ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Coro.pm
(Generate patch)

Comparing Coro/Coro.pm (file contents):
Revision 1.133 by root, Fri Sep 21 01:23:58 2007 UTC vs.
Revision 1.234 by root, Fri Nov 21 06:52:10 2008 UTC

1=head1 NAME 1=head1 NAME
2 2
3Coro - coroutine process abstraction 3Coro - 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 coroutine 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 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
23This module collection manages coroutines. Coroutines are similar 31This module collection manages coroutines, that is, cooperative
24to threads but don't run in parallel at the same time even on SMP 32threads. Coroutines are similar to kernel threads but don't (in general)
25machines. The specific flavor of coroutine used in this module also 33run in parallel at the same time even on SMP machines. The specific flavor
26guarantees you that it will not switch between coroutines unless 34of coroutine used in this module also guarantees you that it will not
27necessary, at easily-identified points in your program, so locking and 35switch between coroutines unless necessary, at easily-identified points
28parallel access are rarely an issue, making coroutine programming much 36in your program, so locking and parallel access are rarely an issue,
29safer than threads programming. 37making coroutine programming much safer and easier than using other thread
38models.
30 39
31(Perl, however, does not natively support real threads but instead does a 40Unlike the so-called "Perl threads" (which are not actually real threads
32very slow and memory-intensive emulation of processes using threads. This 41but only the windows process emulation ported to unix), Coro provides a
33is a performance win on Windows machines, and a loss everywhere else). 42full shared address space, which makes communication between coroutines
43very easy. And coroutines are fast, too: disabling the Windows process
44emulation code in your perl and using Coro can easily result in a two to
45four times speed increase for your programs.
34 46
47Coro achieves that by supporting multiple running interpreters that share
48data, which is especially useful to code pseudo-parallel processes and
49for event-based programming, such as multiple HTTP-GET requests running
50concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro
51into an event-based environment.
52
35In this module, coroutines are defined as "callchain + lexical variables + 53In this module, a coroutines is defined as "callchain + lexical variables
36@_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own callchain, 54+ @_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own
37its own set of lexicals and its own set of perls most important global 55callchain, its own set of lexicals and its own set of perls most important
38variables. 56global variables (see L<Coro::State> for more configuration and background
57info).
58
59See also the C<SEE ALSO> section at the end of this document - the Coro
60module family is quite large.
39 61
40=cut 62=cut
41 63
42package Coro; 64package Coro;
43 65
44use strict; 66use strict qw(vars subs);
45no warnings "uninitialized"; 67no warnings "uninitialized";
46 68
47use Coro::State; 69use Coro::State;
48 70
49use base qw(Coro::State Exporter); 71use base qw(Coro::State Exporter);
50 72
51our $idle; # idle handler 73our $idle; # idle handler
52our $main; # main coroutine 74our $main; # main coroutine
53our $current; # current coroutine 75our $current; # current coroutine
54 76
55our $VERSION = '3.7'; 77our $VERSION = "5.0";
56 78
57our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); 79our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub);
58our %EXPORT_TAGS = ( 80our %EXPORT_TAGS = (
59 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], 81 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
60); 82);
61our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready)); 83our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
62 84
63{ 85=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 86
97=over 4 87=over 4
98 88
99=item $main 89=item $Coro::main
100 90
101This coroutine represents the main program. 91This variable stores the coroutine object that represents the main
92program. While you cna C<ready> it and do most other things you can do to
93coroutines, it is mainly useful to compare again C<$Coro::current>, to see
94whether you are running in the main program or not.
102 95
103=cut 96=cut
104 97
105$main = new Coro; 98# $main is now being initialised by Coro::State
106 99
107=item $current (or as function: current) 100=item $Coro::current
108 101
109The current coroutine (the last coroutine switched to). The initial value 102The coroutine object representing the current coroutine (the last
103coroutine that the Coro scheduler switched to). The initial value is
110is C<$main> (of course). 104C<$Coro::main> (of course).
111 105
112This variable is B<strictly> I<read-only>. It is provided for performance 106This variable is B<strictly> I<read-only>. You can take copies of the
113reasons. If performance is not essential you are encouraged to use the 107value stored in it and use it as any other coroutine object, but you must
114C<Coro::current> function instead. 108not otherwise modify the variable itself.
115 109
116=cut 110=cut
117 111
118$main->{desc} = "[main::]";
119
120# maybe some other module used Coro::Specific before...
121$main->{specific} = $current->{specific}
122 if $current;
123
124_set_current $main;
125
126sub current() { $current } 112sub current() { $current } # [DEPRECATED]
127 113
128=item $idle 114=item $Coro::idle
129 115
130A callback that is called whenever the scheduler finds no ready coroutines 116This variable is mainly useful to integrate Coro into event loops. It is
131to run. The default implementation prints "FATAL: deadlock detected" and 117usually better to rely on L<Coro::AnyEvent> or LC<Coro::EV>, as this is
132exits, because the program has no other way to continue. 118pretty low-level functionality.
119
120This variable stores a callback that is called whenever the scheduler
121finds no ready coroutines to run. The default implementation prints
122"FATAL: deadlock detected" and exits, because the program has no other way
123to continue.
133 124
134This hook is overwritten by modules such as C<Coro::Timer> and 125This hook is overwritten by modules such as C<Coro::Timer> and
135C<Coro::Event> to wait on an external event that hopefully wake up a 126C<Coro::AnyEvent> to wait on an external event that hopefully wake up a
136coroutine so the scheduler can run it. 127coroutine so the scheduler can run it.
137 128
129Note that the callback I<must not>, under any circumstances, block
130the current coroutine. Normally, this is achieved by having an "idle
131coroutine" that calls the event loop and then blocks again, and then
132readying that coroutine in the idle handler.
133
134See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this
135technique.
136
138Please note that if your callback recursively invokes perl (e.g. for event 137Please note that if your callback recursively invokes perl (e.g. for event
139handlers), then it must be prepared to be called recursively. 138handlers), then it must be prepared to be called recursively itself.
140 139
141=cut 140=cut
142 141
143$idle = sub { 142$idle = sub {
144 require Carp; 143 require Carp;
145 Carp::croak ("FATAL: deadlock detected"); 144 Carp::croak ("FATAL: deadlock detected");
146}; 145};
147 146
148sub _cancel {
149 my ($self) = @_;
150
151 # free coroutine data and mark as destructed
152 $self->_destroy
153 or return;
154
155 # call all destruction callbacks
156 $_->(@{$self->{status}})
157 for @{(delete $self->{destroy_cb}) || []};
158}
159
160# this coroutine is necessary because a coroutine 147# this coroutine is necessary because a coroutine
161# cannot destroy itself. 148# cannot destroy itself.
162my @destroy; 149our @destroy;
163my $manager; 150our $manager;
164 151
165$manager = new Coro sub { 152$manager = new Coro sub {
166 while () { 153 while () {
167 (shift @destroy)->_cancel 154 Coro::_cancel shift @destroy
168 while @destroy; 155 while @destroy;
169 156
170 &schedule; 157 &schedule;
171 } 158 }
172}; 159};
173$manager->desc ("[coro manager]"); 160$manager->{desc} = "[coro manager]";
174$manager->prio (PRIO_MAX); 161$manager->prio (PRIO_MAX);
175 162
176# static methods. not really.
177
178=back 163=back
179 164
180=head2 STATIC METHODS 165=head1 SIMPLE COROUTINE CREATION
181
182Static methods are actually functions that operate on the current coroutine only.
183 166
184=over 4 167=over 4
185 168
186=item async { ... } [@args...] 169=item async { ... } [@args...]
187 170
188Create a new asynchronous coroutine and return it's coroutine object 171Create a new coroutine and return it's coroutine object (usually
189(usually unused). When the sub returns the new coroutine is automatically 172unused). The coroutine will be put into the ready queue, so
173it will start running automatically on the next scheduler run.
174
175The first argument is a codeblock/closure that should be executed in the
176coroutine. When it returns argument returns the coroutine is automatically
190terminated. 177terminated.
178
179The remaining arguments are passed as arguments to the closure.
180
181See the C<Coro::State::new> constructor for info about the coroutine
182environment in which coroutines are executed.
191 183
192Calling C<exit> in a coroutine will do the same as calling exit outside 184Calling C<exit> in a coroutine will do the same as calling exit outside
193the coroutine. Likewise, when the coroutine dies, the program will exit, 185the coroutine. Likewise, when the coroutine dies, the program will exit,
194just as it would in the main program. 186just as it would in the main program.
195 187
188If you do not want that, you can provide a default C<die> handler, or
189simply avoid dieing (by use of C<eval>).
190
196 # create a new coroutine that just prints its arguments 191Example: Create a new coroutine that just prints its arguments.
192
197 async { 193 async {
198 print "@_\n"; 194 print "@_\n";
199 } 1,2,3,4; 195 } 1,2,3,4;
200 196
201=cut 197=cut
207} 203}
208 204
209=item async_pool { ... } [@args...] 205=item async_pool { ... } [@args...]
210 206
211Similar to C<async>, but uses a coroutine pool, so you should not call 207Similar to C<async>, but uses a coroutine pool, so you should not call
212terminate or join (although you are allowed to), and you get a coroutine 208terminate or join on it (although you are allowed to), and you get a
213that might have executed other code already (which can be good or bad :). 209coroutine that might have executed other code already (which can be good
210or bad :).
214 211
212On the plus side, this function is about twice as fast as creating (and
213destroying) a completely new coroutine, so if you need a lot of generic
214coroutines in quick successsion, use C<async_pool>, not C<async>.
215
215Also, the block is executed in an C<eval> context and a warning will be 216The code block is executed in an C<eval> context and a warning will be
216issued in case of an exception instead of terminating the program, as 217issued in case of an exception instead of terminating the program, as
217C<async> does. As the coroutine is being reused, stuff like C<on_destroy> 218C<async> does. As the coroutine is being reused, stuff like C<on_destroy>
218will not work in the expected way, unless you call terminate or cancel, 219will not work in the expected way, unless you call terminate or cancel,
219which somehow defeats the purpose of pooling. 220which somehow defeats the purpose of pooling (but is fine in the
221exceptional case).
220 222
221The priority will be reset to C<0> after each job, otherwise the coroutine 223The priority will be reset to C<0> after each run, tracing will be
222will be re-used "as-is". 224disabled, the description will be reset and the default output filehandle
225gets restored, so you can change all these. Otherwise the coroutine will
226be re-used "as-is": most notably if you change other per-coroutine global
227stuff such as C<$/> you I<must needs> revert that change, which is most
228simply done by using local as in: C<< local $/ >>.
223 229
224The pool size is limited to 8 idle coroutines (this can be adjusted by 230The idle pool size is limited to C<8> idle coroutines (this can be
225changing $Coro::POOL_SIZE), and there can be as many non-idle coros as 231adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle
226required. 232coros as required.
227 233
228If you are concerned about pooled coroutines growing a lot because a 234If you are concerned about pooled coroutines growing a lot because a
229single C<async_pool> used a lot of stackspace you can e.g. C<async_pool 235single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
230{ terminate }> once per second or so to slowly replenish the pool. In 236{ terminate }> once per second or so to slowly replenish the pool. In
231addition to that, when the stacks used by a handler grows larger than 16kb 237addition to that, when the stacks used by a handler grows larger than 32kb
232(adjustable with $Coro::MAX_POOL_RSS) it will also exit. 238(adjustable via $Coro::POOL_RSS) it will also be destroyed.
233 239
234=cut 240=cut
235 241
236our $POOL_SIZE = 8; 242our $POOL_SIZE = 8;
237our $MAX_POOL_RSS = 16 * 1024; 243our $POOL_RSS = 32 * 1024;
238our @pool; 244our @async_pool;
239 245
240sub pool_handler { 246sub pool_handler {
241 while () { 247 while () {
242 $current->{desc} = "[async_pool]";
243
244 eval { 248 eval {
245 my ($cb, @arg) = @{ delete $current->{_invoke} or return }; 249 &{&_pool_handler} while 1;
246 $cb->(@arg);
247 }; 250 };
251
248 warn $@ if $@; 252 warn $@ if $@;
249
250 last if @pool >= $POOL_SIZE || $current->rss >= $MAX_POOL_RSS;
251
252 push @pool, $current;
253 $current->{desc} = "[async_pool idle]";
254 $current->save (Coro::State::SAVE_DEF);
255 $current->prio (0);
256 schedule;
257 } 253 }
258} 254}
259 255
260sub async_pool(&@) { 256=back
261 # this is also inlined into the unlock_scheduler
262 my $coro = (pop @pool) || new Coro \&pool_handler;;
263 257
264 $coro->{_invoke} = [@_]; 258=head1 STATIC METHODS
265 $coro->ready;
266 259
267 $coro 260Static methods are actually functions that implicitly operate on the
268} 261current coroutine.
262
263=over 4
269 264
270=item schedule 265=item schedule
271 266
272Calls the scheduler. Please note that the current coroutine will not be put 267Calls the scheduler. The scheduler will find the next coroutine that is
268to be run from the ready queue and switches to it. The next coroutine
269to be run is simply the one with the highest priority that is longest
270in its ready queue. If there is no coroutine ready, it will clal the
271C<$Coro::idle> hook.
272
273Please note that the current coroutine will I<not> be put into the ready
273into the ready queue, so calling this function usually means you will 274queue, so calling this function usually means you will never be called
274never be called again unless something else (e.g. an event handler) calls 275again unless something else (e.g. an event handler) calls C<< ->ready >>,
275ready. 276thus waking you up.
276 277
277The canonical way to wait on external events is this: 278This makes C<schedule> I<the> generic method to use to block the current
279coroutine and wait for events: first you remember the current coroutine in
280a 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
282yourself to sleep. Note that a lot of things can wake your coroutine up,
283so you need to check whether the event indeed happened, e.g. by storing the
284status in a variable.
278 285
286See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks.
287
288=item cede
289
290"Cede" to other coroutines. This function puts the current coroutine into
291the ready queue and calls C<schedule>, which has the effect of giving
292up the current "timeslice" to other coroutines of the same or higher
293priority. Once your coroutine gets its turn again it will automatically be
294resumed.
295
296This function is often called C<yield> in other languages.
297
298=item Coro::cede_notself
299
300Works like cede, but is not exported by default and will cede to I<any>
301coroutine, regardless of priority. This is useful sometimes to ensure
302progress is made.
303
304=item terminate [arg...]
305
306Terminates the current coroutine with the given status values (see L<cancel>).
307
308=item killall
309
310Kills/terminates/cancels all coroutines except the currently running
311one. This is useful after a fork, either in the child or the parent, as
312usually only one of them should inherit the running coroutines.
313
314Note that while this will try to free some of the main programs resources,
315you cannot free all of them, so if a coroutine that is not the main
316program calls this function, there will be some one-time resource leak.
317
318=cut
319
320sub killall {
321 for (Coro::State::list) {
322 $_->cancel
323 if $_ != $current && UNIVERSAL::isa $_, "Coro";
279 { 324 }
280 # remember current coroutine 325}
326
327=back
328
329=head1 COROUTINE OBJECT METHODS
330
331These are the methods you can call on coroutine objects (or to create
332them).
333
334=over 4
335
336=item new Coro \&sub [, @args...]
337
338Create a new coroutine and return it. When the sub returns, the coroutine
339automatically terminates as if C<terminate> with the returned values were
340called. To make the coroutine run you must first put it into the ready
341queue by calling the ready method.
342
343See C<async> and C<Coro::State::new> for additional info about the
344coroutine environment.
345
346=cut
347
348sub _terminate {
349 terminate &{+shift};
350}
351
352=item $success = $coroutine->ready
353
354Put the given coroutine into the end of its ready queue (there is one
355queue for each priority) and return true. If the coroutine is already in
356the ready queue, do nothing and return false.
357
358This ensures that the scheduler will resume this coroutine automatically
359once all the coroutines of higher priority and all coroutines of the same
360priority that were put into the ready queue earlier have been resumed.
361
362=item $is_ready = $coroutine->is_ready
363
364Return whether the coroutine is currently the ready queue or not,
365
366=item $coroutine->cancel (arg...)
367
368Terminates the given coroutine and makes it return the given arguments as
369status (default: the empty list). Never returns if the coroutine is the
370current coroutine.
371
372=cut
373
374sub cancel {
375 my $self = shift;
376
377 if ($current == $self) {
378 terminate @_;
379 } else {
380 $self->{_status} = [@_];
381 $self->_cancel;
382 }
383}
384
385=item $coroutine->schedule_to
386
387Puts the current coroutine to sleep (like C<Coro::schedule>), but instead
388of continuing with the next coro from the ready queue, always switch to
389the given coroutine object (regardless of priority etc.). The readyness
390state of that coroutine isn't changed.
391
392This is an advanced method for special cases - I'd love to hear about any
393uses for this one.
394
395=item $coroutine->cede_to
396
397Like C<schedule_to>, but puts the current coroutine into the ready
398queue. This has the effect of temporarily switching to the given
399coroutine, and continuing some time later.
400
401This is an advanced method for special cases - I'd love to hear about any
402uses for this one.
403
404=item $coroutine->throw ([$scalar])
405
406If C<$throw> is specified and defined, it will be thrown as an exception
407inside the coroutine at the next convenient point in time. Otherwise
408clears the exception object.
409
410Coro will check for the exception each time a schedule-like-function
411returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down
412>>, C<< Coro::Handle->readable >> and so on. Most of these functions
413detect this case and return early in case an exception is pending.
414
415The exception object will be thrown "as is" with the specified scalar in
416C<$@>, i.e. if it is a string, no line number or newline will be appended
417(unlike with C<die>).
418
419This can be used as a softer means than C<cancel> to ask a coroutine to
420end itself, although there is no guarantee that the exception will lead to
421termination, and if the exception isn't caught it might well end the whole
422program.
423
424You might also think of C<throw> as being the moral equivalent of
425C<kill>ing a coroutine with a signal (in this case, a scalar).
426
427=item $coroutine->join
428
429Wait until the coroutine terminates and return any values given to the
430C<terminate> or C<cancel> functions. C<join> can be called concurrently
431from multiple coroutines, and all will be resumed and given the status
432return once the C<$coroutine> terminates.
433
434=cut
435
436sub join {
437 my $self = shift;
438
439 unless ($self->{_status}) {
281 my $current = $Coro::current; 440 my $current = $current;
282 441
283 # register a hypothetical event handler 442 push @{$self->{_on_destroy}}, sub {
284 on_event_invoke sub {
285 # wake up sleeping coroutine
286 $current->ready; 443 $current->ready;
287 undef $current; 444 undef $current;
288 }; 445 };
289 446
290 # call schedule until event occurred.
291 # in case we are woken up for other reasons
292 # (current still defined), loop.
293 Coro::schedule while $current;
294 }
295
296=item cede
297
298"Cede" to other coroutines. This function puts the current coroutine into the
299ready queue and calls C<schedule>, which has the effect of giving up the
300current "timeslice" to other coroutines of the same or higher priority.
301
302Returns true if at least one coroutine switch has happened.
303
304=item Coro::cede_notself
305
306Works like cede, but is not exported by default and will cede to any
307coroutine, regardless of priority, once.
308
309Returns true if at least one coroutine switch has happened.
310
311=item terminate [arg...]
312
313Terminates the current coroutine with the given status values (see L<cancel>).
314
315=cut
316
317sub terminate {
318 $current->cancel (@_);
319}
320
321=back
322
323# dynamic methods
324
325=head2 COROUTINE METHODS
326
327These are the methods you can call on coroutine objects.
328
329=over 4
330
331=item new Coro \&sub [, @args...]
332
333Create a new coroutine and return it. When the sub returns the coroutine
334automatically terminates as if C<terminate> with the returned values were
335called. To make the coroutine run you must first put it into the ready queue
336by calling the ready method.
337
338See C<async> for additional discussion.
339
340=cut
341
342sub _run_coro {
343 terminate &{+shift};
344}
345
346sub new {
347 my $class = shift;
348
349 $class->SUPER::new (\&_run_coro, @_)
350}
351
352=item $success = $coroutine->ready
353
354Put the given coroutine into the ready queue (according to it's priority)
355and return true. If the coroutine is already in the ready queue, do nothing
356and return false.
357
358=item $is_ready = $coroutine->is_ready
359
360Return wether the coroutine is currently the ready queue or not,
361
362=item $coroutine->cancel (arg...)
363
364Terminates the given coroutine and makes it return the given arguments as
365status (default: the empty list). Never returns if the coroutine is the
366current coroutine.
367
368=cut
369
370sub cancel {
371 my $self = shift;
372 $self->{status} = [@_];
373
374 if ($current == $self) {
375 push @destroy, $self;
376 $manager->ready;
377 &schedule while 1;
378 } else {
379 $self->_cancel;
380 }
381}
382
383=item $coroutine->join
384
385Wait until the coroutine terminates and return any values given to the
386C<terminate> or C<cancel> functions. C<join> can be called multiple times
387from multiple coroutine.
388
389=cut
390
391sub join {
392 my $self = shift;
393
394 unless ($self->{status}) {
395 my $current = $current;
396
397 push @{$self->{destroy_cb}}, sub {
398 $current->ready;
399 undef $current;
400 };
401
402 &schedule while $current; 447 &schedule while $current;
403 } 448 }
404 449
405 wantarray ? @{$self->{status}} : $self->{status}[0]; 450 wantarray ? @{$self->{_status}} : $self->{_status}[0];
406} 451}
407 452
408=item $coroutine->on_destroy (\&cb) 453=item $coroutine->on_destroy (\&cb)
409 454
410Registers a callback that is called when this coroutine gets destroyed, 455Registers a callback that is called when this coroutine gets destroyed,
411but before it is joined. The callback gets passed the terminate arguments, 456but before it is joined. The callback gets passed the terminate arguments,
412if any. 457if any, and I<must not> die, under any circumstances.
413 458
414=cut 459=cut
415 460
416sub on_destroy { 461sub on_destroy {
417 my ($self, $cb) = @_; 462 my ($self, $cb) = @_;
418 463
419 push @{ $self->{destroy_cb} }, $cb; 464 push @{ $self->{_on_destroy} }, $cb;
420} 465}
421 466
422=item $oldprio = $coroutine->prio ($newprio) 467=item $oldprio = $coroutine->prio ($newprio)
423 468
424Sets (or gets, if the argument is missing) the priority of the 469Sets (or gets, if the argument is missing) the priority of the
447higher values mean lower priority, just as in unix). 492higher values mean lower priority, just as in unix).
448 493
449=item $olddesc = $coroutine->desc ($newdesc) 494=item $olddesc = $coroutine->desc ($newdesc)
450 495
451Sets (or gets in case the argument is missing) the description for this 496Sets (or gets in case the argument is missing) the description for this
452coroutine. This is just a free-form string you can associate with a coroutine. 497coroutine. This is just a free-form string you can associate with a
498coroutine.
499
500This method simply sets the C<< $coroutine->{desc} >> member to the given
501string. You can modify this member directly if you wish.
453 502
454=cut 503=cut
455 504
456sub desc { 505sub desc {
457 my $old = $_[0]{desc}; 506 my $old = $_[0]{desc};
458 $_[0]{desc} = $_[1] if @_ > 1; 507 $_[0]{desc} = $_[1] if @_ > 1;
459 $old; 508 $old;
460} 509}
461 510
511sub 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");
514}
515
462=back 516=back
463 517
464=head2 GLOBAL FUNCTIONS 518=head1 GLOBAL FUNCTIONS
465 519
466=over 4 520=over 4
467 521
468=item Coro::nready 522=item Coro::nready
469 523
470Returns the number of coroutines that are currently in the ready state, 524Returns the number of coroutines that are currently in the ready state,
471i.e. that can be switched to. The value C<0> means that the only runnable 525i.e. that can be switched to by calling C<schedule> directory or
526indirectly. The value C<0> means that the only runnable coroutine is the
472coroutine is the currently running one, so C<cede> would have no effect, 527currently running one, so C<cede> would have no effect, and C<schedule>
473and C<schedule> would cause a deadlock unless there is an idle handler 528would cause a deadlock unless there is an idle handler that wakes up some
474that wakes up some coroutines. 529coroutines.
475 530
476=item my $guard = Coro::guard { ... } 531=item my $guard = Coro::guard { ... }
477 532
478This creates and returns a guard object. Nothing happens until the object 533This creates and returns a guard object. Nothing happens until the object
479gets destroyed, in which case the codeblock given as argument will be 534gets destroyed, in which case the codeblock given as argument will be
508 563
509 564
510=item unblock_sub { ... } 565=item unblock_sub { ... }
511 566
512This utility function takes a BLOCK or code reference and "unblocks" it, 567This utility function takes a BLOCK or code reference and "unblocks" it,
513returning the new coderef. This means that the new coderef will return 568returning a new coderef. Unblocking means that calling the new coderef
514immediately without blocking, returning nothing, while the original code 569will return immediately without blocking, returning nothing, while the
515ref will be called (with parameters) from within its own coroutine. 570original code ref will be called (with parameters) from within another
571coroutine.
516 572
517The reason this function exists is that many event libraries (such as the 573The reason this function exists is that many event libraries (such as the
518venerable L<Event|Event> module) are not coroutine-safe (a weaker form 574venerable L<Event|Event> module) are not coroutine-safe (a weaker form
519of thread-safety). This means you must not block within event callbacks, 575of thread-safety). This means you must not block within event callbacks,
520otherwise you might suffer from crashes or worse. 576otherwise you might suffer from crashes or worse. The only event library
577currently known that is safe to use without C<unblock_sub> is L<EV>.
521 578
522This function allows your callbacks to block by executing them in another 579This function allows your callbacks to block by executing them in another
523coroutine where it is safe to block. One example where blocking is handy 580coroutine where it is safe to block. One example where blocking is handy
524is when you use the L<Coro::AIO|Coro::AIO> functions to save results to 581is when you use the L<Coro::AIO|Coro::AIO> functions to save results to
525disk. 582disk, for example.
526 583
527In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when 584In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
528creating event callbacks that want to block. 585creating event callbacks that want to block.
586
587If your handler does not plan to block (e.g. simply sends a message to
588another coroutine, or puts some other coroutine into the ready queue),
589there is no reason to use C<unblock_sub>.
590
591Note that you also need to use C<unblock_sub> for any other callbacks that
592are indirectly executed by any C-based event loop. For example, when you
593use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
594provides callbacks that are the result of some event callback, then you
595must not block either, or use C<unblock_sub>.
529 596
530=cut 597=cut
531 598
532our @unblock_queue; 599our @unblock_queue;
533 600
536# return immediately and can be reused) and because we cannot cede 603# return immediately and can be reused) and because we cannot cede
537# inside an event callback. 604# inside an event callback.
538our $unblock_scheduler = new Coro sub { 605our $unblock_scheduler = new Coro sub {
539 while () { 606 while () {
540 while (my $cb = pop @unblock_queue) { 607 while (my $cb = pop @unblock_queue) {
541 # this is an inlined copy of async_pool 608 &async_pool (@$cb);
542 my $coro = (pop @pool or new Coro \&pool_handler);
543 609
544 $coro->{_invoke} = $cb;
545 $coro->ready;
546 cede; # for short-lived callbacks, this reduces pressure on the coro pool 610 # for short-lived callbacks, this reduces pressure on the coro pool
611 # as the chance is very high that the async_poll coro will be back
612 # in the idle state when cede returns
613 cede;
547 } 614 }
548 schedule; # sleep well 615 schedule; # sleep well
549 } 616 }
550}; 617};
551$unblock_scheduler->desc ("[unblock_sub scheduler]"); 618$unblock_scheduler->{desc} = "[unblock_sub scheduler]";
552 619
553sub unblock_sub(&) { 620sub unblock_sub(&) {
554 my $cb = shift; 621 my $cb = shift;
555 622
556 sub { 623 sub {
557 unshift @unblock_queue, [$cb, @_]; 624 unshift @unblock_queue, [$cb, @_];
558 $unblock_scheduler->ready; 625 $unblock_scheduler->ready;
559 } 626 }
560} 627}
561 628
629=item $cb = Coro::rouse_cb
630
631Create and return a "rouse callback". That's a code reference that, when
632called, will save its arguments and notify the owner coroutine of the
633callback.
634
635See the next function.
636
637=item @args = Coro::rouse_wait [$cb]
638
639Wait for the specified rouse callback (or the last one tht was created in
640this coroutine).
641
642As soon as the callback is invoked (or when the calback was invoked before
643C<rouse_wait>), it will return a copy of the arguments originally passed
644to the rouse callback.
645
646See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example.
647
562=back 648=back
563 649
564=cut 650=cut
565 651
5661; 6521;
567 653
654=head1 HOW TO WAIT FOR A CALLBACK
655
656It is very common for a coroutine to wait for some callback to be
657called. This occurs naturally when you use coroutines in an otherwise
658event-based program, or when you use event-based libraries.
659
660These typically register a callback for some event, and call that callback
661when the event occured. In a coroutine, however, you typically want to
662just wait for the event, simplyifying things.
663
664For example C<< AnyEvent->child >> registers a callback to be called when
665a specific child has exited:
666
667 my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });
668
669But from withina coroutine, you often just want to write this:
670
671 my $status = wait_for_child $pid;
672
673Coro offers two functions specifically designed to make this easy,
674C<Coro::rouse_cb> and C<Coro::rouse_wait>.
675
676The first function, C<rouse_cb>, generates and returns a callback that,
677when invoked, will save it's arguments and notify the coroutine that
678created the callback.
679
680The second function, C<rouse_wait>, waits for the callback to be called
681(by calling C<schedule> to go to sleep) and returns the arguments
682originally passed to the callback.
683
684Using these functions, it becomes easy to write the C<wait_for_child>
685function mentioned above:
686
687 sub wait_for_child($) {
688 my ($pid) = @_;
689
690 my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb);
691
692 my ($rpid, $rstatus) = Coro::rouse_wait;
693 $rstatus
694 }
695
696In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough,
697you can roll your own, using C<schedule>:
698
699 sub wait_for_child($) {
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
568=head1 BUGS/LIMITATIONS 720=head1 BUGS/LIMITATIONS
569 721
570 - you must make very sure that no coro is still active on global 722=over 4
571 destruction. very bad things might happen otherwise (usually segfaults).
572 723
724=item fork with pthread backend
725
726When Coro is compiled using the pthread backend (which isn't recommended
727but required on many BSDs as their libcs are completely broken), then
728coroutines will not survive a fork. There is no known workaround except to
729fix your libc and use a saner backend.
730
731=item perl process emulation ("threads")
732
573 - this module is not thread-safe. You should only ever use this module 733This module is not perl-pseudo-thread-safe. You should only ever use this
574 from the same thread (this requirement might be loosened in the future 734module from the same thread (this requirement might be removed in the
575 to allow per-thread schedulers, but Coro::State does not yet allow 735future to allow per-thread schedulers, but Coro::State does not yet allow
576 this). 736this). I recommend disabling thread support and using processes, as having
737the windows process emulation enabled under unix roughly halves perl
738performance, even when not used.
739
740=item coroutine switching not signal safe
741
742You must not switch to another coroutine from within a signal handler
743(only relevant with %SIG - most event libraries provide safe signals).
744
745That means you I<MUST NOT> call any function that might "block" the
746current coroutine - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or
747anything that calls those. Everything else, including calling C<ready>,
748works.
749
750=back
751
577 752
578=head1 SEE ALSO 753=head1 SEE ALSO
579 754
755Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
756
757Debugging: L<Coro::Debug>.
758
580Support/Utility: L<Coro::Cont>, L<Coro::Specific>, L<Coro::State>, L<Coro::Util>. 759Support/Utility: L<Coro::Specific>, L<Coro::Util>.
581 760
582Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>. 761Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>.
583 762
584Event/IO: L<Coro::Timer>, L<Coro::Event>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::Select>. 763IO/Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
585 764
586Embedding: L<Coro:MakeMaker> 765Compatibility: L<Coro::LWP>, L<Coro::BDB>, L<Coro::Storable>, L<Coro::Select>.
766
767XS API: L<Coro::MakeMaker>.
768
769Low level Configuration, Coroutine Environment: L<Coro::State>.
587 770
588=head1 AUTHOR 771=head1 AUTHOR
589 772
590 Marc Lehmann <schmorp@schmorp.de> 773 Marc Lehmann <schmorp@schmorp.de>
591 http://home.schmorp.de/ 774 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines