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

Comparing cvsroot/Coro/Coro.pm (file contents):
Revision 1.178 by root, Thu Apr 17 22:33:10 2008 UTC vs.
Revision 1.237 by root, Sat Nov 22 16:37:11 2008 UTC

1=head1 NAME 1=head1 NAME
2 2
3Coro - coroutine process abstraction 3Coro - real threads in perl
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use Coro; 7 use Coro;
8 8
9 async { 9 async {
10 # some asynchronous thread of execution 10 # some asynchronous thread of execution
11 print "2\n"; 11 print "2\n";
12 cede; # yield back to main 12 cede; # yield back to main
13 print "4\n"; 13 print "4\n";
14 }; 14 };
15 print "1\n"; 15 print "1\n";
16 cede; # yield to coroutine 16 cede; # yield to coroutine
17 print "3\n"; 17 print "3\n";
18 cede; # and again 18 cede; # and again
19 19
20 # use locking 20 # use locking
21 use Coro::Semaphore;
21 my $lock = new Coro::Semaphore; 22 my $lock = new Coro::Semaphore;
22 my $locked; 23 my $locked;
23 24
24 $lock->down; 25 $lock->down;
25 $locked = 1; 26 $locked = 1;
26 $lock->up; 27 $lock->up;
27 28
28=head1 DESCRIPTION 29=head1 DESCRIPTION
29 30
30This module collection manages coroutines. Coroutines are similar 31For a tutorial-style introduction, please read the L<Coro::Intro>
31to threads but don't run in parallel at the same time even on SMP 32manpage. This manpage mainly contains reference information.
32machines. The specific flavor of coroutine used in this module also
33guarantees you that it will not switch between coroutines unless
34necessary, at easily-identified points in your program, so locking and
35parallel access are rarely an issue, making coroutine programming much
36safer than threads programming.
37 33
38(Perl, however, does not natively support real threads but instead does a 34This module collection manages coroutines, that is, cooperative
39very slow and memory-intensive emulation of processes using threads. This 35threads. Coroutines are similar to kernel threads but don't (in general)
40is a performance win on Windows machines, and a loss everywhere else). 36run in parallel at the same time even on SMP machines. The specific flavor
37of coroutine used in this module also guarantees you that it will not
38switch between coroutines unless necessary, at easily-identified points
39in your program, so locking and parallel access are rarely an issue,
40making coroutine programming much safer and easier than using other thread
41models.
41 42
43Unlike the so-called "Perl threads" (which are not actually real threads
44but only the windows process emulation ported to unix), Coro provides a
45full shared address space, which makes communication between coroutines
46very easy. And coroutines are fast, too: disabling the Windows process
47emulation code in your perl and using Coro can easily result in a two to
48four times speed increase for your programs.
49
50Coro achieves that by supporting multiple running interpreters that share
51data, which is especially useful to code pseudo-parallel processes and
52for event-based programming, such as multiple HTTP-GET requests running
53concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro
54into an event-based environment.
55
42In this module, coroutines are defined as "callchain + lexical variables + 56In this module, a coroutines is defined as "callchain + lexical variables
43@_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own callchain, 57+ @_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own
44its own set of lexicals and its own set of perls most important global 58callchain, its own set of lexicals and its own set of perls most important
45variables (see L<Coro::State> for more configuration). 59global variables (see L<Coro::State> for more configuration and background
60info).
61
62See also the C<SEE ALSO> section at the end of this document - the Coro
63module family is quite large.
46 64
47=cut 65=cut
48 66
49package Coro; 67package Coro;
50 68
51use strict; 69use strict qw(vars subs);
52no warnings "uninitialized"; 70no warnings "uninitialized";
53 71
54use Coro::State; 72use Coro::State;
55 73
56use base qw(Coro::State Exporter); 74use base qw(Coro::State Exporter);
57 75
58our $idle; # idle handler 76our $idle; # idle handler
59our $main; # main coroutine 77our $main; # main coroutine
60our $current; # current coroutine 78our $current; # current coroutine
61 79
62our $VERSION = '4.51'; 80our $VERSION = "5.0";
63 81
64our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); 82our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub);
65our %EXPORT_TAGS = ( 83our %EXPORT_TAGS = (
66 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], 84 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
67); 85);
68our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready)); 86our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
69 87
70{ 88=head1 GLOBAL VARIABLES
71 my @async;
72 my $init;
73
74 # this way of handling attributes simply is NOT scalable ;()
75 sub import {
76 no strict 'refs';
77
78 Coro->export_to_level (1, @_);
79
80 my $old = *{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"}{CODE};
81 *{(caller)[0]."::MODIFY_CODE_ATTRIBUTES"} = sub {
82 my ($package, $ref) = (shift, shift);
83 my @attrs;
84 for (@_) {
85 if ($_ eq "Coro") {
86 push @async, $ref;
87 unless ($init++) {
88 eval q{
89 sub INIT {
90 &async(pop @async) while @async;
91 }
92 };
93 }
94 } else {
95 push @attrs, $_;
96 }
97 }
98 return $old ? $old->($package, $ref, @attrs) : @attrs;
99 };
100 }
101
102}
103 89
104=over 4 90=over 4
105 91
106=item $main 92=item $Coro::main
107 93
108This coroutine represents the main program. 94This variable stores the coroutine object that represents the main
95program. While you cna C<ready> it and do most other things you can do to
96coroutines, it is mainly useful to compare again C<$Coro::current>, to see
97whether you are running in the main program or not.
109 98
110=cut 99=cut
111 100
112$main = new Coro; 101# $main is now being initialised by Coro::State
113 102
114=item $current (or as function: current) 103=item $Coro::current
115 104
116The current coroutine (the last coroutine switched to). The initial value 105The coroutine object representing the current coroutine (the last
106coroutine that the Coro scheduler switched to). The initial value is
117is C<$main> (of course). 107C<$Coro::main> (of course).
118 108
119This variable is B<strictly> I<read-only>. It is provided for performance 109This variable is B<strictly> I<read-only>. You can take copies of the
120reasons. If performance is not essential you are encouraged to use the 110value stored in it and use it as any other coroutine object, but you must
121C<Coro::current> function instead. 111not otherwise modify the variable itself.
122 112
123=cut 113=cut
124 114
125$main->{desc} = "[main::]";
126
127# maybe some other module used Coro::Specific before...
128$main->{_specific} = $current->{_specific}
129 if $current;
130
131_set_current $main;
132
133sub current() { $current } 115sub current() { $current } # [DEPRECATED]
134 116
135=item $idle 117=item $Coro::idle
136 118
137A callback that is called whenever the scheduler finds no ready coroutines 119This variable is mainly useful to integrate Coro into event loops. It is
138to run. The default implementation prints "FATAL: deadlock detected" and 120usually better to rely on L<Coro::AnyEvent> or LC<Coro::EV>, as this is
139exits, because the program has no other way to continue. 121pretty low-level functionality.
122
123This variable stores a callback that is called whenever the scheduler
124finds no ready coroutines to run. The default implementation prints
125"FATAL: deadlock detected" and exits, because the program has no other way
126to continue.
140 127
141This hook is overwritten by modules such as C<Coro::Timer> and 128This hook is overwritten by modules such as C<Coro::Timer> and
142C<Coro::Event> to wait on an external event that hopefully wake up a 129C<Coro::AnyEvent> to wait on an external event that hopefully wake up a
143coroutine so the scheduler can run it. 130coroutine so the scheduler can run it.
131
132Note that the callback I<must not>, under any circumstances, block
133the current coroutine. Normally, this is achieved by having an "idle
134coroutine" that calls the event loop and then blocks again, and then
135readying that coroutine in the idle handler.
136
137See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this
138technique.
144 139
145Please note that if your callback recursively invokes perl (e.g. for event 140Please note that if your callback recursively invokes perl (e.g. for event
146handlers), then it must be prepared to be called recursively itself. 141handlers), then it must be prepared to be called recursively itself.
147 142
148=cut 143=cut
… …
150$idle = sub { 145$idle = sub {
151 require Carp; 146 require Carp;
152 Carp::croak ("FATAL: deadlock detected"); 147 Carp::croak ("FATAL: deadlock detected");
153}; 148};
154 149
155sub _cancel {
156 my ($self) = @_;
157
158 # free coroutine data and mark as destructed
159 $self->_destroy
160 or return;
161
162 # call all destruction callbacks
163 $_->(@{$self->{_status}})
164 for @{(delete $self->{_on_destroy}) || []};
165}
166
167# this coroutine is necessary because a coroutine 150# this coroutine is necessary because a coroutine
168# cannot destroy itself. 151# cannot destroy itself.
169my @destroy; 152our @destroy;
170my $manager; 153our $manager;
171 154
172$manager = new Coro sub { 155$manager = new Coro sub {
173 while () { 156 while () {
174 (shift @destroy)->_cancel 157 Coro::_cancel shift @destroy
175 while @destroy; 158 while @destroy;
176 159
177 &schedule; 160 &schedule;
178 } 161 }
179}; 162};
180$manager->desc ("[coro manager]"); 163$manager->{desc} = "[coro manager]";
181$manager->prio (PRIO_MAX); 164$manager->prio (PRIO_MAX);
182 165
183=back 166=back
184 167
185=head2 STATIC METHODS 168=head1 SIMPLE COROUTINE CREATION
186
187Static methods are actually functions that operate on the current coroutine only.
188 169
189=over 4 170=over 4
190 171
191=item async { ... } [@args...] 172=item async { ... } [@args...]
192 173
193Create a new asynchronous coroutine and return it's coroutine object 174Create a new coroutine and return it's coroutine object (usually
194(usually unused). When the sub returns the new coroutine is automatically 175unused). The coroutine will be put into the ready queue, so
176it will start running automatically on the next scheduler run.
177
178The first argument is a codeblock/closure that should be executed in the
179coroutine. When it returns argument returns the coroutine is automatically
195terminated. 180terminated.
196 181
182The remaining arguments are passed as arguments to the closure.
183
197See the C<Coro::State::new> constructor for info about the coroutine 184See the C<Coro::State::new> constructor for info about the coroutine
198environment in which coroutines run. 185environment in which coroutines are executed.
199 186
200Calling C<exit> in a coroutine will do the same as calling exit outside 187Calling C<exit> in a coroutine will do the same as calling exit outside
201the coroutine. Likewise, when the coroutine dies, the program will exit, 188the coroutine. Likewise, when the coroutine dies, the program will exit,
202just as it would in the main program. 189just as it would in the main program.
203 190
191If you do not want that, you can provide a default C<die> handler, or
192simply avoid dieing (by use of C<eval>).
193
204 # create a new coroutine that just prints its arguments 194Example: Create a new coroutine that just prints its arguments.
195
205 async { 196 async {
206 print "@_\n"; 197 print "@_\n";
207 } 1,2,3,4; 198 } 1,2,3,4;
208 199
209=cut 200=cut
… …
215} 206}
216 207
217=item async_pool { ... } [@args...] 208=item async_pool { ... } [@args...]
218 209
219Similar to C<async>, but uses a coroutine pool, so you should not call 210Similar to C<async>, but uses a coroutine pool, so you should not call
220terminate or join (although you are allowed to), and you get a coroutine 211terminate or join on it (although you are allowed to), and you get a
221that might have executed other code already (which can be good or bad :). 212coroutine that might have executed other code already (which can be good
213or bad :).
222 214
215On the plus side, this function is about twice as fast as creating (and
216destroying) a completely new coroutine, so if you need a lot of generic
217coroutines in quick successsion, use C<async_pool>, not C<async>.
218
223Also, the block is executed in an C<eval> context and a warning will be 219The code block is executed in an C<eval> context and a warning will be
224issued in case of an exception instead of terminating the program, as 220issued in case of an exception instead of terminating the program, as
225C<async> does. As the coroutine is being reused, stuff like C<on_destroy> 221C<async> does. As the coroutine is being reused, stuff like C<on_destroy>
226will not work in the expected way, unless you call terminate or cancel, 222will not work in the expected way, unless you call terminate or cancel,
227which somehow defeats the purpose of pooling. 223which somehow defeats the purpose of pooling (but is fine in the
224exceptional case).
228 225
229The priority will be reset to C<0> after each job, tracing will be 226The priority will be reset to C<0> after each run, tracing will be
230disabled, the description will be reset and the default output filehandle 227disabled, the description will be reset and the default output filehandle
231gets restored, so you can change alkl these. Otherwise the coroutine will 228gets restored, so you can change all these. Otherwise the coroutine will
232be re-used "as-is": most notably if you change other per-coroutine global 229be re-used "as-is": most notably if you change other per-coroutine global
233stuff such as C<$/> you need to revert that change, which is most simply 230stuff such as C<$/> you I<must needs> revert that change, which is most
234done by using local as in C< local $/ >. 231simply done by using local as in: C<< local $/ >>.
235 232
236The pool size is limited to 8 idle coroutines (this can be adjusted by 233The idle pool size is limited to C<8> idle coroutines (this can be
237changing $Coro::POOL_SIZE), and there can be as many non-idle coros as 234adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle
238required. 235coros as required.
239 236
240If you are concerned about pooled coroutines growing a lot because a 237If you are concerned about pooled coroutines growing a lot because a
241single C<async_pool> used a lot of stackspace you can e.g. C<async_pool 238single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
242{ terminate }> once per second or so to slowly replenish the pool. In 239{ terminate }> once per second or so to slowly replenish the pool. In
243addition to that, when the stacks used by a handler grows larger than 16kb 240addition to that, when the stacks used by a handler grows larger than 32kb
244(adjustable with $Coro::POOL_RSS) it will also exit. 241(adjustable via $Coro::POOL_RSS) it will also be destroyed.
245 242
246=cut 243=cut
247 244
248our $POOL_SIZE = 8; 245our $POOL_SIZE = 8;
249our $POOL_RSS = 16 * 1024; 246our $POOL_RSS = 32 * 1024;
250our @async_pool; 247our @async_pool;
251 248
252sub pool_handler { 249sub pool_handler {
253 my $cb;
254
255 while () { 250 while () {
256 eval { 251 eval {
257 while () { 252 &{&_pool_handler} while 1;
258 _pool_1 $cb;
259 &$cb;
260 _pool_2 $cb;
261 &schedule;
262 }
263 }; 253 };
264 254
265 last if $@ eq "\3async_pool terminate\2\n";
266 warn $@ if $@; 255 warn $@ if $@;
267 } 256 }
268} 257}
269 258
270sub async_pool(&@) { 259=back
271 # this is also inlined into the unlock_scheduler
272 my $coro = (pop @async_pool) || new Coro \&pool_handler;
273 260
274 $coro->{_invoke} = [@_]; 261=head1 STATIC METHODS
275 $coro->ready;
276 262
277 $coro 263Static methods are actually functions that implicitly operate on the
278} 264current coroutine.
265
266=over 4
279 267
280=item schedule 268=item schedule
281 269
282Calls the scheduler. Please note that the current coroutine will not be put 270Calls the scheduler. The scheduler will find the next coroutine that is
271to be run from the ready queue and switches to it. The next coroutine
272to be run is simply the one with the highest priority that is longest
273in its ready queue. If there is no coroutine ready, it will clal the
274C<$Coro::idle> hook.
275
276Please note that the current coroutine will I<not> be put into the ready
283into the ready queue, so calling this function usually means you will 277queue, so calling this function usually means you will never be called
284never be called again unless something else (e.g. an event handler) calls 278again unless something else (e.g. an event handler) calls C<< ->ready >>,
285ready. 279thus waking you up.
286 280
287The canonical way to wait on external events is this: 281This makes C<schedule> I<the> generic method to use to block the current
282coroutine and wait for events: first you remember the current coroutine in
283a variable, then arrange for some callback of yours to call C<< ->ready
284>> on that once some event happens, and last you call C<schedule> to put
285yourself to sleep. Note that a lot of things can wake your coroutine up,
286so you need to check whether the event indeed happened, e.g. by storing the
287status in a variable.
288 288
289 { 289See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks.
290 # remember current coroutine
291 my $current = $Coro::current;
292
293 # register a hypothetical event handler
294 on_event_invoke sub {
295 # wake up sleeping coroutine
296 $current->ready;
297 undef $current;
298 };
299
300 # call schedule until event occurred.
301 # in case we are woken up for other reasons
302 # (current still defined), loop.
303 Coro::schedule while $current;
304 }
305 290
306=item cede 291=item cede
307 292
308"Cede" to other coroutines. This function puts the current coroutine into the 293"Cede" to other coroutines. This function puts the current coroutine into
309ready queue and calls C<schedule>, which has the effect of giving up the 294the ready queue and calls C<schedule>, which has the effect of giving
310current "timeslice" to other coroutines of the same or higher priority. 295up the current "timeslice" to other coroutines of the same or higher
296priority. Once your coroutine gets its turn again it will automatically be
297resumed.
298
299This function is often called C<yield> in other languages.
311 300
312=item Coro::cede_notself 301=item Coro::cede_notself
313 302
314Works like cede, but is not exported by default and will cede to any 303Works like cede, but is not exported by default and will cede to I<any>
315coroutine, regardless of priority, once. 304coroutine, regardless of priority. This is useful sometimes to ensure
305progress is made.
316 306
317=item terminate [arg...] 307=item terminate [arg...]
318 308
319Terminates the current coroutine with the given status values (see L<cancel>). 309Terminates the current coroutine with the given status values (see L<cancel>).
320 310
… …
322 312
323Kills/terminates/cancels all coroutines except the currently running 313Kills/terminates/cancels all coroutines except the currently running
324one. This is useful after a fork, either in the child or the parent, as 314one. This is useful after a fork, either in the child or the parent, as
325usually only one of them should inherit the running coroutines. 315usually only one of them should inherit the running coroutines.
326 316
327=cut 317Note that while this will try to free some of the main programs resources,
318you cannot free all of them, so if a coroutine that is not the main
319program calls this function, there will be some one-time resource leak.
328 320
329sub terminate { 321=cut
330 $current->cancel (@_);
331}
332 322
333sub killall { 323sub killall {
334 for (Coro::State::list) { 324 for (Coro::State::list) {
335 $_->cancel 325 $_->cancel
336 if $_ != $current && UNIVERSAL::isa $_, "Coro"; 326 if $_ != $current && UNIVERSAL::isa $_, "Coro";
337 } 327 }
338} 328}
339 329
340=back 330=back
341 331
342=head2 COROUTINE METHODS 332=head1 COROUTINE OBJECT METHODS
343 333
344These are the methods you can call on coroutine objects. 334These are the methods you can call on coroutine objects (or to create
335them).
345 336
346=over 4 337=over 4
347 338
348=item new Coro \&sub [, @args...] 339=item new Coro \&sub [, @args...]
349 340
350Create a new coroutine and return it. When the sub returns the coroutine 341Create a new coroutine and return it. When the sub returns, the coroutine
351automatically terminates as if C<terminate> with the returned values were 342automatically terminates as if C<terminate> with the returned values were
352called. To make the coroutine run you must first put it into the ready queue 343called. To make the coroutine run you must first put it into the ready
353by calling the ready method. 344queue by calling the ready method.
354 345
355See C<async> and C<Coro::State::new> for additional info about the 346See C<async> and C<Coro::State::new> for additional info about the
356coroutine environment. 347coroutine environment.
357 348
358=cut 349=cut
359 350
360sub _run_coro { 351sub _terminate {
361 terminate &{+shift}; 352 terminate &{+shift};
362} 353}
363 354
364sub new {
365 my $class = shift;
366
367 $class->SUPER::new (\&_run_coro, @_)
368}
369
370=item $success = $coroutine->ready 355=item $success = $coroutine->ready
371 356
372Put the given coroutine into the ready queue (according to it's priority) 357Put the given coroutine into the end of its ready queue (there is one
373and return true. If the coroutine is already in the ready queue, do nothing 358queue for each priority) and return true. If the coroutine is already in
374and return false. 359the ready queue, do nothing and return false.
360
361This ensures that the scheduler will resume this coroutine automatically
362once all the coroutines of higher priority and all coroutines of the same
363priority that were put into the ready queue earlier have been resumed.
375 364
376=item $is_ready = $coroutine->is_ready 365=item $is_ready = $coroutine->is_ready
377 366
378Return wether the coroutine is currently the ready queue or not, 367Return whether the coroutine is currently the ready queue or not,
379 368
380=item $coroutine->cancel (arg...) 369=item $coroutine->cancel (arg...)
381 370
382Terminates the given coroutine and makes it return the given arguments as 371Terminates the given coroutine and makes it return the given arguments as
383status (default: the empty list). Never returns if the coroutine is the 372status (default: the empty list). Never returns if the coroutine is the
… …
385 374
386=cut 375=cut
387 376
388sub cancel { 377sub cancel {
389 my $self = shift; 378 my $self = shift;
390 $self->{_status} = [@_];
391 379
392 if ($current == $self) { 380 if ($current == $self) {
393 push @destroy, $self; 381 terminate @_;
394 $manager->ready;
395 &schedule while 1;
396 } else { 382 } else {
383 $self->{_status} = [@_];
397 $self->_cancel; 384 $self->_cancel;
398 } 385 }
399} 386}
387
388=item $coroutine->schedule_to
389
390Puts the current coroutine to sleep (like C<Coro::schedule>), but instead
391of continuing with the next coro from the ready queue, always switch to
392the given coroutine object (regardless of priority etc.). The readyness
393state of that coroutine isn't changed.
394
395This is an advanced method for special cases - I'd love to hear about any
396uses for this one.
397
398=item $coroutine->cede_to
399
400Like C<schedule_to>, but puts the current coroutine into the ready
401queue. This has the effect of temporarily switching to the given
402coroutine, and continuing some time later.
403
404This is an advanced method for special cases - I'd love to hear about any
405uses for this one.
406
407=item $coroutine->throw ([$scalar])
408
409If C<$throw> is specified and defined, it will be thrown as an exception
410inside the coroutine at the next convenient point in time. Otherwise
411clears the exception object.
412
413Coro will check for the exception each time a schedule-like-function
414returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down
415>>, C<< Coro::Handle->readable >> and so on. Most of these functions
416detect this case and return early in case an exception is pending.
417
418The exception object will be thrown "as is" with the specified scalar in
419C<$@>, i.e. if it is a string, no line number or newline will be appended
420(unlike with C<die>).
421
422This can be used as a softer means than C<cancel> to ask a coroutine to
423end itself, although there is no guarantee that the exception will lead to
424termination, and if the exception isn't caught it might well end the whole
425program.
426
427You might also think of C<throw> as being the moral equivalent of
428C<kill>ing a coroutine with a signal (in this case, a scalar).
400 429
401=item $coroutine->join 430=item $coroutine->join
402 431
403Wait until the coroutine terminates and return any values given to the 432Wait until the coroutine terminates and return any values given to the
404C<terminate> or C<cancel> functions. C<join> can be called concurrently 433C<terminate> or C<cancel> functions. C<join> can be called concurrently
405from multiple coroutines. 434from multiple coroutines, and all will be resumed and given the status
435return once the C<$coroutine> terminates.
406 436
407=cut 437=cut
408 438
409sub join { 439sub join {
410 my $self = shift; 440 my $self = shift;
… …
425 455
426=item $coroutine->on_destroy (\&cb) 456=item $coroutine->on_destroy (\&cb)
427 457
428Registers a callback that is called when this coroutine gets destroyed, 458Registers a callback that is called when this coroutine gets destroyed,
429but before it is joined. The callback gets passed the terminate arguments, 459but before it is joined. The callback gets passed the terminate arguments,
430if any. 460if any, and I<must not> die, under any circumstances.
431 461
432=cut 462=cut
433 463
434sub on_destroy { 464sub on_destroy {
435 my ($self, $cb) = @_; 465 my ($self, $cb) = @_;
… …
465higher values mean lower priority, just as in unix). 495higher values mean lower priority, just as in unix).
466 496
467=item $olddesc = $coroutine->desc ($newdesc) 497=item $olddesc = $coroutine->desc ($newdesc)
468 498
469Sets (or gets in case the argument is missing) the description for this 499Sets (or gets in case the argument is missing) the description for this
470coroutine. This is just a free-form string you can associate with a coroutine. 500coroutine. This is just a free-form string you can associate with a
501coroutine.
471 502
472This method simply sets the C<< $coroutine->{desc} >> member to the given string. You 503This method simply sets the C<< $coroutine->{desc} >> member to the given
473can modify this member directly if you wish. 504string. You can modify this member directly if you wish.
474
475=item $coroutine->throw ([$scalar])
476
477If C<$throw> is specified and defined, it will be thrown as an exception
478inside the coroutine at the next convinient point in time (usually after
479it gains control at the next schedule/transfer/cede). Otherwise clears the
480exception object.
481
482The exception object will be thrown "as is" with the specified scalar in
483C<$@>, i.e. if it is a string, no line number or newline will be appended
484(unlike with C<die>).
485
486This can be used as a softer means than C<cancel> to ask a coroutine to
487end itself, although there is no guarentee that the exception will lead to
488termination, and if the exception isn't caught it might well end the whole
489program.
490 505
491=cut 506=cut
492 507
493sub desc { 508sub desc {
494 my $old = $_[0]{desc}; 509 my $old = $_[0]{desc};
495 $_[0]{desc} = $_[1] if @_ > 1; 510 $_[0]{desc} = $_[1] if @_ > 1;
496 $old; 511 $old;
497} 512}
498 513
514sub transfer {
515 require Carp;
516 Carp::croak ("You must not call ->transfer on Coro objects. Use Coro::State objects or the ->schedule_to method. Caught");
517}
518
499=back 519=back
500 520
501=head2 GLOBAL FUNCTIONS 521=head1 GLOBAL FUNCTIONS
502 522
503=over 4 523=over 4
504 524
505=item Coro::nready 525=item Coro::nready
506 526
507Returns the number of coroutines that are currently in the ready state, 527Returns the number of coroutines that are currently in the ready state,
508i.e. that can be switched to. The value C<0> means that the only runnable 528i.e. that can be switched to by calling C<schedule> directory or
529indirectly. The value C<0> means that the only runnable coroutine is the
509coroutine is the currently running one, so C<cede> would have no effect, 530currently running one, so C<cede> would have no effect, and C<schedule>
510and C<schedule> would cause a deadlock unless there is an idle handler 531would cause a deadlock unless there is an idle handler that wakes up some
511that wakes up some coroutines. 532coroutines.
512 533
513=item my $guard = Coro::guard { ... } 534=item my $guard = Coro::guard { ... }
514 535
515This creates and returns a guard object. Nothing happens until the object 536This creates and returns a guard object. Nothing happens until the object
516gets destroyed, in which case the codeblock given as argument will be 537gets destroyed, in which case the codeblock given as argument will be
… …
545 566
546 567
547=item unblock_sub { ... } 568=item unblock_sub { ... }
548 569
549This utility function takes a BLOCK or code reference and "unblocks" it, 570This utility function takes a BLOCK or code reference and "unblocks" it,
550returning the new coderef. This means that the new coderef will return 571returning a new coderef. Unblocking means that calling the new coderef
551immediately without blocking, returning nothing, while the original code 572will return immediately without blocking, returning nothing, while the
552ref will be called (with parameters) from within its own coroutine. 573original code ref will be called (with parameters) from within another
574coroutine.
553 575
554The reason this function exists is that many event libraries (such as the 576The reason this function exists is that many event libraries (such as the
555venerable L<Event|Event> module) are not coroutine-safe (a weaker form 577venerable L<Event|Event> module) are not coroutine-safe (a weaker form
556of thread-safety). This means you must not block within event callbacks, 578of thread-safety). This means you must not block within event callbacks,
557otherwise you might suffer from crashes or worse. 579otherwise you might suffer from crashes or worse. The only event library
580currently known that is safe to use without C<unblock_sub> is L<EV>.
558 581
559This function allows your callbacks to block by executing them in another 582This function allows your callbacks to block by executing them in another
560coroutine where it is safe to block. One example where blocking is handy 583coroutine where it is safe to block. One example where blocking is handy
561is when you use the L<Coro::AIO|Coro::AIO> functions to save results to 584is when you use the L<Coro::AIO|Coro::AIO> functions to save results to
562disk. 585disk, for example.
563 586
564In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when 587In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
565creating event callbacks that want to block. 588creating event callbacks that want to block.
589
590If your handler does not plan to block (e.g. simply sends a message to
591another coroutine, or puts some other coroutine into the ready queue),
592there is no reason to use C<unblock_sub>.
593
594Note that you also need to use C<unblock_sub> for any other callbacks that
595are indirectly executed by any C-based event loop. For example, when you
596use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
597provides callbacks that are the result of some event callback, then you
598must not block either, or use C<unblock_sub>.
566 599
567=cut 600=cut
568 601
569our @unblock_queue; 602our @unblock_queue;
570 603
… …
573# return immediately and can be reused) and because we cannot cede 606# return immediately and can be reused) and because we cannot cede
574# inside an event callback. 607# inside an event callback.
575our $unblock_scheduler = new Coro sub { 608our $unblock_scheduler = new Coro sub {
576 while () { 609 while () {
577 while (my $cb = pop @unblock_queue) { 610 while (my $cb = pop @unblock_queue) {
578 # this is an inlined copy of async_pool 611 &async_pool (@$cb);
579 my $coro = (pop @async_pool) || new Coro \&pool_handler;
580 612
581 $coro->{_invoke} = $cb;
582 $coro->ready;
583 cede; # for short-lived callbacks, this reduces pressure on the coro pool 613 # for short-lived callbacks, this reduces pressure on the coro pool
614 # as the chance is very high that the async_poll coro will be back
615 # in the idle state when cede returns
616 cede;
584 } 617 }
585 schedule; # sleep well 618 schedule; # sleep well
586 } 619 }
587}; 620};
588$unblock_scheduler->desc ("[unblock_sub scheduler]"); 621$unblock_scheduler->{desc} = "[unblock_sub scheduler]";
589 622
590sub unblock_sub(&) { 623sub unblock_sub(&) {
591 my $cb = shift; 624 my $cb = shift;
592 625
593 sub { 626 sub {
594 unshift @unblock_queue, [$cb, @_]; 627 unshift @unblock_queue, [$cb, @_];
595 $unblock_scheduler->ready; 628 $unblock_scheduler->ready;
596 } 629 }
597} 630}
598 631
632=item $cb = Coro::rouse_cb
633
634Create and return a "rouse callback". That's a code reference that, when
635called, will save its arguments and notify the owner coroutine of the
636callback.
637
638See the next function.
639
640=item @args = Coro::rouse_wait [$cb]
641
642Wait for the specified rouse callback (or the last one tht was created in
643this coroutine).
644
645As soon as the callback is invoked (or when the calback was invoked before
646C<rouse_wait>), it will return a copy of the arguments originally passed
647to the rouse callback.
648
649See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example.
650
599=back 651=back
600 652
601=cut 653=cut
602 654
6031; 6551;
604 656
657=head1 HOW TO WAIT FOR A CALLBACK
658
659It is very common for a coroutine to wait for some callback to be
660called. This occurs naturally when you use coroutines in an otherwise
661event-based program, or when you use event-based libraries.
662
663These typically register a callback for some event, and call that callback
664when the event occured. In a coroutine, however, you typically want to
665just wait for the event, simplyifying things.
666
667For example C<< AnyEvent->child >> registers a callback to be called when
668a specific child has exited:
669
670 my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });
671
672But from withina coroutine, you often just want to write this:
673
674 my $status = wait_for_child $pid;
675
676Coro offers two functions specifically designed to make this easy,
677C<Coro::rouse_cb> and C<Coro::rouse_wait>.
678
679The first function, C<rouse_cb>, generates and returns a callback that,
680when invoked, will save it's arguments and notify the coroutine that
681created the callback.
682
683The second function, C<rouse_wait>, waits for the callback to be called
684(by calling C<schedule> to go to sleep) and returns the arguments
685originally passed to the callback.
686
687Using these functions, it becomes easy to write the C<wait_for_child>
688function mentioned above:
689
690 sub wait_for_child($) {
691 my ($pid) = @_;
692
693 my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb);
694
695 my ($rpid, $rstatus) = Coro::rouse_wait;
696 $rstatus
697 }
698
699In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough,
700you can roll your own, using C<schedule>:
701
702 sub wait_for_child($) {
703 my ($pid) = @_;
704
705 # store the current coroutine in $current,
706 # and provide result variables for the closure passed to ->child
707 my $current = $Coro::current;
708 my ($done, $rstatus);
709
710 # pass a closure to ->child
711 my $watcher = AnyEvent->child (pid => $pid, cb => sub {
712 $rstatus = $_[1]; # remember rstatus
713 $done = 1; # mark $rstatus as valud
714 });
715
716 # wait until the closure has been called
717 schedule while !$done;
718
719 $rstatus
720 }
721
722
605=head1 BUGS/LIMITATIONS 723=head1 BUGS/LIMITATIONS
606 724
607 - you must make very sure that no coro is still active on global 725=over 4
608 destruction. very bad things might happen otherwise (usually segfaults).
609 726
727=item fork with pthread backend
728
729When Coro is compiled using the pthread backend (which isn't recommended
730but required on many BSDs as their libcs are completely broken), then
731coroutines will not survive a fork. There is no known workaround except to
732fix your libc and use a saner backend.
733
734=item perl process emulation ("threads")
735
610 - this module is not thread-safe. You should only ever use this module 736This module is not perl-pseudo-thread-safe. You should only ever use this
611 from the same thread (this requirement might be loosened in the future 737module from the same thread (this requirement might be removed in the
612 to allow per-thread schedulers, but Coro::State does not yet allow 738future to allow per-thread schedulers, but Coro::State does not yet allow
613 this). 739this). I recommend disabling thread support and using processes, as having
740the windows process emulation enabled under unix roughly halves perl
741performance, even when not used.
742
743=item coroutine switching not signal safe
744
745You must not switch to another coroutine from within a signal handler
746(only relevant with %SIG - most event libraries provide safe signals).
747
748That means you I<MUST NOT> call any function that might "block" the
749current coroutine - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or
750anything that calls those. Everything else, including calling C<ready>,
751works.
752
753=back
754
614 755
615=head1 SEE ALSO 756=head1 SEE ALSO
616 757
617Lower level Configuration, Coroutine Environment: L<Coro::State>. 758Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
618 759
619Debugging: L<Coro::Debug>. 760Debugging: L<Coro::Debug>.
620 761
621Support/Utility: L<Coro::Specific>, L<Coro::Util>. 762Support/Utility: L<Coro::Specific>, L<Coro::Util>.
622 763
623Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>. 764Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>,
765L<Coro::SemaphoreSet>, L<Coro::RWLock>.
624 766
625Event/IO: L<Coro::Timer>, L<Coro::Event>, L<Coro::Handle>, L<Coro::Socket>. 767IO/Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
626 768
627Compatibility: L<Coro::LWP>, L<Coro::Storable>, L<Coro::Select>. 769Compatibility: L<Coro::LWP> (but see also L<AnyEvent::HTTP> for
770a better-working alternative), L<Coro::BDB>, L<Coro::Storable>,
771L<Coro::Select>.
628 772
629Embedding: L<Coro::MakeMaker>. 773XS API: L<Coro::MakeMaker>.
774
775Low level Configuration, Coroutine Environment: L<Coro::State>.
630 776
631=head1 AUTHOR 777=head1 AUTHOR
632 778
633 Marc Lehmann <schmorp@schmorp.de> 779 Marc Lehmann <schmorp@schmorp.de>
634 http://home.schmorp.de/ 780 http://home.schmorp.de/

Diff Legend

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