ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.81 by root, Fri Apr 25 13:05:17 2008 UTC vs.
Revision 1.177 by root, Thu Aug 21 18:45:16 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	EV, Event, Coro::EV, Coro::Event, Glib, Tk, Perl, Event::Lib, Qt, POE - 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 whether 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.
22		37
23	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)	38	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
24		39
25	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen	40	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
26	nowadays. So what is different about AnyEvent?	41	nowadays. So what is different about AnyEvent?
27		42
28	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of	43	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
29	policy> and AnyEvent is I<small and efficient>.	44	policy> and AnyEvent is I<small and efficient>.
30		45
31	First and foremost, I<AnyEvent is not an event model> itself, it only	46	First and foremost, I<AnyEvent is not an event model> itself, it only
32	interfaces to whatever event model the main program happens to use in a	47	interfaces to whatever event model the main program happens to use, in a
33	pragmatic way. For event models and certain classes of immortals alike,	48	pragmatic way. For event models and certain classes of immortals alike,
34	the statement "there can only be one" is a bitter reality: In general,	49	the statement "there can only be one" is a bitter reality: In general,
35	only one event loop can be active at the same time in a process. AnyEvent	50	only one event loop can be active at the same time in a process. AnyEvent
36	helps hiding the differences between those event loops.	51	cannot change this, but it can hide the differences between those event
		52	loops.
37		53
38	The goal of AnyEvent is to offer module authors the ability to do event	54	The goal of AnyEvent is to offer module authors the ability to do event
39	programming (waiting for I/O or timer events) without subscribing to a	55	programming (waiting for I/O or timer events) without subscribing to a
40	religion, a way of living, and most importantly: without forcing your	56	religion, a way of living, and most importantly: without forcing your
41	module users into the same thing by forcing them to use the same event	57	module users into the same thing by forcing them to use the same event
42	model you use.	58	model you use.
43		59
44	For modules like POE or IO::Async (which is a total misnomer as it is	60	For modules like POE or IO::Async (which is a total misnomer as it is
45	actually doing all I/O I<synchronously>...), using them in your module is	61	actually doing all I/O I<synchronously>...), using them in your module is
46	like joining a cult: After you joined, you are dependent on them and you	62	like joining a cult: After you joined, you are dependent on them and you
47	cannot use anything else, as it is simply incompatible to everything that	63	cannot use anything else, as they are simply incompatible to everything
48	isn't itself. What's worse, all the potential users of your module are	64	that isn't them. What's worse, all the potential users of your
49	I<also> forced to use the same event loop you use.	65	module are I<also> forced to use the same event loop you use.
50		66
51	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works	67	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
52	fine. AnyEvent + Tk works fine etc. etc. but none of these work together	68	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
53	with the rest: POE + IO::Async? no go. Tk + Event? no go. Again: if	69	with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
54	your module uses one of those, every user of your module has to use it,	70	your module uses one of those, every user of your module has to use it,
55	too. But if your module uses AnyEvent, it works transparently with all	71	too. But if your module uses AnyEvent, it works transparently with all
56	event models it supports (including stuff like POE and IO::Async, as long	72	event models it supports (including stuff like IO::Async, as long as those
57	as those use one of the supported event loops. It is trivial to add new	73	use one of the supported event loops. It is trivial to add new event loops
58	event loops to AnyEvent, too, so it is future-proof).	74	to AnyEvent, too, so it is future-proof).
59		75
60	In addition to being free of having to use I<the one and only true event	76	In addition to being free of having to use I<the one and only true event
61	model>, AnyEvent also is free of bloat and policy: with POE or similar	77	model>, AnyEvent also is free of bloat and policy: with POE or similar
62	modules, you get an enourmous amount of code and strict rules you have to	78	modules, you get an enormous amount of code and strict rules you have to
63	follow. AnyEvent, on the other hand, is lean and up to the point, by only	79	follow. AnyEvent, on the other hand, is lean and up to the point, by only
64	offering the functionality that is necessary, in as thin as a wrapper as	80	offering the functionality that is necessary, in as thin as a wrapper as
65	technically possible.	81	technically possible.
66		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
67	Of course, if you want lots of policy (this can arguably be somewhat	89	Now, if you I<do want> lots of policy (this can arguably be somewhat
68	useful) and you want to force your users to use the one and only event	90	useful) and you want to force your users to use the one and only event
69	model, you should I<not> use this module.	91	model, you should I<not> use this module.
70
71		92
72	=head1 DESCRIPTION	93	=head1 DESCRIPTION
73		94
74	L<AnyEvent> provides an identical interface to multiple event loops. This	95	L<AnyEvent> provides an identical interface to multiple event loops. This
75	allows module authors to utilise an event loop without forcing module	96	allows module authors to utilise an event loop without forcing module
…		…
79	The interface itself is vaguely similar, but not identical to the L<Event>	100	The interface itself is vaguely similar, but not identical to the L<Event>
80	module.	101	module.
81		102
82	During the first call of any watcher-creation method, the module tries	103	During the first call of any watcher-creation method, the module tries
83	to detect the currently loaded event loop by probing whether one of the	104	to detect the currently loaded event loop by probing whether one of the
84	following modules is already loaded: L<Coro::EV>, L<Coro::Event>, L<EV>,	105	following modules is already loaded: L<EV>,
85	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,	106	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,
86	L<POE>. The first one found is used. If none are found, the module tries	107	L<POE>. The first one found is used. If none are found, the module tries
87	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl	108	to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
88	adaptor should always succeed) in the order given. The first one that can	109	adaptor should always succeed) in the order given. The first one that can
89	be successfully loaded will be used. If, after this, still none could be	110	be successfully loaded will be used. If, after this, still none could be
…		…
103	starts using it, all bets are off. Maybe you should tell their authors to	124	starts using it, all bets are off. Maybe you should tell their authors to
104	use AnyEvent so their modules work together with others seamlessly...	125	use AnyEvent so their modules work together with others seamlessly...
105		126
106	The pure-perl implementation of AnyEvent is called	127	The pure-perl implementation of AnyEvent is called
107	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
108	explicitly.	129	explicitly and enjoy the high availability of that event loop :)
109		130
110	=head1 WATCHERS	131	=head1 WATCHERS
111		132
112	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
113	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
114	the callback to call, the filehandle to watch, etc.	135	the callback to call, the file handle to watch, etc.
115		136
116	These watchers are normal Perl objects with normal Perl lifetime. After	137	These watchers are normal Perl objects with normal Perl lifetime. After
117	creating a watcher it will immediately "watch" for events and invoke the	138	creating a watcher it will immediately "watch" for events and invoke the
118	callback when the event occurs (of course, only when the event model	139	callback when the event occurs (of course, only when the event model
119	is in control).	140	is in control).
…		…
127	Many watchers either are used with "recursion" (repeating timers for	148	Many watchers either are used with "recursion" (repeating timers for
128	example), or need to refer to their watcher object in other ways.	149	example), or need to refer to their watcher object in other ways.
129		150
130	An any way to achieve that is this pattern:	151	An any way to achieve that is this pattern:
131		152
132	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {	153	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
133	# you can use $w here, for example to undef it	154	# you can use $w here, for example to undef it
134	undef $w;	155	undef $w;
135	});	156	});
136		157
137	Note that C<my $w; $w => combination. This is necessary because in Perl,	158	Note that C<my $w; $w => combination. This is necessary because in Perl,
138	my variables are only visible after the statement in which they are	159	my variables are only visible after the statement in which they are
139	declared.	160	declared.
140		161
141	=head2 I/O WATCHERS	162	=head2 I/O WATCHERS
142		163
143	You can create an I/O watcher by calling the C<< AnyEvent->io >> method	164	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
144	with the following mandatory key-value pairs as arguments:	165	with the following mandatory key-value pairs as arguments:
145		166
146	C<fh> the Perl I<file handle> (I<not> file descriptor) 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>
147	events. C<poll> must be a string that is either C<r> or C<w>, which	169	must be a string that is either C<r> or C<w>, which creates a watcher
148	creates a watcher waiting for "r"eadable or "w"ritable events,	170	waiting for "r"eadable or "w"ritable events, respectively. C<cb> is the
149	respectively. C<cb> is the callback to invoke each time the file handle	171	callback to invoke each time the file handle becomes ready.
150	becomes ready.
151		172
152	As long as the I/O watcher exists it will keep the file descriptor or a	173	Although the callback might get passed parameters, their value and
153	copy of it alive/open.	174	presence is undefined and you cannot rely on them. Portable AnyEvent
		175	callbacks cannot use arguments passed to I/O watcher callbacks.
154		176
		177	The I/O watcher might use the underlying file descriptor or a copy of it.
155	It is not allowed to close a file handle as long as any watcher is active	178	You must not close a file handle as long as any watcher is active on the
156	on the underlying file descriptor.	179	underlying file descriptor.
157		180
158	Some event loops issue spurious readyness notifications, so you should	181	Some event loops issue spurious readyness notifications, so you should
159	always use non-blocking calls when reading/writing from/to your file	182	always use non-blocking calls when reading/writing from/to your file
160	handles.	183	handles.
161		184
162	Example:
163
164	# 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
165	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {	188	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
166	chomp (my $input = <STDIN>);	189	chomp (my $input = <STDIN>);
167	warn "read: $input\n";	190	warn "read: $input\n";
168	undef $w;	191	undef $w;
169	});	192	});
…		…
172		195
173	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 >>
174	method with the following mandatory arguments:	197	method with the following mandatory arguments:
175		198
176	C<after> specifies after how many seconds (fractional values are	199	C<after> specifies after how many seconds (fractional values are
177	supported) should the timer activate. C<cb> the callback to invoke in that	200	supported) the callback should be invoked. C<cb> is the callback to invoke
178	case.	201	in that case.
179		202
180	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
181	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
182	and Glib).	205	callbacks cannot use arguments passed to time watcher callbacks.
183		206
184	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.
185		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
186	# fire an event after 7.7 seconds	217	Example: fire an event after 7.7 seconds.
		218
187	my $w = AnyEvent->timer (after => 7.7, cb => sub {	219	my $w = AnyEvent->timer (after => 7.7, cb => sub {
188	warn "timeout\n";	220	warn "timeout\n";
189	});	221	});
190		222
191	# to cancel the timer:	223	# to cancel the timer:
192	undef $w;	224	undef $w;
193		225
194	Example 2:
195
196	# fire an event after 0.5 seconds, then roughly every second	226	Example 2: fire an event after 0.5 seconds, then roughly every second.
197	my $w;
198		227
199	my $cb = sub {
200	# cancel the old timer while creating a new one
201	$w = AnyEvent->timer (after => 1, cb => $cb);	228	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
		229	warn "timeout\n";
202	};	230	};
203
204	# start the "loop" by creating the first watcher
205	$w = AnyEvent->timer (after => 0.5, cb => $cb);
206		231
207	=head3 TIMING ISSUES	232	=head3 TIMING ISSUES
208		233
209	There are two ways to handle timers: based on real time (relative, "fire	234	There are two ways to handle timers: based on real time (relative, "fire
210	in 10 seconds") and based on wallclock time (absolute, "fire at 12	235	in 10 seconds") and based on wallclock time (absolute, "fire at 12
…		…
222	timers.	247	timers.
223		248
224	AnyEvent always prefers relative timers, if available, matching the	249	AnyEvent always prefers relative timers, if available, matching the
225	AnyEvent API.	250	AnyEvent API.
226		251
		252	AnyEvent has two additional methods that return the "current time":
		253
		254	=over 4
		255
		256	=item AnyEvent->time
		257
		258	This returns the "current wallclock time" as a fractional number of
		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).
		261
		262	It progresses independently of any event loop processing, i.e. each call
		263	will check the system clock, which usually gets updated frequently.
		264
		265	=item AnyEvent->now
		266
		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.
		271
		272	I<In almost all cases (in all cases if you don't care), this is the
		273	function to call when you want to know the current time.>
		274
		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
227	=head2 SIGNAL WATCHERS	315	=head2 SIGNAL WATCHERS
228		316
229	You can watch for signals using a signal watcher, C<signal> is the signal	317	You can watch for signals using a signal watcher, C<signal> is the signal
230	I<name> without any C<SIG> prefix, C<cb> is the Perl callback to	318	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
231	be invoked whenever a signal occurs.	319	callback to be invoked whenever a signal occurs.
232		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
233	Multiple signal occurances can be clumped together into one callback	325	Multiple signal occurrences can be clumped together into one callback
234	invocation, and callback invocation will be synchronous. synchronous means	326	invocation, and callback invocation will be synchronous. Synchronous means
235	that it might take a while until the signal gets handled by the process,	327	that it might take a while until the signal gets handled by the process,
236	but it is guarenteed not to interrupt any other callbacks.	328	but it is guaranteed not to interrupt any other callbacks.
237		329
238	The main advantage of using these watchers is that you can share a signal	330	The main advantage of using these watchers is that you can share a signal
239	between multiple watchers.	331	between multiple watchers.
240		332
241	This watcher might use C<%SIG>, so programs overwriting those signals	333	This watcher might use C<%SIG>, so programs overwriting those signals
…		…
251		343
252	The child process is specified by the C<pid> argument (if set to C<0>, it	344	The child process is specified by the C<pid> argument (if set to C<0>, it
253	watches for any child process exit). The watcher will trigger as often	345	watches for any child process exit). The watcher will trigger as often
254	as status change for the child are received. This works by installing a	346	as status change for the child are received. This works by installing a
255	signal handler for C<SIGCHLD>. The callback will be called with the pid	347	signal handler for C<SIGCHLD>. The callback will be called with the pid
256	and exit status (as returned by waitpid).	348	and exit status (as returned by waitpid), so unlike other watcher types,
		349	you I<can> rely on child watcher callback arguments.
257		350
258	Example: wait for pid 1333	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).
259		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
260	my $w = AnyEvent->child (	369	my $w = AnyEvent->child (
261	pid => 1333,	370	pid => $pid,
262	cb => sub {	371	cb => sub {
263	my ($pid, $status) = @_;	372	my ($pid, $status) = @_;
264	warn "pid $pid exited with status $status";	373	warn "pid $pid exited with status $status";
		374	$done->send;
265	},	375	},
266	);	376	);
		377
		378	# do something else, then wait for process exit
		379	$done->recv;
267		380
268	=head2 CONDITION VARIABLES	381	=head2 CONDITION VARIABLES
269		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
270	Condition variables can be created by calling the C<< AnyEvent->condvar >>	393	Condition variables can be created by calling the C<< AnyEvent->condvar
271	method without any arguments.	394	>> method, usually without arguments. The only argument pair allowed is
272		395
273	A condition variable waits for a condition - precisely that the C<<	396	C<cb>, which specifies a callback to be called when the condition variable
274	->broadcast >> method has been called.	397	becomes true, with the condition variable as the first argument (but not
		398	the results).
275		399
276	They are very useful to signal that a condition has been fulfilled, for	400	After creation, the condition variable is "false" until it becomes "true"
		401	by calling the C<send> method (or calling the condition variable as if it
		402	were a callback, read about the caveats in the description for the C<<
		403	->send >> method).
		404
		405	Condition variables are similar to callbacks, except that you can
		406	optionally wait for them. They can also be called merge points - points
		407	in time where multiple outstanding events have been processed. And yet
		408	another way to call them is transactions - each condition variable can be
		409	used to represent a transaction, which finishes at some point and delivers
		410	a result.
		411
		412	Condition variables are very useful to signal that something has finished,
277	example, if you write a module that does asynchronous http requests,	413	for example, if you write a module that does asynchronous http requests,
278	then a condition variable would be the ideal candidate to signal the	414	then a condition variable would be the ideal candidate to signal the
279	availability of results.	415	availability of results. The user can either act when the callback is
		416	called or can synchronously C<< ->recv >> for the results.
280		417
281	You can also use condition variables to block your main program until	418	You can also use them to simulate traditional event loops - for example,
282	an event occurs - for example, you could C<< ->wait >> in your main	419	you can block your main program until an event occurs - for example, you
283	program until the user clicks the Quit button in your app, which would C<<	420	could C<< ->recv >> in your main program until the user clicks the Quit
284	->broadcast >> the "quit" event.	421	button of your app, which would C<< ->send >> the "quit" event.
285		422
286	Note that condition variables recurse into the event loop - if you have	423	Note that condition variables recurse into the event loop - if you have
287	two pirces of code that call C<< ->wait >> in a round-robbin fashion, you	424	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
288	lose. Therefore, condition variables are good to export to your caller, but	425	lose. Therefore, condition variables are good to export to your caller, but
289	you should avoid making a blocking wait yourself, at least in callbacks,	426	you should avoid making a blocking wait yourself, at least in callbacks,
290	as this asks for trouble.	427	as this asks for trouble.
291		428
292	This object has two methods:	429	Condition variables are represented by hash refs in perl, and the keys
		430	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		431	easy (it is often useful to build your own transaction class on top of
		432	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		433	it's C<new> method in your own C<new> method.
		434
		435	There are two "sides" to a condition variable - the "producer side" which
		436	eventually calls C<< -> send >>, and the "consumer side", which waits
		437	for the send to occur.
		438
		439	Example: wait for a timer.
		440
		441	# wait till the result is ready
		442	my $result_ready = AnyEvent->condvar;
		443
		444	# do something such as adding a timer
		445	# or socket watcher the calls $result_ready->send
		446	# when the "result" is ready.
		447	# in this case, we simply use a timer:
		448	my $w = AnyEvent->timer (
		449	after => 1,
		450	cb => sub { $result_ready->send },
		451	);
		452
		453	# this "blocks" (while handling events) till the callback
		454	# calls send
		455	$result_ready->recv;
		456
		457	Example: wait for a timer, but take advantage of the fact that
		458	condition variables are also code references.
		459
		460	my $done = AnyEvent->condvar;
		461	my $delay = AnyEvent->timer (after => 5, cb => $done);
		462	$done->recv;
		463
		464	Example: Imagine an API that returns a condvar and doesn't support
		465	callbacks. This is how you make a synchronous call, for example from
		466	the main program:
		467
		468	use AnyEvent::CouchDB;
		469
		470	...
		471
		472	my @info = $couchdb->info->recv;
		473
		474	And this is how you would just ste a callback to be called whenever the
		475	results are available:
		476
		477	$couchdb->info->cb (sub {
		478	my @info = $_[0]->recv;
		479	});
		480
		481	=head3 METHODS FOR PRODUCERS
		482
		483	These methods should only be used by the producing side, i.e. the
		484	code/module that eventually sends the signal. Note that it is also
		485	the producer side which creates the condvar in most cases, but it isn't
		486	uncommon for the consumer to create it as well.
293		487
294	=over 4	488	=over 4
295		489
		490	=item $cv->send (...)
		491
		492	Flag the condition as ready - a running C<< ->recv >> and all further
		493	calls to C<recv> will (eventually) return after this method has been
		494	called. If nobody is waiting the send will be remembered.
		495
		496	If a callback has been set on the condition variable, it is called
		497	immediately from within send.
		498
		499	Any arguments passed to the C<send> call will be returned by all
		500	future C<< ->recv >> calls.
		501
		502	Condition variables are overloaded so one can call them directly
		503	(as a code reference). Calling them directly is the same as calling
		504	C<send>. Note, however, that many C-based event loops do not handle
		505	overloading, so as tempting as it may be, passing a condition variable
		506	instead of a callback does not work. Both the pure perl and EV loops
		507	support overloading, however, as well as all functions that use perl to
		508	invoke a callback (as in L<AnyEvent::Socket> and L<AnyEvent::DNS> for
		509	example).
		510
		511	=item $cv->croak ($error)
		512
		513	Similar to send, but causes all call's to C<< ->recv >> to invoke
		514	C<Carp::croak> with the given error message/object/scalar.
		515
		516	This can be used to signal any errors to the condition variable
		517	user/consumer.
		518
		519	=item $cv->begin ([group callback])
		520
296	=item $cv->wait	521	=item $cv->end
297		522
298	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	523	These two methods are EXPERIMENTAL and MIGHT CHANGE.
		524
		525	These two methods can be used to combine many transactions/events into
		526	one. For example, a function that pings many hosts in parallel might want
		527	to use a condition variable for the whole process.
		528
		529	Every call to C<< ->begin >> will increment a counter, and every call to
		530	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		531	>>, the (last) callback passed to C<begin> will be executed. That callback
		532	is I<supposed> to call C<< ->send >>, but that is not required. If no
		533	callback was set, C<send> will be called without any arguments.
		534
		535	Let's clarify this with the ping example:
		536
		537	my $cv = AnyEvent->condvar;
		538
		539	my %result;
		540	$cv->begin (sub { $cv->send (\%result) });
		541
		542	for my $host (@list_of_hosts) {
		543	$cv->begin;
		544	ping_host_then_call_callback $host, sub {
		545	$result{$host} = ...;
		546	$cv->end;
		547	};
		548	}
		549
		550	$cv->end;
		551
		552	This code fragment supposedly pings a number of hosts and calls
		553	C<send> after results for all then have have been gathered - in any
		554	order. To achieve this, the code issues a call to C<begin> when it starts
		555	each ping request and calls C<end> when it has received some result for
		556	it. Since C<begin> and C<end> only maintain a counter, the order in which
		557	results arrive is not relevant.
		558
		559	There is an additional bracketing call to C<begin> and C<end> outside the
		560	loop, which serves two important purposes: first, it sets the callback
		561	to be called once the counter reaches C<0>, and second, it ensures that
		562	C<send> is called even when C<no> hosts are being pinged (the loop
		563	doesn't execute once).
		564
		565	This is the general pattern when you "fan out" into multiple subrequests:
		566	use an outer C<begin>/C<end> pair to set the callback and ensure C<end>
		567	is called at least once, and then, for each subrequest you start, call
		568	C<begin> and for each subrequest you finish, call C<end>.
		569
		570	=back
		571
		572	=head3 METHODS FOR CONSUMERS
		573
		574	These methods should only be used by the consuming side, i.e. the
		575	code awaits the condition.
		576
		577	=over 4
		578
		579	=item $cv->recv
		580
		581	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
299	called on c<$cv>, while servicing other watchers normally.	582	>> methods have been called on c<$cv>, while servicing other watchers
		583	normally.
300		584
301	You can only wait once on a condition - additional calls will return	585	You can only wait once on a condition - additional calls are valid but
302	immediately.	586	will return immediately.
		587
		588	If an error condition has been set by calling C<< ->croak >>, then this
		589	function will call C<croak>.
		590
		591	In list context, all parameters passed to C<send> will be returned,
		592	in scalar context only the first one will be returned.
303		593
304	Not all event models support a blocking wait - some die in that case	594	Not all event models support a blocking wait - some die in that case
305	(programs might want to do that to stay interactive), so I<if you are	595	(programs might want to do that to stay interactive), so I<if you are
306	using this from a module, never require a blocking wait>, but let the	596	using this from a module, never require a blocking wait>, but let the
307	caller decide whether the call will block or not (for example, by coupling	597	caller decide whether the call will block or not (for example, by coupling
308	condition variables with some kind of request results and supporting	598	condition variables with some kind of request results and supporting
309	callbacks so the caller knows that getting the result will not block,	599	callbacks so the caller knows that getting the result will not block,
310	while still suppporting blocking waits if the caller so desires).	600	while still supporting blocking waits if the caller so desires).
311		601
312	Another reason I<never> to C<< ->wait >> in a module is that you cannot	602	Another reason I<never> to C<< ->recv >> in a module is that you cannot
313	sensibly have two C<< ->wait >>'s in parallel, as that would require	603	sensibly have two C<< ->recv >>'s in parallel, as that would require
314	multiple interpreters or coroutines/threads, none of which C<AnyEvent>	604	multiple interpreters or coroutines/threads, none of which C<AnyEvent>
315	can supply (the coroutine-aware backends L<AnyEvent::Impl::CoroEV> and	605	can supply.
316	L<AnyEvent::Impl::CoroEvent> explicitly support concurrent C<< ->wait >>'s
317	from different coroutines, however).
318		606
319	=item $cv->broadcast	607	The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
		608	fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
		609	versions and also integrates coroutines into AnyEvent, making blocking
		610	C<< ->recv >> calls perfectly safe as long as they are done from another
		611	coroutine (one that doesn't run the event loop).
320		612
321	Flag the condition as ready - a running C<< ->wait >> and all further	613	You can ensure that C<< -recv >> never blocks by setting a callback and
322	calls to C<wait> will (eventually) return after this method has been	614	only calling C<< ->recv >> from within that callback (or at a later
323	called. If nobody is waiting the broadcast will be remembered..	615	time). This will work even when the event loop does not support blocking
		616	waits otherwise.
		617
		618	=item $bool = $cv->ready
		619
		620	Returns true when the condition is "true", i.e. whether C<send> or
		621	C<croak> have been called.
		622
		623	=item $cb = $cv->cb ($cb->($cv))
		624
		625	This is a mutator function that returns the callback set and optionally
		626	replaces it before doing so.
		627
		628	The callback will be called when the condition becomes "true", i.e. when
		629	C<send> or C<croak> are called, with the only argument being the condition
		630	variable itself. Calling C<recv> inside the callback or at any later time
		631	is guaranteed not to block.
324		632
325	=back	633	=back
326
327	Example:
328
329	# wait till the result is ready
330	my $result_ready = AnyEvent->condvar;
331
332	# do something such as adding a timer
333	# or socket watcher the calls $result_ready->broadcast
334	# when the "result" is ready.
335	# in this case, we simply use a timer:
336	my $w = AnyEvent->timer (
337	after => 1,
338	cb => sub { $result_ready->broadcast },
339	);
340
341	# this "blocks" (while handling events) till the watcher
342	# calls broadcast
343	$result_ready->wait;
344		634
345	=head1 GLOBAL VARIABLES AND FUNCTIONS	635	=head1 GLOBAL VARIABLES AND FUNCTIONS
346		636
347	=over 4	637	=over 4
348		638
…		…
354	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	644	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case
355	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	645	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).
356		646
357	The known classes so far are:	647	The known classes so far are:
358		648
359	AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
360	AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
361	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).	649	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
362	AnyEvent::Impl::Event based on Event, second best choice.	650	AnyEvent::Impl::Event based on Event, second best choice.
		651	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
363	AnyEvent::Impl::Glib based on Glib, third-best choice.	652	AnyEvent::Impl::Glib based on Glib, third-best choice.
364	AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
365	AnyEvent::Impl::Tk based on Tk, very bad choice.	653	AnyEvent::Impl::Tk based on Tk, very bad choice.
366	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).	654	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
367	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.	655	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
368	AnyEvent::Impl::POE based on POE, not generic enough for full support.	656	AnyEvent::Impl::POE based on POE, not generic enough for full support.
369		657
…		…
382	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model	670	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
383	if necessary. You should only call this function right before you would	671	if necessary. You should only call this function right before you would
384	have created an AnyEvent watcher anyway, that is, as late as possible at	672	have created an AnyEvent watcher anyway, that is, as late as possible at
385	runtime.	673	runtime.
386		674
		675	=item $guard = AnyEvent::post_detect { BLOCK }
		676
		677	Arranges for the code block to be executed as soon as the event model is
		678	autodetected (or immediately if this has already happened).
		679
		680	If called in scalar or list context, then it creates and returns an object
		681	that automatically removes the callback again when it is destroyed. See
		682	L<Coro::BDB> for a case where this is useful.
		683
		684	=item @AnyEvent::post_detect
		685
		686	If there are any code references in this array (you can C<push> to it
		687	before or after loading AnyEvent), then they will called directly after
		688	the event loop has been chosen.
		689
		690	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		691	if it contains a true value then the event loop has already been detected,
		692	and the array will be ignored.
		693
		694	Best use C<AnyEvent::post_detect { BLOCK }> instead.
		695
387	=back	696	=back
388		697
389	=head1 WHAT TO DO IN A MODULE	698	=head1 WHAT TO DO IN A MODULE
390		699
391	As a module author, you should C<use AnyEvent> and call AnyEvent methods	700	As a module author, you should C<use AnyEvent> and call AnyEvent methods
…		…
394	Be careful when you create watchers in the module body - AnyEvent will	703	Be careful when you create watchers in the module body - AnyEvent will
395	decide which event module to use as soon as the first method is called, so	704	decide which event module to use as soon as the first method is called, so
396	by calling AnyEvent in your module body you force the user of your module	705	by calling AnyEvent in your module body you force the user of your module
397	to load the event module first.	706	to load the event module first.
398		707
399	Never call C<< ->wait >> on a condition variable unless you I<know> that	708	Never call C<< ->recv >> on a condition variable unless you I<know> that
400	the C<< ->broadcast >> method has been called on it already. This is	709	the C<< ->send >> method has been called on it already. This is
401	because it will stall the whole program, and the whole point of using	710	because it will stall the whole program, and the whole point of using
402	events is to stay interactive.	711	events is to stay interactive.
403		712
404	It is fine, however, to call C<< ->wait >> when the user of your module	713	It is fine, however, to call C<< ->recv >> when the user of your module
405	requests it (i.e. if you create a http request object ad have a method	714	requests it (i.e. if you create a http request object ad have a method
406	called C<results> that returns the results, it should call C<< ->wait >>	715	called C<results> that returns the results, it should call C<< ->recv >>
407	freely, as the user of your module knows what she is doing. always).	716	freely, as the user of your module knows what she is doing. always).
408		717
409	=head1 WHAT TO DO IN THE MAIN PROGRAM	718	=head1 WHAT TO DO IN THE MAIN PROGRAM
410		719
411	There will always be a single main program - the only place that should	720	There will always be a single main program - the only place that should
…		…
413		722
414	If it doesn't care, it can just "use AnyEvent" and use it itself, or not	723	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
415	do anything special (it does not need to be event-based) and let AnyEvent	724	do anything special (it does not need to be event-based) and let AnyEvent
416	decide which implementation to chose if some module relies on it.	725	decide which implementation to chose if some module relies on it.
417		726
418	If the main program relies on a specific event model. For example, in	727	If the main program relies on a specific event model - for example, in
419	Gtk2 programs you have to rely on the Glib module. You should load the	728	Gtk2 programs you have to rely on the Glib module - you should load the
420	event module before loading AnyEvent or any module that uses it: generally	729	event module before loading AnyEvent or any module that uses it: generally
421	speaking, you should load it as early as possible. The reason is that	730	speaking, you should load it as early as possible. The reason is that
422	modules might create watchers when they are loaded, and AnyEvent will	731	modules might create watchers when they are loaded, and AnyEvent will
423	decide on the event model to use as soon as it creates watchers, and it	732	decide on the event model to use as soon as it creates watchers, and it
424	might chose the wrong one unless you load the correct one yourself.	733	might chose the wrong one unless you load the correct one yourself.
425		734
426	You can chose to use a rather inefficient pure-perl implementation by	735	You can chose to use a pure-perl implementation by loading the
427	loading the C<AnyEvent::Impl::Perl> module, which gives you similar	736	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
428	behaviour everywhere, but letting AnyEvent chose is generally better.	737	everywhere, but letting AnyEvent chose the model is generally better.
		738
		739	=head2 MAINLOOP EMULATION
		740
		741	Sometimes (often for short test scripts, or even standalone programs who
		742	only want to use AnyEvent), you do not want to run a specific event loop.
		743
		744	In that case, you can use a condition variable like this:
		745
		746	AnyEvent->condvar->recv;
		747
		748	This has the effect of entering the event loop and looping forever.
		749
		750	Note that usually your program has some exit condition, in which case
		751	it is better to use the "traditional" approach of storing a condition
		752	variable somewhere, waiting for it, and sending it when the program should
		753	exit cleanly.
		754
		755
		756	=head1 OTHER MODULES
		757
		758	The following is a non-exhaustive list of additional modules that use
		759	AnyEvent and can therefore be mixed easily with other AnyEvent modules
		760	in the same program. Some of the modules come with AnyEvent, some are
		761	available via CPAN.
		762
		763	=over 4
		764
		765	=item L<AnyEvent::Util>
		766
		767	Contains various utility functions that replace often-used but blocking
		768	functions such as C<inet_aton> by event-/callback-based versions.
		769
		770	=item L<AnyEvent::Socket>
		771
		772	Provides various utility functions for (internet protocol) sockets,
		773	addresses and name resolution. Also functions to create non-blocking tcp
		774	connections or tcp servers, with IPv6 and SRV record support and more.
		775
		776	=item L<AnyEvent::Handle>
		777
		778	Provide read and write buffers, manages watchers for reads and writes,
		779	supports raw and formatted I/O, I/O queued and fully transparent and
		780	non-blocking SSL/TLS.
		781
		782	=item L<AnyEvent::DNS>
		783
		784	Provides rich asynchronous DNS resolver capabilities.
		785
		786	=item L<AnyEvent::HTTP>
		787
		788	A simple-to-use HTTP library that is capable of making a lot of concurrent
		789	HTTP requests.
		790
		791	=item L<AnyEvent::HTTPD>
		792
		793	Provides a simple web application server framework.
		794
		795	=item L<AnyEvent::FastPing>
		796
		797	The fastest ping in the west.
		798
		799	=item L<AnyEvent::DBI>
		800
		801	Executes L<DBI> requests asynchronously in a proxy process.
		802
		803	=item L<AnyEvent::AIO>
		804
		805	Truly asynchronous I/O, should be in the toolbox of every event
		806	programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
		807	together.
		808
		809	=item L<AnyEvent::BDB>
		810
		811	Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
		812	L<BDB> and AnyEvent together.
		813
		814	=item L<AnyEvent::GPSD>
		815
		816	A non-blocking interface to gpsd, a daemon delivering GPS information.
		817
		818	=item L<AnyEvent::IGS>
		819
		820	A non-blocking interface to the Internet Go Server protocol (used by
		821	L<App::IGS>).
		822
		823	=item L<Net::IRC3>
		824
		825	AnyEvent based IRC client module family.
		826
		827	=item L<Net::XMPP2>
		828
		829	AnyEvent based XMPP (Jabber protocol) module family.
		830
		831	=item L<Net::FCP>
		832
		833	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		834	of AnyEvent.
		835
		836	=item L<Event::ExecFlow>
		837
		838	High level API for event-based execution flow control.
		839
		840	=item L<Coro>
		841
		842	Has special support for AnyEvent via L<Coro::AnyEvent>.
		843
		844	=item L<IO::Lambda>
		845
		846	The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
		847
		848	=back
429		849
430	=cut	850	=cut
431		851
432	package AnyEvent;	852	package AnyEvent;
433		853
434	no warnings;	854	no warnings;
435	use strict;	855	use strict;
436		856
437	use Carp;	857	use Carp;
438		858
439	our $VERSION = '3.3';	859	our $VERSION = 4.232;
440	our $MODEL;	860	our $MODEL;
441		861
442	our $AUTOLOAD;	862	our $AUTOLOAD;
443	our @ISA;	863	our @ISA;
444		864
		865	our @REGISTRY;
		866
		867	our $WIN32;
		868
		869	BEGIN {
		870	my $win32 = ! ! ($^O =~ /mswin32/i);
		871	eval "sub WIN32(){ $win32 }";
		872	}
		873
445	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;	874	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
446		875
447	our @REGISTRY;	876	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		877
		878	{
		879	my $idx;
		880	$PROTOCOL{$_} = ++$idx
		881	for reverse split /\s*,\s*/,
		882	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		883	}
448		884
449	my @models = (	885	my @models = (
450	[Coro::EV:: => AnyEvent::Impl::CoroEV::],
451	[Coro::Event:: => AnyEvent::Impl::CoroEvent::],
452	[EV:: => AnyEvent::Impl::EV::],	886	[EV:: => AnyEvent::Impl::EV::],
453	[Event:: => AnyEvent::Impl::Event::],	887	[Event:: => AnyEvent::Impl::Event::],
454	[Glib:: => AnyEvent::Impl::Glib::],
455	[Tk:: => AnyEvent::Impl::Tk::],
456	[Wx:: => AnyEvent::Impl::POE::],
457	[Prima:: => AnyEvent::Impl::POE::],
458	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	888	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
459	# everything below here will not be autoprobed as the pureperl backend should work everywhere	889	# everything below here will not be autoprobed
		890	# as the pureperl backend should work everywhere
		891	# and is usually faster
		892	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		893	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
460	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy	894	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
461	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program	895	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
462	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza	896	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
		897	[Wx:: => AnyEvent::Impl::POE::],
		898	[Prima:: => AnyEvent::Impl::POE::],
463	);	899	);
464		900
465	our %method = map +($_ => 1), qw(io timer signal child condvar broadcast wait one_event DESTROY);	901	our %method = map +($_ => 1), qw(io timer time now signal child condvar one_event DESTROY);
		902
		903	our @post_detect;
		904
		905	sub post_detect(&) {
		906	my ($cb) = @_;
		907
		908	if ($MODEL) {
		909	$cb->();
		910
		911	1
		912	} else {
		913	push @post_detect, $cb;
		914
		915	defined wantarray
		916	? bless \$cb, "AnyEvent::Util::PostDetect"
		917	: ()
		918	}
		919	}
		920
		921	sub AnyEvent::Util::PostDetect::DESTROY {
		922	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		923	}
466		924
467	sub detect() {	925	sub detect() {
468	unless ($MODEL) {	926	unless ($MODEL) {
469	no strict 'refs';	927	no strict 'refs';
		928	local $SIG{__DIE__};
470		929
471	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {	930	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
472	my $model = "AnyEvent::Impl::$1";	931	my $model = "AnyEvent::Impl::$1";
473	if (eval "require $model") {	932	if (eval "require $model") {
474	$MODEL = $model;	933	$MODEL = $model;
…		…
504	last;	963	last;
505	}	964	}
506	}	965	}
507		966
508	$MODEL	967	$MODEL
509	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV (or Coro+EV), Event (or Coro+Event) or Glib.";	968	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.";
510	}	969	}
511	}	970	}
512		971
		972	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
		973
513	unshift @ISA, $MODEL;	974	unshift @ISA, $MODEL;
514	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	975
		976	require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
		977
		978	(shift @post_detect)->() while @post_detect;
515	}	979	}
516		980
517	$MODEL	981	$MODEL
518	}	982	}
519		983
…		…
527		991
528	my $class = shift;	992	my $class = shift;
529	$class->$func (@_);	993	$class->$func (@_);
530	}	994	}
531		995
		996	# utility function to dup a filehandle. this is used by many backends
		997	# to support binding more than one watcher per filehandle (they usually
		998	# allow only one watcher per fd, so we dup it to get a different one).
		999	sub _dupfh($$$$) {
		1000	my ($poll, $fh, $r, $w) = @_;
		1001
		1002	require Fcntl;
		1003
		1004	# cygwin requires the fh mode to be matching, unix doesn't
		1005	my ($rw, $mode) = $poll eq "r" ? ($r, "<")
		1006	: $poll eq "w" ? ($w, ">")
		1007	: Carp::croak "AnyEvent->io requires poll set to either 'r' or 'w'";
		1008
		1009	open my $fh2, "$mode&" . fileno $fh
		1010	or die "cannot dup() filehandle: $!";
		1011
		1012	# we assume CLOEXEC is already set by perl in all important cases
		1013
		1014	($fh2, $rw)
		1015	}
		1016
532	package AnyEvent::Base;	1017	package AnyEvent::Base;
533		1018
		1019	# default implementation for now and time
		1020
		1021	use Time::HiRes ();
		1022
		1023	sub time { Time::HiRes::time }
		1024	sub now { Time::HiRes::time }
		1025
534	# default implementation for ->condvar, ->wait, ->broadcast	1026	# default implementation for ->condvar
535		1027
536	sub condvar {	1028	sub condvar {
537	bless \my $flag, "AnyEvent::Base::CondVar"	1029	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar::
538	}
539
540	sub AnyEvent::Base::CondVar::broadcast {
541	${$_[0]}++;
542	}
543
544	sub AnyEvent::Base::CondVar::wait {
545	AnyEvent->one_event while !${$_[0]};
546	}	1030	}
547		1031
548	# default implementation for ->signal	1032	# default implementation for ->signal
549		1033
550	our %SIG_CB;	1034	our %SIG_CB;
…		…
566	sub AnyEvent::Base::Signal::DESTROY {	1050	sub AnyEvent::Base::Signal::DESTROY {
567	my ($signal, $cb) = @{$_[0]};	1051	my ($signal, $cb) = @{$_[0]};
568		1052
569	delete $SIG_CB{$signal}{$cb};	1053	delete $SIG_CB{$signal}{$cb};
570		1054
571	$SIG{$signal} = 'DEFAULT' unless keys %{ $SIG_CB{$signal} };	1055	delete $SIG{$signal} unless keys %{ $SIG_CB{$signal} };
572	}	1056	}
573		1057
574	# default implementation for ->child	1058	# default implementation for ->child
575		1059
576	our %PID_CB;	1060	our %PID_CB;
…		…
603	or Carp::croak "required option 'pid' is missing";	1087	or Carp::croak "required option 'pid' is missing";
604		1088
605	$PID_CB{$pid}{$arg{cb}} = $arg{cb};	1089	$PID_CB{$pid}{$arg{cb}} = $arg{cb};
606		1090
607	unless ($WNOHANG) {	1091	unless ($WNOHANG) {
608	$WNOHANG = eval { require POSIX; &POSIX::WNOHANG } || 1;	1092	$WNOHANG = eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
609	}	1093	}
610		1094
611	unless ($CHLD_W) {	1095	unless ($CHLD_W) {
612	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);	1096	$CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
613	# child could be a zombie already, so make at least one round	1097	# child could be a zombie already, so make at least one round
…		…
623	delete $PID_CB{$pid}{$cb};	1107	delete $PID_CB{$pid}{$cb};
624	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };	1108	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
625		1109
626	undef $CHLD_W unless keys %PID_CB;	1110	undef $CHLD_W unless keys %PID_CB;
627	}	1111	}
		1112
		1113	package AnyEvent::CondVar;
		1114
		1115	our @ISA = AnyEvent::CondVar::Base::;
		1116
		1117	package AnyEvent::CondVar::Base;
		1118
		1119	use overload
		1120	'&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1121	fallback => 1;
		1122
		1123	sub _send {
		1124	# nop
		1125	}
		1126
		1127	sub send {
		1128	my $cv = shift;
		1129	$cv->{_ae_sent} = [@_];
		1130	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1131	$cv->_send;
		1132	}
		1133
		1134	sub croak {
		1135	$_[0]{_ae_croak} = $_[1];
		1136	$_[0]->send;
		1137	}
		1138
		1139	sub ready {
		1140	$_[0]{_ae_sent}
		1141	}
		1142
		1143	sub _wait {
		1144	AnyEvent->one_event while !$_[0]{_ae_sent};
		1145	}
		1146
		1147	sub recv {
		1148	$_[0]->_wait;
		1149
		1150	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		1151	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		1152	}
		1153
		1154	sub cb {
		1155	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		1156	$_[0]{_ae_cb}
		1157	}
		1158
		1159	sub begin {
		1160	++$_[0]{_ae_counter};
		1161	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		1162	}
		1163
		1164	sub end {
		1165	return if --$_[0]{_ae_counter};
		1166	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		1167	}
		1168
		1169	# undocumented/compatibility with pre-3.4
		1170	*broadcast = \&send;
		1171	*wait = \&_wait;
628		1172
629	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1173	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
630		1174
631	This is an advanced topic that you do not normally need to use AnyEvent in	1175	This is an advanced topic that you do not normally need to use AnyEvent in
632	a module. This section is only of use to event loop authors who want to	1176	a module. This section is only of use to event loop authors who want to
…		…
686	C<PERL_ANYEVENT_MODEL>.	1230	C<PERL_ANYEVENT_MODEL>.
687		1231
688	When set to C<2> or higher, cause AnyEvent to report to STDERR which event	1232	When set to C<2> or higher, cause AnyEvent to report to STDERR which event
689	model it chooses.	1233	model it chooses.
690		1234
		1235	=item C<PERL_ANYEVENT_STRICT>
		1236
		1237	AnyEvent does not do much argument checking by default, as thorough
		1238	argument checking is very costly. Setting this variable to a true value
		1239	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		1240	check the arguments passed to most method calls. If it finds any problems
		1241	it will croak.
		1242
		1243	In other words, enables "strict" mode.
		1244
		1245	Unlike C<use strict> it is definitely recommended ot keep it off in
		1246	production.
		1247
691	=item C<PERL_ANYEVENT_MODEL>	1248	=item C<PERL_ANYEVENT_MODEL>
692		1249
693	This can be used to specify the event model to be used by AnyEvent, before	1250	This can be used to specify the event model to be used by AnyEvent, before
694	autodetection and -probing kicks in. It must be a string consisting	1251	auto detection and -probing kicks in. It must be a string consisting
695	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended	1252	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
696	and the resulting module name is loaded and if the load was successful,	1253	and the resulting module name is loaded and if the load was successful,
697	used as event model. If it fails to load AnyEvent will proceed with	1254	used as event model. If it fails to load AnyEvent will proceed with
698	autodetection and -probing.	1255	auto detection and -probing.
699		1256
700	This functionality might change in future versions.	1257	This functionality might change in future versions.
701		1258
702	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you	1259	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
703	could start your program like this:	1260	could start your program like this:
704		1261
705	PERL_ANYEVENT_MODEL=Perl perl ...	1262	PERL_ANYEVENT_MODEL=Perl perl ...
		1263
		1264	=item C<PERL_ANYEVENT_PROTOCOLS>
		1265
		1266	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1267	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1268	of auto probing).
		1269
		1270	Must be set to a comma-separated list of protocols or address families,
		1271	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1272	used, and preference will be given to protocols mentioned earlier in the
		1273	list.
		1274
		1275	This variable can effectively be used for denial-of-service attacks
		1276	against local programs (e.g. when setuid), although the impact is likely
		1277	small, as the program has to handle connection errors already-
		1278
		1279	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1280	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1281	- only support IPv4, never try to resolve or contact IPv6
		1282	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1283	IPv6, but prefer IPv6 over IPv4.
		1284
		1285	=item C<PERL_ANYEVENT_EDNS0>
		1286
		1287	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1288	for DNS. This extension is generally useful to reduce DNS traffic, but
		1289	some (broken) firewalls drop such DNS packets, which is why it is off by
		1290	default.
		1291
		1292	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1293	EDNS0 in its DNS requests.
		1294
		1295	=item C<PERL_ANYEVENT_MAX_FORKS>
		1296
		1297	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		1298	will create in parallel.
706		1299
707	=back	1300	=back
708		1301
709	=head1 EXAMPLE PROGRAM	1302	=head1 EXAMPLE PROGRAM
710		1303
…		…
721	poll => 'r',	1314	poll => 'r',
722	cb => sub {	1315	cb => sub {
723	warn "io event <$_[0]>\n"; # will always output <r>	1316	warn "io event <$_[0]>\n"; # will always output <r>
724	chomp (my $input = <STDIN>); # read a line	1317	chomp (my $input = <STDIN>); # read a line
725	warn "read: $input\n"; # output what has been read	1318	warn "read: $input\n"; # output what has been read
726	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1319	$cv->send if $input =~ /^q/i; # quit program if /^q/i
727	},	1320	},
728	);	1321	);
729		1322
730	my $time_watcher; # can only be used once	1323	my $time_watcher; # can only be used once
731		1324
…		…
736	});	1329	});
737	}	1330	}
738		1331
739	new_timer; # create first timer	1332	new_timer; # create first timer
740		1333
741	$cv->wait; # wait until user enters /^q/i	1334	$cv->recv; # wait until user enters /^q/i
742		1335
743	=head1 REAL-WORLD EXAMPLE	1336	=head1 REAL-WORLD EXAMPLE
744		1337
745	Consider the L<Net::FCP> module. It features (among others) the following	1338	Consider the L<Net::FCP> module. It features (among others) the following
746	API calls, which are to freenet what HTTP GET requests are to http:	1339	API calls, which are to freenet what HTTP GET requests are to http:
…		…
796	syswrite $txn->{fh}, $txn->{request}	1389	syswrite $txn->{fh}, $txn->{request}
797	or die "connection or write error";	1390	or die "connection or write error";
798	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	1391	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
799		1392
800	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	1393	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
801	result and signals any possible waiters that the request ahs finished:	1394	result and signals any possible waiters that the request has finished:
802		1395
803	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1396	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
804		1397
805	if (end-of-file or data complete) {	1398	if (end-of-file or data complete) {
806	$txn->{result} = $txn->{buf};	1399	$txn->{result} = $txn->{buf};
807	$txn->{finished}->broadcast;	1400	$txn->{finished}->send;
808	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1401	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
809	}	1402	}
810		1403
811	The C<result> method, finally, just waits for the finished signal (if the	1404	The C<result> method, finally, just waits for the finished signal (if the
812	request was already finished, it doesn't wait, of course, and returns the	1405	request was already finished, it doesn't wait, of course, and returns the
813	data:	1406	data:
814		1407
815	$txn->{finished}->wait;	1408	$txn->{finished}->recv;
816	return $txn->{result};	1409	return $txn->{result};
817		1410
818	The actual code goes further and collects all errors (C<die>s, exceptions)	1411	The actual code goes further and collects all errors (C<die>s, exceptions)
819	that occured during request processing. The C<result> method detects	1412	that occurred during request processing. The C<result> method detects
820	whether an exception as thrown (it is stored inside the $txn object)	1413	whether an exception as thrown (it is stored inside the $txn object)
821	and just throws the exception, which means connection errors and other	1414	and just throws the exception, which means connection errors and other
822	problems get reported tot he code that tries to use the result, not in a	1415	problems get reported tot he code that tries to use the result, not in a
823	random callback.	1416	random callback.
824		1417
…		…
855		1448
856	my $quit = AnyEvent->condvar;	1449	my $quit = AnyEvent->condvar;
857		1450
858	$fcp->txn_client_get ($url)->cb (sub {	1451	$fcp->txn_client_get ($url)->cb (sub {
859	...	1452	...
860	$quit->broadcast;	1453	$quit->send;
861	});	1454	});
862		1455
863	$quit->wait;	1456	$quit->recv;
864		1457
865		1458
866	=head1 BENCHMARK	1459	=head1 BENCHMARKS
867		1460
868	To give you an idea of the performance and overheads that AnyEvent adds	1461	To give you an idea of the performance and overheads that AnyEvent adds
869	over the event loops themselves (and to give you an impression of the	1462	over the event loops themselves and to give you an impression of the speed
870	speed of various event loops), here is a benchmark of various supported	1463	of various event loops I prepared some benchmarks.
871	event models natively and with anyevent. The benchmark creates a lot of	1464
872	timers (with a zero timeout) and I/O watchers (watching STDOUT, a pty, to	1465	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1466
		1467	Here is a benchmark of various supported event models used natively and
		1468	through AnyEvent. The benchmark creates a lot of timers (with a zero
		1469	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
873	become writable, which it is), lets them fire exactly once and destroys	1470	which it is), lets them fire exactly once and destroys them again.
874	them again.
875		1471
876	Rewriting the benchmark to use many different sockets instead of using	1472	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
877	the same filehandle for all I/O watchers results in a much longer runtime	1473	distribution.
878	(socket creation is expensive), but qualitatively the same figures, so it
879	was not used.
880		1474
881	=head2 Explanation of the columns	1475	=head3 Explanation of the columns
882		1476
883	I<watcher> is the number of event watchers created/destroyed. Since	1477	I<watcher> is the number of event watchers created/destroyed. Since
884	different event models feature vastly different performances, each event	1478	different event models feature vastly different performances, each event
885	loop was given a number of watchers so that overall runtime is acceptable	1479	loop was given a number of watchers so that overall runtime is acceptable
886	and similar between tested event loop (and keep them from crashing): Glib	1480	and similar between tested event loop (and keep them from crashing): Glib
…		…
896	all watchers, to avoid adding memory overhead. That means closure creation	1490	all watchers, to avoid adding memory overhead. That means closure creation
897	and memory usage is not included in the figures.	1491	and memory usage is not included in the figures.
898		1492
899	I<invoke> is the time, in microseconds, used to invoke a simple	1493	I<invoke> is the time, in microseconds, used to invoke a simple
900	callback. The callback simply counts down a Perl variable and after it was	1494	callback. The callback simply counts down a Perl variable and after it was
901	invoked "watcher" times, it would C<< ->broadcast >> a condvar once to	1495	invoked "watcher" times, it would C<< ->send >> a condvar once to
902	signal the end of this phase.	1496	signal the end of this phase.
903		1497
904	I<destroy> is the time, in microseconds, that it takes to destroy a single	1498	I<destroy> is the time, in microseconds, that it takes to destroy a single
905	watcher.	1499	watcher.
906		1500
907	=head2 Results	1501	=head3 Results
908		1502
909	name watchers bytes create invoke destroy comment	1503	name watchers bytes create invoke destroy comment
910	EV/EV 400000 244 0.56 0.46 0.31 EV native interface	1504	EV/EV 400000 244 0.56 0.46 0.31 EV native interface
911	EV/Any 100000 610 3.52 0.91 0.75 EV + AnyEvent watchers	1505	EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers
912	CoroEV/Any 100000 610 3.49 0.92 0.75 coroutines + Coro::Signal	1506	CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal
913	Perl/Any 100000 513 4.91 0.92 1.15 pure perl implementation	1507	Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation
914	Event/Event 16000 523 28.05 21.38 0.86 Event native interface	1508	Event/Event 16000 516 31.88 31.30 0.85 Event native interface
915	Event/Any 16000 943 34.43 20.48 1.39 Event + AnyEvent watchers	1509	Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers
916	Glib/Any 16000 1357 96.99 12.55 55.51 quadratic behaviour	1510	Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour
917	Tk/Any 2000 1855 27.01 66.61 14.03 SEGV with >> 2000 watchers	1511	Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers
918	POE/Event 2000 6644 108.15 768.19 14.33 via POE::Loop::Event	1512	POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event
919	POE/Select 2000 6343 94.69 807.65 562.69 via POE::Loop::Select	1513	POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select
920		1514
921	=head2 Discussion	1515	=head3 Discussion
922		1516
923	The benchmark does I<not> measure scalability of the event loop very	1517	The benchmark does I<not> measure scalability of the event loop very
924	well. For example, a select-based event loop (such as the pure perl one)	1518	well. For example, a select-based event loop (such as the pure perl one)
925	can never compete with an event loop that uses epoll when the number of	1519	can never compete with an event loop that uses epoll when the number of
926	file descriptors grows high. In this benchmark, all events become ready at	1520	file descriptors grows high. In this benchmark, all events become ready at
927	the same time, so select/poll-based implementations get an unnatural speed	1521	the same time, so select/poll-based implementations get an unnatural speed
928	boost.	1522	boost.
929		1523
		1524	Also, note that the number of watchers usually has a nonlinear effect on
		1525	overall speed, that is, creating twice as many watchers doesn't take twice
		1526	the time - usually it takes longer. This puts event loops tested with a
		1527	higher number of watchers at a disadvantage.
		1528
		1529	To put the range of results into perspective, consider that on the
		1530	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1531	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1532	cycles with POE.
		1533
930	C<EV> is the sole leader regarding speed and memory use, which are both	1534	C<EV> is the sole leader regarding speed and memory use, which are both
931	maximal/minimal, respectively. Even when going through AnyEvent, there are	1535	maximal/minimal, respectively. Even when going through AnyEvent, it uses
932	only two event loops that use slightly less memory (the C<Event> module	1536	far less memory than any other event loop and is still faster than Event
933	natively and the pure perl backend), and no faster event models, not even	1537	natively.
934	C<Event> natively.
935		1538
936	The pure perl implementation is hit in a few sweet spots (both the	1539	The pure perl implementation is hit in a few sweet spots (both the
937	zero timeout and the use of a single fd hit optimisations in the perl	1540	constant timeout and the use of a single fd hit optimisations in the perl
938	interpreter and the backend itself, and all watchers become ready at the	1541	interpreter and the backend itself). Nevertheless this shows that it
939	same time). Nevertheless this shows that it adds very little overhead in	1542	adds very little overhead in itself. Like any select-based backend its
940	itself. Like any select-based backend its performance becomes really bad	1543	performance becomes really bad with lots of file descriptors (and few of
941	with lots of file descriptors (and few of them active), of course, but	1544	them active), of course, but this was not subject of this benchmark.
942	this was not subject of this benchmark.
943		1545
944	The C<Event> module has a relatively high setup and callback invocation cost,	1546	The C<Event> module has a relatively high setup and callback invocation
945	but overall scores on the third place.	1547	cost, but overall scores in on the third place.
946		1548
947	C<Glib>'s memory usage is quite a bit bit higher, but it features a	1549	C<Glib>'s memory usage is quite a bit higher, but it features a
948	faster callback invocation and overall ends up in the same class as	1550	faster callback invocation and overall ends up in the same class as
949	C<Event>. However, Glib scales extremely badly, doubling the number of	1551	C<Event>. However, Glib scales extremely badly, doubling the number of
950	watchers increases the processing time by more than a factor of four,	1552	watchers increases the processing time by more than a factor of four,
951	making it completely unusable when using larger numbers of watchers	1553	making it completely unusable when using larger numbers of watchers
952	(note that only a single file descriptor was used in the benchmark, so	1554	(note that only a single file descriptor was used in the benchmark, so
…		…
955	The C<Tk> adaptor works relatively well. The fact that it crashes with	1557	The C<Tk> adaptor works relatively well. The fact that it crashes with
956	more than 2000 watchers is a big setback, however, as correctness takes	1558	more than 2000 watchers is a big setback, however, as correctness takes
957	precedence over speed. Nevertheless, its performance is surprising, as the	1559	precedence over speed. Nevertheless, its performance is surprising, as the
958	file descriptor is dup()ed for each watcher. This shows that the dup()	1560	file descriptor is dup()ed for each watcher. This shows that the dup()
959	employed by some adaptors is not a big performance issue (it does incur a	1561	employed by some adaptors is not a big performance issue (it does incur a
960	hidden memory cost inside the kernel, though, that is not reflected in the	1562	hidden memory cost inside the kernel which is not reflected in the figures
961	figures above).	1563	above).
962		1564
963	C<POE>, regardless of underlying event loop (wether using its pure perl	1565	C<POE>, regardless of underlying event loop (whether using its pure perl
964	select-based backend or the Event module) shows abysmal performance and	1566	select-based backend or the Event module, the POE-EV backend couldn't
		1567	be tested because it wasn't working) shows abysmal performance and
965	memory usage: Watchers use almost 30 times as much memory as EV watchers,	1568	memory usage with AnyEvent: Watchers use almost 30 times as much memory
966	and 10 times as much memory as both Event or EV via AnyEvent. Watcher	1569	as EV watchers, and 10 times as much memory as Event (the high memory
		1570	requirements are caused by requiring a session for each watcher). Watcher
967	invocation is almost 900 times slower than with AnyEvent's pure perl	1571	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		1572	implementation.
		1573
968	implementation. The design of the POE adaptor class in AnyEvent can not	1574	The design of the POE adaptor class in AnyEvent can not really account
969	really account for this, as session creation overhead is small compared	1575	for the performance issues, though, as session creation overhead is
970	to execution of the state machine, which is coded pretty optimally within	1576	small compared to execution of the state machine, which is coded pretty
971	L<AnyEvent::Impl::POE>. POE simply seems to be abysmally slow.	1577	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		1578	using multiple sessions is not a good approach, especially regarding
		1579	memory usage, even the author of POE could not come up with a faster
		1580	design).
972		1581
973	=head2 Summary	1582	=head3 Summary
974		1583
		1584	=over 4
		1585
975	Using EV through AnyEvent is faster than any other event loop, but most	1586	=item * Using EV through AnyEvent is faster than any other event loop
976	event loops have acceptable performance with or without AnyEvent.	1587	(even when used without AnyEvent), but most event loops have acceptable
		1588	performance with or without AnyEvent.
977		1589
978	The overhead AnyEvent adds is usually much smaller than the overhead of	1590	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
979	the actual event loop, only with extremely fast event loops such as the EV	1591	the actual event loop, only with extremely fast event loops such as EV
980	adds AnyEvent significant overhead.	1592	adds AnyEvent significant overhead.
981		1593
982	And you should simply avoid POE like the plague if you want performance or	1594	=item * You should avoid POE like the plague if you want performance or
983	reasonable memory usage.	1595	reasonable memory usage.
984		1596
		1597	=back
		1598
		1599	=head2 BENCHMARKING THE LARGE SERVER CASE
		1600
		1601	This benchmark actually benchmarks the event loop itself. It works by
		1602	creating a number of "servers": each server consists of a socket pair, a
		1603	timeout watcher that gets reset on activity (but never fires), and an I/O
		1604	watcher waiting for input on one side of the socket. Each time the socket
		1605	watcher reads a byte it will write that byte to a random other "server".
		1606
		1607	The effect is that there will be a lot of I/O watchers, only part of which
		1608	are active at any one point (so there is a constant number of active
		1609	fds for each loop iteration, but which fds these are is random). The
		1610	timeout is reset each time something is read because that reflects how
		1611	most timeouts work (and puts extra pressure on the event loops).
		1612
		1613	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		1614	(1%) are active. This mirrors the activity of large servers with many
		1615	connections, most of which are idle at any one point in time.
		1616
		1617	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		1618	distribution.
		1619
		1620	=head3 Explanation of the columns
		1621
		1622	I<sockets> is the number of sockets, and twice the number of "servers" (as
		1623	each server has a read and write socket end).
		1624
		1625	I<create> is the time it takes to create a socket pair (which is
		1626	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		1627
		1628	I<request>, the most important value, is the time it takes to handle a
		1629	single "request", that is, reading the token from the pipe and forwarding
		1630	it to another server. This includes deleting the old timeout and creating
		1631	a new one that moves the timeout into the future.
		1632
		1633	=head3 Results
		1634
		1635	name sockets create request
		1636	EV 20000 69.01 11.16
		1637	Perl 20000 73.32 35.87
		1638	Event 20000 212.62 257.32
		1639	Glib 20000 651.16 1896.30
		1640	POE 20000 349.67 12317.24 uses POE::Loop::Event
		1641
		1642	=head3 Discussion
		1643
		1644	This benchmark I<does> measure scalability and overall performance of the
		1645	particular event loop.
		1646
		1647	EV is again fastest. Since it is using epoll on my system, the setup time
		1648	is relatively high, though.
		1649
		1650	Perl surprisingly comes second. It is much faster than the C-based event
		1651	loops Event and Glib.
		1652
		1653	Event suffers from high setup time as well (look at its code and you will
		1654	understand why). Callback invocation also has a high overhead compared to
		1655	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		1656	uses select or poll in basically all documented configurations.
		1657
		1658	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		1659	clearly fails to perform with many filehandles or in busy servers.
		1660
		1661	POE is still completely out of the picture, taking over 1000 times as long
		1662	as EV, and over 100 times as long as the Perl implementation, even though
		1663	it uses a C-based event loop in this case.
		1664
		1665	=head3 Summary
		1666
		1667	=over 4
		1668
		1669	=item * The pure perl implementation performs extremely well.
		1670
		1671	=item * Avoid Glib or POE in large projects where performance matters.
		1672
		1673	=back
		1674
		1675	=head2 BENCHMARKING SMALL SERVERS
		1676
		1677	While event loops should scale (and select-based ones do not...) even to
		1678	large servers, most programs we (or I :) actually write have only a few
		1679	I/O watchers.
		1680
		1681	In this benchmark, I use the same benchmark program as in the large server
		1682	case, but it uses only eight "servers", of which three are active at any
		1683	one time. This should reflect performance for a small server relatively
		1684	well.
		1685
		1686	The columns are identical to the previous table.
		1687
		1688	=head3 Results
		1689
		1690	name sockets create request
		1691	EV 16 20.00 6.54
		1692	Perl 16 25.75 12.62
		1693	Event 16 81.27 35.86
		1694	Glib 16 32.63 15.48
		1695	POE 16 261.87 276.28 uses POE::Loop::Event
		1696
		1697	=head3 Discussion
		1698
		1699	The benchmark tries to test the performance of a typical small
		1700	server. While knowing how various event loops perform is interesting, keep
		1701	in mind that their overhead in this case is usually not as important, due
		1702	to the small absolute number of watchers (that is, you need efficiency and
		1703	speed most when you have lots of watchers, not when you only have a few of
		1704	them).
		1705
		1706	EV is again fastest.
		1707
		1708	Perl again comes second. It is noticeably faster than the C-based event
		1709	loops Event and Glib, although the difference is too small to really
		1710	matter.
		1711
		1712	POE also performs much better in this case, but is is still far behind the
		1713	others.
		1714
		1715	=head3 Summary
		1716
		1717	=over 4
		1718
		1719	=item * C-based event loops perform very well with small number of
		1720	watchers, as the management overhead dominates.
		1721
		1722	=back
		1723
985		1724
986	=head1 FORK	1725	=head1 FORK
987		1726
988	Most event libraries are not fork-safe. The ones who are usually are	1727	Most event libraries are not fork-safe. The ones who are usually are
989	because they are so inefficient. Only L<EV> is fully fork-aware.	1728	because they rely on inefficient but fork-safe C<select> or C<poll>
		1729	calls. Only L<EV> is fully fork-aware.
990		1730
991	If you have to fork, you must either do so I<before> creating your first	1731	If you have to fork, you must either do so I<before> creating your first
992	watcher OR you must not use AnyEvent at all in the child.	1732	watcher OR you must not use AnyEvent at all in the child.
993		1733
994		1734
…		…
1002	specified in the variable.	1742	specified in the variable.
1003		1743
1004	You can make AnyEvent completely ignore this variable by deleting it	1744	You can make AnyEvent completely ignore this variable by deleting it
1005	before the first watcher gets created, e.g. with a C<BEGIN> block:	1745	before the first watcher gets created, e.g. with a C<BEGIN> block:
1006		1746
1007	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }	1747	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
1008		1748
1009	use AnyEvent;	1749	use AnyEvent;
		1750
		1751	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		1752	be used to probe what backend is used and gain other information (which is
		1753	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		1754	$ENV{PERL_ANYEGENT_STRICT}.
		1755
		1756
		1757	=head1 BUGS
		1758
		1759	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		1760	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		1761	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		1762	mamleaks, such as leaking on C<map> and C<grep> but it is usually not as
		1763	pronounced).
1010		1764
1011		1765
1012	=head1 SEE ALSO	1766	=head1 SEE ALSO
1013		1767
1014	Event modules: L<Coro::EV>, L<EV>, L<EV::Glib>, L<Glib::EV>,	1768	Utility functions: L<AnyEvent::Util>.
1015	L<Coro::Event>, L<Event>, L<Glib::Event>, L<Glib>, L<Coro>, L<Tk>,	1769
		1770	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
1016	L<Event::Lib>, L<Qt>, L<POE>.	1771	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
1017		1772
1018	Implementations: L<AnyEvent::Impl::CoroEV>, L<AnyEvent::Impl::EV>,	1773	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
1019	L<AnyEvent::Impl::CoroEvent>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>,	1774	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
1020	L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>, L<AnyEvent::Impl::EventLib>,	1775	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
1021	L<AnyEvent::Impl::Qt>, L<AnyEvent::Impl::POE>.	1776	L<AnyEvent::Impl::POE>.
1022		1777
		1778	Non-blocking file handles, sockets, TCP clients and
		1779	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>.
		1780
		1781	Asynchronous DNS: L<AnyEvent::DNS>.
		1782
		1783	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>,
		1784
1023	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>.	1785	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>.
1024		1786
1025		1787
1026	=head1 AUTHOR	1788	=head1 AUTHOR
1027		1789
1028	Marc Lehmann <schmorp@schmorp.de>	1790	Marc Lehmann <schmorp@schmorp.de>
1029	http://home.schmorp.de/	1791	http://home.schmorp.de/
1030		1792
1031	=cut	1793	=cut
1032		1794
1033	1	1795	1
1034		1796

Diff Legend

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