hosts in parallel might want
		623	to use a condition variable for the whole process.
		624
		625	Every call to C<< ->begin >> will increment a counter, and every call to
		626	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		627	>>, the (last) callback passed to C<begin> will be executed. That callback
		628	is I<supposed> to call C<< ->send >>, but that is not required. If no
		629	callback was set, C<send> will be called without any arguments.
		630
		631	You can think of C<< $cv->send >> giving you an OR condition (one call
		632	sends), while C<< $cv->begin >> and C<< $cv->end >> giving you an AND
		633	condition (all C<begin> calls must be C<end>'ed before the condvar sends).
		634
		635	Let's start with a simple example: you have two I/O watchers (for example,
		636	STDOUT and STDERR for a program), and you want to wait for both streams to
		637	close before activating a condvar:
		638
		639	my $cv = AnyEvent->condvar;
		640
		641	$cv->begin; # first watcher
		642	my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
		643	defined sysread $fh1, my $buf, 4096
		644	or $cv->end;
		645	});
		646
		647	$cv->begin; # second watcher
		648	my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
		649	defined sysread $fh2, my $buf, 4096
		650	or $cv->end;
		651	});
		652
		653	$cv->recv;
		654
		655	This works because for every event source (EOF on file handle), there is
		656	one call to C<begin>, so the condvar waits for all calls to C<end> before
		657	sending.
		658
		659	The ping example mentioned above is slightly more complicated, as the
		660	there are results to be passwd back, and the number of tasks that are
		661	begung can potentially be zero:
		662
		663	my $cv = AnyEvent->condvar;
		664
		665	my %result;
		666	$cv->begin (sub { $cv->send (\%result) });
		667
		668	for my $host (@list_of_hosts) {
		669	$cv->begin;
		670	ping_host_then_call_callback $host, sub {
		671	$result{$host} = ...;
		672	$cv->end;
		673	};
		674	}
		675
		676	$cv->end;
		677
		678	This code fragment supposedly pings a number of hosts and calls
		679	C<send> after results for all then have have been gathered - in any
		680	order. To achieve this, the code issues a call to C<begin> when it starts
		681	each ping request and calls C<end> when it has received some result for
		682	it. Since C<begin> and C<end> only maintain a counter, the order in which
		683	results arrive is not relevant.
		684
		685	There is an additional bracketing call to C<begin> and C<end> outside the
		686	loop, which serves two important purposes: first, it sets the callback
		687	to be called once the counter reaches C<0>, and second, it ensures that
		688	C<send> is called even when C<no> hosts are being pinged (the loop
		689	doesn't execute once).
		690
		691	This is the general pattern when you "fan out" into multiple (but
		692	potentially none) subrequests: use an outer C<begin>/C<end> pair to set
		693	the callback and ensure C<end> is called at least once, and then, for each
		694	subrequest you start, call C<begin> and for each subrequest you finish,
		695	call C<end>.
159		696
160	=back	697	=back
161		698
162	=head2 SIGNAL WATCHERS	699	=head3 METHODS FOR CONSUMERS
163		700
164	You can listen for signals using a signal watcher, C<signal> is the signal	701	These methods should only be used by the consuming side, i.e. the
165	I<name> without any C<SIG> prefix. Multiple signals events can be clumped	702	code awaits the condition.
166	together into one callback invocation, and callback invocation might or
167	might not be asynchronous.
168
169	These watchers might use C<%SIG>, so programs overwriting those signals
170	directly will likely not work correctly.
171
172	Example: exit on SIGINT
173
174	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
175
176	=head2 CHILD PROCESS WATCHERS
177
178	You can also listen for the status of a child process specified by the
179	C<pid> argument (or any child if the pid argument is 0). The watcher will
180	trigger as often as status change for the child are received. This works
181	by installing a signal handler for C<SIGCHLD>. The callback will be called with
182	the pid and exit status (as returned by waitpid).
183
184	Example: wait for pid 1333
185
186	my $w = AnyEvent->child (pid => 1333, cb => sub { warn "exit status $?" });
187
188	=head1 GLOBALS
189		703
190	=over 4	704	=over 4
191		705
		706	=item $cv->recv
		707
		708	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
		709	>> methods have been called on c<$cv>, while servicing other watchers
		710	normally.
		711
		712	You can only wait once on a condition - additional calls are valid but
		713	will return immediately.
		714
		715	If an error condition has been set by calling C<< ->croak >>, then this
		716	function will call C<croak>.
		717
		718	In list context, all parameters passed to C<send> will be returned,
		719	in scalar context only the first one will be returned.
		720
		721	Note that doing a blocking wait in a callback is not supported by any
		722	event loop, that is, recursive invocation of a blocking C<< ->recv
		723	>> is not allowed, and the C<recv> call will C<croak> if such a
		724	condition is detected. This condition can be slightly loosened by using
		725	L<Coro::AnyEvent>, which allows you to do a blocking C<< ->recv >> from
		726	any thread that doesn't run the event loop itself.
		727
		728	Not all event models support a blocking wait - some die in that case
		729	(programs might want to do that to stay interactive), so I<if you are
		730	using this from a module, never require a blocking wait>. Instead, let the
		731	caller decide whether the call will block or not (for example, by coupling
		732	condition variables with some kind of request results and supporting
		733	callbacks so the caller knows that getting the result will not block,
		734	while still supporting blocking waits if the caller so desires).
		735
		736	You can ensure that C<< -recv >> never blocks by setting a callback and
		737	only calling C<< ->recv >> from within that callback (or at a later
		738	time). This will work even when the event loop does not support blocking
		739	waits otherwise.
		740
		741	=item $bool = $cv->ready
		742
		743	Returns true when the condition is "true", i.e. whether C<send> or
		744	C<croak> have been called.
		745
		746	=item $cb = $cv->cb ($cb->($cv))
		747
		748	This is a mutator function that returns the callback set and optionally
		749	replaces it before doing so.
		750
		751	The callback will be called when the condition becomes "true", i.e. when
		752	C<send> or C<croak> are called, with the only argument being the condition
		753	variable itself. Calling C<recv> inside the callback or at any later time
		754	is guaranteed not to block.
		755
		756	=back
		757
		758	=head1 SUPPORTED EVENT LOOPS/BACKENDS
		759
		760	The available backend classes are (every class has its own manpage):
		761
		762	=over 4
		763
		764	=item Backends that are autoprobed when no other event loop can be found.
		765
		766	EV is the preferred backend when no other event loop seems to be in
		767	use. If EV is not installed, then AnyEvent will try Event, and, failing
		768	that, will fall back to its own pure-perl implementation, which is
		769	available everywhere as it comes with AnyEvent itself.
		770
		771	AnyEvent::Impl::EV based on EV (interface to libev, best choice).
		772	AnyEvent::Impl::Event based on Event, very stable, few glitches.
		773	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
		774
		775	=item Backends that are transparently being picked up when they are used.
		776
		777	These will be used when they are currently loaded when the first watcher
		778	is created, in which case it is assumed that the application is using
		779	them. This means that AnyEvent will automatically pick the right backend
		780	when the main program loads an event module before anything starts to
		781	create watchers. Nothing special needs to be done by the main program.
		782
		783	AnyEvent::Impl::Glib based on Glib, slow but very stable.
		784	AnyEvent::Impl::Tk based on Tk, very broken.
		785	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
		786	AnyEvent::Impl::POE based on POE, very slow, some limitations.
		787
		788	=item Backends with special needs.
		789
		790	Qt requires the Qt::Application to be instantiated first, but will
		791	otherwise be picked up automatically. As long as the main program
		792	instantiates the application before any AnyEvent watchers are created,
		793	everything should just work.
		794
		795	AnyEvent::Impl::Qt based on Qt.
		796
		797	Support for IO::Async can only be partial, as it is too broken and
		798	architecturally limited to even support the AnyEvent API. It also
		799	is the only event loop that needs the loop to be set explicitly, so
		800	it can only be used by a main program knowing about AnyEvent. See
		801	L<AnyEvent::Impl::Async> for the gory details.
		802
		803	AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
		804
		805	=item Event loops that are indirectly supported via other backends.
		806
		807	Some event loops can be supported via other modules:
		808
		809	There is no direct support for WxWidgets (L<Wx>) or L<Prima>.
		810
		811	B<WxWidgets> has no support for watching file handles. However, you can
		812	use WxWidgets through the POE adaptor, as POE has a Wx backend that simply
		813	polls 20 times per second, which was considered to be too horrible to even
		814	consider for AnyEvent.
		815
		816	B<Prima> is not supported as nobody seems to be using it, but it has a POE
		817	backend, so it can be supported through POE.
		818
		819	AnyEvent knows about both L<Prima> and L<Wx>, however, and will try to
		820	load L<POE> when detecting them, in the hope that POE will pick them up,
		821	in which case everything will be automatic.
		822
		823	=back
		824
		825	=head1 GLOBAL VARIABLES AND FUNCTIONS
		826
		827	These are not normally required to use AnyEvent, but can be useful to
		828	write AnyEvent extension modules.
		829
		830	=over 4
		831
192	=item $AnyEvent::MODEL	832	=item $AnyEvent::MODEL
193		833
194	Contains C<undef> until the first watcher is being created. Then it	834	Contains C<undef> until the first watcher is being created, before the
		835	backend has been autodetected.
		836
195	contains the event model that is being used, which is the name of the	837	Afterwards it contains the event model that is being used, which is the
196	Perl class implementing the model. This class is usually one of the	838	name of the Perl class implementing the model. This class is usually one
197	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	839	of the C<AnyEvent::Impl:xxx> modules, but can be any other class in the
198	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	840	case AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode> it
199		841	will be C<urxvt::anyevent>).
200	The known classes so far are:
201
202	AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
203	AnyEvent::Impl::EV based on EV (an interface to libev, also best choice).
204	AnyEvent::Impl::Coro based on Coro::Event, second best choice.
205	AnyEvent::Impl::Event based on Event, also second best choice :)
206	AnyEvent::Impl::Glib based on Glib, second-best choice.
207	AnyEvent::Impl::Tk based on Tk, very bad choice.
208	AnyEvent::Impl::Perl pure-perl implementation, inefficient.
209		842
210	=item AnyEvent::detect	843	=item AnyEvent::detect
211		844
212	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model if	845	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
213	necessary. You should only call this function right before you would have	846	if necessary. You should only call this function right before you would
214	created an AnyEvent watcher anyway, that is, very late at runtime.	847	have created an AnyEvent watcher anyway, that is, as late as possible at
		848	runtime, and not e.g. while initialising of your module.
		849
		850	If you need to do some initialisation before AnyEvent watchers are
		851	created, use C<post_detect>.
		852
		853	=item $guard = AnyEvent::post_detect { BLOCK }
		854
		855	Arranges for the code block to be executed as soon as the event model is
		856	autodetected (or immediately if this has already happened).
		857
		858	The block will be executed I<after> the actual backend has been detected
		859	(C<$AnyEvent::MODEL> is set), but I<before> any watchers have been
		860	created, so it is possible to e.g. patch C<@AnyEvent::ISA> or do
		861	other initialisations - see the sources of L<AnyEvent::Strict> or
		862	L<AnyEvent::AIO> to see how this is used.
		863
		864	The most common usage is to create some global watchers, without forcing
		865	event module detection too early, for example, L<AnyEvent::AIO> creates
		866	and installs the global L<IO::AIO> watcher in a C<post_detect> block to
		867	avoid autodetecting the event module at load time.
		868
		869	If called in scalar or list context, then it creates and returns an object
		870	that automatically removes the callback again when it is destroyed. See
		871	L<Coro::BDB> for a case where this is useful.
		872
		873	=item @AnyEvent::post_detect
		874
		875	If there are any code references in this array (you can C<push> to it
		876	before or after loading AnyEvent), then they will called directly after
		877	the event loop has been chosen.
		878
		879	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		880	if it is defined then the event loop has already been detected, and the
		881	array will be ignored.
		882
		883	Best use C<AnyEvent::post_detect { BLOCK }> when your application allows
		884	it,as it takes care of these details.
		885
		886	This variable is mainly useful for modules that can do something useful
		887	when AnyEvent is used and thus want to know when it is initialised, but do
		888	not need to even load it by default. This array provides the means to hook
		889	into AnyEvent passively, without loading it.
215		890
216	=back	891	=back
217		892
218	=head1 WHAT TO DO IN A MODULE	893	=head1 WHAT TO DO IN A MODULE
219		894
220	As a module author, you should "use AnyEvent" and call AnyEvent methods	895	As a module author, you should C<use AnyEvent> and call AnyEvent methods
221	freely, but you should not load a specific event module or rely on it.	896	freely, but you should not load a specific event module or rely on it.
222		897
223	Be careful when you create watchers in the module body - Anyevent will	898	Be careful when you create watchers in the module body - AnyEvent will
224	decide which event module to use as soon as the first method is called, so	899	decide which event module to use as soon as the first method is called, so
225	by calling AnyEvent in your module body you force the user of your module	900	by calling AnyEvent in your module body you force the user of your module
226	to load the event module first.	901	to load the event module first.
227		902
		903	Never call C<< ->recv >> on a condition variable unless you I<know> that
		904	the C<< ->send >> method has been called on it already. This is
		905	because it will stall the whole program, and the whole point of using
		906	events is to stay interactive.
		907
		908	It is fine, however, to call C<< ->recv >> when the user of your module
		909	requests it (i.e. if you create a http request object ad have a method
		910	called C<results> that returns the results, it should call C<< ->recv >>
		911	freely, as the user of your module knows what she is doing. always).
		912
228	=head1 WHAT TO DO IN THE MAIN PROGRAM	913	=head1 WHAT TO DO IN THE MAIN PROGRAM
229		914
230	There will always be a single main program - the only place that should	915	There will always be a single main program - the only place that should
231	dictate which event model to use.	916	dictate which event model to use.
232		917
233	If it doesn't care, it can just "use AnyEvent" and use it itself, or not	918	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
234	do anything special and let AnyEvent decide which implementation to chose.	919	do anything special (it does not need to be event-based) and let AnyEvent
		920	decide which implementation to chose if some module relies on it.
235		921
236	If the main program relies on a specific event model (for example, in Gtk2	922	If the main program relies on a specific event model - for example, in
237	programs you have to rely on either Glib or Glib::Event), you should load	923	Gtk2 programs you have to rely on the Glib module - you should load the
238	it before loading AnyEvent or any module that uses it, generally, as early	924	event module before loading AnyEvent or any module that uses it: generally
239	as possible. The reason is that modules might create watchers when they	925	speaking, you should load it as early as possible. The reason is that
240	are loaded, and AnyEvent will decide on the event model to use as soon as	926	modules might create watchers when they are loaded, and AnyEvent will
241	it creates watchers, and it might chose the wrong one unless you load the	927	decide on the event model to use as soon as it creates watchers, and it
242	correct one yourself.	928	might chose the wrong one unless you load the correct one yourself.
243		929
244	You can chose to use a rather inefficient pure-perl implementation by	930	You can chose to use a pure-perl implementation by loading the
245	loading the C<AnyEvent::Impl::Perl> module, but letting AnyEvent chose is	931	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
246	generally better.	932	everywhere, but letting AnyEvent chose the model is generally better.
		933
		934	=head2 MAINLOOP EMULATION
		935
		936	Sometimes (often for short test scripts, or even standalone programs who
		937	only want to use AnyEvent), you do not want to run a specific event loop.
		938
		939	In that case, you can use a condition variable like this:
		940
		941	AnyEvent->condvar->recv;
		942
		943	This has the effect of entering the event loop and looping forever.
		944
		945	Note that usually your program has some exit condition, in which case
		946	it is better to use the "traditional" approach of storing a condition
		947	variable somewhere, waiting for it, and sending it when the program should
		948	exit cleanly.
		949
		950
		951	=head1 OTHER MODULES
		952
		953	The following is a non-exhaustive list of additional modules that use
		954	AnyEvent as a client and can therefore be mixed easily with other AnyEvent
		955	modules and other event loops in the same program. Some of the modules
		956	come with AnyEvent, most are available via CPAN.
		957
		958	=over 4
		959
		960	=item L<AnyEvent::Util>
		961
		962	Contains various utility functions that replace often-used but blocking
		963	functions such as C<inet_aton> by event-/callback-based versions.
		964
		965	=item L<AnyEvent::Socket>
		966
		967	Provides various utility functions for (internet protocol) sockets,
		968	addresses and name resolution. Also functions to create non-blocking tcp
		969	connections or tcp servers, with IPv6 and SRV record support and more.
		970
		971	=item L<AnyEvent::Handle>
		972
		973	Provide read and write buffers, manages watchers for reads and writes,
		974	supports raw and formatted I/O, I/O queued and fully transparent and
		975	non-blocking SSL/TLS (via L<AnyEvent::TLS>.
		976
		977	=item L<AnyEvent::DNS>
		978
		979	Provides rich asynchronous DNS resolver capabilities.
		980
		981	=item L<AnyEvent::HTTP>
		982
		983	A simple-to-use HTTP library that is capable of making a lot of concurrent
		984	HTTP requests.
		985
		986	=item L<AnyEvent::HTTPD>
		987
		988	Provides a simple web application server framework.
		989
		990	=item L<AnyEvent::FastPing>
		991
		992	The fastest ping in the west.
		993
		994	=item L<AnyEvent::DBI>
		995
		996	Executes L<DBI> requests asynchronously in a proxy process.
		997
		998	=item L<AnyEvent::AIO>
		999
		1000	Truly asynchronous I/O, should be in the toolbox of every event
		1001	programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
		1002	together.
		1003
		1004	=item L<AnyEvent::BDB>
		1005
		1006	Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
		1007	L<BDB> and AnyEvent together.
		1008
		1009	=item L<AnyEvent::GPSD>
		1010
		1011	A non-blocking interface to gpsd, a daemon delivering GPS information.
		1012
		1013	=item L<AnyEvent::IRC>
		1014
		1015	AnyEvent based IRC client module family (replacing the older Net::IRC3).
		1016
		1017	=item L<AnyEvent::XMPP>
		1018
		1019	AnyEvent based XMPP (Jabber protocol) module family (replacing the older
		1020	Net::XMPP2>.
		1021
		1022	=item L<AnyEvent::IGS>
		1023
		1024	A non-blocking interface to the Internet Go Server protocol (used by
		1025	L<App::IGS>).
		1026
		1027	=item L<Net::FCP>
		1028
		1029	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		1030	of AnyEvent.
		1031
		1032	=item L<Event::ExecFlow>
		1033
		1034	High level API for event-based execution flow control.
		1035
		1036	=item L<Coro>
		1037
		1038	Has special support for AnyEvent via L<Coro::AnyEvent>.
		1039
		1040	=back
247		1041
248	=cut	1042	=cut
249		1043
250	package AnyEvent;	1044	package AnyEvent;
251		1045
		1046	# basically a tuned-down version of common::sense
		1047	sub common_sense {
252	no warnings;	1048	# no warnings
253	use strict;	1049	${^WARNING_BITS} ^= ${^WARNING_BITS};
		1050	# use strict vars subs
		1051	$^H |= 0x00000600;
		1052	}
254		1053
		1054	BEGIN { AnyEvent::common_sense }
		1055
255	use Carp;	1056	use Carp ();
256		1057
257	our $VERSION = '2.55';	1058	our $VERSION = 4.83;
258	our $MODEL;	1059	our $MODEL;
259		1060
260	our $AUTOLOAD;	1061	our $AUTOLOAD;
261	our @ISA;	1062	our @ISA;
262		1063
263	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
264
265	our @REGISTRY;	1064	our @REGISTRY;
266		1065
		1066	our $WIN32;
		1067
		1068	our $VERBOSE;
		1069
		1070	BEGIN {
		1071	eval "sub WIN32(){ " . (($^O =~ /mswin32/i)*1) ." }";
		1072	eval "sub TAINT(){ " . (${^TAINT}*1) . " }";
		1073
		1074	delete @ENV{grep /^PERL_ANYEVENT_/, keys %ENV}
		1075	if ${^TAINT};
		1076
		1077	$VERBOSE = $ENV{PERL_ANYEVENT_VERBOSE}*1;
		1078
		1079	}
		1080
		1081	our $MAX_SIGNAL_LATENCY = 10;
		1082
		1083	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		1084
		1085	{
		1086	my $idx;
		1087	$PROTOCOL{$_} = ++$idx
		1088	for reverse split /\s*,\s*/,
		1089	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		1090	}
		1091
267	my @models = (	1092	my @models = (
268	[Coro::EV:: => AnyEvent::Impl::CoroEV::],
269	[EV:: => AnyEvent::Impl::EV::],	1093	[EV:: => AnyEvent::Impl::EV::],
270	[Coro::Event:: => AnyEvent::Impl::Coro::],
271	[Event:: => AnyEvent::Impl::Event::],	1094	[Event:: => AnyEvent::Impl::Event::],
272	[Glib:: => AnyEvent::Impl::Glib::],
273	[Tk:: => AnyEvent::Impl::Tk::],
274	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	1095	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
		1096	# everything below here will not be autoprobed
		1097	# as the pureperl backend should work everywhere
		1098	# and is usually faster
		1099	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
		1100	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
		1101	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		1102	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
		1103	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
		1104	[Wx:: => AnyEvent::Impl::POE::],
		1105	[Prima:: => AnyEvent::Impl::POE::],
		1106	# IO::Async is just too broken - we would need workarounds for its
		1107	# byzantine signal and broken child handling, among others.
		1108	# IO::Async is rather hard to detect, as it doesn't have any
		1109	# obvious default class.
		1110	# [IO::Async:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1111	# [IO::Async::Loop:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1112	# [IO::Async::Notifier:: => AnyEvent::Impl::IOAsync::], # requires special main program
275	);	1113	);
276		1114
277	our %method = map +($_ => 1), qw(io timer condvar broadcast wait signal one_event DESTROY);	1115	our %method = map +($_ => 1),
		1116	qw(io timer time now now_update signal child idle condvar one_event DESTROY);
		1117
		1118	our @post_detect;
		1119
		1120	sub post_detect(&) {
		1121	my ($cb) = @_;
		1122
		1123	if ($MODEL) {
		1124	$cb->();
		1125
		1126	1
		1127	} else {
		1128	push @post_detect, $cb;
		1129
		1130	defined wantarray
		1131	? bless \$cb, "AnyEvent::Util::postdetect"
		1132	: ()
		1133	}
		1134	}
		1135
		1136	sub AnyEvent::Util::postdetect::DESTROY {
		1137	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		1138	}
278		1139
279	sub detect() {	1140	sub detect() {
280	unless ($MODEL) {	1141	unless ($MODEL) {
281	no strict 'refs';	1142	local $SIG{__DIE__};
		1143
		1144	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
		1145	my $model = "AnyEvent::Impl::$1";
		1146	if (eval "require $model") {
		1147	$MODEL = $model;
		1148	warn "AnyEvent: loaded model '$model' (forced by \$ENV{PERL_ANYEVENT_MODEL}), using it.\n" if $VERBOSE >= 2;
		1149	} else {
		1150	warn "AnyEvent: unable to load model '$model' (from \$ENV{PERL_ANYEVENT_MODEL}):\n$@" if $VERBOSE;
		1151	}
		1152	}
282		1153
283	# check for already loaded models	1154	# check for already loaded models
		1155	unless ($MODEL) {
284	for (@REGISTRY, @models) {	1156	for (@REGISTRY, @models) {
285	my ($package, $model) = @$_;	1157	my ($package, $model) = @$_;
286	if (${"$package\::VERSION"} > 0) {	1158	if (${"$package\::VERSION"} > 0) {
287	if (eval "require $model") {	1159	if (eval "require $model") {
288	$MODEL = $model;	1160	$MODEL = $model;
289	warn "AnyEvent: found model '$model', using it.\n" if $verbose > 1;	1161	warn "AnyEvent: autodetected model '$model', using it.\n" if $VERBOSE >= 2;
290	last;	1162	last;
		1163	}
291	}	1164	}
292	}	1165	}
		1166
		1167	unless ($MODEL) {
		1168	# try to load a model
		1169
		1170	for (@REGISTRY, @models) {
		1171	my ($package, $model) = @$_;
		1172	if (eval "require $package"
		1173	and ${"$package\::VERSION"} > 0
		1174	and eval "require $model") {
		1175	$MODEL = $model;
		1176	warn "AnyEvent: autoprobed model '$model', using it.\n" if $VERBOSE >= 2;
		1177	last;
		1178	}
		1179	}
		1180
		1181	$MODEL
		1182	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.\n";
		1183	}
293	}	1184	}
294		1185
295	unless ($MODEL) {	1186	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
296	# try to load a model
297
298	for (@REGISTRY, @models) {
299	my ($package, $model) = @$_;
300	if (eval "require $package"
301	and ${"$package\::VERSION"} > 0
302	and eval "require $model") {
303	$MODEL = $model;
304	warn "AnyEvent: autoprobed and loaded model '$model', using it.\n" if $verbose > 1;
305	last;
306	}
307	}
308
309	$MODEL
310	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: Event (or Coro+Event), Glib or Tk.";
311	}
312		1187
313	unshift @ISA, $MODEL;	1188	unshift @ISA, $MODEL;
314	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	1189
		1190	require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
		1191
		1192	(shift @post_detect)->() while @post_detect;
315	}	1193	}
316		1194
317	$MODEL	1195	$MODEL
318	}	1196	}
319		1197
320	sub AUTOLOAD {	1198	sub AUTOLOAD {
321	(my $func = $AUTOLOAD) =~ s/.*://;	1199	(my $func = $AUTOLOAD) =~ s/.*://;
322		1200
323	$method{$func}	1201	$method{$func}
324	or croak "$func: not a valid method for AnyEvent objects";	1202	or Carp::croak "$func: not a valid method for AnyEvent objects";
325		1203
326	detect unless $MODEL;	1204	detect unless $MODEL;
327		1205
328	my $class = shift;	1206	my $class = shift;
329	$class->$func (@_);	1207	$class->$func (@_);
330	}	1208	}
331		1209
		1210	# utility function to dup a filehandle. this is used by many backends
		1211	# to support binding more than one watcher per filehandle (they usually
		1212	# allow only one watcher per fd, so we dup it to get a different one).
		1213	sub _dupfh($$;$$) {
		1214	my ($poll, $fh, $r, $w) = @_;
		1215
		1216	# cygwin requires the fh mode to be matching, unix doesn't
		1217	my ($rw, $mode) = $poll eq "r" ? ($r, "<&") : ($w, ">&");
		1218
		1219	open my $fh2, $mode, $fh
		1220	or die "AnyEvent->io: cannot dup() filehandle in mode '$poll': $!,";
		1221
		1222	# we assume CLOEXEC is already set by perl in all important cases
		1223
		1224	($fh2, $rw)
		1225	}
		1226
332	package AnyEvent::Base;	1227	package AnyEvent::Base;
333		1228
		1229	# default implementations for many methods
		1230
		1231	sub _time {
		1232	# probe for availability of Time::HiRes
		1233	if (eval "use Time::HiRes (); Time::HiRes::time (); 1") {
		1234	warn "AnyEvent: using Time::HiRes for sub-second timing accuracy.\n" if $VERBOSE >= 8;
		1235	*_time = \&Time::HiRes::time;
		1236	# if (eval "use POSIX (); (POSIX::times())...
		1237	} else {
		1238	warn "AnyEvent: using built-in time(), WARNING, no sub-second resolution!\n" if $VERBOSE;
		1239	*_time = sub { time }; # epic fail
		1240	}
		1241
		1242	&_time
		1243	}
		1244
		1245	sub time { _time }
		1246	sub now { _time }
		1247	sub now_update { }
		1248
334	# default implementation for ->condvar, ->wait, ->broadcast	1249	# default implementation for ->condvar
335		1250
336	sub condvar {	1251	sub condvar {
337	bless \my $flag, "AnyEvent::Base::CondVar"	1252	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, "AnyEvent::CondVar"
338	}
339
340	sub AnyEvent::Base::CondVar::broadcast {
341	${$_[0]}++;
342	}
343
344	sub AnyEvent::Base::CondVar::wait {
345	AnyEvent->one_event while !${$_[0]};
346	}	1253	}
347		1254
348	# default implementation for ->signal	1255	# default implementation for ->signal
349		1256
350	our %SIG_CB;	1257	our $HAVE_ASYNC_INTERRUPT;
		1258	our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
		1259	our (%SIG_ASY, %SIG_ASY_W);
		1260	our ($SIG_COUNT, $SIG_TW);
351		1261
		1262	sub _signal_exec {
		1263	$HAVE_ASYNC_INTERRUPT
		1264	? $SIGPIPE_R->drain
		1265	: sysread $SIGPIPE_R, my $dummy, 9;
		1266
		1267	while (%SIG_EV) {
		1268	for (keys %SIG_EV) {
		1269	delete $SIG_EV{$_};
		1270	$_->() for values %{ $SIG_CB{$_} || {} };
		1271	}
		1272	}
		1273	}
		1274
352	sub signal {	1275	sub _signal {
353	my (undef, %arg) = @_;	1276	my (undef, %arg) = @_;
354		1277
355	my $signal = uc $arg{signal}	1278	my $signal = uc $arg{signal}
356	or Carp::croak "required option 'signal' is missing";	1279	or Carp::croak "required option 'signal' is missing";
357		1280
358	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};	1281	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1282
		1283	if ($HAVE_ASYNC_INTERRUPT) {
		1284	# async::interrupt
		1285
		1286	$SIG_ASY{$signal} ||= do {
		1287	my $asy = new Async::Interrupt
		1288	cb => sub { undef $SIG_EV{$signal} },
		1289	signal => $signal,
		1290	pipe => [$SIGPIPE_R->filenos],
		1291	;
		1292	$asy->pipe_autodrain (0);
		1293
		1294	$asy
		1295	};
		1296
		1297	} else {
		1298	# pure perl
		1299
359	$SIG{$signal} ||= sub {	1300	$SIG{$signal} ||= sub {
360	$_->() for values %{ $SIG_CB{$signal} || {} };	1301	local $!;
		1302	syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
		1303	undef $SIG_EV{$signal};
		1304	};
		1305
		1306	# can't do signal processing without introducing races in pure perl,
		1307	# so limit the signal latency.
		1308	++$SIG_COUNT;
		1309	$SIG_TW ||= AnyEvent->timer (
		1310	after => $MAX_SIGNAL_LATENCY,
		1311	interval => $MAX_SIGNAL_LATENCY,
		1312	cb => sub { }, # just for the PERL_ASYNC_CHECK
		1313	);
361	};	1314	}
362		1315
363	bless [$signal, $arg{cb}], "AnyEvent::Base::Signal"	1316	bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
364	}	1317	}
365		1318
		1319	sub signal {
		1320	# probe for availability of Async::Interrupt
		1321	if (!$ENV{PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT} && eval "use Async::Interrupt 0.6 (); 1") {
		1322	warn "AnyEvent: using Async::Interrupt for race-free signal handling.\n" if $VERBOSE >= 8;
		1323
		1324	$HAVE_ASYNC_INTERRUPT = 1;
		1325	$SIGPIPE_R = new Async::Interrupt::EventPipe;
		1326	$SIG_IO = AnyEvent->io (fh => $SIGPIPE_R->fileno, poll => "r", cb => \&_signal_exec);
		1327
		1328	} else {
		1329	warn "AnyEvent: using emulated perl signal handling with latency timer.\n" if $VERBOSE >= 8;
		1330
		1331	require Fcntl;
		1332
		1333	if (AnyEvent::WIN32) {
		1334	require AnyEvent::Util;
		1335
		1336	($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
		1337	AnyEvent::Util::fh_nonblocking ($SIGPIPE_R) if $SIGPIPE_R;
		1338	AnyEvent::Util::fh_nonblocking ($SIGPIPE_W) if $SIGPIPE_W; # just in case
		1339	} else {
		1340	pipe $SIGPIPE_R, $SIGPIPE_W;
		1341	fcntl $SIGPIPE_R, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_R;
		1342	fcntl $SIGPIPE_W, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_W; # just in case
		1343
		1344	# not strictly required, as $^F is normally 2, but let's make sure...
		1345	fcntl $SIGPIPE_R, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
		1346	fcntl $SIGPIPE_W, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
		1347	}
		1348
		1349	$SIGPIPE_R
		1350	or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
		1351
		1352	$SIG_IO = AnyEvent->io (fh => $SIGPIPE_R, poll => "r", cb => \&_signal_exec);
		1353	}
		1354
		1355	*signal = \&_signal;
		1356	&signal
		1357	}
		1358
366	sub AnyEvent::Base::Signal::DESTROY {	1359	sub AnyEvent::Base::signal::DESTROY {
367	my ($signal, $cb) = @{$_[0]};	1360	my ($signal, $cb) = @{$_[0]};
368		1361
		1362	undef $SIG_TW
		1363	unless --$SIG_COUNT;
		1364
369	delete $SIG_CB{$signal}{$cb};	1365	delete $SIG_CB{$signal}{$cb};
370		1366
371	$SIG{$signal} = 'DEFAULT' unless keys %{ $SIG_CB{$signal} };	1367	# delete doesn't work with older perls - they then
		1368	# print weird messages, or just unconditionally exit
		1369	# instead of getting the default action.
		1370	undef $SIG{$signal}
		1371	unless keys %{ $SIG_CB{$signal} };
372	}	1372	}
373		1373
374	# default implementation for ->child	1374	# default implementation for ->child
375		1375
376	our %PID_CB;	1376	our %PID_CB;
377	our $CHLD_W;	1377	our $CHLD_W;
378	our $PID_IDLE;	1378	our $CHLD_DELAY_W;
379	our $WNOHANG;	1379	our $WNOHANG;
380		1380
381	sub _child_wait {	1381	sub _sigchld {
382	while (0 <= (my $pid = waitpid -1, $WNOHANG)) {	1382	while (0 < (my $pid = waitpid -1, $WNOHANG)) {
		1383	$_->($pid, $?)
383	$_->($pid, $?) for (values %{ $PID_CB{$pid} || {} }),	1384	for values %{ $PID_CB{$pid} || {} },
384	(values %{ $PID_CB{0} || {} });	1385	values %{ $PID_CB{0} || {} };
385	}	1386	}
386
387	undef $PID_IDLE;
388	}	1387	}
389		1388
390	sub child {	1389	sub child {
391	my (undef, %arg) = @_;	1390	my (undef, %arg) = @_;
392		1391
393	defined (my $pid = $arg{pid} + 0)	1392	defined (my $pid = $arg{pid} + 0)
394	or Carp::croak "required option 'pid' is missing";	1393	or Carp::croak "required option 'pid' is missing";
395		1394
396	$PID_CB{$pid}{$arg{cb}} = $arg{cb};	1395	$PID_CB{$pid}{$arg{cb}} = $arg{cb};
397		1396
398	unless ($WNOHANG) {	1397	# WNOHANG is almost cetrainly 1 everywhere
399	$WNOHANG = eval { require POSIX; &POSIX::WNOHANG } || 1;	1398	$WNOHANG ||= $^O =~ /^(?:openbsd|netbsd|linux|freebsd|cygwin|MSWin32)$/
		1399	? 1
		1400	: eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
		1401
		1402	unless ($CHLD_W) {
		1403	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
		1404	# child could be a zombie already, so make at least one round
		1405	&_sigchld;
400	}	1406	}
401		1407
402	unless ($CHLD_W) {
403	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_child_wait);
404	# child could be a zombie already
405	$PID_IDLE ||= AnyEvent->timer (after => 0, cb => \&_child_wait);
406	}
407
408	bless [$pid, $arg{cb}], "AnyEvent::Base::Child"	1408	bless [$pid, $arg{cb}], "AnyEvent::Base::child"
409	}	1409	}
410		1410
411	sub AnyEvent::Base::Child::DESTROY {	1411	sub AnyEvent::Base::child::DESTROY {
412	my ($pid, $cb) = @{$_[0]};	1412	my ($pid, $cb) = @{$_[0]};
413		1413
414	delete $PID_CB{$pid}{$cb};	1414	delete $PID_CB{$pid}{$cb};
415	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };	1415	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
416		1416
417	undef $CHLD_W unless keys %PID_CB;	1417	undef $CHLD_W unless keys %PID_CB;
418	}	1418	}
419		1419
		1420	# idle emulation is done by simply using a timer, regardless
		1421	# of whether the process is idle or not, and not letting
		1422	# the callback use more than 50% of the time.
		1423	sub idle {
		1424	my (undef, %arg) = @_;
		1425
		1426	my ($cb, $w, $rcb) = $arg{cb};
		1427
		1428	$rcb = sub {
		1429	if ($cb) {
		1430	$w = _time;
		1431	&$cb;
		1432	$w = _time - $w;
		1433
		1434	# never use more then 50% of the time for the idle watcher,
		1435	# within some limits
		1436	$w = 0.0001 if $w < 0.0001;
		1437	$w = 5 if $w > 5;
		1438
		1439	$w = AnyEvent->timer (after => $w, cb => $rcb);
		1440	} else {
		1441	# clean up...
		1442	undef $w;
		1443	undef $rcb;
		1444	}
		1445	};
		1446
		1447	$w = AnyEvent->timer (after => 0.05, cb => $rcb);
		1448
		1449	bless \\$cb, "AnyEvent::Base::idle"
		1450	}
		1451
		1452	sub AnyEvent::Base::idle::DESTROY {
		1453	undef $${$_[0]};
		1454	}
		1455
		1456	package AnyEvent::CondVar;
		1457
		1458	our @ISA = AnyEvent::CondVar::Base::;
		1459
		1460	package AnyEvent::CondVar::Base;
		1461
		1462	#use overload
		1463	# '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1464	# fallback => 1;
		1465
		1466	# save 300+ kilobytes by dirtily hardcoding overloading
		1467	${"AnyEvent::CondVar::Base::OVERLOAD"}{dummy}++; # Register with magic by touching.
		1468	*{'AnyEvent::CondVar::Base::()'} = sub { }; # "Make it findable via fetchmethod."
		1469	*{'AnyEvent::CondVar::Base::(&{}'} = sub { my $self = shift; sub { $self->send (@_) } }; # &{}
		1470	${'AnyEvent::CondVar::Base::()'} = 1; # fallback
		1471
		1472	our $WAITING;
		1473
		1474	sub _send {
		1475	# nop
		1476	}
		1477
		1478	sub send {
		1479	my $cv = shift;
		1480	$cv->{_ae_sent} = [@_];
		1481	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1482	$cv->_send;
		1483	}
		1484
		1485	sub croak {
		1486	$_[0]{_ae_croak} = $_[1];
		1487	$_[0]->send;
		1488	}
		1489
		1490	sub ready {
		1491	$_[0]{_ae_sent}
		1492	}
		1493
		1494	sub _wait {
		1495	$WAITING
		1496	and !$_[0]{_ae_sent}
		1497	and Carp::croak "AnyEvent::CondVar: recursive blocking wait detected";
		1498
		1499	local $WAITING = 1;
		1500	AnyEvent->one_event while !$_[0]{_ae_sent};
		1501	}
		1502
		1503	sub recv {
		1504	$_[0]->_wait;
		1505
		1506	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		1507	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		1508	}
		1509
		1510	sub cb {
		1511	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		1512	$_[0]{_ae_cb}
		1513	}
		1514
		1515	sub begin {
		1516	++$_[0]{_ae_counter};
		1517	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		1518	}
		1519
		1520	sub end {
		1521	return if --$_[0]{_ae_counter};
		1522	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		1523	}
		1524
		1525	# undocumented/compatibility with pre-3.4
		1526	*broadcast = \&send;
		1527	*wait = \&_wait;
		1528
		1529	=head1 ERROR AND EXCEPTION HANDLING
		1530
		1531	In general, AnyEvent does not do any error handling - it relies on the
		1532	caller to do that if required. The L<AnyEvent::Strict> module (see also
		1533	the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
		1534	checking of all AnyEvent methods, however, which is highly useful during
		1535	development.
		1536
		1537	As for exception handling (i.e. runtime errors and exceptions thrown while
		1538	executing a callback), this is not only highly event-loop specific, but
		1539	also not in any way wrapped by this module, as this is the job of the main
		1540	program.
		1541
		1542	The pure perl event loop simply re-throws the exception (usually
		1543	within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
		1544	$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
		1545	so on.
		1546
		1547	=head1 ENVIRONMENT VARIABLES
		1548
		1549	The following environment variables are used by this module or its
		1550	submodules.
		1551
		1552	Note that AnyEvent will remove I<all> environment variables starting with
		1553	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		1554	enabled.
		1555
		1556	=over 4
		1557
		1558	=item C<PERL_ANYEVENT_VERBOSE>
		1559
		1560	By default, AnyEvent will be completely silent except in fatal
		1561	conditions. You can set this environment variable to make AnyEvent more
		1562	talkative.
		1563
		1564	When set to C<1> or higher, causes AnyEvent to warn about unexpected
		1565	conditions, such as not being able to load the event model specified by
		1566	C<PERL_ANYEVENT_MODEL>.
		1567
		1568	When set to C<2> or higher, cause AnyEvent to report to STDERR which event
		1569	model it chooses.
		1570
		1571	When set to C<8> or higher, then AnyEvent will report extra information on
		1572	which optional modules it loads and how it implements certain features.
		1573
		1574	=item C<PERL_ANYEVENT_STRICT>
		1575
		1576	AnyEvent does not do much argument checking by default, as thorough
		1577	argument checking is very costly. Setting this variable to a true value
		1578	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		1579	check the arguments passed to most method calls. If it finds any problems,
		1580	it will croak.
		1581
		1582	In other words, enables "strict" mode.
		1583
		1584	Unlike C<use strict> (or it's modern cousin, C<< use L<common::sense>
		1585	>>, it is definitely recommended to keep it off in production. Keeping
		1586	C<PERL_ANYEVENT_STRICT=1> in your environment while developing programs
		1587	can be very useful, however.
		1588
		1589	=item C<PERL_ANYEVENT_MODEL>
		1590
		1591	This can be used to specify the event model to be used by AnyEvent, before
		1592	auto detection and -probing kicks in. It must be a string consisting
		1593	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
		1594	and the resulting module name is loaded and if the load was successful,
		1595	used as event model. If it fails to load AnyEvent will proceed with
		1596	auto detection and -probing.
		1597
		1598	This functionality might change in future versions.
		1599
		1600	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
		1601	could start your program like this:
		1602
		1603	PERL_ANYEVENT_MODEL=Perl perl ...
		1604
		1605	=item C<PERL_ANYEVENT_PROTOCOLS>
		1606
		1607	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1608	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1609	of auto probing).
		1610
		1611	Must be set to a comma-separated list of protocols or address families,
		1612	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1613	used, and preference will be given to protocols mentioned earlier in the
		1614	list.
		1615
		1616	This variable can effectively be used for denial-of-service attacks
		1617	against local programs (e.g. when setuid), although the impact is likely
		1618	small, as the program has to handle conenction and other failures anyways.
		1619
		1620	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1621	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1622	- only support IPv4, never try to resolve or contact IPv6
		1623	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1624	IPv6, but prefer IPv6 over IPv4.
		1625
		1626	=item C<PERL_ANYEVENT_EDNS0>
		1627
		1628	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1629	for DNS. This extension is generally useful to reduce DNS traffic, but
		1630	some (broken) firewalls drop such DNS packets, which is why it is off by
		1631	default.
		1632
		1633	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1634	EDNS0 in its DNS requests.
		1635
		1636	=item C<PERL_ANYEVENT_MAX_FORKS>
		1637
		1638	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		1639	will create in parallel.
		1640
		1641	=item C<PERL_ANYEVENT_MAX_OUTSTANDING_DNS>
		1642
		1643	The default value for the C<max_outstanding> parameter for the default DNS
		1644	resolver - this is the maximum number of parallel DNS requests that are
		1645	sent to the DNS server.
		1646
		1647	=item C<PERL_ANYEVENT_RESOLV_CONF>
		1648
		1649	The file to use instead of F</etc/resolv.conf> (or OS-specific
		1650	configuration) in the default resolver. When set to the empty string, no
		1651	default config will be used.
		1652
		1653	=item C<PERL_ANYEVENT_CA_FILE>, C<PERL_ANYEVENT_CA_PATH>.
		1654
		1655	When neither C<ca_file> nor C<ca_path> was specified during
		1656	L<AnyEvent::TLS> context creation, and either of these environment
		1657	variables exist, they will be used to specify CA certificate locations
		1658	instead of a system-dependent default.
		1659
		1660	=item C<PERL_ANYEVENT_AVOID_GUARD> and C<PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT>
		1661
		1662	When these are set to C<1>, then the respective modules are not
		1663	loaded. Mostly good for testing AnyEvent itself.
		1664
		1665	=back
		1666
420	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1667	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
		1668
		1669	This is an advanced topic that you do not normally need to use AnyEvent in
		1670	a module. This section is only of use to event loop authors who want to
		1671	provide AnyEvent compatibility.
421		1672
422	If you need to support another event library which isn't directly	1673	If you need to support another event library which isn't directly
423	supported by AnyEvent, you can supply your own interface to it by	1674	supported by AnyEvent, you can supply your own interface to it by
424	pushing, before the first watcher gets created, the package name of	1675	pushing, before the first watcher gets created, the package name of
425	the event module and the package name of the interface to use onto	1676	the event module and the package name of the interface to use onto
426	C<@AnyEvent::REGISTRY>. You can do that before and even without loading	1677	C<@AnyEvent::REGISTRY>. You can do that before and even without loading
427	AnyEvent.	1678	AnyEvent, so it is reasonably cheap.
428		1679
429	Example:	1680	Example:
430		1681
431	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];	1682	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];
432		1683
433	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>	1684	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>
434	package/class when it finds the C<urxvt> package/module is loaded. When	1685	package/class when it finds the C<urxvt> package/module is already loaded.
		1686
435	AnyEvent is loaded and asked to find a suitable event model, it will	1687	When AnyEvent is loaded and asked to find a suitable event model, it
436	first check for the presence of urxvt.	1688	will first check for the presence of urxvt by trying to C<use> the
		1689	C<urxvt::anyevent> module.
437		1690
438	The class should provide implementations for all watcher types (see	1691	The class should provide implementations for all watcher types. See
439	L<AnyEvent::Impl::Event> (source code), L<AnyEvent::Impl::Glib>	1692	L<AnyEvent::Impl::EV> (source code), L<AnyEvent::Impl::Glib> (Source code)
440	(Source code) and so on for actual examples, use C<perldoc -m	1693	and so on for actual examples. Use C<perldoc -m AnyEvent::Impl::Glib> to
441	AnyEvent::Impl::Glib> to see the sources).	1694	see the sources.
442		1695
		1696	If you don't provide C<signal> and C<child> watchers than AnyEvent will
		1697	provide suitable (hopefully) replacements.
		1698
443	The above isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)	1699	The above example isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)
444	uses the above line as-is. An interface isn't included in AnyEvent	1700	terminal emulator uses the above line as-is. An interface isn't included
445	because it doesn't make sense outside the embedded interpreter inside	1701	in AnyEvent because it doesn't make sense outside the embedded interpreter
446	I<rxvt-unicode>, and it is updated and maintained as part of the	1702	inside I<rxvt-unicode>, and it is updated and maintained as part of the
447	I<rxvt-unicode> distribution.	1703	I<rxvt-unicode> distribution.
448		1704
449	I<rxvt-unicode> also cheats a bit by not providing blocking access to	1705	I<rxvt-unicode> also cheats a bit by not providing blocking access to
450	condition variables: code blocking while waiting for a condition will	1706	condition variables: code blocking while waiting for a condition will
451	C<die>. This still works with most modules/usages, and blocking calls must	1707	C<die>. This still works with most modules/usages, and blocking calls must
452	not be in an interactive application, so it makes sense.	1708	not be done in an interactive application, so it makes sense.
453		1709
454	=head1 ENVIRONMENT VARIABLES
455
456	The following environment variables are used by this module:
457
458	C<PERL_ANYEVENT_VERBOSE> when set to C<2> or higher, reports which event
459	model gets used.
460
461	=head1 EXAMPLE	1710	=head1 EXAMPLE PROGRAM
462		1711
463	The following program uses an io watcher to read data from stdin, a timer	1712	The following program uses an I/O watcher to read data from STDIN, a timer
464	to display a message once per second, and a condvar to exit the program	1713	to display a message once per second, and a condition variable to quit the
465	when the user enters quit:	1714	program when the user enters quit:
466		1715
467	use AnyEvent;	1716	use AnyEvent;
468		1717
469	my $cv = AnyEvent->condvar;	1718	my $cv = AnyEvent->condvar;
470		1719
471	my $io_watcher = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	1720	my $io_watcher = AnyEvent->io (
		1721	fh => \*STDIN,
		1722	poll => 'r',
		1723	cb => sub {
472	warn "io event <$_[0]>\n"; # will always output <r>	1724	warn "io event <$_[0]>\n"; # will always output <r>
473	chomp (my $input = <STDIN>); # read a line	1725	chomp (my $input = <STDIN>); # read a line
474	warn "read: $input\n"; # output what has been read	1726	warn "read: $input\n"; # output what has been read
475	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1727	$cv->send if $input =~ /^q/i; # quit program if /^q/i
		1728	},
476	});	1729	);
477		1730
478	my $time_watcher; # can only be used once	1731	my $time_watcher; # can only be used once
479		1732
480	sub new_timer {	1733	sub new_timer {
481	$timer = AnyEvent->timer (after => 1, cb => sub {	1734	$timer = AnyEvent->timer (after => 1, cb => sub {
…		…
484	});	1737	});
485	}	1738	}
486		1739
487	new_timer; # create first timer	1740	new_timer; # create first timer
488		1741
489	$cv->wait; # wait until user enters /^q/i	1742	$cv->recv; # wait until user enters /^q/i
490		1743
491	=head1 REAL-WORLD EXAMPLE	1744	=head1 REAL-WORLD EXAMPLE
492		1745
493	Consider the L<Net::FCP> module. It features (among others) the following	1746	Consider the L<Net::FCP> module. It features (among others) the following
494	API calls, which are to freenet what HTTP GET requests are to http:	1747	API calls, which are to freenet what HTTP GET requests are to http:
…		…
544	syswrite $txn->{fh}, $txn->{request}	1797	syswrite $txn->{fh}, $txn->{request}
545	or die "connection or write error";	1798	or die "connection or write error";
546	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	1799	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
547		1800
548	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	1801	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
549	result and signals any possible waiters that the request ahs finished:	1802	result and signals any possible waiters that the request has finished:
550		1803
551	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1804	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
552		1805
553	if (end-of-file or data complete) {	1806	if (end-of-file or data complete) {
554	$txn->{result} = $txn->{buf};	1807	$txn->{result} = $txn->{buf};
555	$txn->{finished}->broadcast;	1808	$txn->{finished}->send;
556	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1809	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
557	}	1810	}
558		1811
559	The C<result> method, finally, just waits for the finished signal (if the	1812	The C<result> method, finally, just waits for the finished signal (if the
560	request was already finished, it doesn't wait, of course, and returns the	1813	request was already finished, it doesn't wait, of course, and returns the
561	data:	1814	data:
562		1815
563	$txn->{finished}->wait;	1816	$txn->{finished}->recv;
564	return $txn->{result};	1817	return $txn->{result};
565		1818
566	The actual code goes further and collects all errors (C<die>s, exceptions)	1819	The actual code goes further and collects all errors (C<die>s, exceptions)
567	that occured during request processing. The C<result> method detects	1820	that occurred during request processing. The C<result> method detects
568	wether an exception as thrown (it is stored inside the $txn object)	1821	whether an exception as thrown (it is stored inside the $txn object)
569	and just throws the exception, which means connection errors and other	1822	and just throws the exception, which means connection errors and other
570	problems get reported tot he code that tries to use the result, not in a	1823	problems get reported tot he code that tries to use the result, not in a
571	random callback.	1824	random callback.
572		1825
573	All of this enables the following usage styles:	1826	All of this enables the following usage styles:
574		1827
575	1. Blocking:	1828	1. Blocking:
576		1829
577	my $data = $fcp->client_get ($url);	1830	my $data = $fcp->client_get ($url);
578		1831
579	2. Blocking, but parallelizing:	1832	2. Blocking, but running in parallel:
580		1833
581	my @datas = map $_->result,	1834	my @datas = map $_->result,
582	map $fcp->txn_client_get ($_),	1835	map $fcp->txn_client_get ($_),
583	@urls;	1836	@urls;
584		1837
585	Both blocking examples work without the module user having to know	1838	Both blocking examples work without the module user having to know
586	anything about events.	1839	anything about events.
587		1840
588	3a. Event-based in a main program, using any support Event module:	1841	3a. Event-based in a main program, using any supported event module:
589		1842
590	use Event;	1843	use EV;
591		1844
592	$fcp->txn_client_get ($url)->cb (sub {	1845	$fcp->txn_client_get ($url)->cb (sub {
593	my $txn = shift;	1846	my $txn = shift;
594	my $data = $txn->result;	1847	my $data = $txn->result;
595	...	1848	...
596	});	1849	});
597		1850
598	Event::loop;	1851	EV::loop;
599		1852
600	3b. The module user could use AnyEvent, too:	1853	3b. The module user could use AnyEvent, too:
601		1854
602	use AnyEvent;	1855	use AnyEvent;
603		1856
604	my $quit = AnyEvent->condvar;	1857	my $quit = AnyEvent->condvar;
605		1858
606	$fcp->txn_client_get ($url)->cb (sub {	1859	$fcp->txn_client_get ($url)->cb (sub {
607	...	1860	...
608	$quit->broadcast;	1861	$quit->send;
609	});	1862	});
610		1863
611	$quit->wait;	1864	$quit->recv;
		1865
		1866
		1867	=head1 BENCHMARKS
		1868
		1869	To give you an idea of the performance and overheads that AnyEvent adds
		1870	over the event loops themselves and to give you an impression of the speed
		1871	of various event loops I prepared some benchmarks.
		1872
		1873	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1874
		1875	Here is a benchmark of various supported event models used natively and
		1876	through AnyEvent. The benchmark creates a lot of timers (with a zero
		1877	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
		1878	which it is), lets them fire exactly once and destroys them again.
		1879
		1880	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
		1881	distribution.
		1882
		1883	=head3 Explanation of the columns
		1884
		1885	I<watcher> is the number of event watchers created/destroyed. Since
		1886	different event models feature vastly different performances, each event
		1887	loop was given a number of watchers so that overall runtime is acceptable
		1888	and similar between tested event loop (and keep them from crashing): Glib
		1889	would probably take thousands of years if asked to process the same number
		1890	of watchers as EV in this benchmark.
		1891
		1892	I<bytes> is the number of bytes (as measured by the resident set size,
		1893	RSS) consumed by each watcher. This method of measuring captures both C
		1894	and Perl-based overheads.
		1895
		1896	I<create> is the time, in microseconds (millionths of seconds), that it
		1897	takes to create a single watcher. The callback is a closure shared between
		1898	all watchers, to avoid adding memory overhead. That means closure creation
		1899	and memory usage is not included in the figures.
		1900
		1901	I<invoke> is the time, in microseconds, used to invoke a simple
		1902	callback. The callback simply counts down a Perl variable and after it was
		1903	invoked "watcher" times, it would C<< ->send >> a condvar once to
		1904	signal the end of this phase.
		1905
		1906	I<destroy> is the time, in microseconds, that it takes to destroy a single
		1907	watcher.
		1908
		1909	=head3 Results
		1910
		1911	name watchers bytes create invoke destroy comment
		1912	EV/EV 400000 224 0.47 0.35 0.27 EV native interface
		1913	EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers
		1914	CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal
		1915	Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation
		1916	Event/Event 16000 517 32.20 31.80 0.81 Event native interface
		1917	Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers
		1918	IOAsync/Any 16000 989 38.10 32.77 11.13 via IO::Async::Loop::IO_Poll
		1919	IOAsync/Any 16000 990 37.59 29.50 10.61 via IO::Async::Loop::Epoll
		1920	Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour
		1921	Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers
		1922	POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event
		1923	POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select
		1924
		1925	=head3 Discussion
		1926
		1927	The benchmark does I<not> measure scalability of the event loop very
		1928	well. For example, a select-based event loop (such as the pure perl one)
		1929	can never compete with an event loop that uses epoll when the number of
		1930	file descriptors grows high. In this benchmark, all events become ready at
		1931	the same time, so select/poll-based implementations get an unnatural speed
		1932	boost.
		1933
		1934	Also, note that the number of watchers usually has a nonlinear effect on
		1935	overall speed, that is, creating twice as many watchers doesn't take twice
		1936	the time - usually it takes longer. This puts event loops tested with a
		1937	higher number of watchers at a disadvantage.
		1938
		1939	To put the range of results into perspective, consider that on the
		1940	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1941	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1942	cycles with POE.
		1943
		1944	C<EV> is the sole leader regarding speed and memory use, which are both
		1945	maximal/minimal, respectively. Even when going through AnyEvent, it uses
		1946	far less memory than any other event loop and is still faster than Event
		1947	natively.
		1948
		1949	The pure perl implementation is hit in a few sweet spots (both the
		1950	constant timeout and the use of a single fd hit optimisations in the perl
		1951	interpreter and the backend itself). Nevertheless this shows that it
		1952	adds very little overhead in itself. Like any select-based backend its
		1953	performance becomes really bad with lots of file descriptors (and few of
		1954	them active), of course, but this was not subject of this benchmark.
		1955
		1956	The C<Event> module has a relatively high setup and callback invocation
		1957	cost, but overall scores in on the third place.
		1958
		1959	C<IO::Async> performs admirably well, about on par with C<Event>, even
		1960	when using its pure perl backend.
		1961
		1962	C<Glib>'s memory usage is quite a bit higher, but it features a
		1963	faster callback invocation and overall ends up in the same class as
		1964	C<Event>. However, Glib scales extremely badly, doubling the number of
		1965	watchers increases the processing time by more than a factor of four,
		1966	making it completely unusable when using larger numbers of watchers
		1967	(note that only a single file descriptor was used in the benchmark, so
		1968	inefficiencies of C<poll> do not account for this).
		1969
		1970	The C<Tk> adaptor works relatively well. The fact that it crashes with
		1971	more than 2000 watchers is a big setback, however, as correctness takes
		1972	precedence over speed. Nevertheless, its performance is surprising, as the
		1973	file descriptor is dup()ed for each watcher. This shows that the dup()
		1974	employed by some adaptors is not a big performance issue (it does incur a
		1975	hidden memory cost inside the kernel which is not reflected in the figures
		1976	above).
		1977
		1978	C<POE>, regardless of underlying event loop (whether using its pure perl
		1979	select-based backend or the Event module, the POE-EV backend couldn't
		1980	be tested because it wasn't working) shows abysmal performance and
		1981	memory usage with AnyEvent: Watchers use almost 30 times as much memory
		1982	as EV watchers, and 10 times as much memory as Event (the high memory
		1983	requirements are caused by requiring a session for each watcher). Watcher
		1984	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		1985	implementation.
		1986
		1987	The design of the POE adaptor class in AnyEvent can not really account
		1988	for the performance issues, though, as session creation overhead is
		1989	small compared to execution of the state machine, which is coded pretty
		1990	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		1991	using multiple sessions is not a good approach, especially regarding
		1992	memory usage, even the author of POE could not come up with a faster
		1993	design).
		1994
		1995	=head3 Summary
		1996
		1997	=over 4
		1998
		1999	=item * Using EV through AnyEvent is faster than any other event loop
		2000	(even when used without AnyEvent), but most event loops have acceptable
		2001	performance with or without AnyEvent.
		2002
		2003	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
		2004	the actual event loop, only with extremely fast event loops such as EV
		2005	adds AnyEvent significant overhead.
		2006
		2007	=item * You should avoid POE like the plague if you want performance or
		2008	reasonable memory usage.
		2009
		2010	=back
		2011
		2012	=head2 BENCHMARKING THE LARGE SERVER CASE
		2013
		2014	This benchmark actually benchmarks the event loop itself. It works by
		2015	creating a number of "servers": each server consists of a socket pair, a
		2016	timeout watcher that gets reset on activity (but never fires), and an I/O
		2017	watcher waiting for input on one side of the socket. Each time the socket
		2018	watcher reads a byte it will write that byte to a random other "server".
		2019
		2020	The effect is that there will be a lot of I/O watchers, only part of which
		2021	are active at any one point (so there is a constant number of active
		2022	fds for each loop iteration, but which fds these are is random). The
		2023	timeout is reset each time something is read because that reflects how
		2024	most timeouts work (and puts extra pressure on the event loops).
		2025
		2026	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		2027	(1%) are active. This mirrors the activity of large servers with many
		2028	connections, most of which are idle at any one point in time.
		2029
		2030	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		2031	distribution.
		2032
		2033	=head3 Explanation of the columns
		2034
		2035	I<sockets> is the number of sockets, and twice the number of "servers" (as
		2036	each server has a read and write socket end).
		2037
		2038	I<create> is the time it takes to create a socket pair (which is
		2039	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		2040
		2041	I<request>, the most important value, is the time it takes to handle a
		2042	single "request", that is, reading the token from the pipe and forwarding
		2043	it to another server. This includes deleting the old timeout and creating
		2044	a new one that moves the timeout into the future.
		2045
		2046	=head3 Results
		2047
		2048	name sockets create request
		2049	EV 20000 69.01 11.16
		2050	Perl 20000 73.32 35.87
		2051	IOAsync 20000 157.00 98.14 epoll
		2052	IOAsync 20000 159.31 616.06 poll
		2053	Event 20000 212.62 257.32
		2054	Glib 20000 651.16 1896.30
		2055	POE 20000 349.67 12317.24 uses POE::Loop::Event
		2056
		2057	=head3 Discussion
		2058
		2059	This benchmark I<does> measure scalability and overall performance of the
		2060	particular event loop.
		2061
		2062	EV is again fastest. Since it is using epoll on my system, the setup time
		2063	is relatively high, though.
		2064
		2065	Perl surprisingly comes second. It is much faster than the C-based event
		2066	loops Event and Glib.
		2067
		2068	IO::Async performs very well when using its epoll backend, and still quite
		2069	good compared to Glib when using its pure perl backend.
		2070
		2071	Event suffers from high setup time as well (look at its code and you will
		2072	understand why). Callback invocation also has a high overhead compared to
		2073	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		2074	uses select or poll in basically all documented configurations.
		2075
		2076	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		2077	clearly fails to perform with many filehandles or in busy servers.
		2078
		2079	POE is still completely out of the picture, taking over 1000 times as long
		2080	as EV, and over 100 times as long as the Perl implementation, even though
		2081	it uses a C-based event loop in this case.
		2082
		2083	=head3 Summary
		2084
		2085	=over 4
		2086
		2087	=item * The pure perl implementation performs extremely well.
		2088
		2089	=item * Avoid Glib or POE in large projects where performance matters.
		2090
		2091	=back
		2092
		2093	=head2 BENCHMARKING SMALL SERVERS
		2094
		2095	While event loops should scale (and select-based ones do not...) even to
		2096	large servers, most programs we (or I :) actually write have only a few
		2097	I/O watchers.
		2098
		2099	In this benchmark, I use the same benchmark program as in the large server
		2100	case, but it uses only eight "servers", of which three are active at any
		2101	one time. This should reflect performance for a small server relatively
		2102	well.
		2103
		2104	The columns are identical to the previous table.
		2105
		2106	=head3 Results
		2107
		2108	name sockets create request
		2109	EV 16 20.00 6.54
		2110	Perl 16 25.75 12.62
		2111	Event 16 81.27 35.86
		2112	Glib 16 32.63 15.48
		2113	POE 16 261.87 276.28 uses POE::Loop::Event
		2114
		2115	=head3 Discussion
		2116
		2117	The benchmark tries to test the performance of a typical small
		2118	server. While knowing how various event loops perform is interesting, keep
		2119	in mind that their overhead in this case is usually not as important, due
		2120	to the small absolute number of watchers (that is, you need efficiency and
		2121	speed most when you have lots of watchers, not when you only have a few of
		2122	them).
		2123
		2124	EV is again fastest.
		2125
		2126	Perl again comes second. It is noticeably faster than the C-based event
		2127	loops Event and Glib, although the difference is too small to really
		2128	matter.
		2129
		2130	POE also performs much better in this case, but is is still far behind the
		2131	others.
		2132
		2133	=head3 Summary
		2134
		2135	=over 4
		2136
		2137	=item * C-based event loops perform very well with small number of
		2138	watchers, as the management overhead dominates.
		2139
		2140	=back
		2141
		2142	=head2 THE IO::Lambda BENCHMARK
		2143
		2144	Recently I was told about the benchmark in the IO::Lambda manpage, which
		2145	could be misinterpreted to make AnyEvent look bad. In fact, the benchmark
		2146	simply compares IO::Lambda with POE, and IO::Lambda looks better (which
		2147	shouldn't come as a surprise to anybody). As such, the benchmark is
		2148	fine, and mostly shows that the AnyEvent backend from IO::Lambda isn't
		2149	very optimal. But how would AnyEvent compare when used without the extra
		2150	baggage? To explore this, I wrote the equivalent benchmark for AnyEvent.
		2151
		2152	The benchmark itself creates an echo-server, and then, for 500 times,
		2153	connects to the echo server, sends a line, waits for the reply, and then
		2154	creates the next connection. This is a rather bad benchmark, as it doesn't
		2155	test the efficiency of the framework or much non-blocking I/O, but it is a
		2156	benchmark nevertheless.
		2157
		2158	name runtime
		2159	Lambda/select 0.330 sec
		2160	+ optimized 0.122 sec
		2161	Lambda/AnyEvent 0.327 sec
		2162	+ optimized 0.138 sec
		2163	Raw sockets/select 0.077 sec
		2164	POE/select, components 0.662 sec
		2165	POE/select, raw sockets 0.226 sec
		2166	POE/select, optimized 0.404 sec
		2167
		2168	AnyEvent/select/nb 0.085 sec
		2169	AnyEvent/EV/nb 0.068 sec
		2170	+state machine 0.134 sec
		2171
		2172	The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
		2173	benchmarks actually make blocking connects and use 100% blocking I/O,
		2174	defeating the purpose of an event-based solution. All of the newly
		2175	written AnyEvent benchmarks use 100% non-blocking connects (using
		2176	AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
		2177	resolver), so AnyEvent is at a disadvantage here, as non-blocking connects
		2178	generally require a lot more bookkeeping and event handling than blocking
		2179	connects (which involve a single syscall only).
		2180
		2181	The last AnyEvent benchmark additionally uses L<AnyEvent::Handle>, which
		2182	offers similar expressive power as POE and IO::Lambda, using conventional
		2183	Perl syntax. This means that both the echo server and the client are 100%
		2184	non-blocking, further placing it at a disadvantage.
		2185
		2186	As you can see, the AnyEvent + EV combination even beats the
		2187	hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
		2188	backend easily beats IO::Lambda and POE.
		2189
		2190	And even the 100% non-blocking version written using the high-level (and
		2191	slow :) L<AnyEvent::Handle> abstraction beats both POE and IO::Lambda by a
		2192	large margin, even though it does all of DNS, tcp-connect and socket I/O
		2193	in a non-blocking way.
		2194
		2195	The two AnyEvent benchmarks programs can be found as F<eg/ae0.pl> and
		2196	F<eg/ae2.pl> in the AnyEvent distribution, the remaining benchmarks are
		2197	part of the IO::lambda distribution and were used without any changes.
		2198
		2199
		2200	=head1 SIGNALS
		2201
		2202	AnyEvent currently installs handlers for these signals:
		2203
		2204	=over 4
		2205
		2206	=item SIGCHLD
		2207
		2208	A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
		2209	emulation for event loops that do not support them natively. Also, some
		2210	event loops install a similar handler.
		2211
		2212	Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, then
		2213	AnyEvent will reset it to default, to avoid losing child exit statuses.
		2214
		2215	=item SIGPIPE
		2216
		2217	A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
		2218	when AnyEvent gets loaded.
		2219
		2220	The rationale for this is that AnyEvent users usually do not really depend
		2221	on SIGPIPE delivery (which is purely an optimisation for shell use, or
		2222	badly-written programs), but C<SIGPIPE> can cause spurious and rare
		2223	program exits as a lot of people do not expect C<SIGPIPE> when writing to
		2224	some random socket.
		2225
		2226	The rationale for installing a no-op handler as opposed to ignoring it is
		2227	that this way, the handler will be restored to defaults on exec.
		2228
		2229	Feel free to install your own handler, or reset it to defaults.
		2230
		2231	=back
		2232
		2233	=cut
		2234
		2235	undef $SIG{CHLD}
		2236	if $SIG{CHLD} eq 'IGNORE';
		2237
		2238	$SIG{PIPE} = sub { }
		2239	unless defined $SIG{PIPE};
		2240
		2241	=head1 RECOMMENDED/OPTIONAL MODULES
		2242
		2243	One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
		2244	it's built-in modules) are required to use it.
		2245
		2246	That does not mean that AnyEvent won't take advantage of some additional
		2247	modules if they are installed.
		2248
		2249	This section epxlains which additional modules will be used, and how they
		2250	affect AnyEvent's operetion.
		2251
		2252	=over 4
		2253
		2254	=item L<Async::Interrupt>
		2255
		2256	This slightly arcane module is used to implement fast signal handling: To
		2257	my knowledge, there is no way to do completely race-free and quick
		2258	signal handling in pure perl. To ensure that signals still get
		2259	delivered, AnyEvent will start an interval timer to wake up perl (and
		2260	catch the signals) with soemd elay (default is 10 seconds, look for
		2261	C<$AnyEvent::MAX_SIGNAL_LATENCY>).
		2262
		2263	If this module is available, then it will be used to implement signal
		2264	catching, which means that signals will not be delayed, and the event loop
		2265	will not be interrupted regularly, which is more efficient (And good for
		2266	battery life on laptops).
		2267
		2268	This affects not just the pure-perl event loop, but also other event loops
		2269	that have no signal handling on their own (e.g. Glib, Tk, Qt).
		2270
		2271	=item L<EV>
		2272
		2273	This module isn't really "optional", as it is simply one of the backend
		2274	event loops that AnyEvent can use. However, it is simply the best event
		2275	loop available in terms of features, speed and stability: It supports
		2276	the AnyEvent API optimally, implements all the watcher types in XS, does
		2277	automatic timer adjustments even when no monotonic clock is available,
		2278	can take avdantage of advanced kernel interfaces such as C<epoll> and
		2279	C<kqueue>, and is the fastest backend I<by far>. You can even embed
		2280	L<Glib>/L<Gtk2> in it (or vice versa, see L<EV::Glib> and L<Glib::EV>).
		2281
		2282	=item L<Guard>
		2283
		2284	The guard module, when used, will be used to implement
		2285	C<AnyEvent::Util::guard>. This speeds up guards considerably (and uses a
		2286	lot less memory), but otherwise doesn't affect guard operation much. It is
		2287	purely used for performance.
		2288
		2289	=item L<JSON> and L<JSON::XS>
		2290
		2291	This module is required when you want to read or write JSON data via
		2292	L<AnyEvent::Handle>. It is also written in pure-perl, but can take
		2293	advantage of the ulta-high-speed L<JSON::XS> module when it is installed.
		2294
		2295	In fact, L<AnyEvent::Handle> will use L<JSON::XS> by default if it is
		2296	installed.
		2297
		2298	=item L<Net::SSLeay>
		2299
		2300	Implementing TLS/SSL in Perl is certainly interesting, but not very
		2301	worthwhile: If this module is installed, then L<AnyEvent::Handle> (with
		2302	the help of L<AnyEvent::TLS>), gains the ability to do TLS/SSL.
		2303
		2304	=item L<Time::HiRes>
		2305
		2306	This module is part of perl since release 5.008. It will be used when the
		2307	chosen event library does not come with a timing source on it's own. The
		2308	pure-perl event loop (L<AnyEvent::Impl::Perl>) will additionally use it to
		2309	try to use a monotonic clock for timing stability.
		2310
		2311	=back
		2312
		2313
		2314	=head1 FORK
		2315
		2316	Most event libraries are not fork-safe. The ones who are usually are
		2317	because they rely on inefficient but fork-safe C<select> or C<poll>
		2318	calls. Only L<EV> is fully fork-aware.
		2319
		2320	If you have to fork, you must either do so I<before> creating your first
		2321	watcher OR you must not use AnyEvent at all in the child OR you must do
		2322	something completely out of the scope of AnyEvent.
		2323
		2324
		2325	=head1 SECURITY CONSIDERATIONS
		2326
		2327	AnyEvent can be forced to load any event model via
		2328	$ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used to
		2329	execute arbitrary code or directly gain access, it can easily be used to
		2330	make the program hang or malfunction in subtle ways, as AnyEvent watchers
		2331	will not be active when the program uses a different event model than
		2332	specified in the variable.
		2333
		2334	You can make AnyEvent completely ignore this variable by deleting it
		2335	before the first watcher gets created, e.g. with a C<BEGIN> block:
		2336
		2337	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
		2338
		2339	use AnyEvent;
		2340
		2341	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		2342	be used to probe what backend is used and gain other information (which is
		2343	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		2344	$ENV{PERL_ANYEVENT_STRICT}.
		2345
		2346	Note that AnyEvent will remove I<all> environment variables starting with
		2347	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		2348	enabled.
		2349
		2350
		2351	=head1 BUGS
		2352
		2353	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		2354	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		2355	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		2356	memleaks, such as leaking on C<map> and C<grep> but it is usually not as
		2357	pronounced).
		2358
612		2359
613	=head1 SEE ALSO	2360	=head1 SEE ALSO
614		2361
615	Event modules: L<Coro::Event>, L<Coro>, L<Event>, L<Glib::Event>, L<Glib>.	2362	Utility functions: L<AnyEvent::Util>.
616		2363
617	Implementations: L<AnyEvent::Impl::Coro>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>.	2364	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
		2365	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
618		2366
619	Nontrivial usage example: L<Net::FCP>.	2367	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
		2368	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
		2369	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
		2370	L<AnyEvent::Impl::POE>, L<AnyEvent::Impl::IOAsync>.
620		2371
621	=head1	2372	Non-blocking file handles, sockets, TCP clients and
		2373	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>, L<AnyEvent::TLS>.
		2374
		2375	Asynchronous DNS: L<AnyEvent::DNS>.
		2376
		2377	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>,
		2378	L<Coro::Event>,
		2379
		2380	Nontrivial usage examples: L<AnyEvent::GPSD>, L<AnyEvent::XMPP>,
		2381	L<AnyEvent::HTTP>.
		2382
		2383
		2384	=head1 AUTHOR
		2385
		2386	Marc Lehmann <schmorp@schmorp.de>
		2387	http://home.schmorp.de/
622		2388
623	=cut	2389	=cut
624		2390
625	1	2391	1
626		2392

Diff Legend

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