ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.23 by root, Wed Mar 7 17:37:24 2007 UTC vs.
Revision 1.252 by root, Tue Jul 21 03:30:02 2009 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent - provide framework for multiple event loops	3	AnyEvent - events independent of event loop implementation
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 SUPPORT
		47
		48	There is a mailinglist for discussing all things AnyEvent, and an IRC
		49	channel, too.
		50
		51	See the AnyEvent project page at the B<Schmorpforge Ta-Sa Software
		52	Respository>, at L<http://anyevent.schmorp.de>, for more info.
		53
		54	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
		55
		56	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
		57	nowadays. So what is different about AnyEvent?
		58
		59	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
		60	policy> and AnyEvent is I<small and efficient>.
		61
		62	First and foremost, I<AnyEvent is not an event model> itself, it only
		63	interfaces to whatever event model the main program happens to use, in a
		64	pragmatic way. For event models and certain classes of immortals alike,
		65	the statement "there can only be one" is a bitter reality: In general,
		66	only one event loop can be active at the same time in a process. AnyEvent
		67	cannot change this, but it can hide the differences between those event
		68	loops.
		69
		70	The goal of AnyEvent is to offer module authors the ability to do event
		71	programming (waiting for I/O or timer events) without subscribing to a
		72	religion, a way of living, and most importantly: without forcing your
		73	module users into the same thing by forcing them to use the same event
		74	model you use.
		75
		76	For modules like POE or IO::Async (which is a total misnomer as it is
		77	actually doing all I/O I<synchronously>...), using them in your module is
		78	like joining a cult: After you joined, you are dependent on them and you
		79	cannot use anything else, as they are simply incompatible to everything
		80	that isn't them. What's worse, all the potential users of your
		81	module are I<also> forced to use the same event loop you use.
		82
		83	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
		84	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
		85	with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
		86	your module uses one of those, every user of your module has to use it,
		87	too. But if your module uses AnyEvent, it works transparently with all
		88	event models it supports (including stuff like IO::Async, as long as those
		89	use one of the supported event loops. It is trivial to add new event loops
		90	to AnyEvent, too, so it is future-proof).
		91
		92	In addition to being free of having to use I<the one and only true event
		93	model>, AnyEvent also is free of bloat and policy: with POE or similar
		94	modules, you get an enormous amount of code and strict rules you have to
		95	follow. AnyEvent, on the other hand, is lean and up to the point, by only
		96	offering the functionality that is necessary, in as thin as a wrapper as
		97	technically possible.
		98
		99	Of course, AnyEvent comes with a big (and fully optional!) toolbox
		100	of useful functionality, such as an asynchronous DNS resolver, 100%
		101	non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
		102	such as Windows) and lots of real-world knowledge and workarounds for
		103	platform bugs and differences.
		104
		105	Now, if you I<do want> lots of policy (this can arguably be somewhat
		106	useful) and you want to force your users to use the one and only event
		107	model, you should I<not> use this module.
22		108
23	=head1 DESCRIPTION	109	=head1 DESCRIPTION
24		110
25	L<AnyEvent> provides an identical interface to multiple event loops. This	111	L<AnyEvent> provides an identical interface to multiple event loops. This
26	allows module authors to utilise an event loop without forcing module	112	allows module authors to utilise an event loop without forcing module
27	users to use the same event loop (as only a single event loop can coexist	113	users to use the same event loop (as only a single event loop can coexist
28	peacefully at any one time).	114	peacefully at any one time).
29		115
30	The interface itself is vaguely similar but not identical to the Event	116	The interface itself is vaguely similar, but not identical to the L<Event>
31	module.	117	module.
32		118
33	On the first call of any method, the module tries to detect the currently	119	During the first call of any watcher-creation method, the module tries
34	loaded event loop by probing wether any of the following modules is	120	to detect the currently loaded event loop by probing whether one of the
35	loaded: L<Coro::Event>, L<Event>, L<Glib>, L<Tk>. The first one found is	121	following modules is already loaded: L<EV>,
36	used. If none is found, the module tries to load these modules in the	122	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,
37	order given. The first one that could be successfully loaded will be	123	L<POE>. The first one found is used. If none are found, the module tries
38	used. If still none could be found, AnyEvent will fall back to a pure-perl	124	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
39	event loop, which is also not very efficient.	125	adaptor should always succeed) in the order given. The first one that can
		126	be successfully loaded will be used. If, after this, still none could be
		127	found, AnyEvent will fall back to a pure-perl event loop, which is not
		128	very efficient, but should work everywhere.
40		129
41	Because AnyEvent first checks for modules that are already loaded, loading	130	Because AnyEvent first checks for modules that are already loaded, loading
42	an Event model explicitly before first using AnyEvent will likely make	131	an event model explicitly before first using AnyEvent will likely make
43	that model the default. For example:	132	that model the default. For example:
44		133
45	use Tk;	134	use Tk;
46	use AnyEvent;	135	use AnyEvent;
47		136
48	# .. AnyEvent will likely default to Tk	137	# .. AnyEvent will likely default to Tk
49		138
		139	The I<likely> means that, if any module loads another event model and
		140	starts using it, all bets are off. Maybe you should tell their authors to
		141	use AnyEvent so their modules work together with others seamlessly...
		142
50	The pure-perl implementation of AnyEvent is called	143	The pure-perl implementation of AnyEvent is called
51	C<AnyEvent::Impl::Perl>. Like other event modules you can load it	144	C<AnyEvent::Impl::Perl>. Like other event modules you can load it
52	explicitly.	145	explicitly and enjoy the high availability of that event loop :)
53		146
54	=head1 WATCHERS	147	=head1 WATCHERS
55		148
56	AnyEvent has the central concept of a I<watcher>, which is an object that	149	AnyEvent has the central concept of a I<watcher>, which is an object that
57	stores relevant data for each kind of event you are waiting for, such as	150	stores relevant data for each kind of event you are waiting for, such as
58	the callback to call, the filehandle to watch, etc.	151	the callback to call, the file handle to watch, etc.
59		152
60	These watchers are normal Perl objects with normal Perl lifetime. After	153	These watchers are normal Perl objects with normal Perl lifetime. After
61	creating a watcher it will immediately "watch" for events and invoke	154	creating a watcher it will immediately "watch" for events and invoke the
		155	callback when the event occurs (of course, only when the event model
		156	is in control).
		157
		158	Note that B<callbacks must not permanently change global variables>
		159	potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
		160	callbacks must not C<die> >>. The former is good programming practise in
		161	Perl and the latter stems from the fact that exception handling differs
		162	widely between event loops.
		163
62	the callback. To disable the watcher you have to destroy it (e.g. by	164	To disable the watcher you have to destroy it (e.g. by setting the
63	setting the variable that stores it to C<undef> or otherwise deleting all	165	variable you store it in to C<undef> or otherwise deleting all references
64	references to it).	166	to it).
65		167
66	All watchers are created by calling a method on the C<AnyEvent> class.	168	All watchers are created by calling a method on the C<AnyEvent> class.
67		169
		170	Many watchers either are used with "recursion" (repeating timers for
		171	example), or need to refer to their watcher object in other ways.
		172
		173	An any way to achieve that is this pattern:
		174
		175	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
		176	# you can use $w here, for example to undef it
		177	undef $w;
		178	});
		179
		180	Note that C<my $w; $w => combination. This is necessary because in Perl,
		181	my variables are only visible after the statement in which they are
		182	declared.
		183
68	=head2 IO WATCHERS	184	=head2 I/O WATCHERS
69		185
70	You can create I/O watcher by calling the C<< AnyEvent->io >> method with	186	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
71	the following mandatory arguments:	187	with the following mandatory key-value pairs as arguments:
72		188
73	C<fh> the Perl I<filehandle> (not filedescriptor) to watch for	189	C<fh> is the Perl I<file handle> (or a naked file descriptor) to watch
		190	for events (AnyEvent might or might not keep a reference to this file
		191	handle). Note that only file handles pointing to things for which
		192	non-blocking operation makes sense are allowed. This includes sockets,
		193	most character devices, pipes, fifos and so on, but not for example files
		194	or block devices.
		195
74	events. C<poll> must be a string that is either C<r> or C<w>, that creates	196	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	197	watcher waiting for "r"eadable or "w"ritable events, respectively.
76	to invoke everytime the filehandle becomes ready.
77		198
78	Only one io watcher per C<fh> and C<poll> combination is allowed (i.e. on	199	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		200
82	Filehandles will be kept alive, so as long as the watcher exists, the	201	Although the callback might get passed parameters, their value and
83	filehandle exists, too.	202	presence is undefined and you cannot rely on them. Portable AnyEvent
		203	callbacks cannot use arguments passed to I/O watcher callbacks.
84		204
85	Example:	205	The I/O watcher might use the underlying file descriptor or a copy of it.
		206	You must not close a file handle as long as any watcher is active on the
		207	underlying file descriptor.
86		208
		209	Some event loops issue spurious readyness notifications, so you should
		210	always use non-blocking calls when reading/writing from/to your file
		211	handles.
		212
87	# wait for readability of STDIN, then read a line and disable the watcher	213	Example: wait for readability of STDIN, then read a line and disable the
		214	watcher.
		215
88	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	216	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
89	chomp (my $input = <STDIN>);	217	chomp (my $input = <STDIN>);
90	warn "read: $input\n";	218	warn "read: $input\n";
91	undef $w;	219	undef $w;
92	});	220	});
…		…
94	=head2 TIME WATCHERS	222	=head2 TIME WATCHERS
95		223
96	You can create a time watcher by calling the C<< AnyEvent->timer >>	224	You can create a time watcher by calling the C<< AnyEvent->timer >>
97	method with the following mandatory arguments:	225	method with the following mandatory arguments:
98		226
99	C<after> after how many seconds (fractions are supported) should the timer	227	C<after> specifies after how many seconds (fractional values are
100	activate. C<cb> the callback to invoke.	228	supported) the callback should be invoked. C<cb> is the callback to invoke
		229	in that case.
101		230
102	The timer callback will be invoked at most once: if you want a repeating	231	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	232	presence is undefined and you cannot rely on them. Portable AnyEvent
104	and Glib).	233	callbacks cannot use arguments passed to time watcher callbacks.
105		234
106	Example:	235	The callback will normally be invoked once only. If you specify another
		236	parameter, C<interval>, as a strictly positive number (> 0), then the
		237	callback will be invoked regularly at that interval (in fractional
		238	seconds) after the first invocation. If C<interval> is specified with a
		239	false value, then it is treated as if it were missing.
107		240
		241	The callback will be rescheduled before invoking the callback, but no
		242	attempt is done to avoid timer drift in most backends, so the interval is
		243	only approximate.
		244
108	# fire an event after 7.7 seconds	245	Example: fire an event after 7.7 seconds.
		246
109	my $w = AnyEvent->timer (after => 7.7, cb => sub {	247	my $w = AnyEvent->timer (after => 7.7, cb => sub {
110	warn "timeout\n";	248	warn "timeout\n";
111	});	249	});
112		250
113	# to cancel the timer:	251	# to cancel the timer:
114	undef $w	252	undef $w;
115		253
116	=head2 CONDITION WATCHERS	254	Example 2: fire an event after 0.5 seconds, then roughly every second.
117		255
118	Condition watchers can be created by calling the C<< AnyEvent->condvar >>	256	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
119	method without any arguments.	257	warn "timeout\n";
		258	};
120		259
121	A condition watcher watches for a condition - precisely that the C<<	260	=head3 TIMING ISSUES
122	->broadcast >> method has been called.
123		261
124	The watcher has only two methods:	262	There are two ways to handle timers: based on real time (relative, "fire
		263	in 10 seconds") and based on wallclock time (absolute, "fire at 12
		264	o'clock").
		265
		266	While most event loops expect timers to specified in a relative way, they
		267	use absolute time internally. This makes a difference when your clock
		268	"jumps", for example, when ntp decides to set your clock backwards from
		269	the wrong date of 2014-01-01 to 2008-01-01, a watcher that is supposed to
		270	fire "after" a second might actually take six years to finally fire.
		271
		272	AnyEvent cannot compensate for this. The only event loop that is conscious
		273	about these issues is L<EV>, which offers both relative (ev_timer, based
		274	on true relative time) and absolute (ev_periodic, based on wallclock time)
		275	timers.
		276
		277	AnyEvent always prefers relative timers, if available, matching the
		278	AnyEvent API.
		279
		280	AnyEvent has two additional methods that return the "current time":
125		281
126	=over 4	282	=over 4
127		283
128	=item $cv->wait	284	=item AnyEvent->time
129		285
130	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	286	This returns the "current wallclock time" as a fractional number of
131	called on c<$cv>, while servicing other watchers normally.	287	seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
		288	return, and the result is guaranteed to be compatible with those).
132		289
133	Not all event models support a blocking wait - some die in that case, so	290	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	291	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		292
140	You can only wait once on a condition - additional calls will return	293	=item AnyEvent->now
141	immediately.
142		294
143	=item $cv->broadcast	295	This also returns the "current wallclock time", but unlike C<time>, above,
		296	this value might change only once per event loop iteration, depending on
		297	the event loop (most return the same time as C<time>, above). This is the
		298	time that AnyEvent's timers get scheduled against.
144		299
145	Flag the condition as ready - a running C<< ->wait >> and all further	300	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	301	function to call when you want to know the current time.>
147	is waiting the broadcast will be remembered..
148		302
149	Example:	303	This function is also often faster then C<< AnyEvent->time >>, and
		304	thus the preferred method if you want some timestamp (for example,
		305	L<AnyEvent::Handle> uses this to update it's activity timeouts).
		306
		307	The rest of this section is only of relevance if you try to be very exact
		308	with your timing, you can skip it without bad conscience.
		309
		310	For a practical example of when these times differ, consider L<Event::Lib>
		311	and L<EV> and the following set-up:
		312
		313	The event loop is running and has just invoked one of your callback at
		314	time=500 (assume no other callbacks delay processing). In your callback,
		315	you wait a second by executing C<sleep 1> (blocking the process for a
		316	second) and then (at time=501) you create a relative timer that fires
		317	after three seconds.
		318
		319	With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
		320	both return C<501>, because that is the current time, and the timer will
		321	be scheduled to fire at time=504 (C<501> + C<3>).
		322
		323	With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
		324	time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
		325	last event processing phase started. With L<EV>, your timer gets scheduled
		326	to run at time=503 (C<500> + C<3>).
		327
		328	In one sense, L<Event::Lib> is more exact, as it uses the current time
		329	regardless of any delays introduced by event processing. However, most
		330	callbacks do not expect large delays in processing, so this causes a
		331	higher drift (and a lot more system calls to get the current time).
		332
		333	In another sense, L<EV> is more exact, as your timer will be scheduled at
		334	the same time, regardless of how long event processing actually took.
		335
		336	In either case, if you care (and in most cases, you don't), then you
		337	can get whatever behaviour you want with any event loop, by taking the
		338	difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
		339	account.
		340
		341	=item AnyEvent->now_update
		342
		343	Some event loops (such as L<EV> or L<AnyEvent::Impl::Perl>) cache
		344	the current time for each loop iteration (see the discussion of L<<
		345	AnyEvent->now >>, above).
		346
		347	When a callback runs for a long time (or when the process sleeps), then
		348	this "current" time will differ substantially from the real time, which
		349	might affect timers and time-outs.
		350
		351	When this is the case, you can call this method, which will update the
		352	event loop's idea of "current time".
		353
		354	Note that updating the time I<might> cause some events to be handled.
		355
		356	=back
		357
		358	=head2 SIGNAL WATCHERS
		359
		360	You can watch for signals using a signal watcher, C<signal> is the signal
		361	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
		362	callback to be invoked whenever a signal occurs.
		363
		364	Although the callback might get passed parameters, their value and
		365	presence is undefined and you cannot rely on them. Portable AnyEvent
		366	callbacks cannot use arguments passed to signal watcher callbacks.
		367
		368	Multiple signal occurrences can be clumped together into one callback
		369	invocation, and callback invocation will be synchronous. Synchronous means
		370	that it might take a while until the signal gets handled by the process,
		371	but it is guaranteed not to interrupt any other callbacks.
		372
		373	The main advantage of using these watchers is that you can share a signal
		374	between multiple watchers, and AnyEvent will ensure that signals will not
		375	interrupt your program at bad times.
		376
		377	This watcher might use C<%SIG> (depending on the event loop used),
		378	so programs overwriting those signals directly will likely not work
		379	correctly.
		380
		381	Example: exit on SIGINT
		382
		383	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
		384
		385	=head3 Signal Races, Delays and Workarounds
		386
		387	Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
		388	callbacks to signals in a generic way, which is a pity, as you cannot do
		389	race-free signal handling in perl. AnyEvent will try to do it's best, but
		390	in some cases, signals will be delayed. The maximum time a signal might
		391	be delayed is specified in C<$AnyEvent::MAX_SIGNAL_LATENCY> (default: 10
		392	seconds). This variable can be changed only before the first signal
		393	watcher is created, and should be left alone otherwise. Higher values
		394	will cause fewer spurious wake-ups, which is better for power and CPU
		395	saving. All these problems can be avoided by installing the optional
		396	L<Async::Interrupt> module. This will not work with inherently broken
		397	event loops such as L<Event> or L<Event::Lib> (and not with L<POE>
		398	currently, as POE does it's own workaround with one-second latency). With
		399	those, you just have to suffer the delays.
		400
		401	=head2 CHILD PROCESS WATCHERS
		402
		403	You can also watch on a child process exit and catch its exit status.
		404
		405	The child process is specified by the C<pid> argument (if set to C<0>, it
		406	watches for any child process exit). The watcher will triggered only when
		407	the child process has finished and an exit status is available, not on
		408	any trace events (stopped/continued).
		409
		410	The callback will be called with the pid and exit status (as returned by
		411	waitpid), so unlike other watcher types, you I<can> rely on child watcher
		412	callback arguments.
		413
		414	This watcher type works by installing a signal handler for C<SIGCHLD>,
		415	and since it cannot be shared, nothing else should use SIGCHLD or reap
		416	random child processes (waiting for specific child processes, e.g. inside
		417	C<system>, is just fine).
		418
		419	There is a slight catch to child watchers, however: you usually start them
		420	I<after> the child process was created, and this means the process could
		421	have exited already (and no SIGCHLD will be sent anymore).
		422
		423	Not all event models handle this correctly (neither POE nor IO::Async do,
		424	see their AnyEvent::Impl manpages for details), but even for event models
		425	that I<do> handle this correctly, they usually need to be loaded before
		426	the process exits (i.e. before you fork in the first place). AnyEvent's
		427	pure perl event loop handles all cases correctly regardless of when you
		428	start the watcher.
		429
		430	This means you cannot create a child watcher as the very first
		431	thing in an AnyEvent program, you I<have> to create at least one
		432	watcher before you C<fork> the child (alternatively, you can call
		433	C<AnyEvent::detect>).
		434
		435	As most event loops do not support waiting for child events, they will be
		436	emulated by AnyEvent in most cases, in which the latency and race problems
		437	mentioned in the description of signal watchers apply.
		438
		439	Example: fork a process and wait for it
		440
		441	my $done = AnyEvent->condvar;
		442
		443	my $pid = fork or exit 5;
		444
		445	my $w = AnyEvent->child (
		446	pid => $pid,
		447	cb => sub {
		448	my ($pid, $status) = @_;
		449	warn "pid $pid exited with status $status";
		450	$done->send;
		451	},
		452	);
		453
		454	# do something else, then wait for process exit
		455	$done->recv;
		456
		457	=head2 IDLE WATCHERS
		458
		459	Sometimes there is a need to do something, but it is not so important
		460	to do it instantly, but only when there is nothing better to do. This
		461	"nothing better to do" is usually defined to be "no other events need
		462	attention by the event loop".
		463
		464	Idle watchers ideally get invoked when the event loop has nothing
		465	better to do, just before it would block the process to wait for new
		466	events. Instead of blocking, the idle watcher is invoked.
		467
		468	Most event loops unfortunately do not really support idle watchers (only
		469	EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
		470	will simply call the callback "from time to time".
		471
		472	Example: read lines from STDIN, but only process them when the
		473	program is otherwise idle:
		474
		475	my @lines; # read data
		476	my $idle_w;
		477	my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
		478	push @lines, scalar <STDIN>;
		479
		480	# start an idle watcher, if not already done
		481	$idle_w ||= AnyEvent->idle (cb => sub {
		482	# handle only one line, when there are lines left
		483	if (my $line = shift @lines) {
		484	print "handled when idle: $line";
		485	} else {
		486	# otherwise disable the idle watcher again
		487	undef $idle_w;
		488	}
		489	});
		490	});
		491
		492	=head2 CONDITION VARIABLES
		493
		494	If you are familiar with some event loops you will know that all of them
		495	require you to run some blocking "loop", "run" or similar function that
		496	will actively watch for new events and call your callbacks.
		497
		498	AnyEvent is slightly different: it expects somebody else to run the event
		499	loop and will only block when necessary (usually when told by the user).
		500
		501	The instrument to do that is called a "condition variable", so called
		502	because they represent a condition that must become true.
		503
		504	Now is probably a good time to look at the examples further below.
		505
		506	Condition variables can be created by calling the C<< AnyEvent->condvar
		507	>> method, usually without arguments. The only argument pair allowed is
		508	C<cb>, which specifies a callback to be called when the condition variable
		509	becomes true, with the condition variable as the first argument (but not
		510	the results).
		511
		512	After creation, the condition variable is "false" until it becomes "true"
		513	by calling the C<send> method (or calling the condition variable as if it
		514	were a callback, read about the caveats in the description for the C<<
		515	->send >> method).
		516
		517	Condition variables are similar to callbacks, except that you can
		518	optionally wait for them. They can also be called merge points - points
		519	in time where multiple outstanding events have been processed. And yet
		520	another way to call them is transactions - each condition variable can be
		521	used to represent a transaction, which finishes at some point and delivers
		522	a result. And yet some people know them as "futures" - a promise to
		523	compute/deliver something that you can wait for.
		524
		525	Condition variables are very useful to signal that something has finished,
		526	for example, if you write a module that does asynchronous http requests,
		527	then a condition variable would be the ideal candidate to signal the
		528	availability of results. The user can either act when the callback is
		529	called or can synchronously C<< ->recv >> for the results.
		530
		531	You can also use them to simulate traditional event loops - for example,
		532	you can block your main program until an event occurs - for example, you
		533	could C<< ->recv >> in your main program until the user clicks the Quit
		534	button of your app, which would C<< ->send >> the "quit" event.
		535
		536	Note that condition variables recurse into the event loop - if you have
		537	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
		538	lose. Therefore, condition variables are good to export to your caller, but
		539	you should avoid making a blocking wait yourself, at least in callbacks,
		540	as this asks for trouble.
		541
		542	Condition variables are represented by hash refs in perl, and the keys
		543	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		544	easy (it is often useful to build your own transaction class on top of
		545	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		546	it's C<new> method in your own C<new> method.
		547
		548	There are two "sides" to a condition variable - the "producer side" which
		549	eventually calls C<< -> send >>, and the "consumer side", which waits
		550	for the send to occur.
		551
		552	Example: wait for a timer.
150		553
151	# wait till the result is ready	554	# wait till the result is ready
152	my $result_ready = AnyEvent->condvar;	555	my $result_ready = AnyEvent->condvar;
153		556
154	# do something such as adding a timer	557	# do something such as adding a timer
155	# or socket watcher the calls $result_ready->broadcast	558	# or socket watcher the calls $result_ready->send
156	# when the "result" is ready.	559	# when the "result" is ready.
		560	# in this case, we simply use a timer:
		561	my $w = AnyEvent->timer (
		562	after => 1,
		563	cb => sub { $result_ready->send },
		564	);
157		565
		566	# this "blocks" (while handling events) till the callback
		567	# calls -<send
158	$result_ready->wait;	568	$result_ready->recv;
		569
		570	Example: wait for a timer, but take advantage of the fact that condition
		571	variables are also callable directly.
		572
		573	my $done = AnyEvent->condvar;
		574	my $delay = AnyEvent->timer (after => 5, cb => $done);
		575	$done->recv;
		576
		577	Example: Imagine an API that returns a condvar and doesn't support
		578	callbacks. This is how you make a synchronous call, for example from
		579	the main program:
		580
		581	use AnyEvent::CouchDB;
		582
		583	...
		584
		585	my @info = $couchdb->info->recv;
		586
		587	And this is how you would just set a callback to be called whenever the
		588	results are available:
		589
		590	$couchdb->info->cb (sub {
		591	my @info = $_[0]->recv;
		592	});
		593
		594	=head3 METHODS FOR PRODUCERS
		595
		596	These methods should only be used by the producing side, i.e. the
		597	code/module that eventually sends the signal. Note that it is also
		598	the producer side which creates the condvar in most cases, but it isn't
		599	uncommon for the consumer to create it as well.
		600
		601	=over 4
		602
		603	=item $cv->send (...)
		604
		605	Flag the condition as ready - a running C<< ->recv >> and all further
		606	calls to C<recv> will (eventually) return after this method has been
		607	called. If nobody is waiting the send will be remembered.
		608
		609	If a callback has been set on the condition variable, it is called
		610	immediately from within send.
		611
		612	Any arguments passed to the C<send> call will be returned by all
		613	future C<< ->recv >> calls.
		614
		615	Condition variables are overloaded so one can call them directly (as if
		616	they were a code reference). Calling them directly is the same as calling
		617	C<send>.
		618
		619	=item $cv->croak ($error)
		620
		621	Similar to send, but causes all call's to C<< ->recv >> to invoke
		622	C<Carp::croak> with the given error message/object/scalar.
		623
		624	This can be used to signal any errors to the condition variable
		625	user/consumer. Doing it this way instead of calling C<croak> directly
		626	delays the error detetcion, but has the overwhelmign advantage that it
		627	diagnoses the error at the place where the result is expected, and not
		628	deep in some event clalback without connection to the actual code causing
		629	the problem.
		630
		631	=item $cv->begin ([group callback])
		632
		633	=item $cv->end
		634
		635	These two methods can be used to combine many transactions/events into
		636	one. For example, a function that pings many hosts in parallel might want
		637	to use a condition variable for the whole process.
		638
		639	Every call to C<< ->begin >> will increment a counter, and every call to
		640	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		641	>>, the (last) callback passed to C<begin> will be executed. That callback
		642	is I<supposed> to call C<< ->send >>, but that is not required. If no
		643	callback was set, C<send> will be called without any arguments.
		644
		645	You can think of C<< $cv->send >> giving you an OR condition (one call
		646	sends), while C<< $cv->begin >> and C<< $cv->end >> giving you an AND
		647	condition (all C<begin> calls must be C<end>'ed before the condvar sends).
		648
		649	Let's start with a simple example: you have two I/O watchers (for example,
		650	STDOUT and STDERR for a program), and you want to wait for both streams to
		651	close before activating a condvar:
		652
		653	my $cv = AnyEvent->condvar;
		654
		655	$cv->begin; # first watcher
		656	my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
		657	defined sysread $fh1, my $buf, 4096
		658	or $cv->end;
		659	});
		660
		661	$cv->begin; # second watcher
		662	my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
		663	defined sysread $fh2, my $buf, 4096
		664	or $cv->end;
		665	});
		666
		667	$cv->recv;
		668
		669	This works because for every event source (EOF on file handle), there is
		670	one call to C<begin>, so the condvar waits for all calls to C<end> before
		671	sending.
		672
		673	The ping example mentioned above is slightly more complicated, as the
		674	there are results to be passwd back, and the number of tasks that are
		675	begung can potentially be zero:
		676
		677	my $cv = AnyEvent->condvar;
		678
		679	my %result;
		680	$cv->begin (sub { $cv->send (\%result) });
		681
		682	for my $host (@list_of_hosts) {
		683	$cv->begin;
		684	ping_host_then_call_callback $host, sub {
		685	$result{$host} = ...;
		686	$cv->end;
		687	};
		688	}
		689
		690	$cv->end;
		691
		692	This code fragment supposedly pings a number of hosts and calls
		693	C<send> after results for all then have have been gathered - in any
		694	order. To achieve this, the code issues a call to C<begin> when it starts
		695	each ping request and calls C<end> when it has received some result for
		696	it. Since C<begin> and C<end> only maintain a counter, the order in which
		697	results arrive is not relevant.
		698
		699	There is an additional bracketing call to C<begin> and C<end> outside the
		700	loop, which serves two important purposes: first, it sets the callback
		701	to be called once the counter reaches C<0>, and second, it ensures that
		702	C<send> is called even when C<no> hosts are being pinged (the loop
		703	doesn't execute once).
		704
		705	This is the general pattern when you "fan out" into multiple (but
		706	potentially none) subrequests: use an outer C<begin>/C<end> pair to set
		707	the callback and ensure C<end> is called at least once, and then, for each
		708	subrequest you start, call C<begin> and for each subrequest you finish,
		709	call C<end>.
159		710
160	=back	711	=back
161		712
162	=head2 SIGNAL WATCHERS	713	=head3 METHODS FOR CONSUMERS
163		714
164	You can listen for signals using a signal watcher, C<signal> is the signal	715	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	716	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. The watcher will only trigger once. This works by
180	installing a signal handler for C<SIGCHLD>.
181
182	Example: wait for pid 1333
183
184	my $w = AnyEvent->child (pid => 1333, cb => sub { warn "exit status $?" });
185
186	=head1 GLOBALS
187		717
188	=over 4	718	=over 4
189		719
		720	=item $cv->recv
		721
		722	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
		723	>> methods have been called on c<$cv>, while servicing other watchers
		724	normally.
		725
		726	You can only wait once on a condition - additional calls are valid but
		727	will return immediately.
		728
		729	If an error condition has been set by calling C<< ->croak >>, then this
		730	function will call C<croak>.
		731
		732	In list context, all parameters passed to C<send> will be returned,
		733	in scalar context only the first one will be returned.
		734
		735	Note that doing a blocking wait in a callback is not supported by any
		736	event loop, that is, recursive invocation of a blocking C<< ->recv
		737	>> is not allowed, and the C<recv> call will C<croak> if such a
		738	condition is detected. This condition can be slightly loosened by using
		739	L<Coro::AnyEvent>, which allows you to do a blocking C<< ->recv >> from
		740	any thread that doesn't run the event loop itself.
		741
		742	Not all event models support a blocking wait - some die in that case
		743	(programs might want to do that to stay interactive), so I<if you are
		744	using this from a module, never require a blocking wait>. Instead, let the
		745	caller decide whether the call will block or not (for example, by coupling
		746	condition variables with some kind of request results and supporting
		747	callbacks so the caller knows that getting the result will not block,
		748	while still supporting blocking waits if the caller so desires).
		749
		750	You can ensure that C<< -recv >> never blocks by setting a callback and
		751	only calling C<< ->recv >> from within that callback (or at a later
		752	time). This will work even when the event loop does not support blocking
		753	waits otherwise.
		754
		755	=item $bool = $cv->ready
		756
		757	Returns true when the condition is "true", i.e. whether C<send> or
		758	C<croak> have been called.
		759
		760	=item $cb = $cv->cb ($cb->($cv))
		761
		762	This is a mutator function that returns the callback set and optionally
		763	replaces it before doing so.
		764
		765	The callback will be called when the condition becomes "true", i.e. when
		766	C<send> or C<croak> are called, with the only argument being the condition
		767	variable itself. Calling C<recv> inside the callback or at any later time
		768	is guaranteed not to block.
		769
		770	=back
		771
		772	=head1 SUPPORTED EVENT LOOPS/BACKENDS
		773
		774	The available backend classes are (every class has its own manpage):
		775
		776	=over 4
		777
		778	=item Backends that are autoprobed when no other event loop can be found.
		779
		780	EV is the preferred backend when no other event loop seems to be in
		781	use. If EV is not installed, then AnyEvent will try Event, and, failing
		782	that, will fall back to its own pure-perl implementation, which is
		783	available everywhere as it comes with AnyEvent itself.
		784
		785	AnyEvent::Impl::EV based on EV (interface to libev, best choice).
		786	AnyEvent::Impl::Event based on Event, very stable, few glitches.
		787	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
		788
		789	=item Backends that are transparently being picked up when they are used.
		790
		791	These will be used when they are currently loaded when the first watcher
		792	is created, in which case it is assumed that the application is using
		793	them. This means that AnyEvent will automatically pick the right backend
		794	when the main program loads an event module before anything starts to
		795	create watchers. Nothing special needs to be done by the main program.
		796
		797	AnyEvent::Impl::Glib based on Glib, slow but very stable.
		798	AnyEvent::Impl::Tk based on Tk, very broken.
		799	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
		800	AnyEvent::Impl::POE based on POE, very slow, some limitations.
		801
		802	=item Backends with special needs.
		803
		804	Qt requires the Qt::Application to be instantiated first, but will
		805	otherwise be picked up automatically. As long as the main program
		806	instantiates the application before any AnyEvent watchers are created,
		807	everything should just work.
		808
		809	AnyEvent::Impl::Qt based on Qt.
		810
		811	Support for IO::Async can only be partial, as it is too broken and
		812	architecturally limited to even support the AnyEvent API. It also
		813	is the only event loop that needs the loop to be set explicitly, so
		814	it can only be used by a main program knowing about AnyEvent. See
		815	L<AnyEvent::Impl::Async> for the gory details.
		816
		817	AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
		818
		819	=item Event loops that are indirectly supported via other backends.
		820
		821	Some event loops can be supported via other modules:
		822
		823	There is no direct support for WxWidgets (L<Wx>) or L<Prima>.
		824
		825	B<WxWidgets> has no support for watching file handles. However, you can
		826	use WxWidgets through the POE adaptor, as POE has a Wx backend that simply
		827	polls 20 times per second, which was considered to be too horrible to even
		828	consider for AnyEvent.
		829
		830	B<Prima> is not supported as nobody seems to be using it, but it has a POE
		831	backend, so it can be supported through POE.
		832
		833	AnyEvent knows about both L<Prima> and L<Wx>, however, and will try to
		834	load L<POE> when detecting them, in the hope that POE will pick them up,
		835	in which case everything will be automatic.
		836
		837	=back
		838
		839	=head1 GLOBAL VARIABLES AND FUNCTIONS
		840
		841	These are not normally required to use AnyEvent, but can be useful to
		842	write AnyEvent extension modules.
		843
		844	=over 4
		845
190	=item $AnyEvent::MODEL	846	=item $AnyEvent::MODEL
191		847
192	Contains C<undef> until the first watcher is being created. Then it	848	Contains C<undef> until the first watcher is being created, before the
		849	backend has been autodetected.
		850
193	contains the event model that is being used, which is the name of the	851	Afterwards it contains the event model that is being used, which is the
194	Perl class implementing the model. This class is usually one of the	852	name of the Perl class implementing the model. This class is usually one
195	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	853	of the C<AnyEvent::Impl:xxx> modules, but can be any other class in the
196	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	854	case AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode> it
197		855	will be C<urxvt::anyevent>).
198	The known classes so far are:
199
200	AnyEvent::Impl::Coro based on Coro::Event, best choise.
201	AnyEvent::Impl::Event based on Event, also best choice :)
202	AnyEvent::Impl::Glib based on Glib, second-best choice.
203	AnyEvent::Impl::Tk based on Tk, very bad choice.
204	AnyEvent::Impl::Perl pure-perl implementation, inefficient.
205		856
206	=item AnyEvent::detect	857	=item AnyEvent::detect
207		858
208	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model if	859	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
209	necessary. You should only call this function right before you would have	860	if necessary. You should only call this function right before you would
210	created an AnyEvent watcher anyway, that is, very late at runtime.	861	have created an AnyEvent watcher anyway, that is, as late as possible at
		862	runtime, and not e.g. while initialising of your module.
		863
		864	If you need to do some initialisation before AnyEvent watchers are
		865	created, use C<post_detect>.
		866
		867	=item $guard = AnyEvent::post_detect { BLOCK }
		868
		869	Arranges for the code block to be executed as soon as the event model is
		870	autodetected (or immediately if this has already happened).
		871
		872	The block will be executed I<after> the actual backend has been detected
		873	(C<$AnyEvent::MODEL> is set), but I<before> any watchers have been
		874	created, so it is possible to e.g. patch C<@AnyEvent::ISA> or do
		875	other initialisations - see the sources of L<AnyEvent::Strict> or
		876	L<AnyEvent::AIO> to see how this is used.
		877
		878	The most common usage is to create some global watchers, without forcing
		879	event module detection too early, for example, L<AnyEvent::AIO> creates
		880	and installs the global L<IO::AIO> watcher in a C<post_detect> block to
		881	avoid autodetecting the event module at load time.
		882
		883	If called in scalar or list context, then it creates and returns an object
		884	that automatically removes the callback again when it is destroyed (or
		885	C<undef> when the hook was immediately executed). See L<AnyEvent::AIO> for
		886	a case where this is useful.
		887
		888	Example: Create a watcher for the IO::AIO module and store it in
		889	C<$WATCHER>. Only do so after the event loop is initialised, though.
		890
		891	our WATCHER;
		892
		893	my $guard = AnyEvent::post_detect {
		894	$WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
		895	};
		896
		897	# the ||= is important in case post_detect immediately runs the block,
		898	# as to not clobber the newly-created watcher. assigning both watcher and
		899	# post_detect guard to the same variable has the advantage of users being
		900	# able to just C<undef $WATCHER> if the watcher causes them grief.
		901
		902	$WATCHER ||= $guard;
		903
		904	=item @AnyEvent::post_detect
		905
		906	If there are any code references in this array (you can C<push> to it
		907	before or after loading AnyEvent), then they will called directly after
		908	the event loop has been chosen.
		909
		910	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		911	if it is defined then the event loop has already been detected, and the
		912	array will be ignored.
		913
		914	Best use C<AnyEvent::post_detect { BLOCK }> when your application allows
		915	it,as it takes care of these details.
		916
		917	This variable is mainly useful for modules that can do something useful
		918	when AnyEvent is used and thus want to know when it is initialised, but do
		919	not need to even load it by default. This array provides the means to hook
		920	into AnyEvent passively, without loading it.
211		921
212	=back	922	=back
213		923
214	=head1 WHAT TO DO IN A MODULE	924	=head1 WHAT TO DO IN A MODULE
215		925
216	As a module author, you should "use AnyEvent" and call AnyEvent methods	926	As a module author, you should C<use AnyEvent> and call AnyEvent methods
217	freely, but you should not load a specific event module or rely on it.	927	freely, but you should not load a specific event module or rely on it.
218		928
219	Be careful when you create watchers in the module body - Anyevent will	929	Be careful when you create watchers in the module body - AnyEvent will
220	decide which event module to use as soon as the first method is called, so	930	decide which event module to use as soon as the first method is called, so
221	by calling AnyEvent in your module body you force the user of your module	931	by calling AnyEvent in your module body you force the user of your module
222	to load the event module first.	932	to load the event module first.
223		933
		934	Never call C<< ->recv >> on a condition variable unless you I<know> that
		935	the C<< ->send >> method has been called on it already. This is
		936	because it will stall the whole program, and the whole point of using
		937	events is to stay interactive.
		938
		939	It is fine, however, to call C<< ->recv >> when the user of your module
		940	requests it (i.e. if you create a http request object ad have a method
		941	called C<results> that returns the results, it should call C<< ->recv >>
		942	freely, as the user of your module knows what she is doing. always).
		943
224	=head1 WHAT TO DO IN THE MAIN PROGRAM	944	=head1 WHAT TO DO IN THE MAIN PROGRAM
225		945
226	There will always be a single main program - the only place that should	946	There will always be a single main program - the only place that should
227	dictate which event model to use.	947	dictate which event model to use.
228		948
229	If it doesn't care, it can just "use AnyEvent" and use it itself, or not	949	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
230	do anything special and let AnyEvent decide which implementation to chose.	950	do anything special (it does not need to be event-based) and let AnyEvent
		951	decide which implementation to chose if some module relies on it.
231		952
232	If the main program relies on a specific event model (for example, in Gtk2	953	If the main program relies on a specific event model - for example, in
233	programs you have to rely on either Glib or Glib::Event), you should load	954	Gtk2 programs you have to rely on the Glib module - you should load the
234	it before loading AnyEvent or any module that uses it, generally, as early	955	event module before loading AnyEvent or any module that uses it: generally
235	as possible. The reason is that modules might create watchers when they	956	speaking, you should load it as early as possible. The reason is that
236	are loaded, and AnyEvent will decide on the event model to use as soon as	957	modules might create watchers when they are loaded, and AnyEvent will
237	it creates watchers, and it might chose the wrong one unless you load the	958	decide on the event model to use as soon as it creates watchers, and it
238	correct one yourself.	959	might chose the wrong one unless you load the correct one yourself.
239		960
240	You can chose to use a rather inefficient pure-perl implementation by	961	You can chose to use a pure-perl implementation by loading the
241	loading the C<AnyEvent::Impl::Perl> module, but letting AnyEvent chose is	962	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
242	generally better.	963	everywhere, but letting AnyEvent chose the model is generally better.
		964
		965	=head2 MAINLOOP EMULATION
		966
		967	Sometimes (often for short test scripts, or even standalone programs who
		968	only want to use AnyEvent), you do not want to run a specific event loop.
		969
		970	In that case, you can use a condition variable like this:
		971
		972	AnyEvent->condvar->recv;
		973
		974	This has the effect of entering the event loop and looping forever.
		975
		976	Note that usually your program has some exit condition, in which case
		977	it is better to use the "traditional" approach of storing a condition
		978	variable somewhere, waiting for it, and sending it when the program should
		979	exit cleanly.
		980
		981
		982	=head1 OTHER MODULES
		983
		984	The following is a non-exhaustive list of additional modules that use
		985	AnyEvent as a client and can therefore be mixed easily with other AnyEvent
		986	modules and other event loops in the same program. Some of the modules
		987	come with AnyEvent, most are available via CPAN.
		988
		989	=over 4
		990
		991	=item L<AnyEvent::Util>
		992
		993	Contains various utility functions that replace often-used but blocking
		994	functions such as C<inet_aton> by event-/callback-based versions.
		995
		996	=item L<AnyEvent::Socket>
		997
		998	Provides various utility functions for (internet protocol) sockets,
		999	addresses and name resolution. Also functions to create non-blocking tcp
		1000	connections or tcp servers, with IPv6 and SRV record support and more.
		1001
		1002	=item L<AnyEvent::Handle>
		1003
		1004	Provide read and write buffers, manages watchers for reads and writes,
		1005	supports raw and formatted I/O, I/O queued and fully transparent and
		1006	non-blocking SSL/TLS (via L<AnyEvent::TLS>.
		1007
		1008	=item L<AnyEvent::DNS>
		1009
		1010	Provides rich asynchronous DNS resolver capabilities.
		1011
		1012	=item L<AnyEvent::HTTP>
		1013
		1014	A simple-to-use HTTP library that is capable of making a lot of concurrent
		1015	HTTP requests.
		1016
		1017	=item L<AnyEvent::HTTPD>
		1018
		1019	Provides a simple web application server framework.
		1020
		1021	=item L<AnyEvent::FastPing>
		1022
		1023	The fastest ping in the west.
		1024
		1025	=item L<AnyEvent::DBI>
		1026
		1027	Executes L<DBI> requests asynchronously in a proxy process.
		1028
		1029	=item L<AnyEvent::AIO>
		1030
		1031	Truly asynchronous I/O, should be in the toolbox of every event
		1032	programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
		1033	together.
		1034
		1035	=item L<AnyEvent::BDB>
		1036
		1037	Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
		1038	L<BDB> and AnyEvent together.
		1039
		1040	=item L<AnyEvent::GPSD>
		1041
		1042	A non-blocking interface to gpsd, a daemon delivering GPS information.
		1043
		1044	=item L<AnyEvent::IRC>
		1045
		1046	AnyEvent based IRC client module family (replacing the older Net::IRC3).
		1047
		1048	=item L<AnyEvent::XMPP>
		1049
		1050	AnyEvent based XMPP (Jabber protocol) module family (replacing the older
		1051	Net::XMPP2>.
		1052
		1053	=item L<AnyEvent::IGS>
		1054
		1055	A non-blocking interface to the Internet Go Server protocol (used by
		1056	L<App::IGS>).
		1057
		1058	=item L<Net::FCP>
		1059
		1060	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		1061	of AnyEvent.
		1062
		1063	=item L<Event::ExecFlow>
		1064
		1065	High level API for event-based execution flow control.
		1066
		1067	=item L<Coro>
		1068
		1069	Has special support for AnyEvent via L<Coro::AnyEvent>.
		1070
		1071	=back
243		1072
244	=cut	1073	=cut
245		1074
246	package AnyEvent;	1075	package AnyEvent;
247		1076
		1077	# basically a tuned-down version of common::sense
		1078	sub common_sense {
248	no warnings;	1079	# no warnings
249	use strict;	1080	${^WARNING_BITS} ^= ${^WARNING_BITS};
		1081	# use strict vars subs
		1082	$^H |= 0x00000600;
		1083	}
		1084
		1085	BEGIN { AnyEvent::common_sense }
		1086
250	use Carp;	1087	use Carp ();
251		1088
252	our $VERSION = '2.52';	1089	our $VERSION = 4.86;
253	our $MODEL;	1090	our $MODEL;
254		1091
255	our $AUTOLOAD;	1092	our $AUTOLOAD;
256	our @ISA;	1093	our @ISA;
257		1094
258	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
259
260	our @REGISTRY;	1095	our @REGISTRY;
261		1096
		1097	our $WIN32;
		1098
		1099	our $VERBOSE;
		1100
		1101	BEGIN {
		1102	eval "sub WIN32(){ " . (($^O =~ /mswin32/i)*1) ." }";
		1103	eval "sub TAINT(){ " . (${^TAINT}*1) . " }";
		1104
		1105	delete @ENV{grep /^PERL_ANYEVENT_/, keys %ENV}
		1106	if ${^TAINT};
		1107
		1108	$VERBOSE = $ENV{PERL_ANYEVENT_VERBOSE}*1;
		1109
		1110	}
		1111
		1112	our $MAX_SIGNAL_LATENCY = 10;
		1113
		1114	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		1115
		1116	{
		1117	my $idx;
		1118	$PROTOCOL{$_} = ++$idx
		1119	for reverse split /\s*,\s*/,
		1120	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		1121	}
		1122
262	my @models = (	1123	my @models = (
263	[Coro::Event:: => AnyEvent::Impl::Coro::],	1124	[EV:: => AnyEvent::Impl::EV::],
264	[Event:: => AnyEvent::Impl::Event::],	1125	[Event:: => AnyEvent::Impl::Event::],
265	[Glib:: => AnyEvent::Impl::Glib::],
266	[Tk:: => AnyEvent::Impl::Tk::],
267	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	1126	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
		1127	# everything below here will not be autoprobed
		1128	# as the pureperl backend should work everywhere
		1129	# and is usually faster
		1130	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
		1131	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
		1132	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		1133	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
		1134	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
		1135	[Wx:: => AnyEvent::Impl::POE::],
		1136	[Prima:: => AnyEvent::Impl::POE::],
		1137	# IO::Async is just too broken - we would need workarounds for its
		1138	# byzantine signal and broken child handling, among others.
		1139	# IO::Async is rather hard to detect, as it doesn't have any
		1140	# obvious default class.
		1141	# [IO::Async:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1142	# [IO::Async::Loop:: => AnyEvent::Impl::IOAsync::], # requires special main program
		1143	# [IO::Async::Notifier:: => AnyEvent::Impl::IOAsync::], # requires special main program
268	);	1144	);
269		1145
270	our %method = map +($_ => 1), qw(io timer condvar broadcast wait signal one_event DESTROY);	1146	our %method = map +($_ => 1),
		1147	qw(io timer time now now_update signal child idle condvar one_event DESTROY);
		1148
		1149	our @post_detect;
		1150
		1151	sub post_detect(&) {
		1152	my ($cb) = @_;
		1153
		1154	if ($MODEL) {
		1155	$cb->();
		1156
		1157	1
		1158	} else {
		1159	push @post_detect, $cb;
		1160
		1161	defined wantarray
		1162	? bless \$cb, "AnyEvent::Util::postdetect"
		1163	: ()
		1164	}
		1165	}
		1166
		1167	sub AnyEvent::Util::postdetect::DESTROY {
		1168	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		1169	}
271		1170
272	sub detect() {	1171	sub detect() {
273	unless ($MODEL) {	1172	unless ($MODEL) {
274	no strict 'refs';	1173	local $SIG{__DIE__};
		1174
		1175	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
		1176	my $model = "AnyEvent::Impl::$1";
		1177	if (eval "require $model") {
		1178	$MODEL = $model;
		1179	warn "AnyEvent: loaded model '$model' (forced by \$ENV{PERL_ANYEVENT_MODEL}), using it.\n" if $VERBOSE >= 2;
		1180	} else {
		1181	warn "AnyEvent: unable to load model '$model' (from \$ENV{PERL_ANYEVENT_MODEL}):\n$@" if $VERBOSE;
		1182	}
		1183	}
275		1184
276	# check for already loaded models	1185	# check for already loaded models
		1186	unless ($MODEL) {
277	for (@REGISTRY, @models) {	1187	for (@REGISTRY, @models) {
278	my ($package, $model) = @$_;	1188	my ($package, $model) = @$_;
279	if (${"$package\::VERSION"} > 0) {	1189	if (${"$package\::VERSION"} > 0) {
280	if (eval "require $model") {	1190	if (eval "require $model") {
281	$MODEL = $model;	1191	$MODEL = $model;
282	warn "AnyEvent: found model '$model', using it.\n" if $verbose > 1;	1192	warn "AnyEvent: autodetected model '$model', using it.\n" if $VERBOSE >= 2;
283	last;	1193	last;
		1194	}
284	}	1195	}
285	}	1196	}
		1197
		1198	unless ($MODEL) {
		1199	# try to load a model
		1200
		1201	for (@REGISTRY, @models) {
		1202	my ($package, $model) = @$_;
		1203	if (eval "require $package"
		1204	and ${"$package\::VERSION"} > 0
		1205	and eval "require $model") {
		1206	$MODEL = $model;
		1207	warn "AnyEvent: autoprobed model '$model', using it.\n" if $VERBOSE >= 2;
		1208	last;
		1209	}
		1210	}
		1211
		1212	$MODEL
		1213	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.\n";
		1214	}
286	}	1215	}
287		1216
288	unless ($MODEL) {	1217	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
289	# try to load a model
290
291	for (@REGISTRY, @models) {
292	my ($package, $model) = @$_;
293	if (eval "require $package"
294	and ${"$package\::VERSION"} > 0
295	and eval "require $model") {
296	$MODEL = $model;
297	warn "AnyEvent: autoprobed and loaded model '$model', using it.\n" if $verbose > 1;
298	last;
299	}
300	}
301
302	$MODEL
303	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: Event (or Coro+Event), Glib or Tk.";
304	}
305		1218
306	unshift @ISA, $MODEL;	1219	unshift @ISA, $MODEL;
307	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	1220
		1221	require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
		1222
		1223	(shift @post_detect)->() while @post_detect;
308	}	1224	}
309		1225
310	$MODEL	1226	$MODEL
311	}	1227	}
312		1228
313	sub AUTOLOAD {	1229	sub AUTOLOAD {
314	(my $func = $AUTOLOAD) =~ s/.*://;	1230	(my $func = $AUTOLOAD) =~ s/.*://;
315		1231
316	$method{$func}	1232	$method{$func}
317	or croak "$func: not a valid method for AnyEvent objects";	1233	or Carp::croak "$func: not a valid method for AnyEvent objects";
318		1234
319	detect unless $MODEL;	1235	detect unless $MODEL;
320		1236
321	my $class = shift;	1237	my $class = shift;
322	$class->$func (@_);	1238	$class->$func (@_);
323	}	1239	}
324		1240
		1241	# utility function to dup a filehandle. this is used by many backends
		1242	# to support binding more than one watcher per filehandle (they usually
		1243	# allow only one watcher per fd, so we dup it to get a different one).
		1244	sub _dupfh($$;$$) {
		1245	my ($poll, $fh, $r, $w) = @_;
		1246
		1247	# cygwin requires the fh mode to be matching, unix doesn't
		1248	my ($rw, $mode) = $poll eq "r" ? ($r, "<&") : ($w, ">&");
		1249
		1250	open my $fh2, $mode, $fh
		1251	or die "AnyEvent->io: cannot dup() filehandle in mode '$poll': $!,";
		1252
		1253	# we assume CLOEXEC is already set by perl in all important cases
		1254
		1255	($fh2, $rw)
		1256	}
		1257
325	package AnyEvent::Base;	1258	package AnyEvent::Base;
326		1259
		1260	# default implementations for many methods
		1261
		1262	sub _time {
		1263	# probe for availability of Time::HiRes
		1264	if (eval "use Time::HiRes (); Time::HiRes::time (); 1") {
		1265	warn "AnyEvent: using Time::HiRes for sub-second timing accuracy.\n" if $VERBOSE >= 8;
		1266	*_time = \&Time::HiRes::time;
		1267	# if (eval "use POSIX (); (POSIX::times())...
		1268	} else {
		1269	warn "AnyEvent: using built-in time(), WARNING, no sub-second resolution!\n" if $VERBOSE;
		1270	*_time = sub { time }; # epic fail
		1271	}
		1272
		1273	&_time
		1274	}
		1275
		1276	sub time { _time }
		1277	sub now { _time }
		1278	sub now_update { }
		1279
327	# default implementation for ->condvar, ->wait, ->broadcast	1280	# default implementation for ->condvar
328		1281
329	sub condvar {	1282	sub condvar {
330	bless \my $flag, "AnyEvent::Base::CondVar"	1283	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, "AnyEvent::CondVar"
331	}
332
333	sub AnyEvent::Base::CondVar::broadcast {
334	${$_[0]}++;
335	}
336
337	sub AnyEvent::Base::CondVar::wait {
338	AnyEvent->one_event while !${$_[0]};
339	}	1284	}
340		1285
341	# default implementation for ->signal	1286	# default implementation for ->signal
342		1287
343	our %SIG_CB;	1288	our $HAVE_ASYNC_INTERRUPT;
		1289	our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
		1290	our (%SIG_ASY, %SIG_ASY_W);
		1291	our ($SIG_COUNT, $SIG_TW);
344		1292
		1293	sub _signal_exec {
		1294	$HAVE_ASYNC_INTERRUPT
		1295	? $SIGPIPE_R->drain
		1296	: sysread $SIGPIPE_R, my $dummy, 9;
		1297
		1298	while (%SIG_EV) {
		1299	for (keys %SIG_EV) {
		1300	delete $SIG_EV{$_};
		1301	$_->() for values %{ $SIG_CB{$_} || {} };
		1302	}
		1303	}
		1304	}
		1305
		1306	# install a dumym wakeupw atcher to reduce signal catching latency
		1307	sub _sig_add() {
		1308	unless ($SIG_COUNT++) {
		1309	# try to align timer on a full-second boundary, if possible
		1310	my $NOW = AnyEvent->now;
		1311
		1312	$SIG_TW = AnyEvent->timer (
		1313	after => $MAX_SIGNAL_LATENCY - ($NOW - int $NOW),
		1314	interval => $MAX_SIGNAL_LATENCY,
		1315	cb => sub { }, # just for the PERL_ASYNC_CHECK
		1316	);
		1317	}
		1318	}
		1319
		1320	sub _sig_del {
		1321	undef $SIG_TW
		1322	unless --$SIG_COUNT;
		1323	}
		1324
345	sub signal {	1325	sub _signal {
346	my (undef, %arg) = @_;	1326	my (undef, %arg) = @_;
347		1327
348	my $signal = uc $arg{signal}	1328	my $signal = uc $arg{signal}
349	or Carp::croak "required option 'signal' is missing";	1329	or Carp::croak "required option 'signal' is missing";
350		1330
351	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};	1331	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1332
		1333	if ($HAVE_ASYNC_INTERRUPT) {
		1334	# async::interrupt
		1335
		1336	$SIG_ASY{$signal} ||= do {
		1337	my $asy = new Async::Interrupt
		1338	cb => sub { undef $SIG_EV{$signal} },
		1339	signal => $signal,
		1340	pipe => [$SIGPIPE_R->filenos],
		1341	;
		1342	$asy->pipe_autodrain (0);
		1343
		1344	$asy
		1345	};
		1346
		1347	} else {
		1348	# pure perl
		1349
352	$SIG{$signal} ||= sub {	1350	$SIG{$signal} ||= sub {
353	$_->() for values %{ $SIG_CB{$signal} || {} };	1351	local $!;
		1352	syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
		1353	undef $SIG_EV{$signal};
		1354	};
		1355
		1356	# can't do signal processing without introducing races in pure perl,
		1357	# so limit the signal latency.
		1358	_sig_add;
354	};	1359	}
355		1360
356	bless [$signal, $arg{cb}], "AnyEvent::Base::Signal"	1361	bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
357	}	1362	}
358		1363
		1364	sub signal {
		1365	# probe for availability of Async::Interrupt
		1366	if (!$ENV{PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT} && eval "use Async::Interrupt 0.6 (); 1") {
		1367	warn "AnyEvent: using Async::Interrupt for race-free signal handling.\n" if $VERBOSE >= 8;
		1368
		1369	$HAVE_ASYNC_INTERRUPT = 1;
		1370	$SIGPIPE_R = new Async::Interrupt::EventPipe;
		1371	$SIG_IO = AnyEvent->io (fh => $SIGPIPE_R->fileno, poll => "r", cb => \&_signal_exec);
		1372
		1373	} else {
		1374	warn "AnyEvent: using emulated perl signal handling with latency timer.\n" if $VERBOSE >= 8;
		1375
		1376	require Fcntl;
		1377
		1378	if (AnyEvent::WIN32) {
		1379	require AnyEvent::Util;
		1380
		1381	($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
		1382	AnyEvent::Util::fh_nonblocking ($SIGPIPE_R) if $SIGPIPE_R;
		1383	AnyEvent::Util::fh_nonblocking ($SIGPIPE_W) if $SIGPIPE_W; # just in case
		1384	} else {
		1385	pipe $SIGPIPE_R, $SIGPIPE_W;
		1386	fcntl $SIGPIPE_R, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_R;
		1387	fcntl $SIGPIPE_W, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_W; # just in case
		1388
		1389	# not strictly required, as $^F is normally 2, but let's make sure...
		1390	fcntl $SIGPIPE_R, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
		1391	fcntl $SIGPIPE_W, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
		1392	}
		1393
		1394	$SIGPIPE_R
		1395	or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
		1396
		1397	$SIG_IO = AnyEvent->io (fh => $SIGPIPE_R, poll => "r", cb => \&_signal_exec);
		1398	}
		1399
		1400	*signal = \&_signal;
		1401	&signal
		1402	}
		1403
359	sub AnyEvent::Base::Signal::DESTROY {	1404	sub AnyEvent::Base::signal::DESTROY {
360	my ($signal, $cb) = @{$_[0]};	1405	my ($signal, $cb) = @{$_[0]};
361		1406
		1407	_sig_del;
		1408
362	delete $SIG_CB{$signal}{$cb};	1409	delete $SIG_CB{$signal}{$cb};
363		1410
364	$SIG{$signal} = 'DEFAULT' unless keys %{ $SIG_CB{$signal} };	1411	$HAVE_ASYNC_INTERRUPT
		1412	? delete $SIG_ASY{$signal}
		1413	: # delete doesn't work with older perls - they then
		1414	# print weird messages, or just unconditionally exit
		1415	# instead of getting the default action.
		1416	undef $SIG{$signal}
		1417	unless keys %{ $SIG_CB{$signal} };
365	}	1418	}
366		1419
367	# default implementation for ->child	1420	# default implementation for ->child
368		1421
369	our %PID_CB;	1422	our %PID_CB;
370	our $CHLD_W;	1423	our $CHLD_W;
371	our $PID_IDLE;	1424	our $CHLD_DELAY_W;
372	our $WNOHANG;	1425	our $WNOHANG;
373		1426
374	sub _child_wait {	1427	sub _sigchld {
375	while (0 < (my $pid = waitpid -1, $WNOHANG)) {	1428	while (0 < (my $pid = waitpid -1, $WNOHANG)) {
		1429	$_->($pid, $?)
376	$_->() for values %{ (delete $PID_CB{$pid}) || {} };	1430	for values %{ $PID_CB{$pid} || {} },
		1431	values %{ $PID_CB{0} || {} };
377	}	1432	}
378
379	undef $PID_IDLE;
380	}	1433	}
381		1434
382	sub child {	1435	sub child {
383	my (undef, %arg) = @_;	1436	my (undef, %arg) = @_;
384		1437
385	my $pid = uc $arg{pid}	1438	defined (my $pid = $arg{pid} + 0)
386	or Carp::croak "required option 'pid' is missing";	1439	or Carp::croak "required option 'pid' is missing";
387		1440
388	$PID_CB{$pid}{$arg{cb}} = $arg{cb};	1441	$PID_CB{$pid}{$arg{cb}} = $arg{cb};
389		1442
390	unless ($WNOHANG) {	1443	# WNOHANG is almost cetrainly 1 everywhere
391	$WNOHANG = eval { require POSIX; &POSIX::WNOHANG } || 1;	1444	$WNOHANG ||= $^O =~ /^(?:openbsd|netbsd|linux|freebsd|cygwin|MSWin32)$/
		1445	? 1
		1446	: eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
		1447
		1448	unless ($CHLD_W) {
		1449	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
		1450	# child could be a zombie already, so make at least one round
		1451	&_sigchld;
392	}	1452	}
393		1453
394	unless ($CHLD_W) {
395	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_child_wait);
396	# child could be a zombie already
397	$PID_IDLE ||= AnyEvent->timer (after => 0, cb => \&_child_wait);
398	}
399
400	bless [$pid, $arg{cb}], "AnyEvent::Base::Child"	1454	bless [$pid, $arg{cb}], "AnyEvent::Base::child"
401	}	1455	}
402		1456
403	sub AnyEvent::Base::Child::DESTROY {	1457	sub AnyEvent::Base::child::DESTROY {
404	my ($pid, $cb) = @{$_[0]};	1458	my ($pid, $cb) = @{$_[0]};
405		1459
406	delete $PID_CB{$pid}{$cb};	1460	delete $PID_CB{$pid}{$cb};
407	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };	1461	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
408		1462
409	undef $CHLD_W unless keys %PID_CB;	1463	undef $CHLD_W unless keys %PID_CB;
410	}	1464	}
411		1465
		1466	# idle emulation is done by simply using a timer, regardless
		1467	# of whether the process is idle or not, and not letting
		1468	# the callback use more than 50% of the time.
		1469	sub idle {
		1470	my (undef, %arg) = @_;
		1471
		1472	my ($cb, $w, $rcb) = $arg{cb};
		1473
		1474	$rcb = sub {
		1475	if ($cb) {
		1476	$w = _time;
		1477	&$cb;
		1478	$w = _time - $w;
		1479
		1480	# never use more then 50% of the time for the idle watcher,
		1481	# within some limits
		1482	$w = 0.0001 if $w < 0.0001;
		1483	$w = 5 if $w > 5;
		1484
		1485	$w = AnyEvent->timer (after => $w, cb => $rcb);
		1486	} else {
		1487	# clean up...
		1488	undef $w;
		1489	undef $rcb;
		1490	}
		1491	};
		1492
		1493	$w = AnyEvent->timer (after => 0.05, cb => $rcb);
		1494
		1495	bless \\$cb, "AnyEvent::Base::idle"
		1496	}
		1497
		1498	sub AnyEvent::Base::idle::DESTROY {
		1499	undef $${$_[0]};
		1500	}
		1501
		1502	package AnyEvent::CondVar;
		1503
		1504	our @ISA = AnyEvent::CondVar::Base::;
		1505
		1506	package AnyEvent::CondVar::Base;
		1507
		1508	#use overload
		1509	# '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1510	# fallback => 1;
		1511
		1512	# save 300+ kilobytes by dirtily hardcoding overloading
		1513	${"AnyEvent::CondVar::Base::OVERLOAD"}{dummy}++; # Register with magic by touching.
		1514	*{'AnyEvent::CondVar::Base::()'} = sub { }; # "Make it findable via fetchmethod."
		1515	*{'AnyEvent::CondVar::Base::(&{}'} = sub { my $self = shift; sub { $self->send (@_) } }; # &{}
		1516	${'AnyEvent::CondVar::Base::()'} = 1; # fallback
		1517
		1518	our $WAITING;
		1519
		1520	sub _send {
		1521	# nop
		1522	}
		1523
		1524	sub send {
		1525	my $cv = shift;
		1526	$cv->{_ae_sent} = [@_];
		1527	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1528	$cv->_send;
		1529	}
		1530
		1531	sub croak {
		1532	$_[0]{_ae_croak} = $_[1];
		1533	$_[0]->send;
		1534	}
		1535
		1536	sub ready {
		1537	$_[0]{_ae_sent}
		1538	}
		1539
		1540	sub _wait {
		1541	$WAITING
		1542	and !$_[0]{_ae_sent}
		1543	and Carp::croak "AnyEvent::CondVar: recursive blocking wait detected";
		1544
		1545	local $WAITING = 1;
		1546	AnyEvent->one_event while !$_[0]{_ae_sent};
		1547	}
		1548
		1549	sub recv {
		1550	$_[0]->_wait;
		1551
		1552	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		1553	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		1554	}
		1555
		1556	sub cb {
		1557	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		1558	$_[0]{_ae_cb}
		1559	}
		1560
		1561	sub begin {
		1562	++$_[0]{_ae_counter};
		1563	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		1564	}
		1565
		1566	sub end {
		1567	return if --$_[0]{_ae_counter};
		1568	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		1569	}
		1570
		1571	# undocumented/compatibility with pre-3.4
		1572	*broadcast = \&send;
		1573	*wait = \&_wait;
		1574
		1575	=head1 ERROR AND EXCEPTION HANDLING
		1576
		1577	In general, AnyEvent does not do any error handling - it relies on the
		1578	caller to do that if required. The L<AnyEvent::Strict> module (see also
		1579	the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
		1580	checking of all AnyEvent methods, however, which is highly useful during
		1581	development.
		1582
		1583	As for exception handling (i.e. runtime errors and exceptions thrown while
		1584	executing a callback), this is not only highly event-loop specific, but
		1585	also not in any way wrapped by this module, as this is the job of the main
		1586	program.
		1587
		1588	The pure perl event loop simply re-throws the exception (usually
		1589	within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
		1590	$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
		1591	so on.
		1592
		1593	=head1 ENVIRONMENT VARIABLES
		1594
		1595	The following environment variables are used by this module or its
		1596	submodules.
		1597
		1598	Note that AnyEvent will remove I<all> environment variables starting with
		1599	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		1600	enabled.
		1601
		1602	=over 4
		1603
		1604	=item C<PERL_ANYEVENT_VERBOSE>
		1605
		1606	By default, AnyEvent will be completely silent except in fatal
		1607	conditions. You can set this environment variable to make AnyEvent more
		1608	talkative.
		1609
		1610	When set to C<1> or higher, causes AnyEvent to warn about unexpected
		1611	conditions, such as not being able to load the event model specified by
		1612	C<PERL_ANYEVENT_MODEL>.
		1613
		1614	When set to C<2> or higher, cause AnyEvent to report to STDERR which event
		1615	model it chooses.
		1616
		1617	When set to C<8> or higher, then AnyEvent will report extra information on
		1618	which optional modules it loads and how it implements certain features.
		1619
		1620	=item C<PERL_ANYEVENT_STRICT>
		1621
		1622	AnyEvent does not do much argument checking by default, as thorough
		1623	argument checking is very costly. Setting this variable to a true value
		1624	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		1625	check the arguments passed to most method calls. If it finds any problems,
		1626	it will croak.
		1627
		1628	In other words, enables "strict" mode.
		1629
		1630	Unlike C<use strict> (or it's modern cousin, C<< use L<common::sense>
		1631	>>, it is definitely recommended to keep it off in production. Keeping
		1632	C<PERL_ANYEVENT_STRICT=1> in your environment while developing programs
		1633	can be very useful, however.
		1634
		1635	=item C<PERL_ANYEVENT_MODEL>
		1636
		1637	This can be used to specify the event model to be used by AnyEvent, before
		1638	auto detection and -probing kicks in. It must be a string consisting
		1639	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
		1640	and the resulting module name is loaded and if the load was successful,
		1641	used as event model. If it fails to load AnyEvent will proceed with
		1642	auto detection and -probing.
		1643
		1644	This functionality might change in future versions.
		1645
		1646	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
		1647	could start your program like this:
		1648
		1649	PERL_ANYEVENT_MODEL=Perl perl ...
		1650
		1651	=item C<PERL_ANYEVENT_PROTOCOLS>
		1652
		1653	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1654	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1655	of auto probing).
		1656
		1657	Must be set to a comma-separated list of protocols or address families,
		1658	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1659	used, and preference will be given to protocols mentioned earlier in the
		1660	list.
		1661
		1662	This variable can effectively be used for denial-of-service attacks
		1663	against local programs (e.g. when setuid), although the impact is likely
		1664	small, as the program has to handle conenction and other failures anyways.
		1665
		1666	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1667	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1668	- only support IPv4, never try to resolve or contact IPv6
		1669	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1670	IPv6, but prefer IPv6 over IPv4.
		1671
		1672	=item C<PERL_ANYEVENT_EDNS0>
		1673
		1674	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1675	for DNS. This extension is generally useful to reduce DNS traffic, but
		1676	some (broken) firewalls drop such DNS packets, which is why it is off by
		1677	default.
		1678
		1679	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1680	EDNS0 in its DNS requests.
		1681
		1682	=item C<PERL_ANYEVENT_MAX_FORKS>
		1683
		1684	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		1685	will create in parallel.
		1686
		1687	=item C<PERL_ANYEVENT_MAX_OUTSTANDING_DNS>
		1688
		1689	The default value for the C<max_outstanding> parameter for the default DNS
		1690	resolver - this is the maximum number of parallel DNS requests that are
		1691	sent to the DNS server.
		1692
		1693	=item C<PERL_ANYEVENT_RESOLV_CONF>
		1694
		1695	The file to use instead of F</etc/resolv.conf> (or OS-specific
		1696	configuration) in the default resolver. When set to the empty string, no
		1697	default config will be used.
		1698
		1699	=item C<PERL_ANYEVENT_CA_FILE>, C<PERL_ANYEVENT_CA_PATH>.
		1700
		1701	When neither C<ca_file> nor C<ca_path> was specified during
		1702	L<AnyEvent::TLS> context creation, and either of these environment
		1703	variables exist, they will be used to specify CA certificate locations
		1704	instead of a system-dependent default.
		1705
		1706	=item C<PERL_ANYEVENT_AVOID_GUARD> and C<PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT>
		1707
		1708	When these are set to C<1>, then the respective modules are not
		1709	loaded. Mostly good for testing AnyEvent itself.
		1710
		1711	=back
		1712
412	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1713	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
		1714
		1715	This is an advanced topic that you do not normally need to use AnyEvent in
		1716	a module. This section is only of use to event loop authors who want to
		1717	provide AnyEvent compatibility.
413		1718
414	If you need to support another event library which isn't directly	1719	If you need to support another event library which isn't directly
415	supported by AnyEvent, you can supply your own interface to it by	1720	supported by AnyEvent, you can supply your own interface to it by
416	pushing, before the first watcher gets created, the package name of	1721	pushing, before the first watcher gets created, the package name of
417	the event module and the package name of the interface to use onto	1722	the event module and the package name of the interface to use onto
418	C<@AnyEvent::REGISTRY>. You can do that before and even without loading	1723	C<@AnyEvent::REGISTRY>. You can do that before and even without loading
419	AnyEvent.	1724	AnyEvent, so it is reasonably cheap.
420		1725
421	Example:	1726	Example:
422		1727
423	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];	1728	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];
424		1729
425	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>	1730	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>
426	package/class when it finds the C<urxvt> package/module is loaded. When	1731	package/class when it finds the C<urxvt> package/module is already loaded.
		1732
427	AnyEvent is loaded and asked to find a suitable event model, it will	1733	When AnyEvent is loaded and asked to find a suitable event model, it
428	first check for the presence of urxvt.	1734	will first check for the presence of urxvt by trying to C<use> the
		1735	C<urxvt::anyevent> module.
429		1736
430	The class should provide implementations for all watcher types (see	1737	The class should provide implementations for all watcher types. See
431	L<AnyEvent::Impl::Event> (source code), L<AnyEvent::Impl::Glib>	1738	L<AnyEvent::Impl::EV> (source code), L<AnyEvent::Impl::Glib> (Source code)
432	(Source code) and so on for actual examples, use C<perldoc -m	1739	and so on for actual examples. Use C<perldoc -m AnyEvent::Impl::Glib> to
433	AnyEvent::Impl::Glib> to see the sources).	1740	see the sources.
434		1741
		1742	If you don't provide C<signal> and C<child> watchers than AnyEvent will
		1743	provide suitable (hopefully) replacements.
		1744
435	The above isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)	1745	The above example isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)
436	uses the above line as-is. An interface isn't included in AnyEvent	1746	terminal emulator uses the above line as-is. An interface isn't included
437	because it doesn't make sense outside the embedded interpreter inside	1747	in AnyEvent because it doesn't make sense outside the embedded interpreter
438	I<rxvt-unicode>, and it is updated and maintained as part of the	1748	inside I<rxvt-unicode>, and it is updated and maintained as part of the
439	I<rxvt-unicode> distribution.	1749	I<rxvt-unicode> distribution.
440		1750
441	I<rxvt-unicode> also cheats a bit by not providing blocking access to	1751	I<rxvt-unicode> also cheats a bit by not providing blocking access to
442	condition variables: code blocking while waiting for a condition will	1752	condition variables: code blocking while waiting for a condition will
443	C<die>. This still works with most modules/usages, and blocking calls must	1753	C<die>. This still works with most modules/usages, and blocking calls must
444	not be in an interactive appliation, so it makes sense.	1754	not be done in an interactive application, so it makes sense.
445		1755
446	=head1 ENVIRONMENT VARIABLES
447
448	The following environment variables are used by this module:
449
450	C<PERL_ANYEVENT_VERBOSE> when set to C<2> or higher, reports which event
451	model gets used.
452
453	=head1 EXAMPLE	1756	=head1 EXAMPLE PROGRAM
454		1757
455	The following program uses an io watcher to read data from stdin, a timer	1758	The following program uses an I/O watcher to read data from STDIN, a timer
456	to display a message once per second, and a condvar to exit the program	1759	to display a message once per second, and a condition variable to quit the
457	when the user enters quit:	1760	program when the user enters quit:
458		1761
459	use AnyEvent;	1762	use AnyEvent;
460		1763
461	my $cv = AnyEvent->condvar;	1764	my $cv = AnyEvent->condvar;
462		1765
463	my $io_watcher = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	1766	my $io_watcher = AnyEvent->io (
		1767	fh => \*STDIN,
		1768	poll => 'r',
		1769	cb => sub {
464	warn "io event <$_[0]>\n"; # will always output <r>	1770	warn "io event <$_[0]>\n"; # will always output <r>
465	chomp (my $input = <STDIN>); # read a line	1771	chomp (my $input = <STDIN>); # read a line
466	warn "read: $input\n"; # output what has been read	1772	warn "read: $input\n"; # output what has been read
467	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1773	$cv->send if $input =~ /^q/i; # quit program if /^q/i
		1774	},
468	});	1775	);
469		1776
470	my $time_watcher; # can only be used once	1777	my $time_watcher; # can only be used once
471		1778
472	sub new_timer {	1779	sub new_timer {
473	$timer = AnyEvent->timer (after => 1, cb => sub {	1780	$timer = AnyEvent->timer (after => 1, cb => sub {
…		…
476	});	1783	});
477	}	1784	}
478		1785
479	new_timer; # create first timer	1786	new_timer; # create first timer
480		1787
481	$cv->wait; # wait until user enters /^q/i	1788	$cv->recv; # wait until user enters /^q/i
482		1789
483	=head1 REAL-WORLD EXAMPLE	1790	=head1 REAL-WORLD EXAMPLE
484		1791
485	Consider the L<Net::FCP> module. It features (among others) the following	1792	Consider the L<Net::FCP> module. It features (among others) the following
486	API calls, which are to freenet what HTTP GET requests are to http:	1793	API calls, which are to freenet what HTTP GET requests are to http:
…		…
536	syswrite $txn->{fh}, $txn->{request}	1843	syswrite $txn->{fh}, $txn->{request}
537	or die "connection or write error";	1844	or die "connection or write error";
538	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	1845	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
539		1846
540	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	1847	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
541	result and signals any possible waiters that the request ahs finished:	1848	result and signals any possible waiters that the request has finished:
542		1849
543	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1850	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
544		1851
545	if (end-of-file or data complete) {	1852	if (end-of-file or data complete) {
546	$txn->{result} = $txn->{buf};	1853	$txn->{result} = $txn->{buf};
547	$txn->{finished}->broadcast;	1854	$txn->{finished}->send;
548	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1855	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
549	}	1856	}
550		1857
551	The C<result> method, finally, just waits for the finished signal (if the	1858	The C<result> method, finally, just waits for the finished signal (if the
552	request was already finished, it doesn't wait, of course, and returns the	1859	request was already finished, it doesn't wait, of course, and returns the
553	data:	1860	data:
554		1861
555	$txn->{finished}->wait;	1862	$txn->{finished}->recv;
556	return $txn->{result};	1863	return $txn->{result};
557		1864
558	The actual code goes further and collects all errors (C<die>s, exceptions)	1865	The actual code goes further and collects all errors (C<die>s, exceptions)
559	that occured during request processing. The C<result> method detects	1866	that occurred during request processing. The C<result> method detects
560	wether an exception as thrown (it is stored inside the $txn object)	1867	whether an exception as thrown (it is stored inside the $txn object)
561	and just throws the exception, which means connection errors and other	1868	and just throws the exception, which means connection errors and other
562	problems get reported tot he code that tries to use the result, not in a	1869	problems get reported tot he code that tries to use the result, not in a
563	random callback.	1870	random callback.
564		1871
565	All of this enables the following usage styles:	1872	All of this enables the following usage styles:
566		1873
567	1. Blocking:	1874	1. Blocking:
568		1875
569	my $data = $fcp->client_get ($url);	1876	my $data = $fcp->client_get ($url);
570		1877
571	2. Blocking, but parallelizing:	1878	2. Blocking, but running in parallel:
572		1879
573	my @datas = map $_->result,	1880	my @datas = map $_->result,
574	map $fcp->txn_client_get ($_),	1881	map $fcp->txn_client_get ($_),
575	@urls;	1882	@urls;
576		1883
577	Both blocking examples work without the module user having to know	1884	Both blocking examples work without the module user having to know
578	anything about events.	1885	anything about events.
579		1886
580	3a. Event-based in a main program, using any support Event module:	1887	3a. Event-based in a main program, using any supported event module:
581		1888
582	use Event;	1889	use EV;
583		1890
584	$fcp->txn_client_get ($url)->cb (sub {	1891	$fcp->txn_client_get ($url)->cb (sub {
585	my $txn = shift;	1892	my $txn = shift;
586	my $data = $txn->result;	1893	my $data = $txn->result;
587	...	1894	...
588	});	1895	});
589		1896
590	Event::loop;	1897	EV::loop;
591		1898
592	3b. The module user could use AnyEvent, too:	1899	3b. The module user could use AnyEvent, too:
593		1900
594	use AnyEvent;	1901	use AnyEvent;
595		1902
596	my $quit = AnyEvent->condvar;	1903	my $quit = AnyEvent->condvar;
597		1904
598	$fcp->txn_client_get ($url)->cb (sub {	1905	$fcp->txn_client_get ($url)->cb (sub {
599	...	1906	...
600	$quit->broadcast;	1907	$quit->send;
601	});	1908	});
602		1909
603	$quit->wait;	1910	$quit->recv;
		1911
		1912
		1913	=head1 BENCHMARKS
		1914
		1915	To give you an idea of the performance and overheads that AnyEvent adds
		1916	over the event loops themselves and to give you an impression of the speed
		1917	of various event loops I prepared some benchmarks.
		1918
		1919	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1920
		1921	Here is a benchmark of various supported event models used natively and
		1922	through AnyEvent. The benchmark creates a lot of timers (with a zero
		1923	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
		1924	which it is), lets them fire exactly once and destroys them again.
		1925
		1926	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
		1927	distribution.
		1928
		1929	=head3 Explanation of the columns
		1930
		1931	I<watcher> is the number of event watchers created/destroyed. Since
		1932	different event models feature vastly different performances, each event
		1933	loop was given a number of watchers so that overall runtime is acceptable
		1934	and similar between tested event loop (and keep them from crashing): Glib
		1935	would probably take thousands of years if asked to process the same number
		1936	of watchers as EV in this benchmark.
		1937
		1938	I<bytes> is the number of bytes (as measured by the resident set size,
		1939	RSS) consumed by each watcher. This method of measuring captures both C
		1940	and Perl-based overheads.
		1941
		1942	I<create> is the time, in microseconds (millionths of seconds), that it
		1943	takes to create a single watcher. The callback is a closure shared between
		1944	all watchers, to avoid adding memory overhead. That means closure creation
		1945	and memory usage is not included in the figures.
		1946
		1947	I<invoke> is the time, in microseconds, used to invoke a simple
		1948	callback. The callback simply counts down a Perl variable and after it was
		1949	invoked "watcher" times, it would C<< ->send >> a condvar once to
		1950	signal the end of this phase.
		1951
		1952	I<destroy> is the time, in microseconds, that it takes to destroy a single
		1953	watcher.
		1954
		1955	=head3 Results
		1956
		1957	name watchers bytes create invoke destroy comment
		1958	EV/EV 400000 224 0.47 0.35 0.27 EV native interface
		1959	EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers
		1960	CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal
		1961	Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation
		1962	Event/Event 16000 517 32.20 31.80 0.81 Event native interface
		1963	Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers
		1964	IOAsync/Any 16000 989 38.10 32.77 11.13 via IO::Async::Loop::IO_Poll
		1965	IOAsync/Any 16000 990 37.59 29.50 10.61 via IO::Async::Loop::Epoll
		1966	Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour
		1967	Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers
		1968	POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event
		1969	POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select
		1970
		1971	=head3 Discussion
		1972
		1973	The benchmark does I<not> measure scalability of the event loop very
		1974	well. For example, a select-based event loop (such as the pure perl one)
		1975	can never compete with an event loop that uses epoll when the number of
		1976	file descriptors grows high. In this benchmark, all events become ready at
		1977	the same time, so select/poll-based implementations get an unnatural speed
		1978	boost.
		1979
		1980	Also, note that the number of watchers usually has a nonlinear effect on
		1981	overall speed, that is, creating twice as many watchers doesn't take twice
		1982	the time - usually it takes longer. This puts event loops tested with a
		1983	higher number of watchers at a disadvantage.
		1984
		1985	To put the range of results into perspective, consider that on the
		1986	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1987	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1988	cycles with POE.
		1989
		1990	C<EV> is the sole leader regarding speed and memory use, which are both
		1991	maximal/minimal, respectively. Even when going through AnyEvent, it uses
		1992	far less memory than any other event loop and is still faster than Event
		1993	natively.
		1994
		1995	The pure perl implementation is hit in a few sweet spots (both the
		1996	constant timeout and the use of a single fd hit optimisations in the perl
		1997	interpreter and the backend itself). Nevertheless this shows that it
		1998	adds very little overhead in itself. Like any select-based backend its
		1999	performance becomes really bad with lots of file descriptors (and few of
		2000	them active), of course, but this was not subject of this benchmark.
		2001
		2002	The C<Event> module has a relatively high setup and callback invocation
		2003	cost, but overall scores in on the third place.
		2004
		2005	C<IO::Async> performs admirably well, about on par with C<Event>, even
		2006	when using its pure perl backend.
		2007
		2008	C<Glib>'s memory usage is quite a bit higher, but it features a
		2009	faster callback invocation and overall ends up in the same class as
		2010	C<Event>. However, Glib scales extremely badly, doubling the number of
		2011	watchers increases the processing time by more than a factor of four,
		2012	making it completely unusable when using larger numbers of watchers
		2013	(note that only a single file descriptor was used in the benchmark, so
		2014	inefficiencies of C<poll> do not account for this).
		2015
		2016	The C<Tk> adaptor works relatively well. The fact that it crashes with
		2017	more than 2000 watchers is a big setback, however, as correctness takes
		2018	precedence over speed. Nevertheless, its performance is surprising, as the
		2019	file descriptor is dup()ed for each watcher. This shows that the dup()
		2020	employed by some adaptors is not a big performance issue (it does incur a
		2021	hidden memory cost inside the kernel which is not reflected in the figures
		2022	above).
		2023
		2024	C<POE>, regardless of underlying event loop (whether using its pure perl
		2025	select-based backend or the Event module, the POE-EV backend couldn't
		2026	be tested because it wasn't working) shows abysmal performance and
		2027	memory usage with AnyEvent: Watchers use almost 30 times as much memory
		2028	as EV watchers, and 10 times as much memory as Event (the high memory
		2029	requirements are caused by requiring a session for each watcher). Watcher
		2030	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		2031	implementation.
		2032
		2033	The design of the POE adaptor class in AnyEvent can not really account
		2034	for the performance issues, though, as session creation overhead is
		2035	small compared to execution of the state machine, which is coded pretty
		2036	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		2037	using multiple sessions is not a good approach, especially regarding
		2038	memory usage, even the author of POE could not come up with a faster
		2039	design).
		2040
		2041	=head3 Summary
		2042
		2043	=over 4
		2044
		2045	=item * Using EV through AnyEvent is faster than any other event loop
		2046	(even when used without AnyEvent), but most event loops have acceptable
		2047	performance with or without AnyEvent.
		2048
		2049	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
		2050	the actual event loop, only with extremely fast event loops such as EV
		2051	adds AnyEvent significant overhead.
		2052
		2053	=item * You should avoid POE like the plague if you want performance or
		2054	reasonable memory usage.
		2055
		2056	=back
		2057
		2058	=head2 BENCHMARKING THE LARGE SERVER CASE
		2059
		2060	This benchmark actually benchmarks the event loop itself. It works by
		2061	creating a number of "servers": each server consists of a socket pair, a
		2062	timeout watcher that gets reset on activity (but never fires), and an I/O
		2063	watcher waiting for input on one side of the socket. Each time the socket
		2064	watcher reads a byte it will write that byte to a random other "server".
		2065
		2066	The effect is that there will be a lot of I/O watchers, only part of which
		2067	are active at any one point (so there is a constant number of active
		2068	fds for each loop iteration, but which fds these are is random). The
		2069	timeout is reset each time something is read because that reflects how
		2070	most timeouts work (and puts extra pressure on the event loops).
		2071
		2072	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		2073	(1%) are active. This mirrors the activity of large servers with many
		2074	connections, most of which are idle at any one point in time.
		2075
		2076	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		2077	distribution.
		2078
		2079	=head3 Explanation of the columns
		2080
		2081	I<sockets> is the number of sockets, and twice the number of "servers" (as
		2082	each server has a read and write socket end).
		2083
		2084	I<create> is the time it takes to create a socket pair (which is
		2085	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		2086
		2087	I<request>, the most important value, is the time it takes to handle a
		2088	single "request", that is, reading the token from the pipe and forwarding
		2089	it to another server. This includes deleting the old timeout and creating
		2090	a new one that moves the timeout into the future.
		2091
		2092	=head3 Results
		2093
		2094	name sockets create request
		2095	EV 20000 69.01 11.16
		2096	Perl 20000 73.32 35.87
		2097	IOAsync 20000 157.00 98.14 epoll
		2098	IOAsync 20000 159.31 616.06 poll
		2099	Event 20000 212.62 257.32
		2100	Glib 20000 651.16 1896.30
		2101	POE 20000 349.67 12317.24 uses POE::Loop::Event
		2102
		2103	=head3 Discussion
		2104
		2105	This benchmark I<does> measure scalability and overall performance of the
		2106	particular event loop.
		2107
		2108	EV is again fastest. Since it is using epoll on my system, the setup time
		2109	is relatively high, though.
		2110
		2111	Perl surprisingly comes second. It is much faster than the C-based event
		2112	loops Event and Glib.
		2113
		2114	IO::Async performs very well when using its epoll backend, and still quite
		2115	good compared to Glib when using its pure perl backend.
		2116
		2117	Event suffers from high setup time as well (look at its code and you will
		2118	understand why). Callback invocation also has a high overhead compared to
		2119	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		2120	uses select or poll in basically all documented configurations.
		2121
		2122	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		2123	clearly fails to perform with many filehandles or in busy servers.
		2124
		2125	POE is still completely out of the picture, taking over 1000 times as long
		2126	as EV, and over 100 times as long as the Perl implementation, even though
		2127	it uses a C-based event loop in this case.
		2128
		2129	=head3 Summary
		2130
		2131	=over 4
		2132
		2133	=item * The pure perl implementation performs extremely well.
		2134
		2135	=item * Avoid Glib or POE in large projects where performance matters.
		2136
		2137	=back
		2138
		2139	=head2 BENCHMARKING SMALL SERVERS
		2140
		2141	While event loops should scale (and select-based ones do not...) even to
		2142	large servers, most programs we (or I :) actually write have only a few
		2143	I/O watchers.
		2144
		2145	In this benchmark, I use the same benchmark program as in the large server
		2146	case, but it uses only eight "servers", of which three are active at any
		2147	one time. This should reflect performance for a small server relatively
		2148	well.
		2149
		2150	The columns are identical to the previous table.
		2151
		2152	=head3 Results
		2153
		2154	name sockets create request
		2155	EV 16 20.00 6.54
		2156	Perl 16 25.75 12.62
		2157	Event 16 81.27 35.86
		2158	Glib 16 32.63 15.48
		2159	POE 16 261.87 276.28 uses POE::Loop::Event
		2160
		2161	=head3 Discussion
		2162
		2163	The benchmark tries to test the performance of a typical small
		2164	server. While knowing how various event loops perform is interesting, keep
		2165	in mind that their overhead in this case is usually not as important, due
		2166	to the small absolute number of watchers (that is, you need efficiency and
		2167	speed most when you have lots of watchers, not when you only have a few of
		2168	them).
		2169
		2170	EV is again fastest.
		2171
		2172	Perl again comes second. It is noticeably faster than the C-based event
		2173	loops Event and Glib, although the difference is too small to really
		2174	matter.
		2175
		2176	POE also performs much better in this case, but is is still far behind the
		2177	others.
		2178
		2179	=head3 Summary
		2180
		2181	=over 4
		2182
		2183	=item * C-based event loops perform very well with small number of
		2184	watchers, as the management overhead dominates.
		2185
		2186	=back
		2187
		2188	=head2 THE IO::Lambda BENCHMARK
		2189
		2190	Recently I was told about the benchmark in the IO::Lambda manpage, which
		2191	could be misinterpreted to make AnyEvent look bad. In fact, the benchmark
		2192	simply compares IO::Lambda with POE, and IO::Lambda looks better (which
		2193	shouldn't come as a surprise to anybody). As such, the benchmark is
		2194	fine, and mostly shows that the AnyEvent backend from IO::Lambda isn't
		2195	very optimal. But how would AnyEvent compare when used without the extra
		2196	baggage? To explore this, I wrote the equivalent benchmark for AnyEvent.
		2197
		2198	The benchmark itself creates an echo-server, and then, for 500 times,
		2199	connects to the echo server, sends a line, waits for the reply, and then
		2200	creates the next connection. This is a rather bad benchmark, as it doesn't
		2201	test the efficiency of the framework or much non-blocking I/O, but it is a
		2202	benchmark nevertheless.
		2203
		2204	name runtime
		2205	Lambda/select 0.330 sec
		2206	+ optimized 0.122 sec
		2207	Lambda/AnyEvent 0.327 sec
		2208	+ optimized 0.138 sec
		2209	Raw sockets/select 0.077 sec
		2210	POE/select, components 0.662 sec
		2211	POE/select, raw sockets 0.226 sec
		2212	POE/select, optimized 0.404 sec
		2213
		2214	AnyEvent/select/nb 0.085 sec
		2215	AnyEvent/EV/nb 0.068 sec
		2216	+state machine 0.134 sec
		2217
		2218	The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
		2219	benchmarks actually make blocking connects and use 100% blocking I/O,
		2220	defeating the purpose of an event-based solution. All of the newly
		2221	written AnyEvent benchmarks use 100% non-blocking connects (using
		2222	AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
		2223	resolver), so AnyEvent is at a disadvantage here, as non-blocking connects
		2224	generally require a lot more bookkeeping and event handling than blocking
		2225	connects (which involve a single syscall only).
		2226
		2227	The last AnyEvent benchmark additionally uses L<AnyEvent::Handle>, which
		2228	offers similar expressive power as POE and IO::Lambda, using conventional
		2229	Perl syntax. This means that both the echo server and the client are 100%
		2230	non-blocking, further placing it at a disadvantage.
		2231
		2232	As you can see, the AnyEvent + EV combination even beats the
		2233	hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
		2234	backend easily beats IO::Lambda and POE.
		2235
		2236	And even the 100% non-blocking version written using the high-level (and
		2237	slow :) L<AnyEvent::Handle> abstraction beats both POE and IO::Lambda by a
		2238	large margin, even though it does all of DNS, tcp-connect and socket I/O
		2239	in a non-blocking way.
		2240
		2241	The two AnyEvent benchmarks programs can be found as F<eg/ae0.pl> and
		2242	F<eg/ae2.pl> in the AnyEvent distribution, the remaining benchmarks are
		2243	part of the IO::lambda distribution and were used without any changes.
		2244
		2245
		2246	=head1 SIGNALS
		2247
		2248	AnyEvent currently installs handlers for these signals:
		2249
		2250	=over 4
		2251
		2252	=item SIGCHLD
		2253
		2254	A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
		2255	emulation for event loops that do not support them natively. Also, some
		2256	event loops install a similar handler.
		2257
		2258	Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, then
		2259	AnyEvent will reset it to default, to avoid losing child exit statuses.
		2260
		2261	=item SIGPIPE
		2262
		2263	A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
		2264	when AnyEvent gets loaded.
		2265
		2266	The rationale for this is that AnyEvent users usually do not really depend
		2267	on SIGPIPE delivery (which is purely an optimisation for shell use, or
		2268	badly-written programs), but C<SIGPIPE> can cause spurious and rare
		2269	program exits as a lot of people do not expect C<SIGPIPE> when writing to
		2270	some random socket.
		2271
		2272	The rationale for installing a no-op handler as opposed to ignoring it is
		2273	that this way, the handler will be restored to defaults on exec.
		2274
		2275	Feel free to install your own handler, or reset it to defaults.
		2276
		2277	=back
		2278
		2279	=cut
		2280
		2281	undef $SIG{CHLD}
		2282	if $SIG{CHLD} eq 'IGNORE';
		2283
		2284	$SIG{PIPE} = sub { }
		2285	unless defined $SIG{PIPE};
		2286
		2287	=head1 RECOMMENDED/OPTIONAL MODULES
		2288
		2289	One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
		2290	it's built-in modules) are required to use it.
		2291
		2292	That does not mean that AnyEvent won't take advantage of some additional
		2293	modules if they are installed.
		2294
		2295	This section epxlains which additional modules will be used, and how they
		2296	affect AnyEvent's operetion.
		2297
		2298	=over 4
		2299
		2300	=item L<Async::Interrupt>
		2301
		2302	This slightly arcane module is used to implement fast signal handling: To
		2303	my knowledge, there is no way to do completely race-free and quick
		2304	signal handling in pure perl. To ensure that signals still get
		2305	delivered, AnyEvent will start an interval timer to wake up perl (and
		2306	catch the signals) with some delay (default is 10 seconds, look for
		2307	C<$AnyEvent::MAX_SIGNAL_LATENCY>).
		2308
		2309	If this module is available, then it will be used to implement signal
		2310	catching, which means that signals will not be delayed, and the event loop
		2311	will not be interrupted regularly, which is more efficient (And good for
		2312	battery life on laptops).
		2313
		2314	This affects not just the pure-perl event loop, but also other event loops
		2315	that have no signal handling on their own (e.g. Glib, Tk, Qt).
		2316
		2317	Some event loops (POE, Event, Event::Lib) offer signal watchers natively,
		2318	and either employ their own workarounds (POE) or use AnyEvent's workaround
		2319	(using C<$AnyEvent::MAX_SIGNAL_LATENCY>). Installing L<Async::Interrupt>
		2320	does nothing for those backends.
		2321
		2322	=item L<EV>
		2323
		2324	This module isn't really "optional", as it is simply one of the backend
		2325	event loops that AnyEvent can use. However, it is simply the best event
		2326	loop available in terms of features, speed and stability: It supports
		2327	the AnyEvent API optimally, implements all the watcher types in XS, does
		2328	automatic timer adjustments even when no monotonic clock is available,
		2329	can take avdantage of advanced kernel interfaces such as C<epoll> and
		2330	C<kqueue>, and is the fastest backend I<by far>. You can even embed
		2331	L<Glib>/L<Gtk2> in it (or vice versa, see L<EV::Glib> and L<Glib::EV>).
		2332
		2333	=item L<Guard>
		2334
		2335	The guard module, when used, will be used to implement
		2336	C<AnyEvent::Util::guard>. This speeds up guards considerably (and uses a
		2337	lot less memory), but otherwise doesn't affect guard operation much. It is
		2338	purely used for performance.
		2339
		2340	=item L<JSON> and L<JSON::XS>
		2341
		2342	This module is required when you want to read or write JSON data via
		2343	L<AnyEvent::Handle>. It is also written in pure-perl, but can take
		2344	advantage of the ultra-high-speed L<JSON::XS> module when it is installed.
		2345
		2346	In fact, L<AnyEvent::Handle> will use L<JSON::XS> by default if it is
		2347	installed.
		2348
		2349	=item L<Net::SSLeay>
		2350
		2351	Implementing TLS/SSL in Perl is certainly interesting, but not very
		2352	worthwhile: If this module is installed, then L<AnyEvent::Handle> (with
		2353	the help of L<AnyEvent::TLS>), gains the ability to do TLS/SSL.
		2354
		2355	=item L<Time::HiRes>
		2356
		2357	This module is part of perl since release 5.008. It will be used when the
		2358	chosen event library does not come with a timing source on it's own. The
		2359	pure-perl event loop (L<AnyEvent::Impl::Perl>) will additionally use it to
		2360	try to use a monotonic clock for timing stability.
		2361
		2362	=back
		2363
		2364
		2365	=head1 FORK
		2366
		2367	Most event libraries are not fork-safe. The ones who are usually are
		2368	because they rely on inefficient but fork-safe C<select> or C<poll>
		2369	calls. Only L<EV> is fully fork-aware.
		2370
		2371	If you have to fork, you must either do so I<before> creating your first
		2372	watcher OR you must not use AnyEvent at all in the child OR you must do
		2373	something completely out of the scope of AnyEvent.
		2374
		2375
		2376	=head1 SECURITY CONSIDERATIONS
		2377
		2378	AnyEvent can be forced to load any event model via
		2379	$ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used to
		2380	execute arbitrary code or directly gain access, it can easily be used to
		2381	make the program hang or malfunction in subtle ways, as AnyEvent watchers
		2382	will not be active when the program uses a different event model than
		2383	specified in the variable.
		2384
		2385	You can make AnyEvent completely ignore this variable by deleting it
		2386	before the first watcher gets created, e.g. with a C<BEGIN> block:
		2387
		2388	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
		2389
		2390	use AnyEvent;
		2391
		2392	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		2393	be used to probe what backend is used and gain other information (which is
		2394	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		2395	$ENV{PERL_ANYEVENT_STRICT}.
		2396
		2397	Note that AnyEvent will remove I<all> environment variables starting with
		2398	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		2399	enabled.
		2400
		2401
		2402	=head1 BUGS
		2403
		2404	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		2405	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		2406	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		2407	memleaks, such as leaking on C<map> and C<grep> but it is usually not as
		2408	pronounced).
		2409
604		2410
605	=head1 SEE ALSO	2411	=head1 SEE ALSO
606		2412
607	Event modules: L<Coro::Event>, L<Coro>, L<Event>, L<Glib::Event>, L<Glib>.	2413	Utility functions: L<AnyEvent::Util>.
608		2414
609	Implementations: L<AnyEvent::Impl::Coro>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>.	2415	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
		2416	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
610		2417
611	Nontrivial usage example: L<Net::FCP>.	2418	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
		2419	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
		2420	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
		2421	L<AnyEvent::Impl::POE>, L<AnyEvent::Impl::IOAsync>.
612		2422
613	=head1	2423	Non-blocking file handles, sockets, TCP clients and
		2424	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>, L<AnyEvent::TLS>.
		2425
		2426	Asynchronous DNS: L<AnyEvent::DNS>.
		2427
		2428	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>,
		2429	L<Coro::Event>,
		2430
		2431	Nontrivial usage examples: L<AnyEvent::GPSD>, L<AnyEvent::XMPP>,
		2432	L<AnyEvent::HTTP>.
		2433
		2434
		2435	=head1 AUTHOR
		2436
		2437	Marc Lehmann <schmorp@schmorp.de>
		2438	http://home.schmorp.de/
614		2439
615	=cut	2440	=cut
616		2441
617	1	2442	1
618		2443

Diff Legend

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