ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent/lib/AnyEvent.pm
(Generate patch)

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.96 by root, Sat Apr 26 11:16:16 2008 UTC vs.
Revision 1.147 by root, Fri May 30 21:43:26 2008 UTC

1=head1 NAME 1=head1 => NAME
2 2
3AnyEvent - provide framework for multiple event loops 3AnyEvent - provide framework for multiple event loops
4 4
5EV, Event, Coro::EV, Coro::Event, Glib, Tk, Perl, Event::Lib, Qt, POE - various supported event loops 5EV, 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
25Glib, POE, IO::Async, Event... CPAN offers event models by the dozen 25Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
26nowadays. So what is different about AnyEvent? 26nowadays. So what is different about AnyEvent?
48isn't itself. What's worse, all the potential users of your module are 48isn't itself. What's worse, all the potential users of your module are
49I<also> forced to use the same event loop you use. 49I<also> forced to use the same event loop you use.
50 50
51AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works 51AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
52fine. AnyEvent + Tk works fine etc. etc. but none of these work together 52fine. AnyEvent + Tk works fine etc. etc. but none of these work together
53with the rest: POE + IO::Async? no go. Tk + Event? no go. Again: if 53with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
54your module uses one of those, every user of your module has to use it, 54your module uses one of those, every user of your module has to use it,
55too. But if your module uses AnyEvent, it works transparently with all 55too. But if your module uses AnyEvent, it works transparently with all
56event models it supports (including stuff like POE and IO::Async, as long 56event models it supports (including stuff like POE and IO::Async, as long
57as those use one of the supported event loops. It is trivial to add new 57as those use one of the supported event loops. It is trivial to add new
58event loops to AnyEvent, too, so it is future-proof). 58event loops to AnyEvent, too, so it is future-proof).
59 59
60In addition to being free of having to use I<the one and only true event 60In addition to being free of having to use I<the one and only true event
61model>, AnyEvent also is free of bloat and policy: with POE or similar 61model>, AnyEvent also is free of bloat and policy: with POE or similar
62modules, you get an enourmous amount of code and strict rules you have to 62modules, you get an enormous amount of code and strict rules you have to
63follow. AnyEvent, on the other hand, is lean and up to the point, by only 63follow. AnyEvent, on the other hand, is lean and up to the point, by only
64offering the functionality that is necessary, in as thin as a wrapper as 64offering the functionality that is necessary, in as thin as a wrapper as
65technically possible. 65technically possible.
66 66
67Of course, AnyEvent comes with a big (and fully optional!) toolbox
68of useful functionality, such as an asynchronous DNS resolver, 100%
69non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
70such as Windows) and lots of real-world knowledge and workarounds for
71platform bugs and differences.
72
67Of course, if you want lots of policy (this can arguably be somewhat 73Now, if you I<do want> lots of policy (this can arguably be somewhat
68useful) and you want to force your users to use the one and only event 74useful) and you want to force your users to use the one and only event
69model, you should I<not> use this module. 75model, you should I<not> use this module.
70
71 76
72=head1 DESCRIPTION 77=head1 DESCRIPTION
73 78
74L<AnyEvent> provides an identical interface to multiple event loops. This 79L<AnyEvent> provides an identical interface to multiple event loops. This
75allows module authors to utilise an event loop without forcing module 80allows module authors to utilise an event loop without forcing module
79The interface itself is vaguely similar, but not identical to the L<Event> 84The interface itself is vaguely similar, but not identical to the L<Event>
80module. 85module.
81 86
82During the first call of any watcher-creation method, the module tries 87During the first call of any watcher-creation method, the module tries
83to detect the currently loaded event loop by probing whether one of the 88to detect the currently loaded event loop by probing whether one of the
84following modules is already loaded: L<Coro::EV>, L<Coro::Event>, L<EV>, 89following modules is already loaded: L<EV>,
85L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>, 90L<Event>, L<Glib>, L<AnyEvent::Impl::Perl>, L<Tk>, L<Event::Lib>, L<Qt>,
86L<POE>. The first one found is used. If none are found, the module tries 91L<POE>. The first one found is used. If none are found, the module tries
87to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl 92to load these modules (excluding Tk, Event::Lib, Qt and POE as the pure perl
88adaptor should always succeed) in the order given. The first one that can 93adaptor should always succeed) in the order given. The first one that can
89be successfully loaded will be used. If, after this, still none could be 94be successfully loaded will be used. If, after this, still none could be
103starts using it, all bets are off. Maybe you should tell their authors to 108starts using it, all bets are off. Maybe you should tell their authors to
104use AnyEvent so their modules work together with others seamlessly... 109use AnyEvent so their modules work together with others seamlessly...
105 110
106The pure-perl implementation of AnyEvent is called 111The pure-perl implementation of AnyEvent is called
107C<AnyEvent::Impl::Perl>. Like other event modules you can load it 112C<AnyEvent::Impl::Perl>. Like other event modules you can load it
108explicitly. 113explicitly and enjoy the high availability of that event loop :)
109 114
110=head1 WATCHERS 115=head1 WATCHERS
111 116
112AnyEvent has the central concept of a I<watcher>, which is an object that 117AnyEvent has the central concept of a I<watcher>, which is an object that
113stores relevant data for each kind of event you are waiting for, such as 118stores relevant data for each kind of event you are waiting for, such as
114the callback to call, the filehandle to watch, etc. 119the callback to call, the file handle to watch, etc.
115 120
116These watchers are normal Perl objects with normal Perl lifetime. After 121These watchers are normal Perl objects with normal Perl lifetime. After
117creating a watcher it will immediately "watch" for events and invoke the 122creating a watcher it will immediately "watch" for events and invoke the
118callback when the event occurs (of course, only when the event model 123callback when the event occurs (of course, only when the event model
119is in control). 124is in control).
228timers. 233timers.
229 234
230AnyEvent always prefers relative timers, if available, matching the 235AnyEvent always prefers relative timers, if available, matching the
231AnyEvent API. 236AnyEvent API.
232 237
238AnyEvent has two additional methods that return the "current time":
239
240=over 4
241
242=item AnyEvent->time
243
244This returns the "current wallclock time" as a fractional number of
245seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
246return, and the result is guaranteed to be compatible with those).
247
248It progresses independently of any event loop processing, i.e. each call
249will check the system clock, which usually gets updated frequently.
250
251=item AnyEvent->now
252
253This also returns the "current wallclock time", but unlike C<time>, above,
254this value might change only once per event loop iteration, depending on
255the event loop (most return the same time as C<time>, above). This is the
256time that AnyEvent's timers get scheduled against.
257
258I<In almost all cases (in all cases if you don't care), this is the
259function to call when you want to know the current time.>
260
261This function is also often faster then C<< AnyEvent->time >>, and
262thus the preferred method if you want some timestamp (for example,
263L<AnyEvent::Handle> uses this to update it's activity timeouts).
264
265The rest of this section is only of relevance if you try to be very exact
266with your timing, you can skip it without bad conscience.
267
268For a practical example of when these times differ, consider L<Event::Lib>
269and L<EV> and the following set-up:
270
271The event loop is running and has just invoked one of your callback at
272time=500 (assume no other callbacks delay processing). In your callback,
273you wait a second by executing C<sleep 1> (blocking the process for a
274second) and then (at time=501) you create a relative timer that fires
275after three seconds.
276
277With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
278both return C<501>, because that is the current time, and the timer will
279be scheduled to fire at time=504 (C<501> + C<3>).
280
281With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
282time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
283last event processing phase started. With L<EV>, your timer gets scheduled
284to run at time=503 (C<500> + C<3>).
285
286In one sense, L<Event::Lib> is more exact, as it uses the current time
287regardless of any delays introduced by event processing. However, most
288callbacks do not expect large delays in processing, so this causes a
289higher drift (and a lot more system calls to get the current time).
290
291In another sense, L<EV> is more exact, as your timer will be scheduled at
292the same time, regardless of how long event processing actually took.
293
294In either case, if you care (and in most cases, you don't), then you
295can get whatever behaviour you want with any event loop, by taking the
296difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
297account.
298
299=back
300
233=head2 SIGNAL WATCHERS 301=head2 SIGNAL WATCHERS
234 302
235You can watch for signals using a signal watcher, C<signal> is the signal 303You can watch for signals using a signal watcher, C<signal> is the signal
236I<name> without any C<SIG> prefix, C<cb> is the Perl callback to 304I<name> without any C<SIG> prefix, C<cb> is the Perl callback to
237be invoked whenever a signal occurs. 305be invoked whenever a signal occurs.
238 306
239Although the callback might get passed parameters, their value and 307Although the callback might get passed parameters, their value and
240presence is undefined and you cannot rely on them. Portable AnyEvent 308presence is undefined and you cannot rely on them. Portable AnyEvent
241callbacks cannot use arguments passed to signal watcher callbacks. 309callbacks cannot use arguments passed to signal watcher callbacks.
242 310
243Multiple signal occurances can be clumped together into one callback 311Multiple signal occurrences can be clumped together into one callback
244invocation, and callback invocation will be synchronous. synchronous means 312invocation, and callback invocation will be synchronous. Synchronous means
245that it might take a while until the signal gets handled by the process, 313that it might take a while until the signal gets handled by the process,
246but it is guarenteed not to interrupt any other callbacks. 314but it is guaranteed not to interrupt any other callbacks.
247 315
248The main advantage of using these watchers is that you can share a signal 316The main advantage of using these watchers is that you can share a signal
249between multiple watchers. 317between multiple watchers.
250 318
251This watcher might use C<%SIG>, so programs overwriting those signals 319This watcher might use C<%SIG>, so programs overwriting those signals
280 348
281Example: fork a process and wait for it 349Example: fork a process and wait for it
282 350
283 my $done = AnyEvent->condvar; 351 my $done = AnyEvent->condvar;
284 352
285 AnyEvent::detect; # force event module to be initialised
286
287 my $pid = fork or exit 5; 353 my $pid = fork or exit 5;
288 354
289 my $w = AnyEvent->child ( 355 my $w = AnyEvent->child (
290 pid => $pid, 356 pid => $pid,
291 cb => sub { 357 cb => sub {
292 my ($pid, $status) = @_; 358 my ($pid, $status) = @_;
293 warn "pid $pid exited with status $status"; 359 warn "pid $pid exited with status $status";
294 $done->broadcast; 360 $done->send;
295 }, 361 },
296 ); 362 );
297 363
298 # do something else, then wait for process exit 364 # do something else, then wait for process exit
299 $done->wait; 365 $done->recv;
300 366
301=head2 CONDITION VARIABLES 367=head2 CONDITION VARIABLES
302 368
369If you are familiar with some event loops you will know that all of them
370require you to run some blocking "loop", "run" or similar function that
371will actively watch for new events and call your callbacks.
372
373AnyEvent is different, it expects somebody else to run the event loop and
374will only block when necessary (usually when told by the user).
375
376The instrument to do that is called a "condition variable", so called
377because they represent a condition that must become true.
378
303Condition variables can be created by calling the C<< AnyEvent->condvar >> 379Condition variables can be created by calling the C<< AnyEvent->condvar
304method without any arguments. 380>> method, usually without arguments. The only argument pair allowed is
381C<cb>, which specifies a callback to be called when the condition variable
382becomes true.
305 383
306A condition variable waits for a condition - precisely that the C<< 384After creation, the condition variable is "false" until it becomes "true"
307->broadcast >> method has been called. 385by calling the C<send> method (or calling the condition variable as if it
386were a callback, read about the caveats in the description for the C<<
387->send >> method).
308 388
309They are very useful to signal that a condition has been fulfilled, for 389Condition variables are similar to callbacks, except that you can
390optionally wait for them. They can also be called merge points - points
391in time where multiple outstanding events have been processed. And yet
392another way to call them is transactions - each condition variable can be
393used to represent a transaction, which finishes at some point and delivers
394a result.
395
396Condition variables are very useful to signal that something has finished,
310example, if you write a module that does asynchronous http requests, 397for example, if you write a module that does asynchronous http requests,
311then a condition variable would be the ideal candidate to signal the 398then a condition variable would be the ideal candidate to signal the
312availability of results. 399availability of results. The user can either act when the callback is
400called or can synchronously C<< ->recv >> for the results.
313 401
314You can also use condition variables to block your main program until 402You can also use them to simulate traditional event loops - for example,
315an event occurs - for example, you could C<< ->wait >> in your main 403you can block your main program until an event occurs - for example, you
316program until the user clicks the Quit button in your app, which would C<< 404could C<< ->recv >> in your main program until the user clicks the Quit
317->broadcast >> the "quit" event. 405button of your app, which would C<< ->send >> the "quit" event.
318 406
319Note that condition variables recurse into the event loop - if you have 407Note that condition variables recurse into the event loop - if you have
320two pirces of code that call C<< ->wait >> in a round-robbin fashion, you 408two pieces of code that call C<< ->recv >> in a round-robin fashion, you
321lose. Therefore, condition variables are good to export to your caller, but 409lose. Therefore, condition variables are good to export to your caller, but
322you should avoid making a blocking wait yourself, at least in callbacks, 410you should avoid making a blocking wait yourself, at least in callbacks,
323as this asks for trouble. 411as this asks for trouble.
324 412
325This object has two methods: 413Condition variables are represented by hash refs in perl, and the keys
414used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
415easy (it is often useful to build your own transaction class on top of
416AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
417it's C<new> method in your own C<new> method.
418
419There are two "sides" to a condition variable - the "producer side" which
420eventually calls C<< -> send >>, and the "consumer side", which waits
421for the send to occur.
422
423Example: wait for a timer.
424
425 # wait till the result is ready
426 my $result_ready = AnyEvent->condvar;
427
428 # do something such as adding a timer
429 # or socket watcher the calls $result_ready->send
430 # when the "result" is ready.
431 # in this case, we simply use a timer:
432 my $w = AnyEvent->timer (
433 after => 1,
434 cb => sub { $result_ready->send },
435 );
436
437 # this "blocks" (while handling events) till the callback
438 # calls send
439 $result_ready->recv;
440
441Example: wait for a timer, but take advantage of the fact that
442condition variables are also code references.
443
444 my $done = AnyEvent->condvar;
445 my $delay = AnyEvent->timer (after => 5, cb => $done);
446 $done->recv;
447
448=head3 METHODS FOR PRODUCERS
449
450These methods should only be used by the producing side, i.e. the
451code/module that eventually sends the signal. Note that it is also
452the producer side which creates the condvar in most cases, but it isn't
453uncommon for the consumer to create it as well.
326 454
327=over 4 455=over 4
328 456
457=item $cv->send (...)
458
459Flag the condition as ready - a running C<< ->recv >> and all further
460calls to C<recv> will (eventually) return after this method has been
461called. If nobody is waiting the send will be remembered.
462
463If a callback has been set on the condition variable, it is called
464immediately from within send.
465
466Any arguments passed to the C<send> call will be returned by all
467future C<< ->recv >> calls.
468
469Condition variables are overloaded so one can call them directly
470(as a code reference). Calling them directly is the same as calling
471C<send>. Note, however, that many C-based event loops do not handle
472overloading, so as tempting as it may be, passing a condition variable
473instead of a callback does not work. Both the pure perl and EV loops
474support overloading, however, as well as all functions that use perl to
475invoke a callback (as in L<AnyEvent::Socket> and L<AnyEvent::DNS> for
476example).
477
478=item $cv->croak ($error)
479
480Similar to send, but causes all call's to C<< ->recv >> to invoke
481C<Carp::croak> with the given error message/object/scalar.
482
483This can be used to signal any errors to the condition variable
484user/consumer.
485
486=item $cv->begin ([group callback])
487
329=item $cv->wait 488=item $cv->end
330 489
331Wait (blocking if necessary) until the C<< ->broadcast >> method has been 490These two methods are EXPERIMENTAL and MIGHT CHANGE.
491
492These two methods can be used to combine many transactions/events into
493one. For example, a function that pings many hosts in parallel might want
494to use a condition variable for the whole process.
495
496Every call to C<< ->begin >> will increment a counter, and every call to
497C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
498>>, the (last) callback passed to C<begin> will be executed. That callback
499is I<supposed> to call C<< ->send >>, but that is not required. If no
500callback was set, C<send> will be called without any arguments.
501
502Let's clarify this with the ping example:
503
504 my $cv = AnyEvent->condvar;
505
506 my %result;
507 $cv->begin (sub { $cv->send (\%result) });
508
509 for my $host (@list_of_hosts) {
510 $cv->begin;
511 ping_host_then_call_callback $host, sub {
512 $result{$host} = ...;
513 $cv->end;
514 };
515 }
516
517 $cv->end;
518
519This code fragment supposedly pings a number of hosts and calls
520C<send> after results for all then have have been gathered - in any
521order. To achieve this, the code issues a call to C<begin> when it starts
522each ping request and calls C<end> when it has received some result for
523it. Since C<begin> and C<end> only maintain a counter, the order in which
524results arrive is not relevant.
525
526There is an additional bracketing call to C<begin> and C<end> outside the
527loop, which serves two important purposes: first, it sets the callback
528to be called once the counter reaches C<0>, and second, it ensures that
529C<send> is called even when C<no> hosts are being pinged (the loop
530doesn't execute once).
531
532This is the general pattern when you "fan out" into multiple subrequests:
533use an outer C<begin>/C<end> pair to set the callback and ensure C<end>
534is called at least once, and then, for each subrequest you start, call
535C<begin> and for each subrequest you finish, call C<end>.
536
537=back
538
539=head3 METHODS FOR CONSUMERS
540
541These methods should only be used by the consuming side, i.e. the
542code awaits the condition.
543
544=over 4
545
546=item $cv->recv
547
548Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
332called on c<$cv>, while servicing other watchers normally. 549>> methods have been called on c<$cv>, while servicing other watchers
550normally.
333 551
334You can only wait once on a condition - additional calls will return 552You can only wait once on a condition - additional calls are valid but
335immediately. 553will return immediately.
554
555If an error condition has been set by calling C<< ->croak >>, then this
556function will call C<croak>.
557
558In list context, all parameters passed to C<send> will be returned,
559in scalar context only the first one will be returned.
336 560
337Not all event models support a blocking wait - some die in that case 561Not 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 562(programs might want to do that to stay interactive), so I<if you are
339using this from a module, never require a blocking wait>, but let the 563using this from a module, never require a blocking wait>, but let the
340caller decide whether the call will block or not (for example, by coupling 564caller decide whether the call will block or not (for example, by coupling
341condition variables with some kind of request results and supporting 565condition variables with some kind of request results and supporting
342callbacks so the caller knows that getting the result will not block, 566callbacks so the caller knows that getting the result will not block,
343while still suppporting blocking waits if the caller so desires). 567while still supporting blocking waits if the caller so desires).
344 568
345Another reason I<never> to C<< ->wait >> in a module is that you cannot 569Another reason I<never> to C<< ->recv >> in a module is that you cannot
346sensibly have two C<< ->wait >>'s in parallel, as that would require 570sensibly have two C<< ->recv >>'s in parallel, as that would require
347multiple interpreters or coroutines/threads, none of which C<AnyEvent> 571multiple interpreters or coroutines/threads, none of which C<AnyEvent>
348can supply (the coroutine-aware backends L<AnyEvent::Impl::CoroEV> and 572can supply.
349L<AnyEvent::Impl::CoroEvent> explicitly support concurrent C<< ->wait >>'s
350from different coroutines, however).
351 573
352=item $cv->broadcast 574The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
575fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
576versions and also integrates coroutines into AnyEvent, making blocking
577C<< ->recv >> calls perfectly safe as long as they are done from another
578coroutine (one that doesn't run the event loop).
353 579
354Flag the condition as ready - a running C<< ->wait >> and all further 580You can ensure that C<< -recv >> never blocks by setting a callback and
355calls to C<wait> will (eventually) return after this method has been 581only calling C<< ->recv >> from within that callback (or at a later
356called. If nobody is waiting the broadcast will be remembered.. 582time). This will work even when the event loop does not support blocking
583waits otherwise.
584
585=item $bool = $cv->ready
586
587Returns true when the condition is "true", i.e. whether C<send> or
588C<croak> have been called.
589
590=item $cb = $cv->cb ([new callback])
591
592This is a mutator function that returns the callback set and optionally
593replaces it before doing so.
594
595The callback will be called when the condition becomes "true", i.e. when
596C<send> or C<croak> are called. Calling C<recv> inside the callback
597or at any later time is guaranteed not to block.
357 598
358=back 599=back
359
360Example:
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 $result_ready->wait;
377 600
378=head1 GLOBAL VARIABLES AND FUNCTIONS 601=head1 GLOBAL VARIABLES AND FUNCTIONS
379 602
380=over 4 603=over 4
381 604
387C<AnyEvent::Impl:xxx> modules, but can be any other class in the case 610C<AnyEvent::Impl:xxx> modules, but can be any other class in the case
388AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>). 611AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>).
389 612
390The known classes so far are: 613The known classes so far are:
391 614
392 AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
393 AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
394 AnyEvent::Impl::EV based on EV (an interface to libev, best choice). 615 AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
395 AnyEvent::Impl::Event based on Event, second best choice. 616 AnyEvent::Impl::Event based on Event, second best choice.
617 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
396 AnyEvent::Impl::Glib based on Glib, third-best choice. 618 AnyEvent::Impl::Glib based on Glib, third-best choice.
397 AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
398 AnyEvent::Impl::Tk based on Tk, very bad choice. 619 AnyEvent::Impl::Tk based on Tk, very bad choice.
399 AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs). 620 AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
400 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse. 621 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
401 AnyEvent::Impl::POE based on POE, not generic enough for full support. 622 AnyEvent::Impl::POE based on POE, not generic enough for full support.
402 623
415Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model 636Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
416if necessary. You should only call this function right before you would 637if necessary. You should only call this function right before you would
417have created an AnyEvent watcher anyway, that is, as late as possible at 638have created an AnyEvent watcher anyway, that is, as late as possible at
418runtime. 639runtime.
419 640
641=item $guard = AnyEvent::post_detect { BLOCK }
642
643Arranges for the code block to be executed as soon as the event model is
644autodetected (or immediately if this has already happened).
645
646If called in scalar or list context, then it creates and returns an object
647that automatically removes the callback again when it is destroyed. See
648L<Coro::BDB> for a case where this is useful.
649
650=item @AnyEvent::post_detect
651
652If there are any code references in this array (you can C<push> to it
653before or after loading AnyEvent), then they will called directly after
654the event loop has been chosen.
655
656You should check C<$AnyEvent::MODEL> before adding to this array, though:
657if it contains a true value then the event loop has already been detected,
658and the array will be ignored.
659
660Best use C<AnyEvent::post_detect { BLOCK }> instead.
661
420=back 662=back
421 663
422=head1 WHAT TO DO IN A MODULE 664=head1 WHAT TO DO IN A MODULE
423 665
424As a module author, you should C<use AnyEvent> and call AnyEvent methods 666As a module author, you should C<use AnyEvent> and call AnyEvent methods
427Be careful when you create watchers in the module body - AnyEvent will 669Be careful when you create watchers in the module body - AnyEvent will
428decide which event module to use as soon as the first method is called, so 670decide which event module to use as soon as the first method is called, so
429by calling AnyEvent in your module body you force the user of your module 671by calling AnyEvent in your module body you force the user of your module
430to load the event module first. 672to load the event module first.
431 673
432Never call C<< ->wait >> on a condition variable unless you I<know> that 674Never call C<< ->recv >> on a condition variable unless you I<know> that
433the C<< ->broadcast >> method has been called on it already. This is 675the C<< ->send >> method has been called on it already. This is
434because it will stall the whole program, and the whole point of using 676because it will stall the whole program, and the whole point of using
435events is to stay interactive. 677events is to stay interactive.
436 678
437It is fine, however, to call C<< ->wait >> when the user of your module 679It is fine, however, to call C<< ->recv >> when the user of your module
438requests it (i.e. if you create a http request object ad have a method 680requests it (i.e. if you create a http request object ad have a method
439called C<results> that returns the results, it should call C<< ->wait >> 681called C<results> that returns the results, it should call C<< ->recv >>
440freely, as the user of your module knows what she is doing. always). 682freely, as the user of your module knows what she is doing. always).
441 683
442=head1 WHAT TO DO IN THE MAIN PROGRAM 684=head1 WHAT TO DO IN THE MAIN PROGRAM
443 685
444There will always be a single main program - the only place that should 686There will always be a single main program - the only place that should
446 688
447If it doesn't care, it can just "use AnyEvent" and use it itself, or not 689If it doesn't care, it can just "use AnyEvent" and use it itself, or not
448do anything special (it does not need to be event-based) and let AnyEvent 690do anything special (it does not need to be event-based) and let AnyEvent
449decide which implementation to chose if some module relies on it. 691decide which implementation to chose if some module relies on it.
450 692
451If the main program relies on a specific event model. For example, in 693If the main program relies on a specific event model - for example, in
452Gtk2 programs you have to rely on the Glib module. You should load the 694Gtk2 programs you have to rely on the Glib module - you should load the
453event module before loading AnyEvent or any module that uses it: generally 695event module before loading AnyEvent or any module that uses it: generally
454speaking, you should load it as early as possible. The reason is that 696speaking, you should load it as early as possible. The reason is that
455modules might create watchers when they are loaded, and AnyEvent will 697modules might create watchers when they are loaded, and AnyEvent will
456decide on the event model to use as soon as it creates watchers, and it 698decide on the event model to use as soon as it creates watchers, and it
457might chose the wrong one unless you load the correct one yourself. 699might chose the wrong one unless you load the correct one yourself.
458 700
459You can chose to use a rather inefficient pure-perl implementation by 701You can chose to use a pure-perl implementation by loading the
460loading the C<AnyEvent::Impl::Perl> module, which gives you similar 702C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
461behaviour everywhere, but letting AnyEvent chose is generally better. 703everywhere, but letting AnyEvent chose the model is generally better.
704
705=head2 MAINLOOP EMULATION
706
707Sometimes (often for short test scripts, or even standalone programs who
708only want to use AnyEvent), you do not want to run a specific event loop.
709
710In that case, you can use a condition variable like this:
711
712 AnyEvent->condvar->recv;
713
714This has the effect of entering the event loop and looping forever.
715
716Note that usually your program has some exit condition, in which case
717it is better to use the "traditional" approach of storing a condition
718variable somewhere, waiting for it, and sending it when the program should
719exit cleanly.
720
721
722=head1 OTHER MODULES
723
724The following is a non-exhaustive list of additional modules that use
725AnyEvent and can therefore be mixed easily with other AnyEvent modules
726in the same program. Some of the modules come with AnyEvent, some are
727available via CPAN.
728
729=over 4
730
731=item L<AnyEvent::Util>
732
733Contains various utility functions that replace often-used but blocking
734functions such as C<inet_aton> by event-/callback-based versions.
735
736=item L<AnyEvent::Handle>
737
738Provide read and write buffers and manages watchers for reads and writes.
739
740=item L<AnyEvent::Socket>
741
742Provides various utility functions for (internet protocol) sockets,
743addresses and name resolution. Also functions to create non-blocking tcp
744connections or tcp servers, with IPv6 and SRV record support and more.
745
746=item L<AnyEvent::DNS>
747
748Provides rich asynchronous DNS resolver capabilities.
749
750=item L<AnyEvent::HTTPD>
751
752Provides a simple web application server framework.
753
754=item L<AnyEvent::FastPing>
755
756The fastest ping in the west.
757
758=item L<Net::IRC3>
759
760AnyEvent based IRC client module family.
761
762=item L<Net::XMPP2>
763
764AnyEvent based XMPP (Jabber protocol) module family.
765
766=item L<Net::FCP>
767
768AnyEvent-based implementation of the Freenet Client Protocol, birthplace
769of AnyEvent.
770
771=item L<Event::ExecFlow>
772
773High level API for event-based execution flow control.
774
775=item L<Coro>
776
777Has special support for AnyEvent via L<Coro::AnyEvent>.
778
779=item L<AnyEvent::AIO>, L<IO::AIO>
780
781Truly asynchronous I/O, should be in the toolbox of every event
782programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
783together.
784
785=item L<AnyEvent::BDB>, L<BDB>
786
787Truly asynchronous Berkeley DB access. AnyEvent::AIO transparently fuses
788IO::AIO and AnyEvent together.
789
790=item L<IO::Lambda>
791
792The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
793
794=back
462 795
463=cut 796=cut
464 797
465package AnyEvent; 798package AnyEvent;
466 799
467no warnings; 800no warnings;
468use strict; 801use strict;
469 802
470use Carp; 803use Carp;
471 804
472our $VERSION = '3.3'; 805our $VERSION = 4.11;
473our $MODEL; 806our $MODEL;
474 807
475our $AUTOLOAD; 808our $AUTOLOAD;
476our @ISA; 809our @ISA;
477 810
811our @REGISTRY;
812
813our $WIN32;
814
815BEGIN {
816 my $win32 = ! ! ($^O =~ /mswin32/i);
817 eval "sub WIN32(){ $win32 }";
818}
819
478our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1; 820our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
479 821
480our @REGISTRY; 822our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
823
824{
825 my $idx;
826 $PROTOCOL{$_} = ++$idx
827 for reverse split /\s*,\s*/,
828 $ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
829}
481 830
482my @models = ( 831my @models = (
483 [Coro::EV:: => AnyEvent::Impl::CoroEV::],
484 [Coro::Event:: => AnyEvent::Impl::CoroEvent::],
485 [EV:: => AnyEvent::Impl::EV::], 832 [EV:: => AnyEvent::Impl::EV::],
486 [Event:: => AnyEvent::Impl::Event::], 833 [Event:: => AnyEvent::Impl::Event::],
487 [Glib:: => AnyEvent::Impl::Glib::],
488 [Tk:: => AnyEvent::Impl::Tk::],
489 [Wx:: => AnyEvent::Impl::POE::],
490 [Prima:: => AnyEvent::Impl::POE::],
491 [AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::], 834 [AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::],
492 # everything below here will not be autoprobed as the pureperl backend should work everywhere 835 # everything below here will not be autoprobed
836 # as the pureperl backend should work everywhere
837 # and is usually faster
838 [Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
839 [Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
493 [Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy 840 [Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
494 [Qt:: => AnyEvent::Impl::Qt::], # requires special main program 841 [Qt:: => AnyEvent::Impl::Qt::], # requires special main program
495 [POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza 842 [POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
843 [Wx:: => AnyEvent::Impl::POE::],
844 [Prima:: => AnyEvent::Impl::POE::],
496); 845);
497 846
498our %method = map +($_ => 1), qw(io timer signal child condvar broadcast wait one_event DESTROY); 847our %method = map +($_ => 1), qw(io timer time now signal child condvar one_event DESTROY);
848
849our @post_detect;
850
851sub post_detect(&) {
852 my ($cb) = @_;
853
854 if ($MODEL) {
855 $cb->();
856
857 1
858 } else {
859 push @post_detect, $cb;
860
861 defined wantarray
862 ? bless \$cb, "AnyEvent::Util::PostDetect"
863 : ()
864 }
865}
866
867sub AnyEvent::Util::PostDetect::DESTROY {
868 @post_detect = grep $_ != ${$_[0]}, @post_detect;
869}
499 870
500sub detect() { 871sub detect() {
501 unless ($MODEL) { 872 unless ($MODEL) {
502 no strict 'refs'; 873 no strict 'refs';
874 local $SIG{__DIE__};
503 875
504 if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) { 876 if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
505 my $model = "AnyEvent::Impl::$1"; 877 my $model = "AnyEvent::Impl::$1";
506 if (eval "require $model") { 878 if (eval "require $model") {
507 $MODEL = $model; 879 $MODEL = $model;
537 last; 909 last;
538 } 910 }
539 } 911 }
540 912
541 $MODEL 913 $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."; 914 or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.";
543 } 915 }
544 } 916 }
545 917
546 unshift @ISA, $MODEL; 918 unshift @ISA, $MODEL;
547 push @{"$MODEL\::ISA"}, "AnyEvent::Base"; 919 push @{"$MODEL\::ISA"}, "AnyEvent::Base";
920
921 (shift @post_detect)->() while @post_detect;
548 } 922 }
549 923
550 $MODEL 924 $MODEL
551} 925}
552 926
562 $class->$func (@_); 936 $class->$func (@_);
563} 937}
564 938
565package AnyEvent::Base; 939package AnyEvent::Base;
566 940
941# default implementation for now and time
942
943use Time::HiRes ();
944
945sub time { Time::HiRes::time }
946sub now { Time::HiRes::time }
947
567# default implementation for ->condvar, ->wait, ->broadcast 948# default implementation for ->condvar
568 949
569sub condvar { 950sub condvar {
570 bless \my $flag, "AnyEvent::Base::CondVar" 951 bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar::
571}
572
573sub AnyEvent::Base::CondVar::broadcast {
574 ${$_[0]}++;
575}
576
577sub AnyEvent::Base::CondVar::wait {
578 AnyEvent->one_event while !${$_[0]};
579} 952}
580 953
581# default implementation for ->signal 954# default implementation for ->signal
582 955
583our %SIG_CB; 956our %SIG_CB;
636 or Carp::croak "required option 'pid' is missing"; 1009 or Carp::croak "required option 'pid' is missing";
637 1010
638 $PID_CB{$pid}{$arg{cb}} = $arg{cb}; 1011 $PID_CB{$pid}{$arg{cb}} = $arg{cb};
639 1012
640 unless ($WNOHANG) { 1013 unless ($WNOHANG) {
641 $WNOHANG = eval { require POSIX; &POSIX::WNOHANG } || 1; 1014 $WNOHANG = eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
642 } 1015 }
643 1016
644 unless ($CHLD_W) { 1017 unless ($CHLD_W) {
645 $CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld); 1018 $CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
646 # child could be a zombie already, so make at least one round 1019 # child could be a zombie already, so make at least one round
656 delete $PID_CB{$pid}{$cb}; 1029 delete $PID_CB{$pid}{$cb};
657 delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} }; 1030 delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
658 1031
659 undef $CHLD_W unless keys %PID_CB; 1032 undef $CHLD_W unless keys %PID_CB;
660} 1033}
1034
1035package AnyEvent::CondVar;
1036
1037our @ISA = AnyEvent::CondVar::Base::;
1038
1039package AnyEvent::CondVar::Base;
1040
1041use overload
1042 '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
1043 fallback => 1;
1044
1045sub _send {
1046 # nop
1047}
1048
1049sub send {
1050 my $cv = shift;
1051 $cv->{_ae_sent} = [@_];
1052 (delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
1053 $cv->_send;
1054}
1055
1056sub croak {
1057 $_[0]{_ae_croak} = $_[1];
1058 $_[0]->send;
1059}
1060
1061sub ready {
1062 $_[0]{_ae_sent}
1063}
1064
1065sub _wait {
1066 AnyEvent->one_event while !$_[0]{_ae_sent};
1067}
1068
1069sub recv {
1070 $_[0]->_wait;
1071
1072 Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
1073 wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
1074}
1075
1076sub cb {
1077 $_[0]{_ae_cb} = $_[1] if @_ > 1;
1078 $_[0]{_ae_cb}
1079}
1080
1081sub begin {
1082 ++$_[0]{_ae_counter};
1083 $_[0]{_ae_end_cb} = $_[1] if @_ > 1;
1084}
1085
1086sub end {
1087 return if --$_[0]{_ae_counter};
1088 &{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
1089}
1090
1091# undocumented/compatibility with pre-3.4
1092*broadcast = \&send;
1093*wait = \&_wait;
661 1094
662=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE 1095=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
663 1096
664This is an advanced topic that you do not normally need to use AnyEvent in 1097This is an advanced topic that you do not normally need to use AnyEvent in
665a module. This section is only of use to event loop authors who want to 1098a module. This section is only of use to event loop authors who want to
722model it chooses. 1155model it chooses.
723 1156
724=item C<PERL_ANYEVENT_MODEL> 1157=item C<PERL_ANYEVENT_MODEL>
725 1158
726This can be used to specify the event model to be used by AnyEvent, before 1159This can be used to specify the event model to be used by AnyEvent, before
727autodetection and -probing kicks in. It must be a string consisting 1160auto detection and -probing kicks in. It must be a string consisting
728entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended 1161entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
729and the resulting module name is loaded and if the load was successful, 1162and the resulting module name is loaded and if the load was successful,
730used as event model. If it fails to load AnyEvent will proceed with 1163used as event model. If it fails to load AnyEvent will proceed with
731autodetection and -probing. 1164auto detection and -probing.
732 1165
733This functionality might change in future versions. 1166This functionality might change in future versions.
734 1167
735For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you 1168For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
736could start your program like this: 1169could start your program like this:
737 1170
738 PERL_ANYEVENT_MODEL=Perl perl ... 1171 PERL_ANYEVENT_MODEL=Perl perl ...
1172
1173=item C<PERL_ANYEVENT_PROTOCOLS>
1174
1175Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
1176for IPv4 or IPv6. The default is unspecified (and might change, or be the result
1177of auto probing).
1178
1179Must be set to a comma-separated list of protocols or address families,
1180current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
1181used, and preference will be given to protocols mentioned earlier in the
1182list.
1183
1184This variable can effectively be used for denial-of-service attacks
1185against local programs (e.g. when setuid), although the impact is likely
1186small, as the program has to handle connection errors already-
1187
1188Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
1189but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
1190- only support IPv4, never try to resolve or contact IPv6
1191addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
1192IPv6, but prefer IPv6 over IPv4.
1193
1194=item C<PERL_ANYEVENT_EDNS0>
1195
1196Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
1197for DNS. This extension is generally useful to reduce DNS traffic, but
1198some (broken) firewalls drop such DNS packets, which is why it is off by
1199default.
1200
1201Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
1202EDNS0 in its DNS requests.
1203
1204=item C<PERL_ANYEVENT_MAX_FORKS>
1205
1206The maximum number of child processes that C<AnyEvent::Util::fork_call>
1207will create in parallel.
739 1208
740=back 1209=back
741 1210
742=head1 EXAMPLE PROGRAM 1211=head1 EXAMPLE PROGRAM
743 1212
754 poll => 'r', 1223 poll => 'r',
755 cb => sub { 1224 cb => sub {
756 warn "io event <$_[0]>\n"; # will always output <r> 1225 warn "io event <$_[0]>\n"; # will always output <r>
757 chomp (my $input = <STDIN>); # read a line 1226 chomp (my $input = <STDIN>); # read a line
758 warn "read: $input\n"; # output what has been read 1227 warn "read: $input\n"; # output what has been read
759 $cv->broadcast if $input =~ /^q/i; # quit program if /^q/i 1228 $cv->send if $input =~ /^q/i; # quit program if /^q/i
760 }, 1229 },
761 ); 1230 );
762 1231
763 my $time_watcher; # can only be used once 1232 my $time_watcher; # can only be used once
764 1233
769 }); 1238 });
770 } 1239 }
771 1240
772 new_timer; # create first timer 1241 new_timer; # create first timer
773 1242
774 $cv->wait; # wait until user enters /^q/i 1243 $cv->recv; # wait until user enters /^q/i
775 1244
776=head1 REAL-WORLD EXAMPLE 1245=head1 REAL-WORLD EXAMPLE
777 1246
778Consider the L<Net::FCP> module. It features (among others) the following 1247Consider the L<Net::FCP> module. It features (among others) the following
779API calls, which are to freenet what HTTP GET requests are to http: 1248API calls, which are to freenet what HTTP GET requests are to http:
829 syswrite $txn->{fh}, $txn->{request} 1298 syswrite $txn->{fh}, $txn->{request}
830 or die "connection or write error"; 1299 or die "connection or write error";
831 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r }); 1300 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
832 1301
833Again, C<fh_ready_r> waits till all data has arrived, and then stores the 1302Again, C<fh_ready_r> waits till all data has arrived, and then stores the
834result and signals any possible waiters that the request ahs finished: 1303result and signals any possible waiters that the request has finished:
835 1304
836 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; 1305 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
837 1306
838 if (end-of-file or data complete) { 1307 if (end-of-file or data complete) {
839 $txn->{result} = $txn->{buf}; 1308 $txn->{result} = $txn->{buf};
840 $txn->{finished}->broadcast; 1309 $txn->{finished}->send;
841 $txb->{cb}->($txn) of $txn->{cb}; # also call callback 1310 $txb->{cb}->($txn) of $txn->{cb}; # also call callback
842 } 1311 }
843 1312
844The C<result> method, finally, just waits for the finished signal (if the 1313The C<result> method, finally, just waits for the finished signal (if the
845request was already finished, it doesn't wait, of course, and returns the 1314request was already finished, it doesn't wait, of course, and returns the
846data: 1315data:
847 1316
848 $txn->{finished}->wait; 1317 $txn->{finished}->recv;
849 return $txn->{result}; 1318 return $txn->{result};
850 1319
851The actual code goes further and collects all errors (C<die>s, exceptions) 1320The actual code goes further and collects all errors (C<die>s, exceptions)
852that occured during request processing. The C<result> method detects 1321that occurred during request processing. The C<result> method detects
853whether an exception as thrown (it is stored inside the $txn object) 1322whether an exception as thrown (it is stored inside the $txn object)
854and just throws the exception, which means connection errors and other 1323and just throws the exception, which means connection errors and other
855problems get reported tot he code that tries to use the result, not in a 1324problems get reported tot he code that tries to use the result, not in a
856random callback. 1325random callback.
857 1326
888 1357
889 my $quit = AnyEvent->condvar; 1358 my $quit = AnyEvent->condvar;
890 1359
891 $fcp->txn_client_get ($url)->cb (sub { 1360 $fcp->txn_client_get ($url)->cb (sub {
892 ... 1361 ...
893 $quit->broadcast; 1362 $quit->send;
894 }); 1363 });
895 1364
896 $quit->wait; 1365 $quit->recv;
897 1366
898 1367
899=head1 BENCHMARKS 1368=head1 BENCHMARKS
900 1369
901To give you an idea of the performance and overheads that AnyEvent adds 1370To give you an idea of the performance and overheads that AnyEvent adds
903of various event loops I prepared some benchmarks. 1372of various event loops I prepared some benchmarks.
904 1373
905=head2 BENCHMARKING ANYEVENT OVERHEAD 1374=head2 BENCHMARKING ANYEVENT OVERHEAD
906 1375
907Here is a benchmark of various supported event models used natively and 1376Here is a benchmark of various supported event models used natively and
908through anyevent. The benchmark creates a lot of timers (with a zero 1377through AnyEvent. The benchmark creates a lot of timers (with a zero
909timeout) and I/O watchers (watching STDOUT, a pty, to become writable, 1378timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
910which it is), lets them fire exactly once and destroys them again. 1379which it is), lets them fire exactly once and destroys them again.
911 1380
912Source code for this benchmark is found as F<eg/bench> in the AnyEvent 1381Source code for this benchmark is found as F<eg/bench> in the AnyEvent
913distribution. 1382distribution.
930all watchers, to avoid adding memory overhead. That means closure creation 1399all watchers, to avoid adding memory overhead. That means closure creation
931and memory usage is not included in the figures. 1400and memory usage is not included in the figures.
932 1401
933I<invoke> is the time, in microseconds, used to invoke a simple 1402I<invoke> is the time, in microseconds, used to invoke a simple
934callback. The callback simply counts down a Perl variable and after it was 1403callback. The callback simply counts down a Perl variable and after it was
935invoked "watcher" times, it would C<< ->broadcast >> a condvar once to 1404invoked "watcher" times, it would C<< ->send >> a condvar once to
936signal the end of this phase. 1405signal the end of this phase.
937 1406
938I<destroy> is the time, in microseconds, that it takes to destroy a single 1407I<destroy> is the time, in microseconds, that it takes to destroy a single
939watcher. 1408watcher.
940 1409
944 EV/EV 400000 244 0.56 0.46 0.31 EV native interface 1413 EV/EV 400000 244 0.56 0.46 0.31 EV native interface
945 EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers 1414 EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers
946 CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal 1415 CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal
947 Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation 1416 Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation
948 Event/Event 16000 516 31.88 31.30 0.85 Event native interface 1417 Event/Event 16000 516 31.88 31.30 0.85 Event native interface
949 Event/Any 16000 936 39.17 33.63 1.43 Event + AnyEvent watchers 1418 Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers
950 Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour 1419 Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour
951 Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers 1420 Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers
952 POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event 1421 POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event
953 POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select 1422 POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select
954 1423
1000file descriptor is dup()ed for each watcher. This shows that the dup() 1469file descriptor is dup()ed for each watcher. This shows that the dup()
1001employed by some adaptors is not a big performance issue (it does incur a 1470employed by some adaptors is not a big performance issue (it does incur a
1002hidden memory cost inside the kernel which is not reflected in the figures 1471hidden memory cost inside the kernel which is not reflected in the figures
1003above). 1472above).
1004 1473
1005C<POE>, regardless of underlying event loop (whether using its pure 1474C<POE>, regardless of underlying event loop (whether using its pure perl
1006perl select-based backend or the Event module, the POE-EV backend 1475select-based backend or the Event module, the POE-EV backend couldn't
1007couldn't be tested because it wasn't working) shows abysmal performance 1476be tested because it wasn't working) shows abysmal performance and
1008and memory usage: Watchers use almost 30 times as much memory as 1477memory usage with AnyEvent: Watchers use almost 30 times as much memory
1009EV watchers, and 10 times as much memory as Event (the high memory 1478as EV watchers, and 10 times as much memory as Event (the high memory
1010requirements are caused by requiring a session for each watcher). Watcher 1479requirements are caused by requiring a session for each watcher). Watcher
1011invocation speed is almost 900 times slower than with AnyEvent's pure perl 1480invocation speed is almost 900 times slower than with AnyEvent's pure perl
1481implementation.
1482
1012implementation. The design of the POE adaptor class in AnyEvent can not 1483The design of the POE adaptor class in AnyEvent can not really account
1013really account for this, as session creation overhead is small compared 1484for the performance issues, though, as session creation overhead is
1014to execution of the state machine, which is coded pretty optimally within 1485small compared to execution of the state machine, which is coded pretty
1015L<AnyEvent::Impl::POE>. POE simply seems to be abysmally slow. 1486optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
1487using multiple sessions is not a good approach, especially regarding
1488memory usage, even the author of POE could not come up with a faster
1489design).
1016 1490
1017=head3 Summary 1491=head3 Summary
1018 1492
1019=over 4 1493=over 4
1020 1494
1031 1505
1032=back 1506=back
1033 1507
1034=head2 BENCHMARKING THE LARGE SERVER CASE 1508=head2 BENCHMARKING THE LARGE SERVER CASE
1035 1509
1036This benchmark atcually benchmarks the event loop itself. It works by 1510This benchmark actually benchmarks the event loop itself. It works by
1037creating a number of "servers": each server consists of a socketpair, a 1511creating a number of "servers": each server consists of a socket pair, a
1038timeout watcher that gets reset on activity (but never fires), and an I/O 1512timeout watcher that gets reset on activity (but never fires), and an I/O
1039watcher waiting for input on one side of the socket. Each time the socket 1513watcher waiting for input on one side of the socket. Each time the socket
1040watcher reads a byte it will write that byte to a random other "server". 1514watcher reads a byte it will write that byte to a random other "server".
1041 1515
1042The effect is that there will be a lot of I/O watchers, only part of which 1516The effect is that there will be a lot of I/O watchers, only part of which
1043are active at any one point (so there is a constant number of active 1517are active at any one point (so there is a constant number of active
1044fds for each loop iterstaion, but which fds these are is random). The 1518fds for each loop iteration, but which fds these are is random). The
1045timeout is reset each time something is read because that reflects how 1519timeout is reset each time something is read because that reflects how
1046most timeouts work (and puts extra pressure on the event loops). 1520most timeouts work (and puts extra pressure on the event loops).
1047 1521
1048In this benchmark, we use 10000 socketpairs (20000 sockets), of which 100 1522In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
1049(1%) are active. This mirrors the activity of large servers with many 1523(1%) are active. This mirrors the activity of large servers with many
1050connections, most of which are idle at any one point in time. 1524connections, most of which are idle at any one point in time.
1051 1525
1052Source code for this benchmark is found as F<eg/bench2> in the AnyEvent 1526Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
1053distribution. 1527distribution.
1055=head3 Explanation of the columns 1529=head3 Explanation of the columns
1056 1530
1057I<sockets> is the number of sockets, and twice the number of "servers" (as 1531I<sockets> is the number of sockets, and twice the number of "servers" (as
1058each server has a read and write socket end). 1532each server has a read and write socket end).
1059 1533
1060I<create> is the time it takes to create a socketpair (which is 1534I<create> is the time it takes to create a socket pair (which is
1061nontrivial) and two watchers: an I/O watcher and a timeout watcher. 1535nontrivial) and two watchers: an I/O watcher and a timeout watcher.
1062 1536
1063I<request>, the most important value, is the time it takes to handle a 1537I<request>, the most important value, is the time it takes to handle a
1064single "request", that is, reading the token from the pipe and forwarding 1538single "request", that is, reading the token from the pipe and forwarding
1065it to another server. This includes deleting the old timeout and creating 1539it to another server. This includes deleting the old timeout and creating
1067 1541
1068=head3 Results 1542=head3 Results
1069 1543
1070 name sockets create request 1544 name sockets create request
1071 EV 20000 69.01 11.16 1545 EV 20000 69.01 11.16
1072 Perl 20000 75.28 112.76 1546 Perl 20000 73.32 35.87
1073 Event 20000 212.62 257.32 1547 Event 20000 212.62 257.32
1074 Glib 20000 651.16 1896.30 1548 Glib 20000 651.16 1896.30
1075 POE 20000 349.67 12317.24 uses POE::Loop::Event 1549 POE 20000 349.67 12317.24 uses POE::Loop::Event
1076 1550
1077=head3 Discussion 1551=head3 Discussion
1099 1573
1100=head3 Summary 1574=head3 Summary
1101 1575
1102=over 4 1576=over 4
1103 1577
1104=item * The pure perl implementation performs extremely well, considering 1578=item * The pure perl implementation performs extremely well.
1105that it uses select.
1106 1579
1107=item * Avoid Glib or POE in large projects where performance matters. 1580=item * Avoid Glib or POE in large projects where performance matters.
1108 1581
1109=back 1582=back
1110 1583
1123 1596
1124=head3 Results 1597=head3 Results
1125 1598
1126 name sockets create request 1599 name sockets create request
1127 EV 16 20.00 6.54 1600 EV 16 20.00 6.54
1601 Perl 16 25.75 12.62
1128 Event 16 81.27 35.86 1602 Event 16 81.27 35.86
1129 Glib 16 32.63 15.48 1603 Glib 16 32.63 15.48
1130 Perl 16 24.62 162.37
1131 POE 16 261.87 276.28 uses POE::Loop::Event 1604 POE 16 261.87 276.28 uses POE::Loop::Event
1132 1605
1133=head3 Discussion 1606=head3 Discussion
1134 1607
1135The benchmark tries to test the performance of a typical small 1608The benchmark tries to test the performance of a typical small
1136server. While knowing how various event loops perform is interesting, keep 1609server. While knowing how various event loops perform is interesting, keep
1137in mind that their overhead in this case is usually not as important, due 1610in mind that their overhead in this case is usually not as important, due
1138to the small absolute number of watchers. 1611to the small absolute number of watchers (that is, you need efficiency and
1612speed most when you have lots of watchers, not when you only have a few of
1613them).
1139 1614
1140EV is again fastest. 1615EV is again fastest.
1141 1616
1142The C-based event loops Event and Glib come in second this time, as the 1617Perl again comes second. It is noticeably faster than the C-based event
1143overhead of running an iteration is much smaller in C than in Perl (little 1618loops Event and Glib, although the difference is too small to really
1144code to execute in the inner loop, and perl's function calling overhead is 1619matter.
1145high, and updating all the data structures is costly).
1146 1620
1147The pure perl event loop is much slower, but still competitive.
1148
1149POE also performs much better in this case, but is is stillf ar behind the 1621POE also performs much better in this case, but is is still far behind the
1150others. 1622others.
1151 1623
1152=head3 Summary 1624=head3 Summary
1153 1625
1154=over 4 1626=over 4
1160 1632
1161 1633
1162=head1 FORK 1634=head1 FORK
1163 1635
1164Most event libraries are not fork-safe. The ones who are usually are 1636Most event libraries are not fork-safe. The ones who are usually are
1165because they are so inefficient. Only L<EV> is fully fork-aware. 1637because they rely on inefficient but fork-safe C<select> or C<poll>
1638calls. Only L<EV> is fully fork-aware.
1166 1639
1167If you have to fork, you must either do so I<before> creating your first 1640If you have to fork, you must either do so I<before> creating your first
1168watcher OR you must not use AnyEvent at all in the child. 1641watcher OR you must not use AnyEvent at all in the child.
1169 1642
1170 1643
1182 1655
1183 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} } 1656 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
1184 1657
1185 use AnyEvent; 1658 use AnyEvent;
1186 1659
1660Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
1661be used to probe what backend is used and gain other information (which is
1662probably even less useful to an attacker than PERL_ANYEVENT_MODEL).
1663
1187 1664
1188=head1 SEE ALSO 1665=head1 SEE ALSO
1189 1666
1190Event modules: L<Coro::EV>, L<EV>, L<EV::Glib>, L<Glib::EV>, 1667Utility functions: L<AnyEvent::Util>.
1191L<Coro::Event>, L<Event>, L<Glib::Event>, L<Glib>, L<Coro>, L<Tk>, 1668
1669Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
1192L<Event::Lib>, L<Qt>, L<POE>. 1670L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
1193 1671
1194Implementations: L<AnyEvent::Impl::CoroEV>, L<AnyEvent::Impl::EV>, 1672Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
1195L<AnyEvent::Impl::CoroEvent>, L<AnyEvent::Impl::Event>, L<AnyEvent::Impl::Glib>, 1673L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
1196L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>, L<AnyEvent::Impl::EventLib>, 1674L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
1197L<AnyEvent::Impl::Qt>, L<AnyEvent::Impl::POE>. 1675L<AnyEvent::Impl::POE>.
1198 1676
1677Non-blocking file handles, sockets, TCP clients and
1678servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>.
1679
1680Asynchronous DNS: L<AnyEvent::DNS>.
1681
1682Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>,
1683
1199Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>. 1684Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>.
1200 1685
1201 1686
1202=head1 AUTHOR 1687=head1 AUTHOR
1203 1688
1204 Marc Lehmann <schmorp@schmorp.de> 1689 Marc Lehmann <schmorp@schmorp.de>

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines