ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent/lib/AnyEvent.pm

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.1 by root, Wed Apr 27 01:26:44 2005 UTC vs.
Revision 1.421 by root, Fri Sep 5 22:24:12 2014 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	AnyEvent - ???	3	AnyEvent - the DBI of event loop programming
		4
		5	EV, Event, Glib, Tk, UV, Perl, Event::Lib, Irssi, rxvt-unicode, IO::Async,
		6	Qt, FLTK and POE are various supported event loops/environments.
4		7
5	=head1 SYNOPSIS	8	=head1 SYNOPSIS
6		9
		10	use AnyEvent;
		11
		12	# if you prefer function calls, look at the AE manpage for
		13	# an alternative API.
		14
		15	# file handle or descriptor readable
		16	my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ... });
		17
		18	# one-shot or repeating timers
		19	my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
		20	my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...);
		21
		22	print AnyEvent->now; # prints current event loop time
		23	print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
		24
		25	# POSIX signal
		26	my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
		27
		28	# child process exit
		29	my $w = AnyEvent->child (pid => $pid, cb => sub {
		30	my ($pid, $status) = @_;
		31	...
		32	});
		33
		34	# called when event loop idle (if applicable)
		35	my $w = AnyEvent->idle (cb => sub { ... });
		36
		37	my $w = AnyEvent->condvar; # stores whether a condition was flagged
		38	$w->send; # wake up current and all future recv's
		39	$w->recv; # enters "main loop" till $condvar gets ->send
		40	# use a condvar in callback mode:
		41	$w->cb (sub { $_[0]->recv });
		42
		43	=head1 INTRODUCTION/TUTORIAL
		44
		45	This manpage is mainly a reference manual. If you are interested
		46	in a tutorial or some gentle introduction, have a look at the
		47	L<AnyEvent::Intro> manpage.
		48
		49	=head1 SUPPORT
		50
		51	An FAQ document is available as L<AnyEvent::FAQ>.
		52
		53	There also is a mailinglist for discussing all things AnyEvent, and an IRC
		54	channel, too.
		55
		56	See the AnyEvent project page at the B<Schmorpforge Ta-Sa Software
		57	Repository>, at L<http://anyevent.schmorp.de>, for more info.
		58
		59	=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
		60
		61	Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
		62	nowadays. So what is different about AnyEvent?
		63
		64	Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
		65	policy> and AnyEvent is I<small and efficient>.
		66
		67	First and foremost, I<AnyEvent is not an event model> itself, it only
		68	interfaces to whatever event model the main program happens to use, in a
		69	pragmatic way. For event models and certain classes of immortals alike,
		70	the statement "there can only be one" is a bitter reality: In general,
		71	only one event loop can be active at the same time in a process. AnyEvent
		72	cannot change this, but it can hide the differences between those event
		73	loops.
		74
		75	The goal of AnyEvent is to offer module authors the ability to do event
		76	programming (waiting for I/O or timer events) without subscribing to a
		77	religion, a way of living, and most importantly: without forcing your
		78	module users into the same thing by forcing them to use the same event
		79	model you use.
		80
		81	For modules like POE or IO::Async (which is a total misnomer as it is
		82	actually doing all I/O I<synchronously>...), using them in your module is
		83	like joining a cult: After you join, you are dependent on them and you
		84	cannot use anything else, as they are simply incompatible to everything
		85	that isn't them. What's worse, all the potential users of your
		86	module are I<also> forced to use the same event loop you use.
		87
		88	AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
		89	fine. AnyEvent + Tk works fine etc. etc. but none of these work together
		90	with the rest: POE + EV? No go. Tk + Event? No go. Again: if your module
		91	uses one of those, every user of your module has to use it, too. But if
		92	your module uses AnyEvent, it works transparently with all event models it
		93	supports (including stuff like IO::Async, as long as those use one of the
		94	supported event loops. It is easy to add new event loops to AnyEvent, too,
		95	so it is future-proof).
		96
		97	In addition to being free of having to use I<the one and only true event
		98	model>, AnyEvent also is free of bloat and policy: with POE or similar
		99	modules, you get an enormous amount of code and strict rules you have to
		100	follow. AnyEvent, on the other hand, is lean and to the point, by only
		101	offering the functionality that is necessary, in as thin as a wrapper as
		102	technically possible.
		103
		104	Of course, AnyEvent comes with a big (and fully optional!) toolbox
		105	of useful functionality, such as an asynchronous DNS resolver, 100%
		106	non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
		107	such as Windows) and lots of real-world knowledge and workarounds for
		108	platform bugs and differences.
		109
		110	Now, if you I<do want> lots of policy (this can arguably be somewhat
		111	useful) and you want to force your users to use the one and only event
		112	model, you should I<not> use this module.
		113
7	=head1 DESCRIPTION	114	=head1 DESCRIPTION
8		115
		116	L<AnyEvent> provides a uniform interface to various event loops. This
		117	allows module authors to use event loop functionality without forcing
		118	module users to use a specific event loop implementation (since more
		119	than one event loop cannot coexist peacefully).
		120
		121	The interface itself is vaguely similar, but not identical to the L<Event>
		122	module.
		123
		124	During the first call of any watcher-creation method, the module tries
		125	to detect the currently loaded event loop by probing whether one of the
		126	following modules is already loaded: L<EV>, L<AnyEvent::Loop>,
		127	L<Event>, L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>. The first one
		128	found is used. If none are detected, the module tries to load the first
		129	four modules in the order given; but note that if L<EV> is not
		130	available, the pure-perl L<AnyEvent::Loop> should always work, so
		131	the other two are not normally tried.
		132
		133	Because AnyEvent first checks for modules that are already loaded, loading
		134	an event model explicitly before first using AnyEvent will likely make
		135	that model the default. For example:
		136
		137	use Tk;
		138	use AnyEvent;
		139
		140	# .. AnyEvent will likely default to Tk
		141
		142	The I<likely> means that, if any module loads another event model and
		143	starts using it, all bets are off - this case should be very rare though,
		144	as very few modules hardcode event loops without announcing this very
		145	loudly.
		146
		147	The pure-perl implementation of AnyEvent is called C<AnyEvent::Loop>. Like
		148	other event modules you can load it explicitly and enjoy the high
		149	availability of that event loop :)
		150
		151	=head1 WATCHERS
		152
		153	AnyEvent has the central concept of a I<watcher>, which is an object that
		154	stores relevant data for each kind of event you are waiting for, such as
		155	the callback to call, the file handle to watch, etc.
		156
		157	These watchers are normal Perl objects with normal Perl lifetime. After
		158	creating a watcher it will immediately "watch" for events and invoke the
		159	callback when the event occurs (of course, only when the event model
		160	is in control).
		161
		162	Note that B<callbacks must not permanently change global variables>
		163	potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
		164	callbacks must not C<die> >>. The former is good programming practice in
		165	Perl and the latter stems from the fact that exception handling differs
		166	widely between event loops.
		167
		168	To disable a watcher you have to destroy it (e.g. by setting the
		169	variable you store it in to C<undef> or otherwise deleting all references
		170	to it).
		171
		172	All watchers are created by calling a method on the C<AnyEvent> class.
		173
		174	Many watchers either are used with "recursion" (repeating timers for
		175	example), or need to refer to their watcher object in other ways.
		176
		177	One way to achieve that is this pattern:
		178
		179	my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
		180	# you can use $w here, for example to undef it
		181	undef $w;
		182	});
		183
		184	Note that C<my $w; $w => combination. This is necessary because in Perl,
		185	my variables are only visible after the statement in which they are
		186	declared.
		187
		188	=head2 I/O WATCHERS
		189
		190	$w = AnyEvent->io (
		191	fh => <filehandle_or_fileno>,
		192	poll => <"r" or "w">,
		193	cb => <callback>,
		194	);
		195
		196	You can create an I/O watcher by calling the C<< AnyEvent->io >> method
		197	with the following mandatory key-value pairs as arguments:
		198
		199	C<fh> is the Perl I<file handle> (or a naked file descriptor) to watch
		200	for events (AnyEvent might or might not keep a reference to this file
		201	handle). Note that only file handles pointing to things for which
		202	non-blocking operation makes sense are allowed. This includes sockets,
		203	most character devices, pipes, fifos and so on, but not for example files
		204	or block devices.
		205
		206	C<poll> must be a string that is either C<r> or C<w>, which creates a
		207	watcher waiting for "r"eadable or "w"ritable events, respectively.
		208
		209	C<cb> is the callback to invoke each time the file handle becomes ready.
		210
		211	Although the callback might get passed parameters, their value and
		212	presence is undefined and you cannot rely on them. Portable AnyEvent
		213	callbacks cannot use arguments passed to I/O watcher callbacks.
		214
		215	The I/O watcher might use the underlying file descriptor or a copy of it.
		216	You must not close a file handle as long as any watcher is active on the
		217	underlying file descriptor.
		218
		219	Some event loops issue spurious readiness notifications, so you should
		220	always use non-blocking calls when reading/writing from/to your file
		221	handles.
		222
		223	Example: wait for readability of STDIN, then read a line and disable the
		224	watcher.
		225
		226	my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
		227	chomp (my $input = <STDIN>);
		228	warn "read: $input\n";
		229	undef $w;
		230	});
		231
		232	=head2 TIME WATCHERS
		233
		234	$w = AnyEvent->timer (after => <seconds>, cb => <callback>);
		235
		236	$w = AnyEvent->timer (
		237	after => <fractional_seconds>,
		238	interval => <fractional_seconds>,
		239	cb => <callback>,
		240	);
		241
		242	You can create a time watcher by calling the C<< AnyEvent->timer >>
		243	method with the following mandatory arguments:
		244
		245	C<after> specifies after how many seconds (fractional values are
		246	supported) the callback should be invoked. C<cb> is the callback to invoke
		247	in that case.
		248
		249	Although the callback might get passed parameters, their value and
		250	presence is undefined and you cannot rely on them. Portable AnyEvent
		251	callbacks cannot use arguments passed to time watcher callbacks.
		252
		253	The callback will normally be invoked only once. If you specify another
		254	parameter, C<interval>, as a strictly positive number (> 0), then the
		255	callback will be invoked regularly at that interval (in fractional
		256	seconds) after the first invocation. If C<interval> is specified with a
		257	false value, then it is treated as if it were not specified at all.
		258
		259	The callback will be rescheduled before invoking the callback, but no
		260	attempt is made to avoid timer drift in most backends, so the interval is
		261	only approximate.
		262
		263	Example: fire an event after 7.7 seconds.
		264
		265	my $w = AnyEvent->timer (after => 7.7, cb => sub {
		266	warn "timeout\n";
		267	});
		268
		269	# to cancel the timer:
		270	undef $w;
		271
		272	Example 2: fire an event after 0.5 seconds, then roughly every second.
		273
		274	my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
		275	warn "timeout\n";
		276	});
		277
		278	=head3 TIMING ISSUES
		279
		280	There are two ways to handle timers: based on real time (relative, "fire
		281	in 10 seconds") and based on wallclock time (absolute, "fire at 12
		282	o'clock").
		283
		284	While most event loops expect timers to specified in a relative way, they
		285	use absolute time internally. This makes a difference when your clock
		286	"jumps", for example, when ntp decides to set your clock backwards from
		287	the wrong date of 2014-01-01 to 2008-01-01, a watcher that is supposed to
		288	fire "after a second" might actually take six years to finally fire.
		289
		290	AnyEvent cannot compensate for this. The only event loop that is conscious
		291	of these issues is L<EV>, which offers both relative (ev_timer, based
		292	on true relative time) and absolute (ev_periodic, based on wallclock time)
		293	timers.
		294
		295	AnyEvent always prefers relative timers, if available, matching the
		296	AnyEvent API.
		297
		298	AnyEvent has two additional methods that return the "current time":
		299
9	=over 4	300	=over 4
10		301
		302	=item AnyEvent->time
		303
		304	This returns the "current wallclock time" as a fractional number of
		305	seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
		306	return, and the result is guaranteed to be compatible with those).
		307
		308	It progresses independently of any event loop processing, i.e. each call
		309	will check the system clock, which usually gets updated frequently.
		310
		311	=item AnyEvent->now
		312
		313	This also returns the "current wallclock time", but unlike C<time>, above,
		314	this value might change only once per event loop iteration, depending on
		315	the event loop (most return the same time as C<time>, above). This is the
		316	time that AnyEvent's timers get scheduled against.
		317
		318	I<In almost all cases (in all cases if you don't care), this is the
		319	function to call when you want to know the current time.>
		320
		321	This function is also often faster then C<< AnyEvent->time >>, and
		322	thus the preferred method if you want some timestamp (for example,
		323	L<AnyEvent::Handle> uses this to update its activity timeouts).
		324
		325	The rest of this section is only of relevance if you try to be very exact
		326	with your timing; you can skip it without a bad conscience.
		327
		328	For a practical example of when these times differ, consider L<Event::Lib>
		329	and L<EV> and the following set-up:
		330
		331	The event loop is running and has just invoked one of your callbacks at
		332	time=500 (assume no other callbacks delay processing). In your callback,
		333	you wait a second by executing C<sleep 1> (blocking the process for a
		334	second) and then (at time=501) you create a relative timer that fires
		335	after three seconds.
		336
		337	With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
		338	both return C<501>, because that is the current time, and the timer will
		339	be scheduled to fire at time=504 (C<501> + C<3>).
		340
		341	With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
		342	time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
		343	last event processing phase started. With L<EV>, your timer gets scheduled
		344	to run at time=503 (C<500> + C<3>).
		345
		346	In one sense, L<Event::Lib> is more exact, as it uses the current time
		347	regardless of any delays introduced by event processing. However, most
		348	callbacks do not expect large delays in processing, so this causes a
		349	higher drift (and a lot more system calls to get the current time).
		350
		351	In another sense, L<EV> is more exact, as your timer will be scheduled at
		352	the same time, regardless of how long event processing actually took.
		353
		354	In either case, if you care (and in most cases, you don't), then you
		355	can get whatever behaviour you want with any event loop, by taking the
		356	difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
		357	account.
		358
		359	=item AnyEvent->now_update
		360
		361	Some event loops (such as L<EV> or L<AnyEvent::Loop>) cache the current
		362	time for each loop iteration (see the discussion of L<< AnyEvent->now >>,
		363	above).
		364
		365	When a callback runs for a long time (or when the process sleeps), then
		366	this "current" time will differ substantially from the real time, which
		367	might affect timers and time-outs.
		368
		369	When this is the case, you can call this method, which will update the
		370	event loop's idea of "current time".
		371
		372	A typical example would be a script in a web server (e.g. C<mod_perl>) -
		373	when mod_perl executes the script, then the event loop will have the wrong
		374	idea about the "current time" (being potentially far in the past, when the
		375	script ran the last time). In that case you should arrange a call to C<<
		376	AnyEvent->now_update >> each time the web server process wakes up again
		377	(e.g. at the start of your script, or in a handler).
		378
		379	Note that updating the time I<might> cause some events to be handled.
		380
		381	=back
		382
		383	=head2 SIGNAL WATCHERS
		384
		385	$w = AnyEvent->signal (signal => <uppercase_signal_name>, cb => <callback>);
		386
		387	You can watch for signals using a signal watcher, C<signal> is the signal
		388	I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
		389	callback to be invoked whenever a signal occurs.
		390
		391	Although the callback might get passed parameters, their value and
		392	presence is undefined and you cannot rely on them. Portable AnyEvent
		393	callbacks cannot use arguments passed to signal watcher callbacks.
		394
		395	Multiple signal occurrences can be clumped together into one callback
		396	invocation, and callback invocation will be synchronous. Synchronous means
		397	that it might take a while until the signal gets handled by the process,
		398	but it is guaranteed not to interrupt any other callbacks.
		399
		400	The main advantage of using these watchers is that you can share a signal
		401	between multiple watchers, and AnyEvent will ensure that signals will not
		402	interrupt your program at bad times.
		403
		404	This watcher might use C<%SIG> (depending on the event loop used),
		405	so programs overwriting those signals directly will likely not work
		406	correctly.
		407
		408	Example: exit on SIGINT
		409
		410	my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
		411
		412	=head3 Restart Behaviour
		413
		414	While restart behaviour is up to the event loop implementation, most will
		415	not restart syscalls (that includes L<Async::Interrupt> and AnyEvent's
		416	pure perl implementation).
		417
		418	=head3 Safe/Unsafe Signals
		419
		420	Perl signals can be either "safe" (synchronous to opcode handling)
		421	or "unsafe" (asynchronous) - the former might delay signal delivery
		422	indefinitely, the latter might corrupt your memory.
		423
		424	AnyEvent signal handlers are, in addition, synchronous to the event loop,
		425	i.e. they will not interrupt your running perl program but will only be
		426	called as part of the normal event handling (just like timer, I/O etc.
		427	callbacks, too).
		428
		429	=head3 Signal Races, Delays and Workarounds
		430
		431	Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support
		432	attaching callbacks to signals in a generic way, which is a pity,
		433	as you cannot do race-free signal handling in perl, requiring
		434	C libraries for this. AnyEvent will try to do its best, which
		435	means in some cases, signals will be delayed. The maximum time
		436	a signal might be delayed is 10 seconds by default, but can
		437	be overriden via C<$ENV{PERL_ANYEVENT_MAX_SIGNAL_LATENCY}> or
		438	C<$AnyEvent::MAX_SIGNAL_LATENCY> - see the L<ENVIRONMENT VARIABLES>
		439	section for details.
		440
		441	All these problems can be avoided by installing the optional
		442	L<Async::Interrupt> module, which works with most event loops. It will not
		443	work with inherently broken event loops such as L<Event> or L<Event::Lib>
		444	(and not with L<POE> currently). For those, you just have to suffer the
		445	delays.
		446
		447	=head2 CHILD PROCESS WATCHERS
		448
		449	$w = AnyEvent->child (pid => <process id>, cb => <callback>);
		450
		451	You can also watch for a child process exit and catch its exit status.
		452
		453	The child process is specified by the C<pid> argument (on some backends,
		454	using C<0> watches for any child process exit, on others this will
		455	croak). The watcher will be triggered only when the child process has
		456	finished and an exit status is available, not on any trace events
		457	(stopped/continued).
		458
		459	The callback will be called with the pid and exit status (as returned by
		460	waitpid), so unlike other watcher types, you I<can> rely on child watcher
		461	callback arguments.
		462
		463	This watcher type works by installing a signal handler for C<SIGCHLD>,
		464	and since it cannot be shared, nothing else should use SIGCHLD or reap
		465	random child processes (waiting for specific child processes, e.g. inside
		466	C<system>, is just fine).
		467
		468	There is a slight catch to child watchers, however: you usually start them
		469	I<after> the child process was created, and this means the process could
		470	have exited already (and no SIGCHLD will be sent anymore).
		471
		472	Not all event models handle this correctly (neither POE nor IO::Async do,
		473	see their AnyEvent::Impl manpages for details), but even for event models
		474	that I<do> handle this correctly, they usually need to be loaded before
		475	the process exits (i.e. before you fork in the first place). AnyEvent's
		476	pure perl event loop handles all cases correctly regardless of when you
		477	start the watcher.
		478
		479	This means you cannot create a child watcher as the very first
		480	thing in an AnyEvent program, you I<have> to create at least one
		481	watcher before you C<fork> the child (alternatively, you can call
		482	C<AnyEvent::detect>).
		483
		484	As most event loops do not support waiting for child events, they will be
		485	emulated by AnyEvent in most cases, in which case the latency and race
		486	problems mentioned in the description of signal watchers apply.
		487
		488	Example: fork a process and wait for it
		489
		490	my $done = AnyEvent->condvar;
		491
		492	# this forks and immediately calls exit in the child. this
		493	# normally has all sorts of bad consequences for your parent,
		494	# so take this as an example only. always fork and exec,
		495	# or call POSIX::_exit, in real code.
		496	my $pid = fork or exit 5;
		497
		498	my $w = AnyEvent->child (
		499	pid => $pid,
		500	cb => sub {
		501	my ($pid, $status) = @_;
		502	warn "pid $pid exited with status $status";
		503	$done->send;
		504	},
		505	);
		506
		507	# do something else, then wait for process exit
		508	$done->recv;
		509
		510	=head2 IDLE WATCHERS
		511
		512	$w = AnyEvent->idle (cb => <callback>);
		513
		514	This will repeatedly invoke the callback after the process becomes idle,
		515	until either the watcher is destroyed or new events have been detected.
		516
		517	Idle watchers are useful when there is a need to do something, but it
		518	is not so important (or wise) to do it instantly. The callback will be
		519	invoked only when there is "nothing better to do", which is usually
		520	defined as "all outstanding events have been handled and no new events
		521	have been detected". That means that idle watchers ideally get invoked
		522	when the event loop has just polled for new events but none have been
		523	detected. Instead of blocking to wait for more events, the idle watchers
		524	will be invoked.
		525
		526	Unfortunately, most event loops do not really support idle watchers (only
		527	EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
		528	will simply call the callback "from time to time".
		529
		530	Example: read lines from STDIN, but only process them when the
		531	program is otherwise idle:
		532
		533	my @lines; # read data
		534	my $idle_w;
		535	my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
		536	push @lines, scalar <STDIN>;
		537
		538	# start an idle watcher, if not already done
		539	$idle_w ||= AnyEvent->idle (cb => sub {
		540	# handle only one line, when there are lines left
		541	if (my $line = shift @lines) {
		542	print "handled when idle: $line";
		543	} else {
		544	# otherwise disable the idle watcher again
		545	undef $idle_w;
		546	}
		547	});
		548	});
		549
		550	=head2 CONDITION VARIABLES
		551
		552	$cv = AnyEvent->condvar;
		553
		554	$cv->send (<list>);
		555	my @res = $cv->recv;
		556
		557	If you are familiar with some event loops you will know that all of them
		558	require you to run some blocking "loop", "run" or similar function that
		559	will actively watch for new events and call your callbacks.
		560
		561	AnyEvent is slightly different: it expects somebody else to run the event
		562	loop and will only block when necessary (usually when told by the user).
		563
		564	The tool to do that is called a "condition variable", so called because
		565	they represent a condition that must become true.
		566
		567	Now is probably a good time to look at the examples further below.
		568
		569	Condition variables can be created by calling the C<< AnyEvent->condvar
		570	>> method, usually without arguments. The only argument pair allowed is
		571	C<cb>, which specifies a callback to be called when the condition variable
		572	becomes true, with the condition variable as the first argument (but not
		573	the results).
		574
		575	After creation, the condition variable is "false" until it becomes "true"
		576	by calling the C<send> method (or calling the condition variable as if it
		577	were a callback, read about the caveats in the description for the C<<
		578	->send >> method).
		579
		580	Since condition variables are the most complex part of the AnyEvent API, here are
		581	some different mental models of what they are - pick the ones you can connect to:
		582
		583	=over 4
		584
		585	=item * Condition variables are like callbacks - you can call them (and pass them instead
		586	of callbacks). Unlike callbacks however, you can also wait for them to be called.
		587
		588	=item * Condition variables are signals - one side can emit or send them,
		589	the other side can wait for them, or install a handler that is called when
		590	the signal fires.
		591
		592	=item * Condition variables are like "Merge Points" - points in your program
		593	where you merge multiple independent results/control flows into one.
		594
		595	=item * Condition variables represent a transaction - functions that start
		596	some kind of transaction can return them, leaving the caller the choice
		597	between waiting in a blocking fashion, or setting a callback.
		598
		599	=item * Condition variables represent future values, or promises to deliver
		600	some result, long before the result is available.
		601
		602	=back
		603
		604	Condition variables are very useful to signal that something has finished,
		605	for example, if you write a module that does asynchronous http requests,
		606	then a condition variable would be the ideal candidate to signal the
		607	availability of results. The user can either act when the callback is
		608	called or can synchronously C<< ->recv >> for the results.
		609
		610	You can also use them to simulate traditional event loops - for example,
		611	you can block your main program until an event occurs - for example, you
		612	could C<< ->recv >> in your main program until the user clicks the Quit
		613	button of your app, which would C<< ->send >> the "quit" event.
		614
		615	Note that condition variables recurse into the event loop - if you have
		616	two pieces of code that call C<< ->recv >> in a round-robin fashion, you
		617	lose. Therefore, condition variables are good to export to your caller, but
		618	you should avoid making a blocking wait yourself, at least in callbacks,
		619	as this asks for trouble.
		620
		621	Condition variables are represented by hash refs in perl, and the keys
		622	used by AnyEvent itself are all named C<_ae_XXX> to make subclassing
		623	easy (it is often useful to build your own transaction class on top of
		624	AnyEvent). To subclass, use C<AnyEvent::CondVar> as base class and call
		625	its C<new> method in your own C<new> method.
		626
		627	There are two "sides" to a condition variable - the "producer side" which
		628	eventually calls C<< -> send >>, and the "consumer side", which waits
		629	for the send to occur.
		630
		631	Example: wait for a timer.
		632
		633	# condition: "wait till the timer is fired"
		634	my $timer_fired = AnyEvent->condvar;
		635
		636	# create the timer - we could wait for, say
		637	# a handle becomign ready, or even an
		638	# AnyEvent::HTTP request to finish, but
		639	# in this case, we simply use a timer:
		640	my $w = AnyEvent->timer (
		641	after => 1,
		642	cb => sub { $timer_fired->send },
		643	);
		644
		645	# this "blocks" (while handling events) till the callback
		646	# calls ->send
		647	$timer_fired->recv;
		648
		649	Example: wait for a timer, but take advantage of the fact that condition
		650	variables are also callable directly.
		651
		652	my $done = AnyEvent->condvar;
		653	my $delay = AnyEvent->timer (after => 5, cb => $done);
		654	$done->recv;
		655
		656	Example: Imagine an API that returns a condvar and doesn't support
		657	callbacks. This is how you make a synchronous call, for example from
		658	the main program:
		659
		660	use AnyEvent::CouchDB;
		661
		662	...
		663
		664	my @info = $couchdb->info->recv;
		665
		666	And this is how you would just set a callback to be called whenever the
		667	results are available:
		668
		669	$couchdb->info->cb (sub {
		670	my @info = $_[0]->recv;
		671	});
		672
		673	=head3 METHODS FOR PRODUCERS
		674
		675	These methods should only be used by the producing side, i.e. the
		676	code/module that eventually sends the signal. Note that it is also
		677	the producer side which creates the condvar in most cases, but it isn't
		678	uncommon for the consumer to create it as well.
		679
		680	=over 4
		681
		682	=item $cv->send (...)
		683
		684	Flag the condition as ready - a running C<< ->recv >> and all further
		685	calls to C<recv> will (eventually) return after this method has been
		686	called. If nobody is waiting the send will be remembered.
		687
		688	If a callback has been set on the condition variable, it is called
		689	immediately from within send.
		690
		691	Any arguments passed to the C<send> call will be returned by all
		692	future C<< ->recv >> calls.
		693
		694	Condition variables are overloaded so one can call them directly (as if
		695	they were a code reference). Calling them directly is the same as calling
		696	C<send>.
		697
		698	=item $cv->croak ($error)
		699
		700	Similar to send, but causes all calls to C<< ->recv >> to invoke
		701	C<Carp::croak> with the given error message/object/scalar.
		702
		703	This can be used to signal any errors to the condition variable
		704	user/consumer. Doing it this way instead of calling C<croak> directly
		705	delays the error detection, but has the overwhelming advantage that it
		706	diagnoses the error at the place where the result is expected, and not
		707	deep in some event callback with no connection to the actual code causing
		708	the problem.
		709
		710	=item $cv->begin ([group callback])
		711
		712	=item $cv->end
		713
		714	These two methods can be used to combine many transactions/events into
		715	one. For example, a function that pings many hosts in parallel might want
		716	to use a condition variable for the whole process.
		717
		718	Every call to C<< ->begin >> will increment a counter, and every call to
		719	C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
		720	>>, the (last) callback passed to C<begin> will be executed, passing the
		721	condvar as first argument. That callback is I<supposed> to call C<< ->send
		722	>>, but that is not required. If no group callback was set, C<send> will
		723	be called without any arguments.
		724
		725	You can think of C<< $cv->send >> giving you an OR condition (one call
		726	sends), while C<< $cv->begin >> and C<< $cv->end >> giving you an AND
		727	condition (all C<begin> calls must be C<end>'ed before the condvar sends).
		728
		729	Let's start with a simple example: you have two I/O watchers (for example,
		730	STDOUT and STDERR for a program), and you want to wait for both streams to
		731	close before activating a condvar:
		732
		733	my $cv = AnyEvent->condvar;
		734
		735	$cv->begin; # first watcher
		736	my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
		737	defined sysread $fh1, my $buf, 4096
		738	or $cv->end;
		739	});
		740
		741	$cv->begin; # second watcher
		742	my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
		743	defined sysread $fh2, my $buf, 4096
		744	or $cv->end;
		745	});
		746
		747	$cv->recv;
		748
		749	This works because for every event source (EOF on file handle), there is
		750	one call to C<begin>, so the condvar waits for all calls to C<end> before
		751	sending.
		752
		753	The ping example mentioned above is slightly more complicated, as the
		754	there are results to be passed back, and the number of tasks that are
		755	begun can potentially be zero:
		756
		757	my $cv = AnyEvent->condvar;
		758
		759	my %result;
		760	$cv->begin (sub { shift->send (\%result) });
		761
		762	for my $host (@list_of_hosts) {
		763	$cv->begin;
		764	ping_host_then_call_callback $host, sub {
		765	$result{$host} = ...;
		766	$cv->end;
		767	};
		768	}
		769
		770	$cv->end;
		771
		772	...
		773
		774	my $results = $cv->recv;
		775
		776	This code fragment supposedly pings a number of hosts and calls
		777	C<send> after results for all then have have been gathered - in any
		778	order. To achieve this, the code issues a call to C<begin> when it starts
		779	each ping request and calls C<end> when it has received some result for
		780	it. Since C<begin> and C<end> only maintain a counter, the order in which
		781	results arrive is not relevant.
		782
		783	There is an additional bracketing call to C<begin> and C<end> outside the
		784	loop, which serves two important purposes: first, it sets the callback
		785	to be called once the counter reaches C<0>, and second, it ensures that
		786	C<send> is called even when C<no> hosts are being pinged (the loop
		787	doesn't execute once).
		788
		789	This is the general pattern when you "fan out" into multiple (but
		790	potentially zero) subrequests: use an outer C<begin>/C<end> pair to set
		791	the callback and ensure C<end> is called at least once, and then, for each
		792	subrequest you start, call C<begin> and for each subrequest you finish,
		793	call C<end>.
		794
		795	=back
		796
		797	=head3 METHODS FOR CONSUMERS
		798
		799	These methods should only be used by the consuming side, i.e. the
		800	code awaits the condition.
		801
		802	=over 4
		803
		804	=item $cv->recv
		805
		806	Wait (blocking if necessary) until the C<< ->send >> or C<< ->croak
		807	>> methods have been called on C<$cv>, while servicing other watchers
		808	normally.
		809
		810	You can only wait once on a condition - additional calls are valid but
		811	will return immediately.
		812
		813	If an error condition has been set by calling C<< ->croak >>, then this
		814	function will call C<croak>.
		815
		816	In list context, all parameters passed to C<send> will be returned,
		817	in scalar context only the first one will be returned.
		818
		819	Note that doing a blocking wait in a callback is not supported by any
		820	event loop, that is, recursive invocation of a blocking C<< ->recv >> is
		821	not allowed and the C<recv> call will C<croak> if such a condition is
		822	detected. This requirement can be dropped by relying on L<Coro::AnyEvent>
		823	, which allows you to do a blocking C<< ->recv >> from any thread
		824	that doesn't run the event loop itself. L<Coro::AnyEvent> is loaded
		825	automatically when L<Coro> is used with L<AnyEvent>, so code does not need
		826	to do anything special to take advantage of that: any code that would
		827	normally block your program because it calls C<recv>, be executed in an
		828	C<async> thread instead without blocking other threads.
		829
		830	Not all event models support a blocking wait - some die in that case
		831	(programs might want to do that to stay interactive), so I<if you are
		832	using this from a module, never require a blocking wait>. Instead, let the
		833	caller decide whether the call will block or not (for example, by coupling
		834	condition variables with some kind of request results and supporting
		835	callbacks so the caller knows that getting the result will not block,
		836	while still supporting blocking waits if the caller so desires).
		837
		838	You can ensure that C<< ->recv >> never blocks by setting a callback and
		839	only calling C<< ->recv >> from within that callback (or at a later
		840	time). This will work even when the event loop does not support blocking
		841	waits otherwise.
		842
		843	=item $bool = $cv->ready
		844
		845	Returns true when the condition is "true", i.e. whether C<send> or
		846	C<croak> have been called.
		847
		848	=item $cb = $cv->cb ($cb->($cv))
		849
		850	This is a mutator function that returns the callback set and optionally
		851	replaces it before doing so.
		852
		853	The callback will be called when the condition becomes "true", i.e. when
		854	C<send> or C<croak> are called, with the only argument being the
		855	condition variable itself. If the condition is already true, the
		856	callback is called immediately when it is set. Calling C<recv> inside
		857	the callback or at any later time is guaranteed not to block.
		858
		859	=back
		860
		861	=head1 SUPPORTED EVENT LOOPS/BACKENDS
		862
		863	The available backend classes are (every class has its own manpage):
		864
		865	=over 4
		866
		867	=item Backends that are autoprobed when no other event loop can be found.
		868
		869	EV is the preferred backend when no other event loop seems to be in
		870	use. If EV is not installed, then AnyEvent will fall back to its own
		871	pure-perl implementation, which is available everywhere as it comes with
		872	AnyEvent itself.
		873
		874	AnyEvent::Impl::EV based on EV (interface to libev, best choice).
		875	AnyEvent::Impl::Perl pure-perl AnyEvent::Loop, fast and portable.
		876
		877	=item Backends that are transparently being picked up when they are used.
		878
		879	These will be used if they are already loaded when the first watcher
		880	is created, in which case it is assumed that the application is using
		881	them. This means that AnyEvent will automatically pick the right backend
		882	when the main program loads an event module before anything starts to
		883	create watchers. Nothing special needs to be done by the main program.
		884
		885	AnyEvent::Impl::Event based on Event, very stable, few glitches.
		886	AnyEvent::Impl::Glib based on Glib, slow but very stable.
		887	AnyEvent::Impl::Tk based on Tk, very broken.
		888	AnyEvent::Impl::UV based on UV, innovated square wheels.
		889	AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
		890	AnyEvent::Impl::POE based on POE, very slow, some limitations.
		891	AnyEvent::Impl::Irssi used when running within irssi.
		892	AnyEvent::Impl::IOAsync based on IO::Async.
		893	AnyEvent::Impl::Cocoa based on Cocoa::EventLoop.
		894	AnyEvent::Impl::FLTK based on FLTK (fltk 2 binding).
		895
		896	=item Backends with special needs.
		897
		898	Qt requires the Qt::Application to be instantiated first, but will
		899	otherwise be picked up automatically. As long as the main program
		900	instantiates the application before any AnyEvent watchers are created,
		901	everything should just work.
		902
		903	AnyEvent::Impl::Qt based on Qt.
		904
		905	=item Event loops that are indirectly supported via other backends.
		906
		907	Some event loops can be supported via other modules:
		908
		909	There is no direct support for WxWidgets (L<Wx>) or L<Prima>.
		910
		911	B<WxWidgets> has no support for watching file handles. However, you can
		912	use WxWidgets through the POE adaptor, as POE has a Wx backend that simply
		913	polls 20 times per second, which was considered to be too horrible to even
		914	consider for AnyEvent.
		915
		916	B<Prima> is not supported as nobody seems to be using it, but it has a POE
		917	backend, so it can be supported through POE.
		918
		919	AnyEvent knows about both L<Prima> and L<Wx>, however, and will try to
		920	load L<POE> when detecting them, in the hope that POE will pick them up,
		921	in which case everything will be automatic.
		922
		923	=back
		924
		925	=head1 GLOBAL VARIABLES AND FUNCTIONS
		926
		927	These are not normally required to use AnyEvent, but can be useful to
		928	write AnyEvent extension modules.
		929
		930	=over 4
		931
		932	=item $AnyEvent::MODEL
		933
		934	Contains C<undef> until the first watcher is being created, before the
		935	backend has been autodetected.
		936
		937	Afterwards it contains the event model that is being used, which is the
		938	name of the Perl class implementing the model. This class is usually one
		939	of the C<AnyEvent::Impl::xxx> modules, but can be any other class in the
		940	case AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode> it
		941	will be C<urxvt::anyevent>).
		942
		943	=item AnyEvent::detect
		944
		945	Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
		946	if necessary. You should only call this function right before you would
		947	have created an AnyEvent watcher anyway, that is, as late as possible at
		948	runtime, and not e.g. during initialisation of your module.
		949
		950	The effect of calling this function is as if a watcher had been created
		951	(specifically, actions that happen "when the first watcher is created"
		952	happen when calling detetc as well).
		953
		954	If you need to do some initialisation before AnyEvent watchers are
		955	created, use C<post_detect>.
		956
		957	=item $guard = AnyEvent::post_detect { BLOCK }
		958
		959	Arranges for the code block to be executed as soon as the event model is
		960	autodetected (or immediately if that has already happened).
		961
		962	The block will be executed I<after> the actual backend has been detected
		963	(C<$AnyEvent::MODEL> is set), but I<before> any watchers have been
		964	created, so it is possible to e.g. patch C<@AnyEvent::ISA> or do
		965	other initialisations - see the sources of L<AnyEvent::Strict> or
		966	L<AnyEvent::AIO> to see how this is used.
		967
		968	The most common usage is to create some global watchers, without forcing
		969	event module detection too early, for example, L<AnyEvent::AIO> creates
		970	and installs the global L<IO::AIO> watcher in a C<post_detect> block to
		971	avoid autodetecting the event module at load time.
		972
		973	If called in scalar or list context, then it creates and returns an object
		974	that automatically removes the callback again when it is destroyed (or
		975	C<undef> when the hook was immediately executed). See L<AnyEvent::AIO> for
		976	a case where this is useful.
		977
		978	Example: Create a watcher for the IO::AIO module and store it in
		979	C<$WATCHER>, but do so only do so after the event loop is initialised.
		980
		981	our WATCHER;
		982
		983	my $guard = AnyEvent::post_detect {
		984	$WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
		985	};
		986
		987	# the ||= is important in case post_detect immediately runs the block,
		988	# as to not clobber the newly-created watcher. assigning both watcher and
		989	# post_detect guard to the same variable has the advantage of users being
		990	# able to just C<undef $WATCHER> if the watcher causes them grief.
		991
		992	$WATCHER ||= $guard;
		993
		994	=item @AnyEvent::post_detect
		995
		996	If there are any code references in this array (you can C<push> to it
		997	before or after loading AnyEvent), then they will be called directly
		998	after the event loop has been chosen.
		999
		1000	You should check C<$AnyEvent::MODEL> before adding to this array, though:
		1001	if it is defined then the event loop has already been detected, and the
		1002	array will be ignored.
		1003
		1004	Best use C<AnyEvent::post_detect { BLOCK }> when your application allows
		1005	it, as it takes care of these details.
		1006
		1007	This variable is mainly useful for modules that can do something useful
		1008	when AnyEvent is used and thus want to know when it is initialised, but do
		1009	not need to even load it by default. This array provides the means to hook
		1010	into AnyEvent passively, without loading it.
		1011
		1012	Example: To load Coro::AnyEvent whenever Coro and AnyEvent are used
		1013	together, you could put this into Coro (this is the actual code used by
		1014	Coro to accomplish this):
		1015
		1016	if (defined $AnyEvent::MODEL) {
		1017	# AnyEvent already initialised, so load Coro::AnyEvent
		1018	require Coro::AnyEvent;
		1019	} else {
		1020	# AnyEvent not yet initialised, so make sure to load Coro::AnyEvent
		1021	# as soon as it is
		1022	push @AnyEvent::post_detect, sub { require Coro::AnyEvent };
		1023	}
		1024
		1025	=item AnyEvent::postpone { BLOCK }
		1026
		1027	Arranges for the block to be executed as soon as possible, but not before
		1028	the call itself returns. In practise, the block will be executed just
		1029	before the event loop polls for new events, or shortly afterwards.
		1030
		1031	This function never returns anything (to make the C<return postpone { ...
		1032	}> idiom more useful.
		1033
		1034	To understand the usefulness of this function, consider a function that
		1035	asynchronously does something for you and returns some transaction
		1036	object or guard to let you cancel the operation. For example,
		1037	C<AnyEvent::Socket::tcp_connect>:
		1038
		1039	# start a connection attempt unless one is active
		1040	$self->{connect_guard} ||= AnyEvent::Socket::tcp_connect "www.example.net", 80, sub {
		1041	delete $self->{connect_guard};
		1042	...
		1043	};
		1044
		1045	Imagine that this function could instantly call the callback, for
		1046	example, because it detects an obvious error such as a negative port
		1047	number. Invoking the callback before the function returns causes problems
		1048	however: the callback will be called and will try to delete the guard
		1049	object. But since the function hasn't returned yet, there is nothing to
		1050	delete. When the function eventually returns it will assign the guard
		1051	object to C<< $self->{connect_guard} >>, where it will likely never be
		1052	deleted, so the program thinks it is still trying to connect.
		1053
		1054	This is where C<AnyEvent::postpone> should be used. Instead of calling the
		1055	callback directly on error:
		1056
		1057	$cb->(undef), return # signal error to callback, BAD!
		1058	if $some_error_condition;
		1059
		1060	It should use C<postpone>:
		1061
		1062	AnyEvent::postpone { $cb->(undef) }, return # signal error to callback, later
		1063	if $some_error_condition;
		1064
		1065	=item AnyEvent::log $level, $msg[, @args]
		1066
		1067	Log the given C<$msg> at the given C<$level>.
		1068
		1069	If L<AnyEvent::Log> is not loaded then this function makes a simple test
		1070	to see whether the message will be logged. If the test succeeds it will
		1071	load AnyEvent::Log and call C<AnyEvent::Log::log> - consequently, look at
		1072	the L<AnyEvent::Log> documentation for details.
		1073
		1074	If the test fails it will simply return. Right now this happens when a
		1075	numerical loglevel is used and it is larger than the level specified via
		1076	C<$ENV{PERL_ANYEVENT_VERBOSE}>.
		1077
		1078	If you want to sprinkle loads of logging calls around your code, consider
		1079	creating a logger callback with the C<AnyEvent::Log::logger> function,
		1080	which can reduce typing, codesize and can reduce the logging overhead
		1081	enourmously.
		1082
		1083	=back
		1084
		1085	=head1 WHAT TO DO IN A MODULE
		1086
		1087	As a module author, you should C<use AnyEvent> and call AnyEvent methods
		1088	freely, but you should not load a specific event module or rely on it.
		1089
		1090	Be careful when you create watchers in the module body - AnyEvent will
		1091	decide which event module to use as soon as the first method is called, so
		1092	by calling AnyEvent in your module body you force the user of your module
		1093	to load the event module first.
		1094
		1095	Never call C<< ->recv >> on a condition variable unless you I<know> that
		1096	the C<< ->send >> method has been called on it already. This is
		1097	because it will stall the whole program, and the whole point of using
		1098	events is to stay interactive.
		1099
		1100	It is fine, however, to call C<< ->recv >> when the user of your module
		1101	requests it (i.e. if you create a http request object ad have a method
		1102	called C<results> that returns the results, it may call C<< ->recv >>
		1103	freely, as the user of your module knows what she is doing. Always).
		1104
		1105	=head1 WHAT TO DO IN THE MAIN PROGRAM
		1106
		1107	There will always be a single main program - the only place that should
		1108	dictate which event model to use.
		1109
		1110	If the program is not event-based, it need not do anything special, even
		1111	when it depends on a module that uses an AnyEvent. If the program itself
		1112	uses AnyEvent, but does not care which event loop is used, all it needs
		1113	to do is C<use AnyEvent>. In either case, AnyEvent will choose the best
		1114	available loop implementation.
		1115
		1116	If the main program relies on a specific event model - for example, in
		1117	Gtk2 programs you have to rely on the Glib module - you should load the
		1118	event module before loading AnyEvent or any module that uses it: generally
		1119	speaking, you should load it as early as possible. The reason is that
		1120	modules might create watchers when they are loaded, and AnyEvent will
		1121	decide on the event model to use as soon as it creates watchers, and it
		1122	might choose the wrong one unless you load the correct one yourself.
		1123
		1124	You can chose to use a pure-perl implementation by loading the
		1125	C<AnyEvent::Loop> module, which gives you similar behaviour
		1126	everywhere, but letting AnyEvent chose the model is generally better.
		1127
		1128	=head2 MAINLOOP EMULATION
		1129
		1130	Sometimes (often for short test scripts, or even standalone programs who
		1131	only want to use AnyEvent), you do not want to run a specific event loop.
		1132
		1133	In that case, you can use a condition variable like this:
		1134
		1135	AnyEvent->condvar->recv;
		1136
		1137	This has the effect of entering the event loop and looping forever.
		1138
		1139	Note that usually your program has some exit condition, in which case
		1140	it is better to use the "traditional" approach of storing a condition
		1141	variable somewhere, waiting for it, and sending it when the program should
		1142	exit cleanly.
		1143
		1144
		1145	=head1 OTHER MODULES
		1146
		1147	The following is a non-exhaustive list of additional modules that use
		1148	AnyEvent as a client and can therefore be mixed easily with other
		1149	AnyEvent modules and other event loops in the same program. Some of the
		1150	modules come as part of AnyEvent, the others are available via CPAN (see
		1151	L<http://search.cpan.org/search?m=module&q=anyevent%3A%3A*> for
		1152	a longer non-exhaustive list), and the list is heavily biased towards
		1153	modules of the AnyEvent author himself :)
		1154
		1155	=over 4
		1156
		1157	=item L<AnyEvent::Util> (part of the AnyEvent distribution)
		1158
		1159	Contains various utility functions that replace often-used blocking
		1160	functions such as C<inet_aton> with event/callback-based versions.
		1161
		1162	=item L<AnyEvent::Socket> (part of the AnyEvent distribution)
		1163
		1164	Provides various utility functions for (internet protocol) sockets,
		1165	addresses and name resolution. Also functions to create non-blocking tcp
		1166	connections or tcp servers, with IPv6 and SRV record support and more.
		1167
		1168	=item L<AnyEvent::Handle> (part of the AnyEvent distribution)
		1169
		1170	Provide read and write buffers, manages watchers for reads and writes,
		1171	supports raw and formatted I/O, I/O queued and fully transparent and
		1172	non-blocking SSL/TLS (via L<AnyEvent::TLS>).
		1173
		1174	=item L<AnyEvent::DNS> (part of the AnyEvent distribution)
		1175
		1176	Provides rich asynchronous DNS resolver capabilities.
		1177
		1178	=item L<AnyEvent::HTTP>, L<AnyEvent::IRC>, L<AnyEvent::XMPP>, L<AnyEvent::GPSD>, L<AnyEvent::IGS>, L<AnyEvent::FCP>
		1179
		1180	Implement event-based interfaces to the protocols of the same name (for
		1181	the curious, IGS is the International Go Server and FCP is the Freenet
		1182	Client Protocol).
		1183
		1184	=item L<AnyEvent::AIO> (part of the AnyEvent distribution)
		1185
		1186	Truly asynchronous (as opposed to non-blocking) I/O, should be in the
		1187	toolbox of every event programmer. AnyEvent::AIO transparently fuses
		1188	L<IO::AIO> and AnyEvent together, giving AnyEvent access to event-based
		1189	file I/O, and much more.
		1190
		1191	=item L<AnyEvent::Fork>, L<AnyEvent::Fork::RPC>, L<AnyEvent::Fork::Pool>, L<AnyEvent::Fork::Remote>
		1192
		1193	These let you safely fork new subprocesses, either locally or
		1194	remotely (e.g.v ia ssh), using some RPC protocol or not, without
		1195	the limitations normally imposed by fork (AnyEvent works fine for
		1196	example). Dynamically-resized worker pools are obviously included as well.
		1197
		1198	And they are quite tiny and fast as well - "abusing" L<AnyEvent::Fork>
		1199	just to exec external programs can easily beat using C<fork> and C<exec>
		1200	(or even C<system>) in most programs.
		1201
		1202	=item L<AnyEvent::Filesys::Notify>
		1203
		1204	AnyEvent is good for non-blocking stuff, but it can't detect file or
		1205	path changes (e.g. "watch this directory for new files", "watch this
		1206	file for changes"). The L<AnyEvent::Filesys::Notify> module promises to
		1207	do just that in a portbale fashion, supporting inotify on GNU/Linux and
		1208	some weird, without doubt broken, stuff on OS X to monitor files. It can
		1209	fall back to blocking scans at regular intervals transparently on other
		1210	platforms, so it's about as portable as it gets.
		1211
		1212	(I haven't used it myself, but it seems the biggest problem with it is
		1213	it quite bad performance).
		1214
		1215	=item L<AnyEvent::DBI>
		1216
		1217	Executes L<DBI> requests asynchronously in a proxy process for you,
		1218	notifying you in an event-based way when the operation is finished.
		1219
		1220	=item L<AnyEvent::FastPing>
		1221
		1222	The fastest ping in the west.
		1223
		1224	=item L<Coro>
		1225
		1226	Has special support for AnyEvent via L<Coro::AnyEvent>, which allows you
		1227	to simply invert the flow control - don't call us, we will call you:
		1228
		1229	async {
		1230	Coro::AnyEvent::sleep 5; # creates a 5s timer and waits for it
		1231	print "5 seconds later!\n";
		1232
		1233	Coro::AnyEvent::readable *STDIN; # uses an I/O watcher
		1234	my $line = <STDIN>; # works for ttys
		1235
		1236	AnyEvent::HTTP::http_get "url", Coro::rouse_cb;
		1237	my ($body, $hdr) = Coro::rouse_wait;
		1238	};
		1239
		1240	=back
		1241
11	=cut	1242	=cut
12		1243
13	package AnyEvent;	1244	package AnyEvent;
14		1245
		1246	BEGIN {
		1247	require "AnyEvent/constants.pl";
		1248	&AnyEvent::common_sense;
		1249	}
		1250
15	use Carp;	1251	use Carp ();
16		1252
17	$VERSION = 0.1;	1253	our $VERSION = '7.08';
		1254	our $MODEL;
		1255	our @ISA;
		1256	our @REGISTRY;
		1257	our $VERBOSE;
		1258	our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
		1259	our $MAX_SIGNAL_LATENCY = $ENV{PERL_ANYEVENT_MAX_SIGNAL_LATENCY} || 10; # executes after the BEGIN block below (tainting!)
18		1260
19	no warnings;	1261	BEGIN {
		1262	eval "sub TAINT (){" . (${^TAINT}*1) . "}";
20		1263
		1264	delete @ENV{grep /^PERL_ANYEVENT_/, keys %ENV}
		1265	if ${^TAINT};
		1266
		1267	$ENV{"PERL_ANYEVENT_$_"} = $ENV{"AE_$_"}
		1268	for grep s/^AE_// && !exists $ENV{"PERL_ANYEVENT_$_"}, keys %ENV;
		1269
		1270	@ENV{grep /^PERL_ANYEVENT_/, keys %ENV} = ()
		1271	if ${^TAINT};
		1272
		1273	# $ENV{PERL_ANYEVENT_xxx} now valid
		1274
		1275	$VERBOSE = length $ENV{PERL_ANYEVENT_VERBOSE} ? $ENV{PERL_ANYEVENT_VERBOSE}*1 : 4;
		1276
		1277	my $idx;
		1278	$PROTOCOL{$_} = ++$idx
		1279	for reverse split /\s*,\s*/,
		1280	$ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
		1281	}
		1282
		1283	our @post_detect;
		1284
		1285	sub post_detect(&) {
		1286	my ($cb) = @_;
		1287
		1288	push @post_detect, $cb;
		1289
		1290	defined wantarray
		1291	? bless \$cb, "AnyEvent::Util::postdetect"
		1292	: ()
		1293	}
		1294
		1295	sub AnyEvent::Util::postdetect::DESTROY {
		1296	@post_detect = grep $_ != ${$_[0]}, @post_detect;
		1297	}
		1298
		1299	our $POSTPONE_W;
		1300	our @POSTPONE;
		1301
		1302	sub _postpone_exec {
		1303	undef $POSTPONE_W;
		1304
		1305	&{ shift @POSTPONE }
		1306	while @POSTPONE;
		1307	}
		1308
		1309	sub postpone(&) {
		1310	push @POSTPONE, shift;
		1311
		1312	$POSTPONE_W ||= AE::timer (0, 0, \&_postpone_exec);
		1313
		1314	()
		1315	}
		1316
		1317	sub log($$;@) {
		1318	# only load the big bloated module when we actually are about to log something
		1319	if ($_[0] <= ($VERBOSE || 1)) { # also catches non-numeric levels(!) and fatal
		1320	local ($!, $@);
		1321	require AnyEvent::Log; # among other things, sets $VERBOSE to 9
		1322	# AnyEvent::Log overwrites this function
		1323	goto &log;
		1324	}
		1325
		1326	0 # not logged
		1327	}
		1328
		1329	sub _logger($;$) {
		1330	my ($level, $renabled) = @_;
		1331
		1332	$$renabled = $level <= $VERBOSE;
		1333
		1334	my $logger = [(caller)[0], $level, $renabled];
		1335
		1336	$AnyEvent::Log::LOGGER{$logger+0} = $logger;
		1337
		1338	# return unless defined wantarray;
		1339	#
		1340	# require AnyEvent::Util;
		1341	# my $guard = AnyEvent::Util::guard (sub {
		1342	# # "clean up"
		1343	# delete $LOGGER{$logger+0};
		1344	# });
		1345	#
		1346	# sub {
		1347	# return 0 unless $$renabled;
		1348	#
		1349	# $guard if 0; # keep guard alive, but don't cause runtime overhead
		1350	# require AnyEvent::Log unless $AnyEvent::Log::VERSION;
		1351	# package AnyEvent::Log;
		1352	# _log ($logger->[0], $level, @_) # logger->[0] has been converted at load time
		1353	# }
		1354	}
		1355
		1356	if (length $ENV{PERL_ANYEVENT_LOG}) {
		1357	require AnyEvent::Log; # AnyEvent::Log does the thing for us
		1358	}
		1359
21	my @models = (	1360	our @models = (
22	[Coro => Coro::Event::],	1361	[EV:: => AnyEvent::Impl::EV::],
23	[Event => Event::],	1362	[AnyEvent::Loop:: => AnyEvent::Impl::Perl::],
24	[Glib => Glib::],	1363	# everything below here will not (normally) be autoprobed
25	[Tk => Tk::],	1364	# as the pure perl backend should work everywhere
		1365	# and is usually faster
		1366	[Irssi:: => AnyEvent::Impl::Irssi::], # Irssi has a bogus "Event" package, so msut be near the top
		1367	[Event:: => AnyEvent::Impl::Event::], # slow, stable
		1368	[Glib:: => AnyEvent::Impl::Glib::], # becomes extremely slow with many watchers
		1369	# everything below here should not be autoloaded
		1370	[Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
		1371	[Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
		1372	[UV:: => AnyEvent::Impl::UV::], # switched from libev, added back all bugs imaginable
		1373	[Qt:: => AnyEvent::Impl::Qt::], # requires special main program
		1374	[POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
		1375	[Wx:: => AnyEvent::Impl::POE::],
		1376	[Prima:: => AnyEvent::Impl::POE::],
		1377	[IO::Async::Loop:: => AnyEvent::Impl::IOAsync::], # a bitch to autodetect
		1378	[Cocoa::EventLoop:: => AnyEvent::Impl::Cocoa::],
		1379	[FLTK:: => AnyEvent::Impl::FLTK::],
26	);	1380	);
27		1381
28	sub AUTOLOAD {	1382	our @isa_hook;
29	$AUTOLOAD =~ s/.*://;
30		1383
31	for (@models) {	1384	sub _isa_set {
32	my ($model, $package) = @$_;	1385	my @pkg = ("AnyEvent", (map $_->[0], grep defined, @isa_hook), $MODEL);
33	if (defined ${"$package\::VERSION"}) {	1386
34	$EVENT = "AnyEvent::Impl::$model";	1387	@{"$pkg[$_-1]::ISA"} = $pkg[$_]
35	eval "require $EVENT"; die if $@;	1388	for 1 .. $#pkg;
36	goto &{"$EVENT\::$AUTOLOAD"};	1389
		1390	grep $_ && $_->[1], @isa_hook
		1391	and AE::_reset ();
		1392	}
		1393
		1394	# used for hooking AnyEvent::Strict and AnyEvent::Debug::Wrap into the class hierarchy
		1395	sub _isa_hook($$;$) {
		1396	my ($i, $pkg, $reset_ae) = @_;
		1397
		1398	$isa_hook[$i] = $pkg ? [$pkg, $reset_ae] : undef;
		1399
		1400	_isa_set;
		1401	}
		1402
		1403	# all autoloaded methods reserve the complete glob, not just the method slot.
		1404	# due to bugs in perls method cache implementation.
		1405	our @methods = qw(io timer time now now_update signal child idle condvar);
		1406
		1407	sub detect() {
		1408	return $MODEL if $MODEL; # some programs keep references to detect
		1409
		1410	# IO::Async::Loop::AnyEvent is extremely evil, refuse to work with it
		1411	# the author knows about the problems and what it does to AnyEvent as a whole
		1412	# (and the ability of others to use AnyEvent), but simply wants to abuse AnyEvent
		1413	# anyway.
		1414	AnyEvent::log fatal => "IO::Async::Loop::AnyEvent detected - that module is broken by\n"
		1415	. "design, abuses internals and breaks AnyEvent - will not continue."
		1416	if exists $INC{"IO/Async/Loop/AnyEvent.pm"};
		1417
		1418	local $!; # for good measure
		1419	local $SIG{__DIE__}; # we use eval
		1420
		1421	# free some memory
		1422	*detect = sub () { $MODEL };
		1423	# undef &func doesn't correctly update the method cache. grmbl.
		1424	# so we delete the whole glob. grmbl.
		1425	# otoh, perl doesn't let me undef an active usb, but it lets me free
		1426	# a glob with an active sub. hrm. i hope it works, but perl is
		1427	# usually buggy in this department. sigh.
		1428	delete @{"AnyEvent::"}{@methods};
		1429	undef @methods;
		1430
		1431	if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z0-9:]+)$/) {
		1432	my $model = $1;
		1433	$model = "AnyEvent::Impl::$model" unless $model =~ s/::$//;
		1434	if (eval "require $model") {
		1435	AnyEvent::log 7 => "Loaded model '$model' (forced by \$ENV{PERL_ANYEVENT_MODEL}), using it.";
		1436	$MODEL = $model;
		1437	} else {
		1438	AnyEvent::log 4 => "Unable to load model '$model' (from \$ENV{PERL_ANYEVENT_MODEL}):\n$@";
37	}	1439	}
38	}	1440	}
39		1441
40	for (@models) {	1442	# check for already loaded models
41	my ($model, $package) = @$_;	1443	unless ($MODEL) {
42	$EVENT = "AnyEvent::Impl::$model";	1444	for (@REGISTRY, @models) {
		1445	my ($package, $model) = @$_;
		1446	if (${"$package\::VERSION"} > 0) {
43	if (eval "require $EVENT") {	1447	if (eval "require $model") {
44	goto &{"$EVENT\::$AUTOLOAD"};	1448	AnyEvent::log 7 => "Autodetected model '$model', using it.";
		1449	$MODEL = $model;
		1450	last;
		1451	} else {
		1452	AnyEvent::log 8 => "Detected event loop $package, but cannot load '$model', skipping: $@";
		1453	}
		1454	}
		1455	}
		1456
		1457	unless ($MODEL) {
		1458	# try to autoload a model
		1459	for (@REGISTRY, @models) {
		1460	my ($package, $model) = @$_;
		1461	if (
		1462	eval "require $package"
		1463	and ${"$package\::VERSION"} > 0
		1464	and eval "require $model"
		1465	) {
		1466	AnyEvent::log 7 => "Autoloaded model '$model', using it.";
		1467	$MODEL = $model;
		1468	last;
		1469	}
		1470	}
		1471
		1472	$MODEL
		1473	or AnyEvent::log fatal => "Backend autodetection failed - did you properly install AnyEvent?";
45	}	1474	}
46	}	1475	}
47		1476
48	die "No event module selected for AnyEvent and autodetect failed. Install any of these: Coro, Event, Glib or Tk.";	1477	# free memory only needed for probing
49	}	1478	undef @models;
		1479	undef @REGISTRY;
50		1480
51	1;	1481	push @{"$MODEL\::ISA"}, "AnyEvent::Base";
52		1482
		1483	# now nuke some methods that are overridden by the backend.
		1484	# SUPER usage is not allowed in these.
		1485	for (qw(time signal child idle)) {
		1486	undef &{"AnyEvent::Base::$_"}
		1487	if defined &{"$MODEL\::$_"};
		1488	}
		1489
		1490	_isa_set;
		1491
		1492	# we're officially open!
		1493
		1494	if ($ENV{PERL_ANYEVENT_STRICT}) {
		1495	require AnyEvent::Strict;
		1496	}
		1497
		1498	if ($ENV{PERL_ANYEVENT_DEBUG_WRAP}) {
		1499	require AnyEvent::Debug;
		1500	AnyEvent::Debug::wrap ($ENV{PERL_ANYEVENT_DEBUG_WRAP});
		1501	}
		1502
		1503	if (length $ENV{PERL_ANYEVENT_DEBUG_SHELL}) {
		1504	require AnyEvent::Socket;
		1505	require AnyEvent::Debug;
		1506
		1507	my $shell = $ENV{PERL_ANYEVENT_DEBUG_SHELL};
		1508	$shell =~ s/\$\$/$$/g;
		1509
		1510	my ($host, $service) = AnyEvent::Socket::parse_hostport ($shell);
		1511	$AnyEvent::Debug::SHELL = AnyEvent::Debug::shell ($host, $service);
		1512	}
		1513
		1514	# now the anyevent environment is set up as the user told us to, so
		1515	# call the actual user code - post detects
		1516
		1517	(shift @post_detect)->() while @post_detect;
		1518	undef @post_detect;
		1519
		1520	*post_detect = sub(&) {
		1521	shift->();
		1522
		1523	undef
		1524	};
		1525
		1526	$MODEL
		1527	}
		1528
		1529	for my $name (@methods) {
		1530	*$name = sub {
		1531	detect;
		1532	# we use goto because
		1533	# a) it makes the thunk more transparent
		1534	# b) it allows us to delete the thunk later
		1535	goto &{ UNIVERSAL::can AnyEvent => "SUPER::$name" }
		1536	};
		1537	}
		1538
		1539	# utility function to dup a filehandle. this is used by many backends
		1540	# to support binding more than one watcher per filehandle (they usually
		1541	# allow only one watcher per fd, so we dup it to get a different one).
		1542	sub _dupfh($$;$$) {
		1543	my ($poll, $fh, $r, $w) = @_;
		1544
		1545	# cygwin requires the fh mode to be matching, unix doesn't
		1546	my ($rw, $mode) = $poll eq "r" ? ($r, "<&") : ($w, ">&");
		1547
		1548	open my $fh2, $mode, $fh
		1549	or die "AnyEvent->io: cannot dup() filehandle in mode '$poll': $!,";
		1550
		1551	# we assume CLOEXEC is already set by perl in all important cases
		1552
		1553	($fh2, $rw)
		1554	}
		1555
		1556	=head1 SIMPLIFIED AE API
		1557
		1558	Starting with version 5.0, AnyEvent officially supports a second, much
		1559	simpler, API that is designed to reduce the calling, typing and memory
		1560	overhead by using function call syntax and a fixed number of parameters.
		1561
		1562	See the L<AE> manpage for details.
		1563
		1564	=cut
		1565
		1566	package AE;
		1567
		1568	our $VERSION = $AnyEvent::VERSION;
		1569
		1570	sub _reset() {
		1571	eval q{
		1572	# fall back to the main API by default - backends and AnyEvent::Base
		1573	# implementations can overwrite these.
		1574
		1575	sub io($$$) {
		1576	AnyEvent->io (fh => $_[0], poll => $_[1] ? "w" : "r", cb => $_[2])
		1577	}
		1578
		1579	sub timer($$$) {
		1580	AnyEvent->timer (after => $_[0], interval => $_[1], cb => $_[2])
		1581	}
		1582
		1583	sub signal($$) {
		1584	AnyEvent->signal (signal => $_[0], cb => $_[1])
		1585	}
		1586
		1587	sub child($$) {
		1588	AnyEvent->child (pid => $_[0], cb => $_[1])
		1589	}
		1590
		1591	sub idle($) {
		1592	AnyEvent->idle (cb => $_[0]);
		1593	}
		1594
		1595	sub cv(;&) {
		1596	AnyEvent->condvar (@_ ? (cb => $_[0]) : ())
		1597	}
		1598
		1599	sub now() {
		1600	AnyEvent->now
		1601	}
		1602
		1603	sub now_update() {
		1604	AnyEvent->now_update
		1605	}
		1606
		1607	sub time() {
		1608	AnyEvent->time
		1609	}
		1610
		1611	*postpone = \&AnyEvent::postpone;
		1612	*log = \&AnyEvent::log;
		1613	};
		1614	die if $@;
		1615	}
		1616
		1617	BEGIN { _reset }
		1618
		1619	package AnyEvent::Base;
		1620
		1621	# default implementations for many methods
		1622
		1623	sub time {
		1624	eval q{ # poor man's autoloading {}
		1625	# probe for availability of Time::HiRes
		1626	if (eval "use Time::HiRes (); Time::HiRes::time (); 1") {
		1627	*time = sub { Time::HiRes::time () };
		1628	*AE::time = \& Time::HiRes::time ;
		1629	*now = \&time;
		1630	AnyEvent::log 8 => "using Time::HiRes for sub-second timing accuracy.";
		1631	# if (eval "use POSIX (); (POSIX::times())...
		1632	} else {
		1633	*time = sub { CORE::time };
		1634	*AE::time = sub (){ CORE::time };
		1635	*now = \&time;
		1636	AnyEvent::log 3 => "Using built-in time(), no sub-second resolution!";
		1637	}
		1638	};
		1639	die if $@;
		1640
		1641	&time
		1642	}
		1643
		1644	*now = \&time;
		1645	sub now_update { }
		1646
		1647	sub _poll {
		1648	Carp::croak "$AnyEvent::MODEL does not support blocking waits. Caught";
		1649	}
		1650
		1651	# default implementation for ->condvar
		1652	# in fact, the default should not be overwritten
		1653
		1654	sub condvar {
		1655	eval q{ # poor man's autoloading {}
		1656	*condvar = sub {
		1657	bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, "AnyEvent::CondVar"
		1658	};
		1659
		1660	*AE::cv = sub (;&) {
		1661	bless { @_ ? (_ae_cb => shift) : () }, "AnyEvent::CondVar"
		1662	};
		1663	};
		1664	die if $@;
		1665
		1666	&condvar
		1667	}
		1668
		1669	# default implementation for ->signal
		1670
		1671	our $HAVE_ASYNC_INTERRUPT;
		1672
		1673	sub _have_async_interrupt() {
		1674	$HAVE_ASYNC_INTERRUPT = 1*(!$ENV{PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT}
		1675	&& eval "use Async::Interrupt 1.02 (); 1")
		1676	unless defined $HAVE_ASYNC_INTERRUPT;
		1677
		1678	$HAVE_ASYNC_INTERRUPT
		1679	}
		1680
		1681	our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
		1682	our (%SIG_ASY, %SIG_ASY_W);
		1683	our ($SIG_COUNT, $SIG_TW);
		1684
		1685	# install a dummy wakeup watcher to reduce signal catching latency
		1686	# used by Impls
		1687	sub _sig_add() {
		1688	unless ($SIG_COUNT++) {
		1689	# try to align timer on a full-second boundary, if possible
		1690	my $NOW = AE::now;
		1691
		1692	$SIG_TW = AE::timer
		1693	$MAX_SIGNAL_LATENCY - ($NOW - int $NOW),
		1694	$MAX_SIGNAL_LATENCY,
		1695	sub { } # just for the PERL_ASYNC_CHECK
		1696	;
		1697	}
		1698	}
		1699
		1700	sub _sig_del {
		1701	undef $SIG_TW
		1702	unless --$SIG_COUNT;
		1703	}
		1704
		1705	our $_sig_name_init; $_sig_name_init = sub {
		1706	eval q{ # poor man's autoloading {}
		1707	undef $_sig_name_init;
		1708
		1709	if (_have_async_interrupt) {
		1710	*sig2num = \&Async::Interrupt::sig2num;
		1711	*sig2name = \&Async::Interrupt::sig2name;
		1712	} else {
		1713	require Config;
		1714
		1715	my %signame2num;
		1716	@signame2num{ split ' ', $Config::Config{sig_name} }
		1717	= split ' ', $Config::Config{sig_num};
		1718
		1719	my @signum2name;
		1720	@signum2name[values %signame2num] = keys %signame2num;
		1721
		1722	*sig2num = sub($) {
		1723	$_[0] > 0 ? shift : $signame2num{+shift}
		1724	};
		1725	*sig2name = sub ($) {
		1726	$_[0] > 0 ? $signum2name[+shift] : shift
		1727	};
		1728	}
		1729	};
		1730	die if $@;
		1731	};
		1732
		1733	sub sig2num ($) { &$_sig_name_init; &sig2num }
		1734	sub sig2name($) { &$_sig_name_init; &sig2name }
		1735
		1736	sub signal {
		1737	eval q{ # poor man's autoloading {}
		1738	# probe for availability of Async::Interrupt
		1739	if (_have_async_interrupt) {
		1740	AnyEvent::log 8 => "Using Async::Interrupt for race-free signal handling.";
		1741
		1742	$SIGPIPE_R = new Async::Interrupt::EventPipe;
		1743	$SIG_IO = AE::io $SIGPIPE_R->fileno, 0, \&_signal_exec;
		1744
		1745	} else {
		1746	AnyEvent::log 8 => "Using emulated perl signal handling with latency timer.";
		1747
		1748	if (AnyEvent::WIN32) {
		1749	require AnyEvent::Util;
		1750
		1751	($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
		1752	AnyEvent::Util::fh_nonblocking ($SIGPIPE_R, 1) if $SIGPIPE_R;
		1753	AnyEvent::Util::fh_nonblocking ($SIGPIPE_W, 1) if $SIGPIPE_W; # just in case
		1754	} else {
		1755	pipe $SIGPIPE_R, $SIGPIPE_W;
		1756	fcntl $SIGPIPE_R, AnyEvent::F_SETFL, AnyEvent::O_NONBLOCK if $SIGPIPE_R;
		1757	fcntl $SIGPIPE_W, AnyEvent::F_SETFL, AnyEvent::O_NONBLOCK if $SIGPIPE_W; # just in case
		1758
		1759	# not strictly required, as $^F is normally 2, but let's make sure...
		1760	fcntl $SIGPIPE_R, AnyEvent::F_SETFD, AnyEvent::FD_CLOEXEC;
		1761	fcntl $SIGPIPE_W, AnyEvent::F_SETFD, AnyEvent::FD_CLOEXEC;
		1762	}
		1763
		1764	$SIGPIPE_R
		1765	or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
		1766
		1767	$SIG_IO = AE::io $SIGPIPE_R, 0, \&_signal_exec;
		1768	}
		1769
		1770	*signal = $HAVE_ASYNC_INTERRUPT
		1771	? sub {
		1772	my (undef, %arg) = @_;
		1773
		1774	# async::interrupt
		1775	my $signal = sig2num $arg{signal};
		1776	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1777
		1778	$SIG_ASY{$signal} ||= new Async::Interrupt
		1779	cb => sub { undef $SIG_EV{$signal} },
		1780	signal => $signal,
		1781	pipe => [$SIGPIPE_R->filenos],
		1782	pipe_autodrain => 0,
		1783	;
		1784
		1785	bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
		1786	}
		1787	: sub {
		1788	my (undef, %arg) = @_;
		1789
		1790	# pure perl
		1791	my $signal = sig2name $arg{signal};
		1792	$SIG_CB{$signal}{$arg{cb}} = $arg{cb};
		1793
		1794	$SIG{$signal} ||= sub {
		1795	local $!;
		1796	syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
		1797	undef $SIG_EV{$signal};
		1798	};
		1799
		1800	# can't do signal processing without introducing races in pure perl,
		1801	# so limit the signal latency.
		1802	_sig_add;
		1803
		1804	bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
		1805	}
		1806	;
		1807
		1808	*AnyEvent::Base::signal::DESTROY = sub {
		1809	my ($signal, $cb) = @{$_[0]};
		1810
		1811	_sig_del;
		1812
		1813	delete $SIG_CB{$signal}{$cb};
		1814
		1815	$HAVE_ASYNC_INTERRUPT
		1816	? delete $SIG_ASY{$signal}
		1817	: # delete doesn't work with older perls - they then
		1818	# print weird messages, or just unconditionally exit
		1819	# instead of getting the default action.
		1820	undef $SIG{$signal}
		1821	unless keys %{ $SIG_CB{$signal} };
		1822	};
		1823
		1824	*_signal_exec = sub {
		1825	$HAVE_ASYNC_INTERRUPT
		1826	? $SIGPIPE_R->drain
		1827	: sysread $SIGPIPE_R, (my $dummy), 9;
		1828
		1829	while (%SIG_EV) {
		1830	for (keys %SIG_EV) {
		1831	delete $SIG_EV{$_};
		1832	&$_ for values %{ $SIG_CB{$_} || {} };
		1833	}
		1834	}
		1835	};
		1836	};
		1837	die if $@;
		1838
		1839	&signal
		1840	}
		1841
		1842	# default implementation for ->child
		1843
		1844	our %PID_CB;
		1845	our $CHLD_W;
		1846	our $CHLD_DELAY_W;
		1847
		1848	# used by many Impl's
		1849	sub _emit_childstatus($$) {
		1850	my (undef, $rpid, $rstatus) = @_;
		1851
		1852	$_->($rpid, $rstatus)
		1853	for values %{ $PID_CB{$rpid} || {} },
		1854	values %{ $PID_CB{0} || {} };
		1855	}
		1856
		1857	sub child {
		1858	eval q{ # poor man's autoloading {}
		1859	*_sigchld = sub {
		1860	my $pid;
		1861
		1862	AnyEvent->_emit_childstatus ($pid, $?)
		1863	while ($pid = waitpid -1, WNOHANG) > 0;
		1864	};
		1865
		1866	*child = sub {
		1867	my (undef, %arg) = @_;
		1868
		1869	my $pid = $arg{pid};
		1870	my $cb = $arg{cb};
		1871
		1872	$PID_CB{$pid}{$cb+0} = $cb;
		1873
		1874	unless ($CHLD_W) {
		1875	$CHLD_W = AE::signal CHLD => \&_sigchld;
		1876	# child could be a zombie already, so make at least one round
		1877	&_sigchld;
		1878	}
		1879
		1880	bless [$pid, $cb+0], "AnyEvent::Base::child"
		1881	};
		1882
		1883	*AnyEvent::Base::child::DESTROY = sub {
		1884	my ($pid, $icb) = @{$_[0]};
		1885
		1886	delete $PID_CB{$pid}{$icb};
		1887	delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
		1888
		1889	undef $CHLD_W unless keys %PID_CB;
		1890	};
		1891	};
		1892	die if $@;
		1893
		1894	&child
		1895	}
		1896
		1897	# idle emulation is done by simply using a timer, regardless
		1898	# of whether the process is idle or not, and not letting
		1899	# the callback use more than 50% of the time.
		1900	sub idle {
		1901	eval q{ # poor man's autoloading {}
		1902	*idle = sub {
		1903	my (undef, %arg) = @_;
		1904
		1905	my ($cb, $w, $rcb) = $arg{cb};
		1906
		1907	$rcb = sub {
		1908	if ($cb) {
		1909	$w = AE::time;
		1910	&$cb;
		1911	$w = AE::time - $w;
		1912
		1913	# never use more then 50% of the time for the idle watcher,
		1914	# within some limits
		1915	$w = 0.0001 if $w < 0.0001;
		1916	$w = 5 if $w > 5;
		1917
		1918	$w = AE::timer $w, 0, $rcb;
		1919	} else {
		1920	# clean up...
		1921	undef $w;
		1922	undef $rcb;
		1923	}
		1924	};
		1925
		1926	$w = AE::timer 0.05, 0, $rcb;
		1927
		1928	bless \\$cb, "AnyEvent::Base::idle"
		1929	};
		1930
		1931	*AnyEvent::Base::idle::DESTROY = sub {
		1932	undef $${$_[0]};
		1933	};
		1934	};
		1935	die if $@;
		1936
		1937	&idle
		1938	}
		1939
		1940	package AnyEvent::CondVar;
		1941
		1942	our @ISA = AnyEvent::CondVar::Base::;
		1943
		1944	# only to be used for subclassing
		1945	sub new {
		1946	my $class = shift;
		1947	bless AnyEvent->condvar (@_), $class
		1948	}
		1949
		1950	package AnyEvent::CondVar::Base;
		1951
		1952	#use overload
		1953	# '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
		1954	# fallback => 1;
		1955
		1956	# save 300+ kilobytes by dirtily hardcoding overloading
		1957	${"AnyEvent::CondVar::Base::OVERLOAD"}{dummy}++; # Register with magic by touching.
		1958	*{'AnyEvent::CondVar::Base::()'} = sub { }; # "Make it findable via fetchmethod."
		1959	*{'AnyEvent::CondVar::Base::(&{}'} = sub { my $self = shift; sub { $self->send (@_) } }; # &{}
		1960	${'AnyEvent::CondVar::Base::()'} = 1; # fallback
		1961
		1962	our $WAITING;
		1963
		1964	sub _send {
		1965	# nop
		1966	}
		1967
		1968	sub _wait {
		1969	AnyEvent->_poll until $_[0]{_ae_sent};
		1970	}
		1971
		1972	sub send {
		1973	my $cv = shift;
		1974	$cv->{_ae_sent} = [@_];
		1975	(delete $cv->{_ae_cb})->($cv) if $cv->{_ae_cb};
		1976	$cv->_send;
		1977	}
		1978
		1979	sub croak {
		1980	$_[0]{_ae_croak} = $_[1];
		1981	$_[0]->send;
		1982	}
		1983
		1984	sub ready {
		1985	$_[0]{_ae_sent}
		1986	}
		1987
		1988	sub recv {
		1989	unless ($_[0]{_ae_sent}) {
		1990	$WAITING
		1991	and Carp::croak "AnyEvent::CondVar: recursive blocking wait attempted";
		1992
		1993	local $WAITING = 1;
		1994	$_[0]->_wait;
		1995	}
		1996
		1997	$_[0]{_ae_croak}
		1998	and Carp::croak $_[0]{_ae_croak};
		1999
		2000	wantarray
		2001	? @{ $_[0]{_ae_sent} }
		2002	: $_[0]{_ae_sent}[0]
		2003	}
		2004
		2005	sub cb {
		2006	my $cv = shift;
		2007
		2008	@_
		2009	and $cv->{_ae_cb} = shift
		2010	and $cv->{_ae_sent}
		2011	and (delete $cv->{_ae_cb})->($cv);
		2012
		2013	$cv->{_ae_cb}
		2014	}
		2015
		2016	sub begin {
		2017	++$_[0]{_ae_counter};
		2018	$_[0]{_ae_end_cb} = $_[1] if @_ > 1;
		2019	}
		2020
		2021	sub end {
		2022	return if --$_[0]{_ae_counter};
		2023	&{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
		2024	}
		2025
		2026	# undocumented/compatibility with pre-3.4
		2027	*broadcast = \&send;
		2028	*wait = \&recv;
		2029
		2030	=head1 ERROR AND EXCEPTION HANDLING
		2031
		2032	In general, AnyEvent does not do any error handling - it relies on the
		2033	caller to do that if required. The L<AnyEvent::Strict> module (see also
		2034	the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
		2035	checking of all AnyEvent methods, however, which is highly useful during
		2036	development.
		2037
		2038	As for exception handling (i.e. runtime errors and exceptions thrown while
		2039	executing a callback), this is not only highly event-loop specific, but
		2040	also not in any way wrapped by this module, as this is the job of the main
		2041	program.
		2042
		2043	The pure perl event loop simply re-throws the exception (usually
		2044	within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
		2045	$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
		2046	so on.
		2047
		2048	=head1 ENVIRONMENT VARIABLES
		2049
		2050	AnyEvent supports a number of environment variables that tune the
		2051	runtime behaviour. They are usually evaluated when AnyEvent is
		2052	loaded, initialised, or a submodule that uses them is loaded. Many of
		2053	them also cause AnyEvent to load additional modules - for example,
		2054	C<PERL_ANYEVENT_DEBUG_WRAP> causes the L<AnyEvent::Debug> module to be
		2055	loaded.
		2056
		2057	All the environment variables documented here start with
		2058	C<PERL_ANYEVENT_>, which is what AnyEvent considers its own
		2059	namespace. Other modules are encouraged (but by no means required) to use
		2060	C<PERL_ANYEVENT_SUBMODULE> if they have registered the AnyEvent::Submodule
		2061	namespace on CPAN, for any submodule. For example, L<AnyEvent::HTTP> could
		2062	be expected to use C<PERL_ANYEVENT_HTTP_PROXY> (it should not access env
		2063	variables starting with C<AE_>, see below).
		2064
		2065	All variables can also be set via the C<AE_> prefix, that is, instead
		2066	of setting C<PERL_ANYEVENT_VERBOSE> you can also set C<AE_VERBOSE>. In
		2067	case there is a clash btween anyevent and another program that uses
		2068	C<AE_something> you can set the corresponding C<PERL_ANYEVENT_something>
		2069	variable to the empty string, as those variables take precedence.
		2070
		2071	When AnyEvent is first loaded, it copies all C<AE_xxx> env variables
		2072	to their C<PERL_ANYEVENT_xxx> counterpart unless that variable already
		2073	exists. If taint mode is on, then AnyEvent will remove I<all> environment
		2074	variables starting with C<PERL_ANYEVENT_> from C<%ENV> (or replace them
		2075	with C<undef> or the empty string, if the corresaponding C<AE_> variable
		2076	is set).
		2077
		2078	The exact algorithm is currently:
		2079
		2080	1. if taint mode enabled, delete all PERL_ANYEVENT_xyz variables from %ENV
		2081	2. copy over AE_xyz to PERL_ANYEVENT_xyz unless the latter alraedy exists
		2082	3. if taint mode enabled, set all PERL_ANYEVENT_xyz variables to undef.
		2083
		2084	This ensures that child processes will not see the C<AE_> variables.
		2085
		2086	The following environment variables are currently known to AnyEvent:
		2087
		2088	=over 4
		2089
		2090	=item C<PERL_ANYEVENT_VERBOSE>
		2091
		2092	By default, AnyEvent will log messages with loglevel C<4> (C<error>) or
		2093	higher (see L<AnyEvent::Log>). You can set this environment variable to a
		2094	numerical loglevel to make AnyEvent more (or less) talkative.
		2095
		2096	If you want to do more than just set the global logging level
		2097	you should have a look at C<PERL_ANYEVENT_LOG>, which allows much more
		2098	complex specifications.
		2099
		2100	When set to C<0> (C<off>), then no messages whatsoever will be logged with
		2101	everything else at defaults.
		2102
		2103	When set to C<5> or higher (C<warn>), AnyEvent warns about unexpected
		2104	conditions, such as not being able to load the event model specified by
		2105	C<PERL_ANYEVENT_MODEL>, or a guard callback throwing an exception - this
		2106	is the minimum recommended level for use during development.
		2107
		2108	When set to C<7> or higher (info), AnyEvent reports which event model it
		2109	chooses.
		2110
		2111	When set to C<8> or higher (debug), then AnyEvent will report extra
		2112	information on which optional modules it loads and how it implements
		2113	certain features.
		2114
		2115	=item C<PERL_ANYEVENT_LOG>
		2116
		2117	Accepts rather complex logging specifications. For example, you could log
		2118	all C<debug> messages of some module to stderr, warnings and above to
		2119	stderr, and errors and above to syslog, with:
		2120
		2121	PERL_ANYEVENT_LOG=Some::Module=debug,+log:filter=warn,+%syslog:%syslog=error,syslog
		2122
		2123	For the rather extensive details, see L<AnyEvent::Log>.
		2124
		2125	This variable is evaluated when AnyEvent (or L<AnyEvent::Log>) is loaded,
		2126	so will take effect even before AnyEvent has initialised itself.
		2127
		2128	Note that specifying this environment variable causes the L<AnyEvent::Log>
		2129	module to be loaded, while C<PERL_ANYEVENT_VERBOSE> does not, so only
		2130	using the latter saves a few hundred kB of memory unless a module
		2131	explicitly needs the extra features of AnyEvent::Log.
		2132
		2133	=item C<PERL_ANYEVENT_STRICT>
		2134
		2135	AnyEvent does not do much argument checking by default, as thorough
		2136	argument checking is very costly. Setting this variable to a true value
		2137	will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
		2138	check the arguments passed to most method calls. If it finds any problems,
		2139	it will croak.
		2140
		2141	In other words, enables "strict" mode.
		2142
		2143	Unlike C<use strict> (or its modern cousin, C<< use L<common::sense>
		2144	>>, it is definitely recommended to keep it off in production. Keeping
		2145	C<PERL_ANYEVENT_STRICT=1> in your environment while developing programs
		2146	can be very useful, however.
		2147
		2148	=item C<PERL_ANYEVENT_DEBUG_SHELL>
		2149
		2150	If this env variable is nonempty, then its contents will be interpreted by
		2151	C<AnyEvent::Socket::parse_hostport> and C<AnyEvent::Debug::shell> (after
		2152	replacing every occurance of C<$$> by the process pid). The shell object
		2153	is saved in C<$AnyEvent::Debug::SHELL>.
		2154
		2155	This happens when the first watcher is created.
		2156
		2157	For example, to bind a debug shell on a unix domain socket in
		2158	F<< /tmp/debug<pid>.sock >>, you could use this:
		2159
		2160	PERL_ANYEVENT_DEBUG_SHELL=/tmp/debug\$\$.sock perlprog
		2161	# connect with e.g.: socat readline /tmp/debug123.sock
		2162
		2163	Or to bind to tcp port 4545 on localhost:
		2164
		2165	PERL_ANYEVENT_DEBUG_SHELL=127.0.0.1:4545 perlprog
		2166	# connect with e.g.: telnet localhost 4545
		2167
		2168	Note that creating sockets in F</tmp> or on localhost is very unsafe on
		2169	multiuser systems.
		2170
		2171	=item C<PERL_ANYEVENT_DEBUG_WRAP>
		2172
		2173	Can be set to C<0>, C<1> or C<2> and enables wrapping of all watchers for
		2174	debugging purposes. See C<AnyEvent::Debug::wrap> for details.
		2175
		2176	=item C<PERL_ANYEVENT_MODEL>
		2177
		2178	This can be used to specify the event model to be used by AnyEvent, before
		2179	auto detection and -probing kicks in.
		2180
		2181	It normally is a string consisting entirely of ASCII letters (e.g. C<EV>
		2182	or C<IOAsync>). The string C<AnyEvent::Impl::> gets prepended and the
		2183	resulting module name is loaded and - if the load was successful - used as
		2184	event model backend. If it fails to load then AnyEvent will proceed with
		2185	auto detection and -probing.
		2186
		2187	If the string ends with C<::> instead (e.g. C<AnyEvent::Impl::EV::>) then
		2188	nothing gets prepended and the module name is used as-is (hint: C<::> at
		2189	the end of a string designates a module name and quotes it appropriately).
		2190
		2191	For example, to force the pure perl model (L<AnyEvent::Loop::Perl>) you
		2192	could start your program like this:
		2193
		2194	PERL_ANYEVENT_MODEL=Perl perl ...
		2195
		2196	=item C<PERL_ANYEVENT_IO_MODEL>
		2197
		2198	The current file I/O model - see L<AnyEvent::IO> for more info.
		2199
		2200	At the moment, only C<Perl> (small, pure-perl, synchronous) and
		2201	C<IOAIO> (truly asynchronous) are supported. The default is C<IOAIO> if
		2202	L<AnyEvent::AIO> can be loaded, otherwise it is C<Perl>.
		2203
		2204	=item C<PERL_ANYEVENT_PROTOCOLS>
		2205
		2206	Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
		2207	for IPv4 or IPv6. The default is unspecified (and might change, or be the result
		2208	of auto probing).
		2209
		2210	Must be set to a comma-separated list of protocols or address families,
		2211	current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
		2212	used, and preference will be given to protocols mentioned earlier in the
		2213	list.
		2214
		2215	This variable can effectively be used for denial-of-service attacks
		2216	against local programs (e.g. when setuid), although the impact is likely
		2217	small, as the program has to handle connection and other failures anyways.
		2218
		2219	Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
		2220	but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
		2221	- only support IPv4, never try to resolve or contact IPv6
		2222	addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
		2223	IPv6, but prefer IPv6 over IPv4.
		2224
		2225	=item C<PERL_ANYEVENT_HOSTS>
		2226
		2227	This variable, if specified, overrides the F</etc/hosts> file used by
		2228	L<AnyEvent::Socket>C<::resolve_sockaddr>, i.e. hosts aliases will be read
		2229	from that file instead.
		2230
		2231	=item C<PERL_ANYEVENT_EDNS0>
		2232
		2233	Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension for
		2234	DNS. This extension is generally useful to reduce DNS traffic, especially
		2235	when DNSSEC is involved, but some (broken) firewalls drop such DNS
		2236	packets, which is why it is off by default.
		2237
		2238	Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
		2239	EDNS0 in its DNS requests.
		2240
		2241	=item C<PERL_ANYEVENT_MAX_FORKS>
		2242
		2243	The maximum number of child processes that C<AnyEvent::Util::fork_call>
		2244	will create in parallel.
		2245
		2246	=item C<PERL_ANYEVENT_MAX_OUTSTANDING_DNS>
		2247
		2248	The default value for the C<max_outstanding> parameter for the default DNS
		2249	resolver - this is the maximum number of parallel DNS requests that are
		2250	sent to the DNS server.
		2251
		2252	=item C<PERL_ANYEVENT_MAX_SIGNAL_LATENCY>
		2253
		2254	Perl has inherently racy signal handling (you can basically choose between
		2255	losing signals and memory corruption) - pure perl event loops (including
		2256	C<AnyEvent::Loop>, when C<Async::Interrupt> isn't available) therefore
		2257	have to poll regularly to avoid losing signals.
		2258
		2259	Some event loops are racy, but don't poll regularly, and some event loops
		2260	are written in C but are still racy. For those event loops, AnyEvent
		2261	installs a timer that regularly wakes up the event loop.
		2262
		2263	By default, the interval for this timer is C<10> seconds, but you can
		2264	override this delay with this environment variable (or by setting
		2265	the C<$AnyEvent::MAX_SIGNAL_LATENCY> variable before creating signal
		2266	watchers).
		2267
		2268	Lower values increase CPU (and energy) usage, higher values can introduce
		2269	long delays when reaping children or waiting for signals.
		2270
		2271	The L<AnyEvent::Async> module, if available, will be used to avoid this
		2272	polling (with most event loops).
		2273
		2274	=item C<PERL_ANYEVENT_RESOLV_CONF>
		2275
		2276	The absolute path to a F<resolv.conf>-style file to use instead of
		2277	F</etc/resolv.conf> (or the OS-specific configuration) in the default
		2278	resolver, or the empty string to select the default configuration.
		2279
		2280	=item C<PERL_ANYEVENT_CA_FILE>, C<PERL_ANYEVENT_CA_PATH>.
		2281
		2282	When neither C<ca_file> nor C<ca_path> was specified during
		2283	L<AnyEvent::TLS> context creation, and either of these environment
		2284	variables are nonempty, they will be used to specify CA certificate
		2285	locations instead of a system-dependent default.
		2286
		2287	=item C<PERL_ANYEVENT_AVOID_GUARD> and C<PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT>
		2288
		2289	When these are set to C<1>, then the respective modules are not
		2290	loaded. Mostly good for testing AnyEvent itself.
		2291
		2292	=back
		2293
		2294	=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
		2295
		2296	This is an advanced topic that you do not normally need to use AnyEvent in
		2297	a module. This section is only of use to event loop authors who want to
		2298	provide AnyEvent compatibility.
		2299
		2300	If you need to support another event library which isn't directly
		2301	supported by AnyEvent, you can supply your own interface to it by
		2302	pushing, before the first watcher gets created, the package name of
		2303	the event module and the package name of the interface to use onto
		2304	C<@AnyEvent::REGISTRY>. You can do that before and even without loading
		2305	AnyEvent, so it is reasonably cheap.
		2306
		2307	Example:
		2308
		2309	push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::];
		2310
		2311	This tells AnyEvent to (literally) use the C<urxvt::anyevent::>
		2312	package/class when it finds the C<urxvt> package/module is already loaded.
		2313
		2314	When AnyEvent is loaded and asked to find a suitable event model, it
		2315	will first check for the presence of urxvt by trying to C<use> the
		2316	C<urxvt::anyevent> module.
		2317
		2318	The class should provide implementations for all watcher types. See
		2319	L<AnyEvent::Impl::EV> (source code), L<AnyEvent::Impl::Glib> (Source code)
		2320	and so on for actual examples. Use C<perldoc -m AnyEvent::Impl::Glib> to
		2321	see the sources.
		2322
		2323	If you don't provide C<signal> and C<child> watchers than AnyEvent will
		2324	provide suitable (hopefully) replacements.
		2325
		2326	The above example isn't fictitious, the I<rxvt-unicode> (a.k.a. urxvt)
		2327	terminal emulator uses the above line as-is. An interface isn't included
		2328	in AnyEvent because it doesn't make sense outside the embedded interpreter
		2329	inside I<rxvt-unicode>, and it is updated and maintained as part of the
		2330	I<rxvt-unicode> distribution.
		2331
		2332	I<rxvt-unicode> also cheats a bit by not providing blocking access to
		2333	condition variables: code blocking while waiting for a condition will
		2334	C<die>. This still works with most modules/usages, and blocking calls must
		2335	not be done in an interactive application, so it makes sense.
		2336
		2337	=head1 EXAMPLE PROGRAM
		2338
		2339	The following program uses an I/O watcher to read data from STDIN, a timer
		2340	to display a message once per second, and a condition variable to quit the
		2341	program when the user enters quit:
		2342
		2343	use AnyEvent;
		2344
		2345	my $cv = AnyEvent->condvar;
		2346
		2347	my $io_watcher = AnyEvent->io (
		2348	fh => \*STDIN,
		2349	poll => 'r',
		2350	cb => sub {
		2351	warn "io event <$_[0]>\n"; # will always output <r>
		2352	chomp (my $input = <STDIN>); # read a line
		2353	warn "read: $input\n"; # output what has been read
		2354	$cv->send if $input =~ /^q/i; # quit program if /^q/i
		2355	},
		2356	);
		2357
		2358	my $time_watcher = AnyEvent->timer (after => 1, interval => 1, cb => sub {
		2359	warn "timeout\n"; # print 'timeout' at most every second
		2360	});
		2361
		2362	$cv->recv; # wait until user enters /^q/i
		2363
		2364	=head1 REAL-WORLD EXAMPLE
		2365
		2366	Consider the L<Net::FCP> module. It features (among others) the following
		2367	API calls, which are to freenet what HTTP GET requests are to http:
		2368
		2369	my $data = $fcp->client_get ($url); # blocks
		2370
		2371	my $transaction = $fcp->txn_client_get ($url); # does not block
		2372	$transaction->cb ( sub { ... } ); # set optional result callback
		2373	my $data = $transaction->result; # possibly blocks
		2374
		2375	The C<client_get> method works like C<LWP::Simple::get>: it requests the
		2376	given URL and waits till the data has arrived. It is defined to be:
		2377
		2378	sub client_get { $_[0]->txn_client_get ($_[1])->result }
		2379
		2380	And in fact is automatically generated. This is the blocking API of
		2381	L<Net::FCP>, and it works as simple as in any other, similar, module.
		2382
		2383	More complicated is C<txn_client_get>: It only creates a transaction
		2384	(completion, result, ...) object and initiates the transaction.
		2385
		2386	my $txn = bless { }, Net::FCP::Txn::;
		2387
		2388	It also creates a condition variable that is used to signal the completion
		2389	of the request:
		2390
		2391	$txn->{finished} = AnyAvent->condvar;
		2392
		2393	It then creates a socket in non-blocking mode.
		2394
		2395	socket $txn->{fh}, ...;
		2396	fcntl $txn->{fh}, F_SETFL, O_NONBLOCK;
		2397	connect $txn->{fh}, ...
		2398	and !$!{EWOULDBLOCK}
		2399	and !$!{EINPROGRESS}
		2400	and Carp::croak "unable to connect: $!\n";
		2401
		2402	Then it creates a write-watcher which gets called whenever an error occurs
		2403	or the connection succeeds:
		2404
		2405	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'w', cb => sub { $txn->fh_ready_w });
		2406
		2407	And returns this transaction object. The C<fh_ready_w> callback gets
		2408	called as soon as the event loop detects that the socket is ready for
		2409	writing.
		2410
		2411	The C<fh_ready_w> method makes the socket blocking again, writes the
		2412	request data and replaces the watcher by a read watcher (waiting for reply
		2413	data). The actual code is more complicated, but that doesn't matter for
		2414	this example:
		2415
		2416	fcntl $txn->{fh}, F_SETFL, 0;
		2417	syswrite $txn->{fh}, $txn->{request}
		2418	or die "connection or write error";
		2419	$txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
		2420
		2421	Again, C<fh_ready_r> waits till all data has arrived, and then stores the
		2422	result and signals any possible waiters that the request has finished:
		2423
		2424	sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
		2425
		2426	if (end-of-file or data complete) {
		2427	$txn->{result} = $txn->{buf};
		2428	$txn->{finished}->send;
		2429	$txb->{cb}->($txn) of $txn->{cb}; # also call callback
		2430	}
		2431
		2432	The C<result> method, finally, just waits for the finished signal (if the
		2433	request was already finished, it doesn't wait, of course, and returns the
		2434	data:
		2435
		2436	$txn->{finished}->recv;
		2437	return $txn->{result};
		2438
		2439	The actual code goes further and collects all errors (C<die>s, exceptions)
		2440	that occurred during request processing. The C<result> method detects
		2441	whether an exception as thrown (it is stored inside the $txn object)
		2442	and just throws the exception, which means connection errors and other
		2443	problems get reported to the code that tries to use the result, not in a
		2444	random callback.
		2445
		2446	All of this enables the following usage styles:
		2447
		2448	1. Blocking:
		2449
		2450	my $data = $fcp->client_get ($url);
		2451
		2452	2. Blocking, but running in parallel:
		2453
		2454	my @datas = map $_->result,
		2455	map $fcp->txn_client_get ($_),
		2456	@urls;
		2457
		2458	Both blocking examples work without the module user having to know
		2459	anything about events.
		2460
		2461	3a. Event-based in a main program, using any supported event module:
		2462
		2463	use EV;
		2464
		2465	$fcp->txn_client_get ($url)->cb (sub {
		2466	my $txn = shift;
		2467	my $data = $txn->result;
		2468	...
		2469	});
		2470
		2471	EV::loop;
		2472
		2473	3b. The module user could use AnyEvent, too:
		2474
		2475	use AnyEvent;
		2476
		2477	my $quit = AnyEvent->condvar;
		2478
		2479	$fcp->txn_client_get ($url)->cb (sub {
		2480	...
		2481	$quit->send;
		2482	});
		2483
		2484	$quit->recv;
		2485
		2486
		2487	=head1 BENCHMARKS
		2488
		2489	To give you an idea of the performance and overheads that AnyEvent adds
		2490	over the event loops themselves and to give you an impression of the speed
		2491	of various event loops I prepared some benchmarks.
		2492
		2493	=head2 BENCHMARKING ANYEVENT OVERHEAD
		2494
		2495	Here is a benchmark of various supported event models used natively and
		2496	through AnyEvent. The benchmark creates a lot of timers (with a zero
		2497	timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
		2498	which it is), lets them fire exactly once and destroys them again.
		2499
		2500	Source code for this benchmark is found as F<eg/bench> in the AnyEvent
		2501	distribution. It uses the L<AE> interface, which makes a real difference
		2502	for the EV and Perl backends only.
		2503
		2504	=head3 Explanation of the columns
		2505
		2506	I<watcher> is the number of event watchers created/destroyed. Since
		2507	different event models feature vastly different performances, each event
		2508	loop was given a number of watchers so that overall runtime is acceptable
		2509	and similar between tested event loop (and keep them from crashing): Glib
		2510	would probably take thousands of years if asked to process the same number
		2511	of watchers as EV in this benchmark.
		2512
		2513	I<bytes> is the number of bytes (as measured by the resident set size,
		2514	RSS) consumed by each watcher. This method of measuring captures both C
		2515	and Perl-based overheads.
		2516
		2517	I<create> is the time, in microseconds (millionths of seconds), that it
		2518	takes to create a single watcher. The callback is a closure shared between
		2519	all watchers, to avoid adding memory overhead. That means closure creation
		2520	and memory usage is not included in the figures.
		2521
		2522	I<invoke> is the time, in microseconds, used to invoke a simple
		2523	callback. The callback simply counts down a Perl variable and after it was
		2524	invoked "watcher" times, it would C<< ->send >> a condvar once to
		2525	signal the end of this phase.
		2526
		2527	I<destroy> is the time, in microseconds, that it takes to destroy a single
		2528	watcher.
		2529
		2530	=head3 Results
		2531
		2532	name watchers bytes create invoke destroy comment
		2533	EV/EV 100000 223 0.47 0.43 0.27 EV native interface
		2534	EV/Any 100000 223 0.48 0.42 0.26 EV + AnyEvent watchers
		2535	Coro::EV/Any 100000 223 0.47 0.42 0.26 coroutines + Coro::Signal
		2536	Perl/Any 100000 431 2.70 0.74 0.92 pure perl implementation
		2537	Event/Event 16000 516 31.16 31.84 0.82 Event native interface
		2538	Event/Any 16000 1203 42.61 34.79 1.80 Event + AnyEvent watchers
		2539	IOAsync/Any 16000 1911 41.92 27.45 16.81 via IO::Async::Loop::IO_Poll
		2540	IOAsync/Any 16000 1726 40.69 26.37 15.25 via IO::Async::Loop::Epoll
		2541	Glib/Any 16000 1118 89.00 12.57 51.17 quadratic behaviour
		2542	Tk/Any 2000 1346 20.96 10.75 8.00 SEGV with >> 2000 watchers
		2543	POE/Any 2000 6951 108.97 795.32 14.24 via POE::Loop::Event
		2544	POE/Any 2000 6648 94.79 774.40 575.51 via POE::Loop::Select
		2545
		2546	=head3 Discussion
		2547
		2548	The benchmark does I<not> measure scalability of the event loop very
		2549	well. For example, a select-based event loop (such as the pure perl one)
		2550	can never compete with an event loop that uses epoll when the number of
		2551	file descriptors grows high. In this benchmark, all events become ready at
		2552	the same time, so select/poll-based implementations get an unnatural speed
		2553	boost.
		2554
		2555	Also, note that the number of watchers usually has a nonlinear effect on
		2556	overall speed, that is, creating twice as many watchers doesn't take twice
		2557	the time - usually it takes longer. This puts event loops tested with a
		2558	higher number of watchers at a disadvantage.
		2559
		2560	To put the range of results into perspective, consider that on the
		2561	benchmark machine, handling an event takes roughly 1600 CPU cycles with
		2562	EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 CPU
		2563	cycles with POE.
		2564
		2565	C<EV> is the sole leader regarding speed and memory use, which are both
		2566	maximal/minimal, respectively. When using the L<AE> API there is zero
		2567	overhead (when going through the AnyEvent API create is about 5-6 times
		2568	slower, with other times being equal, so still uses far less memory than
		2569	any other event loop and is still faster than Event natively).
		2570
		2571	The pure perl implementation is hit in a few sweet spots (both the
		2572	constant timeout and the use of a single fd hit optimisations in the perl
		2573	interpreter and the backend itself). Nevertheless this shows that it
		2574	adds very little overhead in itself. Like any select-based backend its
		2575	performance becomes really bad with lots of file descriptors (and few of
		2576	them active), of course, but this was not subject of this benchmark.
		2577
		2578	The C<Event> module has a relatively high setup and callback invocation
		2579	cost, but overall scores in on the third place.
		2580
		2581	C<IO::Async> performs admirably well, about on par with C<Event>, even
		2582	when using its pure perl backend.
		2583
		2584	C<Glib>'s memory usage is quite a bit higher, but it features a
		2585	faster callback invocation and overall ends up in the same class as
		2586	C<Event>. However, Glib scales extremely badly, doubling the number of
		2587	watchers increases the processing time by more than a factor of four,
		2588	making it completely unusable when using larger numbers of watchers
		2589	(note that only a single file descriptor was used in the benchmark, so
		2590	inefficiencies of C<poll> do not account for this).
		2591
		2592	The C<Tk> adaptor works relatively well. The fact that it crashes with
		2593	more than 2000 watchers is a big setback, however, as correctness takes
		2594	precedence over speed. Nevertheless, its performance is surprising, as the
		2595	file descriptor is dup()ed for each watcher. This shows that the dup()
		2596	employed by some adaptors is not a big performance issue (it does incur a
		2597	hidden memory cost inside the kernel which is not reflected in the figures
		2598	above).
		2599
		2600	C<POE>, regardless of underlying event loop (whether using its pure perl
		2601	select-based backend or the Event module, the POE-EV backend couldn't
		2602	be tested because it wasn't working) shows abysmal performance and
		2603	memory usage with AnyEvent: Watchers use almost 30 times as much memory
		2604	as EV watchers, and 10 times as much memory as Event (the high memory
		2605	requirements are caused by requiring a session for each watcher). Watcher
		2606	invocation speed is almost 900 times slower than with AnyEvent's pure perl
		2607	implementation.
		2608
		2609	The design of the POE adaptor class in AnyEvent can not really account
		2610	for the performance issues, though, as session creation overhead is
		2611	small compared to execution of the state machine, which is coded pretty
		2612	optimally within L<AnyEvent::Impl::POE> (and while everybody agrees that
		2613	using multiple sessions is not a good approach, especially regarding
		2614	memory usage, even the author of POE could not come up with a faster
		2615	design).
		2616
		2617	=head3 Summary
		2618
		2619	=over 4
		2620
		2621	=item * Using EV through AnyEvent is faster than any other event loop
		2622	(even when used without AnyEvent), but most event loops have acceptable
		2623	performance with or without AnyEvent.
		2624
		2625	=item * The overhead AnyEvent adds is usually much smaller than the overhead of
		2626	the actual event loop, only with extremely fast event loops such as EV
		2627	does AnyEvent add significant overhead.
		2628
		2629	=item * You should avoid POE like the plague if you want performance or
		2630	reasonable memory usage.
		2631
		2632	=back
		2633
		2634	=head2 BENCHMARKING THE LARGE SERVER CASE
		2635
		2636	This benchmark actually benchmarks the event loop itself. It works by
		2637	creating a number of "servers": each server consists of a socket pair, a
		2638	timeout watcher that gets reset on activity (but never fires), and an I/O
		2639	watcher waiting for input on one side of the socket. Each time the socket
		2640	watcher reads a byte it will write that byte to a random other "server".
		2641
		2642	The effect is that there will be a lot of I/O watchers, only part of which
		2643	are active at any one point (so there is a constant number of active
		2644	fds for each loop iteration, but which fds these are is random). The
		2645	timeout is reset each time something is read because that reflects how
		2646	most timeouts work (and puts extra pressure on the event loops).
		2647
		2648	In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
		2649	(1%) are active. This mirrors the activity of large servers with many
		2650	connections, most of which are idle at any one point in time.
		2651
		2652	Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
		2653	distribution. It uses the L<AE> interface, which makes a real difference
		2654	for the EV and Perl backends only.
		2655
		2656	=head3 Explanation of the columns
		2657
		2658	I<sockets> is the number of sockets, and twice the number of "servers" (as
		2659	each server has a read and write socket end).
		2660
		2661	I<create> is the time it takes to create a socket pair (which is
		2662	nontrivial) and two watchers: an I/O watcher and a timeout watcher.
		2663
		2664	I<request>, the most important value, is the time it takes to handle a
		2665	single "request", that is, reading the token from the pipe and forwarding
		2666	it to another server. This includes deleting the old timeout and creating
		2667	a new one that moves the timeout into the future.
		2668
		2669	=head3 Results
		2670
		2671	name sockets create request
		2672	EV 20000 62.66 7.99
		2673	Perl 20000 68.32 32.64
		2674	IOAsync 20000 174.06 101.15 epoll
		2675	IOAsync 20000 174.67 610.84 poll
		2676	Event 20000 202.69 242.91
		2677	Glib 20000 557.01 1689.52
		2678	POE 20000 341.54 12086.32 uses POE::Loop::Event
		2679
		2680	=head3 Discussion
		2681
		2682	This benchmark I<does> measure scalability and overall performance of the
		2683	particular event loop.
		2684
		2685	EV is again fastest. Since it is using epoll on my system, the setup time
		2686	is relatively high, though.
		2687
		2688	Perl surprisingly comes second. It is much faster than the C-based event
		2689	loops Event and Glib.
		2690
		2691	IO::Async performs very well when using its epoll backend, and still quite
		2692	good compared to Glib when using its pure perl backend.
		2693
		2694	Event suffers from high setup time as well (look at its code and you will
		2695	understand why). Callback invocation also has a high overhead compared to
		2696	the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
		2697	uses select or poll in basically all documented configurations.
		2698
		2699	Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
		2700	clearly fails to perform with many filehandles or in busy servers.
		2701
		2702	POE is still completely out of the picture, taking over 1000 times as long
		2703	as EV, and over 100 times as long as the Perl implementation, even though
		2704	it uses a C-based event loop in this case.
		2705
		2706	=head3 Summary
		2707
		2708	=over 4
		2709
		2710	=item * The pure perl implementation performs extremely well.
		2711
		2712	=item * Avoid Glib or POE in large projects where performance matters.
		2713
		2714	=back
		2715
		2716	=head2 BENCHMARKING SMALL SERVERS
		2717
		2718	While event loops should scale (and select-based ones do not...) even to
		2719	large servers, most programs we (or I :) actually write have only a few
		2720	I/O watchers.
		2721
		2722	In this benchmark, I use the same benchmark program as in the large server
		2723	case, but it uses only eight "servers", of which three are active at any
		2724	one time. This should reflect performance for a small server relatively
		2725	well.
		2726
		2727	The columns are identical to the previous table.
		2728
		2729	=head3 Results
		2730
		2731	name sockets create request
		2732	EV 16 20.00 6.54
		2733	Perl 16 25.75 12.62
		2734	Event 16 81.27 35.86
		2735	Glib 16 32.63 15.48
		2736	POE 16 261.87 276.28 uses POE::Loop::Event
		2737
		2738	=head3 Discussion
		2739
		2740	The benchmark tries to test the performance of a typical small
		2741	server. While knowing how various event loops perform is interesting, keep
		2742	in mind that their overhead in this case is usually not as important, due
		2743	to the small absolute number of watchers (that is, you need efficiency and
		2744	speed most when you have lots of watchers, not when you only have a few of
		2745	them).
		2746
		2747	EV is again fastest.
		2748
		2749	Perl again comes second. It is noticeably faster than the C-based event
		2750	loops Event and Glib, although the difference is too small to really
		2751	matter.
		2752
		2753	POE also performs much better in this case, but is is still far behind the
		2754	others.
		2755
		2756	=head3 Summary
		2757
		2758	=over 4
		2759
		2760	=item * C-based event loops perform very well with small number of
		2761	watchers, as the management overhead dominates.
		2762
		2763	=back
		2764
		2765	=head2 THE IO::Lambda BENCHMARK
		2766
		2767	Recently I was told about the benchmark in the IO::Lambda manpage, which
		2768	could be misinterpreted to make AnyEvent look bad. In fact, the benchmark
		2769	simply compares IO::Lambda with POE, and IO::Lambda looks better (which
		2770	shouldn't come as a surprise to anybody). As such, the benchmark is
		2771	fine, and mostly shows that the AnyEvent backend from IO::Lambda isn't
		2772	very optimal. But how would AnyEvent compare when used without the extra
		2773	baggage? To explore this, I wrote the equivalent benchmark for AnyEvent.
		2774
		2775	The benchmark itself creates an echo-server, and then, for 500 times,
		2776	connects to the echo server, sends a line, waits for the reply, and then
		2777	creates the next connection. This is a rather bad benchmark, as it doesn't
		2778	test the efficiency of the framework or much non-blocking I/O, but it is a
		2779	benchmark nevertheless.
		2780
		2781	name runtime
		2782	Lambda/select 0.330 sec
		2783	+ optimized 0.122 sec
		2784	Lambda/AnyEvent 0.327 sec
		2785	+ optimized 0.138 sec
		2786	Raw sockets/select 0.077 sec
		2787	POE/select, components 0.662 sec
		2788	POE/select, raw sockets 0.226 sec
		2789	POE/select, optimized 0.404 sec
		2790
		2791	AnyEvent/select/nb 0.085 sec
		2792	AnyEvent/EV/nb 0.068 sec
		2793	+state machine 0.134 sec
		2794
		2795	The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
		2796	benchmarks actually make blocking connects and use 100% blocking I/O,
		2797	defeating the purpose of an event-based solution. All of the newly
		2798	written AnyEvent benchmarks use 100% non-blocking connects (using
		2799	AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
		2800	resolver), so AnyEvent is at a disadvantage here, as non-blocking connects
		2801	generally require a lot more bookkeeping and event handling than blocking
		2802	connects (which involve a single syscall only).
		2803
		2804	The last AnyEvent benchmark additionally uses L<AnyEvent::Handle>, which
		2805	offers similar expressive power as POE and IO::Lambda, using conventional
		2806	Perl syntax. This means that both the echo server and the client are 100%
		2807	non-blocking, further placing it at a disadvantage.
		2808
		2809	As you can see, the AnyEvent + EV combination even beats the
		2810	hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
		2811	backend easily beats IO::Lambda and POE.
		2812
		2813	And even the 100% non-blocking version written using the high-level (and
		2814	slow :) L<AnyEvent::Handle> abstraction beats both POE and IO::Lambda
		2815	higher level ("unoptimised") abstractions by a large margin, even though
		2816	it does all of DNS, tcp-connect and socket I/O in a non-blocking way.
		2817
		2818	The two AnyEvent benchmarks programs can be found as F<eg/ae0.pl> and
		2819	F<eg/ae2.pl> in the AnyEvent distribution, the remaining benchmarks are
		2820	part of the IO::Lambda distribution and were used without any changes.
		2821
		2822
		2823	=head1 SIGNALS
		2824
		2825	AnyEvent currently installs handlers for these signals:
		2826
		2827	=over 4
		2828
		2829	=item SIGCHLD
		2830
		2831	A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
		2832	emulation for event loops that do not support them natively. Also, some
		2833	event loops install a similar handler.
		2834
		2835	Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, then
		2836	AnyEvent will reset it to default, to avoid losing child exit statuses.
		2837
		2838	=item SIGPIPE
		2839
		2840	A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
		2841	when AnyEvent gets loaded.
		2842
		2843	The rationale for this is that AnyEvent users usually do not really depend
		2844	on SIGPIPE delivery (which is purely an optimisation for shell use, or
		2845	badly-written programs), but C<SIGPIPE> can cause spurious and rare
		2846	program exits as a lot of people do not expect C<SIGPIPE> when writing to
		2847	some random socket.
		2848
		2849	The rationale for installing a no-op handler as opposed to ignoring it is
		2850	that this way, the handler will be restored to defaults on exec.
		2851
		2852	Feel free to install your own handler, or reset it to defaults.
		2853
		2854	=back
		2855
		2856	=cut
		2857
		2858	undef $SIG{CHLD}
		2859	if $SIG{CHLD} eq 'IGNORE';
		2860
		2861	$SIG{PIPE} = sub { }
		2862	unless defined $SIG{PIPE};
		2863
		2864	=head1 RECOMMENDED/OPTIONAL MODULES
		2865
		2866	One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
		2867	its built-in modules) are required to use it.
		2868
		2869	That does not mean that AnyEvent won't take advantage of some additional
		2870	modules if they are installed.
		2871
		2872	This section explains which additional modules will be used, and how they
		2873	affect AnyEvent's operation.
		2874
		2875	=over 4
		2876
		2877	=item L<Async::Interrupt>
		2878
		2879	This slightly arcane module is used to implement fast signal handling: To
		2880	my knowledge, there is no way to do completely race-free and quick
		2881	signal handling in pure perl. To ensure that signals still get
		2882	delivered, AnyEvent will start an interval timer to wake up perl (and
		2883	catch the signals) with some delay (default is 10 seconds, look for
		2884	C<$AnyEvent::MAX_SIGNAL_LATENCY>).
		2885
		2886	If this module is available, then it will be used to implement signal
		2887	catching, which means that signals will not be delayed, and the event loop
		2888	will not be interrupted regularly, which is more efficient (and good for
		2889	battery life on laptops).
		2890
		2891	This affects not just the pure-perl event loop, but also other event loops
		2892	that have no signal handling on their own (e.g. Glib, Tk, Qt).
		2893
		2894	Some event loops (POE, Event, Event::Lib) offer signal watchers natively,
		2895	and either employ their own workarounds (POE) or use AnyEvent's workaround
		2896	(using C<$AnyEvent::MAX_SIGNAL_LATENCY>). Installing L<Async::Interrupt>
		2897	does nothing for those backends.
		2898
		2899	=item L<EV>
		2900
		2901	This module isn't really "optional", as it is simply one of the backend
		2902	event loops that AnyEvent can use. However, it is simply the best event
		2903	loop available in terms of features, speed and stability: It supports
		2904	the AnyEvent API optimally, implements all the watcher types in XS, does
		2905	automatic timer adjustments even when no monotonic clock is available,
		2906	can take avdantage of advanced kernel interfaces such as C<epoll> and
		2907	C<kqueue>, and is the fastest backend I<by far>. You can even embed
		2908	L<Glib>/L<Gtk2> in it (or vice versa, see L<EV::Glib> and L<Glib::EV>).
		2909
		2910	If you only use backends that rely on another event loop (e.g. C<Tk>),
		2911	then this module will do nothing for you.
		2912
		2913	=item L<Guard>
		2914
		2915	The guard module, when used, will be used to implement
		2916	C<AnyEvent::Util::guard>. This speeds up guards considerably (and uses a
		2917	lot less memory), but otherwise doesn't affect guard operation much. It is
		2918	purely used for performance.
		2919
		2920	=item L<JSON> and L<JSON::XS>
		2921
		2922	One of these modules is required when you want to read or write JSON data
		2923	via L<AnyEvent::Handle>. L<JSON> is also written in pure-perl, but can take
		2924	advantage of the ultra-high-speed L<JSON::XS> module when it is installed.
		2925
		2926	=item L<Net::SSLeay>
		2927
		2928	Implementing TLS/SSL in Perl is certainly interesting, but not very
		2929	worthwhile: If this module is installed, then L<AnyEvent::Handle> (with
		2930	the help of L<AnyEvent::TLS>), gains the ability to do TLS/SSL.
		2931
		2932	=item L<Time::HiRes>
		2933
		2934	This module is part of perl since release 5.008. It will be used when the
		2935	chosen event library does not come with a timing source of its own. The
		2936	pure-perl event loop (L<AnyEvent::Loop>) will additionally load it to
		2937	try to use a monotonic clock for timing stability.
		2938
		2939	=item L<AnyEvent::AIO> (and L<IO::AIO>)
		2940
		2941	The default implementation of L<AnyEvent::IO> is to do I/O synchronously,
		2942	stopping programs while they access the disk, which is fine for a lot of
		2943	programs.
		2944
		2945	Installing AnyEvent::AIO (and its IO::AIO dependency) makes it switch to
		2946	a true asynchronous implementation, so event processing can continue even
		2947	while waiting for disk I/O.
		2948
		2949	=back
		2950
		2951
		2952	=head1 FORK
		2953
		2954	Most event libraries are not fork-safe. The ones who are usually are
		2955	because they rely on inefficient but fork-safe C<select> or C<poll> calls
		2956	- higher performance APIs such as BSD's kqueue or the dreaded Linux epoll
		2957	are usually badly thought-out hacks that are incompatible with fork in
		2958	one way or another. Only L<EV> is fully fork-aware and ensures that you
		2959	continue event-processing in both parent and child (or both, if you know
		2960	what you are doing).
		2961
		2962	This means that, in general, you cannot fork and do event processing in
		2963	the child if the event library was initialised before the fork (which
		2964	usually happens when the first AnyEvent watcher is created, or the library
		2965	is loaded).
		2966
		2967	If you have to fork, you must either do so I<before> creating your first
		2968	watcher OR you must not use AnyEvent at all in the child OR you must do
		2969	something completely out of the scope of AnyEvent (see below).
		2970
		2971	The problem of doing event processing in the parent I<and> the child
		2972	is much more complicated: even for backends that I<are> fork-aware or
		2973	fork-safe, their behaviour is not usually what you want: fork clones all
		2974	watchers, that means all timers, I/O watchers etc. are active in both
		2975	parent and child, which is almost never what you want. Using C<exec>
		2976	to start worker children from some kind of manage prrocess is usually
		2977	preferred, because it is much easier and cleaner, at the expense of having
		2978	to have another binary.
		2979
		2980	In addition to logical problems with fork, there are also implementation
		2981	problems. For example, on POSIX systems, you cannot fork at all in Perl
		2982	code if a thread (I am talking of pthreads here) was ever created in the
		2983	process, and this is just the tip of the iceberg. In general, using fork
		2984	from Perl is difficult, and attempting to use fork without an exec to
		2985	implement some kind of parallel processing is almost certainly doomed.
		2986
		2987	To safely fork and exec, you should use a module such as
		2988	L<Proc::FastSpawn> that let's you safely fork and exec new processes.
		2989
		2990	If you want to do multiprocessing using processes, you can
		2991	look at the L<AnyEvent::Fork> module (and some related modules
		2992	such as L<AnyEvent::Fork::RPC>, L<AnyEvent::Fork::Pool> and
		2993	L<AnyEvent::Fork::Remote>). This module allows you to safely create
		2994	subprocesses without any limitations - you can use X11 toolkits or
		2995	AnyEvent in the children created by L<AnyEvent::Fork> safely and without
		2996	any special precautions.
		2997
		2998
		2999	=head1 SECURITY CONSIDERATIONS
		3000
		3001	AnyEvent can be forced to load any event model via
		3002	$ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used to
		3003	execute arbitrary code or directly gain access, it can easily be used to
		3004	make the program hang or malfunction in subtle ways, as AnyEvent watchers
		3005	will not be active when the program uses a different event model than
		3006	specified in the variable.
		3007
		3008	You can make AnyEvent completely ignore this variable by deleting it
		3009	before the first watcher gets created, e.g. with a C<BEGIN> block:
		3010
		3011	BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
		3012
		3013	use AnyEvent;
		3014
		3015	Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
		3016	be used to probe what backend is used and gain other information (which is
		3017	probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
		3018	$ENV{PERL_ANYEVENT_STRICT}.
		3019
		3020	Note that AnyEvent will remove I<all> environment variables starting with
		3021	C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
		3022	enabled.
		3023
		3024
		3025	=head1 BUGS
		3026
		3027	Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
		3028	to work around. If you suffer from memleaks, first upgrade to Perl 5.10
		3029	and check wether the leaks still show up. (Perl 5.10.0 has other annoying
		3030	memleaks, such as leaking on C<map> and C<grep> but it is usually not as
		3031	pronounced).
		3032
		3033
		3034	=head1 SEE ALSO
		3035
		3036	Tutorial/Introduction: L<AnyEvent::Intro>.
		3037
		3038	FAQ: L<AnyEvent::FAQ>.
		3039
		3040	Utility functions: L<AnyEvent::Util> (misc. grab-bag), L<AnyEvent::Log>
		3041	(simply logging).
		3042
		3043	Development/Debugging: L<AnyEvent::Strict> (stricter checking),
		3044	L<AnyEvent::Debug> (interactive shell, watcher tracing).
		3045
		3046	Supported event modules: L<AnyEvent::Loop>, L<EV>, L<EV::Glib>,
		3047	L<Glib::EV>, L<Event>, L<Glib::Event>, L<Glib>, L<Tk>, L<Event::Lib>,
		3048	L<Qt>, L<POE>, L<FLTK>.
		3049
		3050	Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
		3051	L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
		3052	L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
		3053	L<AnyEvent::Impl::POE>, L<AnyEvent::Impl::IOAsync>, L<Anyevent::Impl::Irssi>,
		3054	L<AnyEvent::Impl::FLTK>.
		3055
		3056	Non-blocking handles, pipes, stream sockets, TCP clients and
		3057	servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>, L<AnyEvent::TLS>.
		3058
		3059	Asynchronous File I/O: L<AnyEvent::IO>.
		3060
		3061	Asynchronous DNS: L<AnyEvent::DNS>.
		3062
		3063	Thread support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
		3064
		3065	Nontrivial usage examples: L<AnyEvent::GPSD>, L<AnyEvent::IRC>,
		3066	L<AnyEvent::HTTP>.
		3067
		3068
		3069	=head1 AUTHOR
		3070
		3071	Marc Lehmann <schmorp@schmorp.de>
		3072	http://anyevent.schmorp.de
		3073
		3074	=cut
		3075
		3076	1
		3077

Diff Legend

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