ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.29 by root, Wed Oct 31 14:17:43 2007 UTC vs.
Revision 1.173 by root, Mon Jul 21 03:47:22 2008 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent - provide framework for multiple event loops	3	AnyEvent - provide framework for multiple event loops
4		4
5	Event, Coro, Glib, Tk, Perl - various supported event loops	5	EV, Event, Glib, Tk, Perl, Event::Lib, Qt, POE - various supported event loops
6		6
7	=head1 SYNOPSIS	7	=head1 SYNOPSIS
8		8
9	use AnyEvent;	9	use AnyEvent;
10		10
11	my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub {	11	my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { ... });
		12
		13	my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
		14	my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...
		15
		16	print AnyEvent->now; # prints current event loop time
		17	print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
		18
		19	my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
		20
		21	my $w = AnyEvent->child (pid => $pid, cb => sub {
		22	my ($pid, $status) = @_;
12	...	23	...
13	});	24	});
14		25
15	my $w = AnyEvent->timer (after => $seconds, cb => sub {
16	...
17	});
18
19	my $w = AnyEvent->condvar; # stores wether a condition was flagged	26	my $w = AnyEvent->condvar; # stores whether a condition was flagged
		27	$w->send; # wake up current and all future recv's
20	$w->wait; # enters "main loop" till $condvar gets ->broadcast	28	$w->recv; # enters "main loop" till $condvar gets ->send
21	$w->broadcast; # wake up current and all future wait's	29	# use a condvar in callback mode:
		30	$w->cb (sub { $_[0]->recv });
		31
		32	=head1 INTRODUCTION/TUTORIAL
		33
		34	This manpage is mainly a reference manual. If you are interested
		35	in a tutorial or some gentle introduction, have a look at the
		36	L<AnyEvent::Intro> manpage.
		37
		38	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
		39
		40	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
		41	nowadays. So what is different about AnyEvent?
		42
		43	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
		44	policy> and AnyEvent is I<small and efficient>.
		45
		46	First and foremost, I<AnyEvent is not an event model> itself, it only
		47	interfaces to whatever event model the main program happens to use, in a
		48	pragmatic way. For event models and certain classes of immortals alike,
		49	the statement "there can only be one" is a bitter reality: In general,
		50	only one event loop can be active at the same time in a process. AnyEvent
		51	cannot change this, but it can hide the differences between those event
		52	loops.
		53
		54	The goal of AnyEvent is to offer module authors the ability to do event
		55	programming (waiting for I/O or timer events) without subscribing to a
		56	religion, a way of living, and most importantly: without forcing your
		57	module users into the same thing by forcing them to use the same event
		58	model you use.
		59
		60	For modules like POE or IO::Async (which is a total misnomer as it is
		61	actually doing all I/O I<synchronously>...), using them in your module is
		62	like joining a cult: After you joined, you are dependent on them and you
		63	cannot use anything else, as they are simply incompatible to everything
		64	that isn't them. What's worse, all the potential users of your
		65	module are I<also> forced to use the same event loop you use.
		66
		67	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
		68	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
		69	with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
		70	your module uses one of those, every user of your module has to use it,
		71	too. But if your module uses AnyEvent, it works transparently with all
		72	event models it supports (including stuff like IO::Async, as long as those
		73	use one of the supported event loops. It is trivial to add new event loops
		74	to AnyEvent, too, so it is future-proof).
		75
		76	In addition to being free of having to use I<the one and only true event
		77	model>, AnyEvent also is free of bloat and policy: with POE or similar
		78	modules, you get an enormous amount of code and strict rules you have to
		79	follow. AnyEvent, on the other hand, is lean and up to the point, by only
		80	offering the functionality that is necessary, in as thin as a wrapper as
		81	technically possible.
		82
		83	Of course, AnyEvent comes with a big (and fully optional!) toolbox
		84	of useful functionality, such as an asynchronous DNS resolver, 100%
		85	non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
		86	such as Windows) and lots of real-world knowledge and workarounds for
		87	platform bugs and differences.
		88
		89	Now, if you I<do want> lots of policy (this can arguably be somewhat
		90	useful) and you want to force your users to use the one and only event
		91	model, you should I<not> use this module.
22		92
23	=head1 DESCRIPTION	93	=head1 DESCRIPTION
24		94
25	L<AnyEvent> provides an identical interface to multiple event loops. This	95	L<AnyEvent> provides an identical interface to multiple event loops. This
26	allows module authors to utilise an event loop without forcing module	96	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	97	users to use the same event loop (as only a single event loop can coexist
28	peacefully at any one time).	98	peacefully at any one time).
29		99
30	The interface itself is vaguely similar but not identical to the Event	100	The interface itself is vaguely similar, but not identical to the L<Event>
31	module.	101	module.
32		102
33	On the first call of any method, the module tries to detect the currently	103	During the first call of any watcher-creation method, the module tries
34	loaded event loop by probing wether any of the following modules is	104	to detect the currently loaded event loop by probing whether one of the
35	loaded: L<Coro::Event>, L<Event>, L<Glib>, L<Tk>. The first one found is	105	following modules is already loaded: L<EV>,
36	used. If none is found, the module tries to load these modules in the	106	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,
37	order given. The first one that could be successfully loaded will be	107	L<POE>. The first one found is used. If none are found, the module tries
38	used. If still none could be found, AnyEvent will fall back to a pure-perl	108	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
39	event loop, which is also not very efficient.	109	adaptor should always succeed) in the order given. The first one that can
		110	be successfully loaded will be used. If, after this, still none could be
		111	found, AnyEvent will fall back to a pure-perl event loop, which is not
		112	very efficient, but should work everywhere.
40		113
41	Because AnyEvent first checks for modules that are already loaded, loading	114	Because AnyEvent first checks for modules that are already loaded, loading
42	an Event model explicitly before first using AnyEvent will likely make	115	an event model explicitly before first using AnyEvent will likely make
43	that model the default. For example:	116	that model the default. For example:
44		117
45	use Tk;	118	use Tk;
46	use AnyEvent;	119	use AnyEvent;
47		120
48	# .. AnyEvent will likely default to Tk	121	# .. AnyEvent will likely default to Tk
49		122
		123	The I<likely> means that, if any module loads another event model and
		124	starts using it, all bets are off. Maybe you should tell their authors to
		125	use AnyEvent so their modules work together with others seamlessly...
		126
50	The pure-perl implementation of AnyEvent is called	127	The pure-perl implementation of AnyEvent is called
51	C<AnyEvent::Impl::Perl>. Like other event modules you can load it	128	C<AnyEvent::Impl::Perl>. Like other event modules you can load it
52	explicitly.	129	explicitly and enjoy the high availability of that event loop :)
53		130
54	=head1 WATCHERS	131	=head1 WATCHERS
55		132
56	AnyEvent has the central concept of a I<watcher>, which is an object that	133	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	134	stores relevant data for each kind of event you are waiting for, such as
58	the callback to call, the filehandle to watch, etc.	135	the callback to call, the file handle to watch, etc.
59		136
60	These watchers are normal Perl objects with normal Perl lifetime. After	137	These watchers are normal Perl objects with normal Perl lifetime. After
61	creating a watcher it will immediately "watch" for events and invoke	138	creating a watcher it will immediately "watch" for events and invoke the
		139	callback when the event occurs (of course, only when the event model
		140	is in control).
		141
62	the callback. To disable the watcher you have to destroy it (e.g. by	142	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	143	variable you store it in to C<undef> or otherwise deleting all references
64	references to it).	144	to it).
65		145
66	All watchers are created by calling a method on the C<AnyEvent> class.	146	All watchers are created by calling a method on the C<AnyEvent> class.
67		147
		148	Many watchers either are used with "recursion" (repeating timers for
		149	example), or need to refer to their watcher object in other ways.
		150
		151	An any way to achieve that is this pattern:
		152
		153	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
		154	# you can use $w here, for example to undef it
		155	undef $w;
		156	});
		157
		158	Note that C<my $w; $w => combination. This is necessary because in Perl,
		159	my variables are only visible after the statement in which they are
		160	declared.
		161
68	=head2 IO WATCHERS	162	=head2 I/O WATCHERS
69		163
70	You can create I/O watcher by calling the C<< AnyEvent->io >> method with	164	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
71	the following mandatory arguments:	165	with the following mandatory key-value pairs as arguments:
72		166
73	C<fh> the Perl I<filehandle> (not filedescriptor) to watch for	167	C<fh> the Perl I<file handle> (I<not> file descriptor) to watch for events
		168	(AnyEvent might or might not keep a reference to this file handle). C<poll>
74	events. C<poll> must be a string that is either C<r> or C<w>, that creates	169	must be a string that is either C<r> or C<w>, which creates a watcher
75	a watcher waiting for "r"eadable or "w"ritable events. C<cb> teh callback	170	waiting for "r"eadable or "w"ritable events, respectively. C<cb> is the
76	to invoke everytime the filehandle becomes ready.	171	callback to invoke each time the file handle becomes ready.
77		172
78	Only one io watcher per C<fh> and C<poll> combination is allowed (i.e. on	173	Although the callback might get passed parameters, their value and
79	a socket you can have one r + one w, not any more (limitation comes from	174	presence is undefined and you cannot rely on them. Portable AnyEvent
80	Tk - if you are sure you are not using Tk this limitation is gone).	175	callbacks cannot use arguments passed to I/O watcher callbacks.
81		176
82	Filehandles will be kept alive, so as long as the watcher exists, the	177	The I/O watcher might use the underlying file descriptor or a copy of it.
83	filehandle exists, too.	178	You must not close a file handle as long as any watcher is active on the
		179	underlying file descriptor.
84		180
85	Example:	181	Some event loops issue spurious readyness notifications, so you should
		182	always use non-blocking calls when reading/writing from/to your file
		183	handles.
86		184
87	# wait for readability of STDIN, then read a line and disable the watcher	185	Example: wait for readability of STDIN, then read a line and disable the
		186	watcher.
		187
88	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	188	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
89	chomp (my $input = <STDIN>);	189	chomp (my $input = <STDIN>);
90	warn "read: $input\n";	190	warn "read: $input\n";
91	undef $w;	191	undef $w;
92	});	192	});
…		…
94	=head2 TIME WATCHERS	194	=head2 TIME WATCHERS
95		195
96	You can create a time watcher by calling the C<< AnyEvent->timer >>	196	You can create a time watcher by calling the C<< AnyEvent->timer >>
97	method with the following mandatory arguments:	197	method with the following mandatory arguments:
98		198
99	C<after> after how many seconds (fractions are supported) should the timer	199	C<after> specifies after how many seconds (fractional values are
100	activate. C<cb> the callback to invoke.	200	supported) the callback should be invoked. C<cb> is the callback to invoke
		201	in that case.
101		202
102	The timer callback will be invoked at most once: if you want a repeating	203	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	204	presence is undefined and you cannot rely on them. Portable AnyEvent
104	and Glib).	205	callbacks cannot use arguments passed to time watcher callbacks.
105		206
106	Example:	207	The callback will normally be invoked once only. If you specify another
		208	parameter, C<interval>, as a strictly positive number (> 0), then the
		209	callback will be invoked regularly at that interval (in fractional
		210	seconds) after the first invocation. If C<interval> is specified with a
		211	false value, then it is treated as if it were missing.
107		212
		213	The callback will be rescheduled before invoking the callback, but no
		214	attempt is done to avoid timer drift in most backends, so the interval is
		215	only approximate.
		216
108	# fire an event after 7.7 seconds	217	Example: fire an event after 7.7 seconds.
		218
109	my $w = AnyEvent->timer (after => 7.7, cb => sub {	219	my $w = AnyEvent->timer (after => 7.7, cb => sub {
110	warn "timeout\n";	220	warn "timeout\n";
111	});	221	});
112		222
113	# to cancel the timer:	223	# to cancel the timer:
114	undef $w	224	undef $w;
115		225
116	=head2 CONDITION WATCHERS	226	Example 2: fire an event after 0.5 seconds, then roughly every second.
117		227
118	Condition watchers can be created by calling the C<< AnyEvent->condvar >>	228	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
119	method without any arguments.	229	warn "timeout\n";
		230	};
120		231
121	A condition watcher watches for a condition - precisely that the C<<	232	=head3 TIMING ISSUES
122	->broadcast >> method has been called.
123		233
124	The watcher has only two methods:	234	There are two ways to handle timers: based on real time (relative, "fire
		235	in 10 seconds") and based on wallclock time (absolute, "fire at 12
		236	o'clock").
		237
		238	While most event loops expect timers to specified in a relative way, they
		239	use absolute time internally. This makes a difference when your clock
		240	"jumps", for example, when ntp decides to set your clock backwards from
		241	the wrong date of 2014-01-01 to 2008-01-01, a watcher that is supposed to
		242	fire "after" a second might actually take six years to finally fire.
		243
		244	AnyEvent cannot compensate for this. The only event loop that is conscious
		245	about these issues is L<EV>, which offers both relative (ev_timer, based
		246	on true relative time) and absolute (ev_periodic, based on wallclock time)
		247	timers.
		248
		249	AnyEvent always prefers relative timers, if available, matching the
		250	AnyEvent API.
		251
		252	AnyEvent has two additional methods that return the "current time":
125		253
126	=over 4	254	=over 4
127		255
128	=item $cv->wait	256	=item AnyEvent->time
129		257
130	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	258	This returns the "current wallclock time" as a fractional number of
131	called on c<$cv>, while servicing other watchers normally.	259	seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
		260	return, and the result is guaranteed to be compatible with those).
132		261
133	Not all event models support a blocking wait - some die in that case, so	262	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	263	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		264
140	You can only wait once on a condition - additional calls will return	265	=item AnyEvent->now
141	immediately.
142		266
143	=item $cv->broadcast	267	This also returns the "current wallclock time", but unlike C<time>, above,
		268	this value might change only once per event loop iteration, depending on
		269	the event loop (most return the same time as C<time>, above). This is the
		270	time that AnyEvent's timers get scheduled against.
144		271
145	Flag the condition as ready - a running C<< ->wait >> and all further	272	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	273	function to call when you want to know the current time.>
147	is waiting the broadcast will be remembered..
148		274
149	Example:	275	This function is also often faster then C<< AnyEvent->time >>, and
		276	thus the preferred method if you want some timestamp (for example,
		277	L<AnyEvent::Handle> uses this to update it's activity timeouts).
		278
		279	The rest of this section is only of relevance if you try to be very exact
		280	with your timing, you can skip it without bad conscience.
		281
		282	For a practical example of when these times differ, consider L<Event::Lib>
		283	and L<EV> and the following set-up:
		284
		285	The event loop is running and has just invoked one of your callback at
		286	time=500 (assume no other callbacks delay processing). In your callback,
		287	you wait a second by executing C<sleep 1> (blocking the process for a
		288	second) and then (at time=501) you create a relative timer that fires
		289	after three seconds.
		290
		291	With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
		292	both return C<501>, because that is the current time, and the timer will
		293	be scheduled to fire at time=504 (C<501> + C<3>).
		294
		295	With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
		296	time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
		297	last event processing phase started. With L<EV>, your timer gets scheduled
		298	to run at time=503 (C<500> + C<3>).
		299
		300	In one sense, L<Event::Lib> is more exact, as it uses the current time
		301	regardless of any delays introduced by event processing. However, most
		302	callbacks do not expect large delays in processing, so this causes a
		303	higher drift (and a lot more system calls to get the current time).
		304
		305	In another sense, L<EV> is more exact, as your timer will be scheduled at
		306	the same time, regardless of how long event processing actually took.
		307
		308	In either case, if you care (and in most cases, you don't), then you
		309	can get whatever behaviour you want with any event loop, by taking the
		310	difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
		311	account.
		312
		313	=back
		314
		315	=head2 SIGNAL WATCHERS
		316
		317	You can watch for signals using a signal watcher, C<signal> is the signal
		318	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
		319	callback to be invoked whenever a signal occurs.
		320
		321	Although the callback might get passed parameters, their value and
		322	presence is undefined and you cannot rely on them. Portable AnyEvent
		323	callbacks cannot use arguments passed to signal watcher callbacks.
		324
		325	Multiple signal occurrences can be clumped together into one callback
		326	invocation, and callback invocation will be synchronous. Synchronous means
		327	that it might take a while until the signal gets handled by the process,
		328	but it is guaranteed not to interrupt any other callbacks.
		329
		330	The main advantage of using these watchers is that you can share a signal
		331	between multiple watchers.
		332
		333	This watcher might use C<%SIG>, so programs overwriting those signals
		334	directly will likely not work correctly.
		335
		336	Example: exit on SIGINT
		337
		338	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
		339
		340	=head2 CHILD PROCESS WATCHERS
		341
		342	You can also watch on a child process exit and catch its exit status.
		343
		344	The child process is specified by the C<pid> argument (if set to C<0>, it
		345	watches for any child process exit). The watcher will trigger as often
		346	as status change for the child are received. This works by installing a
		347	signal handler for C<SIGCHLD>. The callback will be called with the pid
		348	and exit status (as returned by waitpid), so unlike other watcher types,
		349	you I<can> rely on child watcher callback arguments.
		350
		351	There is a slight catch to child watchers, however: you usually start them
		352	I<after> the child process was created, and this means the process could
		353	have exited already (and no SIGCHLD will be sent anymore).
		354
		355	Not all event models handle this correctly (POE doesn't), but even for
		356	event models that I<do> handle this correctly, they usually need to be
		357	loaded before the process exits (i.e. before you fork in the first place).
		358
		359	This means you cannot create a child watcher as the very first thing in an
		360	AnyEvent program, you I<have> to create at least one watcher before you
		361	C<fork> the child (alternatively, you can call C<AnyEvent::detect>).
		362
		363	Example: fork a process and wait for it
		364
		365	my $done = AnyEvent->condvar;
		366
		367	my $pid = fork or exit 5;
		368
		369	my $w = AnyEvent->child (
		370	pid => $pid,
		371	cb => sub {
		372	my ($pid, $status) = @_;
		373	warn "pid $pid exited with status $status";
		374	$done->send;
		375	},
		376	);
		377
		378	# do something else, then wait for process exit
		379	$done->recv;
		380
		381	=head2 CONDITION VARIABLES
		382
		383	If you are familiar with some event loops you will know that all of them
		384	require you to run some blocking "loop", "run" or similar function that
		385	will actively watch for new events and call your callbacks.
		386
		387	AnyEvent is different, it expects somebody else to run the event loop and
		388	will only block when necessary (usually when told by the user).
		389
		390	The instrument to do that is called a "condition variable", so called
		391	because they represent a condition that must become true.
		392
		393	Condition variables can be created by calling the C<< AnyEvent->condvar
		394
		395	>> method, usually without arguments. The only argument pair allowed is
		396
		397	C<cb>, which specifies a callback to be called when the condition variable
		398	becomes true, with the condition variable as the first argument (but not
		399	the results).
		400
		401	After creation, the condition variable is "false" until it becomes "true"
		402	by calling the C<send> method (or calling the condition variable as if it
		403	were a callback, read about the caveats in the description for the C<<
		404	->send >> method).
		405
		406	Condition variables are similar to callbacks, except that you can
		407	optionally wait for them. They can also be called merge points - points
		408	in time where multiple outstanding events have been processed. And yet
		409	another way to call them is transactions - each condition variable can be
		410	used to represent a transaction, which finishes at some point and delivers
		411	a result.
		412
		413	Condition variables are very useful to signal that something has finished,
		414	for example, if you write a module that does asynchronous http requests,
		415	then a condition variable would be the ideal candidate to signal the
		416	availability of results. The user can either act when the callback is
		417	called or can synchronously C<< ->recv >> for the results.
		418
		419	You can also use them to simulate traditional event loops - for example,
		420	you can block your main program until an event occurs - for example, you
		421	could C<< ->recv >> in your main program until the user clicks the Quit
		422	button of your app, which would C<< ->send >> the "quit" event.
		423
		424	Note that condition variables recurse into the event loop - if you have
		425	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
		426	lose. Therefore, condition variables are good to export to your caller, but
		427	you should avoid making a blocking wait yourself, at least in callbacks,
		428	as this asks for trouble.
		429
		430	Condition variables are represented by hash refs in perl, and the keys
		431	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		432	easy (it is often useful to build your own transaction class on top of
		433	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		434	it's C<new> method in your own C<new> method.
		435
		436	There are two "sides" to a condition variable - the "producer side" which
		437	eventually calls C<< -> send >>, and the "consumer side", which waits
		438	for the send to occur.
		439
		440	Example: wait for a timer.
150		441
151	# wait till the result is ready	442	# wait till the result is ready
152	my $result_ready = AnyEvent->condvar;	443	my $result_ready = AnyEvent->condvar;
153		444
154	# do something such as adding a timer	445	# do something such as adding a timer
155	# or socket watcher the calls $result_ready->broadcast	446	# or socket watcher the calls $result_ready->send
156	# when the "result" is ready.	447	# when the "result" is ready.
		448	# in this case, we simply use a timer:
		449	my $w = AnyEvent->timer (
		450	after => 1,
		451	cb => sub { $result_ready->send },
		452	);
157		453
		454	# this "blocks" (while handling events) till the callback
		455	# calls send
158	$result_ready->wait;	456	$result_ready->recv;
		457
		458	Example: wait for a timer, but take advantage of the fact that
		459	condition variables are also code references.
		460
		461	my $done = AnyEvent->condvar;
		462	my $delay = AnyEvent->timer (after => 5, cb => $done);
		463	$done->recv;
		464
		465	Example: Imagine an API that returns a condvar and doesn't support
		466	callbacks. This is how you make a synchronous call, for example from
		467	the main program:
		468
		469	use AnyEvent::CouchDB;
		470
		471	...
		472
		473	my @info = $couchdb->info->recv;
		474
		475	And this is how you would just ste a callback to be called whenever the
		476	results are available:
		477
		478	$couchdb->info->cb (sub {
		479	my @info = $_[0]->recv;
		480	});
		481
		482	=head3 METHODS FOR PRODUCERS
		483
		484	These methods should only be used by the producing side, i.e. the
		485	code/module that eventually sends the signal. Note that it is also
		486	the producer side which creates the condvar in most cases, but it isn't
		487	uncommon for the consumer to create it as well.
		488
		489	=over 4
		490
		491	=item $cv->send (...)
		492
		493	Flag the condition as ready - a running C<< ->recv >> and all further
		494	calls to C<recv> will (eventually) return after this method has been
		495	called. If nobody is waiting the send will be remembered.
		496
		497	If a callback has been set on the condition variable, it is called
		498	immediately from within send.
		499
		500	Any arguments passed to the C<send> call will be returned by all
		501	future C<< ->recv >> calls.
		502
		503	Condition variables are overloaded so one can call them directly
		504	(as a code reference). Calling them directly is the same as calling
		505	C<send>. Note, however, that many C-based event loops do not handle
		506	overloading, so as tempting as it may be, passing a condition variable
		507	instead of a callback does not work. Both the pure perl and EV loops
		508	support overloading, however, as well as all functions that use perl to
		509	invoke a callback (as in L<AnyEvent::Socket> and L<AnyEvent::DNS> for
		510	example).
		511
		512	=item $cv->croak ($error)
		513
		514	Similar to send, but causes all call's to C<< ->recv >> to invoke
		515	C<Carp::croak> with the given error message/object/scalar.
		516
		517	This can be used to signal any errors to the condition variable
		518	user/consumer.
		519
		520	=item $cv->begin ([group callback])
		521
		522	=item $cv->end
		523
		524	These two methods are EXPERIMENTAL and MIGHT CHANGE.
		525
		526	These two methods can be used to combine many transactions/events into
		527	one. For example, a function that pings many hosts in parallel might want
		528	to use a condition variable for the whole process.
		529
		530	Every call to C<< ->begin >> will increment a counter, and every call to
		531	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		532	>>, the (last) callback passed to C<begin> will be executed. That callback
		533	is I<supposed> to call C<< ->send >>, but that is not required. If no
		534	callback was set, C<send> will be called without any arguments.
		535
		536	Let's clarify this with the ping example:
		537
		538	my $cv = AnyEvent->condvar;
		539
		540	my %result;
		541	$cv->begin (sub { $cv->send (\%result) });
		542
		543	for my $host (@list_of_hosts) {
		544	$cv->begin;
		545	ping_host_then_call_callback $host, sub {
		546	$result{$host} = ...;
		547	$cv->end;
		548	};
		549	}
		550
		551	$cv->end;
		552
		553	This code fragment supposedly pings a number of hosts and calls
		554	C<send> after results for all then have have been gathered - in any
		555	order. To achieve this, the code issues a call to C<begin> when it starts
		556	each ping request and calls C<end> when it has received some result for
		557	it. Since C<begin> and C<end> only maintain a counter, the order in which
		558	results arrive is not relevant.
		559
		560	There is an additional bracketing call to C<begin> and C<end> outside the
		561	loop, which serves two important purposes: first, it sets the callback
		562	to be called once the counter reaches C<0>, and second, it ensures that
		563	C<send> is called even when C<no> hosts are being pinged (the loop
		564	doesn't execute once).
		565
		566	This is the general pattern when you "fan out" into multiple subrequests:
		567	use an outer C<begin>/C<end> pair to set the callback and ensure C<end>
		568	is called at least once, and then, for each subrequest you start, call
		569	C<begin> and for each subrequest you finish, call C<end>.
159		570
160	=back	571	=back
161		572
162	=head2 SIGNAL WATCHERS	573	=head3 METHODS FOR CONSUMERS
163		574
164	You can listen for signals using a signal watcher, C<signal> is the signal	575	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	576	code awaits the condition.
166	together into one callback invocation, and callback invocation might or
167	might not be asynchronous.
168		577
169	These watchers might use C<%SIG>, so programs overwriting those signals	578	=over 4
170	directly will likely not work correctly.
171		579
172	Example: exit on SIGINT	580	=item $cv->recv
173		581
174	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });	582	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
		583	>> methods have been called on c<$cv>, while servicing other watchers
		584	normally.
175		585
176	=head2 CHILD PROCESS WATCHERS	586	You can only wait once on a condition - additional calls are valid but
		587	will return immediately.
177		588
178	You can also listen for the status of a child process specified by the	589	If an error condition has been set by calling C<< ->croak >>, then this
179	C<pid> argument. The watcher will only trigger once. This works by	590	function will call C<croak>.
180	installing a signal handler for C<SIGCHLD>.
181		591
182	Example: wait for pid 1333	592	In list context, all parameters passed to C<send> will be returned,
		593	in scalar context only the first one will be returned.
183		594
184	my $w = AnyEvent->child (pid => 1333, cb => sub { warn "exit status $?" });	595	Not all event models support a blocking wait - some die in that case
		596	(programs might want to do that to stay interactive), so I<if you are
		597	using this from a module, never require a blocking wait>, but let the
		598	caller decide whether the call will block or not (for example, by coupling
		599	condition variables with some kind of request results and supporting
		600	callbacks so the caller knows that getting the result will not block,
		601	while still supporting blocking waits if the caller so desires).
185		602
186	=head1 GLOBALS	603	Another reason I<never> to C<< ->recv >> in a module is that you cannot
		604	sensibly have two C<< ->recv >>'s in parallel, as that would require
		605	multiple interpreters or coroutines/threads, none of which C<AnyEvent>
		606	can supply.
		607
		608	The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
		609	fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
		610	versions and also integrates coroutines into AnyEvent, making blocking
		611	C<< ->recv >> calls perfectly safe as long as they are done from another
		612	coroutine (one that doesn't run the event loop).
		613
		614	You can ensure that C<< -recv >> never blocks by setting a callback and
		615	only calling C<< ->recv >> from within that callback (or at a later
		616	time). This will work even when the event loop does not support blocking
		617	waits otherwise.
		618
		619	=item $bool = $cv->ready
		620
		621	Returns true when the condition is "true", i.e. whether C<send> or
		622	C<croak> have been called.
		623
		624	=item $cb = $cv->cb ($cb->($cv))
		625
		626	This is a mutator function that returns the callback set and optionally
		627	replaces it before doing so.
		628
		629	The callback will be called when the condition becomes "true", i.e. when
		630	C<send> or C<croak> are called, with the only argument being the condition
		631	variable itself. Calling C<recv> inside the callback or at any later time
		632	is guaranteed not to block.
		633
		634	=back
		635
		636	=head1 GLOBAL VARIABLES AND FUNCTIONS
187		637
188	=over 4	638	=over 4
189		639
190	=item $AnyEvent::MODEL	640	=item $AnyEvent::MODEL
191		641
…		…
195	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	645	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case
196	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	646	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).
197		647
198	The known classes so far are:	648	The known classes so far are:
199		649
200	EV::AnyEvent based on EV (an interface to libev, best choice)	650	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
201	AnyEvent::Impl::Coro based on Coro::Event, second best choice.
202	AnyEvent::Impl::Event based on Event, also second best choice :)	651	AnyEvent::Impl::Event based on Event, second best choice.
		652	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
203	AnyEvent::Impl::Glib based on Glib, second-best choice.	653	AnyEvent::Impl::Glib based on Glib, third-best choice.
204	AnyEvent::Impl::Tk based on Tk, very bad choice.	654	AnyEvent::Impl::Tk based on Tk, very bad choice.
205	AnyEvent::Impl::Perl pure-perl implementation, inefficient.	655	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
		656	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
		657	AnyEvent::Impl::POE based on POE, not generic enough for full support.
		658
		659	There is no support for WxWidgets, as WxWidgets has no support for
		660	watching file handles. However, you can use WxWidgets through the
		661	POE Adaptor, as POE has a Wx backend that simply polls 20 times per
		662	second, which was considered to be too horrible to even consider for
		663	AnyEvent. Likewise, other POE backends can be used by AnyEvent by using
		664	it's adaptor.
		665
		666	AnyEvent knows about L<Prima> and L<Wx> and will try to use L<POE> when
		667	autodetecting them.
206		668
207	=item AnyEvent::detect	669	=item AnyEvent::detect
208		670
209	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model if	671	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
210	necessary. You should only call this function right before you would have	672	if necessary. You should only call this function right before you would
211	created an AnyEvent watcher anyway, that is, very late at runtime.	673	have created an AnyEvent watcher anyway, that is, as late as possible at
		674	runtime.
		675
		676	=item $guard = AnyEvent::post_detect { BLOCK }
		677
		678	Arranges for the code block to be executed as soon as the event model is
		679	autodetected (or immediately if this has already happened).
		680
		681	If called in scalar or list context, then it creates and returns an object
		682	that automatically removes the callback again when it is destroyed. See
		683	L<Coro::BDB> for a case where this is useful.
		684
		685	=item @AnyEvent::post_detect
		686
		687	If there are any code references in this array (you can C<push> to it
		688	before or after loading AnyEvent), then they will called directly after
		689	the event loop has been chosen.
		690
		691	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		692	if it contains a true value then the event loop has already been detected,
		693	and the array will be ignored.
		694
		695	Best use C<AnyEvent::post_detect { BLOCK }> instead.
212		696
213	=back	697	=back
214		698
215	=head1 WHAT TO DO IN A MODULE	699	=head1 WHAT TO DO IN A MODULE
216		700
217	As a module author, you should "use AnyEvent" and call AnyEvent methods	701	As a module author, you should C<use AnyEvent> and call AnyEvent methods
218	freely, but you should not load a specific event module or rely on it.	702	freely, but you should not load a specific event module or rely on it.
219		703
220	Be careful when you create watchers in the module body - Anyevent will	704	Be careful when you create watchers in the module body - AnyEvent will
221	decide which event module to use as soon as the first method is called, so	705	decide which event module to use as soon as the first method is called, so
222	by calling AnyEvent in your module body you force the user of your module	706	by calling AnyEvent in your module body you force the user of your module
223	to load the event module first.	707	to load the event module first.
224		708
		709	Never call C<< ->recv >> on a condition variable unless you I<know> that
		710	the C<< ->send >> method has been called on it already. This is
		711	because it will stall the whole program, and the whole point of using
		712	events is to stay interactive.
		713
		714	It is fine, however, to call C<< ->recv >> when the user of your module
		715	requests it (i.e. if you create a http request object ad have a method
		716	called C<results> that returns the results, it should call C<< ->recv >>
		717	freely, as the user of your module knows what she is doing. always).
		718
225	=head1 WHAT TO DO IN THE MAIN PROGRAM	719	=head1 WHAT TO DO IN THE MAIN PROGRAM
226		720
227	There will always be a single main program - the only place that should	721	There will always be a single main program - the only place that should
228	dictate which event model to use.	722	dictate which event model to use.
229		723
230	If it doesn't care, it can just "use AnyEvent" and use it itself, or not	724	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
231	do anything special and let AnyEvent decide which implementation to chose.	725	do anything special (it does not need to be event-based) and let AnyEvent
		726	decide which implementation to chose if some module relies on it.
232		727
233	If the main program relies on a specific event model (for example, in Gtk2	728	If the main program relies on a specific event model - for example, in
234	programs you have to rely on either Glib or Glib::Event), you should load	729	Gtk2 programs you have to rely on the Glib module - you should load the
235	it before loading AnyEvent or any module that uses it, generally, as early	730	event module before loading AnyEvent or any module that uses it: generally
236	as possible. The reason is that modules might create watchers when they	731	speaking, you should load it as early as possible. The reason is that
237	are loaded, and AnyEvent will decide on the event model to use as soon as	732	modules might create watchers when they are loaded, and AnyEvent will
238	it creates watchers, and it might chose the wrong one unless you load the	733	decide on the event model to use as soon as it creates watchers, and it
239	correct one yourself.	734	might chose the wrong one unless you load the correct one yourself.
240		735
241	You can chose to use a rather inefficient pure-perl implementation by	736	You can chose to use a pure-perl implementation by loading the
242	loading the C<AnyEvent::Impl::Perl> module, but letting AnyEvent chose is	737	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
243	generally better.	738	everywhere, but letting AnyEvent chose the model is generally better.
		739
		740	=head2 MAINLOOP EMULATION
		741
		742	Sometimes (often for short test scripts, or even standalone programs who
		743	only want to use AnyEvent), you do not want to run a specific event loop.
		744
		745	In that case, you can use a condition variable like this:
		746
		747	AnyEvent->condvar->recv;
		748
		749	This has the effect of entering the event loop and looping forever.
		750
		751	Note that usually your program has some exit condition, in which case
		752	it is better to use the "traditional" approach of storing a condition
		753	variable somewhere, waiting for it, and sending it when the program should
		754	exit cleanly.
		755
		756
		757	=head1 OTHER MODULES
		758
		759	The following is a non-exhaustive list of additional modules that use
		760	AnyEvent and can therefore be mixed easily with other AnyEvent modules
		761	in the same program. Some of the modules come with AnyEvent, some are
		762	available via CPAN.
		763
		764	=over 4
		765
		766	=item L<AnyEvent::Util>
		767
		768	Contains various utility functions that replace often-used but blocking
		769	functions such as C<inet_aton> by event-/callback-based versions.
		770
		771	=item L<AnyEvent::Socket>
		772
		773	Provides various utility functions for (internet protocol) sockets,
		774	addresses and name resolution. Also functions to create non-blocking tcp
		775	connections or tcp servers, with IPv6 and SRV record support and more.
		776
		777	=item L<AnyEvent::Handle>
		778
		779	Provide read and write buffers, manages watchers for reads and writes,
		780	supports raw and formatted I/O, I/O queued and fully transparent and
		781	non-blocking SSL/TLS.
		782
		783	=item L<AnyEvent::DNS>
		784
		785	Provides rich asynchronous DNS resolver capabilities.
		786
		787	=item L<AnyEvent::HTTP>
		788
		789	A simple-to-use HTTP library that is capable of making a lot of concurrent
		790	HTTP requests.
		791
		792	=item L<AnyEvent::HTTPD>
		793
		794	Provides a simple web application server framework.
		795
		796	=item L<AnyEvent::FastPing>
		797
		798	The fastest ping in the west.
		799
		800	=item L<AnyEvent::DBI>
		801
		802	Executes L<DBI> requests asynchronously in a proxy process.
		803
		804	=item L<AnyEvent::AIO>
		805
		806	Truly asynchronous I/O, should be in the toolbox of every event
		807	programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
		808	together.
		809
		810	=item L<AnyEvent::BDB>
		811
		812	Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
		813	L<BDB> and AnyEvent together.
		814
		815	=item L<AnyEvent::GPSD>
		816
		817	A non-blocking interface to gpsd, a daemon delivering GPS information.
		818
		819	=item L<AnyEvent::IGS>
		820
		821	A non-blocking interface to the Internet Go Server protocol (used by
		822	L<App::IGS>).
		823
		824	=item L<Net::IRC3>
		825
		826	AnyEvent based IRC client module family.
		827
		828	=item L<Net::XMPP2>
		829
		830	AnyEvent based XMPP (Jabber protocol) module family.
		831
		832	=item L<Net::FCP>
		833
		834	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		835	of AnyEvent.
		836
		837	=item L<Event::ExecFlow>
		838
		839	High level API for event-based execution flow control.
		840
		841	=item L<Coro>
		842
		843	Has special support for AnyEvent via L<Coro::AnyEvent>.
		844
		845	=item L<IO::Lambda>
		846
		847	The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
		848
		849	=back
244		850
245	=cut	851	=cut
246		852
247	package AnyEvent;	853	package AnyEvent;
248		854
249	no warnings;	855	no warnings;
250	use strict;	856	use strict;
251		857
252	use Carp;	858	use Carp;
253		859
254	our $VERSION = '2.55';	860	our $VERSION = 4.22;
255	our $MODEL;	861	our $MODEL;
256		862
257	our $AUTOLOAD;	863	our $AUTOLOAD;
258	our @ISA;	864	our @ISA;
259		865
		866	our @REGISTRY;
		867
		868	our $WIN32;
		869
		870	BEGIN {
		871	my $win32 = ! ! ($^O =~ /mswin32/i);
		872	eval "sub WIN32(){ $win32 }";
		873	}
		874
260	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;	875	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
261		876
262	our @REGISTRY;	877	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		878
		879	{
		880	my $idx;
		881	$PROTOCOL{$_} = ++$idx
		882	for reverse split /\s*,\s*/,
		883	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		884	}
263		885
264	my @models = (	886	my @models = (
265	[Coro::Event:: => AnyEvent::Impl::Coro::],
266	[EV:: => EV::AnyEvent::],	887	[EV:: => AnyEvent::Impl::EV::],
267	[Event:: => AnyEvent::Impl::Event::],	888	[Event:: => AnyEvent::Impl::Event::],
268	[Glib:: => AnyEvent::Impl::Glib::],
269	[Tk:: => AnyEvent::Impl::Tk::],
270	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	889	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
		890	# everything below here will not be autoprobed
		891	# as the pureperl backend should work everywhere
		892	# and is usually faster
		893	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		894	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
		895	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
		896	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
		897	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
		898	[Wx:: => AnyEvent::Impl::POE::],
		899	[Prima:: => AnyEvent::Impl::POE::],
271	);	900	);
272		901
273	our %method = map +($_ => 1), qw(io timer condvar broadcast wait signal one_event DESTROY);	902	our %method = map +($_ => 1), qw(io timer time now signal child condvar one_event DESTROY);
		903
		904	our @post_detect;
		905
		906	sub post_detect(&) {
		907	my ($cb) = @_;
		908
		909	if ($MODEL) {
		910	$cb->();
		911
		912	1
		913	} else {
		914	push @post_detect, $cb;
		915
		916	defined wantarray
		917	? bless \$cb, "AnyEvent::Util::PostDetect"
		918	: ()
		919	}
		920	}
		921
		922	sub AnyEvent::Util::PostDetect::DESTROY {
		923	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		924	}
274		925
275	sub detect() {	926	sub detect() {
276	unless ($MODEL) {	927	unless ($MODEL) {
277	no strict 'refs';	928	no strict 'refs';
		929	local $SIG{__DIE__};
		930
		931	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
		932	my $model = "AnyEvent::Impl::$1";
		933	if (eval "require $model") {
		934	$MODEL = $model;
		935	warn "AnyEvent: loaded model '$model' (forced by \$PERL_ANYEVENT_MODEL), using it.\n" if $verbose > 1;
		936	} else {
		937	warn "AnyEvent: unable to load model '$model' (from \$PERL_ANYEVENT_MODEL):\n$@" if $verbose;
		938	}
		939	}
278		940
279	# check for already loaded models	941	# check for already loaded models
		942	unless ($MODEL) {
280	for (@REGISTRY, @models) {	943	for (@REGISTRY, @models) {
281	my ($package, $model) = @$_;	944	my ($package, $model) = @$_;
282	if (${"$package\::VERSION"} > 0) {	945	if (${"$package\::VERSION"} > 0) {
283	if (eval "require $model") {	946	if (eval "require $model") {
284	$MODEL = $model;	947	$MODEL = $model;
285	warn "AnyEvent: found model '$model', using it.\n" if $verbose > 1;	948	warn "AnyEvent: autodetected model '$model', using it.\n" if $verbose > 1;
286	last;	949	last;
		950	}
287	}	951	}
288	}	952	}
		953
		954	unless ($MODEL) {
		955	# try to load a model
		956
		957	for (@REGISTRY, @models) {
		958	my ($package, $model) = @$_;
		959	if (eval "require $package"
		960	and ${"$package\::VERSION"} > 0
		961	and eval "require $model") {
		962	$MODEL = $model;
		963	warn "AnyEvent: autoprobed model '$model', using it.\n" if $verbose > 1;
		964	last;
		965	}
		966	}
		967
		968	$MODEL
		969	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.";
		970	}
289	}	971	}
290		972
291	unless ($MODEL) {	973	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
292	# try to load a model
293
294	for (@REGISTRY, @models) {
295	my ($package, $model) = @$_;
296	if (eval "require $package"
297	and ${"$package\::VERSION"} > 0
298	and eval "require $model") {
299	$MODEL = $model;
300	warn "AnyEvent: autoprobed and loaded model '$model', using it.\n" if $verbose > 1;
301	last;
302	}
303	}
304
305	$MODEL
306	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: Event (or Coro+Event), Glib or Tk.";
307	}
308		974
309	unshift @ISA, $MODEL;	975	unshift @ISA, $MODEL;
310	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	976
		977	require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
		978
		979	(shift @post_detect)->() while @post_detect;
311	}	980	}
312		981
313	$MODEL	982	$MODEL
314	}	983	}
315		984
…		…
323		992
324	my $class = shift;	993	my $class = shift;
325	$class->$func (@_);	994	$class->$func (@_);
326	}	995	}
327		996
		997	# utility function to dup a filehandle. this is used by many backends
		998	# to support binding more than one watcher per filehandle (they usually
		999	# allow only one watcher per fd, so we dup it to get a different one).
		1000	sub _dupfh($$$$) {
		1001	my ($poll, $fh, $r, $w) = @_;
		1002
		1003	require Fcntl;
		1004
		1005	# cygwin requires the fh mode to be matching, unix doesn't
		1006	my ($rw, $mode) = $poll eq "r" ? ($r, "<")
		1007	: $poll eq "w" ? ($w, ">")
		1008	: Carp::croak "AnyEvent->io requires poll set to either 'r' or 'w'";
		1009
		1010	open my $fh2, "$mode&" . fileno $fh
		1011	or die "cannot dup() filehandle: $!";
		1012
		1013	# we assume CLOEXEC is already set by perl in all important cases
		1014
		1015	($fh2, $rw)
		1016	}
		1017
328	package AnyEvent::Base;	1018	package AnyEvent::Base;
329		1019
		1020	# default implementation for now and time
		1021
		1022	use Time::HiRes ();
		1023
		1024	sub time { Time::HiRes::time }
		1025	sub now { Time::HiRes::time }
		1026
330	# default implementation for ->condvar, ->wait, ->broadcast	1027	# default implementation for ->condvar
331		1028
332	sub condvar {	1029	sub condvar {
333	bless \my $flag, "AnyEvent::Base::CondVar"	1030	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar::
334	}
335
336	sub AnyEvent::Base::CondVar::broadcast {
337	${$_[0]}++;
338	}
339
340	sub AnyEvent::Base::CondVar::wait {
341	AnyEvent->one_event while !${$_[0]};
342	}	1031	}
343		1032
344	# default implementation for ->signal	1033	# default implementation for ->signal
345		1034
346	our %SIG_CB;	1035	our %SIG_CB;
…		…
362	sub AnyEvent::Base::Signal::DESTROY {	1051	sub AnyEvent::Base::Signal::DESTROY {
363	my ($signal, $cb) = @{$_[0]};	1052	my ($signal, $cb) = @{$_[0]};
364		1053
365	delete $SIG_CB{$signal}{$cb};	1054	delete $SIG_CB{$signal}{$cb};
366		1055
367	$SIG{$signal} = 'DEFAULT' unless keys %{ $SIG_CB{$signal} };	1056	delete $SIG{$signal} unless keys %{ $SIG_CB{$signal} };
368	}	1057	}
369		1058
370	# default implementation for ->child	1059	# default implementation for ->child
371		1060
372	our %PID_CB;	1061	our %PID_CB;
373	our $CHLD_W;	1062	our $CHLD_W;
		1063	our $CHLD_DELAY_W;
374	our $PID_IDLE;	1064	our $PID_IDLE;
375	our $WNOHANG;	1065	our $WNOHANG;
376		1066
377	sub _child_wait {	1067	sub _child_wait {
378	while (0 < (my $pid = waitpid -1, $WNOHANG)) {	1068	while (0 < (my $pid = waitpid -1, $WNOHANG)) {
379	$_->() for values %{ (delete $PID_CB{$pid}) || {} };	1069	$_->($pid, $?) for (values %{ $PID_CB{$pid} || {} }),
		1070	(values %{ $PID_CB{0} || {} });
380	}	1071	}
381		1072
382	undef $PID_IDLE;	1073	undef $PID_IDLE;
		1074	}
		1075
		1076	sub _sigchld {
		1077	# make sure we deliver these changes "synchronous" with the event loop.
		1078	$CHLD_DELAY_W ||= AnyEvent->timer (after => 0, cb => sub {
		1079	undef $CHLD_DELAY_W;
		1080	&_child_wait;
		1081	});
383	}	1082	}
384		1083
385	sub child {	1084	sub child {
386	my (undef, %arg) = @_;	1085	my (undef, %arg) = @_;
387		1086
388	my $pid = uc $arg{pid}	1087	defined (my $pid = $arg{pid} + 0)
389	or Carp::croak "required option 'pid' is missing";	1088	or Carp::croak "required option 'pid' is missing";
390		1089
391	$PID_CB{$pid}{$arg{cb}} = $arg{cb};	1090	$PID_CB{$pid}{$arg{cb}} = $arg{cb};
392		1091
393	unless ($WNOHANG) {	1092	unless ($WNOHANG) {
394	$WNOHANG = eval { require POSIX; &POSIX::WNOHANG } || 1;	1093	$WNOHANG = eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
395	}	1094	}
396		1095
397	unless ($CHLD_W) {	1096	unless ($CHLD_W) {
398	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_child_wait);	1097	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
399	# child could be a zombie already	1098	# child could be a zombie already, so make at least one round
400	$PID_IDLE ||= AnyEvent->timer (after => 0, cb => \&_child_wait);	1099	&_sigchld;
401	}	1100	}
402		1101
403	bless [$pid, $arg{cb}], "AnyEvent::Base::Child"	1102	bless [$pid, $arg{cb}], "AnyEvent::Base::Child"
404	}	1103	}
405		1104
…		…
410	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };	1109	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
411		1110
412	undef $CHLD_W unless keys %PID_CB;	1111	undef $CHLD_W unless keys %PID_CB;
413	}	1112	}
414		1113
		1114	package AnyEvent::CondVar;
		1115
		1116	our @ISA = AnyEvent::CondVar::Base::;
		1117
		1118	package AnyEvent::CondVar::Base;
		1119
		1120	use overload
		1121	'&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1122	fallback => 1;
		1123
		1124	sub _send {
		1125	# nop
		1126	}
		1127
		1128	sub send {
		1129	my $cv = shift;
		1130	$cv->{_ae_sent} = [@_];
		1131	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1132	$cv->_send;
		1133	}
		1134
		1135	sub croak {
		1136	$_[0]{_ae_croak} = $_[1];
		1137	$_[0]->send;
		1138	}
		1139
		1140	sub ready {
		1141	$_[0]{_ae_sent}
		1142	}
		1143
		1144	sub _wait {
		1145	AnyEvent->one_event while !$_[0]{_ae_sent};
		1146	}
		1147
		1148	sub recv {
		1149	$_[0]->_wait;
		1150
		1151	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		1152	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		1153	}
		1154
		1155	sub cb {
		1156	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		1157	$_[0]{_ae_cb}
		1158	}
		1159
		1160	sub begin {
		1161	++$_[0]{_ae_counter};
		1162	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		1163	}
		1164
		1165	sub end {
		1166	return if --$_[0]{_ae_counter};
		1167	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		1168	}
		1169
		1170	# undocumented/compatibility with pre-3.4
		1171	*broadcast = \&send;
		1172	*wait = \&_wait;
		1173
415	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1174	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
		1175
		1176	This is an advanced topic that you do not normally need to use AnyEvent in
		1177	a module. This section is only of use to event loop authors who want to
		1178	provide AnyEvent compatibility.
416		1179
417	If you need to support another event library which isn't directly	1180	If you need to support another event library which isn't directly
418	supported by AnyEvent, you can supply your own interface to it by	1181	supported by AnyEvent, you can supply your own interface to it by
419	pushing, before the first watcher gets created, the package name of	1182	pushing, before the first watcher gets created, the package name of
420	the event module and the package name of the interface to use onto	1183	the event module and the package name of the interface to use onto
421	C<@AnyEvent::REGISTRY>. You can do that before and even without loading	1184	C<@AnyEvent::REGISTRY>. You can do that before and even without loading
422	AnyEvent.	1185	AnyEvent, so it is reasonably cheap.
423		1186
424	Example:	1187	Example:
425		1188
426	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];	1189	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];
427		1190
428	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>	1191	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>
429	package/class when it finds the C<urxvt> package/module is loaded. When	1192	package/class when it finds the C<urxvt> package/module is already loaded.
		1193
430	AnyEvent is loaded and asked to find a suitable event model, it will	1194	When AnyEvent is loaded and asked to find a suitable event model, it
431	first check for the presence of urxvt.	1195	will first check for the presence of urxvt by trying to C<use> the
		1196	C<urxvt::anyevent> module.
432		1197
433	The class should provide implementations for all watcher types (see	1198	The class should provide implementations for all watcher types. See
434	L<AnyEvent::Impl::Event> (source code), L<AnyEvent::Impl::Glib>	1199	L<AnyEvent::Impl::EV> (source code), L<AnyEvent::Impl::Glib> (Source code)
435	(Source code) and so on for actual examples, use C<perldoc -m	1200	and so on for actual examples. Use C<perldoc -m AnyEvent::Impl::Glib> to
436	AnyEvent::Impl::Glib> to see the sources).	1201	see the sources.
437		1202
		1203	If you don't provide C<signal> and C<child> watchers than AnyEvent will
		1204	provide suitable (hopefully) replacements.
		1205
438	The above isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)	1206	The above example isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)
439	uses the above line as-is. An interface isn't included in AnyEvent	1207	terminal emulator uses the above line as-is. An interface isn't included
440	because it doesn't make sense outside the embedded interpreter inside	1208	in AnyEvent because it doesn't make sense outside the embedded interpreter
441	I<rxvt-unicode>, and it is updated and maintained as part of the	1209	inside I<rxvt-unicode>, and it is updated and maintained as part of the
442	I<rxvt-unicode> distribution.	1210	I<rxvt-unicode> distribution.
443		1211
444	I<rxvt-unicode> also cheats a bit by not providing blocking access to	1212	I<rxvt-unicode> also cheats a bit by not providing blocking access to
445	condition variables: code blocking while waiting for a condition will	1213	condition variables: code blocking while waiting for a condition will
446	C<die>. This still works with most modules/usages, and blocking calls must	1214	C<die>. This still works with most modules/usages, and blocking calls must
447	not be in an interactive application, so it makes sense.	1215	not be done in an interactive application, so it makes sense.
448		1216
449	=head1 ENVIRONMENT VARIABLES	1217	=head1 ENVIRONMENT VARIABLES
450		1218
451	The following environment variables are used by this module:	1219	The following environment variables are used by this module:
452		1220
453	C<PERL_ANYEVENT_VERBOSE> when set to C<2> or higher, reports which event	1221	=over 4
454	model gets used.
455		1222
		1223	=item C<PERL_ANYEVENT_VERBOSE>
		1224
		1225	By default, AnyEvent will be completely silent except in fatal
		1226	conditions. You can set this environment variable to make AnyEvent more
		1227	talkative.
		1228
		1229	When set to C<1> or higher, causes AnyEvent to warn about unexpected
		1230	conditions, such as not being able to load the event model specified by
		1231	C<PERL_ANYEVENT_MODEL>.
		1232
		1233	When set to C<2> or higher, cause AnyEvent to report to STDERR which event
		1234	model it chooses.
		1235
		1236	=item C<PERL_ANYEVENT_STRICT>
		1237
		1238	AnyEvent does not do much argument checking by default, as thorough
		1239	argument checking is very costly. Setting this variable to a true value
		1240	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		1241	check the arguments passed to most method calls. If it finds any problems
		1242	it will croak.
		1243
		1244	In other words, enables "strict" mode.
		1245
		1246	Unlike C<use strict> it is definitely recommended ot keep it off in
		1247	production.
		1248
		1249	=item C<PERL_ANYEVENT_MODEL>
		1250
		1251	This can be used to specify the event model to be used by AnyEvent, before
		1252	auto detection and -probing kicks in. It must be a string consisting
		1253	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
		1254	and the resulting module name is loaded and if the load was successful,
		1255	used as event model. If it fails to load AnyEvent will proceed with
		1256	auto detection and -probing.
		1257
		1258	This functionality might change in future versions.
		1259
		1260	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
		1261	could start your program like this:
		1262
		1263	PERL_ANYEVENT_MODEL=Perl perl ...
		1264
		1265	=item C<PERL_ANYEVENT_PROTOCOLS>
		1266
		1267	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1268	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1269	of auto probing).
		1270
		1271	Must be set to a comma-separated list of protocols or address families,
		1272	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1273	used, and preference will be given to protocols mentioned earlier in the
		1274	list.
		1275
		1276	This variable can effectively be used for denial-of-service attacks
		1277	against local programs (e.g. when setuid), although the impact is likely
		1278	small, as the program has to handle connection errors already-
		1279
		1280	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1281	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1282	- only support IPv4, never try to resolve or contact IPv6
		1283	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1284	IPv6, but prefer IPv6 over IPv4.
		1285
		1286	=item C<PERL_ANYEVENT_EDNS0>
		1287
		1288	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1289	for DNS. This extension is generally useful to reduce DNS traffic, but
		1290	some (broken) firewalls drop such DNS packets, which is why it is off by
		1291	default.
		1292
		1293	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1294	EDNS0 in its DNS requests.
		1295
		1296	=item C<PERL_ANYEVENT_MAX_FORKS>
		1297
		1298	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		1299	will create in parallel.
		1300
		1301	=back
		1302
456	=head1 EXAMPLE	1303	=head1 EXAMPLE PROGRAM
457		1304
458	The following program uses an io watcher to read data from stdin, a timer	1305	The following program uses an I/O watcher to read data from STDIN, a timer
459	to display a message once per second, and a condvar to exit the program	1306	to display a message once per second, and a condition variable to quit the
460	when the user enters quit:	1307	program when the user enters quit:
461		1308
462	use AnyEvent;	1309	use AnyEvent;
463		1310
464	my $cv = AnyEvent->condvar;	1311	my $cv = AnyEvent->condvar;
465		1312
466	my $io_watcher = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	1313	my $io_watcher = AnyEvent->io (
		1314	fh => \*STDIN,
		1315	poll => 'r',
		1316	cb => sub {
467	warn "io event <$_[0]>\n"; # will always output <r>	1317	warn "io event <$_[0]>\n"; # will always output <r>
468	chomp (my $input = <STDIN>); # read a line	1318	chomp (my $input = <STDIN>); # read a line
469	warn "read: $input\n"; # output what has been read	1319	warn "read: $input\n"; # output what has been read
470	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1320	$cv->send if $input =~ /^q/i; # quit program if /^q/i
		1321	},
471	});	1322	);
472		1323
473	my $time_watcher; # can only be used once	1324	my $time_watcher; # can only be used once
474		1325
475	sub new_timer {	1326	sub new_timer {
476	$timer = AnyEvent->timer (after => 1, cb => sub {	1327	$timer = AnyEvent->timer (after => 1, cb => sub {
…		…
479	});	1330	});
480	}	1331	}
481		1332
482	new_timer; # create first timer	1333	new_timer; # create first timer
483		1334
484	$cv->wait; # wait until user enters /^q/i	1335	$cv->recv; # wait until user enters /^q/i
485		1336
486	=head1 REAL-WORLD EXAMPLE	1337	=head1 REAL-WORLD EXAMPLE
487		1338
488	Consider the L<Net::FCP> module. It features (among others) the following	1339	Consider the L<Net::FCP> module. It features (among others) the following
489	API calls, which are to freenet what HTTP GET requests are to http:	1340	API calls, which are to freenet what HTTP GET requests are to http:
…		…
539	syswrite $txn->{fh}, $txn->{request}	1390	syswrite $txn->{fh}, $txn->{request}
540	or die "connection or write error";	1391	or die "connection or write error";
541	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	1392	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
542		1393
543	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	1394	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
544	result and signals any possible waiters that the request ahs finished:	1395	result and signals any possible waiters that the request has finished:
545		1396
546	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1397	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
547		1398
548	if (end-of-file or data complete) {	1399	if (end-of-file or data complete) {
549	$txn->{result} = $txn->{buf};	1400	$txn->{result} = $txn->{buf};
550	$txn->{finished}->broadcast;	1401	$txn->{finished}->send;
551	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1402	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
552	}	1403	}
553		1404
554	The C<result> method, finally, just waits for the finished signal (if the	1405	The C<result> method, finally, just waits for the finished signal (if the
555	request was already finished, it doesn't wait, of course, and returns the	1406	request was already finished, it doesn't wait, of course, and returns the
556	data:	1407	data:
557		1408
558	$txn->{finished}->wait;	1409	$txn->{finished}->recv;
559	return $txn->{result};	1410	return $txn->{result};
560		1411
561	The actual code goes further and collects all errors (C<die>s, exceptions)	1412	The actual code goes further and collects all errors (C<die>s, exceptions)
562	that occured during request processing. The C<result> method detects	1413	that occurred during request processing. The C<result> method detects
563	wether an exception as thrown (it is stored inside the $txn object)	1414	whether an exception as thrown (it is stored inside the $txn object)
564	and just throws the exception, which means connection errors and other	1415	and just throws the exception, which means connection errors and other
565	problems get reported tot he code that tries to use the result, not in a	1416	problems get reported tot he code that tries to use the result, not in a
566	random callback.	1417	random callback.
567		1418
568	All of this enables the following usage styles:	1419	All of this enables the following usage styles:
569		1420
570	1. Blocking:	1421	1. Blocking:
571		1422
572	my $data = $fcp->client_get ($url);	1423	my $data = $fcp->client_get ($url);
573		1424
574	2. Blocking, but parallelizing:	1425	2. Blocking, but running in parallel:
575		1426
576	my @datas = map $_->result,	1427	my @datas = map $_->result,
577	map $fcp->txn_client_get ($_),	1428	map $fcp->txn_client_get ($_),
578	@urls;	1429	@urls;
579		1430
580	Both blocking examples work without the module user having to know	1431	Both blocking examples work without the module user having to know
581	anything about events.	1432	anything about events.
582		1433
583	3a. Event-based in a main program, using any support Event module:	1434	3a. Event-based in a main program, using any supported event module:
584		1435
585	use Event;	1436	use EV;
586		1437
587	$fcp->txn_client_get ($url)->cb (sub {	1438	$fcp->txn_client_get ($url)->cb (sub {
588	my $txn = shift;	1439	my $txn = shift;
589	my $data = $txn->result;	1440	my $data = $txn->result;
590	...	1441	...
591	});	1442	});
592		1443
593	Event::loop;	1444	EV::loop;
594		1445
595	3b. The module user could use AnyEvent, too:	1446	3b. The module user could use AnyEvent, too:
596		1447
597	use AnyEvent;	1448	use AnyEvent;
598		1449
599	my $quit = AnyEvent->condvar;	1450	my $quit = AnyEvent->condvar;
600		1451
601	$fcp->txn_client_get ($url)->cb (sub {	1452	$fcp->txn_client_get ($url)->cb (sub {
602	...	1453	...
603	$quit->broadcast;	1454	$quit->send;
604	});	1455	});
605		1456
606	$quit->wait;	1457	$quit->recv;
		1458
		1459
		1460	=head1 BENCHMARKS
		1461
		1462	To give you an idea of the performance and overheads that AnyEvent adds
		1463	over the event loops themselves and to give you an impression of the speed
		1464	of various event loops I prepared some benchmarks.
		1465
		1466	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1467
		1468	Here is a benchmark of various supported event models used natively and
		1469	through AnyEvent. The benchmark creates a lot of timers (with a zero
		1470	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
		1471	which it is), lets them fire exactly once and destroys them again.
		1472
		1473	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
		1474	distribution.
		1475
		1476	=head3 Explanation of the columns
		1477
		1478	I<watcher> is the number of event watchers created/destroyed. Since
		1479	different event models feature vastly different performances, each event
		1480	loop was given a number of watchers so that overall runtime is acceptable
		1481	and similar between tested event loop (and keep them from crashing): Glib
		1482	would probably take thousands of years if asked to process the same number
		1483	of watchers as EV in this benchmark.
		1484
		1485	I<bytes> is the number of bytes (as measured by the resident set size,
		1486	RSS) consumed by each watcher. This method of measuring captures both C
		1487	and Perl-based overheads.
		1488
		1489	I<create> is the time, in microseconds (millionths of seconds), that it
		1490	takes to create a single watcher. The callback is a closure shared between
		1491	all watchers, to avoid adding memory overhead. That means closure creation
		1492	and memory usage is not included in the figures.
		1493
		1494	I<invoke> is the time, in microseconds, used to invoke a simple
		1495	callback. The callback simply counts down a Perl variable and after it was
		1496	invoked "watcher" times, it would C<< ->send >> a condvar once to
		1497	signal the end of this phase.
		1498
		1499	I<destroy> is the time, in microseconds, that it takes to destroy a single
		1500	watcher.
		1501
		1502	=head3 Results
		1503
		1504	name watchers bytes create invoke destroy comment
		1505	EV/EV 400000 244 0.56 0.46 0.31 EV native interface
		1506	EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers
		1507	CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal
		1508	Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation
		1509	Event/Event 16000 516 31.88 31.30 0.85 Event native interface
		1510	Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers
		1511	Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour
		1512	Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers
		1513	POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event
		1514	POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select
		1515
		1516	=head3 Discussion
		1517
		1518	The benchmark does I<not> measure scalability of the event loop very
		1519	well. For example, a select-based event loop (such as the pure perl one)
		1520	can never compete with an event loop that uses epoll when the number of
		1521	file descriptors grows high. In this benchmark, all events become ready at
		1522	the same time, so select/poll-based implementations get an unnatural speed
		1523	boost.
		1524
		1525	Also, note that the number of watchers usually has a nonlinear effect on
		1526	overall speed, that is, creating twice as many watchers doesn't take twice
		1527	the time - usually it takes longer. This puts event loops tested with a
		1528	higher number of watchers at a disadvantage.
		1529
		1530	To put the range of results into perspective, consider that on the
		1531	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1532	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1533	cycles with POE.
		1534
		1535	C<EV> is the sole leader regarding speed and memory use, which are both
		1536	maximal/minimal, respectively. Even when going through AnyEvent, it uses
		1537	far less memory than any other event loop and is still faster than Event
		1538	natively.
		1539
		1540	The pure perl implementation is hit in a few sweet spots (both the
		1541	constant timeout and the use of a single fd hit optimisations in the perl
		1542	interpreter and the backend itself). Nevertheless this shows that it
		1543	adds very little overhead in itself. Like any select-based backend its
		1544	performance becomes really bad with lots of file descriptors (and few of
		1545	them active), of course, but this was not subject of this benchmark.
		1546
		1547	The C<Event> module has a relatively high setup and callback invocation
		1548	cost, but overall scores in on the third place.
		1549
		1550	C<Glib>'s memory usage is quite a bit higher, but it features a
		1551	faster callback invocation and overall ends up in the same class as
		1552	C<Event>. However, Glib scales extremely badly, doubling the number of
		1553	watchers increases the processing time by more than a factor of four,
		1554	making it completely unusable when using larger numbers of watchers
		1555	(note that only a single file descriptor was used in the benchmark, so
		1556	inefficiencies of C<poll> do not account for this).
		1557
		1558	The C<Tk> adaptor works relatively well. The fact that it crashes with
		1559	more than 2000 watchers is a big setback, however, as correctness takes
		1560	precedence over speed. Nevertheless, its performance is surprising, as the
		1561	file descriptor is dup()ed for each watcher. This shows that the dup()
		1562	employed by some adaptors is not a big performance issue (it does incur a
		1563	hidden memory cost inside the kernel which is not reflected in the figures
		1564	above).
		1565
		1566	C<POE>, regardless of underlying event loop (whether using its pure perl
		1567	select-based backend or the Event module, the POE-EV backend couldn't
		1568	be tested because it wasn't working) shows abysmal performance and
		1569	memory usage with AnyEvent: Watchers use almost 30 times as much memory
		1570	as EV watchers, and 10 times as much memory as Event (the high memory
		1571	requirements are caused by requiring a session for each watcher). Watcher
		1572	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		1573	implementation.
		1574
		1575	The design of the POE adaptor class in AnyEvent can not really account
		1576	for the performance issues, though, as session creation overhead is
		1577	small compared to execution of the state machine, which is coded pretty
		1578	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		1579	using multiple sessions is not a good approach, especially regarding
		1580	memory usage, even the author of POE could not come up with a faster
		1581	design).
		1582
		1583	=head3 Summary
		1584
		1585	=over 4
		1586
		1587	=item * Using EV through AnyEvent is faster than any other event loop
		1588	(even when used without AnyEvent), but most event loops have acceptable
		1589	performance with or without AnyEvent.
		1590
		1591	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
		1592	the actual event loop, only with extremely fast event loops such as EV
		1593	adds AnyEvent significant overhead.
		1594
		1595	=item * You should avoid POE like the plague if you want performance or
		1596	reasonable memory usage.
		1597
		1598	=back
		1599
		1600	=head2 BENCHMARKING THE LARGE SERVER CASE
		1601
		1602	This benchmark actually benchmarks the event loop itself. It works by
		1603	creating a number of "servers": each server consists of a socket pair, a
		1604	timeout watcher that gets reset on activity (but never fires), and an I/O
		1605	watcher waiting for input on one side of the socket. Each time the socket
		1606	watcher reads a byte it will write that byte to a random other "server".
		1607
		1608	The effect is that there will be a lot of I/O watchers, only part of which
		1609	are active at any one point (so there is a constant number of active
		1610	fds for each loop iteration, but which fds these are is random). The
		1611	timeout is reset each time something is read because that reflects how
		1612	most timeouts work (and puts extra pressure on the event loops).
		1613
		1614	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		1615	(1%) are active. This mirrors the activity of large servers with many
		1616	connections, most of which are idle at any one point in time.
		1617
		1618	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		1619	distribution.
		1620
		1621	=head3 Explanation of the columns
		1622
		1623	I<sockets> is the number of sockets, and twice the number of "servers" (as
		1624	each server has a read and write socket end).
		1625
		1626	I<create> is the time it takes to create a socket pair (which is
		1627	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		1628
		1629	I<request>, the most important value, is the time it takes to handle a
		1630	single "request", that is, reading the token from the pipe and forwarding
		1631	it to another server. This includes deleting the old timeout and creating
		1632	a new one that moves the timeout into the future.
		1633
		1634	=head3 Results
		1635
		1636	name sockets create request
		1637	EV 20000 69.01 11.16
		1638	Perl 20000 73.32 35.87
		1639	Event 20000 212.62 257.32
		1640	Glib 20000 651.16 1896.30
		1641	POE 20000 349.67 12317.24 uses POE::Loop::Event
		1642
		1643	=head3 Discussion
		1644
		1645	This benchmark I<does> measure scalability and overall performance of the
		1646	particular event loop.
		1647
		1648	EV is again fastest. Since it is using epoll on my system, the setup time
		1649	is relatively high, though.
		1650
		1651	Perl surprisingly comes second. It is much faster than the C-based event
		1652	loops Event and Glib.
		1653
		1654	Event suffers from high setup time as well (look at its code and you will
		1655	understand why). Callback invocation also has a high overhead compared to
		1656	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		1657	uses select or poll in basically all documented configurations.
		1658
		1659	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		1660	clearly fails to perform with many filehandles or in busy servers.
		1661
		1662	POE is still completely out of the picture, taking over 1000 times as long
		1663	as EV, and over 100 times as long as the Perl implementation, even though
		1664	it uses a C-based event loop in this case.
		1665
		1666	=head3 Summary
		1667
		1668	=over 4
		1669
		1670	=item * The pure perl implementation performs extremely well.
		1671
		1672	=item * Avoid Glib or POE in large projects where performance matters.
		1673
		1674	=back
		1675
		1676	=head2 BENCHMARKING SMALL SERVERS
		1677
		1678	While event loops should scale (and select-based ones do not...) even to
		1679	large servers, most programs we (or I :) actually write have only a few
		1680	I/O watchers.
		1681
		1682	In this benchmark, I use the same benchmark program as in the large server
		1683	case, but it uses only eight "servers", of which three are active at any
		1684	one time. This should reflect performance for a small server relatively
		1685	well.
		1686
		1687	The columns are identical to the previous table.
		1688
		1689	=head3 Results
		1690
		1691	name sockets create request
		1692	EV 16 20.00 6.54
		1693	Perl 16 25.75 12.62
		1694	Event 16 81.27 35.86
		1695	Glib 16 32.63 15.48
		1696	POE 16 261.87 276.28 uses POE::Loop::Event
		1697
		1698	=head3 Discussion
		1699
		1700	The benchmark tries to test the performance of a typical small
		1701	server. While knowing how various event loops perform is interesting, keep
		1702	in mind that their overhead in this case is usually not as important, due
		1703	to the small absolute number of watchers (that is, you need efficiency and
		1704	speed most when you have lots of watchers, not when you only have a few of
		1705	them).
		1706
		1707	EV is again fastest.
		1708
		1709	Perl again comes second. It is noticeably faster than the C-based event
		1710	loops Event and Glib, although the difference is too small to really
		1711	matter.
		1712
		1713	POE also performs much better in this case, but is is still far behind the
		1714	others.
		1715
		1716	=head3 Summary
		1717
		1718	=over 4
		1719
		1720	=item * C-based event loops perform very well with small number of
		1721	watchers, as the management overhead dominates.
		1722
		1723	=back
		1724
		1725
		1726	=head1 FORK
		1727
		1728	Most event libraries are not fork-safe. The ones who are usually are
		1729	because they rely on inefficient but fork-safe C<select> or C<poll>
		1730	calls. Only L<EV> is fully fork-aware.
		1731
		1732	If you have to fork, you must either do so I<before> creating your first
		1733	watcher OR you must not use AnyEvent at all in the child.
		1734
		1735
		1736	=head1 SECURITY CONSIDERATIONS
		1737
		1738	AnyEvent can be forced to load any event model via
		1739	$ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used to
		1740	execute arbitrary code or directly gain access, it can easily be used to
		1741	make the program hang or malfunction in subtle ways, as AnyEvent watchers
		1742	will not be active when the program uses a different event model than
		1743	specified in the variable.
		1744
		1745	You can make AnyEvent completely ignore this variable by deleting it
		1746	before the first watcher gets created, e.g. with a C<BEGIN> block:
		1747
		1748	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
		1749
		1750	use AnyEvent;
		1751
		1752	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		1753	be used to probe what backend is used and gain other information (which is
		1754	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		1755	$ENV{PERL_ANYEGENT_STRICT}.
		1756
		1757
		1758	=head1 BUGS
		1759
		1760	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		1761	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		1762	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		1763	mamleaks, such as leaking on C<map> and C<grep> but it is usually not as
		1764	pronounced).
		1765
607		1766
608	=head1 SEE ALSO	1767	=head1 SEE ALSO
609		1768
610	Event modules: L<Coro::Event>, L<Coro>, L<Event>, L<Glib::Event>, L<Glib>.	1769	Utility functions: L<AnyEvent::Util>.
611		1770
612	Implementations: L<AnyEvent::Impl::Coro>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>.	1771	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
		1772	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
613		1773
614	Nontrivial usage example: L<Net::FCP>.	1774	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
		1775	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
		1776	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
		1777	L<AnyEvent::Impl::POE>.
615		1778
616	=head1	1779	Non-blocking file handles, sockets, TCP clients and
		1780	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>.
		1781
		1782	Asynchronous DNS: L<AnyEvent::DNS>.
		1783
		1784	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>,
		1785
		1786	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>.
		1787
		1788
		1789	=head1 AUTHOR
		1790
		1791	Marc Lehmann <schmorp@schmorp.de>
		1792	http://home.schmorp.de/
617		1793
618	=cut	1794	=cut
619		1795
620	1	1796	1
621		1797

Diff Legend

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