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

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.86 by root, Fri Apr 25 14:01:48 2008 UTC vs.
Revision 1.135 by root, Sun May 25 04:49:01 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?
…		…
57	as those use one of the supported event loops. It is trivial to add new	57	as those use one of the supported event loops. It is trivial to add new
58	event loops to AnyEvent, too, so it is future-proof).	58	event loops to AnyEvent, too, so it is future-proof).
59		59
60	In addition to being free of having to use I<the one and only true event	60	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	61	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	62	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	63	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	64	offering the functionality that is necessary, in as thin as a wrapper as
65	technically possible.	65	technically possible.
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
71		70
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
…		…
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
…		…
109		108
110	=head1 WATCHERS	109	=head1 WATCHERS
111		110
112	AnyEvent has the central concept of a I<watcher>, which is an object that	111	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	112	stores relevant data for each kind of event you are waiting for, such as
114	the callback to call, the filehandle to watch, etc.	113	the callback to call, the file handle to watch, etc.
115		114
116	These watchers are normal Perl objects with normal Perl lifetime. After	115	These watchers are normal Perl objects with normal Perl lifetime. After
117	creating a watcher it will immediately "watch" for events and invoke the	116	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	117	callback when the event occurs (of course, only when the event model
119	is in control).	118	is in control).
…		…
238		237
239	Although the callback might get passed parameters, their value and	238	Although the callback might get passed parameters, their value and
240	presence is undefined and you cannot rely on them. Portable AnyEvent	239	presence is undefined and you cannot rely on them. Portable AnyEvent
241	callbacks cannot use arguments passed to signal watcher callbacks.	240	callbacks cannot use arguments passed to signal watcher callbacks.
242		241
243	Multiple signal occurances can be clumped together into one callback	242	Multiple signal occurrences can be clumped together into one callback
244	invocation, and callback invocation will be synchronous. synchronous means	243	invocation, and callback invocation will be synchronous. Synchronous means
245	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,
246	but it is guarenteed not to interrupt any other callbacks.	245	but it is guaranteed not to interrupt any other callbacks.
247		246
248	The main advantage of using these watchers is that you can share a signal	247	The main advantage of using these watchers is that you can share a signal
249	between multiple watchers.	248	between multiple watchers.
250		249
251	This watcher might use C<%SIG>, so programs overwriting those signals	250	This watcher might use C<%SIG>, so programs overwriting those signals
…		…
280		279
281	Example: fork a process and wait for it	280	Example: fork a process and wait for it
282		281
283	my $done = AnyEvent->condvar;	282	my $done = AnyEvent->condvar;
284		283
285	AnyEvent::detect; # force event module to be initialised
286
287	my $pid = fork or exit 5;	284	my $pid = fork or exit 5;
288		285
289	my $w = AnyEvent->child (	286	my $w = AnyEvent->child (
290	pid => $pid,	287	pid => $pid,
291	cb => sub {	288	cb => sub {
292	my ($pid, $status) = @_;	289	my ($pid, $status) = @_;
293	warn "pid $pid exited with status $status";	290	warn "pid $pid exited with status $status";
294	$done->broadcast;	291	$done->send;
295	},	292	},
296	);	293	);
297		294
298	# do something else, then wait for process exit	295	# do something else, then wait for process exit
299	$done->wait;	296	$done->recv;
300		297
301	=head2 CONDITION VARIABLES	298	=head2 CONDITION VARIABLES
302		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
303	Condition variables can be created by calling the C<< AnyEvent->condvar >>	310	Condition variables can be created by calling the C<< AnyEvent->condvar
304	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.
305		314
306	A condition variable waits for a condition - precisely that the C<<	315	After creation, the condition variable is "false" until it becomes "true"
307	->broadcast >> method has been called.	316	by calling the C<send> method (or calling the condition variable as if it
		317	were a callback, read about the caveats in the description for the C<<
		318	->send >> method).
308		319
309	They are very useful to signal that a condition has been fulfilled, for	320	Condition variables are similar to callbacks, except that you can
		321	optionally wait for them. They can also be called merge points - points
		322	in time where multiple outstanding events have been processed. And yet
		323	another way to call them is transactions - each condition variable can be
		324	used to represent a transaction, which finishes at some point and delivers
		325	a result.
		326
		327	Condition variables are very useful to signal that something has finished,
310	example, if you write a module that does asynchronous http requests,	328	for example, if you write a module that does asynchronous http requests,
311	then a condition variable would be the ideal candidate to signal the	329	then a condition variable would be the ideal candidate to signal the
312	availability of results.	330	availability of results. The user can either act when the callback is
		331	called or can synchronously C<< ->recv >> for the results.
313		332
314	You can also use condition variables to block your main program until	333	You can also use them to simulate traditional event loops - for example,
315	an event occurs - for example, you could C<< ->wait >> in your main	334	you can block your main program until an event occurs - for example, you
316	program until the user clicks the Quit button in your app, which would C<<	335	could C<< ->recv >> in your main program until the user clicks the Quit
317	->broadcast >> the "quit" event.	336	button of your app, which would C<< ->send >> the "quit" event.
318		337
319	Note that condition variables recurse into the event loop - if you have	338	Note that condition variables recurse into the event loop - if you have
320	two pirces of code that call C<< ->wait >> in a round-robbin fashion, you	339	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
321	lose. Therefore, condition variables are good to export to your caller, but	340	lose. Therefore, condition variables are good to export to your caller, but
322	you should avoid making a blocking wait yourself, at least in callbacks,	341	you should avoid making a blocking wait yourself, at least in callbacks,
323	as this asks for trouble.	342	as this asks for trouble.
324		343
325	This object has two methods:	344	Condition variables are represented by hash refs in perl, and the keys
		345	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		346	easy (it is often useful to build your own transaction class on top of
		347	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		348	it's C<new> method in your own C<new> method.
		349
		350	There are two "sides" to a condition variable - the "producer side" which
		351	eventually calls C<< -> send >>, and the "consumer side", which waits
		352	for the send to occur.
		353
		354	Example: wait for a timer.
		355
		356	# wait till the result is ready
		357	my $result_ready = AnyEvent->condvar;
		358
		359	# do something such as adding a timer
		360	# or socket watcher the calls $result_ready->send
		361	# when the "result" is ready.
		362	# in this case, we simply use a timer:
		363	my $w = AnyEvent->timer (
		364	after => 1,
		365	cb => sub { $result_ready->send },
		366	);
		367
		368	# this "blocks" (while handling events) till the callback
		369	# calls send
		370	$result_ready->recv;
		371
		372	Example: wait for a timer, but take advantage of the fact that
		373	condition variables are also code references.
		374
		375	my $done = AnyEvent->condvar;
		376	my $delay = AnyEvent->timer (after => 5, cb => $done);
		377	$done->recv;
		378
		379	=head3 METHODS FOR PRODUCERS
		380
		381	These methods should only be used by the producing side, i.e. the
		382	code/module that eventually sends the signal. Note that it is also
		383	the producer side which creates the condvar in most cases, but it isn't
		384	uncommon for the consumer to create it as well.
326		385
327	=over 4	386	=over 4
328		387
		388	=item $cv->send (...)
		389
		390	Flag the condition as ready - a running C<< ->recv >> and all further
		391	calls to C<recv> will (eventually) return after this method has been
		392	called. If nobody is waiting the send will be remembered.
		393
		394	If a callback has been set on the condition variable, it is called
		395	immediately from within send.
		396
		397	Any arguments passed to the C<send> call will be returned by all
		398	future C<< ->recv >> calls.
		399
		400	Condition variables are overloaded so one can call them directly
		401	(as a code reference). Calling them directly is the same as calling
		402	C<send>. Note, however, that many C-based event loops do not handle
		403	overloading, so as tempting as it may be, passing a condition variable
		404	instead of a callback does not work. Both the pure perl and EV loops
		405	support overloading, however, as well as all functions that use perl to
		406	invoke a callback (as in L<AnyEvent::Socket> and L<AnyEvent::DNS> for
		407	example).
		408
		409	=item $cv->croak ($error)
		410
		411	Similar to send, but causes all call's to C<< ->recv >> to invoke
		412	C<Carp::croak> with the given error message/object/scalar.
		413
		414	This can be used to signal any errors to the condition variable
		415	user/consumer.
		416
		417	=item $cv->begin ([group callback])
		418
329	=item $cv->wait	419	=item $cv->end
330		420
331	Wait (blocking if necessary) until the C<< ->broadcast >> method has been	421	These two methods are EXPERIMENTAL and MIGHT CHANGE.
		422
		423	These two methods can be used to combine many transactions/events into
		424	one. For example, a function that pings many hosts in parallel might want
		425	to use a condition variable for the whole process.
		426
		427	Every call to C<< ->begin >> will increment a counter, and every call to
		428	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		429	>>, the (last) callback passed to C<begin> will be executed. That callback
		430	is I<supposed> to call C<< ->send >>, but that is not required. If no
		431	callback was set, C<send> will be called without any arguments.
		432
		433	Let's clarify this with the ping example:
		434
		435	my $cv = AnyEvent->condvar;
		436
		437	my %result;
		438	$cv->begin (sub { $cv->send (\%result) });
		439
		440	for my $host (@list_of_hosts) {
		441	$cv->begin;
		442	ping_host_then_call_callback $host, sub {
		443	$result{$host} = ...;
		444	$cv->end;
		445	};
		446	}
		447
		448	$cv->end;
		449
		450	This code fragment supposedly pings a number of hosts and calls
		451	C<send> after results for all then have have been gathered - in any
		452	order. To achieve this, the code issues a call to C<begin> when it starts
		453	each ping request and calls C<end> when it has received some result for
		454	it. Since C<begin> and C<end> only maintain a counter, the order in which
		455	results arrive is not relevant.
		456
		457	There is an additional bracketing call to C<begin> and C<end> outside the
		458	loop, which serves two important purposes: first, it sets the callback
		459	to be called once the counter reaches C<0>, and second, it ensures that
		460	C<send> is called even when C<no> hosts are being pinged (the loop
		461	doesn't execute once).
		462
		463	This is the general pattern when you "fan out" into multiple subrequests:
		464	use an outer C<begin>/C<end> pair to set the callback and ensure C<end>
		465	is called at least once, and then, for each subrequest you start, call
		466	C<begin> and for each subrequest you finish, call C<end>.
		467
		468	=back
		469
		470	=head3 METHODS FOR CONSUMERS
		471
		472	These methods should only be used by the consuming side, i.e. the
		473	code awaits the condition.
		474
		475	=over 4
		476
		477	=item $cv->recv
		478
		479	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
332	called on c<$cv>, while servicing other watchers normally.	480	>> methods have been called on c<$cv>, while servicing other watchers
		481	normally.
333		482
334	You can only wait once on a condition - additional calls will return	483	You can only wait once on a condition - additional calls are valid but
335	immediately.	484	will return immediately.
		485
		486	If an error condition has been set by calling C<< ->croak >>, then this
		487	function will call C<croak>.
		488
		489	In list context, all parameters passed to C<send> will be returned,
		490	in scalar context only the first one will be returned.
336		491
337	Not all event models support a blocking wait - some die in that case	492	Not all event models support a blocking wait - some die in that case
338	(programs might want to do that to stay interactive), so I<if you are	493	(programs might want to do that to stay interactive), so I<if you are
339	using this from a module, never require a blocking wait>, but let the	494	using this from a module, never require a blocking wait>, but let the
340	caller decide whether the call will block or not (for example, by coupling	495	caller decide whether the call will block or not (for example, by coupling
341	condition variables with some kind of request results and supporting	496	condition variables with some kind of request results and supporting
342	callbacks so the caller knows that getting the result will not block,	497	callbacks so the caller knows that getting the result will not block,
343	while still suppporting blocking waits if the caller so desires).	498	while still supporting blocking waits if the caller so desires).
344		499
345	Another reason I<never> to C<< ->wait >> in a module is that you cannot	500	Another reason I<never> to C<< ->recv >> in a module is that you cannot
346	sensibly have two C<< ->wait >>'s in parallel, as that would require	501	sensibly have two C<< ->recv >>'s in parallel, as that would require
347	multiple interpreters or coroutines/threads, none of which C<AnyEvent>	502	multiple interpreters or coroutines/threads, none of which C<AnyEvent>
348	can supply (the coroutine-aware backends L<AnyEvent::Impl::CoroEV> and	503	can supply.
349	L<AnyEvent::Impl::CoroEvent> explicitly support concurrent C<< ->wait >>'s
350	from different coroutines, however).
351		504
352	=item $cv->broadcast	505	The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
		506	fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
		507	versions and also integrates coroutines into AnyEvent, making blocking
		508	C<< ->recv >> calls perfectly safe as long as they are done from another
		509	coroutine (one that doesn't run the event loop).
353		510
354	Flag the condition as ready - a running C<< ->wait >> and all further	511	You can ensure that C<< -recv >> never blocks by setting a callback and
355	calls to C<wait> will (eventually) return after this method has been	512	only calling C<< ->recv >> from within that callback (or at a later
356	called. If nobody is waiting the broadcast will be remembered..	513	time). This will work even when the event loop does not support blocking
		514	waits otherwise.
		515
		516	=item $bool = $cv->ready
		517
		518	Returns true when the condition is "true", i.e. whether C<send> or
		519	C<croak> have been called.
		520
		521	=item $cb = $cv->cb ([new callback])
		522
		523	This is a mutator function that returns the callback set and optionally
		524	replaces it before doing so.
		525
		526	The callback will be called when the condition becomes "true", i.e. when
		527	C<send> or C<croak> are called. Calling C<recv> inside the callback
		528	or at any later time is guaranteed not to block.
357		529
358	=back	530	=back
359
360	Example:
361
362	# wait till the result is ready
363	my $result_ready = AnyEvent->condvar;
364
365	# do something such as adding a timer
366	# or socket watcher the calls $result_ready->broadcast
367	# when the "result" is ready.
368	# in this case, we simply use a timer:
369	my $w = AnyEvent->timer (
370	after => 1,
371	cb => sub { $result_ready->broadcast },
372	);
373
374	# this "blocks" (while handling events) till the watcher
375	# calls broadcast
376	$result_ready->wait;
377		531
378	=head1 GLOBAL VARIABLES AND FUNCTIONS	532	=head1 GLOBAL VARIABLES AND FUNCTIONS
379		533
380	=over 4	534	=over 4
381		535
…		…
387	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case	541	C<AnyEvent::Impl:xxx> modules, but can be any other class in the case
388	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).	542	AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).
389		543
390	The known classes so far are:	544	The known classes so far are:
391		545
392	AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
393	AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
394	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).	546	AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
395	AnyEvent::Impl::Event based on Event, second best choice.	547	AnyEvent::Impl::Event based on Event, second best choice.
		548	AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
396	AnyEvent::Impl::Glib based on Glib, third-best choice.	549	AnyEvent::Impl::Glib based on Glib, third-best choice.
397	AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
398	AnyEvent::Impl::Tk based on Tk, very bad choice.	550	AnyEvent::Impl::Tk based on Tk, very bad choice.
399	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).	551	AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
400	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.	552	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
401	AnyEvent::Impl::POE based on POE, not generic enough for full support.	553	AnyEvent::Impl::POE based on POE, not generic enough for full support.
402		554
…		…
415	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model	567	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
416	if necessary. You should only call this function right before you would	568	if necessary. You should only call this function right before you would
417	have created an AnyEvent watcher anyway, that is, as late as possible at	569	have created an AnyEvent watcher anyway, that is, as late as possible at
418	runtime.	570	runtime.
419		571
		572	=item $guard = AnyEvent::post_detect { BLOCK }
		573
		574	Arranges for the code block to be executed as soon as the event model is
		575	autodetected (or immediately if this has already happened).
		576
		577	If called in scalar or list context, then it creates and returns an object
		578	that automatically removes the callback again when it is destroyed. See
		579	L<Coro::BDB> for a case where this is useful.
		580
		581	=item @AnyEvent::post_detect
		582
		583	If there are any code references in this array (you can C<push> to it
		584	before or after loading AnyEvent), then they will called directly after
		585	the event loop has been chosen.
		586
		587	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		588	if it contains a true value then the event loop has already been detected,
		589	and the array will be ignored.
		590
		591	Best use C<AnyEvent::post_detect { BLOCK }> instead.
		592
420	=back	593	=back
421		594
422	=head1 WHAT TO DO IN A MODULE	595	=head1 WHAT TO DO IN A MODULE
423		596
424	As a module author, you should C<use AnyEvent> and call AnyEvent methods	597	As a module author, you should C<use AnyEvent> and call AnyEvent methods
…		…
427	Be careful when you create watchers in the module body - AnyEvent will	600	Be careful when you create watchers in the module body - AnyEvent will
428	decide which event module to use as soon as the first method is called, so	601	decide which event module to use as soon as the first method is called, so
429	by calling AnyEvent in your module body you force the user of your module	602	by calling AnyEvent in your module body you force the user of your module
430	to load the event module first.	603	to load the event module first.
431		604
432	Never call C<< ->wait >> on a condition variable unless you I<know> that	605	Never call C<< ->recv >> on a condition variable unless you I<know> that
433	the C<< ->broadcast >> method has been called on it already. This is	606	the C<< ->send >> method has been called on it already. This is
434	because it will stall the whole program, and the whole point of using	607	because it will stall the whole program, and the whole point of using
435	events is to stay interactive.	608	events is to stay interactive.
436		609
437	It is fine, however, to call C<< ->wait >> when the user of your module	610	It is fine, however, to call C<< ->recv >> when the user of your module
438	requests it (i.e. if you create a http request object ad have a method	611	requests it (i.e. if you create a http request object ad have a method
439	called C<results> that returns the results, it should call C<< ->wait >>	612	called C<results> that returns the results, it should call C<< ->recv >>
440	freely, as the user of your module knows what she is doing. always).	613	freely, as the user of your module knows what she is doing. always).
441		614
442	=head1 WHAT TO DO IN THE MAIN PROGRAM	615	=head1 WHAT TO DO IN THE MAIN PROGRAM
443		616
444	There will always be a single main program - the only place that should	617	There will always be a single main program - the only place that should
…		…
446		619
447	If it doesn't care, it can just "use AnyEvent" and use it itself, or not	620	If it doesn't care, it can just "use AnyEvent" and use it itself, or not
448	do anything special (it does not need to be event-based) and let AnyEvent	621	do anything special (it does not need to be event-based) and let AnyEvent
449	decide which implementation to chose if some module relies on it.	622	decide which implementation to chose if some module relies on it.
450		623
451	If the main program relies on a specific event model. For example, in	624	If the main program relies on a specific event model - for example, in
452	Gtk2 programs you have to rely on the Glib module. You should load the	625	Gtk2 programs you have to rely on the Glib module - you should load the
453	event module before loading AnyEvent or any module that uses it: generally	626	event module before loading AnyEvent or any module that uses it: generally
454	speaking, you should load it as early as possible. The reason is that	627	speaking, you should load it as early as possible. The reason is that
455	modules might create watchers when they are loaded, and AnyEvent will	628	modules might create watchers when they are loaded, and AnyEvent will
456	decide on the event model to use as soon as it creates watchers, and it	629	decide on the event model to use as soon as it creates watchers, and it
457	might chose the wrong one unless you load the correct one yourself.	630	might chose the wrong one unless you load the correct one yourself.
458		631
459	You can chose to use a rather inefficient pure-perl implementation by	632	You can chose to use a pure-perl implementation by loading the
460	loading the C<AnyEvent::Impl::Perl> module, which gives you similar	633	C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
461	behaviour everywhere, but letting AnyEvent chose is generally better.	634	everywhere, but letting AnyEvent chose the model is generally better.
		635
		636	=head2 MAINLOOP EMULATION
		637
		638	Sometimes (often for short test scripts, or even standalone programs who
		639	only want to use AnyEvent), you do not want to run a specific event loop.
		640
		641	In that case, you can use a condition variable like this:
		642
		643	AnyEvent->condvar->recv;
		644
		645	This has the effect of entering the event loop and looping forever.
		646
		647	Note that usually your program has some exit condition, in which case
		648	it is better to use the "traditional" approach of storing a condition
		649	variable somewhere, waiting for it, and sending it when the program should
		650	exit cleanly.
		651
		652
		653	=head1 OTHER MODULES
		654
		655	The following is a non-exhaustive list of additional modules that use
		656	AnyEvent and can therefore be mixed easily with other AnyEvent modules
		657	in the same program. Some of the modules come with AnyEvent, some are
		658	available via CPAN.
		659
		660	=over 4
		661
		662	=item L<AnyEvent::Util>
		663
		664	Contains various utility functions that replace often-used but blocking
		665	functions such as C<inet_aton> by event-/callback-based versions.
		666
		667	=item L<AnyEvent::Handle>
		668
		669	Provide read and write buffers and manages watchers for reads and writes.
		670
		671	=item L<AnyEvent::Socket>
		672
		673	Provides various utility functions for (internet protocol) sockets,
		674	addresses and name resolution. Also functions to create non-blocking tcp
		675	connections or tcp servers, with IPv6 and SRV record support and more.
		676
		677	=item L<AnyEvent::DNS>
		678
		679	Provides rich asynchronous DNS resolver capabilities.
		680
		681	=item L<AnyEvent::HTTPD>
		682
		683	Provides a simple web application server framework.
		684
		685	=item L<AnyEvent::FastPing>
		686
		687	The fastest ping in the west.
		688
		689	=item L<Net::IRC3>
		690
		691	AnyEvent based IRC client module family.
		692
		693	=item L<Net::XMPP2>
		694
		695	AnyEvent based XMPP (Jabber protocol) module family.
		696
		697	=item L<Net::FCP>
		698
		699	AnyEvent-based implementation of the Freenet Client Protocol, birthplace
		700	of AnyEvent.
		701
		702	=item L<Event::ExecFlow>
		703
		704	High level API for event-based execution flow control.
		705
		706	=item L<Coro>
		707
		708	Has special support for AnyEvent via L<Coro::AnyEvent>.
		709
		710	=item L<AnyEvent::AIO>, L<IO::AIO>
		711
		712	Truly asynchronous I/O, should be in the toolbox of every event
		713	programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
		714	together.
		715
		716	=item L<AnyEvent::BDB>, L<BDB>
		717
		718	Truly asynchronous Berkeley DB access. AnyEvent::AIO transparently fuses
		719	IO::AIO and AnyEvent together.
		720
		721	=item L<IO::Lambda>
		722
		723	The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
		724
		725	=back
462		726
463	=cut	727	=cut
464		728
465	package AnyEvent;	729	package AnyEvent;
466		730
467	no warnings;	731	no warnings;
468	use strict;	732	use strict;
469		733
470	use Carp;	734	use Carp;
471		735
472	our $VERSION = '3.3';	736	our $VERSION = '4.03';
473	our $MODEL;	737	our $MODEL;
474		738
475	our $AUTOLOAD;	739	our $AUTOLOAD;
476	our @ISA;	740	our @ISA;
477		741
		742	our @REGISTRY;
		743
478	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;	744	our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
479		745
480	our @REGISTRY;	746	our %PROTOCOL; # (ipv4\|ipv6) => (1\|2)
		747
		748	{
		749	my $idx;
		750	$PROTOCOL{$_} = ++$idx
		751	for split /\s,\s/, $ENV{PERL_ANYEVENT_PROTOCOLS} \|\| "ipv4,ipv6";
		752	}
481		753
482	my @models = (	754	my @models = (
483	[Coro::EV:: => AnyEvent::Impl::CoroEV::],
484	[Coro::Event:: => AnyEvent::Impl::CoroEvent::],
485	[EV:: => AnyEvent::Impl::EV::],	755	[EV:: => AnyEvent::Impl::EV::],
486	[Event:: => AnyEvent::Impl::Event::],	756	[Event:: => AnyEvent::Impl::Event::],
487	[Glib:: => AnyEvent::Impl::Glib::],
488	[Tk:: => AnyEvent::Impl::Tk::],
489	[Wx:: => AnyEvent::Impl::POE::],
490	[Prima:: => AnyEvent::Impl::POE::],
491	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],	757	[AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
492	# everything below here will not be autoprobed as the pureperl backend should work everywhere	758	# everything below here will not be autoprobed
		759	# as the pureperl backend should work everywhere
		760	# and is usually faster
		761	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		762	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
493	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy	763	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
494	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program	764	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
495	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza	765	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
		766	[Wx:: => AnyEvent::Impl::POE::],
		767	[Prima:: => AnyEvent::Impl::POE::],
496	);	768	);
497		769
498	our %method = map +($_ => 1), qw(io timer signal child condvar broadcast wait one_event DESTROY);	770	our %method = map +($_ => 1), qw(io timer signal child condvar one_event DESTROY);
		771
		772	our @post_detect;
		773
		774	sub post_detect(&) {
		775	my ($cb) = @_;
		776
		777	if ($MODEL) {
		778	$cb->();
		779
		780	1
		781	} else {
		782	push @post_detect, $cb;
		783
		784	defined wantarray
		785	? bless \$cb, "AnyEvent::Util::PostDetect"
		786	: ()
		787	}
		788	}
		789
		790	sub AnyEvent::Util::PostDetect::DESTROY {
		791	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		792	}
499		793
500	sub detect() {	794	sub detect() {
501	unless ($MODEL) {	795	unless ($MODEL) {
502	no strict 'refs';	796	no strict 'refs';
503		797
…		…
537	last;	831	last;
538	}	832	}
539	}	833	}
540		834
541	$MODEL	835	$MODEL
542	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.";	836	or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.";
543	}	837	}
544	}	838	}
545		839
546	unshift @ISA, $MODEL;	840	unshift @ISA, $MODEL;
547	push @{"$MODEL\::ISA"}, "AnyEvent::Base";	841	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
		842
		843	(shift @post_detect)->() while @post_detect;
548	}	844	}
549		845
550	$MODEL	846	$MODEL
551	}	847	}
552		848
…		…
562	$class->$func (@_);	858	$class->$func (@_);
563	}	859	}
564		860
565	package AnyEvent::Base;	861	package AnyEvent::Base;
566		862
567	# default implementation for ->condvar, ->wait, ->broadcast	863	# default implementation for ->condvar
568		864
569	sub condvar {	865	sub condvar {
570	bless \my $flag, "AnyEvent::Base::CondVar"	866	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar::
571	}
572
573	sub AnyEvent::Base::CondVar::broadcast {
574	${$_[0]}++;
575	}
576
577	sub AnyEvent::Base::CondVar::wait {
578	AnyEvent->one_event while !${$_[0]};
579	}	867	}
580		868
581	# default implementation for ->signal	869	# default implementation for ->signal
582		870
583	our %SIG_CB;	871	our %SIG_CB;
…		…
657	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };	945	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
658		946
659	undef $CHLD_W unless keys %PID_CB;	947	undef $CHLD_W unless keys %PID_CB;
660	}	948	}
661		949
		950	package AnyEvent::CondVar;
		951
		952	our @ISA = AnyEvent::CondVar::Base::;
		953
		954	package AnyEvent::CondVar::Base;
		955
		956	use overload
		957	'&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		958	fallback => 1;
		959
		960	sub _send {
		961	# nop
		962	}
		963
		964	sub send {
		965	my $cv = shift;
		966	$cv->{_ae_sent} = [@_];
		967	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		968	$cv->_send;
		969	}
		970
		971	sub croak {
		972	$_[0]{_ae_croak} = $_[1];
		973	$_[0]->send;
		974	}
		975
		976	sub ready {
		977	$_[0]{_ae_sent}
		978	}
		979
		980	sub _wait {
		981	AnyEvent->one_event while !$_[0]{_ae_sent};
		982	}
		983
		984	sub recv {
		985	$_[0]->_wait;
		986
		987	Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
		988	wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
		989	}
		990
		991	sub cb {
		992	$_[0]{_ae_cb} = $_[1] if @_ > 1;
		993	$_[0]{_ae_cb}
		994	}
		995
		996	sub begin {
		997	++$_[0]{_ae_counter};
		998	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		999	}
		1000
		1001	sub end {
		1002	return if --$_[0]{_ae_counter};
		1003	&{ $_[0]{_ae_end_cb} \|\| sub { $_[0]->send } };
		1004	}
		1005
		1006	# undocumented/compatibility with pre-3.4
		1007	*broadcast = \&send;
		1008	*wait = \&_wait;
		1009
662	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE	1010	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
663		1011
664	This is an advanced topic that you do not normally need to use AnyEvent in	1012	This is an advanced topic that you do not normally need to use AnyEvent in
665	a module. This section is only of use to event loop authors who want to	1013	a module. This section is only of use to event loop authors who want to
666	provide AnyEvent compatibility.	1014	provide AnyEvent compatibility.
…		…
722	model it chooses.	1070	model it chooses.
723		1071
724	=item C<PERL_ANYEVENT_MODEL>	1072	=item C<PERL_ANYEVENT_MODEL>
725		1073
726	This can be used to specify the event model to be used by AnyEvent, before	1074	This can be used to specify the event model to be used by AnyEvent, before
727	autodetection and -probing kicks in. It must be a string consisting	1075	auto detection and -probing kicks in. It must be a string consisting
728	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended	1076	entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
729	and the resulting module name is loaded and if the load was successful,	1077	and the resulting module name is loaded and if the load was successful,
730	used as event model. If it fails to load AnyEvent will proceed with	1078	used as event model. If it fails to load AnyEvent will proceed with
731	autodetection and -probing.	1079	auto detection and -probing.
732		1080
733	This functionality might change in future versions.	1081	This functionality might change in future versions.
734		1082
735	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you	1083	For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
736	could start your program like this:	1084	could start your program like this:
737		1085
738	PERL_ANYEVENT_MODEL=Perl perl ...	1086	PERL_ANYEVENT_MODEL=Perl perl ...
		1087
		1088	=item C<PERL_ANYEVENT_PROTOCOLS>
		1089
		1090	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		1091	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		1092	of auto probing).
		1093
		1094	Must be set to a comma-separated list of protocols or address families,
		1095	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		1096	used, and preference will be given to protocols mentioned earlier in the
		1097	list.
		1098
		1099	This variable can effectively be used for denial-of-service attacks
		1100	against local programs (e.g. when setuid), although the impact is likely
		1101	small, as the program has to handle connection errors already-
		1102
		1103	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		1104	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		1105	- only support IPv4, never try to resolve or contact IPv6
		1106	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		1107	IPv6, but prefer IPv6 over IPv4.
		1108
		1109	=item C<PERL_ANYEVENT_EDNS0>
		1110
		1111	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
		1112	for DNS. This extension is generally useful to reduce DNS traffic, but
		1113	some (broken) firewalls drop such DNS packets, which is why it is off by
		1114	default.
		1115
		1116	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		1117	EDNS0 in its DNS requests.
739		1118
740	=back	1119	=back
741		1120
742	=head1 EXAMPLE PROGRAM	1121	=head1 EXAMPLE PROGRAM
743		1122
…		…
754	poll => 'r',	1133	poll => 'r',
755	cb => sub {	1134	cb => sub {
756	warn "io event <$_[0]>\n"; # will always output <r>	1135	warn "io event <$_[0]>\n"; # will always output <r>
757	chomp (my $input = <STDIN>); # read a line	1136	chomp (my $input = <STDIN>); # read a line
758	warn "read: $input\n"; # output what has been read	1137	warn "read: $input\n"; # output what has been read
759	$cv->broadcast if $input =~ /^q/i; # quit program if /^q/i	1138	$cv->send if $input =~ /^q/i; # quit program if /^q/i
760	},	1139	},
761	);	1140	);
762		1141
763	my $time_watcher; # can only be used once	1142	my $time_watcher; # can only be used once
764		1143
…		…
769	});	1148	});
770	}	1149	}
771		1150
772	new_timer; # create first timer	1151	new_timer; # create first timer
773		1152
774	$cv->wait; # wait until user enters /^q/i	1153	$cv->recv; # wait until user enters /^q/i
775		1154
776	=head1 REAL-WORLD EXAMPLE	1155	=head1 REAL-WORLD EXAMPLE
777		1156
778	Consider the L<Net::FCP> module. It features (among others) the following	1157	Consider the L<Net::FCP> module. It features (among others) the following
779	API calls, which are to freenet what HTTP GET requests are to http:	1158	API calls, which are to freenet what HTTP GET requests are to http:
…		…
829	syswrite $txn->{fh}, $txn->{request}	1208	syswrite $txn->{fh}, $txn->{request}
830	or die "connection or write error";	1209	or die "connection or write error";
831	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });	1210	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
832		1211
833	Again, C<fh_ready_r> waits till all data has arrived, and then stores the	1212	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
834	result and signals any possible waiters that the request ahs finished:	1213	result and signals any possible waiters that the request has finished:
835		1214
836	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};	1215	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
837		1216
838	if (end-of-file or data complete) {	1217	if (end-of-file or data complete) {
839	$txn->{result} = $txn->{buf};	1218	$txn->{result} = $txn->{buf};
840	$txn->{finished}->broadcast;	1219	$txn->{finished}->send;
841	$txb->{cb}->($txn) of $txn->{cb}; # also call callback	1220	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
842	}	1221	}
843		1222
844	The C<result> method, finally, just waits for the finished signal (if the	1223	The C<result> method, finally, just waits for the finished signal (if the
845	request was already finished, it doesn't wait, of course, and returns the	1224	request was already finished, it doesn't wait, of course, and returns the
846	data:	1225	data:
847		1226
848	$txn->{finished}->wait;	1227	$txn->{finished}->recv;
849	return $txn->{result};	1228	return $txn->{result};
850		1229
851	The actual code goes further and collects all errors (C<die>s, exceptions)	1230	The actual code goes further and collects all errors (C<die>s, exceptions)
852	that occured during request processing. The C<result> method detects	1231	that occurred during request processing. The C<result> method detects
853	whether an exception as thrown (it is stored inside the $txn object)	1232	whether an exception as thrown (it is stored inside the $txn object)
854	and just throws the exception, which means connection errors and other	1233	and just throws the exception, which means connection errors and other
855	problems get reported tot he code that tries to use the result, not in a	1234	problems get reported tot he code that tries to use the result, not in a
856	random callback.	1235	random callback.
857		1236
…		…
888		1267
889	my $quit = AnyEvent->condvar;	1268	my $quit = AnyEvent->condvar;
890		1269
891	$fcp->txn_client_get ($url)->cb (sub {	1270	$fcp->txn_client_get ($url)->cb (sub {
892	...	1271	...
893	$quit->broadcast;	1272	$quit->send;
894	});	1273	});
895		1274
896	$quit->wait;	1275	$quit->recv;
897		1276
898		1277
899	=head1 BENCHMARK	1278	=head1 BENCHMARKS
900		1279
901	To give you an idea of the performance and overheads that AnyEvent adds	1280	To give you an idea of the performance and overheads that AnyEvent adds
902	over the event loops themselves (and to give you an impression of the	1281	over the event loops themselves and to give you an impression of the speed
903	speed of various event loops), here is a benchmark of various supported	1282	of various event loops I prepared some benchmarks.
904	event models natively and with anyevent. The benchmark creates a lot of	1283
905	timers (with a zero timeout) and I/O watchers (watching STDOUT, a pty, to	1284	=head2 BENCHMARKING ANYEVENT OVERHEAD
		1285
		1286	Here is a benchmark of various supported event models used natively and
		1287	through AnyEvent. The benchmark creates a lot of timers (with a zero
		1288	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
906	become writable, which it is), lets them fire exactly once and destroys	1289	which it is), lets them fire exactly once and destroys them again.
907	them again.
908		1290
909	Rewriting the benchmark to use many different sockets instead of using	1291	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
910	the same filehandle for all I/O watchers results in a much longer runtime	1292	distribution.
911	(socket creation is expensive), but qualitatively the same figures, so it
912	was not used.
913		1293
914	=head2 Explanation of the columns	1294	=head3 Explanation of the columns
915		1295
916	I<watcher> is the number of event watchers created/destroyed. Since	1296	I<watcher> is the number of event watchers created/destroyed. Since
917	different event models feature vastly different performances, each event	1297	different event models feature vastly different performances, each event
918	loop was given a number of watchers so that overall runtime is acceptable	1298	loop was given a number of watchers so that overall runtime is acceptable
919	and similar between tested event loop (and keep them from crashing): Glib	1299	and similar between tested event loop (and keep them from crashing): Glib
…		…
929	all watchers, to avoid adding memory overhead. That means closure creation	1309	all watchers, to avoid adding memory overhead. That means closure creation
930	and memory usage is not included in the figures.	1310	and memory usage is not included in the figures.
931		1311
932	I<invoke> is the time, in microseconds, used to invoke a simple	1312	I<invoke> is the time, in microseconds, used to invoke a simple
933	callback. The callback simply counts down a Perl variable and after it was	1313	callback. The callback simply counts down a Perl variable and after it was
934	invoked "watcher" times, it would C<< ->broadcast >> a condvar once to	1314	invoked "watcher" times, it would C<< ->send >> a condvar once to
935	signal the end of this phase.	1315	signal the end of this phase.
936		1316
937	I<destroy> is the time, in microseconds, that it takes to destroy a single	1317	I<destroy> is the time, in microseconds, that it takes to destroy a single
938	watcher.	1318	watcher.
939		1319
940	=head2 Results	1320	=head3 Results
941		1321
942	name watchers bytes create invoke destroy comment	1322	name watchers bytes create invoke destroy comment
943	EV/EV 400000 244 0.56 0.46 0.31 EV native interface	1323	EV/EV 400000 244 0.56 0.46 0.31 EV native interface
944	EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers	1324	EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers
945	CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal	1325	CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal
946	Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation	1326	Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation
947	Event/Event 16000 516 31.88 31.30 0.85 Event native interface	1327	Event/Event 16000 516 31.88 31.30 0.85 Event native interface
948	Event/Any 16000 936 39.17 33.63 1.43 Event + AnyEvent watchers	1328	Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers
949	Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour	1329	Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour
950	Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers	1330	Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers
951	POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event	1331	POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event
952	POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select	1332	POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select
953		1333
954	=head2 Discussion	1334	=head3 Discussion
955		1335
956	The benchmark does I<not> measure scalability of the event loop very	1336	The benchmark does I<not> measure scalability of the event loop very
957	well. For example, a select-based event loop (such as the pure perl one)	1337	well. For example, a select-based event loop (such as the pure perl one)
958	can never compete with an event loop that uses epoll when the number of	1338	can never compete with an event loop that uses epoll when the number of
959	file descriptors grows high. In this benchmark, all events become ready at	1339	file descriptors grows high. In this benchmark, all events become ready at
960	the same time, so select/poll-based implementations get an unnatural speed	1340	the same time, so select/poll-based implementations get an unnatural speed
961	boost.	1341	boost.
962		1342
		1343	Also, note that the number of watchers usually has a nonlinear effect on
		1344	overall speed, that is, creating twice as many watchers doesn't take twice
		1345	the time - usually it takes longer. This puts event loops tested with a
		1346	higher number of watchers at a disadvantage.
		1347
		1348	To put the range of results into perspective, consider that on the
		1349	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		1350	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		1351	cycles with POE.
		1352
963	C<EV> is the sole leader regarding speed and memory use, which are both	1353	C<EV> is the sole leader regarding speed and memory use, which are both
964	maximal/minimal, respectively. Even when going through AnyEvent, it uses	1354	maximal/minimal, respectively. Even when going through AnyEvent, it uses
965	far less memory than any other event loop and is still faster than Event	1355	far less memory than any other event loop and is still faster than Event
966	natively.	1356	natively.
967		1357
…		…
970	interpreter and the backend itself). Nevertheless this shows that it	1360	interpreter and the backend itself). Nevertheless this shows that it
971	adds very little overhead in itself. Like any select-based backend its	1361	adds very little overhead in itself. Like any select-based backend its
972	performance becomes really bad with lots of file descriptors (and few of	1362	performance becomes really bad with lots of file descriptors (and few of
973	them active), of course, but this was not subject of this benchmark.	1363	them active), of course, but this was not subject of this benchmark.
974		1364
975	The C<Event> module has a relatively high setup and callback invocation cost,	1365	The C<Event> module has a relatively high setup and callback invocation
976	but overall scores on the third place.	1366	cost, but overall scores in on the third place.
977		1367
978	C<Glib>'s memory usage is quite a bit bit higher, but it features a	1368	C<Glib>'s memory usage is quite a bit higher, but it features a
979	faster callback invocation and overall ends up in the same class as	1369	faster callback invocation and overall ends up in the same class as
980	C<Event>. However, Glib scales extremely badly, doubling the number of	1370	C<Event>. However, Glib scales extremely badly, doubling the number of
981	watchers increases the processing time by more than a factor of four,	1371	watchers increases the processing time by more than a factor of four,
982	making it completely unusable when using larger numbers of watchers	1372	making it completely unusable when using larger numbers of watchers
983	(note that only a single file descriptor was used in the benchmark, so	1373	(note that only a single file descriptor was used in the benchmark, so
…		…
986	The C<Tk> adaptor works relatively well. The fact that it crashes with	1376	The C<Tk> adaptor works relatively well. The fact that it crashes with
987	more than 2000 watchers is a big setback, however, as correctness takes	1377	more than 2000 watchers is a big setback, however, as correctness takes
988	precedence over speed. Nevertheless, its performance is surprising, as the	1378	precedence over speed. Nevertheless, its performance is surprising, as the
989	file descriptor is dup()ed for each watcher. This shows that the dup()	1379	file descriptor is dup()ed for each watcher. This shows that the dup()
990	employed by some adaptors is not a big performance issue (it does incur a	1380	employed by some adaptors is not a big performance issue (it does incur a
991	hidden memory cost inside the kernel, though, that is not reflected in the	1381	hidden memory cost inside the kernel which is not reflected in the figures
992	figures above).	1382	above).
993		1383
994	C<POE>, regardless of underlying event loop (whether using its pure perl	1384	C<POE>, regardless of underlying event loop (whether using its pure perl
995	select-based backend or the Event module) shows abysmal performance and	1385	select-based backend or the Event module, the POE-EV backend couldn't
		1386	be tested because it wasn't working) shows abysmal performance and
996	memory usage: Watchers use almost 30 times as much memory as EV watchers,	1387	memory usage with AnyEvent: Watchers use almost 30 times as much memory
997	and 10 times as much memory as both Event or EV via AnyEvent. Watcher	1388	as EV watchers, and 10 times as much memory as Event (the high memory
		1389	requirements are caused by requiring a session for each watcher). Watcher
998	invocation is almost 900 times slower than with AnyEvent's pure perl	1390	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		1391	implementation.
		1392
999	implementation. The design of the POE adaptor class in AnyEvent can not	1393	The design of the POE adaptor class in AnyEvent can not really account
1000	really account for this, as session creation overhead is small compared	1394	for the performance issues, though, as session creation overhead is
1001	to execution of the state machine, which is coded pretty optimally within	1395	small compared to execution of the state machine, which is coded pretty
1002	L<AnyEvent::Impl::POE>. POE simply seems to be abysmally slow.	1396	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		1397	using multiple sessions is not a good approach, especially regarding
		1398	memory usage, even the author of POE could not come up with a faster
		1399	design).
1003		1400
1004	=head2 Summary	1401	=head3 Summary
1005		1402
		1403	=over 4
		1404
1006	Using EV through AnyEvent is faster than any other event loop, but most	1405	=item * Using EV through AnyEvent is faster than any other event loop
1007	event loops have acceptable performance with or without AnyEvent.	1406	(even when used without AnyEvent), but most event loops have acceptable
		1407	performance with or without AnyEvent.
1008		1408
1009	The overhead AnyEvent adds is usually much smaller than the overhead of	1409	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
1010	the actual event loop, only with extremely fast event loops such as the EV	1410	the actual event loop, only with extremely fast event loops such as EV
1011	adds AnyEvent significant overhead.	1411	adds AnyEvent significant overhead.
1012		1412
1013	And you should simply avoid POE like the plague if you want performance or	1413	=item * You should avoid POE like the plague if you want performance or
1014	reasonable memory usage.	1414	reasonable memory usage.
1015		1415
		1416	=back
		1417
		1418	=head2 BENCHMARKING THE LARGE SERVER CASE
		1419
		1420	This benchmark actually benchmarks the event loop itself. It works by
		1421	creating a number of "servers": each server consists of a socket pair, a
		1422	timeout watcher that gets reset on activity (but never fires), and an I/O
		1423	watcher waiting for input on one side of the socket. Each time the socket
		1424	watcher reads a byte it will write that byte to a random other "server".
		1425
		1426	The effect is that there will be a lot of I/O watchers, only part of which
		1427	are active at any one point (so there is a constant number of active
		1428	fds for each loop iteration, but which fds these are is random). The
		1429	timeout is reset each time something is read because that reflects how
		1430	most timeouts work (and puts extra pressure on the event loops).
		1431
		1432	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		1433	(1%) are active. This mirrors the activity of large servers with many
		1434	connections, most of which are idle at any one point in time.
		1435
		1436	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		1437	distribution.
		1438
		1439	=head3 Explanation of the columns
		1440
		1441	I<sockets> is the number of sockets, and twice the number of "servers" (as
		1442	each server has a read and write socket end).
		1443
		1444	I<create> is the time it takes to create a socket pair (which is
		1445	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		1446
		1447	I<request>, the most important value, is the time it takes to handle a
		1448	single "request", that is, reading the token from the pipe and forwarding
		1449	it to another server. This includes deleting the old timeout and creating
		1450	a new one that moves the timeout into the future.
		1451
		1452	=head3 Results
		1453
		1454	name sockets create request
		1455	EV 20000 69.01 11.16
		1456	Perl 20000 73.32 35.87
		1457	Event 20000 212.62 257.32
		1458	Glib 20000 651.16 1896.30
		1459	POE 20000 349.67 12317.24 uses POE::Loop::Event
		1460
		1461	=head3 Discussion
		1462
		1463	This benchmark I<does> measure scalability and overall performance of the
		1464	particular event loop.
		1465
		1466	EV is again fastest. Since it is using epoll on my system, the setup time
		1467	is relatively high, though.
		1468
		1469	Perl surprisingly comes second. It is much faster than the C-based event
		1470	loops Event and Glib.
		1471
		1472	Event suffers from high setup time as well (look at its code and you will
		1473	understand why). Callback invocation also has a high overhead compared to
		1474	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		1475	uses select or poll in basically all documented configurations.
		1476
		1477	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		1478	clearly fails to perform with many filehandles or in busy servers.
		1479
		1480	POE is still completely out of the picture, taking over 1000 times as long
		1481	as EV, and over 100 times as long as the Perl implementation, even though
		1482	it uses a C-based event loop in this case.
		1483
		1484	=head3 Summary
		1485
		1486	=over 4
		1487
		1488	=item * The pure perl implementation performs extremely well.
		1489
		1490	=item * Avoid Glib or POE in large projects where performance matters.
		1491
		1492	=back
		1493
		1494	=head2 BENCHMARKING SMALL SERVERS
		1495
		1496	While event loops should scale (and select-based ones do not...) even to
		1497	large servers, most programs we (or I :) actually write have only a few
		1498	I/O watchers.
		1499
		1500	In this benchmark, I use the same benchmark program as in the large server
		1501	case, but it uses only eight "servers", of which three are active at any
		1502	one time. This should reflect performance for a small server relatively
		1503	well.
		1504
		1505	The columns are identical to the previous table.
		1506
		1507	=head3 Results
		1508
		1509	name sockets create request
		1510	EV 16 20.00 6.54
		1511	Perl 16 25.75 12.62
		1512	Event 16 81.27 35.86
		1513	Glib 16 32.63 15.48
		1514	POE 16 261.87 276.28 uses POE::Loop::Event
		1515
		1516	=head3 Discussion
		1517
		1518	The benchmark tries to test the performance of a typical small
		1519	server. While knowing how various event loops perform is interesting, keep
		1520	in mind that their overhead in this case is usually not as important, due
		1521	to the small absolute number of watchers (that is, you need efficiency and
		1522	speed most when you have lots of watchers, not when you only have a few of
		1523	them).
		1524
		1525	EV is again fastest.
		1526
		1527	Perl again comes second. It is noticeably faster than the C-based event
		1528	loops Event and Glib, although the difference is too small to really
		1529	matter.
		1530
		1531	POE also performs much better in this case, but is is still far behind the
		1532	others.
		1533
		1534	=head3 Summary
		1535
		1536	=over 4
		1537
		1538	=item * C-based event loops perform very well with small number of
		1539	watchers, as the management overhead dominates.
		1540
		1541	=back
		1542
1016		1543
1017	=head1 FORK	1544	=head1 FORK
1018		1545
1019	Most event libraries are not fork-safe. The ones who are usually are	1546	Most event libraries are not fork-safe. The ones who are usually are
1020	because they are so inefficient. Only L<EV> is fully fork-aware.	1547	because they rely on inefficient but fork-safe C<select> or C<poll>
		1548	calls. Only L<EV> is fully fork-aware.
1021		1549
1022	If you have to fork, you must either do so I<before> creating your first	1550	If you have to fork, you must either do so I<before> creating your first
1023	watcher OR you must not use AnyEvent at all in the child.	1551	watcher OR you must not use AnyEvent at all in the child.
1024		1552
1025		1553
…		…
1037		1565
1038	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }	1566	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
1039		1567
1040	use AnyEvent;	1568	use AnyEvent;
1041		1569
		1570	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		1571	be used to probe what backend is used and gain other information (which is
		1572	probably even less useful to an attacker than PERL_ANYEVENT_MODEL).
		1573
1042		1574
1043	=head1 SEE ALSO	1575	=head1 SEE ALSO
1044		1576
1045	Event modules: L<Coro::EV>, L<EV>, L<EV::Glib>, L<Glib::EV>,	1577	Utility functions: L<AnyEvent::Util>.
1046	L<Coro::Event>, L<Event>, L<Glib::Event>, L<Glib>, L<Coro>, L<Tk>,	1578
		1579	Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
1047	L<Event::Lib>, L<Qt>, L<POE>.	1580	L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
1048		1581
1049	Implementations: L<AnyEvent::Impl::CoroEV>, L<AnyEvent::Impl::EV>,	1582	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
1050	L<AnyEvent::Impl::CoroEvent>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>,	1583	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
1051	L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>, L<AnyEvent::Impl::EventLib>,	1584	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
1052	L<AnyEvent::Impl::Qt>, L<AnyEvent::Impl::POE>.	1585	L<AnyEvent::Impl::POE>.
1053		1586
		1587	Non-blocking file handles, sockets, TCP clients and
		1588	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>.
		1589
		1590	Asynchronous DNS: L<AnyEvent::DNS>.
		1591
		1592	Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>,
		1593
1054	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>.	1594	Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>.
1055		1595
1056		1596
1057	=head1 AUTHOR	1597	=head1 AUTHOR
1058		1598
1059	Marc Lehmann <schmorp@schmorp.de>	1599	Marc Lehmann <schmorp@schmorp.de>

Diff Legend

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

Comparing AnyEvent/lib/AnyEvent.pm (file contents): Revision 1.86 by root, Fri Apr 25 14:01:48 2008 UTC vs. Revision 1.135 by root, Sun May 25 04:49:01 2008 UTC

Diff Legend

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.86 by root, Fri Apr 25 14:01:48 2008 UTC vs.
Revision 1.135 by root, Sun May 25 04:49:01 2008 UTC