ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.13 by root, Thu Jul 20 08:26:00 2006 UTC vs.
Revision 1.196 by root, Thu Mar 26 07:47:42 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 - various supported event loops	5	EV, Event, Glib, Tk, Perl, Event::Lib, Qt, POE - various supported event loops
6		6
7	=head1 SYNOPSIS	7	=head1 SYNOPSIS
8		8
9	use AnyEvent;	9	use AnyEvent;
10		10
11	my $w = AnyEvent->io (fh => ..., poll => "[rw]+", cb => sub {	11	my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { ... });
12	my ($poll_got) = @_;	12
		13	my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
		14	my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...
		15
		16	print AnyEvent->now; # prints current event loop time
		17	print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
		18
		19	my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
		20
		21	my $w = AnyEvent->child (pid => $pid, cb => sub {
		22	my ($pid, $status) = @_;
13	...	23	...
14	});	24	});
15		25
16	* only one io watcher per $fh and $poll type is allowed (i.e. on a socket	26	my $w = AnyEvent->condvar; # stores whether a condition was flagged
17	you can have one r + one w or one rw watcher, not any more (limitation by	27	$w->send; # wake up current and all future recv's
18	Tk).
19
20	* the C<$poll_got> passed to the handler needs to be checked by looking
21	for single characters (e.g. with a regex), as it can contain more event
22	types than were requested (e.g. a 'w' watcher might generate 'rw' events,
23	limitation by Glib).
24
25	* AnyEvent will keep filehandles alive, so as long as the watcher exists,
26	the filehandle exists.
27
28	my $w = AnyEvent->timer (after => $seconds, cb => sub {
29	...
30	});
31
32	* io and time watchers get canceled whenever $w is destroyed, so keep a copy
33
34	* timers can only be used once and must be recreated for repeated
35	operation (limitation by Glib and Tk).
36
37	my $w = AnyEvent->condvar; # kind of main loop replacement
38	$w->wait; # enters main loop till $condvar gets ->broadcast	28	$w->recv; # enters "main loop" till $condvar gets ->send
39	$w->broadcast; # wake up current and all future wait's	29	# use a condvar in callback mode:
		30	$w->cb (sub { $_[0]->recv });
40		31
41	* condvars are used to give blocking behaviour when neccessary. Create	32	=head1 INTRODUCTION/TUTORIAL
42	a condvar for any "request" or "event" your module might create, C<<	33
43	->broadcast >> it when the event happens and provide a function that calls	34	This manpage is mainly a reference manual. If you are interested
44	C<< ->wait >> for it. See the examples below.	35	in a tutorial or some gentle introduction, have a look at the
		36	L<AnyEvent::Intro> manpage.
		37
		38	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
		39
		40	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
		41	nowadays. So what is different about AnyEvent?
		42
		43	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
		44	policy> and AnyEvent is I<small and efficient>.
		45
		46	First and foremost, I<AnyEvent is not an event model> itself, it only
		47	interfaces to whatever event model the main program happens to use, in a
		48	pragmatic way. For event models and certain classes of immortals alike,
		49	the statement "there can only be one" is a bitter reality: In general,
		50	only one event loop can be active at the same time in a process. AnyEvent
		51	cannot change this, but it can hide the differences between those event
		52	loops.
		53
		54	The goal of AnyEvent is to offer module authors the ability to do event
		55	programming (waiting for I/O or timer events) without subscribing to a
		56	religion, a way of living, and most importantly: without forcing your
		57	module users into the same thing by forcing them to use the same event
		58	model you use.
		59
		60	For modules like POE or IO::Async (which is a total misnomer as it is
		61	actually doing all I/O I<synchronously>...), using them in your module is
		62	like joining a cult: After you joined, you are dependent on them and you
		63	cannot use anything else, as they are simply incompatible to everything
		64	that isn't them. What's worse, all the potential users of your
		65	module are I<also> forced to use the same event loop you use.
		66
		67	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
		68	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
		69	with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
		70	your module uses one of those, every user of your module has to use it,
		71	too. But if your module uses AnyEvent, it works transparently with all
		72	event models it supports (including stuff like IO::Async, as long as those
		73	use one of the supported event loops. It is trivial to add new event loops
		74	to AnyEvent, too, so it is future-proof).
		75
		76	In addition to being free of having to use I<the one and only true event
		77	model>, AnyEvent also is free of bloat and policy: with POE or similar
		78	modules, you get an enormous amount of code and strict rules you have to
		79	follow. AnyEvent, on the other hand, is lean and up to the point, by only
		80	offering the functionality that is necessary, in as thin as a wrapper as
		81	technically possible.
		82
		83	Of course, AnyEvent comes with a big (and fully optional!) toolbox
		84	of useful functionality, such as an asynchronous DNS resolver, 100%
		85	non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
		86	such as Windows) and lots of real-world knowledge and workarounds for
		87	platform bugs and differences.
		88
		89	Now, if you I<do want> lots of policy (this can arguably be somewhat
		90	useful) and you want to force your users to use the one and only event
		91	model, you should I<not> use this module.
45		92
46	=head1 DESCRIPTION	93	=head1 DESCRIPTION
47		94
48	L<AnyEvent> provides an identical interface to multiple event loops. This	95	L<AnyEvent> provides an identical interface to multiple event loops. This
49	allows module authors to utilise an event loop without forcing module	96	allows module authors to utilise an event loop without forcing module
50	users to use the same event loop (as only a single event loop can coexist	97	users to use the same event loop (as only a single event loop can coexist
51	peacefully at any one time).	98	peacefully at any one time).
52		99
53	The interface itself is vaguely similar but not identical to the Event	100	The interface itself is vaguely similar, but not identical to the L<Event>
54	module.	101	module.
55		102
56	On the first call of any method, the module tries to detect the currently	103	During the first call of any watcher-creation method, the module tries
57	loaded event loop by probing wether any of the following modules is	104	to detect the currently loaded event loop by probing whether one of the
58	loaded: L<Coro::Event>, L<Event>, L<Glib>, L<Tk>. The first one found is	105	following modules is already loaded: L<EV>,
59	used. If none is found, the module tries to load these modules in the	106	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,
60	order given. The first one that could be successfully loaded will be	107	L<POE>. The first one found is used. If none are found, the module tries
61	used. If still none could be found, it will issue an error.	108	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
		109	adaptor should always succeed) in the order given. The first one that can
		110	be successfully loaded will be used. If, after this, still none could be
		111	found, AnyEvent will fall back to a pure-perl event loop, which is not
		112	very efficient, but should work everywhere.
		113
		114	Because AnyEvent first checks for modules that are already loaded, loading
		115	an event model explicitly before first using AnyEvent will likely make
		116	that model the default. For example:
		117
		118	use Tk;
		119	use AnyEvent;
		120
		121	# .. AnyEvent will likely default to Tk
		122
		123	The I<likely> means that, if any module loads another event model and
		124	starts using it, all bets are off. Maybe you should tell their authors to
		125	use AnyEvent so their modules work together with others seamlessly...
		126
		127	The pure-perl implementation of AnyEvent is called
		128	C<AnyEvent::Impl::Perl>. Like other event modules you can load it
		129	explicitly and enjoy the high availability of that event loop :)
		130
		131	=head1 WATCHERS
		132
		133	AnyEvent has the central concept of a I<watcher>, which is an object that
		134	stores relevant data for each kind of event you are waiting for, such as
		135	the callback to call, the file handle to watch, etc.
		136
		137	These watchers are normal Perl objects with normal Perl lifetime. After
		138	creating a watcher it will immediately "watch" for events and invoke the
		139	callback when the event occurs (of course, only when the event model
		140	is in control).
		141
		142	Note that B<callbacks must not permanently change global variables>
		143	potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
		144	callbacks must not C<die> >>. The former is good programming practise in
		145	Perl and the latter stems from the fact that exception handling differs
		146	widely between event loops.
		147
		148	To disable the watcher you have to destroy it (e.g. by setting the
		149	variable you store it in to C<undef> or otherwise deleting all references
		150	to it).
		151
		152	All watchers are created by calling a method on the C<AnyEvent> class.
		153
		154	Many watchers either are used with "recursion" (repeating timers for
		155	example), or need to refer to their watcher object in other ways.
		156
		157	An any way to achieve that is this pattern:
		158
		159	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
		160	# you can use $w here, for example to undef it
		161	undef $w;
		162	});
		163
		164	Note that C<my $w; $w => combination. This is necessary because in Perl,
		165	my variables are only visible after the statement in which they are
		166	declared.
		167
		168	=head2 I/O WATCHERS
		169
		170	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
		171	with the following mandatory key-value pairs as arguments:
		172
		173	C<fh> the Perl I<file handle> (I<not> file descriptor) to watch for events
		174	(AnyEvent might or might not keep a reference to this file handle). C<poll>
		175	must be a string that is either C<r> or C<w>, which creates a watcher
		176	waiting for "r"eadable or "w"ritable events, respectively. C<cb> is the
		177	callback to invoke each time the file handle becomes ready.
		178
		179	Although the callback might get passed parameters, their value and
		180	presence is undefined and you cannot rely on them. Portable AnyEvent
		181	callbacks cannot use arguments passed to I/O watcher callbacks.
		182
		183	The I/O watcher might use the underlying file descriptor or a copy of it.
		184	You must not close a file handle as long as any watcher is active on the
		185	underlying file descriptor.
		186
		187	Some event loops issue spurious readyness notifications, so you should
		188	always use non-blocking calls when reading/writing from/to your file
		189	handles.
		190
		191	Example: wait for readability of STDIN, then read a line and disable the
		192	watcher.
		193
		194	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
		195	chomp (my $input = <STDIN>);
		196	warn "read: $input\n";
		197	undef $w;
		198	});
		199
		200	=head2 TIME WATCHERS
		201
		202	You can create a time watcher by calling the C<< AnyEvent->timer >>
		203	method with the following mandatory arguments:
		204
		205	C<after> specifies after how many seconds (fractional values are
		206	supported) the callback should be invoked. C<cb> is the callback to invoke
		207	in that case.
		208
		209	Although the callback might get passed parameters, their value and
		210	presence is undefined and you cannot rely on them. Portable AnyEvent
		211	callbacks cannot use arguments passed to time watcher callbacks.
		212
		213	The callback will normally be invoked once only. If you specify another
		214	parameter, C<interval>, as a strictly positive number (> 0), then the
		215	callback will be invoked regularly at that interval (in fractional
		216	seconds) after the first invocation. If C<interval> is specified with a
		217	false value, then it is treated as if it were missing.
		218
		219	The callback will be rescheduled before invoking the callback, but no
		220	attempt is done to avoid timer drift in most backends, so the interval is
		221	only approximate.
		222
		223	Example: fire an event after 7.7 seconds.
		224
		225	my $w = AnyEvent->timer (after => 7.7, cb => sub {
		226	warn "timeout\n";
		227	});
		228
		229	# to cancel the timer:
		230	undef $w;
		231
		232	Example 2: fire an event after 0.5 seconds, then roughly every second.
		233
		234	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
		235	warn "timeout\n";
		236	};
		237
		238	=head3 TIMING ISSUES
		239
		240	There are two ways to handle timers: based on real time (relative, "fire
		241	in 10 seconds") and based on wallclock time (absolute, "fire at 12
		242	o'clock").
		243
		244	While most event loops expect timers to specified in a relative way, they
		245	use absolute time internally. This makes a difference when your clock
		246	"jumps", for example, when ntp decides to set your clock backwards from
		247	the wrong date of 2014-01-01 to 2008-01-01, a watcher that is supposed to
		248	fire "after" a second might actually take six years to finally fire.
		249
		250	AnyEvent cannot compensate for this. The only event loop that is conscious
		251	about these issues is L<EV>, which offers both relative (ev_timer, based
		252	on true relative time) and absolute (ev_periodic, based on wallclock time)
		253	timers.
		254
		255	AnyEvent always prefers relative timers, if available, matching the
		256	AnyEvent API.
		257
		258	AnyEvent has two additional methods that return the "current time":
62		259
63	=over 4	260	=over 4
64		261
		262	=item AnyEvent->time
		263
		264	This returns the "current wallclock time" as a fractional number of
		265	seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
		266	return, and the result is guaranteed to be compatible with those).
		267
		268	It progresses independently of any event loop processing, i.e. each call
		269	will check the system clock, which usually gets updated frequently.
		270
		271	=item AnyEvent->now
		272
		273	This also returns the "current wallclock time", but unlike C<time>, above,
		274	this value might change only once per event loop iteration, depending on
		275	the event loop (most return the same time as C<time>, above). This is the
		276	time that AnyEvent's timers get scheduled against.
		277
		278	I<In almost all cases (in all cases if you don't care), this is the
		279	function to call when you want to know the current time.>
		280
		281	This function is also often faster then C<< AnyEvent->time >>, and
		282	thus the preferred method if you want some timestamp (for example,
		283	L<AnyEvent::Handle> uses this to update it's activity timeouts).
		284
		285	The rest of this section is only of relevance if you try to be very exact
		286	with your timing, you can skip it without bad conscience.
		287
		288	For a practical example of when these times differ, consider L<Event::Lib>
		289	and L<EV> and the following set-up:
		290
		291	The event loop is running and has just invoked one of your callback at
		292	time=500 (assume no other callbacks delay processing). In your callback,
		293	you wait a second by executing C<sleep 1> (blocking the process for a
		294	second) and then (at time=501) you create a relative timer that fires
		295	after three seconds.
		296
		297	With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
		298	both return C<501>, because that is the current time, and the timer will
		299	be scheduled to fire at time=504 (C<501> + C<3>).
		300
		301	With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
		302	time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
		303	last event processing phase started. With L<EV>, your timer gets scheduled
		304	to run at time=503 (C<500> + C<3>).
		305
		306	In one sense, L<Event::Lib> is more exact, as it uses the current time
		307	regardless of any delays introduced by event processing. However, most
		308	callbacks do not expect large delays in processing, so this causes a
		309	higher drift (and a lot more system calls to get the current time).
		310
		311	In another sense, L<EV> is more exact, as your timer will be scheduled at
		312	the same time, regardless of how long event processing actually took.
		313
		314	In either case, if you care (and in most cases, you don't), then you
		315	can get whatever behaviour you want with any event loop, by taking the
		316	difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
		317	account.
		318
		319	=back
		320
		321	=head2 SIGNAL WATCHERS
		322
		323	You can watch for signals using a signal watcher, C<signal> is the signal
		324	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
		325	callback to be invoked whenever a signal occurs.
		326
		327	Although the callback might get passed parameters, their value and
		328	presence is undefined and you cannot rely on them. Portable AnyEvent
		329	callbacks cannot use arguments passed to signal watcher callbacks.
		330
		331	Multiple signal occurrences can be clumped together into one callback
		332	invocation, and callback invocation will be synchronous. Synchronous means
		333	that it might take a while until the signal gets handled by the process,
		334	but it is guaranteed not to interrupt any other callbacks.
		335
		336	The main advantage of using these watchers is that you can share a signal
		337	between multiple watchers.
		338
		339	This watcher might use C<%SIG>, so programs overwriting those signals
		340	directly will likely not work correctly.
		341
		342	Example: exit on SIGINT
		343
		344	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
		345
		346	=head2 CHILD PROCESS WATCHERS
		347
		348	You can also watch on a child process exit and catch its exit status.
		349
		350	The child process is specified by the C<pid> argument (if set to C<0>, it
		351	watches for any child process exit). The watcher will triggered only when
		352	the child process has finished and an exit status is available, not on
		353	any trace events (stopped/continued).
		354
		355	The callback will be called with the pid and exit status (as returned by
		356	waitpid), so unlike other watcher types, you I<can> rely on child watcher
		357	callback arguments.
		358
		359	This watcher type works by installing a signal handler for C<SIGCHLD>,
		360	and since it cannot be shared, nothing else should use SIGCHLD or reap
		361	random child processes (waiting for specific child processes, e.g. inside
		362	C<system>, is just fine).
		363
		364	There is a slight catch to child watchers, however: you usually start them
		365	I<after> the child process was created, and this means the process could
		366	have exited already (and no SIGCHLD will be sent anymore).
		367
		368	Not all event models handle this correctly (POE doesn't), but even for
		369	event models that I<do> handle this correctly, they usually need to be
		370	loaded before the process exits (i.e. before you fork in the first place).
		371
		372	This means you cannot create a child watcher as the very first thing in an
		373	AnyEvent program, you I<have> to create at least one watcher before you
		374	C<fork> the child (alternatively, you can call C<AnyEvent::detect>).
		375
		376	Example: fork a process and wait for it
		377
		378	my $done = AnyEvent->condvar;
		379
		380	my $pid = fork or exit 5;
		381
		382	my $w = AnyEvent->child (
		383	pid => $pid,
		384	cb => sub {
		385	my ($pid, $status) = @_;
		386	warn "pid $pid exited with status $status";
		387	$done->send;
		388	},
		389	);
		390
		391	# do something else, then wait for process exit
		392	$done->recv;
		393
		394	=head2 CONDITION VARIABLES
		395
		396	If you are familiar with some event loops you will know that all of them
		397	require you to run some blocking "loop", "run" or similar function that
		398	will actively watch for new events and call your callbacks.
		399
		400	AnyEvent is different, it expects somebody else to run the event loop and
		401	will only block when necessary (usually when told by the user).
		402
		403	The instrument to do that is called a "condition variable", so called
		404	because they represent a condition that must become true.
		405
		406	Condition variables can be created by calling the C<< AnyEvent->condvar
		407	>> method, usually without arguments. The only argument pair allowed is
		408
		409	C<cb>, which specifies a callback to be called when the condition variable
		410	becomes true, with the condition variable as the first argument (but not
		411	the results).
		412
		413	After creation, the condition variable is "false" until it becomes "true"
		414	by calling the C<send> method (or calling the condition variable as if it
		415	were a callback, read about the caveats in the description for the C<<
		416	->send >> method).
		417
		418	Condition variables are similar to callbacks, except that you can
		419	optionally wait for them. They can also be called merge points - points
		420	in time where multiple outstanding events have been processed. And yet
		421	another way to call them is transactions - each condition variable can be
		422	used to represent a transaction, which finishes at some point and delivers
		423	a result.
		424
		425	Condition variables are very useful to signal that something has finished,
		426	for example, if you write a module that does asynchronous http requests,
		427	then a condition variable would be the ideal candidate to signal the
		428	availability of results. The user can either act when the callback is
		429	called or can synchronously C<< ->recv >> for the results.
		430
		431	You can also use them to simulate traditional event loops - for example,
		432	you can block your main program until an event occurs - for example, you
		433	could C<< ->recv >> in your main program until the user clicks the Quit
		434	button of your app, which would C<< ->send >> the "quit" event.
		435
		436	Note that condition variables recurse into the event loop - if you have
		437	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
		438	lose. Therefore, condition variables are good to export to your caller, but
		439	you should avoid making a blocking wait yourself, at least in callbacks,
		440	as this asks for trouble.
		441
		442	Condition variables are represented by hash refs in perl, and the keys
		443	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		444	easy (it is often useful to build your own transaction class on top of
		445	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		446	it's C<new> method in your own C<new> method.
		447
		448	There are two "sides" to a condition variable - the "producer side" which
		449	eventually calls C<< -> send >>, and the "consumer side", which waits
		450	for the send to occur.
		451
		452	Example: wait for a timer.
		453
		454	# wait till the result is ready
		455	my $result_ready = AnyEvent->condvar;
		456
		457	# do something such as adding a timer
		458	# or socket watcher the calls $result_ready->send
		459	# when the "result" is ready.
		460	# in this case, we simply use a timer:
		461	my $w = AnyEvent->timer (
		462	after => 1,
		463	cb => sub { $result_ready->send },
		464	);
		465
		466	# this "blocks" (while handling events) till the callback
		467	# calls send
		468	$result_ready->recv;
		469
		470	Example: wait for a timer, but take advantage of the fact that
		471	condition variables are also code references.
		472
		473	my $done = AnyEvent->condvar;
		474	my $delay = AnyEvent->timer (after => 5, cb => $done);
		475	$done->recv;
		476
		477	Example: Imagine an API that returns a condvar and doesn't support
		478	callbacks. This is how you make a synchronous call, for example from
		479	the main program:
		480
		481	use AnyEvent::CouchDB;
		482
		483	...
		484
		485	my @info = $couchdb->info->recv;
		486
		487	And this is how you would just ste a callback to be called whenever the
		488	results are available:
		489
		490	$couchdb->info->cb (sub {
		491	my @info = $_[0]->recv;
		492	});
		493
		494	=head3 METHODS FOR PRODUCERS
		495
		496	These methods should only be used by the producing side, i.e. the
		497	code/module that eventually sends the signal. Note that it is also
		498	the producer side which creates the condvar in most cases, but it isn't
		499	uncommon for the consumer to create it as well.
		500
		501	=over 4
		502
		503	=item $cv->send (...)
		504
		505	Flag the condition as ready - a running C<< ->recv >> and all further
		506	calls to C<recv> will (eventually) return after this method has been
		507	called. If nobody is waiting the send will be remembered.
		508
		509	If a callback has been set on the condition variable, it is called
		510	immediately from within send.
		511
		512	Any arguments passed to the C<send> call will be returned by all
		513	future C<< ->recv >> calls.
		514
		515	Condition variables are overloaded so one can call them directly
		516	(as a code reference). Calling them directly is the same as calling
		517	C<send>. Note, however, that many C-based event loops do not handle
		518	overloading, so as tempting as it may be, passing a condition variable
		519	instead of a callback does not work. Both the pure perl and EV loops
		520	support overloading, however, as well as all functions that use perl to
		521	invoke a callback (as in L<AnyEvent::Socket> and L<AnyEvent::DNS> for
		522	example).
		523
		524	=item $cv->croak ($error)
		525
		526	Similar to send, but causes all call's to C<< ->recv >> to invoke
		527	C<Carp::croak> with the given error message/object/scalar.
		528
		529	This can be used to signal any errors to the condition variable
		530	user/consumer.
		531
		532	=item $cv->begin ([group callback])
		533
		534	=item $cv->end
		535
		536	These two methods are EXPERIMENTAL and MIGHT CHANGE.
		537
		538	These two methods can be used to combine many transactions/events into
		539	one. For example, a function that pings many hosts in parallel might want
		540	to use a condition variable for the whole process.
		541
		542	Every call to C<< ->begin >> will increment a counter, and every call to
		543	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		544	>>, the (last) callback passed to C<begin> will be executed. That callback
		545	is I<supposed> to call C<< ->send >>, but that is not required. If no
		546	callback was set, C<send> will be called without any arguments.
		547
		548	Let's clarify this with the ping example:
		549
		550	my $cv = AnyEvent->condvar;
		551
		552	my %result;
		553	$cv->begin (sub { $cv->send (\%result) });
		554
		555	for my $host (@list_of_hosts) {
		556	$cv->begin;
		557	ping_host_then_call_callback $host, sub {
		558	$result{$host} = ...;
		559	$cv->end;
		560	};
		561	}
		562
		563	$cv->end;
		564
		565	This code fragment supposedly pings a number of hosts and calls
		566	C<send> after results for all then have have been gathered - in any
		567	order. To achieve this, the code issues a call to C<begin> when it starts
		568	each ping request and calls C<end> when it has received some result for
		569	it. Since C<begin> and C<end> only maintain a counter, the order in which
		570	results arrive is not relevant.
		571
		572	There is an additional bracketing call to C<begin> and C<end> outside the
		573	loop, which serves two important purposes: first, it sets the callback
		574	to be called once the counter reaches C<0>, and second, it ensures that
		575	C<send> is called even when C<no> hosts are being pinged (the loop
		576	doesn't execute once).
		577
		578	This is the general pattern when you "fan out" into multiple subrequests:
		579	use an outer C<begin>/C<end> pair to set the callback and ensure C<end>
		580	is called at least once, and then, for each subrequest you start, call
		581	C<begin> and for each subrequest you finish, call C<end>.
		582
		583	=back
		584
		585	=head3 METHODS FOR CONSUMERS
		586
		587	These methods should only be used by the consuming side, i.e. the
		588	code awaits the condition.
		589
		590	=over 4
		591
		592	=item $cv->recv
		593
		594	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
		595	>> methods have been called on c<$cv>, while servicing other watchers
		596	normally.
		597
		598	You can only wait once on a condition - additional calls are valid but
		599	will return immediately.
		600
		601	If an error condition has been set by calling C<< ->croak >>, then this
		602	function will call C<croak>.
		603
		604	In list context, all parameters passed to C<send> will be returned,
		605	in scalar context only the first one will be returned.
		606
		607	Not all event models support a blocking wait - some die in that case
		608	(programs might want to do that to stay interactive), so I<if you are
		609	using this from a module, never require a blocking wait>, but let the
		610	caller decide whether the call will block or not (for example, by coupling
		611	condition variables with some kind of request results and supporting
		612	callbacks so the caller knows that getting the result will not block,
		613	while still supporting blocking waits if the caller so desires).
		614
		615	Another reason I<never> to C<< ->recv >> in a module is that you cannot
		616	sensibly have two C<< ->recv >>'s in parallel, as that would require
		617	multiple interpreters or coroutines/threads, none of which C<AnyEvent>
		618	can supply.
		619
		620	The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
		621	fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
		622	versions and also integrates coroutines into AnyEvent, making blocking
		623	C<< ->recv >> calls perfectly safe as long as they are done from another
		624	coroutine (one that doesn't run the event loop).
		625
		626	You can ensure that C<< -recv >> never blocks by setting a callback and
		627	only calling C<< ->recv >> from within that callback (or at a later
		628	time). This will work even when the event loop does not support blocking
		629	waits otherwise.
		630
		631	=item $bool = $cv->ready
		632
		633	Returns true when the condition is "true", i.e. whether C<send> or
		634	C<croak> have been called.
		635
		636	=item $cb = $cv->cb ($cb->($cv))
		637
		638	This is a mutator function that returns the callback set and optionally
		639	replaces it before doing so.
		640
		641	The callback will be called when the condition becomes "true", i.e. when
		642	C<send> or C<croak> are called, with the only argument being the condition
		643	variable itself. Calling C<recv> inside the callback or at any later time
		644	is guaranteed not to block.
		645
		646	=back
		647
		648	=head1 GLOBAL VARIABLES AND FUNCTIONS
		649
		650	=over 4
		651
		652	=item $AnyEvent::MODEL
		653
		654	Contains C<undef> until the first watcher is being created. Then it
		655	contains the event model that is being used, which is the name of the
		656	Perl class implementing the model. This class is usually one of the
		657	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case
		658	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).
		659
		660	The known classes so far are:
		661
		662	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
		663	AnyEvent::Impl::Event based on Event, second best choice.
		664	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
		665	AnyEvent::Impl::Glib based on Glib, third-best choice.
		666	AnyEvent::Impl::Tk based on Tk, very bad choice.
		667	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
		668	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
		669	AnyEvent::Impl::POE based on POE, not generic enough for full support.
		670
		671	There is no support for WxWidgets, as WxWidgets has no support for
		672	watching file handles. However, you can use WxWidgets through the
		673	POE Adaptor, as POE has a Wx backend that simply polls 20 times per
		674	second, which was considered to be too horrible to even consider for
		675	AnyEvent. Likewise, other POE backends can be used by AnyEvent by using
		676	it's adaptor.
		677
		678	AnyEvent knows about L<Prima> and L<Wx> and will try to use L<POE> when
		679	autodetecting them.
		680
		681	=item AnyEvent::detect
		682
		683	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
		684	if necessary. You should only call this function right before you would
		685	have created an AnyEvent watcher anyway, that is, as late as possible at
		686	runtime.
		687
		688	=item $guard = AnyEvent::post_detect { BLOCK }
		689
		690	Arranges for the code block to be executed as soon as the event model is
		691	autodetected (or immediately if this has already happened).
		692
		693	If called in scalar or list context, then it creates and returns an object
		694	that automatically removes the callback again when it is destroyed. See
		695	L<Coro::BDB> for a case where this is useful.
		696
		697	=item @AnyEvent::post_detect
		698
		699	If there are any code references in this array (you can C<push> to it
		700	before or after loading AnyEvent), then they will called directly after
		701	the event loop has been chosen.
		702
		703	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		704	if it contains a true value then the event loop has already been detected,
		705	and the array will be ignored.
		706
		707	Best use C<AnyEvent::post_detect { BLOCK }> instead.
		708
		709	=back
		710
		711	=head1 WHAT TO DO IN A MODULE
		712
		713	As a module author, you should C<use AnyEvent> and call AnyEvent methods
		714	freely, but you should not load a specific event module or rely on it.
		715
		716	Be careful when you create watchers in the module body - AnyEvent will
		717	decide which event module to use as soon as the first method is called, so
		718	by calling AnyEvent in your module body you force the user of your module
		719	to load the event module first.
		720
		721	Never call C<< ->recv >> on a condition variable unless you I<know> that
		722	the C<< ->send >> method has been called on it already. This is
		723	because it will stall the whole program, and the whole point of using
		724	events is to stay interactive.
		725
		726	It is fine, however, to call C<< ->recv >> when the user of your module
		727	requests it (i.e. if you create a http request object ad have a method
		728	called C<results> that returns the results, it should call C<< ->recv >>
		729	freely, as the user of your module knows what she is doing. always).
		730
		731	=head1 WHAT TO DO IN THE MAIN PROGRAM
		732
		733	There will always be a single main program - the only place that should
		734	dictate which event model to use.
		735
		736	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
		737	do anything special (it does not need to be event-based) and let AnyEvent
		738	decide which implementation to chose if some module relies on it.
		739
		740	If the main program relies on a specific event model - for example, in
		741	Gtk2 programs you have to rely on the Glib module - you should load the
		742	event module before loading AnyEvent or any module that uses it: generally
		743	speaking, you should load it as early as possible. The reason is that
		744	modules might create watchers when they are loaded, and AnyEvent will
		745	decide on the event model to use as soon as it creates watchers, and it
		746	might chose the wrong one unless you load the correct one yourself.
		747
		748	You can chose to use a pure-perl implementation by loading the
		749	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
		750	everywhere, but letting AnyEvent chose the model is generally better.
		751
		752	=head2 MAINLOOP EMULATION
		753
		754	Sometimes (often for short test scripts, or even standalone programs who
		755	only want to use AnyEvent), you do not want to run a specific event loop.
		756
		757	In that case, you can use a condition variable like this:
		758
		759	AnyEvent->condvar->recv;
		760
		761	This has the effect of entering the event loop and looping forever.
		762
		763	Note that usually your program has some exit condition, in which case
		764	it is better to use the "traditional" approach of storing a condition
		765	variable somewhere, waiting for it, and sending it when the program should
		766	exit cleanly.
		767
		768
		769	=head1 OTHER MODULES
		770
		771	The following is a non-exhaustive list of additional modules that use
		772	AnyEvent and can therefore be mixed easily with other AnyEvent modules
		773	in the same program. Some of the modules come with AnyEvent, some are
		774	available via CPAN.
		775
		776	=over 4
		777
		778	=item L<AnyEvent::Util>
		779
		780	Contains various utility functions that replace often-used but blocking
		781	functions such as C<inet_aton> by event-/callback-based versions.
		782
		783	=item L<AnyEvent::Socket>
		784
		785	Provides various utility functions for (internet protocol) sockets,
		786	addresses and name resolution. Also functions to create non-blocking tcp
		787	connections or tcp servers, with IPv6 and SRV record support and more.
		788
		789	=item L<AnyEvent::Handle>
		790
		791	Provide read and write buffers, manages watchers for reads and writes,
		792	supports raw and formatted I/O, I/O queued and fully transparent and
		793	non-blocking SSL/TLS.
		794
		795	=item L<AnyEvent::DNS>
		796
		797	Provides rich asynchronous DNS resolver capabilities.
		798
		799	=item L<AnyEvent::HTTP>
		800
		801	A simple-to-use HTTP library that is capable of making a lot of concurrent
		802	HTTP requests.
		803
		804	=item L<AnyEvent::HTTPD>
		805
		806	Provides a simple web application server framework.
		807
		808	=item L<AnyEvent::FastPing>
		809
		810	The fastest ping in the west.
		811
		812	=item L<AnyEvent::DBI>
		813
		814	Executes L<DBI> requests asynchronously in a proxy process.
		815
		816	=item L<AnyEvent::AIO>
		817
		818	Truly asynchronous I/O, should be in the toolbox of every event
		819	programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
		820	together.
		821
		822	=item L<AnyEvent::BDB>
		823
		824	Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
		825	L<BDB> and AnyEvent together.
		826
		827	=item L<AnyEvent::GPSD>
		828
		829	A non-blocking interface to gpsd, a daemon delivering GPS information.
		830
		831	=item L<AnyEvent::IGS>
		832
		833	A non-blocking interface to the Internet Go Server protocol (used by
		834	L<App::IGS>).
		835
		836	=item L<AnyEvent::IRC>
		837
		838	AnyEvent based IRC client module family (replacing the older Net::IRC3).
		839
		840	=item L<Net::XMPP2>
		841
		842	AnyEvent based XMPP (Jabber protocol) module family.
		843
		844	=item L<Net::FCP>
		845
		846	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		847	of AnyEvent.
		848
		849	=item L<Event::ExecFlow>
		850
		851	High level API for event-based execution flow control.
		852
		853	=item L<Coro>
		854
		855	Has special support for AnyEvent via L<Coro::AnyEvent>.
		856
		857	=item L<IO::Lambda>
		858
		859	The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
		860
		861	=back
		862
65	=cut	863	=cut
66		864
67	package AnyEvent;	865	package AnyEvent;
68		866
69	no warnings;	867	no warnings;
70	use strict 'vars';	868	use strict qw(vars subs);
		869
71	use Carp;	870	use Carp;
72		871
73	our $VERSION = '1.02';	872	our $VERSION = 4.341;
74	our $MODEL;	873	our $MODEL;
75		874
76	our $AUTOLOAD;	875	our $AUTOLOAD;
77	our @ISA;	876	our @ISA;
78		877
		878	our @REGISTRY;
		879
		880	our $WIN32;
		881
		882	BEGIN {
		883	my $win32 = ! ! ($^O =~ /mswin32/i);
		884	eval "sub WIN32(){ $win32 }";
		885	}
		886
79	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;	887	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
80		888
81	our @REGISTRY;	889	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		890
		891	{
		892	my $idx;
		893	$PROTOCOL{$_} = ++$idx
		894	for reverse split /\s*,\s*/,
		895	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		896	}
82		897
83	my @models = (	898	my @models = (
84	[Coro::Event:: => AnyEvent::Impl::Coro::],	899	[EV:: => AnyEvent::Impl::EV::],
85	[Event:: => AnyEvent::Impl::Event::],	900	[Event:: => AnyEvent::Impl::Event::],
86	[Glib:: => AnyEvent::Impl::Glib::],	901	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
		902	# everything below here will not be autoprobed
		903	# as the pureperl backend should work everywhere
		904	# and is usually faster
		905	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		906	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
		907	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
		908	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
		909	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
87	[Tk:: => AnyEvent::Impl::Tk::],	910	[Wx:: => AnyEvent::Impl::POE::],
		911	[Prima:: => AnyEvent::Impl::POE::],
88	);	912	);
89		913
90	our %method = map +($_ => 1), qw(io timer condvar broadcast wait cancel DESTROY);	914	our %method = map +($_ => 1), qw(io timer time now signal child condvar one_event DESTROY);
91		915
92	sub AUTOLOAD {	916	our @post_detect;
93	$AUTOLOAD =~ s/.*://;
94		917
95	$method{$AUTOLOAD}	918	sub post_detect(&) {
96	or croak "$AUTOLOAD: not a valid method for AnyEvent objects";	919	my ($cb) = @_;
97		920
		921	if ($MODEL) {
		922	$cb->();
		923
		924	1
		925	} else {
		926	push @post_detect, $cb;
		927
		928	defined wantarray
		929	? bless \$cb, "AnyEvent::Util::PostDetect"
		930	: ()
		931	}
		932	}
		933
		934	sub AnyEvent::Util::PostDetect::DESTROY {
		935	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		936	}
		937
		938	sub detect() {
98	unless ($MODEL) {	939	unless ($MODEL) {
		940	no strict 'refs';
		941	local $SIG{__DIE__};
		942
		943	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
		944	my $model = "AnyEvent::Impl::$1";
		945	if (eval "require $model") {
		946	$MODEL = $model;
		947	warn "AnyEvent: loaded model '$model' (forced by \$PERL_ANYEVENT_MODEL), using it.\n" if $verbose > 1;
		948	} else {
		949	warn "AnyEvent: unable to load model '$model' (from \$PERL_ANYEVENT_MODEL):\n$@" if $verbose;
		950	}
		951	}
		952
99	# check for already loaded models	953	# check for already loaded models
		954	unless ($MODEL) {
100	for (@REGISTRY, @models) {	955	for (@REGISTRY, @models) {
101	my ($package, $model) = @$_;	956	my ($package, $model) = @$_;
102	if (${"$package\::VERSION"} > 0) {	957	if (${"$package\::VERSION"} > 0) {
103	if (eval "require $model") {	958	if (eval "require $model") {
104	$MODEL = $model;	959	$MODEL = $model;
105	warn "AnyEvent: found model '$model', using it.\n" if $verbose > 1;	960	warn "AnyEvent: autodetected model '$model', using it.\n" if $verbose > 1;
106	last;	961	last;
		962	}
107	}	963	}
108	}	964	}
		965
		966	unless ($MODEL) {
		967	# try to load a model
		968
		969	for (@REGISTRY, @models) {
		970	my ($package, $model) = @$_;
		971	if (eval "require $package"
		972	and ${"$package\::VERSION"} > 0
		973	and eval "require $model") {
		974	$MODEL = $model;
		975	warn "AnyEvent: autoprobed model '$model', using it.\n" if $verbose > 1;
		976	last;
		977	}
		978	}
		979
		980	$MODEL
		981	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.";
		982	}
109	}	983	}
110		984
111	unless ($MODEL) {	985	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
112	# try to load a model
113		986
114	for (@REGISTRY, @models) {	987	unshift @ISA, $MODEL;
115	my ($package, $model) = @$_;
116	if (eval "require $model") {
117	$MODEL = $model;
118	warn "AnyEvent: autoprobed and loaded model '$model', using it.\n" if $verbose > 1;
119	last;
120	}
121	}
122		988
		989	require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
		990
		991	(shift @post_detect)->() while @post_detect;
		992	}
		993
123	$MODEL	994	$MODEL
124	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: Coro, Event, Glib or Tk.";	995	}
		996
		997	sub AUTOLOAD {
		998	(my $func = $AUTOLOAD) =~ s/.*://;
		999
		1000	$method{$func}
		1001	or croak "$func: not a valid method for AnyEvent objects";
		1002
		1003	detect unless $MODEL;
		1004
		1005	my $class = shift;
		1006	$class->$func (@_);
		1007	}
		1008
		1009	# utility function to dup a filehandle. this is used by many backends
		1010	# to support binding more than one watcher per filehandle (they usually
		1011	# allow only one watcher per fd, so we dup it to get a different one).
		1012	sub _dupfh($$$$) {
		1013	my ($poll, $fh, $r, $w) = @_;
		1014
		1015	# cygwin requires the fh mode to be matching, unix doesn't
		1016	my ($rw, $mode) = $poll eq "r" ? ($r, "<")
		1017	: $poll eq "w" ? ($w, ">")
		1018	: Carp::croak "AnyEvent->io requires poll set to either 'r' or 'w'";
		1019
		1020	open my $fh2, "$mode&" . fileno $fh
		1021	or die "cannot dup() filehandle: $!";
		1022
		1023	# we assume CLOEXEC is already set by perl in all important cases
		1024
		1025	($fh2, $rw)
		1026	}
		1027
		1028	package AnyEvent::Base;
		1029
		1030	# default implementation for now and time
		1031
		1032	BEGIN {
		1033	if (eval "use Time::HiRes (); time (); 1") {
		1034	*_time = \&Time::HiRes::time;
		1035	# if (eval "use POSIX (); (POSIX::times())...
		1036	} else {
		1037	*_time = sub { time }; # epic fail
		1038	}
		1039	}
		1040
		1041	sub time { _time }
		1042	sub now { _time }
		1043
		1044	# default implementation for ->condvar
		1045
		1046	sub condvar {
		1047	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar::
		1048	}
		1049
		1050	# default implementation for ->signal
		1051
		1052	our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
		1053
		1054	sub _signal_exec {
		1055	while (%SIG_EV) {
		1056	sysread $SIGPIPE_R, my $dummy, 4;
		1057	for (keys %SIG_EV) {
		1058	delete $SIG_EV{$_};
		1059	$_->() for values %{ $SIG_CB{$_} || {} };
125	}	1060	}
126	}	1061	}
		1062	}
127		1063
128	@ISA = $MODEL;	1064	sub signal {
		1065	my (undef, %arg) = @_;
129		1066
		1067	unless ($SIGPIPE_R) {
		1068	if (AnyEvent::WIN32) {
		1069	($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
		1070	AnyEvent::Util::fh_nonblocking ($SIGPIPE_R) if $SIGPIPE_R;
		1071	AnyEvent::Util::fh_nonblocking ($SIGPIPE_W) if $SIGPIPE_W; # just in case
		1072	} else {
		1073	pipe $SIGPIPE_R, $SIGPIPE_W;
		1074	require Fcntl;
		1075	fcntl $SIGPIPE_R, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_R;
		1076	fcntl $SIGPIPE_W, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_W; # just in case
		1077	}
		1078
		1079	$SIGPIPE_R
		1080	or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
		1081
		1082	$SIG_IO = AnyEvent->io (fh => $SIGPIPE_R, poll => "r", cb => \&_signal_exec);
		1083	}
		1084
		1085	my $signal = uc $arg{signal}
		1086	or Carp::croak "required option 'signal' is missing";
		1087
		1088	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1089	$SIG{$signal} ||= sub {
		1090	syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
		1091	undef $SIG_EV{$signal};
		1092	};
		1093
		1094	bless [$signal, $arg{cb}], "AnyEvent::Base::Signal"
		1095	}
		1096
		1097	sub AnyEvent::Base::Signal::DESTROY {
		1098	my ($signal, $cb) = @{$_[0]};
		1099
		1100	delete $SIG_CB{$signal}{$cb};
		1101
		1102	delete $SIG{$signal} unless keys %{ $SIG_CB{$signal} };
		1103	}
		1104
		1105	# default implementation for ->child
		1106
		1107	our %PID_CB;
		1108	our $CHLD_W;
		1109	our $CHLD_DELAY_W;
		1110	our $PID_IDLE;
		1111	our $WNOHANG;
		1112
		1113	sub _child_wait {
		1114	while (0 < (my $pid = waitpid -1, $WNOHANG)) {
		1115	$_->($pid, $?) for (values %{ $PID_CB{$pid} || {} }),
		1116	(values %{ $PID_CB{0} || {} });
		1117	}
		1118
		1119	undef $PID_IDLE;
		1120	}
		1121
		1122	sub _sigchld {
		1123	# make sure we deliver these changes "synchronous" with the event loop.
		1124	$CHLD_DELAY_W ||= AnyEvent->timer (after => 0, cb => sub {
		1125	undef $CHLD_DELAY_W;
		1126	&_child_wait;
		1127	});
		1128	}
		1129
		1130	sub child {
		1131	my (undef, %arg) = @_;
		1132
		1133	defined (my $pid = $arg{pid} + 0)
		1134	or Carp::croak "required option 'pid' is missing";
		1135
		1136	$PID_CB{$pid}{$arg{cb}} = $arg{cb};
		1137
		1138	unless ($WNOHANG) {
		1139	$WNOHANG = eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
		1140	}
		1141
		1142	unless ($CHLD_W) {
		1143	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
		1144	# child could be a zombie already, so make at least one round
		1145	&_sigchld;
		1146	}
		1147
		1148	bless [$pid, $arg{cb}], "AnyEvent::Base::Child"
		1149	}
		1150
		1151	sub AnyEvent::Base::Child::DESTROY {
		1152	my ($pid, $cb) = @{$_[0]};
		1153
		1154	delete $PID_CB{$pid}{$cb};
		1155	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
		1156
		1157	undef $CHLD_W unless keys %PID_CB;
		1158	}
		1159
		1160	package AnyEvent::CondVar;
		1161
		1162	our @ISA = AnyEvent::CondVar::Base::;
		1163
		1164	package AnyEvent::CondVar::Base;
		1165
		1166	use overload
		1167	'&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1168	fallback => 1;
		1169
		1170	sub _send {
		1171	# nop
		1172	}
		1173
		1174	sub send {
130	my $class = shift;	1175	my $cv = shift;
131	$class->$AUTOLOAD (@_);	1176	$cv->{_ae_sent} = [@_];
		1177	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1178	$cv->_send;
132	}	1179	}
		1180
		1181	sub croak {
		1182	$_[0]{_ae_croak} = $_[1];
		1183	$_[0]->send;
		1184	}
		1185
		1186	sub ready {
		1187	$_[0]{_ae_sent}
		1188	}
		1189
		1190	sub _wait {
		1191	AnyEvent->one_event while !$_[0]{_ae_sent};
		1192	}
		1193
		1194	sub recv {
		1195	$_[0]->_wait;
		1196
		1197	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		1198	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		1199	}
		1200
		1201	sub cb {
		1202	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		1203	$_[0]{_ae_cb}
		1204	}
		1205
		1206	sub begin {
		1207	++$_[0]{_ae_counter};
		1208	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		1209	}
		1210
		1211	sub end {
		1212	return if --$_[0]{_ae_counter};
		1213	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		1214	}
		1215
		1216	# undocumented/compatibility with pre-3.4
		1217	*broadcast = \&send;
		1218	*wait = \&_wait;
		1219
		1220	=head1 ERROR AND EXCEPTION HANDLING
		1221
		1222	In general, AnyEvent does not do any error handling - it relies on the
		1223	caller to do that if required. The L<AnyEvent::Strict> module (see also
		1224	the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
		1225	checking of all AnyEvent methods, however, which is highly useful during
		1226	development.
		1227
		1228	As for exception handling (i.e. runtime errors and exceptions thrown while
		1229	executing a callback), this is not only highly event-loop specific, but
		1230	also not in any way wrapped by this module, as this is the job of the main
		1231	program.
		1232
		1233	The pure perl event loop simply re-throws the exception (usually
		1234	within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
		1235	$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
		1236	so on.
		1237
		1238	=head1 ENVIRONMENT VARIABLES
		1239
		1240	The following environment variables are used by this module or its
		1241	submodules:
		1242
		1243	=over 4
		1244
		1245	=item C<PERL_ANYEVENT_VERBOSE>
		1246
		1247	By default, AnyEvent will be completely silent except in fatal
		1248	conditions. You can set this environment variable to make AnyEvent more
		1249	talkative.
		1250
		1251	When set to C<1> or higher, causes AnyEvent to warn about unexpected
		1252	conditions, such as not being able to load the event model specified by
		1253	C<PERL_ANYEVENT_MODEL>.
		1254
		1255	When set to C<2> or higher, cause AnyEvent to report to STDERR which event
		1256	model it chooses.
		1257
		1258	=item C<PERL_ANYEVENT_STRICT>
		1259
		1260	AnyEvent does not do much argument checking by default, as thorough
		1261	argument checking is very costly. Setting this variable to a true value
		1262	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		1263	check the arguments passed to most method calls. If it finds any problems
		1264	it will croak.
		1265
		1266	In other words, enables "strict" mode.
		1267
		1268	Unlike C<use strict>, it is definitely recommended ot keep it off in
		1269	production. Keeping C<PERL_ANYEVENT_STRICT=1> in your environment while
		1270	developing programs can be very useful, however.
		1271
		1272	=item C<PERL_ANYEVENT_MODEL>
		1273
		1274	This can be used to specify the event model to be used by AnyEvent, before
		1275	auto detection and -probing kicks in. It must be a string consisting
		1276	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
		1277	and the resulting module name is loaded and if the load was successful,
		1278	used as event model. If it fails to load AnyEvent will proceed with
		1279	auto detection and -probing.
		1280
		1281	This functionality might change in future versions.
		1282
		1283	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
		1284	could start your program like this:
		1285
		1286	PERL_ANYEVENT_MODEL=Perl perl ...
		1287
		1288	=item C<PERL_ANYEVENT_PROTOCOLS>
		1289
		1290	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1291	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1292	of auto probing).
		1293
		1294	Must be set to a comma-separated list of protocols or address families,
		1295	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1296	used, and preference will be given to protocols mentioned earlier in the
		1297	list.
		1298
		1299	This variable can effectively be used for denial-of-service attacks
		1300	against local programs (e.g. when setuid), although the impact is likely
		1301	small, as the program has to handle conenction and other failures anyways.
		1302
		1303	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1304	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1305	- only support IPv4, never try to resolve or contact IPv6
		1306	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1307	IPv6, but prefer IPv6 over IPv4.
		1308
		1309	=item C<PERL_ANYEVENT_EDNS0>
		1310
		1311	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1312	for DNS. This extension is generally useful to reduce DNS traffic, but
		1313	some (broken) firewalls drop such DNS packets, which is why it is off by
		1314	default.
		1315
		1316	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1317	EDNS0 in its DNS requests.
		1318
		1319	=item C<PERL_ANYEVENT_MAX_FORKS>
		1320
		1321	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		1322	will create in parallel.
133		1323
134	=back	1324	=back
135		1325
136	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1326	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
		1327
		1328	This is an advanced topic that you do not normally need to use AnyEvent in
		1329	a module. This section is only of use to event loop authors who want to
		1330	provide AnyEvent compatibility.
137		1331
138	If you need to support another event library which isn't directly	1332	If you need to support another event library which isn't directly
139	supported by AnyEvent, you can supply your own interface to it by	1333	supported by AnyEvent, you can supply your own interface to it by
140	pushing, before the first watcher gets created, the package name of	1334	pushing, before the first watcher gets created, the package name of
141	the event module and the package name of the interface to use onto	1335	the event module and the package name of the interface to use onto
142	C<@AnyEvent::REGISTRY>. You can do that before and even without loading	1336	C<@AnyEvent::REGISTRY>. You can do that before and even without loading
143	AnyEvent.	1337	AnyEvent, so it is reasonably cheap.
144		1338
145	Example:	1339	Example:
146		1340
147	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];	1341	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];
148		1342
149	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>	1343	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>
150	package/class when it finds the C<urxvt> package/module is loaded. When	1344	package/class when it finds the C<urxvt> package/module is already loaded.
		1345
151	AnyEvent is loaded and asked to find a suitable event model, it will	1346	When AnyEvent is loaded and asked to find a suitable event model, it
152	first check for the presence of urxvt.	1347	will first check for the presence of urxvt by trying to C<use> the
		1348	C<urxvt::anyevent> module.
153		1349
154	The class should prove implementations for all watcher types (see	1350	The class should provide implementations for all watcher types. See
155	L<AnyEvent::Impl::Event> (source code), L<AnyEvent::Impl::Glib>	1351	L<AnyEvent::Impl::EV> (source code), L<AnyEvent::Impl::Glib> (Source code)
156	(Source code) and so on for actual examples, use C<perldoc -m	1352	and so on for actual examples. Use C<perldoc -m AnyEvent::Impl::Glib> to
157	AnyEvent::Impl::Glib> to see the sources).	1353	see the sources.
158		1354
		1355	If you don't provide C<signal> and C<child> watchers than AnyEvent will
		1356	provide suitable (hopefully) replacements.
		1357
159	The above isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)	1358	The above example isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)
160	uses the above line as-is. An interface isn't included in AnyEvent	1359	terminal emulator uses the above line as-is. An interface isn't included
161	because it doesn't make sense outside the embedded interpreter inside	1360	in AnyEvent because it doesn't make sense outside the embedded interpreter
162	I<rxvt-unicode>, and it is updated and maintained as part of the	1361	inside I<rxvt-unicode>, and it is updated and maintained as part of the
163	I<rxvt-unicode> distribution.	1362	I<rxvt-unicode> distribution.
164		1363
165	I<rxvt-unicode> also cheats a bit by not providing blocking access to	1364	I<rxvt-unicode> also cheats a bit by not providing blocking access to
166	condition variables: code blocking while waiting for a condition will	1365	condition variables: code blocking while waiting for a condition will
167	C<die>. This still works with most modules/usages, and blocking calls must	1366	C<die>. This still works with most modules/usages, and blocking calls must
168	not be in an interactive appliation, so it makes sense.	1367	not be done in an interactive application, so it makes sense.
169		1368
170	=head1 ENVIRONMENT VARIABLES
171
172	The following environment variables are used by this module:
173
174	C<PERL_ANYEVENT_VERBOSE> when set to C<2> or higher, reports which event
175	model gets used.
176
177	=head1 EXAMPLE	1369	=head1 EXAMPLE PROGRAM
178		1370
179	The following program uses an io watcher to read data from stdin, a timer	1371	The following program uses an I/O watcher to read data from STDIN, a timer
180	to display a message once per second, and a condvar to exit the program	1372	to display a message once per second, and a condition variable to quit the
181	when the user enters quit:	1373	program when the user enters quit:
182		1374
183	use AnyEvent;	1375	use AnyEvent;
184		1376
185	my $cv = AnyEvent->condvar;	1377	my $cv = AnyEvent->condvar;
186		1378
187	my $io_watcher = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	1379	my $io_watcher = AnyEvent->io (
		1380	fh => \*STDIN,
		1381	poll => 'r',
		1382	cb => sub {
188	warn "io event <$_[0]>\n"; # will always output <r>	1383	warn "io event <$_[0]>\n"; # will always output <r>
189	chomp (my $input = <STDIN>); # read a line	1384	chomp (my $input = <STDIN>); # read a line
190	warn "read: $input\n"; # output what has been read	1385	warn "read: $input\n"; # output what has been read
191	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1386	$cv->send if $input =~ /^q/i; # quit program if /^q/i
		1387	},
192	});	1388	);
193		1389
194	my $time_watcher; # can only be used once	1390	my $time_watcher; # can only be used once
195		1391
196	sub new_timer {	1392	sub new_timer {
197	$timer = AnyEvent->timer (after => 1, cb => sub {	1393	$timer = AnyEvent->timer (after => 1, cb => sub {
…		…
200	});	1396	});
201	}	1397	}
202		1398
203	new_timer; # create first timer	1399	new_timer; # create first timer
204		1400
205	$cv->wait; # wait until user enters /^q/i	1401	$cv->recv; # wait until user enters /^q/i
206		1402
207	=head1 REAL-WORLD EXAMPLE	1403	=head1 REAL-WORLD EXAMPLE
208		1404
209	Consider the L<Net::FCP> module. It features (among others) the following	1405	Consider the L<Net::FCP> module. It features (among others) the following
210	API calls, which are to freenet what HTTP GET requests are to http:	1406	API calls, which are to freenet what HTTP GET requests are to http:
…		…
260	syswrite $txn->{fh}, $txn->{request}	1456	syswrite $txn->{fh}, $txn->{request}
261	or die "connection or write error";	1457	or die "connection or write error";
262	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	1458	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
263		1459
264	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	1460	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
265	result and signals any possible waiters that the request ahs finished:	1461	result and signals any possible waiters that the request has finished:
266		1462
267	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1463	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
268		1464
269	if (end-of-file or data complete) {	1465	if (end-of-file or data complete) {
270	$txn->{result} = $txn->{buf};	1466	$txn->{result} = $txn->{buf};
271	$txn->{finished}->broadcast;	1467	$txn->{finished}->send;
272	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1468	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
273	}	1469	}
274		1470
275	The C<result> method, finally, just waits for the finished signal (if the	1471	The C<result> method, finally, just waits for the finished signal (if the
276	request was already finished, it doesn't wait, of course, and returns the	1472	request was already finished, it doesn't wait, of course, and returns the
277	data:	1473	data:
278		1474
279	$txn->{finished}->wait;	1475	$txn->{finished}->recv;
280	return $txn->{result};	1476	return $txn->{result};
281		1477
282	The actual code goes further and collects all errors (C<die>s, exceptions)	1478	The actual code goes further and collects all errors (C<die>s, exceptions)
283	that occured during request processing. The C<result> method detects	1479	that occurred during request processing. The C<result> method detects
284	wether an exception as thrown (it is stored inside the $txn object)	1480	whether an exception as thrown (it is stored inside the $txn object)
285	and just throws the exception, which means connection errors and other	1481	and just throws the exception, which means connection errors and other
286	problems get reported tot he code that tries to use the result, not in a	1482	problems get reported tot he code that tries to use the result, not in a
287	random callback.	1483	random callback.
288		1484
289	All of this enables the following usage styles:	1485	All of this enables the following usage styles:
290		1486
291	1. Blocking:	1487	1. Blocking:
292		1488
293	my $data = $fcp->client_get ($url);	1489	my $data = $fcp->client_get ($url);
294		1490
295	2. Blocking, but parallelizing:	1491	2. Blocking, but running in parallel:
296		1492
297	my @datas = map $_->result,	1493	my @datas = map $_->result,
298	map $fcp->txn_client_get ($_),	1494	map $fcp->txn_client_get ($_),
299	@urls;	1495	@urls;
300		1496
301	Both blocking examples work without the module user having to know	1497	Both blocking examples work without the module user having to know
302	anything about events.	1498	anything about events.
303		1499
304	3a. Event-based in a main program, using any support Event module:	1500	3a. Event-based in a main program, using any supported event module:
305		1501
306	use Event;	1502	use EV;
307		1503
308	$fcp->txn_client_get ($url)->cb (sub {	1504	$fcp->txn_client_get ($url)->cb (sub {
309	my $txn = shift;	1505	my $txn = shift;
310	my $data = $txn->result;	1506	my $data = $txn->result;
311	...	1507	...
312	});	1508	});
313		1509
314	Event::loop;	1510	EV::loop;
315		1511
316	3b. The module user could use AnyEvent, too:	1512	3b. The module user could use AnyEvent, too:
317		1513
318	use AnyEvent;	1514	use AnyEvent;
319		1515
320	my $quit = AnyEvent->condvar;	1516	my $quit = AnyEvent->condvar;
321		1517
322	$fcp->txn_client_get ($url)->cb (sub {	1518	$fcp->txn_client_get ($url)->cb (sub {
323	...	1519	...
324	$quit->broadcast;	1520	$quit->send;
325	});	1521	});
326		1522
327	$quit->wait;	1523	$quit->recv;
		1524
		1525
		1526	=head1 BENCHMARKS
		1527
		1528	To give you an idea of the performance and overheads that AnyEvent adds
		1529	over the event loops themselves and to give you an impression of the speed
		1530	of various event loops I prepared some benchmarks.
		1531
		1532	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1533
		1534	Here is a benchmark of various supported event models used natively and
		1535	through AnyEvent. The benchmark creates a lot of timers (with a zero
		1536	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
		1537	which it is), lets them fire exactly once and destroys them again.
		1538
		1539	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
		1540	distribution.
		1541
		1542	=head3 Explanation of the columns
		1543
		1544	I<watcher> is the number of event watchers created/destroyed. Since
		1545	different event models feature vastly different performances, each event
		1546	loop was given a number of watchers so that overall runtime is acceptable
		1547	and similar between tested event loop (and keep them from crashing): Glib
		1548	would probably take thousands of years if asked to process the same number
		1549	of watchers as EV in this benchmark.
		1550
		1551	I<bytes> is the number of bytes (as measured by the resident set size,
		1552	RSS) consumed by each watcher. This method of measuring captures both C
		1553	and Perl-based overheads.
		1554
		1555	I<create> is the time, in microseconds (millionths of seconds), that it
		1556	takes to create a single watcher. The callback is a closure shared between
		1557	all watchers, to avoid adding memory overhead. That means closure creation
		1558	and memory usage is not included in the figures.
		1559
		1560	I<invoke> is the time, in microseconds, used to invoke a simple
		1561	callback. The callback simply counts down a Perl variable and after it was
		1562	invoked "watcher" times, it would C<< ->send >> a condvar once to
		1563	signal the end of this phase.
		1564
		1565	I<destroy> is the time, in microseconds, that it takes to destroy a single
		1566	watcher.
		1567
		1568	=head3 Results
		1569
		1570	name watchers bytes create invoke destroy comment
		1571	EV/EV 400000 224 0.47 0.35 0.27 EV native interface
		1572	EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers
		1573	CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal
		1574	Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation
		1575	Event/Event 16000 517 32.20 31.80 0.81 Event native interface
		1576	Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers
		1577	Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour
		1578	Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers
		1579	POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event
		1580	POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select
		1581
		1582	=head3 Discussion
		1583
		1584	The benchmark does I<not> measure scalability of the event loop very
		1585	well. For example, a select-based event loop (such as the pure perl one)
		1586	can never compete with an event loop that uses epoll when the number of
		1587	file descriptors grows high. In this benchmark, all events become ready at
		1588	the same time, so select/poll-based implementations get an unnatural speed
		1589	boost.
		1590
		1591	Also, note that the number of watchers usually has a nonlinear effect on
		1592	overall speed, that is, creating twice as many watchers doesn't take twice
		1593	the time - usually it takes longer. This puts event loops tested with a
		1594	higher number of watchers at a disadvantage.
		1595
		1596	To put the range of results into perspective, consider that on the
		1597	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1598	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1599	cycles with POE.
		1600
		1601	C<EV> is the sole leader regarding speed and memory use, which are both
		1602	maximal/minimal, respectively. Even when going through AnyEvent, it uses
		1603	far less memory than any other event loop and is still faster than Event
		1604	natively.
		1605
		1606	The pure perl implementation is hit in a few sweet spots (both the
		1607	constant timeout and the use of a single fd hit optimisations in the perl
		1608	interpreter and the backend itself). Nevertheless this shows that it
		1609	adds very little overhead in itself. Like any select-based backend its
		1610	performance becomes really bad with lots of file descriptors (and few of
		1611	them active), of course, but this was not subject of this benchmark.
		1612
		1613	The C<Event> module has a relatively high setup and callback invocation
		1614	cost, but overall scores in on the third place.
		1615
		1616	C<Glib>'s memory usage is quite a bit higher, but it features a
		1617	faster callback invocation and overall ends up in the same class as
		1618	C<Event>. However, Glib scales extremely badly, doubling the number of
		1619	watchers increases the processing time by more than a factor of four,
		1620	making it completely unusable when using larger numbers of watchers
		1621	(note that only a single file descriptor was used in the benchmark, so
		1622	inefficiencies of C<poll> do not account for this).
		1623
		1624	The C<Tk> adaptor works relatively well. The fact that it crashes with
		1625	more than 2000 watchers is a big setback, however, as correctness takes
		1626	precedence over speed. Nevertheless, its performance is surprising, as the
		1627	file descriptor is dup()ed for each watcher. This shows that the dup()
		1628	employed by some adaptors is not a big performance issue (it does incur a
		1629	hidden memory cost inside the kernel which is not reflected in the figures
		1630	above).
		1631
		1632	C<POE>, regardless of underlying event loop (whether using its pure perl
		1633	select-based backend or the Event module, the POE-EV backend couldn't
		1634	be tested because it wasn't working) shows abysmal performance and
		1635	memory usage with AnyEvent: Watchers use almost 30 times as much memory
		1636	as EV watchers, and 10 times as much memory as Event (the high memory
		1637	requirements are caused by requiring a session for each watcher). Watcher
		1638	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		1639	implementation.
		1640
		1641	The design of the POE adaptor class in AnyEvent can not really account
		1642	for the performance issues, though, as session creation overhead is
		1643	small compared to execution of the state machine, which is coded pretty
		1644	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		1645	using multiple sessions is not a good approach, especially regarding
		1646	memory usage, even the author of POE could not come up with a faster
		1647	design).
		1648
		1649	=head3 Summary
		1650
		1651	=over 4
		1652
		1653	=item * Using EV through AnyEvent is faster than any other event loop
		1654	(even when used without AnyEvent), but most event loops have acceptable
		1655	performance with or without AnyEvent.
		1656
		1657	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
		1658	the actual event loop, only with extremely fast event loops such as EV
		1659	adds AnyEvent significant overhead.
		1660
		1661	=item * You should avoid POE like the plague if you want performance or
		1662	reasonable memory usage.
		1663
		1664	=back
		1665
		1666	=head2 BENCHMARKING THE LARGE SERVER CASE
		1667
		1668	This benchmark actually benchmarks the event loop itself. It works by
		1669	creating a number of "servers": each server consists of a socket pair, a
		1670	timeout watcher that gets reset on activity (but never fires), and an I/O
		1671	watcher waiting for input on one side of the socket. Each time the socket
		1672	watcher reads a byte it will write that byte to a random other "server".
		1673
		1674	The effect is that there will be a lot of I/O watchers, only part of which
		1675	are active at any one point (so there is a constant number of active
		1676	fds for each loop iteration, but which fds these are is random). The
		1677	timeout is reset each time something is read because that reflects how
		1678	most timeouts work (and puts extra pressure on the event loops).
		1679
		1680	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		1681	(1%) are active. This mirrors the activity of large servers with many
		1682	connections, most of which are idle at any one point in time.
		1683
		1684	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		1685	distribution.
		1686
		1687	=head3 Explanation of the columns
		1688
		1689	I<sockets> is the number of sockets, and twice the number of "servers" (as
		1690	each server has a read and write socket end).
		1691
		1692	I<create> is the time it takes to create a socket pair (which is
		1693	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		1694
		1695	I<request>, the most important value, is the time it takes to handle a
		1696	single "request", that is, reading the token from the pipe and forwarding
		1697	it to another server. This includes deleting the old timeout and creating
		1698	a new one that moves the timeout into the future.
		1699
		1700	=head3 Results
		1701
		1702	name sockets create request
		1703	EV 20000 69.01 11.16
		1704	Perl 20000 73.32 35.87
		1705	Event 20000 212.62 257.32
		1706	Glib 20000 651.16 1896.30
		1707	POE 20000 349.67 12317.24 uses POE::Loop::Event
		1708
		1709	=head3 Discussion
		1710
		1711	This benchmark I<does> measure scalability and overall performance of the
		1712	particular event loop.
		1713
		1714	EV is again fastest. Since it is using epoll on my system, the setup time
		1715	is relatively high, though.
		1716
		1717	Perl surprisingly comes second. It is much faster than the C-based event
		1718	loops Event and Glib.
		1719
		1720	Event suffers from high setup time as well (look at its code and you will
		1721	understand why). Callback invocation also has a high overhead compared to
		1722	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		1723	uses select or poll in basically all documented configurations.
		1724
		1725	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		1726	clearly fails to perform with many filehandles or in busy servers.
		1727
		1728	POE is still completely out of the picture, taking over 1000 times as long
		1729	as EV, and over 100 times as long as the Perl implementation, even though
		1730	it uses a C-based event loop in this case.
		1731
		1732	=head3 Summary
		1733
		1734	=over 4
		1735
		1736	=item * The pure perl implementation performs extremely well.
		1737
		1738	=item * Avoid Glib or POE in large projects where performance matters.
		1739
		1740	=back
		1741
		1742	=head2 BENCHMARKING SMALL SERVERS
		1743
		1744	While event loops should scale (and select-based ones do not...) even to
		1745	large servers, most programs we (or I :) actually write have only a few
		1746	I/O watchers.
		1747
		1748	In this benchmark, I use the same benchmark program as in the large server
		1749	case, but it uses only eight "servers", of which three are active at any
		1750	one time. This should reflect performance for a small server relatively
		1751	well.
		1752
		1753	The columns are identical to the previous table.
		1754
		1755	=head3 Results
		1756
		1757	name sockets create request
		1758	EV 16 20.00 6.54
		1759	Perl 16 25.75 12.62
		1760	Event 16 81.27 35.86
		1761	Glib 16 32.63 15.48
		1762	POE 16 261.87 276.28 uses POE::Loop::Event
		1763
		1764	=head3 Discussion
		1765
		1766	The benchmark tries to test the performance of a typical small
		1767	server. While knowing how various event loops perform is interesting, keep
		1768	in mind that their overhead in this case is usually not as important, due
		1769	to the small absolute number of watchers (that is, you need efficiency and
		1770	speed most when you have lots of watchers, not when you only have a few of
		1771	them).
		1772
		1773	EV is again fastest.
		1774
		1775	Perl again comes second. It is noticeably faster than the C-based event
		1776	loops Event and Glib, although the difference is too small to really
		1777	matter.
		1778
		1779	POE also performs much better in this case, but is is still far behind the
		1780	others.
		1781
		1782	=head3 Summary
		1783
		1784	=over 4
		1785
		1786	=item * C-based event loops perform very well with small number of
		1787	watchers, as the management overhead dominates.
		1788
		1789	=back
		1790
		1791
		1792	=head1 SIGNALS
		1793
		1794	AnyEvent currently installs handlers for these signals:
		1795
		1796	=over 4
		1797
		1798	=item SIGCHLD
		1799
		1800	A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
		1801	emulation for event loops that do not support them natively. Also, some
		1802	event loops install a similar handler.
		1803
		1804	=item SIGPIPE
		1805
		1806	A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
		1807	when AnyEvent gets loaded.
		1808
		1809	The rationale for this is that AnyEvent users usually do not really depend
		1810	on SIGPIPE delivery (which is purely an optimisation for shell use, or
		1811	badly-written programs), but C<SIGPIPE> can cause spurious and rare
		1812	program exits as a lot of people do not expect C<SIGPIPE> when writing to
		1813	some random socket.
		1814
		1815	The rationale for installing a no-op handler as opposed to ignoring it is
		1816	that this way, the handler will be restored to defaults on exec.
		1817
		1818	Feel free to install your own handler, or reset it to defaults.
		1819
		1820	=back
		1821
		1822	=cut
		1823
		1824	$SIG{PIPE} = sub { }
		1825	unless defined $SIG{PIPE};
		1826
		1827
		1828	=head1 FORK
		1829
		1830	Most event libraries are not fork-safe. The ones who are usually are
		1831	because they rely on inefficient but fork-safe C<select> or C<poll>
		1832	calls. Only L<EV> is fully fork-aware.
		1833
		1834	If you have to fork, you must either do so I<before> creating your first
		1835	watcher OR you must not use AnyEvent at all in the child.
		1836
		1837
		1838	=head1 SECURITY CONSIDERATIONS
		1839
		1840	AnyEvent can be forced to load any event model via
		1841	$ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used to
		1842	execute arbitrary code or directly gain access, it can easily be used to
		1843	make the program hang or malfunction in subtle ways, as AnyEvent watchers
		1844	will not be active when the program uses a different event model than
		1845	specified in the variable.
		1846
		1847	You can make AnyEvent completely ignore this variable by deleting it
		1848	before the first watcher gets created, e.g. with a C<BEGIN> block:
		1849
		1850	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
		1851
		1852	use AnyEvent;
		1853
		1854	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		1855	be used to probe what backend is used and gain other information (which is
		1856	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		1857	$ENV{PERL_ANYEGENT_STRICT}.
		1858
		1859
		1860	=head1 BUGS
		1861
		1862	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		1863	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		1864	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		1865	mamleaks, such as leaking on C<map> and C<grep> but it is usually not as
		1866	pronounced).
		1867
328		1868
329	=head1 SEE ALSO	1869	=head1 SEE ALSO
330		1870
331	Event modules: L<Coro::Event>, L<Coro>, L<Event>, L<Glib::Event>, L<Glib>.	1871	Utility functions: L<AnyEvent::Util>.
332		1872
333	Implementations: L<AnyEvent::Impl::Coro>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>.	1873	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
		1874	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
334		1875
335	Nontrivial usage example: L<Net::FCP>.	1876	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
		1877	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
		1878	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
		1879	L<AnyEvent::Impl::POE>.
336		1880
337	=head1	1881	Non-blocking file handles, sockets, TCP clients and
		1882	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>.
		1883
		1884	Asynchronous DNS: L<AnyEvent::DNS>.
		1885
		1886	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>,
		1887
		1888	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>.
		1889
		1890
		1891	=head1 AUTHOR
		1892
		1893	Marc Lehmann <schmorp@schmorp.de>
		1894	http://home.schmorp.de/
338		1895
339	=cut	1896	=cut
340		1897
341	1	1898	1
342		1899

Diff Legend

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