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

Comparing Coro/Coro.pm (file contents):
Revision 1.130 by root, Thu Sep 20 12:02:25 2007 UTC vs.
Revision 1.186 by root, Sun May 25 01:32:36 2008 UTC

2 2
3Coro - coroutine process abstraction 3Coro - coroutine process abstraction
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 my $lock = new Coro::Semaphore;
19 cede; 22 my $locked;
23
24 $lock->down;
25 $locked = 1;
26 $lock->up;
20 27
21=head1 DESCRIPTION 28=head1 DESCRIPTION
22 29
23This module collection manages coroutines. Coroutines are similar 30This module collection manages coroutines. Coroutines are similar to
24to threads but don't run in parallel at the same time even on SMP 31threads but don't (in general) run in parallel at the same time even
25machines. The specific flavor of coroutine used in this module also 32on SMP machines. The specific flavor of coroutine used in this module
26guarantees you that it will not switch between coroutines unless 33also guarantees you that it will not switch between coroutines unless
27necessary, at easily-identified points in your program, so locking and 34necessary, at easily-identified points in your program, so locking and
28parallel access are rarely an issue, making coroutine programming much 35parallel access are rarely an issue, making coroutine programming much
29safer than threads programming. 36safer and easier than threads programming.
30 37
31(Perl, however, does not natively support real threads but instead does a 38Unlike a normal perl program, however, coroutines allow you to have
32very slow and memory-intensive emulation of processes using threads. This 39multiple running interpreters that share data, which is especially useful
33is a performance win on Windows machines, and a loss everywhere else). 40to code pseudo-parallel processes and for event-based programming, such as
41multiple HTTP-GET requests running concurrently. See L<Coro::AnyEvent> to
42learn more.
43
44Coroutines are also useful because Perl has no support for threads (the so
45called "threads" that perl offers are nothing more than the (bad) process
46emulation coming from the Windows platform: On standard operating systems
47they serve no purpose whatsoever, except by making your programs slow and
48making them use a lot of memory. Best disable them when building perl, or
49aks your software vendor/distributor to do it for you).
34 50
35In this module, coroutines are defined as "callchain + lexical variables + 51In this module, coroutines are defined as "callchain + lexical variables +
36@_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own callchain, 52@_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own callchain,
37its own set of lexicals and its own set of perls most important global 53its own set of lexicals and its own set of perls most important global
38variables. 54variables (see L<Coro::State> for more configuration).
39 55
40=cut 56=cut
41 57
42package Coro; 58package Coro;
43 59
50 66
51our $idle; # idle handler 67our $idle; # idle handler
52our $main; # main coroutine 68our $main; # main coroutine
53our $current; # current coroutine 69our $current; # current coroutine
54 70
55our $VERSION = '3.7'; 71our $VERSION = '4.72';
56 72
57our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); 73our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub);
58our %EXPORT_TAGS = ( 74our %EXPORT_TAGS = (
59 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], 75 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
60); 76);
61our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready)); 77our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
62 78
63{
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
97=over 4 79=over 4
98 80
99=item $main 81=item $Coro::main
100 82
101This coroutine represents the main program. 83This variable stores the coroutine object that represents the main
84program. While you cna C<ready> it and do most other things you can do to
85coroutines, it is mainly useful to compare again C<$Coro::current>, to see
86wether you are running in the main program or not.
102 87
103=cut 88=cut
104 89
105$main = new Coro; 90$main = new Coro;
106 91
107=item $current (or as function: current) 92=item $Coro::current
108 93
109The current coroutine (the last coroutine switched to). The initial value 94The coroutine object representing the current coroutine (the last
95coroutine that the Coro scheduler switched to). The initial value is
110is C<$main> (of course). 96C<$main> (of course).
111 97
112This variable is B<strictly> I<read-only>. It is provided for performance 98This variable is B<strictly> I<read-only>. You can take copies of the
113reasons. If performance is not essential you are encouraged to use the 99value stored in it and use it as any other coroutine object, but you must
114C<Coro::current> function instead. 100not otherwise modify the variable itself.
115 101
116=cut 102=cut
103
104$main->{desc} = "[main::]";
117 105
118# maybe some other module used Coro::Specific before... 106# maybe some other module used Coro::Specific before...
119$main->{specific} = $current->{specific} 107$main->{_specific} = $current->{_specific}
120 if $current; 108 if $current;
121 109
122_set_current $main; 110_set_current $main;
123 111
124sub current() { $current } 112sub current() { $current } # [DEPRECATED]
125 113
126=item $idle 114=item $Coro::idle
127 115
128A callback that is called whenever the scheduler finds no ready coroutines 116This variable is mainly useful to integrate Coro into event loops. It is
129to run. The default implementation prints "FATAL: deadlock detected" and 117usually better to rely on L<Coro::AnyEvent> or LC<Coro::EV>, as this is
130exits, because the program has no other way to continue. 118pretty low-level functionality.
119
120This variable stores a callback that is called whenever the scheduler
121finds no ready coroutines to run. The default implementation prints
122"FATAL: deadlock detected" and exits, because the program has no other way
123to continue.
131 124
132This hook is overwritten by modules such as C<Coro::Timer> and 125This hook is overwritten by modules such as C<Coro::Timer> and
133C<Coro::Event> to wait on an external event that hopefully wake up a 126C<Coro::AnyEvent> to wait on an external event that hopefully wake up a
134coroutine so the scheduler can run it. 127coroutine so the scheduler can run it.
135 128
129Note that the callback I<must not>, under any circumstances, block
130the current coroutine. Normally, this is achieved by having an "idle
131coroutine" that calls the event loop and then blocks again, and then
132readying that coroutine in the idle handler.
133
134See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this
135technique.
136
136Please note that if your callback recursively invokes perl (e.g. for event 137Please note that if your callback recursively invokes perl (e.g. for event
137handlers), then it must be prepared to be called recursively. 138handlers), then it must be prepared to be called recursively itself.
138 139
139=cut 140=cut
140 141
141$idle = sub { 142$idle = sub {
142 require Carp; 143 require Carp;
149 # free coroutine data and mark as destructed 150 # free coroutine data and mark as destructed
150 $self->_destroy 151 $self->_destroy
151 or return; 152 or return;
152 153
153 # call all destruction callbacks 154 # call all destruction callbacks
154 $_->(@{$self->{status}}) 155 $_->(@{$self->{_status}})
155 for @{(delete $self->{destroy_cb}) || []}; 156 for @{(delete $self->{_on_destroy}) || []};
156} 157}
157 158
158# this coroutine is necessary because a coroutine 159# this coroutine is necessary because a coroutine
159# cannot destroy itself. 160# cannot destroy itself.
160my @destroy; 161my @destroy;
161my $manager; 162my $manager;
162 163
163$manager = new Coro sub { 164$manager = new Coro sub {
164 $current->desc ("[coro manager]");
165
166 while () { 165 while () {
167 (shift @destroy)->_cancel 166 (shift @destroy)->_cancel
168 while @destroy; 167 while @destroy;
169 168
170 &schedule; 169 &schedule;
171 } 170 }
172}; 171};
173 172$manager->desc ("[coro manager]");
174$manager->prio (PRIO_MAX); 173$manager->prio (PRIO_MAX);
175 174
176# static methods. not really.
177
178=back 175=back
179 176
180=head2 STATIC METHODS 177=head2 SIMPLE COROUTINE CREATION
181
182Static methods are actually functions that operate on the current coroutine only.
183 178
184=over 4 179=over 4
185 180
186=item async { ... } [@args...] 181=item async { ... } [@args...]
187 182
188Create a new asynchronous coroutine and return it's coroutine object 183Create a new coroutine and return it's coroutine object (usually
189(usually unused). When the sub returns the new coroutine is automatically 184unused). The coroutine will be put into the ready queue, so
185it will start running automatically on the next scheduler run.
186
187The first argument is a codeblock/closure that should be executed in the
188coroutine. When it returns argument returns the coroutine is automatically
190terminated. 189terminated.
190
191The remaining arguments are passed as arguments to the closure.
192
193See the C<Coro::State::new> constructor for info about the coroutine
194environment in which coroutines are executed.
191 195
192Calling C<exit> in a coroutine will do the same as calling exit outside 196Calling C<exit> in a coroutine will do the same as calling exit outside
193the coroutine. Likewise, when the coroutine dies, the program will exit, 197the coroutine. Likewise, when the coroutine dies, the program will exit,
194just as it would in the main program. 198just as it would in the main program.
195 199
200If you do not want that, you can provide a default C<die> handler, or
201simply avoid dieing (by use of C<eval>).
202
196 # create a new coroutine that just prints its arguments 203Example: Create a new coroutine that just prints its arguments.
204
197 async { 205 async {
198 print "@_\n"; 206 print "@_\n";
199 } 1,2,3,4; 207 } 1,2,3,4;
200 208
201=cut 209=cut
207} 215}
208 216
209=item async_pool { ... } [@args...] 217=item async_pool { ... } [@args...]
210 218
211Similar to C<async>, but uses a coroutine pool, so you should not call 219Similar 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 220terminate 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 :). 221coroutine that might have executed other code already (which can be good
222or bad :).
214 223
224On the plus side, this function is faster than creating (and destroying)
225a completely new coroutine, so if you need a lot of generic coroutines in
226quick successsion, use C<async_pool>, not C<async>.
227
215Also, the block is executed in an C<eval> context and a warning will be 228The 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 229issued 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> 230C<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, 231will not work in the expected way, unless you call terminate or cancel,
219which somehow defeats the purpose of pooling. 232which somehow defeats the purpose of pooling (but is fine in the
233exceptional case).
220 234
221The priority will be reset to C<0> after each job, otherwise the coroutine 235The priority will be reset to C<0> after each run, tracing will be
222will be re-used "as-is". 236disabled, the description will be reset and the default output filehandle
237gets restored, so you can change all these. Otherwise the coroutine will
238be re-used "as-is": most notably if you change other per-coroutine global
239stuff such as C<$/> you I<must needs> to revert that change, which is most
240simply done by using local as in: C< local $/ >.
223 241
224The pool size is limited to 8 idle coroutines (this can be adjusted by 242The pool size is limited to C<8> idle coroutines (this can be adjusted by
225changing $Coro::POOL_SIZE), and there can be as many non-idle coros as 243changing $Coro::POOL_SIZE), and there can be as many non-idle coros as
226required. 244required.
227 245
228If you are concerned about pooled coroutines growing a lot because a 246If 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 { 247single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
230terminate }> once per second or so to slowly replenish the pool. 248{ terminate }> once per second or so to slowly replenish the pool. In
249addition to that, when the stacks used by a handler grows larger than 16kb
250(adjustable via $Coro::POOL_RSS) it will also be destroyed.
231 251
232=cut 252=cut
233 253
234our $POOL_SIZE = 8; 254our $POOL_SIZE = 8;
235our $MAX_POOL_RSS = 64 * 1024; 255our $POOL_RSS = 16 * 1024;
236our @pool; 256our @async_pool;
237 257
238sub pool_handler { 258sub pool_handler {
259 my $cb;
260
239 while () { 261 while () {
240 $current->{desc} = "[async_pool]";
241
242 eval { 262 eval {
243 my ($cb, @arg) = @{ delete $current->{_invoke} or return }; 263 while () {
244 $cb->(@arg); 264 _pool_1 $cb;
265 &$cb;
266 _pool_2 $cb;
267 &schedule;
268 }
245 }; 269 };
270
271 last if $@ eq "\3async_pool terminate\2\n";
246 warn $@ if $@; 272 warn $@ if $@;
247
248 last if @pool >= $POOL_SIZE || $current->rss >= $MAX_POOL_RSS;
249
250 push @pool, $current;
251 $current->{desc} = "[async_pool idle]";
252 $current->save (Coro::State::SAVE_DEF);
253 $current->prio (0);
254 schedule;
255 } 273 }
256} 274}
257 275
258sub async_pool(&@) { 276sub async_pool(&@) {
259 # this is also inlined into the unlock_scheduler 277 # this is also inlined into the unlock_scheduler
260 my $coro = (pop @pool) || new Coro \&pool_handler;; 278 my $coro = (pop @async_pool) || new Coro \&pool_handler;
261 279
262 $coro->{_invoke} = [@_]; 280 $coro->{_invoke} = [@_];
263 $coro->ready; 281 $coro->ready;
264 282
265 $coro 283 $coro
266} 284}
267 285
286=back
287
288=head2 STATIC METHODS
289
290Static methods are actually functions that operate on the current coroutine.
291
292=over 4
293
268=item schedule 294=item schedule
269 295
270Calls the scheduler. Please note that the current coroutine will not be put 296Calls the scheduler. The scheduler will find the next coroutine that is
297to be run from the ready queue and switches to it. The next coroutine
298to be run is simply the one with the highest priority that is longest
299in its ready queue. If there is no coroutine ready, it will clal the
300C<$Coro::idle> hook.
301
302Please note that the current coroutine will I<not> be put into the ready
271into the ready queue, so calling this function usually means you will 303queue, so calling this function usually means you will never be called
272never be called again unless something else (e.g. an event handler) calls 304again unless something else (e.g. an event handler) calls C<< ->ready >>,
273ready. 305thus waking you up.
306
307This makes C<schedule> I<the> generic method to use to block the current
308coroutine and wait for events: first you remember the current coroutine in
309a variable, then arrange for some callback of yours to call C<< ->ready
310>> on that once some event happens, and last you call C<schedule> to put
311yourself to sleep. Note that a lot of things can wake your coroutine up,
312so you need to check wether the event indeed happened, e.g. by storing the
313status in a variable.
274 314
275The canonical way to wait on external events is this: 315The canonical way to wait on external events is this:
276 316
277 { 317 {
278 # remember current coroutine 318 # remember current coroutine
291 Coro::schedule while $current; 331 Coro::schedule while $current;
292 } 332 }
293 333
294=item cede 334=item cede
295 335
296"Cede" to other coroutines. This function puts the current coroutine into the 336"Cede" to other coroutines. This function puts the current coroutine into
297ready queue and calls C<schedule>, which has the effect of giving up the 337the ready queue and calls C<schedule>, which has the effect of giving
298current "timeslice" to other coroutines of the same or higher priority. 338up the current "timeslice" to other coroutines of the same or higher
339priority. Once your coroutine gets its turn again it will automatically be
340resumed.
299 341
300Returns true if at least one coroutine switch has happened. 342This function is often called C<yield> in other languages.
301 343
302=item Coro::cede_notself 344=item Coro::cede_notself
303 345
304Works like cede, but is not exported by default and will cede to any 346Works like cede, but is not exported by default and will cede to I<any>
305coroutine, regardless of priority, once. 347coroutine, regardless of priority. This is useful sometimes to ensure
306 348progress is made.
307Returns true if at least one coroutine switch has happened.
308 349
309=item terminate [arg...] 350=item terminate [arg...]
310 351
311Terminates the current coroutine with the given status values (see L<cancel>). 352Terminates the current coroutine with the given status values (see L<cancel>).
353
354=item killall
355
356Kills/terminates/cancels all coroutines except the currently running
357one. This is useful after a fork, either in the child or the parent, as
358usually only one of them should inherit the running coroutines.
359
360Note that while this will try to free some of the main programs resources,
361you cnanot free all of them, so if a coroutine that is not the main
362program calls this function, there will be some one-time resource leak.
312 363
313=cut 364=cut
314 365
315sub terminate { 366sub terminate {
316 $current->cancel (@_); 367 $current->cancel (@_);
317} 368}
318 369
370sub killall {
371 for (Coro::State::list) {
372 $_->cancel
373 if $_ != $current && UNIVERSAL::isa $_, "Coro";
374 }
375}
376
319=back 377=back
320 378
321# dynamic methods
322
323=head2 COROUTINE METHODS 379=head2 COROUTINE METHODS
324 380
325These are the methods you can call on coroutine objects. 381These are the methods you can call on coroutine objects (or to create
382them).
326 383
327=over 4 384=over 4
328 385
329=item new Coro \&sub [, @args...] 386=item new Coro \&sub [, @args...]
330 387
331Create a new coroutine and return it. When the sub returns the coroutine 388Create a new coroutine and return it. When the sub returns, the coroutine
332automatically terminates as if C<terminate> with the returned values were 389automatically terminates as if C<terminate> with the returned values were
333called. To make the coroutine run you must first put it into the ready queue 390called. To make the coroutine run you must first put it into the ready
334by calling the ready method. 391queue by calling the ready method.
335 392
336See C<async> for additional discussion. 393See C<async> and C<Coro::State::new> for additional info about the
394coroutine environment.
337 395
338=cut 396=cut
339 397
340sub _run_coro { 398sub _run_coro {
341 terminate &{+shift}; 399 terminate &{+shift};
347 $class->SUPER::new (\&_run_coro, @_) 405 $class->SUPER::new (\&_run_coro, @_)
348} 406}
349 407
350=item $success = $coroutine->ready 408=item $success = $coroutine->ready
351 409
352Put the given coroutine into the ready queue (according to it's priority) 410Put the given coroutine into the end of its ready queue (there is one
353and return true. If the coroutine is already in the ready queue, do nothing 411queue for each priority) and return true. If the coroutine is already in
354and return false. 412the ready queue, do nothing and return false.
413
414This ensures that the scheduler will resume this coroutine automatically
415once all the coroutines of higher priority and all coroutines of the same
416priority that were put into the ready queue earlier have been resumed.
355 417
356=item $is_ready = $coroutine->is_ready 418=item $is_ready = $coroutine->is_ready
357 419
358Return wether the coroutine is currently the ready queue or not, 420Return wether the coroutine is currently the ready queue or not,
359 421
365 427
366=cut 428=cut
367 429
368sub cancel { 430sub cancel {
369 my $self = shift; 431 my $self = shift;
370 $self->{status} = [@_]; 432 $self->{_status} = [@_];
371 433
372 if ($current == $self) { 434 if ($current == $self) {
373 push @destroy, $self; 435 push @destroy, $self;
374 $manager->ready; 436 $manager->ready;
375 &schedule while 1; 437 &schedule while 1;
379} 441}
380 442
381=item $coroutine->join 443=item $coroutine->join
382 444
383Wait until the coroutine terminates and return any values given to the 445Wait until the coroutine terminates and return any values given to the
384C<terminate> or C<cancel> functions. C<join> can be called multiple times 446C<terminate> or C<cancel> functions. C<join> can be called concurrently
385from multiple coroutine. 447from multiple coroutines, and all will be resumed and given the status
448return once the C<$coroutine> terminates.
386 449
387=cut 450=cut
388 451
389sub join { 452sub join {
390 my $self = shift; 453 my $self = shift;
391 454
392 unless ($self->{status}) { 455 unless ($self->{_status}) {
393 my $current = $current; 456 my $current = $current;
394 457
395 push @{$self->{destroy_cb}}, sub { 458 push @{$self->{_on_destroy}}, sub {
396 $current->ready; 459 $current->ready;
397 undef $current; 460 undef $current;
398 }; 461 };
399 462
400 &schedule while $current; 463 &schedule while $current;
401 } 464 }
402 465
403 wantarray ? @{$self->{status}} : $self->{status}[0]; 466 wantarray ? @{$self->{_status}} : $self->{_status}[0];
404} 467}
405 468
406=item $coroutine->on_destroy (\&cb) 469=item $coroutine->on_destroy (\&cb)
407 470
408Registers a callback that is called when this coroutine gets destroyed, 471Registers a callback that is called when this coroutine gets destroyed,
409but before it is joined. The callback gets passed the terminate arguments, 472but before it is joined. The callback gets passed the terminate arguments,
410if any. 473if any, and I<must not> die, under any circumstances.
411 474
412=cut 475=cut
413 476
414sub on_destroy { 477sub on_destroy {
415 my ($self, $cb) = @_; 478 my ($self, $cb) = @_;
416 479
417 push @{ $self->{destroy_cb} }, $cb; 480 push @{ $self->{_on_destroy} }, $cb;
418} 481}
419 482
420=item $oldprio = $coroutine->prio ($newprio) 483=item $oldprio = $coroutine->prio ($newprio)
421 484
422Sets (or gets, if the argument is missing) the priority of the 485Sets (or gets, if the argument is missing) the priority of the
447=item $olddesc = $coroutine->desc ($newdesc) 510=item $olddesc = $coroutine->desc ($newdesc)
448 511
449Sets (or gets in case the argument is missing) the description for this 512Sets (or gets in case the argument is missing) the description for this
450coroutine. This is just a free-form string you can associate with a coroutine. 513coroutine. This is just a free-form string you can associate with a coroutine.
451 514
515This method simply sets the C<< $coroutine->{desc} >> member to the given string. You
516can modify this member directly if you wish.
517
518=item $coroutine->throw ([$scalar])
519
520If C<$throw> is specified and defined, it will be thrown as an exception
521inside the coroutine at the next convinient point in time (usually after
522it gains control at the next schedule/transfer/cede). Otherwise clears the
523exception object.
524
525The exception object will be thrown "as is" with the specified scalar in
526C<$@>, i.e. if it is a string, no line number or newline will be appended
527(unlike with C<die>).
528
529This can be used as a softer means than C<cancel> to ask a coroutine to
530end itself, although there is no guarentee that the exception will lead to
531termination, and if the exception isn't caught it might well end the whole
532program.
533
452=cut 534=cut
453 535
454sub desc { 536sub desc {
455 my $old = $_[0]{desc}; 537 my $old = $_[0]{desc};
456 $_[0]{desc} = $_[1] if @_ > 1; 538 $_[0]{desc} = $_[1] if @_ > 1;
464=over 4 546=over 4
465 547
466=item Coro::nready 548=item Coro::nready
467 549
468Returns the number of coroutines that are currently in the ready state, 550Returns the number of coroutines that are currently in the ready state,
469i.e. that can be switched to. The value C<0> means that the only runnable 551i.e. that can be switched to by calling C<schedule> directory or
552indirectly. The value C<0> means that the only runnable coroutine is the
470coroutine is the currently running one, so C<cede> would have no effect, 553currently running one, so C<cede> would have no effect, and C<schedule>
471and C<schedule> would cause a deadlock unless there is an idle handler 554would cause a deadlock unless there is an idle handler that wakes up some
472that wakes up some coroutines. 555coroutines.
473 556
474=item my $guard = Coro::guard { ... } 557=item my $guard = Coro::guard { ... }
475 558
476This creates and returns a guard object. Nothing happens until the object 559This creates and returns a guard object. Nothing happens until the object
477gets destroyed, in which case the codeblock given as argument will be 560gets destroyed, in which case the codeblock given as argument will be
506 589
507 590
508=item unblock_sub { ... } 591=item unblock_sub { ... }
509 592
510This utility function takes a BLOCK or code reference and "unblocks" it, 593This utility function takes a BLOCK or code reference and "unblocks" it,
511returning the new coderef. This means that the new coderef will return 594returning a new coderef. Unblocking means that calling the new coderef
512immediately without blocking, returning nothing, while the original code 595will return immediately without blocking, returning nothing, while the
513ref will be called (with parameters) from within its own coroutine. 596original code ref will be called (with parameters) from within another
597coroutine.
514 598
515The reason this function exists is that many event libraries (such as the 599The reason this function exists is that many event libraries (such as the
516venerable L<Event|Event> module) are not coroutine-safe (a weaker form 600venerable L<Event|Event> module) are not coroutine-safe (a weaker form
517of thread-safety). This means you must not block within event callbacks, 601of thread-safety). This means you must not block within event callbacks,
518otherwise you might suffer from crashes or worse. 602otherwise you might suffer from crashes or worse. The only event library
603currently known that is safe to use without C<unblock_sub> is L<EV>.
519 604
520This function allows your callbacks to block by executing them in another 605This function allows your callbacks to block by executing them in another
521coroutine where it is safe to block. One example where blocking is handy 606coroutine where it is safe to block. One example where blocking is handy
522is when you use the L<Coro::AIO|Coro::AIO> functions to save results to 607is when you use the L<Coro::AIO|Coro::AIO> functions to save results to
523disk. 608disk, for example.
524 609
525In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when 610In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
526creating event callbacks that want to block. 611creating event callbacks that want to block.
612
613If your handler does not plan to block (e.g. simply sends a message to
614another coroutine, or puts some other coroutine into the ready queue),
615there is no reason to use C<unblock_sub>.
616
617Note that you also need to use C<unblock_sub> for any other callbacks that
618are indirectly executed by any C-based event loop. For example, when you
619use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
620provides callbacks that are the result of some event callback, then you
621must not block either, or use C<unblock_sub>.
527 622
528=cut 623=cut
529 624
530our @unblock_queue; 625our @unblock_queue;
531 626
532# we create a special coro because we want to cede, 627# we create a special coro because we want to cede,
533# to reduce pressure on the coro pool (because most callbacks 628# to reduce pressure on the coro pool (because most callbacks
534# return immediately and can be reused) and because we cannot cede 629# return immediately and can be reused) and because we cannot cede
535# inside an event callback. 630# inside an event callback.
536our $unblock_scheduler = async { 631our $unblock_scheduler = new Coro sub {
537 $current->desc ("[unblock_sub scheduler]");
538 while () { 632 while () {
539 while (my $cb = pop @unblock_queue) { 633 while (my $cb = pop @unblock_queue) {
540 # this is an inlined copy of async_pool 634 # this is an inlined copy of async_pool
541 my $coro = (pop @pool or new Coro \&pool_handler); 635 my $coro = (pop @async_pool) || new Coro \&pool_handler;
542 636
543 $coro->{_invoke} = $cb; 637 $coro->{_invoke} = $cb;
544 $coro->ready; 638 $coro->ready;
545 cede; # for short-lived callbacks, this reduces pressure on the coro pool 639 cede; # for short-lived callbacks, this reduces pressure on the coro pool
546 } 640 }
547 schedule; # sleep well 641 schedule; # sleep well
548 } 642 }
549}; 643};
644$unblock_scheduler->desc ("[unblock_sub scheduler]");
550 645
551sub unblock_sub(&) { 646sub unblock_sub(&) {
552 my $cb = shift; 647 my $cb = shift;
553 648
554 sub { 649 sub {
563 658
5641; 6591;
565 660
566=head1 BUGS/LIMITATIONS 661=head1 BUGS/LIMITATIONS
567 662
568 - you must make very sure that no coro is still active on global
569 destruction. very bad things might happen otherwise (usually segfaults).
570
571 - this module is not thread-safe. You should only ever use this module 663This module is not perl-pseudo-thread-safe. You should only ever use this
572 from the same thread (this requirement might be loosened in the future 664module from the same thread (this requirement might be removed in the
573 to allow per-thread schedulers, but Coro::State does not yet allow 665future to allow per-thread schedulers, but Coro::State does not yet allow
574 this). 666this). I recommend disabling thread support and using processes, as this
667is much faster and uses less memory.
575 668
576=head1 SEE ALSO 669=head1 SEE ALSO
577 670
671Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
672
673Debugging: L<Coro::Debug>.
674
578Support/Utility: L<Coro::Cont>, L<Coro::Specific>, L<Coro::State>, L<Coro::Util>. 675Support/Utility: L<Coro::Specific>, L<Coro::Util>.
579 676
580Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>. 677Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>.
581 678
582Event/IO: L<Coro::Timer>, L<Coro::Event>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::Select>. 679IO/Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
583 680
584Embedding: L<Coro:MakeMaker> 681Compatibility: L<Coro::LWP>, L<Coro::BDB>, L<Coro::Storable>, L<Coro::Select>.
682
683XS API: L<Coro::MakeMaker>.
684
685Low level Configuration, Coroutine Environment: L<Coro::State>.
585 686
586=head1 AUTHOR 687=head1 AUTHOR
587 688
588 Marc Lehmann <schmorp@schmorp.de> 689 Marc Lehmann <schmorp@schmorp.de>
589 http://home.schmorp.de/ 690 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines