ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/Coro/Coro.pm

Comparing Coro/Coro.pm (file contents):
Revision 1.128 by root, Wed Sep 19 21:39:15 2007 UTC vs.
Revision 1.297 by root, Thu May 12 23:55:39 2011 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 coroutine 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	cede;	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
23	This module collection manages coroutines. Coroutines are similar	31	For a tutorial-style introduction, please read the L<Coro::Intro>
24	to threads but don't run in parallel at the same time even on SMP	32	manpage. This manpage mainly contains reference information.
25	machines. The specific flavor of coroutine used in this module also
26	guarantees you that it will not switch between coroutines unless
27	necessary, at easily-identified points in your program, so locking and
28	parallel access are rarely an issue, making coroutine programming much
29	safer than threads programming.
30		33
31	(Perl, however, does not natively support real threads but instead does a	34	This module collection manages continuations in general, most often in
32	very slow and memory-intensive emulation of processes using threads. This	35	the form of cooperative threads (also called coros, or simply "coro"
33	is a performance win on Windows machines, and a loss everywhere else).	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.
34		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
		46	more details) ported to UNIX, and as such act as processes), Coro
		47	provides a full shared address space, which makes communication between
		48	threads very easy. And coro 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 (very communication-intensive) runs over 300
		52	times faster on a single core than perls pseudo-threads on a quad core
		53	using all four cores.
		54
		55	Coro achieves that by supporting multiple running interpreters that share
		56	data, which is especially useful to code pseudo-parallel processes and
		57	for event-based programming, such as multiple HTTP-GET requests running
		58	concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro
		59	into an event-based environment.
		60
35	In this module, coroutines are defined as "callchain + lexical variables +	61	In this module, a thread is defined as "callchain + lexical variables +
36	@_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own callchain,	62	some package variables + C stack), that is, a thread has its own callchain,
37	its own set of lexicals and its own set of perls most important global	63	its own set of lexicals and its own set of perls most important global
		64	variables (see L<Coro::State> for more configuration and background info).
		65
		66	See also the C<SEE ALSO> section at the end of this document - the Coro
		67	module family is quite large.
		68
		69	=head1 CORO THREAD LIFE CYCLE
		70
		71	During the long and exciting (or not) life of a coro thread, it goes
		72	through a number of states:
		73
		74	=over 4
		75
		76	=item 1. Creation
		77
		78	The first thing in the life of a coro thread is it's creation -
		79	obviously. The typical way to create a thread is to call the C<async
		80	BLOCK> function:
		81
		82	async {
		83	# thread code goes here
		84	};
		85
		86	You can also pass arguments, which are put in C<@_>:
		87
		88	async {
		89	print $_[1]; # prints 2
		90	} 1, 2, 3;
		91
		92	This creates a new coro thread and puts it into the ready queue, meaning
		93	it will run as soon as the CPU is free for it.
		94
		95	C<async> will return a Coro object - you can store this for future
		96	reference or ignore it - a thread that is running, ready to run or waiting
		97	for some event is alive on it's own.
		98
		99	Another way to create a thread is to call the C<new> constructor with a
		100	code-reference:
		101
		102	new Coro sub {
		103	# thread code goes here
		104	}, @optional_arguments;
		105
		106	This is quite similar to calling C<async>, but the important difference is
		107	that the new thread is not put into the ready queue, so the thread will
		108	not run until somebody puts it there. C<async> is, therefore, identical to
		109	this sequence:
		110
		111	my $coro = new Coro sub {
		112	# thread code goes here
		113	};
		114	$coro->ready;
		115	return $coro;
		116
		117	=item 2. Startup
		118
		119	When a new coro thread is created, only a copy of the code reference
		120	and the arguments are stored, no extra memory for stacks and so on is
		121	allocated, keeping the coro thread in a low-memory state.
		122
		123	Only when it actually starts executing will all the resources be finally
		124	allocated.
		125
		126	The optional arguments specified at coro creation are available in C<@_>,
		127	similar to function calls.
		128
		129	=item 3. Running / Blocking
		130
		131	A lot can happen after the coro thread has started running. Quite usually,
		132	it will not run to the end in one go (because you could use a function
		133	instead), but it will give up the CPU regularly because it waits for
		134	external events.
		135
		136	As long as a coro thread runs, its Coro object is available in the global
		137	variable C<$Coro::current>.
		138
		139	The low-level way to give up the CPU is to call the scheduler, which
		140	selects a new coro thread to run:
		141
		142	Coro::schedule;
		143
		144	Since running threads are not in the ready queue, calling the scheduler
		145	without doing anything else will block the coro thread forever - you need
		146	to arrange either for the coro to put woken up (readied) by some other
		147	event or some other thread, or you can put it into the ready queue before
		148	scheduling:
		149
		150	# this is exactly what Coro::cede does
		151	$Coro::current->ready;
		152	Coro::schedule;
		153
		154	All the higher-level synchronisation methods (Coro::Semaphore,
		155	Coro::rouse_*...) are actually implemented via C<< ->ready >> and C<<
		156	Coro::schedule >>.
		157
		158	While the coro thread is running it also might get assigned a C-level
		159	thread, or the C-level thread might be unassigned from it, as the Coro
		160	runtime wishes. A C-level thread needs to be assigned when your perl
		161	thread calls into some C-level function and that function in turn calls
		162	perl and perl then wants to switch coroutines. This happens most often
		163	when you run an event loop and block in the callback, or when perl
		164	itself calls some function such as C<AUTOLOAD> or methods via the C<tie>
		165	mechanism.
		166
		167	=item 4. Termination
		168
		169	Many threads actually terminate after some time. There are a number of
		170	ways to terminate a coro thread, the simplest is returning from the
		171	top-level code reference:
		172
		173	async {
		174	# after returning from here, the coro thread is terminated
		175	};
		176
		177	async {
		178	return if 0.5 < rand; # terminate a little earlier, maybe
		179	print "got a chance to print this\n";
		180	# or here
		181	};
		182
		183	Any values returned from the coroutine can be recovered using C<< ->join
		184	>>:
		185
		186	my $coro = async {
		187	"hello, world\n" # return a string
		188	};
		189
		190	my $hello_world = $coro->join;
		191
		192	print $hello_world;
		193
		194	Another way to terminate is to call C<< Coro::terminate >>, which at any
		195	subroutine call nesting level:
		196
		197	async {
		198	Coro::terminate "return value 1", "return value 2";
		199	};
		200
		201	And yet another way is to C<< ->cancel >> (or C<< ->safe_cancel >>) the
		202	coro thread from another thread:
		203
		204	my $coro = async {
		205	exit 1;
		206	};
		207
		208	$coro->cancel; # also accepts values for ->join to retrieve
		209
		210	Cancellation I<can> be dangerous - it's a bit like calling C<exit> without
		211	actually exiting, and might leave C libraries and XS modules in a weird
		212	state. Unlike other thread implementations, however, Coro is exceptionally
		213	safe with regards to cancellation, as perl will always be in a consistent
		214	state, and for those cases where you want to do truly marvellous things
		215	with your coro while it is being cancelled - that is, make sure all
		216	cleanup code is executed from the thread being cancelled - there is even a
		217	C<< ->safe_cancel >> method.
		218
		219	So, cancelling a thread that runs in an XS event loop might not be the
		220	best idea, but any other combination that deals with perl only (cancelling
		221	when a thread is in a C<tie> method or an C<AUTOLOAD> for example) is
		222	safe.
		223
		224	Lastly, a coro thread object that isn't referenced is C<< ->cancel >>'ed
		225	automatically - just like other objects in Perl. This is not such a common
		226	case, however - a running thread is referencedy b C<$Coro::current>, a
		227	thread ready to run is referenced by the ready queue, a thread waiting
		228	on a lock or semaphore is referenced by being in some wait list and so
		229	on. But a thread that isn't in any of those queues gets cancelled:
		230
		231	async {
		232	schedule; # cede to other coros, don't go into the ready queue
		233	};
		234
		235	cede;
		236	# now the async above is destroyed, as it is not referenced by anything.
		237
		238	=item 5. Cleanup
		239
		240	Threads will allocate various resources. Most but not all will be returned
		241	when a thread terminates, during clean-up.
		242
		243	Cleanup is quite similar to throwing an uncaught exception: perl will
		244	work it's way up through all subroutine calls and blocks. On it's way, it
		245	will release all C<my> variables, undo all C<local>'s and free any other
		246	resources truly local to the thread.
		247
		248	So, a common way to free resources is to keep them referenced only by my
38	variables.	249	variables:
		250
		251	async {
		252	my $big_cache = new Cache ...;
		253	};
		254
		255	If there are no other references, then the C<$big_cache> object will be
		256	freed when the thread terminates, regardless of how it does so.
		257
		258	What it does C<NOT> do is unlock any Coro::Semaphores or similar
		259	resources, but that's where the C<guard> methods come in handy:
		260
		261	my $sem = new Coro::Semaphore;
		262
		263	async {
		264	my $lock_guard = $sem->guard;
		265	# if we reutrn, or die or get cancelled, here,
		266	# then the semaphore will be "up"ed.
		267	};
		268
		269	The C<Guard::guard> function comes in handy for any custom cleanup you
		270	might want to do (but you cannot switch to other coroutines form those
		271	code blocks):
		272
		273	async {
		274	my $window = new Gtk2::Window "toplevel";
		275	# The window will not be cleaned up automatically, even when $window
		276	# gets freed, so use a guard to ensure it's destruction
		277	# in case of an error:
		278	my $window_guard = Guard::guard { $window->destroy };
		279
		280	# we are safe here
		281	};
		282
		283	Last not least, C<local> can often be handy, too, e.g. when temporarily
		284	replacing the coro thread description:
		285
		286	sub myfunction {
		287	local $Coro::current->{desc} = "inside myfunction(@_)";
		288
		289	# if we return or die here, the description will be restored
		290	}
		291
		292	=item 6. Viva La Zombie Muerte
		293
		294	Even after a thread has terminated and cleaned up its resources, the Coro
		295	object still is there and stores the return values of the thread.
		296
		297	The means the Coro object gets freed automatically when the thread has
		298	terminated and cleaned up and there arenot other references.
		299
		300	If there are, the Coro object will stay around, and you can call C<<
		301	->join >> as many times as you wish to retrieve the result values:
		302
		303	async {
		304	print "hi\n";
		305	1
		306	};
		307
		308	# run the async above, and free everything before returning
		309	# from Coro::cede:
		310	Coro::cede;
		311
		312	{
		313	my $coro = async {
		314	print "hi\n";
		315	1
		316	};
		317
		318	# run the async above, and clean up, but do not free the coro
		319	# object:
		320	Coro::cede;
		321
		322	# optionally retrieve the result values
		323	my @results = $coro->join;
		324
		325	# now $coro goes out of scope, and presumably gets freed
		326	};
		327
		328	=back
39		329
40	=cut	330	=cut
41		331
42	package Coro;	332	package Coro;
43		333
44	use strict;	334	use common::sense;
45	no warnings "uninitialized";	335
		336	use Carp ();
		337
		338	use Guard ();
46		339
47	use Coro::State;	340	use Coro::State;
48		341
49	use base qw(Coro::State Exporter);	342	use base qw(Coro::State Exporter);
50		343
51	our $idle; # idle handler	344	our $idle; # idle handler
52	our $main; # main coroutine	345	our $main; # main coro
53	our $current; # current coroutine	346	our $current; # current coro
54		347
55	our $VERSION = '3.7';	348	our $VERSION = 5.372;
56		349
57	our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub);	350	our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub rouse_cb rouse_wait);
58	our %EXPORT_TAGS = (	351	our %EXPORT_TAGS = (
59	prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],	352	prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
60	);	353	);
61	our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));	354	our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
62		355
63	{	356	=head1 GLOBAL VARIABLES
64	my @async;
65	my $init;
66
67	# this way of handling attributes simply is NOT scalable ;()
68	sub import {
69	no strict 'refs';
70
71	Coro->export_to_level (1, @_);
72
73	my $old = *{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"}{CODE};
74	*{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"} = sub {
75	my ($package, $ref) = (shift, shift);
76	my @attrs;
77	for (@_) {
78	if ($_ eq "Coro") {
79	push @async, $ref;
80	unless ($init++) {
81	eval q{
82	sub INIT {
83	&async(pop @async) while @async;
84	}
85	};
86	}
87	} else {
88	push @attrs, $_;
89	}
90	}
91	return $old ? $old->($package, $ref, @attrs) : @attrs;
92	};
93	}
94
95	}
96		357
97	=over 4	358	=over 4
98		359
99	=item $main	360	=item $Coro::main
100		361
101	This coroutine represents the main program.	362	This variable stores the Coro object that represents the main
		363	program. While you cna C<ready> it and do most other things you can do to
		364	coro, it is mainly useful to compare again C<$Coro::current>, to see
		365	whether you are running in the main program or not.
102		366
103	=cut	367	=cut
104		368
105	$main = new Coro;	369	# $main is now being initialised by Coro::State
106		370
107	=item $current (or as function: current)	371	=item $Coro::current
108		372
109	The current coroutine (the last coroutine switched to). The initial value	373	The Coro object representing the current coro (the last
		374	coro that the Coro scheduler switched to). The initial value is
110	is C<$main> (of course).	375	C<$Coro::main> (of course).
111		376
112	This variable is B<strictly> I<read-only>. It is provided for performance	377	This variable is B<strictly> I<read-only>. You can take copies of the
113	reasons. If performance is not essential you are encouraged to use the	378	value stored in it and use it as any other Coro object, but you must
114	C<Coro::current> function instead.	379	not otherwise modify the variable itself.
115		380
116	=cut	381	=cut
117		382
118	# maybe some other module used Coro::Specific before...
119	$main->{specific} = $current->{specific}
120	if $current;
121
122	_set_current $main;
123
124	sub current() { $current }	383	sub current() { $current } # [DEPRECATED]
125		384
126	=item $idle	385	=item $Coro::idle
127		386
128	A callback that is called whenever the scheduler finds no ready coroutines	387	This variable is mainly useful to integrate Coro into event loops. It is
		388	usually better to rely on L<Coro::AnyEvent> or L<Coro::EV>, as this is
		389	pretty low-level functionality.
		390
		391	This variable stores a Coro object that is put into the ready queue when
		392	there are no other ready threads (without invoking any ready hooks).
		393
129	to run. The default implementation prints "FATAL: deadlock detected" and	394	The default implementation dies with "FATAL: deadlock detected.", followed
130	exits, because the program has no other way to continue.	395	by a thread listing, because the program has no other way to continue.
131		396
132	This hook is overwritten by modules such as C<Coro::Timer> and	397	This hook is overwritten by modules such as C<Coro::EV> and
133	C<Coro::Event> to wait on an external event that hopefully wake up a	398	C<Coro::AnyEvent> to wait on an external event that hopefully wakes up a
134	coroutine so the scheduler can run it.	399	coro so the scheduler can run it.
135		400
136	Please note that if your callback recursively invokes perl (e.g. for event	401	See L<Coro::EV> or L<Coro::AnyEvent> for examples of using this technique.
137	handlers), then it must be prepared to be called recursively.
138		402
139	=cut	403	=cut
140		404
141	$idle = sub {	405	# ||= because other modules could have provided their own by now
142	require Carp;	406	$idle ||= new Coro sub {
143	Carp::croak ("FATAL: deadlock detected");	407	require Coro::Debug;
		408	die "FATAL: deadlock detected.\n"
		409	. Coro::Debug::ps_listing ();
144	};	410	};
145		411
146	sub _cancel {
147	my ($self) = @_;
148
149	# free coroutine data and mark as destructed
150	$self->_destroy
151	or return;
152
153	# call all destruction callbacks
154	$_->(@{$self->{status}})
155	for @{(delete $self->{destroy_cb}) || []};
156	}
157
158	# this coroutine is necessary because a coroutine	412	# this coro is necessary because a coro
159	# cannot destroy itself.	413	# cannot destroy itself.
160	my @destroy;	414	our @destroy;
161	my $manager;	415	our $manager;
162		416
163	$manager = new Coro sub {	417	$manager = new Coro sub {
164	while () {	418	while () {
165	(shift @destroy)->_cancel	419	_destroy shift @destroy
166	while @destroy;	420	while @destroy;
167		421
168	&schedule;	422	&schedule;
169	}	423	}
170	};	424	};
171		425	$manager->{desc} = "[coro manager]";
172	$manager->prio (PRIO_MAX);	426	$manager->prio (PRIO_MAX);
173		427
174	# static methods. not really.
175
176	=back	428	=back
177		429
178	=head2 STATIC METHODS	430	=head1 SIMPLE CORO CREATION
179
180	Static methods are actually functions that operate on the current coroutine only.
181		431
182	=over 4	432	=over 4
183		433
184	=item async { ... } [@args...]	434	=item async { ... } [@args...]
185		435
186	Create a new asynchronous coroutine and return it's coroutine object	436	Create a new coro and return its Coro object (usually
187	(usually unused). When the sub returns the new coroutine is automatically	437	unused). The coro will be put into the ready queue, so
		438	it will start running automatically on the next scheduler run.
		439
		440	The first argument is a codeblock/closure that should be executed in the
		441	coro. When it returns argument returns the coro is automatically
188	terminated.	442	terminated.
189		443
		444	The remaining arguments are passed as arguments to the closure.
		445
		446	See the C<Coro::State::new> constructor for info about the coro
		447	environment in which coro are executed.
		448
190	Calling C<exit> in a coroutine will do the same as calling exit outside	449	Calling C<exit> in a coro will do the same as calling exit outside
191	the coroutine. Likewise, when the coroutine dies, the program will exit,	450	the coro. Likewise, when the coro dies, the program will exit,
192	just as it would in the main program.	451	just as it would in the main program.
193		452
		453	If you do not want that, you can provide a default C<die> handler, or
		454	simply avoid dieing (by use of C<eval>).
		455
194	# create a new coroutine that just prints its arguments	456	Example: Create a new coro that just prints its arguments.
		457
195	async {	458	async {
196	print "@_\n";	459	print "@_\n";
197	} 1,2,3,4;	460	} 1,2,3,4;
198		461
		462	=item async_pool { ... } [@args...]
		463
		464	Similar to C<async>, but uses a coro pool, so you should not call
		465	terminate or join on it (although you are allowed to), and you get a
		466	coro that might have executed other code already (which can be good
		467	or bad :).
		468
		469	On the plus side, this function is about twice as fast as creating (and
		470	destroying) a completely new coro, so if you need a lot of generic
		471	coros in quick successsion, use C<async_pool>, not C<async>.
		472
		473	The code block is executed in an C<eval> context and a warning will be
		474	issued in case of an exception instead of terminating the program, as
		475	C<async> does. As the coro is being reused, stuff like C<on_destroy>
		476	will not work in the expected way, unless you call terminate or cancel,
		477	which somehow defeats the purpose of pooling (but is fine in the
		478	exceptional case).
		479
		480	The priority will be reset to C<0> after each run, tracing will be
		481	disabled, the description will be reset and the default output filehandle
		482	gets restored, so you can change all these. Otherwise the coro will
		483	be re-used "as-is": most notably if you change other per-coro global
		484	stuff such as C<$/> you I<must needs> revert that change, which is most
		485	simply done by using local as in: C<< local $/ >>.
		486
		487	The idle pool size is limited to C<8> idle coros (this can be
		488	adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle
		489	coros as required.
		490
		491	If you are concerned about pooled coros growing a lot because a
		492	single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
		493	{ terminate }> once per second or so to slowly replenish the pool. In
		494	addition to that, when the stacks used by a handler grows larger than 32kb
		495	(adjustable via $Coro::POOL_RSS) it will also be destroyed.
		496
199	=cut	497	=cut
200		498
201	sub async(&@) {
202	my $coro = new Coro @_;
203	$coro->ready;
204	$coro
205	}
206
207	=item async_pool { ... } [@args...]
208
209	Similar to C<async>, but uses a coroutine pool, so you should not call
210	terminate or join (although you are allowed to), and you get a coroutine
211	that might have executed other code already (which can be good or bad :).
212
213	Also, the block is executed in an C<eval> context and a warning will be
214	issued in case of an exception instead of terminating the program, as
215	C<async> does. As the coroutine is being reused, stuff like C<on_destroy>
216	will not work in the expected way, unless you call terminate or cancel,
217	which somehow defeats the purpose of pooling.
218
219	The priority will be reset to C<0> after each job, otherwise the coroutine
220	will be re-used "as-is".
221
222	The pool size is limited to 8 idle coroutines (this can be adjusted by
223	changing $Coro::POOL_SIZE), and there can be as many non-idle coros as
224	required.
225
226	If you are concerned about pooled coroutines growing a lot because a
227	single C<async_pool> used a lot of stackspace you can e.g. C<async_pool {
228	terminate }> once per second or so to slowly replenish the pool.
229
230	=cut
231
232	our $POOL_SIZE = 8;	499	our $POOL_SIZE = 8;
		500	our $POOL_RSS = 32 * 1024;
233	our @pool;	501	our @async_pool;
234		502
235	sub pool_handler {	503	sub pool_handler {
236	while () {	504	while () {
237	eval {	505	eval {
238	my ($cb, @arg) = @{ delete $current->{_invoke} or return };	506	&{&_pool_handler} while 1;
239	$cb->(@arg);
240	};	507	};
		508
241	warn $@ if $@;	509	warn $@ if $@;
242
243	last if @pool >= $POOL_SIZE;
244	push @pool, $current;
245
246	$current->save (Coro::State::SAVE_DEF);
247	$current->prio (0);
248	schedule;
249	}	510	}
250	}	511	}
251		512
252	sub async_pool(&@) {	513	=back
253	# this is also inlined into the unlock_scheduler	514
254	my $coro = (pop @pool) || do {	515	=head1 STATIC METHODS
255	my $coro = new Coro \&pool_handler;	516
256	$coro->{desc} = "async_pool";	517	Static methods are actually functions that implicitly operate on the
257	$coro	518	current coro.
		519
		520	=over 4
		521
		522	=item schedule
		523
		524	Calls the scheduler. The scheduler will find the next coro that is
		525	to be run from the ready queue and switches to it. The next coro
		526	to be run is simply the one with the highest priority that is longest
		527	in its ready queue. If there is no coro ready, it will call the
		528	C<$Coro::idle> hook.
		529
		530	Please note that the current coro will I<not> be put into the ready
		531	queue, so calling this function usually means you will never be called
		532	again unless something else (e.g. an event handler) calls C<< ->ready >>,
		533	thus waking you up.
		534
		535	This makes C<schedule> I<the> generic method to use to block the current
		536	coro and wait for events: first you remember the current coro in
		537	a variable, then arrange for some callback of yours to call C<< ->ready
		538	>> on that once some event happens, and last you call C<schedule> to put
		539	yourself to sleep. Note that a lot of things can wake your coro up,
		540	so you need to check whether the event indeed happened, e.g. by storing the
		541	status in a variable.
		542
		543	See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks.
		544
		545	=item cede
		546
		547	"Cede" to other coros. This function puts the current coro into
		548	the ready queue and calls C<schedule>, which has the effect of giving
		549	up the current "timeslice" to other coros of the same or higher
		550	priority. Once your coro gets its turn again it will automatically be
		551	resumed.
		552
		553	This function is often called C<yield> in other languages.
		554
		555	=item Coro::cede_notself
		556
		557	Works like cede, but is not exported by default and will cede to I<any>
		558	coro, regardless of priority. This is useful sometimes to ensure
		559	progress is made.
		560
		561	=item terminate [arg...]
		562
		563	Terminates the current coro with the given status values (see
		564	L<cancel>). The values will not be copied, but referenced directly.
		565
		566	=item Coro::on_enter BLOCK, Coro::on_leave BLOCK
		567
		568	These function install enter and leave winders in the current scope. The
		569	enter block will be executed when on_enter is called and whenever the
		570	current coro is re-entered by the scheduler, while the leave block is
		571	executed whenever the current coro is blocked by the scheduler, and
		572	also when the containing scope is exited (by whatever means, be it exit,
		573	die, last etc.).
		574
		575	I<Neither invoking the scheduler, nor exceptions, are allowed within those
		576	BLOCKs>. That means: do not even think about calling C<die> without an
		577	eval, and do not even think of entering the scheduler in any way.
		578
		579	Since both BLOCKs are tied to the current scope, they will automatically
		580	be removed when the current scope exits.
		581
		582	These functions implement the same concept as C<dynamic-wind> in scheme
		583	does, and are useful when you want to localise some resource to a specific
		584	coro.
		585
		586	They slow down thread switching considerably for coros that use them
		587	(about 40% for a BLOCK with a single assignment, so thread switching is
		588	still reasonably fast if the handlers are fast).
		589
		590	These functions are best understood by an example: The following function
		591	will change the current timezone to "Antarctica/South_Pole", which
		592	requires a call to C<tzset>, but by using C<on_enter> and C<on_leave>,
		593	which remember/change the current timezone and restore the previous
		594	value, respectively, the timezone is only changed for the coro that
		595	installed those handlers.
		596
		597	use POSIX qw(tzset);
		598
		599	async {
		600	my $old_tz; # store outside TZ value here
		601
		602	Coro::on_enter {
		603	$old_tz = $ENV{TZ}; # remember the old value
		604
		605	$ENV{TZ} = "Antarctica/South_Pole";
		606	tzset; # enable new value
		607	};
		608
		609	Coro::on_leave {
		610	$ENV{TZ} = $old_tz;
		611	tzset; # restore old value
		612	};
		613
		614	# at this place, the timezone is Antarctica/South_Pole,
		615	# without disturbing the TZ of any other coro.
		616	};
		617
		618	This can be used to localise about any resource (locale, uid, current
		619	working directory etc.) to a block, despite the existance of other
		620	coros.
		621
		622	Another interesting example implements time-sliced multitasking using
		623	interval timers (this could obviously be optimised, but does the job):
		624
		625	# "timeslice" the given block
		626	sub timeslice(&) {
		627	use Time::HiRes ();
		628
		629	Coro::on_enter {
		630	# on entering the thread, we set an VTALRM handler to cede
		631	$SIG{VTALRM} = sub { cede };
		632	# and then start the interval timer
		633	Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0.01, 0.01;
		634	};
		635	Coro::on_leave {
		636	# on leaving the thread, we stop the interval timer again
		637	Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0, 0;
		638	};
		639
		640	&{+shift};
		641	}
		642
		643	# use like this:
		644	timeslice {
		645	# The following is an endless loop that would normally
		646	# monopolise the process. Since it runs in a timesliced
		647	# environment, it will regularly cede to other threads.
		648	while () { }
258	};	649	};
259		650
260	$coro->{_invoke} = [@_];
261	$coro->ready;
262		651
263	$coro	652	=item killall
		653
		654	Kills/terminates/cancels all coros except the currently running one.
		655
		656	Note that while this will try to free some of the main interpreter
		657	resources if the calling coro isn't the main coro, but one
		658	cannot free all of them, so if a coro that is not the main coro
		659	calls this function, there will be some one-time resource leak.
		660
		661	=cut
		662
		663	sub killall {
		664	for (Coro::State::list) {
		665	$_->cancel
		666	if $_ != $current && UNIVERSAL::isa $_, "Coro";
		667	}
264	}	668	}
265		669
266	=item schedule	670	=back
267		671
268	Calls the scheduler. Please note that the current coroutine will not be put	672	=head1 CORO OBJECT METHODS
269	into the ready queue, so calling this function usually means you will
270	never be called again unless something else (e.g. an event handler) calls
271	ready.
272		673
273	The canonical way to wait on external events is this:	674	These are the methods you can call on coro objects (or to create
		675	them).
274		676
275	{	677	=over 4
276	# remember current coroutine
277	my $current = $Coro::current;
278		678
279	# register a hypothetical event handler	679	=item new Coro \&sub [, @args...]
280	on_event_invoke sub {
281	# wake up sleeping coroutine
282	$current->ready;
283	undef $current;
284	};
285		680
286	# call schedule until event occurred.	681	Create a new coro and return it. When the sub returns, the coro
287	# in case we are woken up for other reasons	682	automatically terminates as if C<terminate> with the returned values were
288	# (current still defined), loop.	683	called. To make the coro run you must first put it into the ready
289	Coro::schedule while $current;	684	queue by calling the ready method.
290	}
291		685
292	=item cede	686	See C<async> and C<Coro::State::new> for additional info about the
293		687	coro environment.
294	"Cede" to other coroutines. This function puts the current coroutine into the
295	ready queue and calls C<schedule>, which has the effect of giving up the
296	current "timeslice" to other coroutines of the same or higher priority.
297
298	Returns true if at least one coroutine switch has happened.
299
300	=item Coro::cede_notself
301
302	Works like cede, but is not exported by default and will cede to any
303	coroutine, regardless of priority, once.
304
305	Returns true if at least one coroutine switch has happened.
306
307	=item terminate [arg...]
308
309	Terminates the current coroutine with the given status values (see L<cancel>).
310		688
311	=cut	689	=cut
312		690
313	sub terminate {	691	sub _coro_run {
314	$current->cancel (@_);
315	}
316
317	=back
318
319	# dynamic methods
320
321	=head2 COROUTINE METHODS
322
323	These are the methods you can call on coroutine objects.
324
325	=over 4
326
327	=item new Coro \&sub [, @args...]
328
329	Create a new coroutine and return it. When the sub returns the coroutine
330	automatically terminates as if C<terminate> with the returned values were
331	called. To make the coroutine run you must first put it into the ready queue
332	by calling the ready method.
333
334	See C<async> for additional discussion.
335
336	=cut
337
338	sub _run_coro {
339	terminate &{+shift};	692	terminate &{+shift};
340	}	693	}
341		694
342	sub new {
343	my $class = shift;
344
345	$class->SUPER::new (\&_run_coro, @_)
346	}
347
348	=item $success = $coroutine->ready	695	=item $success = $coro->ready
349		696
350	Put the given coroutine into the ready queue (according to it's priority)	697	Put the given coro into the end of its ready queue (there is one
351	and return true. If the coroutine is already in the ready queue, do nothing	698	queue for each priority) and return true. If the coro is already in
352	and return false.	699	the ready queue, do nothing and return false.
353		700
		701	This ensures that the scheduler will resume this coro automatically
		702	once all the coro of higher priority and all coro of the same
		703	priority that were put into the ready queue earlier have been resumed.
		704
		705	=item $coro->suspend
		706
		707	Suspends the specified coro. A suspended coro works just like any other
		708	coro, except that the scheduler will not select a suspended coro for
		709	execution.
		710
		711	Suspending a coro can be useful when you want to keep the coro from
		712	running, but you don't want to destroy it, or when you want to temporarily
		713	freeze a coro (e.g. for debugging) to resume it later.
		714
		715	A scenario for the former would be to suspend all (other) coros after a
		716	fork and keep them alive, so their destructors aren't called, but new
		717	coros can be created.
		718
		719	=item $coro->resume
		720
		721	If the specified coro was suspended, it will be resumed. Note that when
		722	the coro was in the ready queue when it was suspended, it might have been
		723	unreadied by the scheduler, so an activation might have been lost.
		724
		725	To avoid this, it is best to put a suspended coro into the ready queue
		726	unconditionally, as every synchronisation mechanism must protect itself
		727	against spurious wakeups, and the one in the Coro family certainly do
		728	that.
		729
		730	=item $state->is_new
		731
		732	Returns true iff this Coro object is "new", i.e. has never been run
		733	yet. Those states basically consist of only the code reference to call and
		734	the arguments, but consumes very little other resources. New states will
		735	automatically get assigned a perl interpreter when they are transfered to.
		736
		737	=item $state->is_zombie
		738
		739	Returns true iff the Coro object has been cancelled, i.e.
		740	it's resources freed because they were C<cancel>'ed, C<terminate>'d,
		741	C<safe_cancel>'ed or simply went out of scope.
		742
		743	The name "zombie" stems from UNIX culture, where a process that has
		744	exited and only stores and exit status and no other resources is called a
		745	"zombie".
		746
354	=item $is_ready = $coroutine->is_ready	747	=item $is_ready = $coro->is_ready
355		748
356	Return wether the coroutine is currently the ready queue or not,	749	Returns true iff the Coro object is in the ready queue. Unless the Coro
		750	object gets destroyed, it will eventually be scheduled by the scheduler.
357		751
		752	=item $is_running = $coro->is_running
		753
		754	Returns true iff the Coro object is currently running. Only one Coro object
		755	can ever be in the running state (but it currently is possible to have
		756	multiple running Coro::States).
		757
		758	=item $is_suspended = $coro->is_suspended
		759
		760	Returns true iff this Coro object has been suspended. Suspended Coros will
		761	not ever be scheduled.
		762
358	=item $coroutine->cancel (arg...)	763	=item $coro->cancel (arg...)
359		764
360	Terminates the given coroutine and makes it return the given arguments as	765	Terminates the given Coro thread and makes it return the given arguments as
361	status (default: the empty list). Never returns if the coroutine is the	766	status (default: an empty list). Never returns if the Coro is the
362	current coroutine.	767	current Coro.
363		768
364	=cut	769	This is a rather brutal way to free a coro, with some limitations - if
		770	the thread is inside a C callback that doesn't expect to be canceled,
		771	bad things can happen, or if the cancelled thread insists on running
		772	complicated cleanup handlers that rely on it'S thread context, things will
		773	not work.
365		774
366	sub cancel {	775	Any cleanup code being run (e.g. from C<guard> blocks) will be run without
367	my $self = shift;	776	a thread context, and is not allowed to switch to other threads. On the
368	$self->{status} = [@_];	777	plus side, C<< ->cancel >> will always clean up the thread, no matter
		778	what. If your cleanup code is complex or you want to avoid cancelling a
		779	C-thread that doesn't know how to clean up itself, it can be better to C<<
		780	->throw >> an exception, or use C<< ->safe_cancel >>.
369		781
370	if ($current == $self) {	782	The arguments to C<< ->cancel >> are not copied, but instead will
371	push @destroy, $self;	783	be referenced directly (e.g. if you pass C<$var> and after the call
372	$manager->ready;	784	change that variable, then you might change the return values passed to
373	&schedule while 1;	785	e.g. C<join>, so don't do that).
374	} else {	786
375	$self->_cancel;	787	The resources of the Coro are usually freed (or destructed) before this
		788	call returns, but this can be delayed for an indefinite amount of time, as
		789	in some cases the manager thread has to run first to actually destruct the
		790	Coro object.
		791
		792	=item $coro->safe_cancel ($arg...)
		793
		794	Works mostly like C<< ->cancel >>, but is inherently "safer", and
		795	consequently, can fail with an exception in cases the thread is not in a
		796	cancellable state.
		797
		798	This method works a bit like throwing an exception that cannot be caught
		799	- specifically, it will clean up the thread from within itself, so
		800	all cleanup handlers (e.g. C<guard> blocks) are run with full thread
		801	context and can block if they wish. The downside is that there is no
		802	guarantee that the thread can be cancelled when you call this method, and
		803	therefore, it might fail. It is also considerably slower than C<cancel> or
		804	C<terminate>.
		805
		806	A thread is in a safe-cancellable state if it either hasn't been run yet,
		807	or it has no C context attached and is inside an SLF function.
		808
		809	The latter two basically mean that the thread isn't currently inside a
		810	perl callback called from some C function (usually via some XS modules)
		811	and isn't currently executing inside some C function itself (via Coro's XS
		812	API).
		813
		814	This call returns true when it could cancel the thread, or croaks with an
		815	error otherwise (i.e. it either returns true or doesn't return at all).
		816
		817	Why the weird interface? Well, there are two common models on how and
		818	when to cancel things. In the first, you have the expectation that your
		819	coro thread can be cancelled when you want to cancel it - if the thread
		820	isn't cancellable, this would be a bug somewhere, so C<< ->safe_cancel >>
		821	croaks to notify of the bug.
		822
		823	In the second model you sometimes want to ask nicely to cancel a thread,
		824	but if it's not a good time, well, then don't cancel. This can be done
		825	relatively easy like this:
		826
		827	if (! eval { $coro->safe_cancel }) {
		828	warn "unable to cancel thread: $@";
376	}	829	}
377	}
378		830
		831	However, what you never should do is first try to cancel "safely" and
		832	if that fails, cancel the "hard" way with C<< ->cancel >>. That makes
		833	no sense: either you rely on being able to execute cleanup code in your
		834	thread context, or you don't. If you do, then C<< ->safe_cancel >> is the
		835	only way, and if you don't, then C<< ->cancel >> is always faster and more
		836	direct.
		837
		838	=item $coro->schedule_to
		839
		840	Puts the current coro to sleep (like C<Coro::schedule>), but instead
		841	of continuing with the next coro from the ready queue, always switch to
		842	the given coro object (regardless of priority etc.). The readyness
		843	state of that coro isn't changed.
		844
		845	This is an advanced method for special cases - I'd love to hear about any
		846	uses for this one.
		847
		848	=item $coro->cede_to
		849
		850	Like C<schedule_to>, but puts the current coro into the ready
		851	queue. This has the effect of temporarily switching to the given
		852	coro, and continuing some time later.
		853
		854	This is an advanced method for special cases - I'd love to hear about any
		855	uses for this one.
		856
		857	=item $coro->throw ([$scalar])
		858
		859	If C<$throw> is specified and defined, it will be thrown as an exception
		860	inside the coro at the next convenient point in time. Otherwise
		861	clears the exception object.
		862
		863	Coro will check for the exception each time a schedule-like-function
		864	returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down
		865	>>, C<< Coro::Handle->readable >> and so on. Most of those functions (all
		866	that are part of Coro itself) detect this case and return early in case an
		867	exception is pending.
		868
		869	The exception object will be thrown "as is" with the specified scalar in
		870	C<$@>, i.e. if it is a string, no line number or newline will be appended
		871	(unlike with C<die>).
		872
		873	This can be used as a softer means than either C<cancel> or C<safe_cancel
		874	>to ask a coro to end itself, although there is no guarantee that the
		875	exception will lead to termination, and if the exception isn't caught it
		876	might well end the whole program.
		877
		878	You might also think of C<throw> as being the moral equivalent of
		879	C<kill>ing a coro with a signal (in this case, a scalar).
		880
379	=item $coroutine->join	881	=item $coro->join
380		882
381	Wait until the coroutine terminates and return any values given to the	883	Wait until the coro terminates and return any values given to the
382	C<terminate> or C<cancel> functions. C<join> can be called multiple times	884	C<terminate> or C<cancel> functions. C<join> can be called concurrently
383	from multiple coroutine.	885	from multiple threads, and all will be resumed and given the status
		886	return once the C<$coro> terminates.
384		887
385	=cut
386
387	sub join {
388	my $self = shift;
389
390	unless ($self->{status}) {
391	my $current = $current;
392
393	push @{$self->{destroy_cb}}, sub {
394	$current->ready;
395	undef $current;
396	};
397
398	&schedule while $current;
399	}
400
401	wantarray ? @{$self->{status}} : $self->{status}[0];
402	}
403
404	=item $coroutine->on_destroy (\&cb)	888	=item $coro->on_destroy (\&cb)
405		889
406	Registers a callback that is called when this coroutine gets destroyed,	890	Registers a callback that is called when this coro thread gets destroyed,
407	but before it is joined. The callback gets passed the terminate arguments,	891	that is, after it's resources have been freed but before it is joined. The
408	if any.	892	callback gets passed the terminate/cancel arguments, if any, and I<must
		893	not> die, under any circumstances.
409		894
410	=cut	895	There can be any number of C<on_destroy> callbacks per coro, and there is
		896	no way currently to remove a callback once added.
411		897
412	sub on_destroy {
413	my ($self, $cb) = @_;
414
415	push @{ $self->{destroy_cb} }, $cb;
416	}
417
418	=item $oldprio = $coroutine->prio ($newprio)	898	=item $oldprio = $coro->prio ($newprio)
419		899
420	Sets (or gets, if the argument is missing) the priority of the	900	Sets (or gets, if the argument is missing) the priority of the
421	coroutine. Higher priority coroutines get run before lower priority	901	coro thread. Higher priority coro get run before lower priority
422	coroutines. Priorities are small signed integers (currently -4 .. +3),	902	coros. Priorities are small signed integers (currently -4 .. +3),
423	that you can refer to using PRIO_xxx constants (use the import tag :prio	903	that you can refer to using PRIO_xxx constants (use the import tag :prio
424	to get then):	904	to get then):
425		905
426	PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN	906	PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN
427	3 > 1 > 0 > -1 > -3 > -4	907	3 > 1 > 0 > -1 > -3 > -4
428		908
429	# set priority to HIGH	909	# set priority to HIGH
430	current->prio(PRIO_HIGH);	910	current->prio (PRIO_HIGH);
431		911
432	The idle coroutine ($Coro::idle) always has a lower priority than any	912	The idle coro thread ($Coro::idle) always has a lower priority than any
433	existing coroutine.	913	existing coro.
434		914
435	Changing the priority of the current coroutine will take effect immediately,	915	Changing the priority of the current coro will take effect immediately,
436	but changing the priority of coroutines in the ready queue (but not	916	but changing the priority of a coro in the ready queue (but not running)
437	running) will only take effect after the next schedule (of that	917	will only take effect after the next schedule (of that coro). This is a
438	coroutine). This is a bug that will be fixed in some future version.	918	bug that will be fixed in some future version.
439		919
440	=item $newprio = $coroutine->nice ($change)	920	=item $newprio = $coro->nice ($change)
441		921
442	Similar to C<prio>, but subtract the given value from the priority (i.e.	922	Similar to C<prio>, but subtract the given value from the priority (i.e.
443	higher values mean lower priority, just as in unix).	923	higher values mean lower priority, just as in UNIX's nice command).
444		924
445	=item $olddesc = $coroutine->desc ($newdesc)	925	=item $olddesc = $coro->desc ($newdesc)
446		926
447	Sets (or gets in case the argument is missing) the description for this	927	Sets (or gets in case the argument is missing) the description for this
448	coroutine. This is just a free-form string you can associate with a coroutine.	928	coro thread. This is just a free-form string you can associate with a
		929	coro.
		930
		931	This method simply sets the C<< $coro->{desc} >> member to the given
		932	string. You can modify this member directly if you wish, and in fact, this
		933	is often preferred to indicate major processing states that cna then be
		934	seen for example in a L<Coro::Debug> session:
		935
		936	sub my_long_function {
		937	local $Coro::current->{desc} = "now in my_long_function";
		938	...
		939	$Coro::current->{desc} = "my_long_function: phase 1";
		940	...
		941	$Coro::current->{desc} = "my_long_function: phase 2";
		942	...
		943	}
449		944
450	=cut	945	=cut
451		946
452	sub desc {	947	sub desc {
453	my $old = $_[0]{desc};	948	my $old = $_[0]{desc};
454	$_[0]{desc} = $_[1] if @_ > 1;	949	$_[0]{desc} = $_[1] if @_ > 1;
455	$old;	950	$old;
456	}	951	}
457		952
		953	sub transfer {
		954	require Carp;
		955	Carp::croak ("You must not call ->transfer on Coro objects. Use Coro::State objects or the ->schedule_to method. Caught");
		956	}
		957
458	=back	958	=back
459		959
460	=head2 GLOBAL FUNCTIONS	960	=head1 GLOBAL FUNCTIONS
461		961
462	=over 4	962	=over 4
463		963
464	=item Coro::nready	964	=item Coro::nready
465		965
466	Returns the number of coroutines that are currently in the ready state,	966	Returns the number of coro that are currently in the ready state,
467	i.e. that can be switched to. The value C<0> means that the only runnable	967	i.e. that can be switched to by calling C<schedule> directory or
		968	indirectly. The value C<0> means that the only runnable coro is the
468	coroutine is the currently running one, so C<cede> would have no effect,	969	currently running one, so C<cede> would have no effect, and C<schedule>
469	and C<schedule> would cause a deadlock unless there is an idle handler	970	would cause a deadlock unless there is an idle handler that wakes up some
470	that wakes up some coroutines.	971	coro.
471		972
472	=item my $guard = Coro::guard { ... }	973	=item my $guard = Coro::guard { ... }
473		974
474	This creates and returns a guard object. Nothing happens until the object	975	This function still exists, but is deprecated. Please use the
475	gets destroyed, in which case the codeblock given as argument will be	976	C<Guard::guard> function instead.
476	executed. This is useful to free locks or other resources in case of a
477	runtime error or when the coroutine gets canceled, as in both cases the
478	guard block will be executed. The guard object supports only one method,
479	C<< ->cancel >>, which will keep the codeblock from being executed.
480
481	Example: set some flag and clear it again when the coroutine gets canceled
482	or the function returns:
483
484	sub do_something {
485	my $guard = Coro::guard { $busy = 0 };
486	$busy = 1;
487
488	# do something that requires $busy to be true
489	}
490		977
491	=cut	978	=cut
492		979
493	sub guard(&) {	980	BEGIN { *guard = \&Guard::guard }
494	bless \(my $cb = $_[0]), "Coro::guard"
495	}
496
497	sub Coro::guard::cancel {
498	${$_[0]} = sub { };
499	}
500
501	sub Coro::guard::DESTROY {
502	${$_[0]}->();
503	}
504
505		981
506	=item unblock_sub { ... }	982	=item unblock_sub { ... }
507		983
508	This utility function takes a BLOCK or code reference and "unblocks" it,	984	This utility function takes a BLOCK or code reference and "unblocks" it,
509	returning the new coderef. This means that the new coderef will return	985	returning a new coderef. Unblocking means that calling the new coderef
510	immediately without blocking, returning nothing, while the original code	986	will return immediately without blocking, returning nothing, while the
511	ref will be called (with parameters) from within its own coroutine.	987	original code ref will be called (with parameters) from within another
		988	coro.
512		989
513	The reason this function exists is that many event libraries (such as the	990	The reason this function exists is that many event libraries (such as
514	venerable L<Event|Event> module) are not coroutine-safe (a weaker form	991	the venerable L<Event|Event> module) are not thread-safe (a weaker form
515	of thread-safety). This means you must not block within event callbacks,	992	of reentrancy). This means you must not block within event callbacks,
516	otherwise you might suffer from crashes or worse.	993	otherwise you might suffer from crashes or worse. The only event library
		994	currently known that is safe to use without C<unblock_sub> is L<EV> (but
		995	you might still run into deadlocks if all event loops are blocked).
		996
		997	Coro will try to catch you when you block in the event loop
		998	("FATAL:$Coro::IDLE blocked itself"), but this is just best effort and
		999	only works when you do not run your own event loop.
517		1000
518	This function allows your callbacks to block by executing them in another	1001	This function allows your callbacks to block by executing them in another
519	coroutine where it is safe to block. One example where blocking is handy	1002	coro where it is safe to block. One example where blocking is handy
520	is when you use the L<Coro::AIO|Coro::AIO> functions to save results to	1003	is when you use the L<Coro::AIO|Coro::AIO> functions to save results to
521	disk.	1004	disk, for example.
522		1005
523	In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when	1006	In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
524	creating event callbacks that want to block.	1007	creating event callbacks that want to block.
		1008
		1009	If your handler does not plan to block (e.g. simply sends a message to
		1010	another coro, or puts some other coro into the ready queue), there is
		1011	no reason to use C<unblock_sub>.
		1012
		1013	Note that you also need to use C<unblock_sub> for any other callbacks that
		1014	are indirectly executed by any C-based event loop. For example, when you
		1015	use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
		1016	provides callbacks that are the result of some event callback, then you
		1017	must not block either, or use C<unblock_sub>.
525		1018
526	=cut	1019	=cut
527		1020
528	our @unblock_queue;	1021	our @unblock_queue;
529		1022
530	# we create a special coro because we want to cede,	1023	# we create a special coro because we want to cede,
531	# to reduce pressure on the coro pool (because most callbacks	1024	# to reduce pressure on the coro pool (because most callbacks
532	# return immediately and can be reused) and because we cannot cede	1025	# return immediately and can be reused) and because we cannot cede
533	# inside an event callback.	1026	# inside an event callback.
534	our $unblock_scheduler = async {	1027	our $unblock_scheduler = new Coro sub {
535	while () {	1028	while () {
536	while (my $cb = pop @unblock_queue) {	1029	while (my $cb = pop @unblock_queue) {
537	# this is an inlined copy of async_pool	1030	&async_pool (@$cb);
538	my $coro = (pop @pool or new Coro \&pool_handler);
539		1031
540	$coro->{_invoke} = $cb;
541	$coro->ready;
542	cede; # for short-lived callbacks, this reduces pressure on the coro pool	1032	# for short-lived callbacks, this reduces pressure on the coro pool
		1033	# as the chance is very high that the async_poll coro will be back
		1034	# in the idle state when cede returns
		1035	cede;
543	}	1036	}
544	schedule; # sleep well	1037	schedule; # sleep well
545	}	1038	}
546	};	1039	};
		1040	$unblock_scheduler->{desc} = "[unblock_sub scheduler]";
547		1041
548	sub unblock_sub(&) {	1042	sub unblock_sub(&) {
549	my $cb = shift;	1043	my $cb = shift;
550		1044
551	sub {	1045	sub {
552	unshift @unblock_queue, [$cb, @_];	1046	unshift @unblock_queue, [$cb, @_];
553	$unblock_scheduler->ready;	1047	$unblock_scheduler->ready;
554	}	1048	}
555	}	1049	}
556		1050
		1051	=item $cb = rouse_cb
		1052
		1053	Create and return a "rouse callback". That's a code reference that,
		1054	when called, will remember a copy of its arguments and notify the owner
		1055	coro of the callback.
		1056
		1057	See the next function.
		1058
		1059	=item @args = rouse_wait [$cb]
		1060
		1061	Wait for the specified rouse callback (or the last one that was created in
		1062	this coro).
		1063
		1064	As soon as the callback is invoked (or when the callback was invoked
		1065	before C<rouse_wait>), it will return the arguments originally passed to
		1066	the rouse callback. In scalar context, that means you get the I<last>
		1067	argument, just as if C<rouse_wait> had a C<return ($a1, $a2, $a3...)>
		1068	statement at the end.
		1069
		1070	See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example.
		1071
557	=back	1072	=back
558		1073
559	=cut	1074	=cut
560		1075
		1076	for my $module (qw(Channel RWLock Semaphore SemaphoreSet Signal Specific)) {
		1077	my $old = defined &{"Coro::$module\::new"} && \&{"Coro::$module\::new"};
		1078
		1079	*{"Coro::$module\::new"} = sub {
		1080	require "Coro/$module.pm";
		1081
		1082	# some modules have their new predefined in State.xs, some don't
		1083	*{"Coro::$module\::new"} = $old
		1084	if $old;
		1085
		1086	goto &{"Coro::$module\::new"};
		1087	};
		1088	}
		1089
561	1;	1090	1;
562		1091
		1092	=head1 HOW TO WAIT FOR A CALLBACK
		1093
		1094	It is very common for a coro to wait for some callback to be
		1095	called. This occurs naturally when you use coro in an otherwise
		1096	event-based program, or when you use event-based libraries.
		1097
		1098	These typically register a callback for some event, and call that callback
		1099	when the event occured. In a coro, however, you typically want to
		1100	just wait for the event, simplyifying things.
		1101
		1102	For example C<< AnyEvent->child >> registers a callback to be called when
		1103	a specific child has exited:
		1104
		1105	my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });
		1106
		1107	But from within a coro, you often just want to write this:
		1108
		1109	my $status = wait_for_child $pid;
		1110
		1111	Coro offers two functions specifically designed to make this easy,
		1112	C<Coro::rouse_cb> and C<Coro::rouse_wait>.
		1113
		1114	The first function, C<rouse_cb>, generates and returns a callback that,
		1115	when invoked, will save its arguments and notify the coro that
		1116	created the callback.
		1117
		1118	The second function, C<rouse_wait>, waits for the callback to be called
		1119	(by calling C<schedule> to go to sleep) and returns the arguments
		1120	originally passed to the callback.
		1121
		1122	Using these functions, it becomes easy to write the C<wait_for_child>
		1123	function mentioned above:
		1124
		1125	sub wait_for_child($) {
		1126	my ($pid) = @_;
		1127
		1128	my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb);
		1129
		1130	my ($rpid, $rstatus) = Coro::rouse_wait;
		1131	$rstatus
		1132	}
		1133
		1134	In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough,
		1135	you can roll your own, using C<schedule>:
		1136
		1137	sub wait_for_child($) {
		1138	my ($pid) = @_;
		1139
		1140	# store the current coro in $current,
		1141	# and provide result variables for the closure passed to ->child
		1142	my $current = $Coro::current;
		1143	my ($done, $rstatus);
		1144
		1145	# pass a closure to ->child
		1146	my $watcher = AnyEvent->child (pid => $pid, cb => sub {
		1147	$rstatus = $_[1]; # remember rstatus
		1148	$done = 1; # mark $rstatus as valud
		1149	});
		1150
		1151	# wait until the closure has been called
		1152	schedule while !$done;
		1153
		1154	$rstatus
		1155	}
		1156
		1157
563	=head1 BUGS/LIMITATIONS	1158	=head1 BUGS/LIMITATIONS
564		1159
565	- you must make very sure that no coro is still active on global	1160	=over 4
566	destruction. very bad things might happen otherwise (usually segfaults).
567		1161
		1162	=item fork with pthread backend
		1163
		1164	When Coro is compiled using the pthread backend (which isn't recommended
		1165	but required on many BSDs as their libcs are completely broken), then
		1166	coro will not survive a fork. There is no known workaround except to
		1167	fix your libc and use a saner backend.
		1168
		1169	=item perl process emulation ("threads")
		1170
568	- this module is not thread-safe. You should only ever use this module	1171	This module is not perl-pseudo-thread-safe. You should only ever use this
569	from the same thread (this requirement might be loosened in the future	1172	module from the first thread (this requirement might be removed in the
570	to allow per-thread schedulers, but Coro::State does not yet allow	1173	future to allow per-thread schedulers, but Coro::State does not yet allow
571	this).	1174	this). I recommend disabling thread support and using processes, as having
		1175	the windows process emulation enabled under unix roughly halves perl
		1176	performance, even when not used.
		1177
		1178	Attempts to use threads created in another emulated process will crash
		1179	("cleanly", with a null pointer exception).
		1180
		1181	=item coro switching is not signal safe
		1182
		1183	You must not switch to another coro from within a signal handler (only
		1184	relevant with %SIG - most event libraries provide safe signals), I<unless>
		1185	you are sure you are not interrupting a Coro function.
		1186
		1187	That means you I<MUST NOT> call any function that might "block" the
		1188	current coro - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or
		1189	anything that calls those. Everything else, including calling C<ready>,
		1190	works.
		1191
		1192	=back
		1193
		1194
		1195	=head1 WINDOWS PROCESS EMULATION
		1196
		1197	A great many people seem to be confused about ithreads (for example, Chip
		1198	Salzenberg called me unintelligent, incapable, stupid and gullible,
		1199	while in the same mail making rather confused statements about perl
		1200	ithreads (for example, that memory or files would be shared), showing his
		1201	lack of understanding of this area - if it is hard to understand for Chip,
		1202	it is probably not obvious to everybody).
		1203
		1204	What follows is an ultra-condensed version of my talk about threads in
		1205	scripting languages given on the perl workshop 2009:
		1206
		1207	The so-called "ithreads" were originally implemented for two reasons:
		1208	first, to (badly) emulate unix processes on native win32 perls, and
		1209	secondly, to replace the older, real thread model ("5.005-threads").
		1210
		1211	It does that by using threads instead of OS processes. The difference
		1212	between processes and threads is that threads share memory (and other
		1213	state, such as files) between threads within a single process, while
		1214	processes do not share anything (at least not semantically). That
		1215	means that modifications done by one thread are seen by others, while
		1216	modifications by one process are not seen by other processes.
		1217
		1218	The "ithreads" work exactly like that: when creating a new ithreads
		1219	process, all state is copied (memory is copied physically, files and code
		1220	is copied logically). Afterwards, it isolates all modifications. On UNIX,
		1221	the same behaviour can be achieved by using operating system processes,
		1222	except that UNIX typically uses hardware built into the system to do this
		1223	efficiently, while the windows process emulation emulates this hardware in
		1224	software (rather efficiently, but of course it is still much slower than
		1225	dedicated hardware).
		1226
		1227	As mentioned before, loading code, modifying code, modifying data
		1228	structures and so on is only visible in the ithreads process doing the
		1229	modification, not in other ithread processes within the same OS process.
		1230
		1231	This is why "ithreads" do not implement threads for perl at all, only
		1232	processes. What makes it so bad is that on non-windows platforms, you can
		1233	actually take advantage of custom hardware for this purpose (as evidenced
		1234	by the forks module, which gives you the (i-) threads API, just much
		1235	faster).
		1236
		1237	Sharing data is in the i-threads model is done by transfering data
		1238	structures between threads using copying semantics, which is very slow -
		1239	shared data simply does not exist. Benchmarks using i-threads which are
		1240	communication-intensive show extremely bad behaviour with i-threads (in
		1241	fact, so bad that Coro, which cannot take direct advantage of multiple
		1242	CPUs, is often orders of magnitude faster because it shares data using
		1243	real threads, refer to my talk for details).
		1244
		1245	As summary, i-threads *use* threads to implement processes, while
		1246	the compatible forks module *uses* processes to emulate, uhm,
		1247	processes. I-threads slow down every perl program when enabled, and
		1248	outside of windows, serve no (or little) practical purpose, but
		1249	disadvantages every single-threaded Perl program.
		1250
		1251	This is the reason that I try to avoid the name "ithreads", as it is
		1252	misleading as it implies that it implements some kind of thread model for
		1253	perl, and prefer the name "windows process emulation", which describes the
		1254	actual use and behaviour of it much better.
572		1255
573	=head1 SEE ALSO	1256	=head1 SEE ALSO
574		1257
		1258	Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
		1259
		1260	Debugging: L<Coro::Debug>.
		1261
575	Support/Utility: L<Coro::Cont>, L<Coro::Specific>, L<Coro::State>, L<Coro::Util>.	1262	Support/Utility: L<Coro::Specific>, L<Coro::Util>.
576		1263
577	Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>.	1264	Locking and IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>,
		1265	L<Coro::SemaphoreSet>, L<Coro::RWLock>.
578		1266
579	Event/IO: L<Coro::Timer>, L<Coro::Event>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::Select>.	1267	I/O and Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
580		1268
581	Embedding: L<Coro:MakeMaker>	1269	Compatibility with other modules: L<Coro::LWP> (but see also L<AnyEvent::HTTP> for
		1270	a better-working alternative), L<Coro::BDB>, L<Coro::Storable>,
		1271	L<Coro::Select>.
		1272
		1273	XS API: L<Coro::MakeMaker>.
		1274
		1275	Low level Configuration, Thread Environment, Continuations: L<Coro::State>.
582		1276
583	=head1 AUTHOR	1277	=head1 AUTHOR
584		1278
585	Marc Lehmann <schmorp@schmorp.de>	1279	Marc Lehmann <schmorp@schmorp.de>
586	http://home.schmorp.de/	1280	http://home.schmorp.de/

Diff Legend

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