[ViewVC] Diff of: cvs/Coro/Coro.pm

Comparing Coro/Coro.pm (file contents):
Revision 1.11 by root, Sun Jul 15 03:24:18 2001 UTC vs.
Revision 1.274 by root, Sat Dec 12 01:30:26 2009 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	Coro - coroutine process abstraction	3	Coro - the only real threads in perl
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use Coro;	7	use Coro;
8		8
9	async {	9	async {
10	# some asynchronous thread of execution	10	# some asynchronous thread of execution
		11	print "2\n";
		12	cede; # yield back to main
		13	print "4\n";
11	};	14	};
12		15	print "1\n";
13	# alternatively create an async process like this:	16	cede; # yield to coro
14		17	print "3\n";
15	sub some_func : Coro {	18	cede; # and again
16	# some more async code	19
17	}	20	# use locking
18		21	use Coro::Semaphore;
19	yield;	22	my $lock = new Coro::Semaphore;
		23	my $locked;
		24
		25	$lock->down;
		26	$locked = 1;
		27	$lock->up;
20		28
21	=head1 DESCRIPTION	29	=head1 DESCRIPTION
22		30
		31	For a tutorial-style introduction, please read the L<Coro::Intro>
		32	manpage. This manpage mainly contains reference information.
		33
		34	This module collection manages continuations in general, most often in
		35	the form of cooperative threads (also called coros, or simply "coro"
		36	in the documentation). They are similar to kernel threads but don't (in
		37	general) run in parallel at the same time even on SMP machines. The
		38	specific flavor of thread offered by this module also guarantees you that
		39	it will not switch between threads unless necessary, at easily-identified
		40	points in your program, so locking and parallel access are rarely an
		41	issue, making thread programming much safer and easier than using other
		42	thread models.
		43
		44	Unlike the so-called "Perl threads" (which are not actually real threads
		45	but only the windows process emulation (see section of same name for more
		46	details) ported to unix, and as such act as processes), Coro provides
		47	a full shared address space, which makes communication between threads
		48	very easy. And Coro's threads are fast, too: disabling the Windows
		49	process emulation code in your perl and using Coro can easily result in
		50	a two to four times speed increase for your programs. A parallel matrix
		51	multiplication benchmark runs over 300 times faster on a single core than
		52	perl's pseudo-threads on a quad core using all four cores.
		53
		54	Coro achieves that by supporting multiple running interpreters that share
		55	data, which is especially useful to code pseudo-parallel processes and
		56	for event-based programming, such as multiple HTTP-GET requests running
		57	concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro
		58	into an event-based environment.
		59
		60	In this module, a thread is defined as "callchain + lexical variables +
		61	some package variables + C stack), that is, a thread has its own callchain,
		62	its own set of lexicals and its own set of perls most important global
		63	variables (see L<Coro::State> for more configuration and background info).
		64
		65	See also the C<SEE ALSO> section at the end of this document - the Coro
		66	module family is quite large.
		67
23	=cut	68	=cut
24		69
25	package Coro;	70	package Coro;
26		71
		72	use common::sense;
		73
		74	use Carp ();
		75
		76	use Guard ();
		77
27	use Coro::State;	78	use Coro::State;
28		79
29	use base Exporter;	80	use base qw(Coro::State Exporter);
30		81
31	$VERSION = 0.04;	82	our $idle; # idle handler
		83	our $main; # main coro
		84	our $current; # current coro
32		85
33	@EXPORT = qw(async yield schedule);	86	our $VERSION = 5.21;
34	@EXPORT_OK = qw($current);
35		87
36	{	88	our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub rouse_cb rouse_wait);
37	use subs 'async';	89	our %EXPORT_TAGS = (
		90	prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
		91	);
		92	our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
38		93
39	my @async;	94	=head1 GLOBAL VARIABLES
40		95
41	# this way of handling attributes simply is NOT scalable ;()	96	=over 4
42	sub import {	97
43	Coro->export_to_level(1, @_);	98	=item $Coro::main
44	my $old = *{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"}{CODE};	99
45	*{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"} = sub {	100	This variable stores the Coro object that represents the main
46	my ($package, $ref) = (shift, shift);	101	program. While you cna C<ready> it and do most other things you can do to
47	my @attrs;	102	coro, it is mainly useful to compare again C<$Coro::current>, to see
48	for (@_) {	103	whether you are running in the main program or not.
49	if ($_ eq "Coro") {	104
50	push @async, $ref;	105	=cut
51	} else {	106
52	push @attrs, @_;	107	# $main is now being initialised by Coro::State
53	}	108
54	}	109	=item $Coro::current
55	return $old ? $old->($package, $name, @attrs) : @attrs;	110
		111	The Coro object representing the current coro (the last
		112	coro that the Coro scheduler switched to). The initial value is
		113	C<$Coro::main> (of course).
		114
		115	This variable is B<strictly> I<read-only>. You can take copies of the
		116	value stored in it and use it as any other Coro object, but you must
		117	not otherwise modify the variable itself.
		118
		119	=cut
		120
		121	sub current() { $current } # [DEPRECATED]
		122
		123	=item $Coro::idle
		124
		125	This variable is mainly useful to integrate Coro into event loops. It is
		126	usually better to rely on L<Coro::AnyEvent> or L<Coro::EV>, as this is
		127	pretty low-level functionality.
		128
		129	This variable stores a Coro object that is put into the ready queue when
		130	there are no other ready threads (without invoking any ready hooks).
		131
		132	The default implementation dies with "FATAL: deadlock detected.", followed
		133	by a thread listing, because the program has no other way to continue.
		134
		135	This hook is overwritten by modules such as C<Coro::EV> and
		136	C<Coro::AnyEvent> to wait on an external event that hopefully wake up a
		137	coro so the scheduler can run it.
		138
		139	See L<Coro::EV> or L<Coro::AnyEvent> for examples of using this technique.
		140
		141	=cut
		142
		143	$idle = new Coro sub {
		144	require Coro::Debug;
		145	die "FATAL: deadlock detected.\n"
		146	. Coro::Debug::ps_listing ();
		147	};
		148
		149	# this coro is necessary because a coro
		150	# cannot destroy itself.
		151	our @destroy;
		152	our $manager;
		153
		154	$manager = new Coro sub {
		155	while () {
		156	Coro::State::cancel shift @destroy
		157	while @destroy;
		158
		159	&schedule;
		160	}
		161	};
		162	$manager->{desc} = "[coro manager]";
		163	$manager->prio (PRIO_MAX);
		164
		165	=back
		166
		167	=head1 SIMPLE CORO CREATION
		168
		169	=over 4
		170
		171	=item async { ... } [@args...]
		172
		173	Create a new coro and return its Coro object (usually
		174	unused). The coro will be put into the ready queue, so
		175	it will start running automatically on the next scheduler run.
		176
		177	The first argument is a codeblock/closure that should be executed in the
		178	coro. When it returns argument returns the coro is automatically
		179	terminated.
		180
		181	The remaining arguments are passed as arguments to the closure.
		182
		183	See the C<Coro::State::new> constructor for info about the coro
		184	environment in which coro are executed.
		185
		186	Calling C<exit> in a coro will do the same as calling exit outside
		187	the coro. Likewise, when the coro dies, the program will exit,
		188	just as it would in the main program.
		189
		190	If you do not want that, you can provide a default C<die> handler, or
		191	simply avoid dieing (by use of C<eval>).
		192
		193	Example: Create a new coro that just prints its arguments.
		194
		195	async {
		196	print "@_\n";
		197	} 1,2,3,4;
		198
		199	=item async_pool { ... } [@args...]
		200
		201	Similar to C<async>, but uses a coro pool, so you should not call
		202	terminate or join on it (although you are allowed to), and you get a
		203	coro that might have executed other code already (which can be good
		204	or bad :).
		205
		206	On the plus side, this function is about twice as fast as creating (and
		207	destroying) a completely new coro, so if you need a lot of generic
		208	coros in quick successsion, use C<async_pool>, not C<async>.
		209
		210	The code block is executed in an C<eval> context and a warning will be
		211	issued in case of an exception instead of terminating the program, as
		212	C<async> does. As the coro is being reused, stuff like C<on_destroy>
		213	will not work in the expected way, unless you call terminate or cancel,
		214	which somehow defeats the purpose of pooling (but is fine in the
		215	exceptional case).
		216
		217	The priority will be reset to C<0> after each run, tracing will be
		218	disabled, the description will be reset and the default output filehandle
		219	gets restored, so you can change all these. Otherwise the coro will
		220	be re-used "as-is": most notably if you change other per-coro global
		221	stuff such as C<$/> you I<must needs> revert that change, which is most
		222	simply done by using local as in: C<< local $/ >>.
		223
		224	The idle pool size is limited to C<8> idle coros (this can be
		225	adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle
		226	coros as required.
		227
		228	If you are concerned about pooled coros growing a lot because a
		229	single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
		230	{ terminate }> once per second or so to slowly replenish the pool. In
		231	addition to that, when the stacks used by a handler grows larger than 32kb
		232	(adjustable via $Coro::POOL_RSS) it will also be destroyed.
		233
		234	=cut
		235
		236	our $POOL_SIZE = 8;
		237	our $POOL_RSS = 32 * 1024;
		238	our @async_pool;
		239
		240	sub pool_handler {
		241	while () {
		242	eval {
		243	&{&_pool_handler} while 1;
56	};	244	};
57	}
58		245
59	sub INIT {	246	warn $@ if $@;
60	async pop @async while @async;
61	}	247	}
62	}	248	}
63		249
64	=item $main	250	=back
65		251
66	This coroutine represents the main program.	252	=head1 STATIC METHODS
67		253
68	=cut	254	Static methods are actually functions that implicitly operate on the
		255	current coro.
69		256
70	our $main = new Coro;	257	=over 4
71		258
72	=item $current	259	=item schedule
73		260
74	The current coroutine (the last coroutine switched to). The initial value is C<$main> (of course).	261	Calls the scheduler. The scheduler will find the next coro that is
		262	to be run from the ready queue and switches to it. The next coro
		263	to be run is simply the one with the highest priority that is longest
		264	in its ready queue. If there is no coro ready, it will call the
		265	C<$Coro::idle> hook.
75		266
76	=cut	267	Please note that the current coro will I<not> be put into the ready
		268	queue, so calling this function usually means you will never be called
		269	again unless something else (e.g. an event handler) calls C<< ->ready >>,
		270	thus waking you up.
77		271
78	# maybe some other module used Coro::Specific before...	272	This makes C<schedule> I<the> generic method to use to block the current
79	if ($current) {	273	coro and wait for events: first you remember the current coro in
80	$main->{specific} = $current->{specific};	274	a variable, then arrange for some callback of yours to call C<< ->ready
		275	>> on that once some event happens, and last you call C<schedule> to put
		276	yourself to sleep. Note that a lot of things can wake your coro up,
		277	so you need to check whether the event indeed happened, e.g. by storing the
		278	status in a variable.
		279
		280	See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks.
		281
		282	=item cede
		283
		284	"Cede" to other coros. This function puts the current coro into
		285	the ready queue and calls C<schedule>, which has the effect of giving
		286	up the current "timeslice" to other coros of the same or higher
		287	priority. Once your coro gets its turn again it will automatically be
		288	resumed.
		289
		290	This function is often called C<yield> in other languages.
		291
		292	=item Coro::cede_notself
		293
		294	Works like cede, but is not exported by default and will cede to I<any>
		295	coro, regardless of priority. This is useful sometimes to ensure
		296	progress is made.
		297
		298	=item terminate [arg...]
		299
		300	Terminates the current coro with the given status values (see L<cancel>).
		301
		302	=item Coro::on_enter BLOCK, Coro::on_leave BLOCK
		303
		304	These function install enter and leave winders in the current scope. The
		305	enter block will be executed when on_enter is called and whenever the
		306	current coro is re-entered by the scheduler, while the leave block is
		307	executed whenever the current coro is blocked by the scheduler, and
		308	also when the containing scope is exited (by whatever means, be it exit,
		309	die, last etc.).
		310
		311	I<Neither invoking the scheduler, nor exceptions, are allowed within those
		312	BLOCKs>. That means: do not even think about calling C<die> without an
		313	eval, and do not even think of entering the scheduler in any way.
		314
		315	Since both BLOCKs are tied to the current scope, they will automatically
		316	be removed when the current scope exits.
		317
		318	These functions implement the same concept as C<dynamic-wind> in scheme
		319	does, and are useful when you want to localise some resource to a specific
		320	coro.
		321
		322	They slow down thread switching considerably for coros that use them
		323	(about 40% for a BLOCK with a single assignment, so thread switching is
		324	still reasonably fast if the handlers are fast).
		325
		326	These functions are best understood by an example: The following function
		327	will change the current timezone to "Antarctica/South_Pole", which
		328	requires a call to C<tzset>, but by using C<on_enter> and C<on_leave>,
		329	which remember/change the current timezone and restore the previous
		330	value, respectively, the timezone is only changed for the coro that
		331	installed those handlers.
		332
		333	use POSIX qw(tzset);
		334
		335	async {
		336	my $old_tz; # store outside TZ value here
		337
		338	Coro::on_enter {
		339	$old_tz = $ENV{TZ}; # remember the old value
		340
		341	$ENV{TZ} = "Antarctica/South_Pole";
		342	tzset; # enable new value
		343	};
		344
		345	Coro::on_leave {
		346	$ENV{TZ} = $old_tz;
		347	tzset; # restore old value
		348	};
		349
		350	# at this place, the timezone is Antarctica/South_Pole,
		351	# without disturbing the TZ of any other coro.
		352	};
		353
		354	This can be used to localise about any resource (locale, uid, current
		355	working directory etc.) to a block, despite the existance of other
		356	coros.
		357
		358	Another interesting example implements time-sliced multitasking using
		359	interval timers (this could obviously be optimised, but does the job):
		360
		361	# "timeslice" the given block
		362	sub timeslice(&) {
		363	use Time::HiRes ();
		364
		365	Coro::on_enter {
		366	# on entering the thread, we set an VTALRM handler to cede
		367	$SIG{VTALRM} = sub { cede };
		368	# and then start the interval timer
		369	Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0.01, 0.01;
		370	};
		371	Coro::on_leave {
		372	# on leaving the thread, we stop the interval timer again
		373	Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0, 0;
		374	};
		375
		376	&{+shift};
		377	}
		378
		379	# use like this:
		380	timeslice {
		381	# The following is an endless loop that would normally
		382	# monopolise the process. Since it runs in a timesliced
		383	# environment, it will regularly cede to other threads.
		384	while () { }
		385	};
		386
		387
		388	=item killall
		389
		390	Kills/terminates/cancels all coros except the currently running one.
		391
		392	Note that while this will try to free some of the main interpreter
		393	resources if the calling coro isn't the main coro, but one
		394	cannot free all of them, so if a coro that is not the main coro
		395	calls this function, there will be some one-time resource leak.
		396
		397	=cut
		398
		399	sub killall {
		400	for (Coro::State::list) {
		401	$_->cancel
		402	if $_ != $current && UNIVERSAL::isa $_, "Coro";
		403	}
81	}	404	}
82		405
83	our $current = $main;	406	=back
84		407
85	=item $idle	408	=head1 CORO OBJECT METHODS
86		409
87	The coroutine to switch to when no other coroutine is running. The default	410	These are the methods you can call on coro objects (or to create
88	implementation prints "FATAL: deadlock detected" and exits.	411	them).
89		412
90	=cut	413	=over 4
91		414
92	# should be done using priorities :(	415	=item new Coro \&sub [, @args...]
		416
		417	Create a new coro and return it. When the sub returns, the coro
		418	automatically terminates as if C<terminate> with the returned values were
		419	called. To make the coro run you must first put it into the ready
		420	queue by calling the ready method.
		421
		422	See C<async> and C<Coro::State::new> for additional info about the
		423	coro environment.
		424
		425	=cut
		426
		427	sub _coro_run {
		428	terminate &{+shift};
		429	}
		430
		431	=item $success = $coro->ready
		432
		433	Put the given coro into the end of its ready queue (there is one
		434	queue for each priority) and return true. If the coro is already in
		435	the ready queue, do nothing and return false.
		436
		437	This ensures that the scheduler will resume this coro automatically
		438	once all the coro of higher priority and all coro of the same
		439	priority that were put into the ready queue earlier have been resumed.
		440
		441	=item $coro->suspend
		442
		443	Suspends the specified coro. A suspended coro works just like any other
		444	coro, except that the scheduler will not select a suspended coro for
		445	execution.
		446
		447	Suspending a coro can be useful when you want to keep the coro from
		448	running, but you don't want to destroy it, or when you want to temporarily
		449	freeze a coro (e.g. for debugging) to resume it later.
		450
		451	A scenario for the former would be to suspend all (other) coros after a
		452	fork and keep them alive, so their destructors aren't called, but new
		453	coros can be created.
		454
		455	=item $coro->resume
		456
		457	If the specified coro was suspended, it will be resumed. Note that when
		458	the coro was in the ready queue when it was suspended, it might have been
		459	unreadied by the scheduler, so an activation might have been lost.
		460
		461	To avoid this, it is best to put a suspended coro into the ready queue
		462	unconditionally, as every synchronisation mechanism must protect itself
		463	against spurious wakeups, and the one in the Coro family certainly do
		464	that.
		465
		466	=item $is_ready = $coro->is_ready
		467
		468	Returns true iff the Coro object is in the ready queue. Unless the Coro
		469	object gets destroyed, it will eventually be scheduled by the scheduler.
		470
		471	=item $is_running = $coro->is_running
		472
		473	Returns true iff the Coro object is currently running. Only one Coro object
		474	can ever be in the running state (but it currently is possible to have
		475	multiple running Coro::States).
		476
		477	=item $is_suspended = $coro->is_suspended
		478
		479	Returns true iff this Coro object has been suspended. Suspended Coros will
		480	not ever be scheduled.
		481
		482	=item $coro->cancel (arg...)
		483
		484	Terminates the given Coro and makes it return the given arguments as
		485	status (default: the empty list). Never returns if the Coro is the
		486	current Coro.
		487
		488	=cut
		489
		490	sub cancel {
		491	my $self = shift;
		492
		493	if ($current == $self) {
		494	terminate @_;
		495	} else {
		496	$self->{_status} = [@_];
		497	Coro::State::cancel $self;
		498	}
		499	}
		500
		501	=item $coro->schedule_to
		502
		503	Puts the current coro to sleep (like C<Coro::schedule>), but instead
		504	of continuing with the next coro from the ready queue, always switch to
		505	the given coro object (regardless of priority etc.). The readyness
		506	state of that coro isn't changed.
		507
		508	This is an advanced method for special cases - I'd love to hear about any
		509	uses for this one.
		510
		511	=item $coro->cede_to
		512
		513	Like C<schedule_to>, but puts the current coro into the ready
		514	queue. This has the effect of temporarily switching to the given
		515	coro, and continuing some time later.
		516
		517	This is an advanced method for special cases - I'd love to hear about any
		518	uses for this one.
		519
		520	=item $coro->throw ([$scalar])
		521
		522	If C<$throw> is specified and defined, it will be thrown as an exception
		523	inside the coro at the next convenient point in time. Otherwise
		524	clears the exception object.
		525
		526	Coro will check for the exception each time a schedule-like-function
		527	returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down
		528	>>, C<< Coro::Handle->readable >> and so on. Most of these functions
		529	detect this case and return early in case an exception is pending.
		530
		531	The exception object will be thrown "as is" with the specified scalar in
		532	C<$@>, i.e. if it is a string, no line number or newline will be appended
		533	(unlike with C<die>).
		534
		535	This can be used as a softer means than C<cancel> to ask a coro to
		536	end itself, although there is no guarantee that the exception will lead to
		537	termination, and if the exception isn't caught it might well end the whole
		538	program.
		539
		540	You might also think of C<throw> as being the moral equivalent of
		541	C<kill>ing a coro with a signal (in this case, a scalar).
		542
		543	=item $coro->join
		544
		545	Wait until the coro terminates and return any values given to the
		546	C<terminate> or C<cancel> functions. C<join> can be called concurrently
		547	from multiple coro, and all will be resumed and given the status
		548	return once the C<$coro> terminates.
		549
		550	=cut
		551
		552	sub join {
		553	my $self = shift;
		554
		555	unless ($self->{_status}) {
		556	my $current = $current;
		557
		558	push @{$self->{_on_destroy}}, sub {
		559	$current->ready;
		560	undef $current;
		561	};
		562
		563	&schedule while $current;
		564	}
		565
		566	wantarray ? @{$self->{_status}} : $self->{_status}[0];
		567	}
		568
		569	=item $coro->on_destroy (\&cb)
		570
		571	Registers a callback that is called when this coro gets destroyed,
		572	but before it is joined. The callback gets passed the terminate arguments,
		573	if any, and I<must not> die, under any circumstances.
		574
		575	=cut
		576
		577	sub on_destroy {
		578	my ($self, $cb) = @_;
		579
		580	push @{ $self->{_on_destroy} }, $cb;
		581	}
		582
		583	=item $oldprio = $coro->prio ($newprio)
		584
		585	Sets (or gets, if the argument is missing) the priority of the
		586	coro. Higher priority coro get run before lower priority
		587	coro. Priorities are small signed integers (currently -4 .. +3),
		588	that you can refer to using PRIO_xxx constants (use the import tag :prio
		589	to get then):
		590
		591	PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN
		592	3 > 1 > 0 > -1 > -3 > -4
		593
		594	# set priority to HIGH
		595	current->prio (PRIO_HIGH);
		596
		597	The idle coro ($Coro::idle) always has a lower priority than any
		598	existing coro.
		599
		600	Changing the priority of the current coro will take effect immediately,
		601	but changing the priority of coro in the ready queue (but not
		602	running) will only take effect after the next schedule (of that
		603	coro). This is a bug that will be fixed in some future version.
		604
		605	=item $newprio = $coro->nice ($change)
		606
		607	Similar to C<prio>, but subtract the given value from the priority (i.e.
		608	higher values mean lower priority, just as in unix).
		609
		610	=item $olddesc = $coro->desc ($newdesc)
		611
		612	Sets (or gets in case the argument is missing) the description for this
		613	coro. This is just a free-form string you can associate with a
		614	coro.
		615
		616	This method simply sets the C<< $coro->{desc} >> member to the given
		617	string. You can modify this member directly if you wish.
		618
		619	=cut
		620
		621	sub desc {
		622	my $old = $_[0]{desc};
		623	$_[0]{desc} = $_[1] if @_ > 1;
		624	$old;
		625	}
		626
		627	sub transfer {
		628	require Carp;
		629	Carp::croak ("You must not call ->transfer on Coro objects. Use Coro::State objects or the ->schedule_to method. Caught");
		630	}
		631
		632	=back
		633
		634	=head1 GLOBAL FUNCTIONS
		635
		636	=over 4
		637
		638	=item Coro::nready
		639
		640	Returns the number of coro that are currently in the ready state,
		641	i.e. that can be switched to by calling C<schedule> directory or
		642	indirectly. The value C<0> means that the only runnable coro is the
		643	currently running one, so C<cede> would have no effect, and C<schedule>
		644	would cause a deadlock unless there is an idle handler that wakes up some
		645	coro.
		646
		647	=item my $guard = Coro::guard { ... }
		648
		649	This function still exists, but is deprecated. Please use the
		650	C<Guard::guard> function instead.
		651
		652	=cut
		653
		654	BEGIN { *guard = \&Guard::guard }
		655
		656	=item unblock_sub { ... }
		657
		658	This utility function takes a BLOCK or code reference and "unblocks" it,
		659	returning a new coderef. Unblocking means that calling the new coderef
		660	will return immediately without blocking, returning nothing, while the
		661	original code ref will be called (with parameters) from within another
		662	coro.
		663
		664	The reason this function exists is that many event libraries (such as the
		665	venerable L<Event\|Event> module) are not thread-safe (a weaker form
		666	of reentrancy). This means you must not block within event callbacks,
		667	otherwise you might suffer from crashes or worse. The only event library
		668	currently known that is safe to use without C<unblock_sub> is L<EV>.
		669
		670	Coro will try to catch you when you block in the event loop
		671	("FATAL:$Coro::IDLE blocked itself"), but this is just best effort and
		672	only works when you do not run your own event loop.
		673
		674	This function allows your callbacks to block by executing them in another
		675	coro where it is safe to block. One example where blocking is handy
		676	is when you use the L<Coro::AIO\|Coro::AIO> functions to save results to
		677	disk, for example.
		678
		679	In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
		680	creating event callbacks that want to block.
		681
		682	If your handler does not plan to block (e.g. simply sends a message to
		683	another coro, or puts some other coro into the ready queue), there is
		684	no reason to use C<unblock_sub>.
		685
		686	Note that you also need to use C<unblock_sub> for any other callbacks that
		687	are indirectly executed by any C-based event loop. For example, when you
		688	use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
		689	provides callbacks that are the result of some event callback, then you
		690	must not block either, or use C<unblock_sub>.
		691
		692	=cut
		693
		694	our @unblock_queue;
		695
		696	# we create a special coro because we want to cede,
		697	# to reduce pressure on the coro pool (because most callbacks
		698	# return immediately and can be reused) and because we cannot cede
		699	# inside an event callback.
93	our $idle = new Coro sub {	700	our $unblock_scheduler = new Coro sub {
94	print STDERR "FATAL: deadlock detected\n";	701	while () {
95	exit(51);	702	while (my $cb = pop @unblock_queue) {
		703	&async_pool (@$cb);
		704
		705	# for short-lived callbacks, this reduces pressure on the coro pool
		706	# as the chance is very high that the async_poll coro will be back
		707	# in the idle state when cede returns
		708	cede;
		709	}
		710	schedule; # sleep well
		711	}
96	};	712	};
		713	$unblock_scheduler->{desc} = "[unblock_sub scheduler]";
97		714
98	# we really need priorities...	715	sub unblock_sub(&) {
99	my @ready = (); # the ready queue. hehe, rather broken ;)	716	my $cb = shift;
100		717
101	# static methods. not really.	718	sub {
		719	unshift @unblock_queue, [$cb, @_];
		720	$unblock_scheduler->ready;
		721	}
		722	}
102		723
103	=head2 STATIC METHODS	724	=item $cb = rouse_cb
104		725
105	Static methods are actually functions that operate on the current process only.	726	Create and return a "rouse callback". That's a code reference that,
		727	when called, will remember a copy of its arguments and notify the owner
		728	coro of the callback.
		729
		730	See the next function.
		731
		732	=item @args = rouse_wait [$cb]
		733
		734	Wait for the specified rouse callback (or the last one that was created in
		735	this coro).
		736
		737	As soon as the callback is invoked (or when the callback was invoked
		738	before C<rouse_wait>), it will return the arguments originally passed to
		739	the rouse callback. In scalar context, that means you get the I<last>
		740	argument, just as if C<rouse_wait> had a C<return ($a1, $a2, $a3...)>
		741	statement at the end.
		742
		743	See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example.
		744
		745	=back
		746
		747	=cut
		748
		749	1;
		750
		751	=head1 HOW TO WAIT FOR A CALLBACK
		752
		753	It is very common for a coro to wait for some callback to be
		754	called. This occurs naturally when you use coro in an otherwise
		755	event-based program, or when you use event-based libraries.
		756
		757	These typically register a callback for some event, and call that callback
		758	when the event occured. In a coro, however, you typically want to
		759	just wait for the event, simplyifying things.
		760
		761	For example C<< AnyEvent->child >> registers a callback to be called when
		762	a specific child has exited:
		763
		764	my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });
		765
		766	But from within a coro, you often just want to write this:
		767
		768	my $status = wait_for_child $pid;
		769
		770	Coro offers two functions specifically designed to make this easy,
		771	C<Coro::rouse_cb> and C<Coro::rouse_wait>.
		772
		773	The first function, C<rouse_cb>, generates and returns a callback that,
		774	when invoked, will save its arguments and notify the coro that
		775	created the callback.
		776
		777	The second function, C<rouse_wait>, waits for the callback to be called
		778	(by calling C<schedule> to go to sleep) and returns the arguments
		779	originally passed to the callback.
		780
		781	Using these functions, it becomes easy to write the C<wait_for_child>
		782	function mentioned above:
		783
		784	sub wait_for_child($) {
		785	my ($pid) = @_;
		786
		787	my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb);
		788
		789	my ($rpid, $rstatus) = Coro::rouse_wait;
		790	$rstatus
		791	}
		792
		793	In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough,
		794	you can roll your own, using C<schedule>:
		795
		796	sub wait_for_child($) {
		797	my ($pid) = @_;
		798
		799	# store the current coro in $current,
		800	# and provide result variables for the closure passed to ->child
		801	my $current = $Coro::current;
		802	my ($done, $rstatus);
		803
		804	# pass a closure to ->child
		805	my $watcher = AnyEvent->child (pid => $pid, cb => sub {
		806	$rstatus = $_[1]; # remember rstatus
		807	$done = 1; # mark $rstatus as valud
		808	});
		809
		810	# wait until the closure has been called
		811	schedule while !$done;
		812
		813	$rstatus
		814	}
		815
		816
		817	=head1 BUGS/LIMITATIONS
106		818
107	=over 4	819	=over 4
108		820
109	=item async { ... };	821	=item fork with pthread backend
110		822
111	Create a new asynchronous process and return it's process object	823	When Coro is compiled using the pthread backend (which isn't recommended
112	(usually unused). When the sub returns the new process is automatically	824	but required on many BSDs as their libcs are completely broken), then
113	terminated.	825	coro will not survive a fork. There is no known workaround except to
		826	fix your libc and use a saner backend.
114		827
115	=cut	828	=item perl process emulation ("threads")
116		829
117	sub async(&) {	830	This module is not perl-pseudo-thread-safe. You should only ever use this
118	my $pid = new Coro $_[0];	831	module from the first thread (this requirement might be removed in the
119	$pid->ready;	832	future to allow per-thread schedulers, but Coro::State does not yet allow
120	$pid;	833	this). I recommend disabling thread support and using processes, as having
121	}	834	the windows process emulation enabled under unix roughly halves perl
		835	performance, even when not used.
122		836
123	=item schedule	837	=item coro switching is not signal safe
124		838
125	Calls the scheduler. Please note that the current process will not be put	839	You must not switch to another coro from within a signal handler (only
126	into the ready queue, so calling this function usually means you will	840	relevant with %SIG - most event libraries provide safe signals), I<unless>
127	never be called again.	841	you are sure you are not interrupting a Coro function.
128		842
129	=cut	843	That means you I<MUST NOT> call any function that might "block" the
130		844	current coro - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or
131	my $prev;	845	anything that calls those. Everything else, including calling C<ready>,
132		846	works.
133	sub schedule {
134	# should be done using priorities :(
135	($prev, $current) = ($current, shift @ready \|\| $idle);
136	Coro::State::transfer($prev, $current);
137	}
138
139	=item yield
140
141	Yield to other processes. This function puts the current process into the
142	ready queue and calls C<schedule>.
143
144	=cut
145
146	sub yield {
147	$current->ready;
148	&schedule;
149	}
150
151	=item terminate
152
153	Terminates the current process.
154
155	=cut
156
157	sub terminate {
158	&schedule;
159	}
160		847
161	=back	848	=back
162		849
163	# dynamic methods
164		850
165	=head2 PROCESS METHODS	851	=head1 WINDOWS PROCESS EMULATION
166		852
167	These are the methods you can call on process objects.	853	A great many people seem to be confused about ithreads (for example, Chip
		854	Salzenberg called me unintelligent, incapable, stupid and gullible,
		855	while in the same mail making rather confused statements about perl
		856	ithreads (for example, that memory or files would be shared), showing his
		857	lack of understanding of this area - if it is hard to understand for Chip,
		858	it is probably not obvious to everybody).
168		859
169	=over 4	860	What follows is an ultra-condensed version of my talk about threads in
		861	scripting languages given onthe perl workshop 2009:
170		862
171	=item new Coro \⊂	863	The so-called "ithreads" were originally implemented for two reasons:
		864	first, to (badly) emulate unix processes on native win32 perls, and
		865	secondly, to replace the older, real thread model ("5.005-threads").
172		866
173	Create a new process and return it. When the sub returns the process	867	It does that by using threads instead of OS processes. The difference
174	automatically terminates. To start the process you must first put it into	868	between processes and threads is that threads share memory (and other
175	the ready queue by calling the ready method.	869	state, such as files) between threads within a single process, while
		870	processes do not share anything (at least not semantically). That
		871	means that modifications done by one thread are seen by others, while
		872	modifications by one process are not seen by other processes.
176		873
177	=cut	874	The "ithreads" work exactly like that: when creating a new ithreads
		875	process, all state is copied (memory is copied physically, files and code
		876	is copied logically). Afterwards, it isolates all modifications. On UNIX,
		877	the same behaviour can be achieved by using operating system processes,
		878	except that UNIX typically uses hardware built into the system to do this
		879	efficiently, while the windows process emulation emulates this hardware in
		880	software (rather efficiently, but of course it is still much slower than
		881	dedicated hardware).
178		882
179	sub new {	883	As mentioned before, loading code, modifying code, modifying data
180	my $class = shift;	884	structures and so on is only visible in the ithreads process doing the
181	my $proc = $_[0];	885	modification, not in other ithread processes within the same OS process.
182	bless {
183	_coro_state => new Coro::State ($proc ? sub { &$proc; &terminate } : $proc),
184	}, $class;
185	}
186		886
187	=item $process->ready	887	This is why "ithreads" do not implement threads for perl at all, only
		888	processes. What makes it so bad is that on non-windows platforms, you can
		889	actually take advantage of custom hardware for this purpose (as evidenced
		890	by the forks module, which gives you the (i-) threads API, just much
		891	faster).
188		892
189	Put the current process into the ready queue.	893	Sharing data is in the i-threads model is done by transfering data
		894	structures between threads using copying semantics, which is very slow -
		895	shared data simply does not exist. Benchmarks using i-threads which are
		896	communication-intensive show extremely bad behaviour with i-threads (in
		897	fact, so bad that Coro, which cannot take direct advantage of multiple
		898	CPUs, is often orders of magnitude faster because it shares data using
		899	real threads, refer to my talk for details).
190		900
191	=cut	901	As summary, i-threads use threads to implement processes, while
		902	the compatible forks module uses processes to emulate, uhm,
		903	processes. I-threads slow down every perl program when enabled, and
		904	outside of windows, serve no (or little) practical purpose, but
		905	disadvantages every single-threaded Perl program.
192		906
193	sub ready {	907	This is the reason that I try to avoid the name "ithreads", as it is
194	push @ready, $_[0];	908	misleading as it implies that it implements some kind of thread model for
195	}	909	perl, and prefer the name "windows process emulation", which describes the
196		910	actual use and behaviour of it much better.
197	=back
198
199	=cut
200
201	1;
202		911
203	=head1 SEE ALSO	912	=head1 SEE ALSO
204		913
205	L<Coro::Channel>, L<Coro::Cont>, L<Coro::Specific>, L<Coro::Semaphore>,	914	Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
206	L<Coro::Signal>, L<Coro::State>, L<Coro::Event>.	915
		916	Debugging: L<Coro::Debug>.
		917
		918	Support/Utility: L<Coro::Specific>, L<Coro::Util>.
		919
		920	Locking and IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>,
		921	L<Coro::SemaphoreSet>, L<Coro::RWLock>.
		922
		923	I/O and Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
		924
		925	Compatibility with other modules: L<Coro::LWP> (but see also L<AnyEvent::HTTP> for
		926	a better-working alternative), L<Coro::BDB>, L<Coro::Storable>,
		927	L<Coro::Select>.
		928
		929	XS API: L<Coro::MakeMaker>.
		930
		931	Low level Configuration, Thread Environment, Continuations: L<Coro::State>.
207		932
208	=head1 AUTHOR	933	=head1 AUTHOR
209		934
210	Marc Lehmann <pcg@goof.com>	935	Marc Lehmann <schmorp@schmorp.de>
211	http://www.goof.com/pcg/marc/	936	http://home.schmorp.de/
212		937
213	=cut	938	=cut
214		939

Diff Legend

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

Comparing Coro/Coro.pm (file contents): Revision 1.11 by root, Sun Jul 15 03:24:18 2001 UTC vs. Revision 1.274 by root, Sat Dec 12 01:30:26 2009 UTC

Diff Legend

Comparing Coro/Coro.pm (file contents):
Revision 1.11 by root, Sun Jul 15 03:24:18 2001 UTC vs.
Revision 1.274 by root, Sat Dec 12 01:30:26 2009 UTC