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

Comparing Coro/Coro.pm (file contents):
Revision 1.16 by root, Tue Jul 17 15:42:28 2001 UTC vs.
Revision 1.234 by root, Fri Nov 21 06:52:10 2008 UTC

1	=head1 NAME	1	=head1 NAME
2		2
3	Coro - coroutine process abstraction	3	Coro - the real perl threads
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 coroutine
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
23	This module collection manages coroutines. Coroutines are similar to	31	This module collection manages coroutines, that is, cooperative
24	Threads but don't run in parallel.	32	threads. Coroutines are similar to kernel threads but don't (in general)
		33	run in parallel at the same time even on SMP machines. The specific flavor
		34	of coroutine used in this module also guarantees you that it will not
		35	switch between coroutines unless necessary, at easily-identified points
		36	in your program, so locking and parallel access are rarely an issue,
		37	making coroutine programming much safer and easier than using other thread
		38	models.
25		39
26	This module is still experimental, see the BUGS section below.	40	Unlike the so-called "Perl threads" (which are not actually real threads
		41	but only the windows process emulation ported to unix), Coro provides a
		42	full shared address space, which makes communication between coroutines
		43	very easy. And coroutines are fast, too: disabling the Windows process
		44	emulation code in your perl and using Coro can easily result in a two to
		45	four times speed increase for your programs.
		46
		47	Coro achieves that by supporting multiple running interpreters that share
		48	data, which is especially useful to code pseudo-parallel processes and
		49	for event-based programming, such as multiple HTTP-GET requests running
		50	concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro
		51	into an event-based environment.
		52
		53	In this module, a coroutines is defined as "callchain + lexical variables
		54	+ @_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own
		55	callchain, its own set of lexicals and its own set of perls most important
		56	global variables (see L<Coro::State> for more configuration and background
		57	info).
		58
		59	See also the C<SEE ALSO> section at the end of this document - the Coro
		60	module family is quite large.
27		61
28	=cut	62	=cut
29		63
30	package Coro;	64	package Coro;
31		65
		66	use strict qw(vars subs);
		67	no warnings "uninitialized";
		68
32	use Coro::State;	69	use Coro::State;
33		70
34	use base Exporter;	71	use base qw(Coro::State Exporter);
35		72
		73	our $idle; # idle handler
		74	our $main; # main coroutine
		75	our $current; # current coroutine
		76
36	$VERSION = 0.07;	77	our $VERSION = "5.0";
37		78
38	@EXPORT = qw(async yield schedule terminate);	79	our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub);
39	@EXPORT_OK = qw($current);	80	our %EXPORT_TAGS = (
		81	prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
		82	);
		83	our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
40		84
41	{	85	=head1 GLOBAL VARIABLES
42	use subs 'async';
43		86
44	my @async;	87	=over 4
45		88
46	# this way of handling attributes simply is NOT scalable ;()
47	sub import {
48	Coro->export_to_level(1, @_);
49	my $old = *{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"}{CODE};
50	*{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"} = sub {
51	my ($package, $ref) = (shift, shift);
52	my @attrs;
53	for (@_) {
54	if ($_ eq "Coro") {
55	push @async, $ref;
56	} else {
57	push @attrs, @_;
58	}
59	}
60	return $old ? $old->($package, $name, @attrs) : @attrs;
61	};
62	}
63
64	sub INIT {
65	async pop @async while @async;
66	}
67	}
68
69	=item $main	89	=item $Coro::main
70		90
71	This coroutine represents the main program.	91	This variable stores the coroutine object that represents the main
		92	program. While you cna C<ready> it and do most other things you can do to
		93	coroutines, it is mainly useful to compare again C<$Coro::current>, to see
		94	whether you are running in the main program or not.
72		95
73	=cut	96	=cut
74		97
75	our $main = new Coro;	98	# $main is now being initialised by Coro::State
76		99
77	=item $current	100	=item $Coro::current
78		101
79	The current coroutine (the last coroutine switched to). The initial value is C<$main> (of course).	102	The coroutine object representing the current coroutine (the last
		103	coroutine that the Coro scheduler switched to). The initial value is
		104	C<$Coro::main> (of course).
80		105
81	=cut	106	This variable is B<strictly> I<read-only>. You can take copies of the
		107	value stored in it and use it as any other coroutine object, but you must
		108	not otherwise modify the variable itself.
82		109
83	# maybe some other module used Coro::Specific before...	110	=cut
84	if ($current) {
85	$main->{specific} = $current->{specific};
86	}
87		111
88	our $current = $main;	112	sub current() { $current } # [DEPRECATED]
89		113
90	=item $idle	114	=item $Coro::idle
91		115
92	The coroutine to switch to when no other coroutine is running. The default	116	This variable is mainly useful to integrate Coro into event loops. It is
93	implementation prints "FATAL: deadlock detected" and exits.	117	usually better to rely on L<Coro::AnyEvent> or LC<Coro::EV>, as this is
		118	pretty low-level functionality.
94		119
95	=cut	120	This variable stores a callback that is called whenever the scheduler
		121	finds no ready coroutines to run. The default implementation prints
		122	"FATAL: deadlock detected" and exits, because the program has no other way
		123	to continue.
96		124
97	# should be done using priorities :(	125	This hook is overwritten by modules such as C<Coro::Timer> and
98	our $idle = new Coro sub {	126	C<Coro::AnyEvent> to wait on an external event that hopefully wake up a
99	print STDERR "FATAL: deadlock detected\n";	127	coroutine so the scheduler can run it.
100	exit(51);	128
		129	Note that the callback I<must not>, under any circumstances, block
		130	the current coroutine. Normally, this is achieved by having an "idle
		131	coroutine" that calls the event loop and then blocks again, and then
		132	readying that coroutine in the idle handler.
		133
		134	See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this
		135	technique.
		136
		137	Please note that if your callback recursively invokes perl (e.g. for event
		138	handlers), then it must be prepared to be called recursively itself.
		139
		140	=cut
		141
		142	$idle = sub {
		143	require Carp;
		144	Carp::croak ("FATAL: deadlock detected");
101	};	145	};
102		146
103	# we really need priorities...	147	# this coroutine is necessary because a coroutine
104	my @ready; # the ready queue. hehe, rather broken ;)	148	# cannot destroy itself.
		149	our @destroy;
		150	our $manager;
105		151
106	# static methods. not really.	152	$manager = new Coro sub {
		153	while () {
		154	Coro::_cancel shift @destroy
		155	while @destroy;
107		156
108	=head2 STATIC METHODS	157	&schedule;
		158	}
		159	};
		160	$manager->{desc} = "[coro manager]";
		161	$manager->prio (PRIO_MAX);
109		162
110	Static methods are actually functions that operate on the current process only.	163	=back
		164
		165	=head1 SIMPLE COROUTINE CREATION
111		166
112	=over 4	167	=over 4
113		168
114	=item async { ... } [@args...]	169	=item async { ... } [@args...]
115		170
116	Create a new asynchronous process and return it's process object	171	Create a new coroutine and return it's coroutine object (usually
117	(usually unused). When the sub returns the new process is automatically	172	unused). The coroutine will be put into the ready queue, so
		173	it will start running automatically on the next scheduler run.
		174
		175	The first argument is a codeblock/closure that should be executed in the
		176	coroutine. When it returns argument returns the coroutine is automatically
118	terminated.	177	terminated.
119		178
		179	The remaining arguments are passed as arguments to the closure.
		180
		181	See the C<Coro::State::new> constructor for info about the coroutine
		182	environment in which coroutines are executed.
		183
		184	Calling C<exit> in a coroutine will do the same as calling exit outside
		185	the coroutine. Likewise, when the coroutine dies, the program will exit,
		186	just as it would in the main program.
		187
		188	If you do not want that, you can provide a default C<die> handler, or
		189	simply avoid dieing (by use of C<eval>).
		190
120	# create a new coroutine that just prints its arguments	191	Example: Create a new coroutine that just prints its arguments.
		192
121	async {	193	async {
122	print "@_\n";	194	print "@_\n";
123	} 1,2,3,4;	195	} 1,2,3,4;
124		196
125	The coderef you submit MUST NOT be a closure that refers to variables
126	in an outer scope. This does NOT work. Pass arguments into it instead.
127
128	=cut	197	=cut
129		198
130	sub async(&@) {	199	sub async(&@) {
131	my $pid = new Coro @_;	200	my $coro = new Coro @_;
132	$pid->ready;	201	$coro->ready;
133	$pid;	202	$coro
134	}	203	}
		204
		205	=item async_pool { ... } [@args...]
		206
		207	Similar to C<async>, but uses a coroutine pool, so you should not call
		208	terminate or join on it (although you are allowed to), and you get a
		209	coroutine that might have executed other code already (which can be good
		210	or bad :).
		211
		212	On the plus side, this function is about twice as fast as creating (and
		213	destroying) a completely new coroutine, so if you need a lot of generic
		214	coroutines in quick successsion, use C<async_pool>, not C<async>.
		215
		216	The code block is executed in an C<eval> context and a warning will be
		217	issued in case of an exception instead of terminating the program, as
		218	C<async> does. As the coroutine is being reused, stuff like C<on_destroy>
		219	will not work in the expected way, unless you call terminate or cancel,
		220	which somehow defeats the purpose of pooling (but is fine in the
		221	exceptional case).
		222
		223	The priority will be reset to C<0> after each run, tracing will be
		224	disabled, the description will be reset and the default output filehandle
		225	gets restored, so you can change all these. Otherwise the coroutine will
		226	be re-used "as-is": most notably if you change other per-coroutine global
		227	stuff such as C<$/> you I<must needs> revert that change, which is most
		228	simply done by using local as in: C<< local $/ >>.
		229
		230	The idle pool size is limited to C<8> idle coroutines (this can be
		231	adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle
		232	coros as required.
		233
		234	If you are concerned about pooled coroutines growing a lot because a
		235	single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
		236	{ terminate }> once per second or so to slowly replenish the pool. In
		237	addition to that, when the stacks used by a handler grows larger than 32kb
		238	(adjustable via $Coro::POOL_RSS) it will also be destroyed.
		239
		240	=cut
		241
		242	our $POOL_SIZE = 8;
		243	our $POOL_RSS = 32 * 1024;
		244	our @async_pool;
		245
		246	sub pool_handler {
		247	while () {
		248	eval {
		249	&{&_pool_handler} while 1;
		250	};
		251
		252	warn $@ if $@;
		253	}
		254	}
		255
		256	=back
		257
		258	=head1 STATIC METHODS
		259
		260	Static methods are actually functions that implicitly operate on the
		261	current coroutine.
		262
		263	=over 4
135		264
136	=item schedule	265	=item schedule
137		266
138	Calls the scheduler. Please note that the current process will not be put	267	Calls the scheduler. The scheduler will find the next coroutine that is
		268	to be run from the ready queue and switches to it. The next coroutine
		269	to be run is simply the one with the highest priority that is longest
		270	in its ready queue. If there is no coroutine ready, it will clal the
		271	C<$Coro::idle> hook.
		272
		273	Please note that the current coroutine will I<not> be put into the ready
139	into the ready queue, so calling this function usually means you will	274	queue, so calling this function usually means you will never be called
140	never be called again.	275	again unless something else (e.g. an event handler) calls C<< ->ready >>,
		276	thus waking you up.
141		277
142	=cut	278	This makes C<schedule> I<the> generic method to use to block the current
		279	coroutine and wait for events: first you remember the current coroutine in
		280	a variable, then arrange for some callback of yours to call C<< ->ready
		281	>> on that once some event happens, and last you call C<schedule> to put
		282	yourself to sleep. Note that a lot of things can wake your coroutine up,
		283	so you need to check whether the event indeed happened, e.g. by storing the
		284	status in a variable.
143		285
144	my $prev;	286	See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks.
145		287
146	sub schedule {
147	# should be done using priorities :(
148	($prev, $current) = ($current, shift @ready \|\| $idle);
149	Coro::State::transfer($prev, $current);
150	}
151
152	=item yield	288	=item cede
153		289
154	Yield to other processes. This function puts the current process into the	290	"Cede" to other coroutines. This function puts the current coroutine into
155	ready queue and calls C<schedule>.	291	the ready queue and calls C<schedule>, which has the effect of giving
		292	up the current "timeslice" to other coroutines of the same or higher
		293	priority. Once your coroutine gets its turn again it will automatically be
		294	resumed.
156		295
157	=cut	296	This function is often called C<yield> in other languages.
158		297
159	sub yield {	298	=item Coro::cede_notself
160	$current->ready;
161	&schedule;
162	}
163		299
		300	Works like cede, but is not exported by default and will cede to I<any>
		301	coroutine, regardless of priority. This is useful sometimes to ensure
		302	progress is made.
		303
164	=item terminate	304	=item terminate [arg...]
165		305
166	Terminates the current process.	306	Terminates the current coroutine with the given status values (see L<cancel>).
167		307
168	Future versions of this function will allow result arguments.	308	=item killall
169		309
170	=cut	310	Kills/terminates/cancels all coroutines except the currently running
		311	one. This is useful after a fork, either in the child or the parent, as
		312	usually only one of them should inherit the running coroutines.
171		313
172	sub terminate {	314	Note that while this will try to free some of the main programs resources,
173	$current->{_results} = [@_];	315	you cannot free all of them, so if a coroutine that is not the main
174	&schedule;	316	program calls this function, there will be some one-time resource leak.
		317
		318	=cut
		319
		320	sub killall {
		321	for (Coro::State::list) {
		322	$_->cancel
		323	if $_ != $current && UNIVERSAL::isa $_, "Coro";
		324	}
175	}	325	}
176		326
177	=back	327	=back
178		328
179	# dynamic methods	329	=head1 COROUTINE OBJECT METHODS
180		330
181	=head2 PROCESS METHODS
182
183	These are the methods you can call on process objects.	331	These are the methods you can call on coroutine objects (or to create
		332	them).
184		333
185	=over 4	334	=over 4
186		335
187	=item new Coro \&sub [, @args...]	336	=item new Coro \&sub [, @args...]
188		337
189	Create a new process and return it. When the sub returns the process	338	Create a new coroutine and return it. When the sub returns, the coroutine
190	automatically terminates. To start the process you must first put it into	339	automatically terminates as if C<terminate> with the returned values were
		340	called. To make the coroutine run you must first put it into the ready
191	the ready queue by calling the ready method.	341	queue by calling the ready method.
192		342
193	The coderef you submit MUST NOT be a closure that refers to variables	343	See C<async> and C<Coro::State::new> for additional info about the
194	in an outer scope. This does NOT work. Pass arguments into it instead.	344	coroutine environment.
195		345
196	=cut	346	=cut
197		347
198	sub _newcoro {	348	sub _terminate {
199	terminate &{+shift};	349	terminate &{+shift};
200	}	350	}
201		351
		352	=item $success = $coroutine->ready
		353
		354	Put the given coroutine into the end of its ready queue (there is one
		355	queue for each priority) and return true. If the coroutine is already in
		356	the ready queue, do nothing and return false.
		357
		358	This ensures that the scheduler will resume this coroutine automatically
		359	once all the coroutines of higher priority and all coroutines of the same
		360	priority that were put into the ready queue earlier have been resumed.
		361
		362	=item $is_ready = $coroutine->is_ready
		363
		364	Return whether the coroutine is currently the ready queue or not,
		365
		366	=item $coroutine->cancel (arg...)
		367
		368	Terminates the given coroutine and makes it return the given arguments as
		369	status (default: the empty list). Never returns if the coroutine is the
		370	current coroutine.
		371
		372	=cut
		373
202	sub new {	374	sub cancel {
203	my $class = shift;	375	my $self = shift;
204	bless {
205	_coro_state => (new Coro::State $_[0] && \&_newcoro, @_),
206	}, $class;
207	}
208		376
209	=item $process->ready	377	if ($current == $self) {
		378	terminate @_;
		379	} else {
		380	$self->{_status} = [@_];
		381	$self->_cancel;
		382	}
		383	}
210		384
211	Put the current process into the ready queue.	385	=item $coroutine->schedule_to
212		386
213	=cut	387	Puts the current coroutine to sleep (like C<Coro::schedule>), but instead
		388	of continuing with the next coro from the ready queue, always switch to
		389	the given coroutine object (regardless of priority etc.). The readyness
		390	state of that coroutine isn't changed.
214		391
215	sub ready {	392	This is an advanced method for special cases - I'd love to hear about any
216	push @ready, $_[0];	393	uses for this one.
		394
		395	=item $coroutine->cede_to
		396
		397	Like C<schedule_to>, but puts the current coroutine into the ready
		398	queue. This has the effect of temporarily switching to the given
		399	coroutine, and continuing some time later.
		400
		401	This is an advanced method for special cases - I'd love to hear about any
		402	uses for this one.
		403
		404	=item $coroutine->throw ([$scalar])
		405
		406	If C<$throw> is specified and defined, it will be thrown as an exception
		407	inside the coroutine at the next convenient point in time. Otherwise
		408	clears the exception object.
		409
		410	Coro will check for the exception each time a schedule-like-function
		411	returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down
		412	>>, C<< Coro::Handle->readable >> and so on. Most of these functions
		413	detect this case and return early in case an exception is pending.
		414
		415	The exception object will be thrown "as is" with the specified scalar in
		416	C<$@>, i.e. if it is a string, no line number or newline will be appended
		417	(unlike with C<die>).
		418
		419	This can be used as a softer means than C<cancel> to ask a coroutine to
		420	end itself, although there is no guarantee that the exception will lead to
		421	termination, and if the exception isn't caught it might well end the whole
		422	program.
		423
		424	You might also think of C<throw> as being the moral equivalent of
		425	C<kill>ing a coroutine with a signal (in this case, a scalar).
		426
		427	=item $coroutine->join
		428
		429	Wait until the coroutine terminates and return any values given to the
		430	C<terminate> or C<cancel> functions. C<join> can be called concurrently
		431	from multiple coroutines, and all will be resumed and given the status
		432	return once the C<$coroutine> terminates.
		433
		434	=cut
		435
		436	sub join {
		437	my $self = shift;
		438
		439	unless ($self->{_status}) {
		440	my $current = $current;
		441
		442	push @{$self->{_on_destroy}}, sub {
		443	$current->ready;
		444	undef $current;
		445	};
		446
		447	&schedule while $current;
		448	}
		449
		450	wantarray ? @{$self->{_status}} : $self->{_status}[0];
		451	}
		452
		453	=item $coroutine->on_destroy (\&cb)
		454
		455	Registers a callback that is called when this coroutine gets destroyed,
		456	but before it is joined. The callback gets passed the terminate arguments,
		457	if any, and I<must not> die, under any circumstances.
		458
		459	=cut
		460
		461	sub on_destroy {
		462	my ($self, $cb) = @_;
		463
		464	push @{ $self->{_on_destroy} }, $cb;
		465	}
		466
		467	=item $oldprio = $coroutine->prio ($newprio)
		468
		469	Sets (or gets, if the argument is missing) the priority of the
		470	coroutine. Higher priority coroutines get run before lower priority
		471	coroutines. Priorities are small signed integers (currently -4 .. +3),
		472	that you can refer to using PRIO_xxx constants (use the import tag :prio
		473	to get then):
		474
		475	PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN
		476	3 > 1 > 0 > -1 > -3 > -4
		477
		478	# set priority to HIGH
		479	current->prio(PRIO_HIGH);
		480
		481	The idle coroutine ($Coro::idle) always has a lower priority than any
		482	existing coroutine.
		483
		484	Changing the priority of the current coroutine will take effect immediately,
		485	but changing the priority of coroutines in the ready queue (but not
		486	running) will only take effect after the next schedule (of that
		487	coroutine). This is a bug that will be fixed in some future version.
		488
		489	=item $newprio = $coroutine->nice ($change)
		490
		491	Similar to C<prio>, but subtract the given value from the priority (i.e.
		492	higher values mean lower priority, just as in unix).
		493
		494	=item $olddesc = $coroutine->desc ($newdesc)
		495
		496	Sets (or gets in case the argument is missing) the description for this
		497	coroutine. This is just a free-form string you can associate with a
		498	coroutine.
		499
		500	This method simply sets the C<< $coroutine->{desc} >> member to the given
		501	string. You can modify this member directly if you wish.
		502
		503	=cut
		504
		505	sub desc {
		506	my $old = $_[0]{desc};
		507	$_[0]{desc} = $_[1] if @_ > 1;
		508	$old;
		509	}
		510
		511	sub transfer {
		512	require Carp;
		513	Carp::croak ("You must not call ->transfer on Coro objects. Use Coro::State objects or the ->schedule_to method. Caught");
217	}	514	}
218		515
219	=back	516	=back
220		517
		518	=head1 GLOBAL FUNCTIONS
		519
		520	=over 4
		521
		522	=item Coro::nready
		523
		524	Returns the number of coroutines that are currently in the ready state,
		525	i.e. that can be switched to by calling C<schedule> directory or
		526	indirectly. The value C<0> means that the only runnable coroutine is the
		527	currently running one, so C<cede> would have no effect, and C<schedule>
		528	would cause a deadlock unless there is an idle handler that wakes up some
		529	coroutines.
		530
		531	=item my $guard = Coro::guard { ... }
		532
		533	This creates and returns a guard object. Nothing happens until the object
		534	gets destroyed, in which case the codeblock given as argument will be
		535	executed. This is useful to free locks or other resources in case of a
		536	runtime error or when the coroutine gets canceled, as in both cases the
		537	guard block will be executed. The guard object supports only one method,
		538	C<< ->cancel >>, which will keep the codeblock from being executed.
		539
		540	Example: set some flag and clear it again when the coroutine gets canceled
		541	or the function returns:
		542
		543	sub do_something {
		544	my $guard = Coro::guard { $busy = 0 };
		545	$busy = 1;
		546
		547	# do something that requires $busy to be true
		548	}
		549
		550	=cut
		551
		552	sub guard(&) {
		553	bless \(my $cb = $_[0]), "Coro::guard"
		554	}
		555
		556	sub Coro::guard::cancel {
		557	${$_[0]} = sub { };
		558	}
		559
		560	sub Coro::guard::DESTROY {
		561	${$_[0]}->();
		562	}
		563
		564
		565	=item unblock_sub { ... }
		566
		567	This utility function takes a BLOCK or code reference and "unblocks" it,
		568	returning a new coderef. Unblocking means that calling the new coderef
		569	will return immediately without blocking, returning nothing, while the
		570	original code ref will be called (with parameters) from within another
		571	coroutine.
		572
		573	The reason this function exists is that many event libraries (such as the
		574	venerable L<Event\|Event> module) are not coroutine-safe (a weaker form
		575	of thread-safety). This means you must not block within event callbacks,
		576	otherwise you might suffer from crashes or worse. The only event library
		577	currently known that is safe to use without C<unblock_sub> is L<EV>.
		578
		579	This function allows your callbacks to block by executing them in another
		580	coroutine where it is safe to block. One example where blocking is handy
		581	is when you use the L<Coro::AIO\|Coro::AIO> functions to save results to
		582	disk, for example.
		583
		584	In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
		585	creating event callbacks that want to block.
		586
		587	If your handler does not plan to block (e.g. simply sends a message to
		588	another coroutine, or puts some other coroutine into the ready queue),
		589	there is no reason to use C<unblock_sub>.
		590
		591	Note that you also need to use C<unblock_sub> for any other callbacks that
		592	are indirectly executed by any C-based event loop. For example, when you
		593	use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
		594	provides callbacks that are the result of some event callback, then you
		595	must not block either, or use C<unblock_sub>.
		596
		597	=cut
		598
		599	our @unblock_queue;
		600
		601	# we create a special coro because we want to cede,
		602	# to reduce pressure on the coro pool (because most callbacks
		603	# return immediately and can be reused) and because we cannot cede
		604	# inside an event callback.
		605	our $unblock_scheduler = new Coro sub {
		606	while () {
		607	while (my $cb = pop @unblock_queue) {
		608	&async_pool (@$cb);
		609
		610	# for short-lived callbacks, this reduces pressure on the coro pool
		611	# as the chance is very high that the async_poll coro will be back
		612	# in the idle state when cede returns
		613	cede;
		614	}
		615	schedule; # sleep well
		616	}
		617	};
		618	$unblock_scheduler->{desc} = "[unblock_sub scheduler]";
		619
		620	sub unblock_sub(&) {
		621	my $cb = shift;
		622
		623	sub {
		624	unshift @unblock_queue, [$cb, @_];
		625	$unblock_scheduler->ready;
		626	}
		627	}
		628
		629	=item $cb = Coro::rouse_cb
		630
		631	Create and return a "rouse callback". That's a code reference that, when
		632	called, will save its arguments and notify the owner coroutine of the
		633	callback.
		634
		635	See the next function.
		636
		637	=item @args = Coro::rouse_wait [$cb]
		638
		639	Wait for the specified rouse callback (or the last one tht was created in
		640	this coroutine).
		641
		642	As soon as the callback is invoked (or when the calback was invoked before
		643	C<rouse_wait>), it will return a copy of the arguments originally passed
		644	to the rouse callback.
		645
		646	See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example.
		647
		648	=back
		649
221	=cut	650	=cut
222		651
223	1;	652	1;
224		653
225	=head1 BUGS	654	=head1 HOW TO WAIT FOR A CALLBACK
226		655
227	- could be faster, especially when the core would introduce special	656	It is very common for a coroutine to wait for some callback to be
228	support for coroutines (like it does for threads).	657	called. This occurs naturally when you use coroutines in an otherwise
229	- there is still a memleak on coroutine termination that I could not	658	event-based program, or when you use event-based libraries.
230	identify. Could be as small as a single SV.	659
231	- this module is not well-tested.	660	These typically register a callback for some event, and call that callback
		661	when the event occured. In a coroutine, however, you typically want to
		662	just wait for the event, simplyifying things.
		663
		664	For example C<< AnyEvent->child >> registers a callback to be called when
		665	a specific child has exited:
		666
		667	my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });
		668
		669	But from withina coroutine, you often just want to write this:
		670
		671	my $status = wait_for_child $pid;
		672
		673	Coro offers two functions specifically designed to make this easy,
		674	C<Coro::rouse_cb> and C<Coro::rouse_wait>.
		675
		676	The first function, C<rouse_cb>, generates and returns a callback that,
		677	when invoked, will save it's arguments and notify the coroutine that
		678	created the callback.
		679
		680	The second function, C<rouse_wait>, waits for the callback to be called
		681	(by calling C<schedule> to go to sleep) and returns the arguments
		682	originally passed to the callback.
		683
		684	Using these functions, it becomes easy to write the C<wait_for_child>
		685	function mentioned above:
		686
		687	sub wait_for_child($) {
		688	my ($pid) = @_;
		689
		690	my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb);
		691
		692	my ($rpid, $rstatus) = Coro::rouse_wait;
		693	$rstatus
		694	}
		695
		696	In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough,
		697	you can roll your own, using C<schedule>:
		698
		699	sub wait_for_child($) {
		700	my ($pid) = @_;
		701
		702	# store the current coroutine in $current,
		703	# and provide result variables for the closure passed to ->child
		704	my $current = $Coro::current;
		705	my ($done, $rstatus);
		706
		707	# pass a closure to ->child
		708	my $watcher = AnyEvent->child (pid => $pid, cb => sub {
		709	$rstatus = $_[1]; # remember rstatus
		710	$done = 1; # mark $rstatus as valud
		711	});
		712
		713	# wait until the closure has been called
		714	schedule while !$done;
		715
		716	$rstatus
		717	}
		718
		719
		720	=head1 BUGS/LIMITATIONS
		721
		722	=over 4
		723
		724	=item fork with pthread backend
		725
		726	When Coro is compiled using the pthread backend (which isn't recommended
		727	but required on many BSDs as their libcs are completely broken), then
		728	coroutines will not survive a fork. There is no known workaround except to
		729	fix your libc and use a saner backend.
		730
		731	=item perl process emulation ("threads")
		732
		733	This module is not perl-pseudo-thread-safe. You should only ever use this
		734	module from the same thread (this requirement might be removed in the
		735	future to allow per-thread schedulers, but Coro::State does not yet allow
		736	this). I recommend disabling thread support and using processes, as having
		737	the windows process emulation enabled under unix roughly halves perl
		738	performance, even when not used.
		739
		740	=item coroutine switching not signal safe
		741
		742	You must not switch to another coroutine from within a signal handler
		743	(only relevant with %SIG - most event libraries provide safe signals).
		744
		745	That means you I<MUST NOT> call any function that might "block" the
		746	current coroutine - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or
		747	anything that calls those. Everything else, including calling C<ready>,
		748	works.
		749
		750	=back
		751
232		752
233	=head1 SEE ALSO	753	=head1 SEE ALSO
234		754
235	L<Coro::Channel>, L<Coro::Cont>, L<Coro::Specific>, L<Coro::Semaphore>,	755	Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
236	L<Coro::Signal>, L<Coro::State>, L<Coro::Event>.	756
		757	Debugging: L<Coro::Debug>.
		758
		759	Support/Utility: L<Coro::Specific>, L<Coro::Util>.
		760
		761	Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>.
		762
		763	IO/Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
		764
		765	Compatibility: L<Coro::LWP>, L<Coro::BDB>, L<Coro::Storable>, L<Coro::Select>.
		766
		767	XS API: L<Coro::MakeMaker>.
		768
		769	Low level Configuration, Coroutine Environment: L<Coro::State>.
237		770
238	=head1 AUTHOR	771	=head1 AUTHOR
239		772
240	Marc Lehmann <pcg@goof.com>	773	Marc Lehmann <schmorp@schmorp.de>
241	http://www.goof.com/pcg/marc/	774	http://home.schmorp.de/
242		775
243	=cut	776	=cut
244		777

Diff Legend

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

Comparing Coro/Coro.pm (file contents): Revision 1.16 by root, Tue Jul 17 15:42:28 2001 UTC vs. Revision 1.234 by root, Fri Nov 21 06:52:10 2008 UTC

Diff Legend

Comparing Coro/Coro.pm (file contents):
Revision 1.16 by root, Tue Jul 17 15:42:28 2001 UTC vs.
Revision 1.234 by root, Fri Nov 21 06:52:10 2008 UTC