[ViewVC] Diff of: cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.82 by root, Fri Apr 25 13:32:39 2008 UTC vs.
Revision 1.126 by root, Fri May 23 23:44:55 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
15	my $w = AnyEvent->timer (after => $seconds, cb => sub {	15	my $w = AnyEvent->timer (after => $seconds, cb => sub {
16	...	16	...
17	});	17	});
18		18
19	my $w = AnyEvent->condvar; # stores whether a condition was flagged	19	my $w = AnyEvent->condvar; # stores whether a condition was flagged
		20	$w->send; # wake up current and all future recv's
20	$w->wait; # enters "main loop" till $condvar gets ->broadcast	21	$w->recv; # enters "main loop" till $condvar gets ->send
21	$w->broadcast; # wake up current and all future wait's
22		22
23	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)	23	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
24		24
25	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen	25	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
26	nowadays. So what is different about AnyEvent?	26	nowadays. So what is different about AnyEvent?
…		…
66		66
67	Of course, if you want lots of policy (this can arguably be somewhat	67	Of course, if you 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	68	useful) and you want to force your users to use the one and only event
69	model, you should I<not> use this module.	69	model, you should I<not> use this module.
70		70
71
72	=head1 DESCRIPTION	71	=head1 DESCRIPTION
73		72
74	L<AnyEvent> provides an identical interface to multiple event loops. This	73	L<AnyEvent> provides an identical interface to multiple event loops. This
75	allows module authors to utilise an event loop without forcing module	74	allows module authors to utilise an event loop without forcing module
76	users to use the same event loop (as only a single event loop can coexist	75	users to use the same event loop (as only a single event loop can coexist
…		…
79	The interface itself is vaguely similar, but not identical to the L<Event>	78	The interface itself is vaguely similar, but not identical to the L<Event>
80	module.	79	module.
81		80
82	During the first call of any watcher-creation method, the module tries	81	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	82	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>,	83	following modules is already loaded: L<EV>,
85	L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,	84	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	85	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	86	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	87	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	88	be successfully loaded will be used. If, after this, still none could be
…		…
141	=head2 I/O WATCHERS	140	=head2 I/O WATCHERS
142		141
143	You can create an I/O watcher by calling the C<< AnyEvent->io >> method	142	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
144	with the following mandatory key-value pairs as arguments:	143	with the following mandatory key-value pairs as arguments:
145		144
146	C<fh> the Perl I<file handle> (I<not> file descriptor) to watch for	145	C<fh> the Perl I<file handle> (I<not> file descriptor) to watch
147	events. C<poll> must be a string that is either C<r> or C<w>, which	146	for events. C<poll> must be a string that is either C<r> or C<w>,
148	creates a watcher waiting for "r"eadable or "w"ritable events,	147	which creates a watcher waiting for "r"eadable or "w"ritable events,
149	respectively. C<cb> is the callback to invoke each time the file handle	148	respectively. C<cb> is the callback to invoke each time the file handle
150	becomes ready.	149	becomes ready.
151		150
		151	Although the callback might get passed parameters, their value and
		152	presence is undefined and you cannot rely on them. Portable AnyEvent
		153	callbacks cannot use arguments passed to I/O watcher callbacks.
		154
152	The I/O watcher might use the underlying file descriptor or a copy of it.	155	The I/O watcher might use the underlying file descriptor or a copy of it.
153	It is not allowed to close a file handle as long as any watcher is active	156	You must not close a file handle as long as any watcher is active on the
154	on the underlying file descriptor.	157	underlying file descriptor.
155		158
156	Some event loops issue spurious readyness notifications, so you should	159	Some event loops issue spurious readyness notifications, so you should
157	always use non-blocking calls when reading/writing from/to your file	160	always use non-blocking calls when reading/writing from/to your file
158	handles.	161	handles.
159		162
…		…
170		173
171	You can create a time watcher by calling the C<< AnyEvent->timer >>	174	You can create a time watcher by calling the C<< AnyEvent->timer >>
172	method with the following mandatory arguments:	175	method with the following mandatory arguments:
173		176
174	C<after> specifies after how many seconds (fractional values are	177	C<after> specifies after how many seconds (fractional values are
175	supported) should the timer activate. C<cb> the callback to invoke in that	178	supported) the callback should be invoked. C<cb> is the callback to invoke
176	case.	179	in that case.
		180
		181	Although the callback might get passed parameters, their value and
		182	presence is undefined and you cannot rely on them. Portable AnyEvent
		183	callbacks cannot use arguments passed to time watcher callbacks.
177		184
178	The timer callback will be invoked at most once: if you want a repeating	185	The timer callback will be invoked at most once: if you want a repeating
179	timer you have to create a new watcher (this is a limitation by both Tk	186	timer you have to create a new watcher (this is a limitation by both Tk
180	and Glib).	187	and Glib).
181		188
…		…
226		233
227	You can watch for signals using a signal watcher, C<signal> is the signal	234	You can watch for signals using a signal watcher, C<signal> is the signal
228	I<name> without any C<SIG> prefix, C<cb> is the Perl callback to	235	I<name> without any C<SIG> prefix, C<cb> is the Perl callback to
229	be invoked whenever a signal occurs.	236	be invoked whenever a signal occurs.
230		237
		238	Although the callback might get passed parameters, their value and
		239	presence is undefined and you cannot rely on them. Portable AnyEvent
		240	callbacks cannot use arguments passed to signal watcher callbacks.
		241
231	Multiple signal occurances can be clumped together into one callback	242	Multiple signal occurances can be clumped together into one callback
232	invocation, and callback invocation will be synchronous. synchronous means	243	invocation, and callback invocation will be synchronous. synchronous means
233	that it might take a while until the signal gets handled by the process,	244	that it might take a while until the signal gets handled by the process,
234	but it is guarenteed not to interrupt any other callbacks.	245	but it is guarenteed not to interrupt any other callbacks.
235		246
…		…
249		260
250	The child process is specified by the C<pid> argument (if set to C<0>, it	261	The child process is specified by the C<pid> argument (if set to C<0>, it
251	watches for any child process exit). The watcher will trigger as often	262	watches for any child process exit). The watcher will trigger as often
252	as status change for the child are received. This works by installing a	263	as status change for the child are received. This works by installing a
253	signal handler for C<SIGCHLD>. The callback will be called with the pid	264	signal handler for C<SIGCHLD>. The callback will be called with the pid
254	and exit status (as returned by waitpid).	265	and exit status (as returned by waitpid), so unlike other watcher types,
		266	you I<can> rely on child watcher callback arguments.
255		267
256	There is a slight catch to child watchers, however: you usually start them	268	There is a slight catch to child watchers, however: you usually start them
257	I<after> the child process was created, and this means the process could	269	I<after> the child process was created, and this means the process could
258	have exited already (and no SIGCHLD will be sent anymore).	270	have exited already (and no SIGCHLD will be sent anymore).
259		271
…		…
266	C<fork> the child (alternatively, you can call C<AnyEvent::detect>).	278	C<fork> the child (alternatively, you can call C<AnyEvent::detect>).
267		279
268	Example: fork a process and wait for it	280	Example: fork a process and wait for it
269		281
270	my $done = AnyEvent->condvar;	282	my $done = AnyEvent->condvar;
271
272	AnyEvent::detect; # force event module to be initialised
273		283
274	my $pid = fork or exit 5;	284	my $pid = fork or exit 5;
275		285
276	my $w = AnyEvent->child (	286	my $w = AnyEvent->child (
277	pid => $pid,	287	pid => $pid,
278	cb => sub {	288	cb => sub {
279	my ($pid, $status) = @_;	289	my ($pid, $status) = @_;
280	warn "pid $pid exited with status $status";	290	warn "pid $pid exited with status $status";
281	$done->broadcast;	291	$done->send;
282	},	292	},
283	);	293	);
284		294
285	# do something else, then wait for process exit	295	# do something else, then wait for process exit
286	$done->wait;	296	$done->recv;
287		297
288	=head2 CONDITION VARIABLES	298	=head2 CONDITION VARIABLES
289		299
		300	If you are familiar with some event loops you will know that all of them
		301	require you to run some blocking "loop", "run" or similar function that
		302	will actively watch for new events and call your callbacks.
		303
		304	AnyEvent is different, it expects somebody else to run the event loop and
		305	will only block when necessary (usually when told by the user).
		306
		307	The instrument to do that is called a "condition variable", so called
		308	because they represent a condition that must become true.
		309
290	Condition variables can be created by calling the C<< AnyEvent->condvar >>	310	Condition variables can be created by calling the C<< AnyEvent->condvar
291	method without any arguments.	311	>> method, usually without arguments. The only argument pair allowed is
		312	C<cb>, which specifies a callback to be called when the condition variable
		313	becomes true.
292		314
293	A condition variable waits for a condition - precisely that the C<<	315	After creation, the conditon variable is "false" until it becomes "true"
294	->broadcast >> method has been called.	316	by calling the C<send> method.
295		317
296	They are very useful to signal that a condition has been fulfilled, for	318	Condition variables are similar to callbacks, except that you can
		319	optionally wait for them. They can also be called merge points - points
		320	in time where multiple outstandign events have been processed. And yet
		321	another way to call them is transations - each condition variable can be
		322	used to represent a transaction, which finishes at some point and delivers
		323	a result.
		324
		325	Condition variables are very useful to signal that something has finished,
297	example, if you write a module that does asynchronous http requests,	326	for example, if you write a module that does asynchronous http requests,
298	then a condition variable would be the ideal candidate to signal the	327	then a condition variable would be the ideal candidate to signal the
299	availability of results.	328	availability of results. The user can either act when the callback is
		329	called or can synchronously C<< ->recv >> for the results.
300		330
301	You can also use condition variables to block your main program until	331	You can also use them to simulate traditional event loops - for example,
302	an event occurs - for example, you could C<< ->wait >> in your main	332	you can block your main program until an event occurs - for example, you
303	program until the user clicks the Quit button in your app, which would C<<	333	could C<< ->recv >> in your main program until the user clicks the Quit
304	->broadcast >> the "quit" event.	334	button of your app, which would C<< ->send >> the "quit" event.
305		335
306	Note that condition variables recurse into the event loop - if you have	336	Note that condition variables recurse into the event loop - if you have
307	two pirces of code that call C<< ->wait >> in a round-robbin fashion, you	337	two pieces of code that call C<< ->recv >> in a round-robbin fashion, you
308	lose. Therefore, condition variables are good to export to your caller, but	338	lose. Therefore, condition variables are good to export to your caller, but
309	you should avoid making a blocking wait yourself, at least in callbacks,	339	you should avoid making a blocking wait yourself, at least in callbacks,
310	as this asks for trouble.	340	as this asks for trouble.
311		341
312	This object has two methods:	342	Condition variables are represented by hash refs in perl, and the keys
		343	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		344	easy (it is often useful to build your own transaction class on top of
		345	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		346	it's C<new> method in your own C<new> method.
		347
		348	There are two "sides" to a condition variable - the "producer side" which
		349	eventually calls C<< -> send >>, and the "consumer side", which waits
		350	for the send to occur.
		351
		352	Example:
		353
		354	# wait till the result is ready
		355	my $result_ready = AnyEvent->condvar;
		356
		357	# do something such as adding a timer
		358	# or socket watcher the calls $result_ready->send
		359	# when the "result" is ready.
		360	# in this case, we simply use a timer:
		361	my $w = AnyEvent->timer (
		362	after => 1,
		363	cb => sub { $result_ready->send },
		364	);
		365
		366	# this "blocks" (while handling events) till the callback
		367	# calls send
		368	$result_ready->recv;
		369
		370	=head3 METHODS FOR PRODUCERS
		371
		372	These methods should only be used by the producing side, i.e. the
		373	code/module that eventually sends the signal. Note that it is also
		374	the producer side which creates the condvar in most cases, but it isn't
		375	uncommon for the consumer to create it as well.
313		376
314	=over 4	377	=over 4
315		378
		379	=item $cv->send (...)
		380
		381	Flag the condition as ready - a running C<< ->recv >> and all further
		382	calls to C<recv> will (eventually) return after this method has been
		383	called. If nobody is waiting the send will be remembered.
		384
		385	If a callback has been set on the condition variable, it is called
		386	immediately from within send.
		387
		388	Any arguments passed to the C<send> call will be returned by all
		389	future C<< ->recv >> calls.
		390
		391	=item $cv->croak ($error)
		392
		393	Similar to send, but causes all call's to C<< ->recv >> to invoke
		394	C<Carp::croak> with the given error message/object/scalar.
		395
		396	This can be used to signal any errors to the condition variable
		397	user/consumer.
		398
		399	=item $cv->begin ([group callback])
		400
316	=item $cv->wait	401	=item $cv->end
317		402
318	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	403	These two methods are EXPERIMENTAL and MIGHT CHANGE.
		404
		405	These two methods can be used to combine many transactions/events into
		406	one. For example, a function that pings many hosts in parallel might want
		407	to use a condition variable for the whole process.
		408
		409	Every call to C<< ->begin >> will increment a counter, and every call to
		410	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		411	>>, the (last) callback passed to C<begin> will be executed. That callback
		412	is I<supposed> to call C<< ->send >>, but that is not required. If no
		413	callback was set, C<send> will be called without any arguments.
		414
		415	Let's clarify this with the ping example:
		416
		417	my $cv = AnyEvent->condvar;
		418
		419	my %result;
		420	$cv->begin (sub { $cv->send (\%result) });
		421
		422	for my $host (@list_of_hosts) {
		423	$cv->begin;
		424	ping_host_then_call_callback $host, sub {
		425	$result{$host} = ...;
		426	$cv->end;
		427	};
		428	}
		429
		430	$cv->end;
		431
		432	This code fragment supposedly pings a number of hosts and calls
		433	C<send> after results for all then have have been gathered - in any
		434	order. To achieve this, the code issues a call to C<begin> when it starts
		435	each ping request and calls C<end> when it has received some result for
		436	it. Since C<begin> and C<end> only maintain a counter, the order in which
		437	results arrive is not relevant.
		438
		439	There is an additional bracketing call to C<begin> and C<end> outside the
		440	loop, which serves two important purposes: first, it sets the callback
		441	to be called once the counter reaches C<0>, and second, it ensures that
		442	C<send> is called even when C<no> hosts are being pinged (the loop
		443	doesn't execute once).
		444
		445	This is the general pattern when you "fan out" into multiple subrequests:
		446	use an outer C<begin>/C<end> pair to set the callback and ensure C<end>
		447	is called at least once, and then, for each subrequest you start, call
		448	C<begin> and for eahc subrequest you finish, call C<end>.
		449
		450	=back
		451
		452	=head3 METHODS FOR CONSUMERS
		453
		454	These methods should only be used by the consuming side, i.e. the
		455	code awaits the condition.
		456
		457	=over 4
		458
		459	=item $cv->recv
		460
		461	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
319	called on c<$cv>, while servicing other watchers normally.	462	>> methods have been called on c<$cv>, while servicing other watchers
		463	normally.
320		464
321	You can only wait once on a condition - additional calls will return	465	You can only wait once on a condition - additional calls are valid but
322	immediately.	466	will return immediately.
		467
		468	If an error condition has been set by calling C<< ->croak >>, then this
		469	function will call C<croak>.
		470
		471	In list context, all parameters passed to C<send> will be returned,
		472	in scalar context only the first one will be returned.
323		473
324	Not all event models support a blocking wait - some die in that case	474	Not all event models support a blocking wait - some die in that case
325	(programs might want to do that to stay interactive), so I<if you are	475	(programs might want to do that to stay interactive), so I<if you are
326	using this from a module, never require a blocking wait>, but let the	476	using this from a module, never require a blocking wait>, but let the
327	caller decide whether the call will block or not (for example, by coupling	477	caller decide whether the call will block or not (for example, by coupling
328	condition variables with some kind of request results and supporting	478	condition variables with some kind of request results and supporting
329	callbacks so the caller knows that getting the result will not block,	479	callbacks so the caller knows that getting the result will not block,
330	while still suppporting blocking waits if the caller so desires).	480	while still suppporting blocking waits if the caller so desires).
331		481
332	Another reason I<never> to C<< ->wait >> in a module is that you cannot	482	Another reason I<never> to C<< ->recv >> in a module is that you cannot
333	sensibly have two C<< ->wait >>'s in parallel, as that would require	483	sensibly have two C<< ->recv >>'s in parallel, as that would require
334	multiple interpreters or coroutines/threads, none of which C<AnyEvent>	484	multiple interpreters or coroutines/threads, none of which C<AnyEvent>
335	can supply (the coroutine-aware backends L<AnyEvent::Impl::CoroEV> and	485	can supply.
336	L<AnyEvent::Impl::CoroEvent> explicitly support concurrent C<< ->wait >>'s
337	from different coroutines, however).
338		486
339	=item $cv->broadcast	487	The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
		488	fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
		489	versions and also integrates coroutines into AnyEvent, making blocking
		490	C<< ->recv >> calls perfectly safe as long as they are done from another
		491	coroutine (one that doesn't run the event loop).
340		492
341	Flag the condition as ready - a running C<< ->wait >> and all further	493	You can ensure that C<< -recv >> never blocks by setting a callback and
342	calls to C<wait> will (eventually) return after this method has been	494	only calling C<< ->recv >> from within that callback (or at a later
343	called. If nobody is waiting the broadcast will be remembered..	495	time). This will work even when the event loop does not support blocking
		496	waits otherwise.
		497
		498	=item $bool = $cv->ready
		499
		500	Returns true when the condition is "true", i.e. whether C<send> or
		501	C<croak> have been called.
		502
		503	=item $cb = $cv->cb ([new callback])
		504
		505	This is a mutator function that returns the callback set and optionally
		506	replaces it before doing so.
		507
		508	The callback will be called when the condition becomes "true", i.e. when
		509	C<send> or C<croak> are called. Calling C<recv> inside the callback
		510	or at any later time is guaranteed not to block.
344		511
345	=back	512	=back
346
347	Example:
348
349	# wait till the result is ready
350	my $result_ready = AnyEvent->condvar;
351
352	# do something such as adding a timer
353	# or socket watcher the calls $result_ready->broadcast
354	# when the "result" is ready.
355	# in this case, we simply use a timer:
356	my $w = AnyEvent->timer (
357	after => 1,
358	cb => sub { $result_ready->broadcast },
359	);
360
361	# this "blocks" (while handling events) till the watcher
362	# calls broadcast
363	$result_ready->wait;
364		513
365	=head1 GLOBAL VARIABLES AND FUNCTIONS	514	=head1 GLOBAL VARIABLES AND FUNCTIONS
366		515
367	=over 4	516	=over 4
368		517
…		…
374	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	523	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case
375	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	524	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).
376		525
377	The known classes so far are:	526	The known classes so far are:
378		527
379	AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
380	AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
381	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).	528	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
382	AnyEvent::Impl::Event based on Event, second best choice.	529	AnyEvent::Impl::Event based on Event, second best choice.
		530	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
383	AnyEvent::Impl::Glib based on Glib, third-best choice.	531	AnyEvent::Impl::Glib based on Glib, third-best choice.
384	AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
385	AnyEvent::Impl::Tk based on Tk, very bad choice.	532	AnyEvent::Impl::Tk based on Tk, very bad choice.
386	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).	533	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
387	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.	534	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
388	AnyEvent::Impl::POE based on POE, not generic enough for full support.	535	AnyEvent::Impl::POE based on POE, not generic enough for full support.
389		536
…		…
402	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model	549	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
403	if necessary. You should only call this function right before you would	550	if necessary. You should only call this function right before you would
404	have created an AnyEvent watcher anyway, that is, as late as possible at	551	have created an AnyEvent watcher anyway, that is, as late as possible at
405	runtime.	552	runtime.
406		553
		554	=item $guard = AnyEvent::post_detect { BLOCK }
		555
		556	Arranges for the code block to be executed as soon as the event model is
		557	autodetected (or immediately if this has already happened).
		558
		559	If called in scalar or list context, then it creates and returns an object
		560	that automatically removes the callback again when it is destroyed. See
		561	L<Coro::BDB> for a case where this is useful.
		562
		563	=item @AnyEvent::post_detect
		564
		565	If there are any code references in this array (you can C<push> to it
		566	before or after loading AnyEvent), then they will called directly after
		567	the event loop has been chosen.
		568
		569	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		570	if it contains a true value then the event loop has already been detected,
		571	and the array will be ignored.
		572
		573	Best use C<AnyEvent::post_detect { BLOCK }> instead.
		574
407	=back	575	=back
408		576
409	=head1 WHAT TO DO IN A MODULE	577	=head1 WHAT TO DO IN A MODULE
410		578
411	As a module author, you should C<use AnyEvent> and call AnyEvent methods	579	As a module author, you should C<use AnyEvent> and call AnyEvent methods
…		…
414	Be careful when you create watchers in the module body - AnyEvent will	582	Be careful when you create watchers in the module body - AnyEvent will
415	decide which event module to use as soon as the first method is called, so	583	decide which event module to use as soon as the first method is called, so
416	by calling AnyEvent in your module body you force the user of your module	584	by calling AnyEvent in your module body you force the user of your module
417	to load the event module first.	585	to load the event module first.
418		586
419	Never call C<< ->wait >> on a condition variable unless you I<know> that	587	Never call C<< ->recv >> on a condition variable unless you I<know> that
420	the C<< ->broadcast >> method has been called on it already. This is	588	the C<< ->send >> method has been called on it already. This is
421	because it will stall the whole program, and the whole point of using	589	because it will stall the whole program, and the whole point of using
422	events is to stay interactive.	590	events is to stay interactive.
423		591
424	It is fine, however, to call C<< ->wait >> when the user of your module	592	It is fine, however, to call C<< ->recv >> when the user of your module
425	requests it (i.e. if you create a http request object ad have a method	593	requests it (i.e. if you create a http request object ad have a method
426	called C<results> that returns the results, it should call C<< ->wait >>	594	called C<results> that returns the results, it should call C<< ->recv >>
427	freely, as the user of your module knows what she is doing. always).	595	freely, as the user of your module knows what she is doing. always).
428		596
429	=head1 WHAT TO DO IN THE MAIN PROGRAM	597	=head1 WHAT TO DO IN THE MAIN PROGRAM
430		598
431	There will always be a single main program - the only place that should	599	There will always be a single main program - the only place that should
…		…
445		613
446	You can chose to use a rather inefficient pure-perl implementation by	614	You can chose to use a rather inefficient pure-perl implementation by
447	loading the C<AnyEvent::Impl::Perl> module, which gives you similar	615	loading the C<AnyEvent::Impl::Perl> module, which gives you similar
448	behaviour everywhere, but letting AnyEvent chose is generally better.	616	behaviour everywhere, but letting AnyEvent chose is generally better.
449		617
		618	=head1 OTHER MODULES
		619
		620	The following is a non-exhaustive list of additional modules that use
		621	AnyEvent and can therefore be mixed easily with other AnyEvent modules
		622	in the same program. Some of the modules come with AnyEvent, some are
		623	available via CPAN.
		624
		625	=over 4
		626
		627	=item L<AnyEvent::Util>
		628
		629	Contains various utility functions that replace often-used but blocking
		630	functions such as C<inet_aton> by event-/callback-based versions.
		631
		632	=item L<AnyEvent::Handle>
		633
		634	Provide read and write buffers and manages watchers for reads and writes.
		635
		636	=item L<AnyEvent::Socket>
		637
		638	Provides various utility functions for (internet protocol) sockets,
		639	addresses and name resolution. Also functions to create non-blocking tcp
		640	connections or tcp servers, with IPv6 and SRV record support and more.
		641
		642	=item L<AnyEvent::HTTPD>
		643
		644	Provides a simple web application server framework.
		645
		646	=item L<AnyEvent::DNS>
		647
		648	Provides rich asynchronous DNS resolver capabilities.
		649
		650	=item L<AnyEvent::FastPing>
		651
		652	The fastest ping in the west.
		653
		654	=item L<Net::IRC3>
		655
		656	AnyEvent based IRC client module family.
		657
		658	=item L<Net::XMPP2>
		659
		660	AnyEvent based XMPP (Jabber protocol) module family.
		661
		662	=item L<Net::FCP>
		663
		664	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		665	of AnyEvent.
		666
		667	=item L<Event::ExecFlow>
		668
		669	High level API for event-based execution flow control.
		670
		671	=item L<Coro>
		672
		673	Has special support for AnyEvent via L<Coro::AnyEvent>.
		674
		675	=item L<AnyEvent::AIO>, L<IO::AIO>
		676
		677	Truly asynchronous I/O, should be in the toolbox of every event
		678	programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
		679	together.
		680
		681	=item L<AnyEvent::BDB>, L<BDB>
		682
		683	Truly asynchronous Berkeley DB access. AnyEvent::AIO transparently fuses
		684	IO::AIO and AnyEvent together.
		685
		686	=item L<IO::Lambda>
		687
		688	The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
		689
		690	=back
		691
450	=cut	692	=cut
451		693
452	package AnyEvent;	694	package AnyEvent;
453		695
454	no warnings;	696	no warnings;
455	use strict;	697	use strict;
456		698
457	use Carp;	699	use Carp;
458		700
459	our $VERSION = '3.3';	701	our $VERSION = '3.6';
460	our $MODEL;	702	our $MODEL;
461		703
462	our $AUTOLOAD;	704	our $AUTOLOAD;
463	our @ISA;	705	our @ISA;
464		706
465	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;	707	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
466		708
467	our @REGISTRY;	709	our @REGISTRY;
468		710
		711	our %PROTOCOL; # (ipv4\|ipv6) => (1\|2)
		712
		713	{
		714	my $idx;
		715	$PROTOCOL{$_} = ++$idx
		716	for split /\s,\s/, $ENV{PERL_ANYEVENT_PROTOCOLS} \|\| "ipv4,ipv6";
		717	}
		718
469	my @models = (	719	my @models = (
470	[Coro::EV:: => AnyEvent::Impl::CoroEV::],
471	[Coro::Event:: => AnyEvent::Impl::CoroEvent::],
472	[EV:: => AnyEvent::Impl::EV::],	720	[EV:: => AnyEvent::Impl::EV::],
473	[Event:: => AnyEvent::Impl::Event::],	721	[Event:: => AnyEvent::Impl::Event::],
474	[Glib:: => AnyEvent::Impl::Glib::],
475	[Tk:: => AnyEvent::Impl::Tk::],	722	[Tk:: => AnyEvent::Impl::Tk::],
476	[Wx:: => AnyEvent::Impl::POE::],	723	[Wx:: => AnyEvent::Impl::POE::],
477	[Prima:: => AnyEvent::Impl::POE::],	724	[Prima:: => AnyEvent::Impl::POE::],
478	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	725	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
479	# everything below here will not be autoprobed as the pureperl backend should work everywhere	726	# everything below here will not be autoprobed as the pureperl backend should work everywhere
		727	[Glib:: => AnyEvent::Impl::Glib::],
480	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy	728	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
481	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program	729	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
482	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza	730	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
483	);	731	);
484		732
485	our %method = map +($_ => 1), qw(io timer signal child condvar broadcast wait one_event DESTROY);	733	our %method = map +($_ => 1), qw(io timer signal child condvar one_event DESTROY);
		734
		735	our @post_detect;
		736
		737	sub post_detect(&) {
		738	my ($cb) = @_;
		739
		740	if ($MODEL) {
		741	$cb->();
		742
		743	1
		744	} else {
		745	push @post_detect, $cb;
		746
		747	defined wantarray
		748	? bless \$cb, "AnyEvent::Util::PostDetect"
		749	: ()
		750	}
		751	}
		752
		753	sub AnyEvent::Util::PostDetect::DESTROY {
		754	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		755	}
486		756
487	sub detect() {	757	sub detect() {
488	unless ($MODEL) {	758	unless ($MODEL) {
489	no strict 'refs';	759	no strict 'refs';
490		760
…		…
524	last;	794	last;
525	}	795	}
526	}	796	}
527		797
528	$MODEL	798	$MODEL
529	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.";	799	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.";
530	}	800	}
531	}	801	}
532		802
533	unshift @ISA, $MODEL;	803	unshift @ISA, $MODEL;
534	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	804	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
		805
		806	(shift @post_detect)->() while @post_detect;
535	}	807	}
536		808
537	$MODEL	809	$MODEL
538	}	810	}
539		811
…		…
549	$class->$func (@_);	821	$class->$func (@_);
550	}	822	}
551		823
552	package AnyEvent::Base;	824	package AnyEvent::Base;
553		825
554	# default implementation for ->condvar, ->wait, ->broadcast	826	# default implementation for ->condvar
555		827
556	sub condvar {	828	sub condvar {
557	bless \my $flag, "AnyEvent::Base::CondVar"	829	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar::
558	}
559
560	sub AnyEvent::Base::CondVar::broadcast {
561	${$_[0]}++;
562	}
563
564	sub AnyEvent::Base::CondVar::wait {
565	AnyEvent->one_event while !${$_[0]};
566	}	830	}
567		831
568	# default implementation for ->signal	832	# default implementation for ->signal
569		833
570	our %SIG_CB;	834	our %SIG_CB;
…		…
644	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };	908	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
645		909
646	undef $CHLD_W unless keys %PID_CB;	910	undef $CHLD_W unless keys %PID_CB;
647	}	911	}
648		912
		913	package AnyEvent::CondVar;
		914
		915	our @ISA = AnyEvent::CondVar::Base::;
		916
		917	package AnyEvent::CondVar::Base;
		918
		919	sub _send {
		920	# nop
		921	}
		922
		923	sub send {
		924	my $cv = shift;
		925	$cv->{_ae_sent} = [@_];
		926	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		927	$cv->_send;
		928	}
		929
		930	sub croak {
		931	$_[0]{_ae_croak} = $_[1];
		932	$_[0]->send;
		933	}
		934
		935	sub ready {
		936	$_[0]{_ae_sent}
		937	}
		938
		939	sub _wait {
		940	AnyEvent->one_event while !$_[0]{_ae_sent};
		941	}
		942
		943	sub recv {
		944	$_[0]->_wait;
		945
		946	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		947	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		948	}
		949
		950	sub cb {
		951	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		952	$_[0]{_ae_cb}
		953	}
		954
		955	sub begin {
		956	++$_[0]{_ae_counter};
		957	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		958	}
		959
		960	sub end {
		961	return if --$_[0]{_ae_counter};
		962	&{ $_[0]{_ae_end_cb} \|\| sub { $_[0]->send } };
		963	}
		964
		965	# undocumented/compatibility with pre-3.4
		966	*broadcast = \&send;
		967	*wait = \&_wait;
		968
649	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	969	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
650		970
651	This is an advanced topic that you do not normally need to use AnyEvent in	971	This is an advanced topic that you do not normally need to use AnyEvent in
652	a module. This section is only of use to event loop authors who want to	972	a module. This section is only of use to event loop authors who want to
653	provide AnyEvent compatibility.	973	provide AnyEvent compatibility.
…		…
721		1041
722	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you	1042	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
723	could start your program like this:	1043	could start your program like this:
724		1044
725	PERL_ANYEVENT_MODEL=Perl perl ...	1045	PERL_ANYEVENT_MODEL=Perl perl ...
		1046
		1047	=item C<PERL_ANYEVENT_PROTOCOLS>
		1048
		1049	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1050	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1051	of autoprobing).
		1052
		1053	Must be set to a comma-separated list of protocols or address families,
		1054	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1055	used, and preference will be given to protocols mentioned earlier in the
		1056	list.
		1057
		1058	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1059	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1060	- only support IPv4, never try to resolve or contact IPv6
		1061	addressses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1062	IPv6, but prefer IPv6 over IPv4.
726		1063
727	=back	1064	=back
728		1065
729	=head1 EXAMPLE PROGRAM	1066	=head1 EXAMPLE PROGRAM
730		1067
…		…
741	poll => 'r',	1078	poll => 'r',
742	cb => sub {	1079	cb => sub {
743	warn "io event <$_[0]>\n"; # will always output <r>	1080	warn "io event <$_[0]>\n"; # will always output <r>
744	chomp (my $input = <STDIN>); # read a line	1081	chomp (my $input = <STDIN>); # read a line
745	warn "read: $input\n"; # output what has been read	1082	warn "read: $input\n"; # output what has been read
746	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1083	$cv->send if $input =~ /^q/i; # quit program if /^q/i
747	},	1084	},
748	);	1085	);
749		1086
750	my $time_watcher; # can only be used once	1087	my $time_watcher; # can only be used once
751		1088
…		…
756	});	1093	});
757	}	1094	}
758		1095
759	new_timer; # create first timer	1096	new_timer; # create first timer
760		1097
761	$cv->wait; # wait until user enters /^q/i	1098	$cv->recv; # wait until user enters /^q/i
762		1099
763	=head1 REAL-WORLD EXAMPLE	1100	=head1 REAL-WORLD EXAMPLE
764		1101
765	Consider the L<Net::FCP> module. It features (among others) the following	1102	Consider the L<Net::FCP> module. It features (among others) the following
766	API calls, which are to freenet what HTTP GET requests are to http:	1103	API calls, which are to freenet what HTTP GET requests are to http:
…		…
822		1159
823	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1160	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
824		1161
825	if (end-of-file or data complete) {	1162	if (end-of-file or data complete) {
826	$txn->{result} = $txn->{buf};	1163	$txn->{result} = $txn->{buf};
827	$txn->{finished}->broadcast;	1164	$txn->{finished}->send;
828	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1165	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
829	}	1166	}
830		1167
831	The C<result> method, finally, just waits for the finished signal (if the	1168	The C<result> method, finally, just waits for the finished signal (if the
832	request was already finished, it doesn't wait, of course, and returns the	1169	request was already finished, it doesn't wait, of course, and returns the
833	data:	1170	data:
834		1171
835	$txn->{finished}->wait;	1172	$txn->{finished}->recv;
836	return $txn->{result};	1173	return $txn->{result};
837		1174
838	The actual code goes further and collects all errors (C<die>s, exceptions)	1175	The actual code goes further and collects all errors (C<die>s, exceptions)
839	that occured during request processing. The C<result> method detects	1176	that occured during request processing. The C<result> method detects
840	whether an exception as thrown (it is stored inside the $txn object)	1177	whether an exception as thrown (it is stored inside the $txn object)
…		…
875		1212
876	my $quit = AnyEvent->condvar;	1213	my $quit = AnyEvent->condvar;
877		1214
878	$fcp->txn_client_get ($url)->cb (sub {	1215	$fcp->txn_client_get ($url)->cb (sub {
879	...	1216	...
880	$quit->broadcast;	1217	$quit->send;
881	});	1218	});
882		1219
883	$quit->wait;	1220	$quit->recv;
884		1221
885		1222
886	=head1 BENCHMARK	1223	=head1 BENCHMARKS
887		1224
888	To give you an idea of the performance and overheads that AnyEvent adds	1225	To give you an idea of the performance and overheads that AnyEvent adds
889	over the event loops themselves (and to give you an impression of the	1226	over the event loops themselves and to give you an impression of the speed
890	speed of various event loops), here is a benchmark of various supported	1227	of various event loops I prepared some benchmarks.
891	event models natively and with anyevent. The benchmark creates a lot of	1228
892	timers (with a zero timeout) and I/O watchers (watching STDOUT, a pty, to	1229	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1230
		1231	Here is a benchmark of various supported event models used natively and
		1232	through anyevent. The benchmark creates a lot of timers (with a zero
		1233	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
893	become writable, which it is), lets them fire exactly once and destroys	1234	which it is), lets them fire exactly once and destroys them again.
894	them again.
895		1235
896	Rewriting the benchmark to use many different sockets instead of using	1236	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
897	the same filehandle for all I/O watchers results in a much longer runtime	1237	distribution.
898	(socket creation is expensive), but qualitatively the same figures, so it
899	was not used.
900		1238
901	=head2 Explanation of the columns	1239	=head3 Explanation of the columns
902		1240
903	I<watcher> is the number of event watchers created/destroyed. Since	1241	I<watcher> is the number of event watchers created/destroyed. Since
904	different event models feature vastly different performances, each event	1242	different event models feature vastly different performances, each event
905	loop was given a number of watchers so that overall runtime is acceptable	1243	loop was given a number of watchers so that overall runtime is acceptable
906	and similar between tested event loop (and keep them from crashing): Glib	1244	and similar between tested event loop (and keep them from crashing): Glib
…		…
916	all watchers, to avoid adding memory overhead. That means closure creation	1254	all watchers, to avoid adding memory overhead. That means closure creation
917	and memory usage is not included in the figures.	1255	and memory usage is not included in the figures.
918		1256
919	I<invoke> is the time, in microseconds, used to invoke a simple	1257	I<invoke> is the time, in microseconds, used to invoke a simple
920	callback. The callback simply counts down a Perl variable and after it was	1258	callback. The callback simply counts down a Perl variable and after it was
921	invoked "watcher" times, it would C<< ->broadcast >> a condvar once to	1259	invoked "watcher" times, it would C<< ->send >> a condvar once to
922	signal the end of this phase.	1260	signal the end of this phase.
923		1261
924	I<destroy> is the time, in microseconds, that it takes to destroy a single	1262	I<destroy> is the time, in microseconds, that it takes to destroy a single
925	watcher.	1263	watcher.
926		1264
927	=head2 Results	1265	=head3 Results
928		1266
929	name watchers bytes create invoke destroy comment	1267	name watchers bytes create invoke destroy comment
930	EV/EV 400000 244 0.56 0.46 0.31 EV native interface	1268	EV/EV 400000 244 0.56 0.46 0.31 EV native interface
931	EV/Any 100000 610 3.52 0.91 0.75 EV + AnyEvent watchers	1269	EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers
932	CoroEV/Any 100000 610 3.49 0.92 0.75 coroutines + Coro::Signal	1270	CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal
933	Perl/Any 100000 513 4.91 0.92 1.15 pure perl implementation	1271	Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation
934	Event/Event 16000 523 28.05 21.38 0.86 Event native interface	1272	Event/Event 16000 516 31.88 31.30 0.85 Event native interface
935	Event/Any 16000 943 34.43 20.48 1.39 Event + AnyEvent watchers	1273	Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers
936	Glib/Any 16000 1357 96.99 12.55 55.51 quadratic behaviour	1274	Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour
937	Tk/Any 2000 1855 27.01 66.61 14.03 SEGV with >> 2000 watchers	1275	Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers
938	POE/Event 2000 6644 108.15 768.19 14.33 via POE::Loop::Event	1276	POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event
939	POE/Select 2000 6343 94.69 807.65 562.69 via POE::Loop::Select	1277	POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select
940		1278
941	=head2 Discussion	1279	=head3 Discussion
942		1280
943	The benchmark does I<not> measure scalability of the event loop very	1281	The benchmark does I<not> measure scalability of the event loop very
944	well. For example, a select-based event loop (such as the pure perl one)	1282	well. For example, a select-based event loop (such as the pure perl one)
945	can never compete with an event loop that uses epoll when the number of	1283	can never compete with an event loop that uses epoll when the number of
946	file descriptors grows high. In this benchmark, all events become ready at	1284	file descriptors grows high. In this benchmark, all events become ready at
947	the same time, so select/poll-based implementations get an unnatural speed	1285	the same time, so select/poll-based implementations get an unnatural speed
948	boost.	1286	boost.
949		1287
		1288	Also, note that the number of watchers usually has a nonlinear effect on
		1289	overall speed, that is, creating twice as many watchers doesn't take twice
		1290	the time - usually it takes longer. This puts event loops tested with a
		1291	higher number of watchers at a disadvantage.
		1292
		1293	To put the range of results into perspective, consider that on the
		1294	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1295	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1296	cycles with POE.
		1297
950	C<EV> is the sole leader regarding speed and memory use, which are both	1298	C<EV> is the sole leader regarding speed and memory use, which are both
951	maximal/minimal, respectively. Even when going through AnyEvent, there are	1299	maximal/minimal, respectively. Even when going through AnyEvent, it uses
952	only two event loops that use slightly less memory (the C<Event> module	1300	far less memory than any other event loop and is still faster than Event
953	natively and the pure perl backend), and no faster event models, not even	1301	natively.
954	C<Event> natively.
955		1302
956	The pure perl implementation is hit in a few sweet spots (both the	1303	The pure perl implementation is hit in a few sweet spots (both the
957	zero timeout and the use of a single fd hit optimisations in the perl	1304	constant timeout and the use of a single fd hit optimisations in the perl
958	interpreter and the backend itself, and all watchers become ready at the	1305	interpreter and the backend itself). Nevertheless this shows that it
959	same time). Nevertheless this shows that it adds very little overhead in	1306	adds very little overhead in itself. Like any select-based backend its
960	itself. Like any select-based backend its performance becomes really bad	1307	performance becomes really bad with lots of file descriptors (and few of
961	with lots of file descriptors (and few of them active), of course, but	1308	them active), of course, but this was not subject of this benchmark.
962	this was not subject of this benchmark.
963		1309
964	The C<Event> module has a relatively high setup and callback invocation cost,	1310	The C<Event> module has a relatively high setup and callback invocation
965	but overall scores on the third place.	1311	cost, but overall scores in on the third place.
966		1312
967	C<Glib>'s memory usage is quite a bit bit higher, but it features a	1313	C<Glib>'s memory usage is quite a bit higher, but it features a
968	faster callback invocation and overall ends up in the same class as	1314	faster callback invocation and overall ends up in the same class as
969	C<Event>. However, Glib scales extremely badly, doubling the number of	1315	C<Event>. However, Glib scales extremely badly, doubling the number of
970	watchers increases the processing time by more than a factor of four,	1316	watchers increases the processing time by more than a factor of four,
971	making it completely unusable when using larger numbers of watchers	1317	making it completely unusable when using larger numbers of watchers
972	(note that only a single file descriptor was used in the benchmark, so	1318	(note that only a single file descriptor was used in the benchmark, so
…		…
975	The C<Tk> adaptor works relatively well. The fact that it crashes with	1321	The C<Tk> adaptor works relatively well. The fact that it crashes with
976	more than 2000 watchers is a big setback, however, as correctness takes	1322	more than 2000 watchers is a big setback, however, as correctness takes
977	precedence over speed. Nevertheless, its performance is surprising, as the	1323	precedence over speed. Nevertheless, its performance is surprising, as the
978	file descriptor is dup()ed for each watcher. This shows that the dup()	1324	file descriptor is dup()ed for each watcher. This shows that the dup()
979	employed by some adaptors is not a big performance issue (it does incur a	1325	employed by some adaptors is not a big performance issue (it does incur a
980	hidden memory cost inside the kernel, though, that is not reflected in the	1326	hidden memory cost inside the kernel which is not reflected in the figures
981	figures above).	1327	above).
982		1328
983	C<POE>, regardless of underlying event loop (wether using its pure perl	1329	C<POE>, regardless of underlying event loop (whether using its pure perl
984	select-based backend or the Event module) shows abysmal performance and	1330	select-based backend or the Event module, the POE-EV backend couldn't
		1331	be tested because it wasn't working) shows abysmal performance and
985	memory usage: Watchers use almost 30 times as much memory as EV watchers,	1332	memory usage with AnyEvent: Watchers use almost 30 times as much memory
986	and 10 times as much memory as both Event or EV via AnyEvent. Watcher	1333	as EV watchers, and 10 times as much memory as Event (the high memory
		1334	requirements are caused by requiring a session for each watcher). Watcher
987	invocation is almost 900 times slower than with AnyEvent's pure perl	1335	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		1336	implementation.
		1337
988	implementation. The design of the POE adaptor class in AnyEvent can not	1338	The design of the POE adaptor class in AnyEvent can not really account
989	really account for this, as session creation overhead is small compared	1339	for the performance issues, though, as session creation overhead is
990	to execution of the state machine, which is coded pretty optimally within	1340	small compared to execution of the state machine, which is coded pretty
991	L<AnyEvent::Impl::POE>. POE simply seems to be abysmally slow.	1341	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		1342	using multiple sessions is not a good approach, especially regarding
		1343	memory usage, even the author of POE could not come up with a faster
		1344	design).
992		1345
993	=head2 Summary	1346	=head3 Summary
994		1347
		1348	=over 4
		1349
995	Using EV through AnyEvent is faster than any other event loop, but most	1350	=item * Using EV through AnyEvent is faster than any other event loop
996	event loops have acceptable performance with or without AnyEvent.	1351	(even when used without AnyEvent), but most event loops have acceptable
		1352	performance with or without AnyEvent.
997		1353
998	The overhead AnyEvent adds is usually much smaller than the overhead of	1354	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
999	the actual event loop, only with extremely fast event loops such as the EV	1355	the actual event loop, only with extremely fast event loops such as EV
1000	adds AnyEvent significant overhead.	1356	adds AnyEvent significant overhead.
1001		1357
1002	And you should simply avoid POE like the plague if you want performance or	1358	=item * You should avoid POE like the plague if you want performance or
1003	reasonable memory usage.	1359	reasonable memory usage.
1004		1360
		1361	=back
		1362
		1363	=head2 BENCHMARKING THE LARGE SERVER CASE
		1364
		1365	This benchmark atcually benchmarks the event loop itself. It works by
		1366	creating a number of "servers": each server consists of a socketpair, a
		1367	timeout watcher that gets reset on activity (but never fires), and an I/O
		1368	watcher waiting for input on one side of the socket. Each time the socket
		1369	watcher reads a byte it will write that byte to a random other "server".
		1370
		1371	The effect is that there will be a lot of I/O watchers, only part of which
		1372	are active at any one point (so there is a constant number of active
		1373	fds for each loop iterstaion, but which fds these are is random). The
		1374	timeout is reset each time something is read because that reflects how
		1375	most timeouts work (and puts extra pressure on the event loops).
		1376
		1377	In this benchmark, we use 10000 socketpairs (20000 sockets), of which 100
		1378	(1%) are active. This mirrors the activity of large servers with many
		1379	connections, most of which are idle at any one point in time.
		1380
		1381	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		1382	distribution.
		1383
		1384	=head3 Explanation of the columns
		1385
		1386	I<sockets> is the number of sockets, and twice the number of "servers" (as
		1387	each server has a read and write socket end).
		1388
		1389	I<create> is the time it takes to create a socketpair (which is
		1390	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		1391
		1392	I<request>, the most important value, is the time it takes to handle a
		1393	single "request", that is, reading the token from the pipe and forwarding
		1394	it to another server. This includes deleting the old timeout and creating
		1395	a new one that moves the timeout into the future.
		1396
		1397	=head3 Results
		1398
		1399	name sockets create request
		1400	EV 20000 69.01 11.16
		1401	Perl 20000 73.32 35.87
		1402	Event 20000 212.62 257.32
		1403	Glib 20000 651.16 1896.30
		1404	POE 20000 349.67 12317.24 uses POE::Loop::Event
		1405
		1406	=head3 Discussion
		1407
		1408	This benchmark I<does> measure scalability and overall performance of the
		1409	particular event loop.
		1410
		1411	EV is again fastest. Since it is using epoll on my system, the setup time
		1412	is relatively high, though.
		1413
		1414	Perl surprisingly comes second. It is much faster than the C-based event
		1415	loops Event and Glib.
		1416
		1417	Event suffers from high setup time as well (look at its code and you will
		1418	understand why). Callback invocation also has a high overhead compared to
		1419	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		1420	uses select or poll in basically all documented configurations.
		1421
		1422	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		1423	clearly fails to perform with many filehandles or in busy servers.
		1424
		1425	POE is still completely out of the picture, taking over 1000 times as long
		1426	as EV, and over 100 times as long as the Perl implementation, even though
		1427	it uses a C-based event loop in this case.
		1428
		1429	=head3 Summary
		1430
		1431	=over 4
		1432
		1433	=item * The pure perl implementation performs extremely well.
		1434
		1435	=item * Avoid Glib or POE in large projects where performance matters.
		1436
		1437	=back
		1438
		1439	=head2 BENCHMARKING SMALL SERVERS
		1440
		1441	While event loops should scale (and select-based ones do not...) even to
		1442	large servers, most programs we (or I :) actually write have only a few
		1443	I/O watchers.
		1444
		1445	In this benchmark, I use the same benchmark program as in the large server
		1446	case, but it uses only eight "servers", of which three are active at any
		1447	one time. This should reflect performance for a small server relatively
		1448	well.
		1449
		1450	The columns are identical to the previous table.
		1451
		1452	=head3 Results
		1453
		1454	name sockets create request
		1455	EV 16 20.00 6.54
		1456	Perl 16 25.75 12.62
		1457	Event 16 81.27 35.86
		1458	Glib 16 32.63 15.48
		1459	POE 16 261.87 276.28 uses POE::Loop::Event
		1460
		1461	=head3 Discussion
		1462
		1463	The benchmark tries to test the performance of a typical small
		1464	server. While knowing how various event loops perform is interesting, keep
		1465	in mind that their overhead in this case is usually not as important, due
		1466	to the small absolute number of watchers (that is, you need efficiency and
		1467	speed most when you have lots of watchers, not when you only have a few of
		1468	them).
		1469
		1470	EV is again fastest.
		1471
		1472	Perl again comes second. It is noticably faster than the C-based event
		1473	loops Event and Glib, although the difference is too small to really
		1474	matter.
		1475
		1476	POE also performs much better in this case, but is is still far behind the
		1477	others.
		1478
		1479	=head3 Summary
		1480
		1481	=over 4
		1482
		1483	=item * C-based event loops perform very well with small number of
		1484	watchers, as the management overhead dominates.
		1485
		1486	=back
		1487
1005		1488
1006	=head1 FORK	1489	=head1 FORK
1007		1490
1008	Most event libraries are not fork-safe. The ones who are usually are	1491	Most event libraries are not fork-safe. The ones who are usually are
1009	because they are so inefficient. Only L<EV> is fully fork-aware.	1492	because they rely on inefficient but fork-safe C<select> or C<poll>
		1493	calls. Only L<EV> is fully fork-aware.
1010		1494
1011	If you have to fork, you must either do so I<before> creating your first	1495	If you have to fork, you must either do so I<before> creating your first
1012	watcher OR you must not use AnyEvent at all in the child.	1496	watcher OR you must not use AnyEvent at all in the child.
1013		1497
1014		1498
…		…
1026		1510
1027	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }	1511	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
1028		1512
1029	use AnyEvent;	1513	use AnyEvent;
1030		1514
		1515	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		1516	be used to probe what backend is used and gain other information (which is
		1517	probably even less useful to an attacker than PERL_ANYEVENT_MODEL).
		1518
1031		1519
1032	=head1 SEE ALSO	1520	=head1 SEE ALSO
1033		1521
1034	Event modules: L<Coro::EV>, L<EV>, L<EV::Glib>, L<Glib::EV>,	1522	Utility functions: L<AnyEvent::Util>.
1035	L<Coro::Event>, L<Event>, L<Glib::Event>, L<Glib>, L<Coro>, L<Tk>,	1523
		1524	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
1036	L<Event::Lib>, L<Qt>, L<POE>.	1525	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
1037		1526
1038	Implementations: L<AnyEvent::Impl::CoroEV>, L<AnyEvent::Impl::EV>,	1527	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
1039	L<AnyEvent::Impl::CoroEvent>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>,	1528	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
1040	L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>, L<AnyEvent::Impl::EventLib>,	1529	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
1041	L<AnyEvent::Impl::Qt>, L<AnyEvent::Impl::POE>.	1530	L<AnyEvent::Impl::POE>.
1042		1531
		1532	Non-blocking file handles, sockets, TCP clients and
		1533	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>.
		1534
		1535	Asynchronous DNS: L<AnyEvent::DNS>.
		1536
		1537	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>,
		1538
1043	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>.	1539	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>.
1044		1540
1045		1541
1046	=head1 AUTHOR	1542	=head1 AUTHOR
1047		1543
1048	Marc Lehmann <schmorp@schmorp.de>	1544	Marc Lehmann <schmorp@schmorp.de>

Diff Legend

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

Comparing AnyEvent/lib/AnyEvent.pm (file contents): Revision 1.82 by root, Fri Apr 25 13:32:39 2008 UTC vs. Revision 1.126 by root, Fri May 23 23:44:55 2008 UTC

Diff Legend

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.82 by root, Fri Apr 25 13:32:39 2008 UTC vs.
Revision 1.126 by root, Fri May 23 23:44:55 2008 UTC