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 1.171 by root, Wed Jul 16 21:17:59 2008 UTC vs.
Revision 1.317 by root, Wed Mar 24 21:22:57 2010 UTC

1=head1 NAME 1=head1 NAME
2 2
3AnyEvent - provide framework for multiple event loops 3AnyEvent - the DBI of event loop programming
4 4
5EV, Event, Glib, Tk, Perl, Event::Lib, Qt, POE - various supported event loops 5EV, Event, Glib, Tk, Perl, Event::Lib, Irssi, rxvt-unicode, IO::Async, Qt
6and POE are various supported event loops/environments.
6 7
7=head1 SYNOPSIS 8=head1 SYNOPSIS
8 9
9 use AnyEvent; 10 use AnyEvent;
10 11
12 # file descriptor readable
11 my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { 13 my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ... });
14
15 # one-shot or repeating timers
16 my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
17 my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...
18
19 print AnyEvent->now; # prints current event loop time
20 print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
21
22 # POSIX signal
23 my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
24
25 # child process exit
26 my $w = AnyEvent->child (pid => $pid, cb => sub {
27 my ($pid, $status) = @_;
12 ... 28 ...
13 }); 29 });
14 30
15 my $w = AnyEvent->timer (after => $seconds, cb => sub { 31 # called when event loop idle (if applicable)
16 ... 32 my $w = AnyEvent->idle (cb => sub { ... });
17 });
18 33
19 my $w = AnyEvent->condvar; # stores whether a condition was flagged 34 my $w = AnyEvent->condvar; # stores whether a condition was flagged
20 $w->send; # wake up current and all future recv's 35 $w->send; # wake up current and all future recv's
21 $w->recv; # enters "main loop" till $condvar gets ->send 36 $w->recv; # enters "main loop" till $condvar gets ->send
37 # use a condvar in callback mode:
38 $w->cb (sub { $_[0]->recv });
22 39
23=head1 INTRODUCTION/TUTORIAL 40=head1 INTRODUCTION/TUTORIAL
24 41
25This manpage is mainly a reference manual. If you are interested 42This manpage is mainly a reference manual. If you are interested
26in a tutorial or some gentle introduction, have a look at the 43in a tutorial or some gentle introduction, have a look at the
27L<AnyEvent::Intro> manpage. 44L<AnyEvent::Intro> manpage.
45
46=head1 SUPPORT
47
48There is a mailinglist for discussing all things AnyEvent, and an IRC
49channel, too.
50
51See the AnyEvent project page at the B<Schmorpforge Ta-Sa Software
52Repository>, at L<http://anyevent.schmorp.de>, for more info.
28 53
29=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT) 54=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
30 55
31Glib, POE, IO::Async, Event... CPAN offers event models by the dozen 56Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
32nowadays. So what is different about AnyEvent? 57nowadays. So what is different about AnyEvent?
128These watchers are normal Perl objects with normal Perl lifetime. After 153These watchers are normal Perl objects with normal Perl lifetime. After
129creating a watcher it will immediately "watch" for events and invoke the 154creating a watcher it will immediately "watch" for events and invoke the
130callback when the event occurs (of course, only when the event model 155callback when the event occurs (of course, only when the event model
131is in control). 156is in control).
132 157
158Note that B<callbacks must not permanently change global variables>
159potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
160callbacks must not C<die> >>. The former is good programming practise in
161Perl and the latter stems from the fact that exception handling differs
162widely between event loops.
163
133To disable the watcher you have to destroy it (e.g. by setting the 164To disable the watcher you have to destroy it (e.g. by setting the
134variable you store it in to C<undef> or otherwise deleting all references 165variable you store it in to C<undef> or otherwise deleting all references
135to it). 166to it).
136 167
137All watchers are created by calling a method on the C<AnyEvent> class. 168All watchers are created by calling a method on the C<AnyEvent> class.
150my variables are only visible after the statement in which they are 181my variables are only visible after the statement in which they are
151declared. 182declared.
152 183
153=head2 I/O WATCHERS 184=head2 I/O WATCHERS
154 185
186 $w = AnyEvent->io (
187 fh => <filehandle_or_fileno>,
188 poll => <"r" or "w">,
189 cb => <callback>,
190 );
191
155You can create an I/O watcher by calling the C<< AnyEvent->io >> method 192You can create an I/O watcher by calling the C<< AnyEvent->io >> method
156with the following mandatory key-value pairs as arguments: 193with the following mandatory key-value pairs as arguments:
157 194
158C<fh> the Perl I<file handle> (I<not> file descriptor) to watch for events 195C<fh> is the Perl I<file handle> (or a naked file descriptor) to watch
159(AnyEvent might or might not keep a reference to this file handle). C<poll> 196for events (AnyEvent might or might not keep a reference to this file
197handle). Note that only file handles pointing to things for which
198non-blocking operation makes sense are allowed. This includes sockets,
199most character devices, pipes, fifos and so on, but not for example files
200or block devices.
201
160must be a string that is either C<r> or C<w>, which creates a watcher 202C<poll> must be a string that is either C<r> or C<w>, which creates a
161waiting for "r"eadable or "w"ritable events, respectively. C<cb> is the 203watcher waiting for "r"eadable or "w"ritable events, respectively.
204
162callback to invoke each time the file handle becomes ready. 205C<cb> is the callback to invoke each time the file handle becomes ready.
163 206
164Although the callback might get passed parameters, their value and 207Although the callback might get passed parameters, their value and
165presence is undefined and you cannot rely on them. Portable AnyEvent 208presence is undefined and you cannot rely on them. Portable AnyEvent
166callbacks cannot use arguments passed to I/O watcher callbacks. 209callbacks cannot use arguments passed to I/O watcher callbacks.
167 210
182 undef $w; 225 undef $w;
183 }); 226 });
184 227
185=head2 TIME WATCHERS 228=head2 TIME WATCHERS
186 229
230 $w = AnyEvent->timer (after => <seconds>, cb => <callback>);
231
232 $w = AnyEvent->timer (
233 after => <fractional_seconds>,
234 interval => <fractional_seconds>,
235 cb => <callback>,
236 );
237
187You can create a time watcher by calling the C<< AnyEvent->timer >> 238You can create a time watcher by calling the C<< AnyEvent->timer >>
188method with the following mandatory arguments: 239method with the following mandatory arguments:
189 240
190C<after> specifies after how many seconds (fractional values are 241C<after> specifies after how many seconds (fractional values are
191supported) the callback should be invoked. C<cb> is the callback to invoke 242supported) the callback should be invoked. C<cb> is the callback to invoke
299In either case, if you care (and in most cases, you don't), then you 350In either case, if you care (and in most cases, you don't), then you
300can get whatever behaviour you want with any event loop, by taking the 351can get whatever behaviour you want with any event loop, by taking the
301difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into 352difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
302account. 353account.
303 354
355=item AnyEvent->now_update
356
357Some event loops (such as L<EV> or L<AnyEvent::Impl::Perl>) cache
358the current time for each loop iteration (see the discussion of L<<
359AnyEvent->now >>, above).
360
361When a callback runs for a long time (or when the process sleeps), then
362this "current" time will differ substantially from the real time, which
363might affect timers and time-outs.
364
365When this is the case, you can call this method, which will update the
366event loop's idea of "current time".
367
368A typical example would be a script in a web server (e.g. C<mod_perl>) -
369when mod_perl executes the script, then the event loop will have the wrong
370idea about the "current time" (being potentially far in the past, when the
371script ran the last time). In that case you should arrange a call to C<<
372AnyEvent->now_update >> each time the web server process wakes up again
373(e.g. at the start of your script, or in a handler).
374
375Note that updating the time I<might> cause some events to be handled.
376
304=back 377=back
305 378
306=head2 SIGNAL WATCHERS 379=head2 SIGNAL WATCHERS
380
381 $w = AnyEvent->signal (signal => <uppercase_signal_name>, cb => <callback>);
307 382
308You can watch for signals using a signal watcher, C<signal> is the signal 383You can watch for signals using a signal watcher, C<signal> is the signal
309I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl 384I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
310callback to be invoked whenever a signal occurs. 385callback to be invoked whenever a signal occurs.
311 386
317invocation, and callback invocation will be synchronous. Synchronous means 392invocation, and callback invocation will be synchronous. Synchronous means
318that it might take a while until the signal gets handled by the process, 393that it might take a while until the signal gets handled by the process,
319but it is guaranteed not to interrupt any other callbacks. 394but it is guaranteed not to interrupt any other callbacks.
320 395
321The main advantage of using these watchers is that you can share a signal 396The main advantage of using these watchers is that you can share a signal
322between multiple watchers. 397between multiple watchers, and AnyEvent will ensure that signals will not
398interrupt your program at bad times.
323 399
324This watcher might use C<%SIG>, so programs overwriting those signals 400This watcher might use C<%SIG> (depending on the event loop used),
325directly will likely not work correctly. 401so programs overwriting those signals directly will likely not work
402correctly.
326 403
327Example: exit on SIGINT 404Example: exit on SIGINT
328 405
329 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 }); 406 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
330 407
408=head3 Restart Behaviour
409
410While restart behaviour is up to the event loop implementation, most will
411not restart syscalls (that includes L<Async::Interrupt> and AnyEvent's
412pure perl implementation).
413
414=head3 Safe/Unsafe Signals
415
416Perl signals can be either "safe" (synchronous to opcode handling) or
417"unsafe" (asynchronous) - the former might get delayed indefinitely, the
418latter might corrupt your memory.
419
420AnyEvent signal handlers are, in addition, synchronous to the event loop,
421i.e. they will not interrupt your running perl program but will only be
422called as part of the normal event handling (just like timer, I/O etc.
423callbacks, too).
424
425=head3 Signal Races, Delays and Workarounds
426
427Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
428callbacks to signals in a generic way, which is a pity, as you cannot
429do race-free signal handling in perl, requiring C libraries for
430this. AnyEvent will try to do it's best, which means in some cases,
431signals will be delayed. The maximum time a signal might be delayed is
432specified in C<$AnyEvent::MAX_SIGNAL_LATENCY> (default: 10 seconds). This
433variable can be changed only before the first signal watcher is created,
434and should be left alone otherwise. This variable determines how often
435AnyEvent polls for signals (in case a wake-up was missed). Higher values
436will cause fewer spurious wake-ups, which is better for power and CPU
437saving.
438
439All these problems can be avoided by installing the optional
440L<Async::Interrupt> module, which works with most event loops. It will not
441work with inherently broken event loops such as L<Event> or L<Event::Lib>
442(and not with L<POE> currently, as POE does it's own workaround with
443one-second latency). For those, you just have to suffer the delays.
444
331=head2 CHILD PROCESS WATCHERS 445=head2 CHILD PROCESS WATCHERS
332 446
447 $w = AnyEvent->child (pid => <process id>, cb => <callback>);
448
333You can also watch on a child process exit and catch its exit status. 449You can also watch on a child process exit and catch its exit status.
334 450
335The child process is specified by the C<pid> argument (if set to C<0>, it 451The child process is specified by the C<pid> argument (one some backends,
336watches for any child process exit). The watcher will trigger as often 452using C<0> watches for any child process exit, on others this will
337as status change for the child are received. This works by installing a 453croak). The watcher will be triggered only when the child process has
338signal handler for C<SIGCHLD>. The callback will be called with the pid 454finished and an exit status is available, not on any trace events
339and exit status (as returned by waitpid), so unlike other watcher types, 455(stopped/continued).
340you I<can> rely on child watcher callback arguments. 456
457The callback will be called with the pid and exit status (as returned by
458waitpid), so unlike other watcher types, you I<can> rely on child watcher
459callback arguments.
460
461This watcher type works by installing a signal handler for C<SIGCHLD>,
462and since it cannot be shared, nothing else should use SIGCHLD or reap
463random child processes (waiting for specific child processes, e.g. inside
464C<system>, is just fine).
341 465
342There is a slight catch to child watchers, however: you usually start them 466There is a slight catch to child watchers, however: you usually start them
343I<after> the child process was created, and this means the process could 467I<after> the child process was created, and this means the process could
344have exited already (and no SIGCHLD will be sent anymore). 468have exited already (and no SIGCHLD will be sent anymore).
345 469
346Not all event models handle this correctly (POE doesn't), but even for 470Not all event models handle this correctly (neither POE nor IO::Async do,
471see their AnyEvent::Impl manpages for details), but even for event models
347event models that I<do> handle this correctly, they usually need to be 472that I<do> handle this correctly, they usually need to be loaded before
348loaded before the process exits (i.e. before you fork in the first place). 473the process exits (i.e. before you fork in the first place). AnyEvent's
474pure perl event loop handles all cases correctly regardless of when you
475start the watcher.
349 476
350This means you cannot create a child watcher as the very first thing in an 477This means you cannot create a child watcher as the very first
351AnyEvent program, you I<have> to create at least one watcher before you 478thing in an AnyEvent program, you I<have> to create at least one
352C<fork> the child (alternatively, you can call C<AnyEvent::detect>). 479watcher before you C<fork> the child (alternatively, you can call
480C<AnyEvent::detect>).
481
482As most event loops do not support waiting for child events, they will be
483emulated by AnyEvent in most cases, in which the latency and race problems
484mentioned in the description of signal watchers apply.
353 485
354Example: fork a process and wait for it 486Example: fork a process and wait for it
355 487
356 my $done = AnyEvent->condvar; 488 my $done = AnyEvent->condvar;
357 489
367 ); 499 );
368 500
369 # do something else, then wait for process exit 501 # do something else, then wait for process exit
370 $done->recv; 502 $done->recv;
371 503
504=head2 IDLE WATCHERS
505
506 $w = AnyEvent->idle (cb => <callback>);
507
508Repeatedly invoke the callback after the process becomes idle, until
509either the watcher is destroyed or new events have been detected.
510
511Idle watchers are useful when there is a need to do something, but it
512is not so important (or wise) to do it instantly. The callback will be
513invoked only when there is "nothing better to do", which is usually
514defined as "all outstanding events have been handled and no new events
515have been detected". That means that idle watchers ideally get invoked
516when the event loop has just polled for new events but none have been
517detected. Instead of blocking to wait for more events, the idle watchers
518will be invoked.
519
520Unfortunately, most event loops do not really support idle watchers (only
521EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
522will simply call the callback "from time to time".
523
524Example: read lines from STDIN, but only process them when the
525program is otherwise idle:
526
527 my @lines; # read data
528 my $idle_w;
529 my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
530 push @lines, scalar <STDIN>;
531
532 # start an idle watcher, if not already done
533 $idle_w ||= AnyEvent->idle (cb => sub {
534 # handle only one line, when there are lines left
535 if (my $line = shift @lines) {
536 print "handled when idle: $line";
537 } else {
538 # otherwise disable the idle watcher again
539 undef $idle_w;
540 }
541 });
542 });
543
372=head2 CONDITION VARIABLES 544=head2 CONDITION VARIABLES
545
546 $cv = AnyEvent->condvar;
547
548 $cv->send (<list>);
549 my @res = $cv->recv;
373 550
374If you are familiar with some event loops you will know that all of them 551If you are familiar with some event loops you will know that all of them
375require you to run some blocking "loop", "run" or similar function that 552require you to run some blocking "loop", "run" or similar function that
376will actively watch for new events and call your callbacks. 553will actively watch for new events and call your callbacks.
377 554
378AnyEvent is different, it expects somebody else to run the event loop and 555AnyEvent is slightly different: it expects somebody else to run the event
379will only block when necessary (usually when told by the user). 556loop and will only block when necessary (usually when told by the user).
380 557
381The instrument to do that is called a "condition variable", so called 558The instrument to do that is called a "condition variable", so called
382because they represent a condition that must become true. 559because they represent a condition that must become true.
560
561Now is probably a good time to look at the examples further below.
383 562
384Condition variables can be created by calling the C<< AnyEvent->condvar 563Condition variables can be created by calling the C<< AnyEvent->condvar
385>> method, usually without arguments. The only argument pair allowed is 564>> method, usually without arguments. The only argument pair allowed is
386C<cb>, which specifies a callback to be called when the condition variable 565C<cb>, which specifies a callback to be called when the condition variable
387becomes true. 566becomes true, with the condition variable as the first argument (but not
567the results).
388 568
389After creation, the condition variable is "false" until it becomes "true" 569After creation, the condition variable is "false" until it becomes "true"
390by calling the C<send> method (or calling the condition variable as if it 570by calling the C<send> method (or calling the condition variable as if it
391were a callback, read about the caveats in the description for the C<< 571were a callback, read about the caveats in the description for the C<<
392->send >> method). 572->send >> method).
394Condition variables are similar to callbacks, except that you can 574Condition variables are similar to callbacks, except that you can
395optionally wait for them. They can also be called merge points - points 575optionally wait for them. They can also be called merge points - points
396in time where multiple outstanding events have been processed. And yet 576in time where multiple outstanding events have been processed. And yet
397another way to call them is transactions - each condition variable can be 577another way to call them is transactions - each condition variable can be
398used to represent a transaction, which finishes at some point and delivers 578used to represent a transaction, which finishes at some point and delivers
399a result. 579a result. And yet some people know them as "futures" - a promise to
580compute/deliver something that you can wait for.
400 581
401Condition variables are very useful to signal that something has finished, 582Condition variables are very useful to signal that something has finished,
402for example, if you write a module that does asynchronous http requests, 583for example, if you write a module that does asynchronous http requests,
403then a condition variable would be the ideal candidate to signal the 584then a condition variable would be the ideal candidate to signal the
404availability of results. The user can either act when the callback is 585availability of results. The user can either act when the callback is
438 after => 1, 619 after => 1,
439 cb => sub { $result_ready->send }, 620 cb => sub { $result_ready->send },
440 ); 621 );
441 622
442 # this "blocks" (while handling events) till the callback 623 # this "blocks" (while handling events) till the callback
443 # calls send 624 # calls ->send
444 $result_ready->recv; 625 $result_ready->recv;
445 626
446Example: wait for a timer, but take advantage of the fact that 627Example: wait for a timer, but take advantage of the fact that condition
447condition variables are also code references. 628variables are also callable directly.
448 629
449 my $done = AnyEvent->condvar; 630 my $done = AnyEvent->condvar;
450 my $delay = AnyEvent->timer (after => 5, cb => $done); 631 my $delay = AnyEvent->timer (after => 5, cb => $done);
451 $done->recv; 632 $done->recv;
633
634Example: Imagine an API that returns a condvar and doesn't support
635callbacks. This is how you make a synchronous call, for example from
636the main program:
637
638 use AnyEvent::CouchDB;
639
640 ...
641
642 my @info = $couchdb->info->recv;
643
644And this is how you would just set a callback to be called whenever the
645results are available:
646
647 $couchdb->info->cb (sub {
648 my @info = $_[0]->recv;
649 });
452 650
453=head3 METHODS FOR PRODUCERS 651=head3 METHODS FOR PRODUCERS
454 652
455These methods should only be used by the producing side, i.e. the 653These methods should only be used by the producing side, i.e. the
456code/module that eventually sends the signal. Note that it is also 654code/module that eventually sends the signal. Note that it is also
469immediately from within send. 667immediately from within send.
470 668
471Any arguments passed to the C<send> call will be returned by all 669Any arguments passed to the C<send> call will be returned by all
472future C<< ->recv >> calls. 670future C<< ->recv >> calls.
473 671
474Condition variables are overloaded so one can call them directly 672Condition variables are overloaded so one can call them directly (as if
475(as a code reference). Calling them directly is the same as calling 673they were a code reference). Calling them directly is the same as calling
476C<send>. Note, however, that many C-based event loops do not handle 674C<send>.
477overloading, so as tempting as it may be, passing a condition variable
478instead of a callback does not work. Both the pure perl and EV loops
479support overloading, however, as well as all functions that use perl to
480invoke a callback (as in L<AnyEvent::Socket> and L<AnyEvent::DNS> for
481example).
482 675
483=item $cv->croak ($error) 676=item $cv->croak ($error)
484 677
485Similar to send, but causes all call's to C<< ->recv >> to invoke 678Similar to send, but causes all call's to C<< ->recv >> to invoke
486C<Carp::croak> with the given error message/object/scalar. 679C<Carp::croak> with the given error message/object/scalar.
487 680
488This can be used to signal any errors to the condition variable 681This can be used to signal any errors to the condition variable
489user/consumer. 682user/consumer. Doing it this way instead of calling C<croak> directly
683delays the error detetcion, but has the overwhelmign advantage that it
684diagnoses the error at the place where the result is expected, and not
685deep in some event clalback without connection to the actual code causing
686the problem.
490 687
491=item $cv->begin ([group callback]) 688=item $cv->begin ([group callback])
492 689
493=item $cv->end 690=item $cv->end
494
495These two methods are EXPERIMENTAL and MIGHT CHANGE.
496 691
497These two methods can be used to combine many transactions/events into 692These two methods can be used to combine many transactions/events into
498one. For example, a function that pings many hosts in parallel might want 693one. For example, a function that pings many hosts in parallel might want
499to use a condition variable for the whole process. 694to use a condition variable for the whole process.
500 695
501Every call to C<< ->begin >> will increment a counter, and every call to 696Every call to C<< ->begin >> will increment a counter, and every call to
502C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end 697C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
503>>, the (last) callback passed to C<begin> will be executed. That callback 698>>, the (last) callback passed to C<begin> will be executed, passing the
504is I<supposed> to call C<< ->send >>, but that is not required. If no 699condvar as first argument. That callback is I<supposed> to call C<< ->send
505callback was set, C<send> will be called without any arguments. 700>>, but that is not required. If no group callback was set, C<send> will
701be called without any arguments.
506 702
507Let's clarify this with the ping example: 703You can think of C<< $cv->send >> giving you an OR condition (one call
704sends), while C<< $cv->begin >> and C<< $cv->end >> giving you an AND
705condition (all C<begin> calls must be C<end>'ed before the condvar sends).
706
707Let's start with a simple example: you have two I/O watchers (for example,
708STDOUT and STDERR for a program), and you want to wait for both streams to
709close before activating a condvar:
508 710
509 my $cv = AnyEvent->condvar; 711 my $cv = AnyEvent->condvar;
510 712
713 $cv->begin; # first watcher
714 my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
715 defined sysread $fh1, my $buf, 4096
716 or $cv->end;
717 });
718
719 $cv->begin; # second watcher
720 my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
721 defined sysread $fh2, my $buf, 4096
722 or $cv->end;
723 });
724
725 $cv->recv;
726
727This works because for every event source (EOF on file handle), there is
728one call to C<begin>, so the condvar waits for all calls to C<end> before
729sending.
730
731The ping example mentioned above is slightly more complicated, as the
732there are results to be passwd back, and the number of tasks that are
733begung can potentially be zero:
734
735 my $cv = AnyEvent->condvar;
736
511 my %result; 737 my %result;
512 $cv->begin (sub { $cv->send (\%result) }); 738 $cv->begin (sub { shift->send (\%result) });
513 739
514 for my $host (@list_of_hosts) { 740 for my $host (@list_of_hosts) {
515 $cv->begin; 741 $cv->begin;
516 ping_host_then_call_callback $host, sub { 742 ping_host_then_call_callback $host, sub {
517 $result{$host} = ...; 743 $result{$host} = ...;
532loop, which serves two important purposes: first, it sets the callback 758loop, which serves two important purposes: first, it sets the callback
533to be called once the counter reaches C<0>, and second, it ensures that 759to be called once the counter reaches C<0>, and second, it ensures that
534C<send> is called even when C<no> hosts are being pinged (the loop 760C<send> is called even when C<no> hosts are being pinged (the loop
535doesn't execute once). 761doesn't execute once).
536 762
537This is the general pattern when you "fan out" into multiple subrequests: 763This is the general pattern when you "fan out" into multiple (but
538use an outer C<begin>/C<end> pair to set the callback and ensure C<end> 764potentially none) subrequests: use an outer C<begin>/C<end> pair to set
539is called at least once, and then, for each subrequest you start, call 765the callback and ensure C<end> is called at least once, and then, for each
540C<begin> and for each subrequest you finish, call C<end>. 766subrequest you start, call C<begin> and for each subrequest you finish,
767call C<end>.
541 768
542=back 769=back
543 770
544=head3 METHODS FOR CONSUMERS 771=head3 METHODS FOR CONSUMERS
545 772
561function will call C<croak>. 788function will call C<croak>.
562 789
563In list context, all parameters passed to C<send> will be returned, 790In list context, all parameters passed to C<send> will be returned,
564in scalar context only the first one will be returned. 791in scalar context only the first one will be returned.
565 792
793Note that doing a blocking wait in a callback is not supported by any
794event loop, that is, recursive invocation of a blocking C<< ->recv
795>> is not allowed, and the C<recv> call will C<croak> if such a
796condition is detected. This condition can be slightly loosened by using
797L<Coro::AnyEvent>, which allows you to do a blocking C<< ->recv >> from
798any thread that doesn't run the event loop itself.
799
566Not all event models support a blocking wait - some die in that case 800Not all event models support a blocking wait - some die in that case
567(programs might want to do that to stay interactive), so I<if you are 801(programs might want to do that to stay interactive), so I<if you are
568using this from a module, never require a blocking wait>, but let the 802using this from a module, never require a blocking wait>. Instead, let the
569caller decide whether the call will block or not (for example, by coupling 803caller decide whether the call will block or not (for example, by coupling
570condition variables with some kind of request results and supporting 804condition variables with some kind of request results and supporting
571callbacks so the caller knows that getting the result will not block, 805callbacks so the caller knows that getting the result will not block,
572while still supporting blocking waits if the caller so desires). 806while still supporting blocking waits if the caller so desires).
573 807
574Another reason I<never> to C<< ->recv >> in a module is that you cannot
575sensibly have two C<< ->recv >>'s in parallel, as that would require
576multiple interpreters or coroutines/threads, none of which C<AnyEvent>
577can supply.
578
579The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
580fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
581versions and also integrates coroutines into AnyEvent, making blocking
582C<< ->recv >> calls perfectly safe as long as they are done from another
583coroutine (one that doesn't run the event loop).
584
585You can ensure that C<< -recv >> never blocks by setting a callback and 808You can ensure that C<< -recv >> never blocks by setting a callback and
586only calling C<< ->recv >> from within that callback (or at a later 809only calling C<< ->recv >> from within that callback (or at a later
587time). This will work even when the event loop does not support blocking 810time). This will work even when the event loop does not support blocking
588waits otherwise. 811waits otherwise.
589 812
590=item $bool = $cv->ready 813=item $bool = $cv->ready
591 814
592Returns true when the condition is "true", i.e. whether C<send> or 815Returns true when the condition is "true", i.e. whether C<send> or
593C<croak> have been called. 816C<croak> have been called.
594 817
595=item $cb = $cv->cb ([new callback]) 818=item $cb = $cv->cb ($cb->($cv))
596 819
597This is a mutator function that returns the callback set and optionally 820This is a mutator function that returns the callback set and optionally
598replaces it before doing so. 821replaces it before doing so.
599 822
600The callback will be called when the condition becomes "true", i.e. when 823The callback will be called when the condition becomes (or already was)
601C<send> or C<croak> are called, with the only argument being the condition 824"true", i.e. when C<send> or C<croak> are called (or were called), with
602variable itself. Calling C<recv> inside the callback or at any later time 825the only argument being the condition variable itself. Calling C<recv>
603is guaranteed not to block. 826inside the callback or at any later time is guaranteed not to block.
604 827
605=back 828=back
606 829
830=head1 SUPPORTED EVENT LOOPS/BACKENDS
831
832The available backend classes are (every class has its own manpage):
833
834=over 4
835
836=item Backends that are autoprobed when no other event loop can be found.
837
838EV is the preferred backend when no other event loop seems to be in
839use. If EV is not installed, then AnyEvent will fall back to its own
840pure-perl implementation, which is available everywhere as it comes with
841AnyEvent itself.
842
843 AnyEvent::Impl::EV based on EV (interface to libev, best choice).
844 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
845
846=item Backends that are transparently being picked up when they are used.
847
848These will be used when they are currently loaded when the first watcher
849is created, in which case it is assumed that the application is using
850them. This means that AnyEvent will automatically pick the right backend
851when the main program loads an event module before anything starts to
852create watchers. Nothing special needs to be done by the main program.
853
854 AnyEvent::Impl::Event based on Event, very stable, few glitches.
855 AnyEvent::Impl::Glib based on Glib, slow but very stable.
856 AnyEvent::Impl::Tk based on Tk, very broken.
857 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
858 AnyEvent::Impl::POE based on POE, very slow, some limitations.
859 AnyEvent::Impl::Irssi used when running within irssi.
860
861=item Backends with special needs.
862
863Qt requires the Qt::Application to be instantiated first, but will
864otherwise be picked up automatically. As long as the main program
865instantiates the application before any AnyEvent watchers are created,
866everything should just work.
867
868 AnyEvent::Impl::Qt based on Qt.
869
870Support for IO::Async can only be partial, as it is too broken and
871architecturally limited to even support the AnyEvent API. It also
872is the only event loop that needs the loop to be set explicitly, so
873it can only be used by a main program knowing about AnyEvent. See
874L<AnyEvent::Impl::Async> for the gory details.
875
876 AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
877
878=item Event loops that are indirectly supported via other backends.
879
880Some event loops can be supported via other modules:
881
882There is no direct support for WxWidgets (L<Wx>) or L<Prima>.
883
884B<WxWidgets> has no support for watching file handles. However, you can
885use WxWidgets through the POE adaptor, as POE has a Wx backend that simply
886polls 20 times per second, which was considered to be too horrible to even
887consider for AnyEvent.
888
889B<Prima> is not supported as nobody seems to be using it, but it has a POE
890backend, so it can be supported through POE.
891
892AnyEvent knows about both L<Prima> and L<Wx>, however, and will try to
893load L<POE> when detecting them, in the hope that POE will pick them up,
894in which case everything will be automatic.
895
896=back
897
607=head1 GLOBAL VARIABLES AND FUNCTIONS 898=head1 GLOBAL VARIABLES AND FUNCTIONS
608 899
900These are not normally required to use AnyEvent, but can be useful to
901write AnyEvent extension modules.
902
609=over 4 903=over 4
610 904
611=item $AnyEvent::MODEL 905=item $AnyEvent::MODEL
612 906
613Contains C<undef> until the first watcher is being created. Then it 907Contains C<undef> until the first watcher is being created, before the
908backend has been autodetected.
909
614contains the event model that is being used, which is the name of the 910Afterwards it contains the event model that is being used, which is the
615Perl class implementing the model. This class is usually one of the 911name of the Perl class implementing the model. This class is usually one
616C<AnyEvent::Impl:xxx> modules, but can be any other class in the case 912of the C<AnyEvent::Impl:xxx> modules, but can be any other class in the
617AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>). 913case AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode> it
618 914will be C<urxvt::anyevent>).
619The known classes so far are:
620
621 AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
622 AnyEvent::Impl::Event based on Event, second best choice.
623 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
624 AnyEvent::Impl::Glib based on Glib, third-best choice.
625 AnyEvent::Impl::Tk based on Tk, very bad choice.
626 AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
627 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
628 AnyEvent::Impl::POE based on POE, not generic enough for full support.
629
630There is no support for WxWidgets, as WxWidgets has no support for
631watching file handles. However, you can use WxWidgets through the
632POE Adaptor, as POE has a Wx backend that simply polls 20 times per
633second, which was considered to be too horrible to even consider for
634AnyEvent. Likewise, other POE backends can be used by AnyEvent by using
635it's adaptor.
636
637AnyEvent knows about L<Prima> and L<Wx> and will try to use L<POE> when
638autodetecting them.
639 915
640=item AnyEvent::detect 916=item AnyEvent::detect
641 917
642Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model 918Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
643if necessary. You should only call this function right before you would 919if necessary. You should only call this function right before you would
644have created an AnyEvent watcher anyway, that is, as late as possible at 920have created an AnyEvent watcher anyway, that is, as late as possible at
645runtime. 921runtime, and not e.g. while initialising of your module.
922
923If you need to do some initialisation before AnyEvent watchers are
924created, use C<post_detect>.
646 925
647=item $guard = AnyEvent::post_detect { BLOCK } 926=item $guard = AnyEvent::post_detect { BLOCK }
648 927
649Arranges for the code block to be executed as soon as the event model is 928Arranges for the code block to be executed as soon as the event model is
650autodetected (or immediately if this has already happened). 929autodetected (or immediately if this has already happened).
651 930
931The block will be executed I<after> the actual backend has been detected
932(C<$AnyEvent::MODEL> is set), but I<before> any watchers have been
933created, so it is possible to e.g. patch C<@AnyEvent::ISA> or do
934other initialisations - see the sources of L<AnyEvent::Strict> or
935L<AnyEvent::AIO> to see how this is used.
936
937The most common usage is to create some global watchers, without forcing
938event module detection too early, for example, L<AnyEvent::AIO> creates
939and installs the global L<IO::AIO> watcher in a C<post_detect> block to
940avoid autodetecting the event module at load time.
941
652If called in scalar or list context, then it creates and returns an object 942If called in scalar or list context, then it creates and returns an object
653that automatically removes the callback again when it is destroyed. See 943that automatically removes the callback again when it is destroyed (or
944C<undef> when the hook was immediately executed). See L<AnyEvent::AIO> for
654L<Coro::BDB> for a case where this is useful. 945a case where this is useful.
946
947Example: Create a watcher for the IO::AIO module and store it in
948C<$WATCHER>. Only do so after the event loop is initialised, though.
949
950 our WATCHER;
951
952 my $guard = AnyEvent::post_detect {
953 $WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
954 };
955
956 # the ||= is important in case post_detect immediately runs the block,
957 # as to not clobber the newly-created watcher. assigning both watcher and
958 # post_detect guard to the same variable has the advantage of users being
959 # able to just C<undef $WATCHER> if the watcher causes them grief.
960
961 $WATCHER ||= $guard;
655 962
656=item @AnyEvent::post_detect 963=item @AnyEvent::post_detect
657 964
658If there are any code references in this array (you can C<push> to it 965If there are any code references in this array (you can C<push> to it
659before or after loading AnyEvent), then they will called directly after 966before or after loading AnyEvent), then they will called directly after
660the event loop has been chosen. 967the event loop has been chosen.
661 968
662You should check C<$AnyEvent::MODEL> before adding to this array, though: 969You should check C<$AnyEvent::MODEL> before adding to this array, though:
663if it contains a true value then the event loop has already been detected, 970if it is defined then the event loop has already been detected, and the
664and the array will be ignored. 971array will be ignored.
665 972
666Best use C<AnyEvent::post_detect { BLOCK }> instead. 973Best use C<AnyEvent::post_detect { BLOCK }> when your application allows
974it, as it takes care of these details.
975
976This variable is mainly useful for modules that can do something useful
977when AnyEvent is used and thus want to know when it is initialised, but do
978not need to even load it by default. This array provides the means to hook
979into AnyEvent passively, without loading it.
980
981Example: To load Coro::AnyEvent whenever Coro and AnyEvent are used
982together, you could put this into Coro (this is the actual code used by
983Coro to accomplish this):
984
985 if (defined $AnyEvent::MODEL) {
986 # AnyEvent already initialised, so load Coro::AnyEvent
987 require Coro::AnyEvent;
988 } else {
989 # AnyEvent not yet initialised, so make sure to load Coro::AnyEvent
990 # as soon as it is
991 push @AnyEvent::post_detect, sub { require Coro::AnyEvent };
992 }
667 993
668=back 994=back
669 995
670=head1 WHAT TO DO IN A MODULE 996=head1 WHAT TO DO IN A MODULE
671 997
726 1052
727 1053
728=head1 OTHER MODULES 1054=head1 OTHER MODULES
729 1055
730The following is a non-exhaustive list of additional modules that use 1056The following is a non-exhaustive list of additional modules that use
731AnyEvent and can therefore be mixed easily with other AnyEvent modules 1057AnyEvent as a client and can therefore be mixed easily with other AnyEvent
732in the same program. Some of the modules come with AnyEvent, some are 1058modules and other event loops in the same program. Some of the modules
733available via CPAN. 1059come with AnyEvent, most are available via CPAN.
734 1060
735=over 4 1061=over 4
736 1062
737=item L<AnyEvent::Util> 1063=item L<AnyEvent::Util>
738 1064
747 1073
748=item L<AnyEvent::Handle> 1074=item L<AnyEvent::Handle>
749 1075
750Provide read and write buffers, manages watchers for reads and writes, 1076Provide read and write buffers, manages watchers for reads and writes,
751supports raw and formatted I/O, I/O queued and fully transparent and 1077supports raw and formatted I/O, I/O queued and fully transparent and
752non-blocking SSL/TLS. 1078non-blocking SSL/TLS (via L<AnyEvent::TLS>.
753 1079
754=item L<AnyEvent::DNS> 1080=item L<AnyEvent::DNS>
755 1081
756Provides rich asynchronous DNS resolver capabilities. 1082Provides rich asynchronous DNS resolver capabilities.
757 1083
785 1111
786=item L<AnyEvent::GPSD> 1112=item L<AnyEvent::GPSD>
787 1113
788A non-blocking interface to gpsd, a daemon delivering GPS information. 1114A non-blocking interface to gpsd, a daemon delivering GPS information.
789 1115
1116=item L<AnyEvent::IRC>
1117
1118AnyEvent based IRC client module family (replacing the older Net::IRC3).
1119
1120=item L<AnyEvent::XMPP>
1121
1122AnyEvent based XMPP (Jabber protocol) module family (replacing the older
1123Net::XMPP2>.
1124
790=item L<AnyEvent::IGS> 1125=item L<AnyEvent::IGS>
791 1126
792A non-blocking interface to the Internet Go Server protocol (used by 1127A non-blocking interface to the Internet Go Server protocol (used by
793L<App::IGS>). 1128L<App::IGS>).
794 1129
795=item L<Net::IRC3>
796
797AnyEvent based IRC client module family.
798
799=item L<Net::XMPP2>
800
801AnyEvent based XMPP (Jabber protocol) module family.
802
803=item L<Net::FCP> 1130=item L<Net::FCP>
804 1131
805AnyEvent-based implementation of the Freenet Client Protocol, birthplace 1132AnyEvent-based implementation of the Freenet Client Protocol, birthplace
806of AnyEvent. 1133of AnyEvent.
807 1134
811 1138
812=item L<Coro> 1139=item L<Coro>
813 1140
814Has special support for AnyEvent via L<Coro::AnyEvent>. 1141Has special support for AnyEvent via L<Coro::AnyEvent>.
815 1142
816=item L<IO::Lambda>
817
818The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
819
820=back 1143=back
821 1144
822=cut 1145=cut
823 1146
824package AnyEvent; 1147package AnyEvent;
825 1148
826no warnings; 1149# basically a tuned-down version of common::sense
827use strict; 1150sub common_sense {
1151 # from common:.sense 1.0
1152 ${^WARNING_BITS} = "\xfc\x3f\x33\x00\x0f\xf3\xcf\xc0\xf3\xfc\x33\x00";
1153 # use strict vars subs - NO UTF-8, as Util.pm doesn't like this atm. (uts46data.pl)
1154 $^H |= 0x00000600;
1155}
828 1156
1157BEGIN { AnyEvent::common_sense }
1158
829use Carp; 1159use Carp ();
830 1160
831our $VERSION = 4.21; 1161our $VERSION = '5.251';
832our $MODEL; 1162our $MODEL;
833 1163
834our $AUTOLOAD; 1164our $AUTOLOAD;
835our @ISA; 1165our @ISA;
836 1166
837our @REGISTRY; 1167our @REGISTRY;
838 1168
839our $WIN32; 1169our $VERBOSE;
840 1170
841BEGIN { 1171BEGIN {
842 my $win32 = ! ! ($^O =~ /mswin32/i); 1172 require "AnyEvent/constants.pl";
843 eval "sub WIN32(){ $win32 }";
844}
845 1173
1174 eval "sub TAINT (){" . (${^TAINT}*1) . "}";
1175
1176 delete @ENV{grep /^PERL_ANYEVENT_/, keys %ENV}
1177 if ${^TAINT};
1178
846our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1; 1179 $VERBOSE = $ENV{PERL_ANYEVENT_VERBOSE}*1;
1180
1181}
1182
1183our $MAX_SIGNAL_LATENCY = 10;
847 1184
848our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred 1185our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
849 1186
850{ 1187{
851 my $idx; 1188 my $idx;
853 for reverse split /\s*,\s*/, 1190 for reverse split /\s*,\s*/,
854 $ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6"; 1191 $ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
855} 1192}
856 1193
857my @models = ( 1194my @models = (
858 [EV:: => AnyEvent::Impl::EV::], 1195 [EV:: => AnyEvent::Impl::EV:: , 1],
859 [Event:: => AnyEvent::Impl::Event::],
860 [AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::], 1196 [AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl:: , 1],
861 # everything below here will not be autoprobed 1197 # everything below here will not (normally) be autoprobed
862 # as the pureperl backend should work everywhere 1198 # as the pureperl backend should work everywhere
863 # and is usually faster 1199 # and is usually faster
1200 [Event:: => AnyEvent::Impl::Event::, 1],
1201 [Glib:: => AnyEvent::Impl::Glib:: , 1], # becomes extremely slow with many watchers
1202 [Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
1203 [Irssi:: => AnyEvent::Impl::Irssi::], # Irssi has a bogus "Event" package
864 [Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles 1204 [Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
865 [Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
866 [Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
867 [Qt:: => AnyEvent::Impl::Qt::], # requires special main program 1205 [Qt:: => AnyEvent::Impl::Qt::], # requires special main program
868 [POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza 1206 [POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
869 [Wx:: => AnyEvent::Impl::POE::], 1207 [Wx:: => AnyEvent::Impl::POE::],
870 [Prima:: => AnyEvent::Impl::POE::], 1208 [Prima:: => AnyEvent::Impl::POE::],
1209 # IO::Async is just too broken - we would need workarounds for its
1210 # byzantine signal and broken child handling, among others.
1211 # IO::Async is rather hard to detect, as it doesn't have any
1212 # obvious default class.
1213 [IO::Async:: => AnyEvent::Impl::IOAsync::], # requires special main program
1214 [IO::Async::Loop:: => AnyEvent::Impl::IOAsync::], # requires special main program
1215 [IO::Async::Notifier:: => AnyEvent::Impl::IOAsync::], # requires special main program
1216 [AnyEvent::Impl::IOAsync:: => AnyEvent::Impl::IOAsync::], # requires special main program
871); 1217);
872 1218
873our %method = map +($_ => 1), qw(io timer time now signal child condvar one_event DESTROY); 1219our %method = map +($_ => 1),
1220 qw(io timer time now now_update signal child idle condvar one_event DESTROY);
874 1221
875our @post_detect; 1222our @post_detect;
876 1223
877sub post_detect(&) { 1224sub post_detect(&) {
878 my ($cb) = @_; 1225 my ($cb) = @_;
879 1226
880 if ($MODEL) {
881 $cb->();
882
883 1
884 } else {
885 push @post_detect, $cb; 1227 push @post_detect, $cb;
886 1228
887 defined wantarray 1229 defined wantarray
888 ? bless \$cb, "AnyEvent::Util::PostDetect" 1230 ? bless \$cb, "AnyEvent::Util::postdetect"
889 : () 1231 : ()
1232}
1233
1234sub AnyEvent::Util::postdetect::DESTROY {
1235 @post_detect = grep $_ != ${$_[0]}, @post_detect;
1236}
1237
1238sub detect() {
1239 # free some memory
1240 *detect = sub () { $MODEL };
1241
1242 local $!; # for good measure
1243 local $SIG{__DIE__};
1244
1245 if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
1246 my $model = "AnyEvent::Impl::$1";
1247 if (eval "require $model") {
1248 $MODEL = $model;
1249 warn "AnyEvent: loaded model '$model' (forced by \$ENV{PERL_ANYEVENT_MODEL}), using it.\n" if $VERBOSE >= 2;
1250 } else {
1251 warn "AnyEvent: unable to load model '$model' (from \$ENV{PERL_ANYEVENT_MODEL}):\n$@" if $VERBOSE;
1252 }
890 } 1253 }
891}
892 1254
893sub AnyEvent::Util::PostDetect::DESTROY { 1255 # check for already loaded models
894 @post_detect = grep $_ != ${$_[0]}, @post_detect;
895}
896
897sub detect() {
898 unless ($MODEL) { 1256 unless ($MODEL) {
899 no strict 'refs'; 1257 for (@REGISTRY, @models) {
900 local $SIG{__DIE__}; 1258 my ($package, $model) = @$_;
901 1259 if (${"$package\::VERSION"} > 0) {
902 if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
903 my $model = "AnyEvent::Impl::$1";
904 if (eval "require $model") { 1260 if (eval "require $model") {
905 $MODEL = $model; 1261 $MODEL = $model;
906 warn "AnyEvent: loaded model '$model' (forced by \$PERL_ANYEVENT_MODEL), using it.\n" if $verbose > 1; 1262 warn "AnyEvent: autodetected model '$model', using it.\n" if $VERBOSE >= 2;
907 } else { 1263 last;
908 warn "AnyEvent: unable to load model '$model' (from \$PERL_ANYEVENT_MODEL):\n$@" if $verbose; 1264 }
909 } 1265 }
910 } 1266 }
911 1267
912 # check for already loaded models
913 unless ($MODEL) { 1268 unless ($MODEL) {
1269 # try to autoload a model
914 for (@REGISTRY, @models) { 1270 for (@REGISTRY, @models) {
915 my ($package, $model) = @$_; 1271 my ($package, $model, $autoload) = @$_;
1272 if (
1273 $autoload
1274 and eval "require $package"
916 if (${"$package\::VERSION"} > 0) { 1275 and ${"$package\::VERSION"} > 0
917 if (eval "require $model") { 1276 and eval "require $model"
1277 ) {
918 $MODEL = $model; 1278 $MODEL = $model;
919 warn "AnyEvent: autodetected model '$model', using it.\n" if $verbose > 1; 1279 warn "AnyEvent: autoloaded model '$model', using it.\n" if $VERBOSE >= 2;
920 last; 1280 last;
921 }
922 } 1281 }
923 } 1282 }
924 1283
925 unless ($MODEL) {
926 # try to load a model
927
928 for (@REGISTRY, @models) {
929 my ($package, $model) = @$_;
930 if (eval "require $package"
931 and ${"$package\::VERSION"} > 0
932 and eval "require $model") {
933 $MODEL = $model;
934 warn "AnyEvent: autoprobed model '$model', using it.\n" if $verbose > 1;
935 last;
936 }
937 }
938
939 $MODEL 1284 $MODEL
940 or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib."; 1285 or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.\n";
941 }
942 } 1286 }
943
944 push @{"$MODEL\::ISA"}, "AnyEvent::Base";
945
946 unshift @ISA, $MODEL;
947
948 require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
949
950 (shift @post_detect)->() while @post_detect;
951 } 1287 }
1288
1289 @models = (); # free probe data
1290
1291 push @{"$MODEL\::ISA"}, "AnyEvent::Base";
1292 unshift @ISA, $MODEL;
1293
1294 # now nuke some methods that are overriden by the backend.
1295 # SUPER is not allowed.
1296 for (qw(time signal child idle)) {
1297 undef &{"AnyEvent::Base::$_"}
1298 if defined &{"$MODEL\::$_"};
1299 }
1300
1301 require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
1302
1303 (shift @post_detect)->() while @post_detect;
1304
1305 *post_detect = sub(&) {
1306 shift->();
1307
1308 undef
1309 };
952 1310
953 $MODEL 1311 $MODEL
954} 1312}
955 1313
956sub AUTOLOAD { 1314sub AUTOLOAD {
957 (my $func = $AUTOLOAD) =~ s/.*://; 1315 (my $func = $AUTOLOAD) =~ s/.*://;
958 1316
959 $method{$func} 1317 $method{$func}
960 or croak "$func: not a valid method for AnyEvent objects"; 1318 or Carp::croak "$func: not a valid AnyEvent class method";
961 1319
962 detect unless $MODEL; 1320 detect;
963 1321
964 my $class = shift; 1322 my $class = shift;
965 $class->$func (@_); 1323 $class->$func (@_);
966} 1324}
967 1325
968# utility function to dup a filehandle. this is used by many backends 1326# utility function to dup a filehandle. this is used by many backends
969# to support binding more than one watcher per filehandle (they usually 1327# to support binding more than one watcher per filehandle (they usually
970# allow only one watcher per fd, so we dup it to get a different one). 1328# allow only one watcher per fd, so we dup it to get a different one).
971sub _dupfh($$$$) { 1329sub _dupfh($$;$$) {
972 my ($poll, $fh, $r, $w) = @_; 1330 my ($poll, $fh, $r, $w) = @_;
973 1331
974 require Fcntl;
975
976 # cygwin requires the fh mode to be matching, unix doesn't 1332 # cygwin requires the fh mode to be matching, unix doesn't
977 my ($rw, $mode) = $poll eq "r" ? ($r, "<") 1333 my ($rw, $mode) = $poll eq "r" ? ($r, "<&") : ($w, ">&");
978 : $poll eq "w" ? ($w, ">")
979 : Carp::croak "AnyEvent->io requires poll set to either 'r' or 'w'";
980 1334
981 open my $fh2, "$mode&" . fileno $fh 1335 open my $fh2, $mode, $fh
982 or die "cannot dup() filehandle: $!"; 1336 or die "AnyEvent->io: cannot dup() filehandle in mode '$poll': $!,";
983 1337
984 # we assume CLOEXEC is already set by perl in all important cases 1338 # we assume CLOEXEC is already set by perl in all important cases
985 1339
986 ($fh2, $rw) 1340 ($fh2, $rw)
987} 1341}
988 1342
1343=head1 SIMPLIFIED AE API
1344
1345Starting with version 5.0, AnyEvent officially supports a second, much
1346simpler, API that is designed to reduce the calling, typing and memory
1347overhead.
1348
1349See the L<AE> manpage for details.
1350
1351=cut
1352
1353package AE;
1354
1355our $VERSION = $AnyEvent::VERSION;
1356
1357sub io($$$) {
1358 AnyEvent->io (fh => $_[0], poll => $_[1] ? "w" : "r", cb => $_[2])
1359}
1360
1361sub timer($$$) {
1362 AnyEvent->timer (after => $_[0], interval => $_[1], cb => $_[2])
1363}
1364
1365sub signal($$) {
1366 AnyEvent->signal (signal => $_[0], cb => $_[1])
1367}
1368
1369sub child($$) {
1370 AnyEvent->child (pid => $_[0], cb => $_[1])
1371}
1372
1373sub idle($) {
1374 AnyEvent->idle (cb => $_[0])
1375}
1376
1377sub cv(;&) {
1378 AnyEvent->condvar (@_ ? (cb => $_[0]) : ())
1379}
1380
1381sub now() {
1382 AnyEvent->now
1383}
1384
1385sub now_update() {
1386 AnyEvent->now_update
1387}
1388
1389sub time() {
1390 AnyEvent->time
1391}
1392
989package AnyEvent::Base; 1393package AnyEvent::Base;
990 1394
991# default implementation for now and time 1395# default implementations for many methods
992 1396
993use Time::HiRes (); 1397sub time {
1398 eval q{ # poor man's autoloading {}
1399 # probe for availability of Time::HiRes
1400 if (eval "use Time::HiRes (); Time::HiRes::time (); 1") {
1401 warn "AnyEvent: using Time::HiRes for sub-second timing accuracy.\n" if $VERBOSE >= 8;
1402 *AE::time = \&Time::HiRes::time;
1403 # if (eval "use POSIX (); (POSIX::times())...
1404 } else {
1405 warn "AnyEvent: using built-in time(), WARNING, no sub-second resolution!\n" if $VERBOSE;
1406 *AE::time = sub (){ time }; # epic fail
1407 }
994 1408
995sub time { Time::HiRes::time } 1409 *time = sub { AE::time }; # different prototypes
996sub now { Time::HiRes::time } 1410 };
1411 die if $@;
1412
1413 &time
1414}
1415
1416*now = \&time;
1417
1418sub now_update { }
997 1419
998# default implementation for ->condvar 1420# default implementation for ->condvar
999 1421
1000sub condvar { 1422sub condvar {
1423 eval q{ # poor man's autoloading {}
1424 *condvar = sub {
1001 bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, AnyEvent::CondVar:: 1425 bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, "AnyEvent::CondVar"
1426 };
1427
1428 *AE::cv = sub (;&) {
1429 bless { @_ ? (_ae_cb => shift) : () }, "AnyEvent::CondVar"
1430 };
1431 };
1432 die if $@;
1433
1434 &condvar
1002} 1435}
1003 1436
1004# default implementation for ->signal 1437# default implementation for ->signal
1005 1438
1006our %SIG_CB; 1439our $HAVE_ASYNC_INTERRUPT;
1440
1441sub _have_async_interrupt() {
1442 $HAVE_ASYNC_INTERRUPT = 1*(!$ENV{PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT}
1443 && eval "use Async::Interrupt 1.02 (); 1")
1444 unless defined $HAVE_ASYNC_INTERRUPT;
1445
1446 $HAVE_ASYNC_INTERRUPT
1447}
1448
1449our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
1450our (%SIG_ASY, %SIG_ASY_W);
1451our ($SIG_COUNT, $SIG_TW);
1452
1453# install a dummy wakeup watcher to reduce signal catching latency
1454# used by Impls
1455sub _sig_add() {
1456 unless ($SIG_COUNT++) {
1457 # try to align timer on a full-second boundary, if possible
1458 my $NOW = AE::now;
1459
1460 $SIG_TW = AE::timer
1461 $MAX_SIGNAL_LATENCY - ($NOW - int $NOW),
1462 $MAX_SIGNAL_LATENCY,
1463 sub { } # just for the PERL_ASYNC_CHECK
1464 ;
1465 }
1466}
1467
1468sub _sig_del {
1469 undef $SIG_TW
1470 unless --$SIG_COUNT;
1471}
1472
1473our $_sig_name_init; $_sig_name_init = sub {
1474 eval q{ # poor man's autoloading {}
1475 undef $_sig_name_init;
1476
1477 if (_have_async_interrupt) {
1478 *sig2num = \&Async::Interrupt::sig2num;
1479 *sig2name = \&Async::Interrupt::sig2name;
1480 } else {
1481 require Config;
1482
1483 my %signame2num;
1484 @signame2num{ split ' ', $Config::Config{sig_name} }
1485 = split ' ', $Config::Config{sig_num};
1486
1487 my @signum2name;
1488 @signum2name[values %signame2num] = keys %signame2num;
1489
1490 *sig2num = sub($) {
1491 $_[0] > 0 ? shift : $signame2num{+shift}
1492 };
1493 *sig2name = sub ($) {
1494 $_[0] > 0 ? $signum2name[+shift] : shift
1495 };
1496 }
1497 };
1498 die if $@;
1499};
1500
1501sub sig2num ($) { &$_sig_name_init; &sig2num }
1502sub sig2name($) { &$_sig_name_init; &sig2name }
1007 1503
1008sub signal { 1504sub signal {
1505 eval q{ # poor man's autoloading {}
1506 # probe for availability of Async::Interrupt
1507 if (_have_async_interrupt) {
1508 warn "AnyEvent: using Async::Interrupt for race-free signal handling.\n" if $VERBOSE >= 8;
1509
1510 $SIGPIPE_R = new Async::Interrupt::EventPipe;
1511 $SIG_IO = AE::io $SIGPIPE_R->fileno, 0, \&_signal_exec;
1512
1513 } else {
1514 warn "AnyEvent: using emulated perl signal handling with latency timer.\n" if $VERBOSE >= 8;
1515
1516 if (AnyEvent::WIN32) {
1517 require AnyEvent::Util;
1518
1519 ($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
1520 AnyEvent::Util::fh_nonblocking ($SIGPIPE_R, 1) if $SIGPIPE_R;
1521 AnyEvent::Util::fh_nonblocking ($SIGPIPE_W, 1) if $SIGPIPE_W; # just in case
1522 } else {
1523 pipe $SIGPIPE_R, $SIGPIPE_W;
1524 fcntl $SIGPIPE_R, AnyEvent::F_SETFL, AnyEvent::O_NONBLOCK if $SIGPIPE_R;
1525 fcntl $SIGPIPE_W, AnyEvent::F_SETFL, AnyEvent::O_NONBLOCK if $SIGPIPE_W; # just in case
1526
1527 # not strictly required, as $^F is normally 2, but let's make sure...
1528 fcntl $SIGPIPE_R, AnyEvent::F_SETFD, AnyEvent::FD_CLOEXEC;
1529 fcntl $SIGPIPE_W, AnyEvent::F_SETFD, AnyEvent::FD_CLOEXEC;
1530 }
1531
1532 $SIGPIPE_R
1533 or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
1534
1535 $SIG_IO = AE::io $SIGPIPE_R, 0, \&_signal_exec;
1536 }
1537
1538 *signal = $HAVE_ASYNC_INTERRUPT
1539 ? sub {
1009 my (undef, %arg) = @_; 1540 my (undef, %arg) = @_;
1010 1541
1542 # async::interrupt
1011 my $signal = uc $arg{signal} 1543 my $signal = sig2num $arg{signal};
1012 or Carp::croak "required option 'signal' is missing";
1013
1014 $SIG_CB{$signal}{$arg{cb}} = $arg{cb}; 1544 $SIG_CB{$signal}{$arg{cb}} = $arg{cb};
1545
1546 $SIG_ASY{$signal} ||= new Async::Interrupt
1547 cb => sub { undef $SIG_EV{$signal} },
1548 signal => $signal,
1549 pipe => [$SIGPIPE_R->filenos],
1550 pipe_autodrain => 0,
1551 ;
1552
1553 bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
1554 }
1555 : sub {
1556 my (undef, %arg) = @_;
1557
1558 # pure perl
1559 my $signal = sig2name $arg{signal};
1560 $SIG_CB{$signal}{$arg{cb}} = $arg{cb};
1561
1015 $SIG{$signal} ||= sub { 1562 $SIG{$signal} ||= sub {
1563 local $!;
1564 syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
1565 undef $SIG_EV{$signal};
1566 };
1567
1568 # can't do signal processing without introducing races in pure perl,
1569 # so limit the signal latency.
1570 _sig_add;
1571
1572 bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
1573 }
1574 ;
1575
1576 *AnyEvent::Base::signal::DESTROY = sub {
1577 my ($signal, $cb) = @{$_[0]};
1578
1579 _sig_del;
1580
1581 delete $SIG_CB{$signal}{$cb};
1582
1583 $HAVE_ASYNC_INTERRUPT
1584 ? delete $SIG_ASY{$signal}
1585 : # delete doesn't work with older perls - they then
1586 # print weird messages, or just unconditionally exit
1587 # instead of getting the default action.
1588 undef $SIG{$signal}
1589 unless keys %{ $SIG_CB{$signal} };
1590 };
1591
1592 *_signal_exec = sub {
1593 $HAVE_ASYNC_INTERRUPT
1594 ? $SIGPIPE_R->drain
1595 : sysread $SIGPIPE_R, (my $dummy), 9;
1596
1597 while (%SIG_EV) {
1598 for (keys %SIG_EV) {
1599 delete $SIG_EV{$_};
1016 $_->() for values %{ $SIG_CB{$signal} || {} }; 1600 $_->() for values %{ $SIG_CB{$_} || {} };
1601 }
1602 }
1603 };
1017 }; 1604 };
1605 die if $@;
1018 1606
1019 bless [$signal, $arg{cb}], "AnyEvent::Base::Signal" 1607 &signal
1020}
1021
1022sub AnyEvent::Base::Signal::DESTROY {
1023 my ($signal, $cb) = @{$_[0]};
1024
1025 delete $SIG_CB{$signal}{$cb};
1026
1027 delete $SIG{$signal} unless keys %{ $SIG_CB{$signal} };
1028} 1608}
1029 1609
1030# default implementation for ->child 1610# default implementation for ->child
1031 1611
1032our %PID_CB; 1612our %PID_CB;
1033our $CHLD_W; 1613our $CHLD_W;
1034our $CHLD_DELAY_W; 1614our $CHLD_DELAY_W;
1035our $PID_IDLE;
1036our $WNOHANG; 1615our $WNOHANG;
1037 1616
1038sub _child_wait { 1617# used by many Impl's
1039 while (0 < (my $pid = waitpid -1, $WNOHANG)) { 1618sub _emit_childstatus($$) {
1619 my (undef, $rpid, $rstatus) = @_;
1620
1621 $_->($rpid, $rstatus)
1040 $_->($pid, $?) for (values %{ $PID_CB{$pid} || {} }), 1622 for values %{ $PID_CB{$rpid} || {} },
1041 (values %{ $PID_CB{0} || {} }); 1623 values %{ $PID_CB{0} || {} };
1042 }
1043
1044 undef $PID_IDLE;
1045}
1046
1047sub _sigchld {
1048 # make sure we deliver these changes "synchronous" with the event loop.
1049 $CHLD_DELAY_W ||= AnyEvent->timer (after => 0, cb => sub {
1050 undef $CHLD_DELAY_W;
1051 &_child_wait;
1052 });
1053} 1624}
1054 1625
1055sub child { 1626sub child {
1627 eval q{ # poor man's autoloading {}
1628 *_sigchld = sub {
1629 my $pid;
1630
1631 AnyEvent->_emit_childstatus ($pid, $?)
1632 while ($pid = waitpid -1, $WNOHANG) > 0;
1633 };
1634
1635 *child = sub {
1056 my (undef, %arg) = @_; 1636 my (undef, %arg) = @_;
1057 1637
1058 defined (my $pid = $arg{pid} + 0) 1638 defined (my $pid = $arg{pid} + 0)
1059 or Carp::croak "required option 'pid' is missing"; 1639 or Carp::croak "required option 'pid' is missing";
1060 1640
1061 $PID_CB{$pid}{$arg{cb}} = $arg{cb}; 1641 $PID_CB{$pid}{$arg{cb}} = $arg{cb};
1062 1642
1063 unless ($WNOHANG) { 1643 # WNOHANG is almost cetrainly 1 everywhere
1644 $WNOHANG ||= $^O =~ /^(?:openbsd|netbsd|linux|freebsd|cygwin|MSWin32)$/
1645 ? 1
1064 $WNOHANG = eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1; 1646 : eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
1065 }
1066 1647
1067 unless ($CHLD_W) { 1648 unless ($CHLD_W) {
1068 $CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld); 1649 $CHLD_W = AE::signal CHLD => \&_sigchld;
1069 # child could be a zombie already, so make at least one round 1650 # child could be a zombie already, so make at least one round
1070 &_sigchld; 1651 &_sigchld;
1071 } 1652 }
1072 1653
1073 bless [$pid, $arg{cb}], "AnyEvent::Base::Child" 1654 bless [$pid, $arg{cb}], "AnyEvent::Base::child"
1074} 1655 };
1075 1656
1076sub AnyEvent::Base::Child::DESTROY { 1657 *AnyEvent::Base::child::DESTROY = sub {
1077 my ($pid, $cb) = @{$_[0]}; 1658 my ($pid, $cb) = @{$_[0]};
1078 1659
1079 delete $PID_CB{$pid}{$cb}; 1660 delete $PID_CB{$pid}{$cb};
1080 delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} }; 1661 delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
1081 1662
1082 undef $CHLD_W unless keys %PID_CB; 1663 undef $CHLD_W unless keys %PID_CB;
1664 };
1665 };
1666 die if $@;
1667
1668 &child
1669}
1670
1671# idle emulation is done by simply using a timer, regardless
1672# of whether the process is idle or not, and not letting
1673# the callback use more than 50% of the time.
1674sub idle {
1675 eval q{ # poor man's autoloading {}
1676 *idle = sub {
1677 my (undef, %arg) = @_;
1678
1679 my ($cb, $w, $rcb) = $arg{cb};
1680
1681 $rcb = sub {
1682 if ($cb) {
1683 $w = _time;
1684 &$cb;
1685 $w = _time - $w;
1686
1687 # never use more then 50% of the time for the idle watcher,
1688 # within some limits
1689 $w = 0.0001 if $w < 0.0001;
1690 $w = 5 if $w > 5;
1691
1692 $w = AE::timer $w, 0, $rcb;
1693 } else {
1694 # clean up...
1695 undef $w;
1696 undef $rcb;
1697 }
1698 };
1699
1700 $w = AE::timer 0.05, 0, $rcb;
1701
1702 bless \\$cb, "AnyEvent::Base::idle"
1703 };
1704
1705 *AnyEvent::Base::idle::DESTROY = sub {
1706 undef $${$_[0]};
1707 };
1708 };
1709 die if $@;
1710
1711 &idle
1083} 1712}
1084 1713
1085package AnyEvent::CondVar; 1714package AnyEvent::CondVar;
1086 1715
1087our @ISA = AnyEvent::CondVar::Base::; 1716our @ISA = AnyEvent::CondVar::Base::;
1088 1717
1089package AnyEvent::CondVar::Base; 1718package AnyEvent::CondVar::Base;
1090 1719
1091use overload 1720#use overload
1092 '&{}' => sub { my $self = shift; sub { $self->send (@_) } }, 1721# '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
1093 fallback => 1; 1722# fallback => 1;
1723
1724# save 300+ kilobytes by dirtily hardcoding overloading
1725${"AnyEvent::CondVar::Base::OVERLOAD"}{dummy}++; # Register with magic by touching.
1726*{'AnyEvent::CondVar::Base::()'} = sub { }; # "Make it findable via fetchmethod."
1727*{'AnyEvent::CondVar::Base::(&{}'} = sub { my $self = shift; sub { $self->send (@_) } }; # &{}
1728${'AnyEvent::CondVar::Base::()'} = 1; # fallback
1729
1730our $WAITING;
1094 1731
1095sub _send { 1732sub _send {
1096 # nop 1733 # nop
1097} 1734}
1098 1735
1111sub ready { 1748sub ready {
1112 $_[0]{_ae_sent} 1749 $_[0]{_ae_sent}
1113} 1750}
1114 1751
1115sub _wait { 1752sub _wait {
1753 $WAITING
1754 and !$_[0]{_ae_sent}
1755 and Carp::croak "AnyEvent::CondVar: recursive blocking wait detected";
1756
1757 local $WAITING = 1;
1116 AnyEvent->one_event while !$_[0]{_ae_sent}; 1758 AnyEvent->one_event while !$_[0]{_ae_sent};
1117} 1759}
1118 1760
1119sub recv { 1761sub recv {
1120 $_[0]->_wait; 1762 $_[0]->_wait;
1122 Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak}; 1764 Carp::croak $_[0]{_ae_croak} if $_[0]{_ae_croak};
1123 wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0] 1765 wantarray ? @{ $_[0]{_ae_sent} } : $_[0]{_ae_sent}[0]
1124} 1766}
1125 1767
1126sub cb { 1768sub cb {
1127 $_[0]{_ae_cb} = $_[1] if @_ > 1; 1769 my $cv = shift;
1770
1771 @_
1772 and $cv->{_ae_cb} = shift
1773 and $cv->{_ae_sent}
1774 and (delete $cv->{_ae_cb})->($cv);
1775
1128 $_[0]{_ae_cb} 1776 $cv->{_ae_cb}
1129} 1777}
1130 1778
1131sub begin { 1779sub begin {
1132 ++$_[0]{_ae_counter}; 1780 ++$_[0]{_ae_counter};
1133 $_[0]{_ae_end_cb} = $_[1] if @_ > 1; 1781 $_[0]{_ae_end_cb} = $_[1] if @_ > 1;
1139} 1787}
1140 1788
1141# undocumented/compatibility with pre-3.4 1789# undocumented/compatibility with pre-3.4
1142*broadcast = \&send; 1790*broadcast = \&send;
1143*wait = \&_wait; 1791*wait = \&_wait;
1792
1793=head1 ERROR AND EXCEPTION HANDLING
1794
1795In general, AnyEvent does not do any error handling - it relies on the
1796caller to do that if required. The L<AnyEvent::Strict> module (see also
1797the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
1798checking of all AnyEvent methods, however, which is highly useful during
1799development.
1800
1801As for exception handling (i.e. runtime errors and exceptions thrown while
1802executing a callback), this is not only highly event-loop specific, but
1803also not in any way wrapped by this module, as this is the job of the main
1804program.
1805
1806The pure perl event loop simply re-throws the exception (usually
1807within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
1808$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
1809so on.
1810
1811=head1 ENVIRONMENT VARIABLES
1812
1813The following environment variables are used by this module or its
1814submodules.
1815
1816Note that AnyEvent will remove I<all> environment variables starting with
1817C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
1818enabled.
1819
1820=over 4
1821
1822=item C<PERL_ANYEVENT_VERBOSE>
1823
1824By default, AnyEvent will be completely silent except in fatal
1825conditions. You can set this environment variable to make AnyEvent more
1826talkative.
1827
1828When set to C<1> or higher, causes AnyEvent to warn about unexpected
1829conditions, such as not being able to load the event model specified by
1830C<PERL_ANYEVENT_MODEL>.
1831
1832When set to C<2> or higher, cause AnyEvent to report to STDERR which event
1833model it chooses.
1834
1835When set to C<8> or higher, then AnyEvent will report extra information on
1836which optional modules it loads and how it implements certain features.
1837
1838=item C<PERL_ANYEVENT_STRICT>
1839
1840AnyEvent does not do much argument checking by default, as thorough
1841argument checking is very costly. Setting this variable to a true value
1842will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
1843check the arguments passed to most method calls. If it finds any problems,
1844it will croak.
1845
1846In other words, enables "strict" mode.
1847
1848Unlike C<use strict> (or it's modern cousin, C<< use L<common::sense>
1849>>, it is definitely recommended to keep it off in production. Keeping
1850C<PERL_ANYEVENT_STRICT=1> in your environment while developing programs
1851can be very useful, however.
1852
1853=item C<PERL_ANYEVENT_MODEL>
1854
1855This can be used to specify the event model to be used by AnyEvent, before
1856auto detection and -probing kicks in. It must be a string consisting
1857entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
1858and the resulting module name is loaded and if the load was successful,
1859used as event model. If it fails to load AnyEvent will proceed with
1860auto detection and -probing.
1861
1862This functionality might change in future versions.
1863
1864For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
1865could start your program like this:
1866
1867 PERL_ANYEVENT_MODEL=Perl perl ...
1868
1869=item C<PERL_ANYEVENT_PROTOCOLS>
1870
1871Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
1872for IPv4 or IPv6. The default is unspecified (and might change, or be the result
1873of auto probing).
1874
1875Must be set to a comma-separated list of protocols or address families,
1876current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
1877used, and preference will be given to protocols mentioned earlier in the
1878list.
1879
1880This variable can effectively be used for denial-of-service attacks
1881against local programs (e.g. when setuid), although the impact is likely
1882small, as the program has to handle conenction and other failures anyways.
1883
1884Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
1885but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
1886- only support IPv4, never try to resolve or contact IPv6
1887addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
1888IPv6, but prefer IPv6 over IPv4.
1889
1890=item C<PERL_ANYEVENT_EDNS0>
1891
1892Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
1893for DNS. This extension is generally useful to reduce DNS traffic, but
1894some (broken) firewalls drop such DNS packets, which is why it is off by
1895default.
1896
1897Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
1898EDNS0 in its DNS requests.
1899
1900=item C<PERL_ANYEVENT_MAX_FORKS>
1901
1902The maximum number of child processes that C<AnyEvent::Util::fork_call>
1903will create in parallel.
1904
1905=item C<PERL_ANYEVENT_MAX_OUTSTANDING_DNS>
1906
1907The default value for the C<max_outstanding> parameter for the default DNS
1908resolver - this is the maximum number of parallel DNS requests that are
1909sent to the DNS server.
1910
1911=item C<PERL_ANYEVENT_RESOLV_CONF>
1912
1913The file to use instead of F</etc/resolv.conf> (or OS-specific
1914configuration) in the default resolver. When set to the empty string, no
1915default config will be used.
1916
1917=item C<PERL_ANYEVENT_CA_FILE>, C<PERL_ANYEVENT_CA_PATH>.
1918
1919When neither C<ca_file> nor C<ca_path> was specified during
1920L<AnyEvent::TLS> context creation, and either of these environment
1921variables exist, they will be used to specify CA certificate locations
1922instead of a system-dependent default.
1923
1924=item C<PERL_ANYEVENT_AVOID_GUARD> and C<PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT>
1925
1926When these are set to C<1>, then the respective modules are not
1927loaded. Mostly good for testing AnyEvent itself.
1928
1929=back
1144 1930
1145=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE 1931=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
1146 1932
1147This is an advanced topic that you do not normally need to use AnyEvent in 1933This is an advanced topic that you do not normally need to use AnyEvent in
1148a module. This section is only of use to event loop authors who want to 1934a module. This section is only of use to event loop authors who want to
1182 1968
1183I<rxvt-unicode> also cheats a bit by not providing blocking access to 1969I<rxvt-unicode> also cheats a bit by not providing blocking access to
1184condition variables: code blocking while waiting for a condition will 1970condition variables: code blocking while waiting for a condition will
1185C<die>. This still works with most modules/usages, and blocking calls must 1971C<die>. This still works with most modules/usages, and blocking calls must
1186not be done in an interactive application, so it makes sense. 1972not be done in an interactive application, so it makes sense.
1187
1188=head1 ENVIRONMENT VARIABLES
1189
1190The following environment variables are used by this module:
1191
1192=over 4
1193
1194=item C<PERL_ANYEVENT_VERBOSE>
1195
1196By default, AnyEvent will be completely silent except in fatal
1197conditions. You can set this environment variable to make AnyEvent more
1198talkative.
1199
1200When set to C<1> or higher, causes AnyEvent to warn about unexpected
1201conditions, such as not being able to load the event model specified by
1202C<PERL_ANYEVENT_MODEL>.
1203
1204When set to C<2> or higher, cause AnyEvent to report to STDERR which event
1205model it chooses.
1206
1207=item C<PERL_ANYEVENT_STRICT>
1208
1209AnyEvent does not do much argument checking by default, as thorough
1210argument checking is very costly. Setting this variable to a true value
1211will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
1212check the arguments passed to most method calls. If it finds any problems
1213it will croak.
1214
1215In other words, enables "strict" mode.
1216
1217Unlike C<use strict> it is definitely recommended ot keep it off in
1218production.
1219
1220=item C<PERL_ANYEVENT_MODEL>
1221
1222This can be used to specify the event model to be used by AnyEvent, before
1223auto detection and -probing kicks in. It must be a string consisting
1224entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
1225and the resulting module name is loaded and if the load was successful,
1226used as event model. If it fails to load AnyEvent will proceed with
1227auto detection and -probing.
1228
1229This functionality might change in future versions.
1230
1231For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
1232could start your program like this:
1233
1234 PERL_ANYEVENT_MODEL=Perl perl ...
1235
1236=item C<PERL_ANYEVENT_PROTOCOLS>
1237
1238Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
1239for IPv4 or IPv6. The default is unspecified (and might change, or be the result
1240of auto probing).
1241
1242Must be set to a comma-separated list of protocols or address families,
1243current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
1244used, and preference will be given to protocols mentioned earlier in the
1245list.
1246
1247This variable can effectively be used for denial-of-service attacks
1248against local programs (e.g. when setuid), although the impact is likely
1249small, as the program has to handle connection errors already-
1250
1251Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
1252but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
1253- only support IPv4, never try to resolve or contact IPv6
1254addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
1255IPv6, but prefer IPv6 over IPv4.
1256
1257=item C<PERL_ANYEVENT_EDNS0>
1258
1259Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
1260for DNS. This extension is generally useful to reduce DNS traffic, but
1261some (broken) firewalls drop such DNS packets, which is why it is off by
1262default.
1263
1264Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
1265EDNS0 in its DNS requests.
1266
1267=item C<PERL_ANYEVENT_MAX_FORKS>
1268
1269The maximum number of child processes that C<AnyEvent::Util::fork_call>
1270will create in parallel.
1271
1272=back
1273 1973
1274=head1 EXAMPLE PROGRAM 1974=head1 EXAMPLE PROGRAM
1275 1975
1276The following program uses an I/O watcher to read data from STDIN, a timer 1976The following program uses an I/O watcher to read data from STDIN, a timer
1277to display a message once per second, and a condition variable to quit the 1977to display a message once per second, and a condition variable to quit the
1290 warn "read: $input\n"; # output what has been read 1990 warn "read: $input\n"; # output what has been read
1291 $cv->send if $input =~ /^q/i; # quit program if /^q/i 1991 $cv->send if $input =~ /^q/i; # quit program if /^q/i
1292 }, 1992 },
1293 ); 1993 );
1294 1994
1295 my $time_watcher; # can only be used once
1296
1297 sub new_timer {
1298 $timer = AnyEvent->timer (after => 1, cb => sub { 1995 my $time_watcher = AnyEvent->timer (after => 1, interval => 1, cb => sub {
1299 warn "timeout\n"; # print 'timeout' about every second 1996 warn "timeout\n"; # print 'timeout' at most every second
1300 &new_timer; # and restart the time
1301 }); 1997 });
1302 }
1303
1304 new_timer; # create first timer
1305 1998
1306 $cv->recv; # wait until user enters /^q/i 1999 $cv->recv; # wait until user enters /^q/i
1307 2000
1308=head1 REAL-WORLD EXAMPLE 2001=head1 REAL-WORLD EXAMPLE
1309 2002
1440through AnyEvent. The benchmark creates a lot of timers (with a zero 2133through AnyEvent. The benchmark creates a lot of timers (with a zero
1441timeout) and I/O watchers (watching STDOUT, a pty, to become writable, 2134timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
1442which it is), lets them fire exactly once and destroys them again. 2135which it is), lets them fire exactly once and destroys them again.
1443 2136
1444Source code for this benchmark is found as F<eg/bench> in the AnyEvent 2137Source code for this benchmark is found as F<eg/bench> in the AnyEvent
1445distribution. 2138distribution. It uses the L<AE> interface, which makes a real difference
2139for the EV and Perl backends only.
1446 2140
1447=head3 Explanation of the columns 2141=head3 Explanation of the columns
1448 2142
1449I<watcher> is the number of event watchers created/destroyed. Since 2143I<watcher> is the number of event watchers created/destroyed. Since
1450different event models feature vastly different performances, each event 2144different event models feature vastly different performances, each event
1471watcher. 2165watcher.
1472 2166
1473=head3 Results 2167=head3 Results
1474 2168
1475 name watchers bytes create invoke destroy comment 2169 name watchers bytes create invoke destroy comment
1476 EV/EV 400000 244 0.56 0.46 0.31 EV native interface 2170 EV/EV 100000 223 0.47 0.43 0.27 EV native interface
1477 EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers 2171 EV/Any 100000 223 0.48 0.42 0.26 EV + AnyEvent watchers
1478 CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal 2172 Coro::EV/Any 100000 223 0.47 0.42 0.26 coroutines + Coro::Signal
1479 Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation 2173 Perl/Any 100000 431 2.70 0.74 0.92 pure perl implementation
1480 Event/Event 16000 516 31.88 31.30 0.85 Event native interface 2174 Event/Event 16000 516 31.16 31.84 0.82 Event native interface
1481 Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers 2175 Event/Any 16000 1203 42.61 34.79 1.80 Event + AnyEvent watchers
2176 IOAsync/Any 16000 1911 41.92 27.45 16.81 via IO::Async::Loop::IO_Poll
2177 IOAsync/Any 16000 1726 40.69 26.37 15.25 via IO::Async::Loop::Epoll
1482 Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour 2178 Glib/Any 16000 1118 89.00 12.57 51.17 quadratic behaviour
1483 Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers 2179 Tk/Any 2000 1346 20.96 10.75 8.00 SEGV with >> 2000 watchers
1484 POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event 2180 POE/Any 2000 6951 108.97 795.32 14.24 via POE::Loop::Event
1485 POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select 2181 POE/Any 2000 6648 94.79 774.40 575.51 via POE::Loop::Select
1486 2182
1487=head3 Discussion 2183=head3 Discussion
1488 2184
1489The benchmark does I<not> measure scalability of the event loop very 2185The benchmark does I<not> measure scalability of the event loop very
1490well. For example, a select-based event loop (such as the pure perl one) 2186well. For example, a select-based event loop (such as the pure perl one)
1502benchmark machine, handling an event takes roughly 1600 CPU cycles with 2198benchmark machine, handling an event takes roughly 1600 CPU cycles with
1503EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU 2199EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
1504cycles with POE. 2200cycles with POE.
1505 2201
1506C<EV> is the sole leader regarding speed and memory use, which are both 2202C<EV> is the sole leader regarding speed and memory use, which are both
1507maximal/minimal, respectively. Even when going through AnyEvent, it uses 2203maximal/minimal, respectively. When using the L<AE> API there is zero
2204overhead (when going through the AnyEvent API create is about 5-6 times
2205slower, with other times being equal, so still uses far less memory than
1508far less memory than any other event loop and is still faster than Event 2206any other event loop and is still faster than Event natively).
1509natively.
1510 2207
1511The pure perl implementation is hit in a few sweet spots (both the 2208The pure perl implementation is hit in a few sweet spots (both the
1512constant timeout and the use of a single fd hit optimisations in the perl 2209constant timeout and the use of a single fd hit optimisations in the perl
1513interpreter and the backend itself). Nevertheless this shows that it 2210interpreter and the backend itself). Nevertheless this shows that it
1514adds very little overhead in itself. Like any select-based backend its 2211adds very little overhead in itself. Like any select-based backend its
1515performance becomes really bad with lots of file descriptors (and few of 2212performance becomes really bad with lots of file descriptors (and few of
1516them active), of course, but this was not subject of this benchmark. 2213them active), of course, but this was not subject of this benchmark.
1517 2214
1518The C<Event> module has a relatively high setup and callback invocation 2215The C<Event> module has a relatively high setup and callback invocation
1519cost, but overall scores in on the third place. 2216cost, but overall scores in on the third place.
2217
2218C<IO::Async> performs admirably well, about on par with C<Event>, even
2219when using its pure perl backend.
1520 2220
1521C<Glib>'s memory usage is quite a bit higher, but it features a 2221C<Glib>'s memory usage is quite a bit higher, but it features a
1522faster callback invocation and overall ends up in the same class as 2222faster callback invocation and overall ends up in the same class as
1523C<Event>. However, Glib scales extremely badly, doubling the number of 2223C<Event>. However, Glib scales extremely badly, doubling the number of
1524watchers increases the processing time by more than a factor of four, 2224watchers increases the processing time by more than a factor of four,
1585In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100 2285In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
1586(1%) are active. This mirrors the activity of large servers with many 2286(1%) are active. This mirrors the activity of large servers with many
1587connections, most of which are idle at any one point in time. 2287connections, most of which are idle at any one point in time.
1588 2288
1589Source code for this benchmark is found as F<eg/bench2> in the AnyEvent 2289Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
1590distribution. 2290distribution. It uses the L<AE> interface, which makes a real difference
2291for the EV and Perl backends only.
1591 2292
1592=head3 Explanation of the columns 2293=head3 Explanation of the columns
1593 2294
1594I<sockets> is the number of sockets, and twice the number of "servers" (as 2295I<sockets> is the number of sockets, and twice the number of "servers" (as
1595each server has a read and write socket end). 2296each server has a read and write socket end).
1602it to another server. This includes deleting the old timeout and creating 2303it to another server. This includes deleting the old timeout and creating
1603a new one that moves the timeout into the future. 2304a new one that moves the timeout into the future.
1604 2305
1605=head3 Results 2306=head3 Results
1606 2307
1607 name sockets create request 2308 name sockets create request
1608 EV 20000 69.01 11.16 2309 EV 20000 62.66 7.99
1609 Perl 20000 73.32 35.87 2310 Perl 20000 68.32 32.64
1610 Event 20000 212.62 257.32 2311 IOAsync 20000 174.06 101.15 epoll
1611 Glib 20000 651.16 1896.30 2312 IOAsync 20000 174.67 610.84 poll
2313 Event 20000 202.69 242.91
2314 Glib 20000 557.01 1689.52
1612 POE 20000 349.67 12317.24 uses POE::Loop::Event 2315 POE 20000 341.54 12086.32 uses POE::Loop::Event
1613 2316
1614=head3 Discussion 2317=head3 Discussion
1615 2318
1616This benchmark I<does> measure scalability and overall performance of the 2319This benchmark I<does> measure scalability and overall performance of the
1617particular event loop. 2320particular event loop.
1619EV is again fastest. Since it is using epoll on my system, the setup time 2322EV is again fastest. Since it is using epoll on my system, the setup time
1620is relatively high, though. 2323is relatively high, though.
1621 2324
1622Perl surprisingly comes second. It is much faster than the C-based event 2325Perl surprisingly comes second. It is much faster than the C-based event
1623loops Event and Glib. 2326loops Event and Glib.
2327
2328IO::Async performs very well when using its epoll backend, and still quite
2329good compared to Glib when using its pure perl backend.
1624 2330
1625Event suffers from high setup time as well (look at its code and you will 2331Event suffers from high setup time as well (look at its code and you will
1626understand why). Callback invocation also has a high overhead compared to 2332understand why). Callback invocation also has a high overhead compared to
1627the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event 2333the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
1628uses select or poll in basically all documented configurations. 2334uses select or poll in basically all documented configurations.
1691=item * C-based event loops perform very well with small number of 2397=item * C-based event loops perform very well with small number of
1692watchers, as the management overhead dominates. 2398watchers, as the management overhead dominates.
1693 2399
1694=back 2400=back
1695 2401
2402=head2 THE IO::Lambda BENCHMARK
2403
2404Recently I was told about the benchmark in the IO::Lambda manpage, which
2405could be misinterpreted to make AnyEvent look bad. In fact, the benchmark
2406simply compares IO::Lambda with POE, and IO::Lambda looks better (which
2407shouldn't come as a surprise to anybody). As such, the benchmark is
2408fine, and mostly shows that the AnyEvent backend from IO::Lambda isn't
2409very optimal. But how would AnyEvent compare when used without the extra
2410baggage? To explore this, I wrote the equivalent benchmark for AnyEvent.
2411
2412The benchmark itself creates an echo-server, and then, for 500 times,
2413connects to the echo server, sends a line, waits for the reply, and then
2414creates the next connection. This is a rather bad benchmark, as it doesn't
2415test the efficiency of the framework or much non-blocking I/O, but it is a
2416benchmark nevertheless.
2417
2418 name runtime
2419 Lambda/select 0.330 sec
2420 + optimized 0.122 sec
2421 Lambda/AnyEvent 0.327 sec
2422 + optimized 0.138 sec
2423 Raw sockets/select 0.077 sec
2424 POE/select, components 0.662 sec
2425 POE/select, raw sockets 0.226 sec
2426 POE/select, optimized 0.404 sec
2427
2428 AnyEvent/select/nb 0.085 sec
2429 AnyEvent/EV/nb 0.068 sec
2430 +state machine 0.134 sec
2431
2432The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
2433benchmarks actually make blocking connects and use 100% blocking I/O,
2434defeating the purpose of an event-based solution. All of the newly
2435written AnyEvent benchmarks use 100% non-blocking connects (using
2436AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
2437resolver), so AnyEvent is at a disadvantage here, as non-blocking connects
2438generally require a lot more bookkeeping and event handling than blocking
2439connects (which involve a single syscall only).
2440
2441The last AnyEvent benchmark additionally uses L<AnyEvent::Handle>, which
2442offers similar expressive power as POE and IO::Lambda, using conventional
2443Perl syntax. This means that both the echo server and the client are 100%
2444non-blocking, further placing it at a disadvantage.
2445
2446As you can see, the AnyEvent + EV combination even beats the
2447hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
2448backend easily beats IO::Lambda and POE.
2449
2450And even the 100% non-blocking version written using the high-level (and
2451slow :) L<AnyEvent::Handle> abstraction beats both POE and IO::Lambda
2452higher level ("unoptimised") abstractions by a large margin, even though
2453it does all of DNS, tcp-connect and socket I/O in a non-blocking way.
2454
2455The two AnyEvent benchmarks programs can be found as F<eg/ae0.pl> and
2456F<eg/ae2.pl> in the AnyEvent distribution, the remaining benchmarks are
2457part of the IO::Lambda distribution and were used without any changes.
2458
2459
2460=head1 SIGNALS
2461
2462AnyEvent currently installs handlers for these signals:
2463
2464=over 4
2465
2466=item SIGCHLD
2467
2468A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
2469emulation for event loops that do not support them natively. Also, some
2470event loops install a similar handler.
2471
2472Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, then
2473AnyEvent will reset it to default, to avoid losing child exit statuses.
2474
2475=item SIGPIPE
2476
2477A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
2478when AnyEvent gets loaded.
2479
2480The rationale for this is that AnyEvent users usually do not really depend
2481on SIGPIPE delivery (which is purely an optimisation for shell use, or
2482badly-written programs), but C<SIGPIPE> can cause spurious and rare
2483program exits as a lot of people do not expect C<SIGPIPE> when writing to
2484some random socket.
2485
2486The rationale for installing a no-op handler as opposed to ignoring it is
2487that this way, the handler will be restored to defaults on exec.
2488
2489Feel free to install your own handler, or reset it to defaults.
2490
2491=back
2492
2493=cut
2494
2495undef $SIG{CHLD}
2496 if $SIG{CHLD} eq 'IGNORE';
2497
2498$SIG{PIPE} = sub { }
2499 unless defined $SIG{PIPE};
2500
2501=head1 RECOMMENDED/OPTIONAL MODULES
2502
2503One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
2504it's built-in modules) are required to use it.
2505
2506That does not mean that AnyEvent won't take advantage of some additional
2507modules if they are installed.
2508
2509This section explains which additional modules will be used, and how they
2510affect AnyEvent's operation.
2511
2512=over 4
2513
2514=item L<Async::Interrupt>
2515
2516This slightly arcane module is used to implement fast signal handling: To
2517my knowledge, there is no way to do completely race-free and quick
2518signal handling in pure perl. To ensure that signals still get
2519delivered, AnyEvent will start an interval timer to wake up perl (and
2520catch the signals) with some delay (default is 10 seconds, look for
2521C<$AnyEvent::MAX_SIGNAL_LATENCY>).
2522
2523If this module is available, then it will be used to implement signal
2524catching, which means that signals will not be delayed, and the event loop
2525will not be interrupted regularly, which is more efficient (and good for
2526battery life on laptops).
2527
2528This affects not just the pure-perl event loop, but also other event loops
2529that have no signal handling on their own (e.g. Glib, Tk, Qt).
2530
2531Some event loops (POE, Event, Event::Lib) offer signal watchers natively,
2532and either employ their own workarounds (POE) or use AnyEvent's workaround
2533(using C<$AnyEvent::MAX_SIGNAL_LATENCY>). Installing L<Async::Interrupt>
2534does nothing for those backends.
2535
2536=item L<EV>
2537
2538This module isn't really "optional", as it is simply one of the backend
2539event loops that AnyEvent can use. However, it is simply the best event
2540loop available in terms of features, speed and stability: It supports
2541the AnyEvent API optimally, implements all the watcher types in XS, does
2542automatic timer adjustments even when no monotonic clock is available,
2543can take avdantage of advanced kernel interfaces such as C<epoll> and
2544C<kqueue>, and is the fastest backend I<by far>. You can even embed
2545L<Glib>/L<Gtk2> in it (or vice versa, see L<EV::Glib> and L<Glib::EV>).
2546
2547If you only use backends that rely on another event loop (e.g. C<Tk>),
2548then this module will do nothing for you.
2549
2550=item L<Guard>
2551
2552The guard module, when used, will be used to implement
2553C<AnyEvent::Util::guard>. This speeds up guards considerably (and uses a
2554lot less memory), but otherwise doesn't affect guard operation much. It is
2555purely used for performance.
2556
2557=item L<JSON> and L<JSON::XS>
2558
2559One of these modules is required when you want to read or write JSON data
2560via L<AnyEvent::Handle>. L<JSON> is also written in pure-perl, but can take
2561advantage of the ultra-high-speed L<JSON::XS> module when it is installed.
2562
2563=item L<Net::SSLeay>
2564
2565Implementing TLS/SSL in Perl is certainly interesting, but not very
2566worthwhile: If this module is installed, then L<AnyEvent::Handle> (with
2567the help of L<AnyEvent::TLS>), gains the ability to do TLS/SSL.
2568
2569=item L<Time::HiRes>
2570
2571This module is part of perl since release 5.008. It will be used when the
2572chosen event library does not come with a timing source on it's own. The
2573pure-perl event loop (L<AnyEvent::Impl::Perl>) will additionally use it to
2574try to use a monotonic clock for timing stability.
2575
2576=back
2577
1696 2578
1697=head1 FORK 2579=head1 FORK
1698 2580
1699Most event libraries are not fork-safe. The ones who are usually are 2581Most event libraries are not fork-safe. The ones who are usually are
1700because they rely on inefficient but fork-safe C<select> or C<poll> 2582because they rely on inefficient but fork-safe C<select> or C<poll> calls
1701calls. Only L<EV> is fully fork-aware. 2583- higher performance APIs such as BSD's kqueue or the dreaded Linux epoll
2584are usually badly thought-out hacks that are incompatible with fork in
2585one way or another. Only L<EV> is fully fork-aware and ensures that you
2586continue event-processing in both parent and child (or both, if you know
2587what you are doing).
2588
2589This means that, in general, you cannot fork and do event processing in
2590the child if the event library was initialised before the fork (which
2591usually happens when the first AnyEvent watcher is created, or the library
2592is loaded).
1702 2593
1703If you have to fork, you must either do so I<before> creating your first 2594If you have to fork, you must either do so I<before> creating your first
1704watcher OR you must not use AnyEvent at all in the child. 2595watcher OR you must not use AnyEvent at all in the child OR you must do
2596something completely out of the scope of AnyEvent.
2597
2598The problem of doing event processing in the parent I<and> the child
2599is much more complicated: even for backends that I<are> fork-aware or
2600fork-safe, their behaviour is not usually what you want: fork clones all
2601watchers, that means all timers, I/O watchers etc. are active in both
2602parent and child, which is almost never what you want. USing C<exec>
2603to start worker children from some kind of manage rprocess is usually
2604preferred, because it is much easier and cleaner, at the expense of having
2605to have another binary.
1705 2606
1706 2607
1707=head1 SECURITY CONSIDERATIONS 2608=head1 SECURITY CONSIDERATIONS
1708 2609
1709AnyEvent can be forced to load any event model via 2610AnyEvent can be forced to load any event model via
1721 use AnyEvent; 2622 use AnyEvent;
1722 2623
1723Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can 2624Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
1724be used to probe what backend is used and gain other information (which is 2625be used to probe what backend is used and gain other information (which is
1725probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and 2626probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
1726$ENV{PERL_ANYEGENT_STRICT}. 2627$ENV{PERL_ANYEVENT_STRICT}.
2628
2629Note that AnyEvent will remove I<all> environment variables starting with
2630C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
2631enabled.
1727 2632
1728 2633
1729=head1 BUGS 2634=head1 BUGS
1730 2635
1731Perl 5.8 has numerous memleaks that sometimes hit this module and are hard 2636Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
1732to work around. If you suffer from memleaks, first upgrade to Perl 5.10 2637to work around. If you suffer from memleaks, first upgrade to Perl 5.10
1733and check wether the leaks still show up. (Perl 5.10.0 has other annoying 2638and check wether the leaks still show up. (Perl 5.10.0 has other annoying
1734mamleaks, such as leaking on C<map> and C<grep> but it is usually not as 2639memleaks, such as leaking on C<map> and C<grep> but it is usually not as
1735pronounced). 2640pronounced).
1736 2641
1737 2642
1738=head1 SEE ALSO 2643=head1 SEE ALSO
1739 2644
1743L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>. 2648L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
1744 2649
1745Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>, 2650Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
1746L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>, 2651L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
1747L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>, 2652L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
1748L<AnyEvent::Impl::POE>. 2653L<AnyEvent::Impl::POE>, L<AnyEvent::Impl::IOAsync>, L<Anyevent::Impl::Irssi>.
1749 2654
1750Non-blocking file handles, sockets, TCP clients and 2655Non-blocking file handles, sockets, TCP clients and
1751servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>. 2656servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>, L<AnyEvent::TLS>.
1752 2657
1753Asynchronous DNS: L<AnyEvent::DNS>. 2658Asynchronous DNS: L<AnyEvent::DNS>.
1754 2659
1755Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>, 2660Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>,
2661L<Coro::Event>,
1756 2662
1757Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>, L<AnyEvent::DNS>. 2663Nontrivial usage examples: L<AnyEvent::GPSD>, L<AnyEvent::XMPP>,
2664L<AnyEvent::HTTP>.
1758 2665
1759 2666
1760=head1 AUTHOR 2667=head1 AUTHOR
1761 2668
1762 Marc Lehmann <schmorp@schmorp.de> 2669 Marc Lehmann <schmorp@schmorp.de>

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines